Bayesian Lasso Regression for Identifying Home and Parental Predictors of Reading Achievement: Evidence from PIRLS 2021 Türkiye

Bayesian Lasso Regression for Identifying Home and Parental Predictors of Reading Achievement: Evidence from PIRLS 2021 Türkiye | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Bayesian Lasso Regression for Identifying Home and Parental Predictors of Reading Achievement: Evidence from PIRLS 2021 Türkiye Fatih Ozkan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7474211/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Reading proficiency is foundational to future educational success and socio‑economic mobility. Despite widespread recognition that children’s reading outcomes are shaped by both the home environment and broader socio‑economic contexts, the relative contributions and complex interactions of these factors remain incompletely understood. This study analyses data from the fourth‑grade sample of the 2021 Progress in International Reading Literacy Study (PIRLS) for Türkiye. We examine associations between reading achievement and a comprehensive set of home, parental and socio‑economic variables. The data are cross‑sectional, exposures and outcomes are measured at the same time, so our findings describe associations rather than causal effects. We employ Bayesian LASSO regression to handle multicollinearity and identify a parsimonious set of variables. Consistent with PIRLS methodological guidelines, we estimate separate models for each of the five plausible values of reading achievement and pool the posterior distributions using Rubin’s rules. Analyses incorporate PIRLS student sampling weights and account for clustering at the school level through random intercepts. Posterior summaries include medians, 95% credible intervals, the probability of direction (PD)—the proportion of the posterior distribution on the median’s side of zero—and the percentage of posterior mass lying within a region of practical equivalence (ROPE) set to ± 0.1 standardised units. Findings indicate that girls outperform boys by roughly 12 points after accounting for school‑level clustering; the availability of children’s books, parental education and study supports show the clearest positive associations with reading scores. Many other variables (e.g., socio‑economic status categories, parental attitudes and school‑facing parental involvement) display wide credible intervals, PD values near 0.55–0.70 and substantial ROPE mass, signalling indeterminate associations. We discuss these results in light of theoretical frameworks and the international literature, highlight policy implications for Türkiye and other contexts, and outline directions for future research. Home literacy environment Bayesian LASSO PIRLS 2021 reading achievement Türkiye Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Being able to read fluently and understand written text is a fundamental skill that enables individuals to access knowledge, participate fully in society, and shape their own lives. Illiteracy and low literacy not only limit personal opportunities but also contribute to entrenched socioeconomic disparities within and between countries. In this regard, elementary school is a critical stage: research has demonstrated that reading proficiency by the end of third grade is strongly predictive of subsequent academic achievement, high school graduation, and even employment prospects (Lesnick et al., 2010 ). Given the importance of early literacy, educators, policymakers, and researchers have invested significant effort in understanding the factors that contribute to variation in reading outcomes. One strand of research emphasizes the influence of the home environment. A substantial body of literature shows that children who grow up in homes with plentiful reading materials, positive attitudes toward reading, and frequent shared reading sessions outperform peers lacking these resources (Alston-Abel & Berninger, 2018 ; Leseman & De Jong, 1998 ). However, access to literacy resources is shaped by structural inequalities such as parental education, income, urban or rural location, and cultural norms (Zhang, 2006 ). Another strand investigates the school environment, examining how instructional practices, school resources, and teacher quality influence literacy development. There is growing recognition that literacy outcomes result from interactions between home and school domains: schools can compensate for deficits in the home environment or magnify inequalities. Türkiye provides a compelling context for studying these issues. Although the country has made strides in expanding access to education and improving literacy rates, large disparities persist across regions and socioeconomic groups (Ataç, 2017 ). Educational reforms over the past two decades have aimed to modernize curricula, professionalize the teaching force, and provide resources to disadvantaged schools (Aksit, 2007 ). Despite these reforms, national assessments and international comparisons continue to reveal gaps in reading achievement (Baysu, 2022 ). Understanding home and parental factors associated with reading outcomes in Türkiye can inform targeted interventions and complement system-level reforms. Research in cognitive psychology notes the “Matthew effect,” whereby early success in reading leads to more reading and greater skill accumulation, while early difficulties produce a downward spiral of disengagement and skill gaps (Merga, 2019 ). By fourth grade, these trajectories are often well established (Stanovich, 2009 ). This underscores the need to identify modifiable environmental factors that could shift children onto more positive paths. Literacy is not only an individual outcome but a key indicator of human development: countries with higher average literacy rates tend to enjoy greater economic productivity, lower crime rates, better health outcomes, and more robust democratic participation (Baum & Lake, 2003 ). Consequently, international organizations such as UNESCO and the World Bank have positioned literacy promotion at the center of sustainable development goals (Wickens & Sandlin, 2007 ). In Türkiye, policy documents recognize that improving literacy is essential for competitiveness in a global knowledge economy, cultural preservation, and social cohesion. Thus, understanding the determinants of reading achievement is both an educational and societal imperative. Emergent literacy theory emphasizes that literacy development begins well before formal schooling. Children’s early exposure to language, print, and narrative shapes their phonological awareness, vocabulary, and comprehension skills (Rohde, 2015 ). Home literacy practices thus play a critical role in shaping readiness for school and subsequent achievement. However, the increasingly digital nature of children’s environments introduces new complexities. During the COVID-19 pandemic, many Turkish schools shifted to online learning, and digital devices became primary platforms for reading, homework, and communication. Access to computers, tablets, and internet connectivity became critical determinants of educational continuity (Çakmak, 2022 ). At the same time, concerns about screen time and digital distraction raised questions about the quality of reading engagement. Our study, focused on traditional measures of print exposure, must be interpreted in the context of a rapidly evolving literacy landscape that includes both print and digital mediums. Moreover, there is a growing discourse on equity in education that moves beyond resource allocation to consider cultural responsiveness and inclusion. In linguistically diverse contexts like Türkiye, where dialects and minority languages are present, literacy instruction may need adaptation to reflect students’ home languages and cultural practices (Yağmur & Akoğlu, 2016 ). Understanding the home literacy environment requires sensitivity to language use, cultural norms, and parental literacy in those languages. Interventions providing books without linguistic and cultural relevance may be less effective. Our research contributes to this nuanced understanding by examining a wide range of home and parental variables and situating findings within broader sociocultural dynamics. Qualitative studies from Turkish regions complement this quantitative perspective: Parents often view reading as a school requirement rather than leisure, with gendered beliefs discouraging boys from home reading while expecting girls to focus on chores (Ozturk et al., 2016 ). Government programs like the Fatih Project and mobile libraries aim to expand access to books and digital resources yet impacts vary across regions and depend on family engagement (Yavuzalp & Gürer, 2015 ). By considering these dynamics alongside statistical findings, we provide a holistic view of Türkiye’s home literacy environment. Theoretical Frameworks The study of home and parental influences on literacy is grounded in complementary frameworks. The Family Literacy Model posits literacy development as a social process through family interactions, emphasizing material resources like books and relational practices like shared reading (Taylor, 1983 ). It includes siblings, grandparents, and caregivers in Turkish multi-generational households, enriching exposure but potentially introducing inconsistencies (Gedik, 2025 ). Bronfenbrenner’s Ecological Systems Theory views child development within nested systems: microsystem (family, school), mesosystem (parent-teacher links), exosystem (work conditions, resources), and macrosystem (policies, norms) (Bronfenbrenner, 1979 ). In Türkiye, initiatives like Education Vision 2023 reflect macrosystem priorities, while urban-rural disparities influence the exosystem (Ministry of National Education, 2018 ). Our analysis isolates microsystem contributions while acknowledging nested operations. Cultural Capital Theory argues parents transmit capital through practices and resources, with books and reading enjoyment as proxies (Bourdieu, 1986 ; Sullivan, 2001 ). In Türkiye, fluency in Turkish confers advantages, and variables like parental education proxy cultural capital, though ethnographic work is needed for community manifestations (Yağmur & Akoğlu, 2016 ). Human Capital Theory highlights parental education as key, providing stimulating environments and support (Becker, 1964 ). Criticized for economic focus, it overlooks emotional aspects; parents with low education may still invest heavily (Marginson, 2019 ). We find strong monotonic associations but interpret alongside theories emphasizing transmission complexities. Interventions providing tools for all education levels are actionable. These frameworks underscore examining broad variables simultaneously, noting confounding and cautioning causal claims. Review of international evidence Research across countries consistently identifies the availability of books as one of the strongest home predictors of reading achievement. Evans and colleagues ( 2010 ) conducted a meta‑analysis of 27 studies and found that the number of books in the home explained substantial variance in literacy scores across socio‑economic groups. The effect was particularly pronounced at low levels of book access: moving from no books to a modest collection was associated with large gains, while differences at higher levels of access were smaller. Sénéchal and LeFevre ( 2014 ) distinguished between receptive (listening to stories) and expressive (dialogic reading) activities and concluded that both contribute to literacy but through different pathways: receptive activities build vocabulary and listening comprehension, while expressive activities foster emergent reading skills and phonemic awareness. Evidence on parental education and socio‑economic status is also robust. A cross‑national analysis of PIRLS 2011 data showed that parental education was positively associated with reading achievement in all participating countries, but the strength of the association differed substantially: it was strongest in countries with greater income inequality and weaker in countries with comprehensive social welfare systems. Studies in Scandinavia, for instance, report that differences in reading outcomes by parental education are relatively small, whereas studies in Brazil, Chile and South Africa find large gaps. Parental involvement findings are more mixed. Boonk et al. ( 2018 ) reviewed 75 studies and concluded that home‑based involvement (e.g., reading together) had a stronger and more consistent positive association with academic outcomes than school‑based involvement (e.g., attending meetings). However, Hill and Tyson ( 2009 ) found that the effectiveness of parental involvement depends on children’s age and the type of involvement. They argued that autonomy‑supportive behaviours (e.g., encouraging independent reading) were more beneficial than controlling behaviours (e.g., monitoring homework), particularly in adolescence. These nuances underscore the need to consider the content and quality of parental engagement rather than frequency alone. Empirical findings in Türkiye Research on reading achievement in Türkiye is less extensive than in some other countries, but it is growing. Gür et al. ( 2020 ) analysed the 2016 PIRLS data and found that parental education and the number of books at home were positively associated with reading scores, even after controlling for school characteristics. They also noted that parental support variables (e.g., helping with homework) showed weaker associations. A qualitative study by Sad ( 2012 ) highlighted cultural practices of storytelling within extended families and community reading groups that may complement or substitute for parental involvement. Additionally, Toran and Özgen ( 2018 ) documented regional disparities in access to libraries and reading materials; children in rural areas often rely on mobile libraries or school libraries, which may mitigate the absence of home books. These findings suggest that the role of home literacy resources may differ across contexts within Türkiye. Methodological contributions of the present study Previous quantitative studies in Türkiye and elsewhere often use ordinary least squares regression or logistic regression, including a limited number of predictors. Such approaches can suffer from multicollinearity and overfitting when many correlated variables are included. Few studies have exploited the full richness of PIRLS data using modern regularization techniques. Our study contributes methodologically by applying Bayesian LASSO regression to PIRLS 2021 data, incorporating sampling weights and clustering and correctly handling plausible values (Tibshirani, 1996 ; Park & Casella, 2008 ; Hans, 2009 ; Yin et al., 2023 ). The Bayesian framework provides full posterior distributions, allowing us to assess both effect direction and practical importance using PD and ROPE indices. For this purpose, the following research questions guided the study: What are the key home and parental factors associated with fourth-grade reading achievement in Türkiye, and how do they vary in strength after accounting for multicollinearity? To what extent do socioeconomic and cultural variables interact with home literacy resources in predicting reading outcomes among Turkish students, and what are the implications for targeted interventions? Methods Data Analysis PIRLS 2021 design and sample weights The International Association for the Evaluation of Educational Achievement (IEA) developed PIRLS to monitor reading achievement trends. In each participating country, a target population of fourth‑grade students is defined, and schools are selected with probability proportional to size from strata based on geographic region, school type, and language of instruction. Within schools, one or two fourth‑grade classes are randomly sampled. Student weights compensate for unequal probabilities of selection and non‑response. Following IEA guidelines, our analysis uses the student weight as a probability weight and accounts for clustering by specifying schools as random intercepts. We also conducted a sensitivity analysis including class‑level clustering (cross‑classified with schools), but found that random intercepts at the class level contributed little additional variance relative to the school level. To contextualize the sample, the 2021 cycle of PIRLS involved 57 participating countries and education systems. In Türkiye, the sample frame comprised roughly 11,000 primary schools stratified by geographic region, urbanicity and school type. From this frame, 166 schools were selected with probability proportional to enrolment, and 176 fourth‑grade classes were sampled within these schools, yielding 5,284 student participants. After accounting for student and school non‑response, the analytic sample consisted of 4,883 students. Sampling weights adjust for differential selection probabilities at each stage and for non‑response, ensuring that estimates generalize to the population of fourth‑grade students. Stratification variables (region, urbanicity, school type) enhance the precision of estimates but make the sample design complex. We therefore use multilevel models with random intercepts to capture between‑school variance; the intraclass correlation coefficient (ICC) for reading achievement was about 0.113 (range 0.107–0.116), indicating that roughly 11% of variance lies between schools. Including a random intercept at the class level added little additional variance (ICC ≈ 0.02), so we focus on the school level. Although PIRLS provides replicate weights for variance estimation via jackknife or balanced repeated replication, we did not incorporate them directly because our Bayesian multilevel model accounts for sampling weights and clustering. Nevertheless, we compared our variance estimates with those obtained using replicate weights in a design‑based analysis and found similar magnitudes, suggesting robustness. Future work could explore Bayesian methods that integrate replicate weights into the likelihood function. Achievement scaling and plausible values PIRLS uses item response theory (IRT) scaling to place student scores on a common scale (Bezirhan et al., 2023 ). Students answer a subset of items from a larger pool; this matrix sampling design reduces test burden but introduces missingness in item responses (Foy & Yin, 2017 ). IRT modelling produces proficiency estimates and associated uncertainty (Bezirhan et al., 2023 ). To reflect this uncertainty in public use files, PIRLS provides multiple plausible values for each student (von Davier et al., 2009 ). Each plausible value is a random draw from the posterior distribution of the student’s proficiency given their item responses and background variables (von Davier et al., 2009 ). Using a single plausible value or their mean would underestimate standard errors and bias inference, particularly for covariate analyses (Wu, 2005 ). Therefore, the IEA recommends fitting the model separately for each plausible value and pooling results using Rubin’s combination rules (Yin et al., 2023 ). The rationale for plausible values is grounded in the principle that secondary analyses should reflect measurement uncertainty inherent in largescale assessments (von Davier et al., 2009 ). Because each student answers only a subset of test items, a single maximumlikelihood estimate of ability would be overly precise (Laukaityte & von Davier, 2021 ). By drawing multiple plausible values from the posterior distribution of ability, we propagate measurement error into subsequent analyses (von Davier et al., 2009 ). For each plausible value, we fit a separate regression model and combine estimates using Rubin’s rules: the pooled estimate is the average of the individual estimates, and the total variance is the sum of the withinimputation variance and the betweenimputation variance (Rubin, 1987 ). This procedure ensures that both sampling variability and imputation uncertainty are reflected in credible intervals (Rubin, 1987 ). We concatenated posterior draws from all plausible values and multiple imputations to approximate the pooled posterior distribution (Scharl & Nestler, 2021 ). Although some researchers advocate for joint models that estimate item parameters, abilities, and regression coefficients simultaneously, such approaches require restricted use ofitem-level data and sophisticated modelling (Scharl & Nestler, 2021 ). Given our use of publicly available data, we adhered to the standard practice of analyzing plausible values separately (Yin et al., 2023 ). Sensitivity analyses comparing our pooled results with those from a joint model (using a smaller dataset with available item responses) yielded similar substantive conclusions (Scharl & Nestler, 2021 ). Predictor variables and coding We assembled a comprehensive set of predictors from the student, parent and school questionnaires. Table 1 lists all variables, their coding schemes and descriptive statistics. Here we describe key categories: Table 1 Descriptive statistics for the analytic sample (PIRLS 2021 Türkiye) Dataset variable Variable (label) Level % ASBH13 Children’s books at home > 200 28 101–200 23 51–100 23 26–50 12 11–25 8 0–10 6 ASBH12 Non-children’s books at home > 200 28* 101–200 23* 51–100 23* 26–50 12* 11–25 8* 0–10 6* ASDGHRL Home resources index Many 40 Some 42 Few 18 ATBR16 Homework reading time (per day) ≤ 15 minutes 15 16–30 minutes 36 31–60 minutes 28 > 60 minutes 21 ASDG05S (categorized) Study supports Low 18† Moderate 50† High 32† ASDHEDUP Parental education Some primary or no school 11 Lower secondary 23 Upper secondary 29 Post-secondary (non-university) 15 University or higher 22 ASDHSES Household SES (self-report) Lower 38 Middle 40 Upper 22 ACBG03A School composition: economically disadvantaged > 50% (Yes) 24 ACBG03B School composition: economically affluent > 50% (Yes) 16 ASDHPLR Parents like reading Very much 63 Somewhat 27 Do not like 10 ASDGSLR Students like reading Very much 58 Somewhat 32 Do not like 10 Home‑literacy resources. The number of children’s books and non‑children’s books at home were assessed separately; both variables had six categories representing increasing ranges of book counts. We treat the highest category (> 200 books) as the reference. Home resources (few, some, many) measure the availability of educational materials such as a desk, dictionaries and reading technology. Homework reading time captures how long students spend reading outside of school (four categories). Study supports are based on parent responses about the availability of quiet study space, materials and parental assistance; categories are low, moderate or high. Socio‑economic indicators. Parental education is captured in five ordered categories, from no schooling/primary school to university or higher. SES is a composite measure of household possessions and services; the PIRLS database provides a categorical variable representing lower, middle or upper SES. School economic composition is derived from school‑administrator reports of the percentage of students considered economically disadvantaged or affluent; we created categories for schools where > 50% of students are disadvantaged or affluent. Attitudes and engagement. Parents’ enjoyment of reading and students’ enjoyment of reading are both measured on a three‑point Likert scale (do not like, somewhat like, very much like). Parental engagement variables include the frequency of communicating with teachers, monitoring homework, and participating in school events. These variables are categorical with two to three levels. We also include a gender indicator. All predictor variables were treated in their raw or categorical forms to preserve original scales and facilitate direct interpretation in the context of PIRLS reading achievement scores. Continuous or interval-like variables, such as study supports (ASDG05S), were not standardized, so coefficients represent changes in raw score units rather than standardized deviations. This approach maintains the substantive meaning of effects on the PIRLS scale (international mean of 500, SD of 100) but may limit direct comparisons across variables with different metrics. Categorical variables were coded as factors with descriptive labels and reference levels chosen to reflect theoretical baselines (e.g., highest category for books and education as reference). This uses R's default treatment contrasts, where coefficients indicate differences from the reference level, which can introduce some multicollinearity among dummy variables but is mitigated by the LASSO regularization. For ordered categories (e.g., parental education, book counts), this coding captures cumulative deficits relative to the highest level, though it does not explicitly test incremental trends via specialized contrasts. For nominal variables without inherent order, such as gender or SES categories, the same treatment coding applies, with estimates as deviations from the reference. In preliminary analyses, we considered alternative coding, such as raw numeric treatment for categorical or collapsing adjacent categories when sample sizes were small. For instance, we tested combining the 101–200 and > 200 book categories, but this led to loss of information about potential nonlinearities. We also explored creating composite indices from principal components or factor analysis (e.g., combining study supports, homework time, and home resources), but we found that individual components had distinct associations with reading and that composites obscured these nuances. Therefore, we retained separate variables to maintain granularity. These coding decisions highlight the importance of aligning statistical procedures with theoretical expectations and substantive interpretability, prioritizing simplicity and direct reference comparisons in the Bayesian multilevel framework. Data Analysis Bayesian LASSO model specification Bayesian LASSO applies a Laplace prior to each regression coefficient, which induces shrinkage towards zero and effectively performs variable selection (Park & Casella, 2008 ). In our hierarchical model, the response for student i in school j is modeled as $$\:{y}_{ij}=\:{\sum\:}_{k=1}^{K}{X}_{\left(ijk\right){\beta\:}_{k}}+\:{u}_{j}+\:{ϵ}_{ij}$$ where \(\:{y}_{ij}\) is the outcome, \(\:{X}_{\left(ijk\right)}\) are predictor values, \(\:{\beta\:}_{k}\) have Laplace priors, \(\:{u}_{j}\) ~ N(0, τ²) are school-level random effects, and \(\:{ϵ}_{ij}\) ~ N(0, σ²) are residual errors (Li & Lin, 2010 ). The global shrinkage parameter λ controls the degree of shrinkage; we place an exponential prior on λ with mean λ₀ (Hans, 2009 ). We set λ₀ = 0.1 and conduct sensitivity analyses. Variance components τ² and σ² receive half-Student-t priors with 3 df and scale 1 (Gelman, 2006 ). Priors are weakly informative, allowing data to drive inference while providing regularization. Models were fitted in Stan via the RStan interface (Stan Development Team, 2023 ). For each plausible value and imputed dataset, we ran four chains of 5000 iterations, discarding the first 2000 as warm-up. We assessed convergence using Gelman–Rubin R̂ statistics (R̂ 400) and trace plots (Gelman & Rubin, 1992 ). Posterior predictive checks indicated satisfactory model fit (Gelman et al., 1996 ). We pooled posterior samples across plausible values and imputations by concatenating draws and applying Rubin’s rules for variance components (Rubin, 1987 ). The choice of a Laplace prior corresponds to the Bayesian analogue of the L1 penalty in frequentist LASSO. It exerts a constant pressure toward zero on all coefficients, resulting in some being exactly or nearly zero. This property is beneficial when the number of predictors is large relative to the sample size or when multicollinearity is present, as it reduces overfitting and improves out-of-sample predictive performance. However, the shrinkage is uniform across coefficients, meaning that truly large effects may also be pulled toward zero more than desirable. The horseshoe prior, by contrast, uses a global local shrinkage structure with heavy tails, allowing large signals to be less shrunk. In addition, the spike and slab prior explicitly models a mixture of zero and non-zero coefficients, providing probabilistic variable selection (Bhadra et al., 2019 ). We tested these alternative priors and found that substantive conclusions were consistent, though some additional small effects emerged under the horseshoe and spike and slab, reinforcing the robustness of our findings. Choosing hyperparameters such as the mean of the exponential prior on λ involves a trade-off between shrinkage and model complexity. A smaller λ₀ induces stronger shrinkage, leading to sparser models, while a larger value allows more variables to remain in the model. Our sensitivity analysis varied λ₀ from 0.05 to 0.5; the relative ranking of key predictors remained stable, but the exact magnitude of coefficients changed (Li et al., 2011 ). To ensure reproducibility, we report our chosen hyperparameters and provide code in the Supplementary Materials. We also experimented with cross-validation to select λ , but cross-validation is more natural in a frequentist context; in the Bayesian framework, prior and posterior predictive checks provide analogous information. Posterior predictive checks using leave-one-out cross-validation (LOO-CV) confirmed good model fit (Vehtari et al., 2017 ). Interpretation indices: PD and ROPE Traditional frequentist analyses often rely on p-values and significance thresholds. In the Bayesian framework, p-values do not exist; instead, we summarize the posterior distribution (Makowski et al., 2019 ). The probability of direction (PD) is the proportion of posterior samples that have the same sign as the posterior median. PD ranges from 50% (complete uncertainty) to 100% (complete certainty). It provides an intuitive measure of how confident we are that the effect is positive or negative (Makowski et al., 2019 ). The region of practical equivalence (ROPE) is an interval around zero deemed practically negligible. By default, we set the ROPE to ± 0.1 standardized units, following guidelines for small effect sizes (Makowski et al., 2019 ). The percentage of posterior samples within the ROPE indicates whether the effect is practically equivalent to zero (Kruschke, 2018). By considering PD and ROPE together, we avoid binary significance decisions and instead convey graded evidence (Makowski et al., 2019 ). Missing data and multiple imputation Missing data are unavoidable in large-scale assessments. In our dataset, the proportion of missing values ranged from < 1% for gender to about 8% for parental engagement variables. Assuming data are missing at random, we used multiple imputation by chained equations (MICE) to impute missing values (MAR conditional on observed covariates; Little & Rubin, 2002 ; van Buuren & Groothuis-Oudshoorn, 2011 ). The imputation model included all variables used in the analysis, the outcome plausible values (PV), and school identifiers and used type-appropriate conditional models (predictive mean matching for continuous variables, logistic regression for binary indicators, and proportional-odds models for ordered categorical items). To better respect the multilevel structure, we also included cluster indicators and key cluster-level means, and we mirrored any interactions/nonlinear terms from the analysis model to maintain congeniality (Meng, 1994 ; White, Royston, & Wood, 2011 ). We generated five imputed datasets, consistent with recommendations that the number of imputations roughly match the percentage of missingness (≈ 5–10 when overall missingness is ~ 8%; White et al., 2011 ). We checked the convergence of imputation chains and inspected distributions of imputed values for plausibility (examining chain histories and overlays of observed vs. imputed distributions). Analyses were conducted in R using rstan, mice, loo, bayesplot, and tidybayes. Seeds, code, and output summaries are provided in the Supplementary Materials to ensure full reproducibility. Results Descriptive statistics Before presenting model results, we describe the distributions of key variables. Table 1 provides descriptive statistics for home literacy resources. Approximately 28% of students reported having more than 200 children’s books at home. By contrast, 6% reported having 0–10 children’s books, 8% 11–25 books, 12% 26–50 books, 23% 51–100 books and 23% 101–200 books. The distribution of non‑children’s books was similar. Around 40% of students reported “many” home resources, 42% reported “some” and 18% reported “few.” In terms of homework reading time, 15% reported spending ≤ 15 minutes per day on reading homework, 36% 16–30 minutes, 28% 31–60 minutes and 21% >60 minutes. Study supports were distributed as 18% low, 50% moderate and 32% high. Parental education levels were: 11% some primary or no school, 23% lower secondary, 29% upper secondary, 15% post‑secondary non‑university and 22% university or higher. Household SES categories were 38% lower, 40% middle and 22% upper. Approximately 24% of students attended schools where more than half of their peers were considered economically disadvantaged; conversely, 16% attended schools where more than half were affluent. Attitudinal variables indicated generally positive outlooks on reading: 63% of parents and 58% of students reported enjoying reading “very much,” while 27% and 32% reported “somewhat,” and 10% and 10% reported “do not like,” respectively. Parental engagement behaviours showed moderate variation: 46% of parents reported high frequency of communication with teachers, 44% moderate, and 10% low; similar distributions were observed for homework monitoring. Correlations among predictors were moderate, reflecting relationships such as the positive association between parental education and the number of books, and justified the use of regularisation to address multicollinearity. Descriptive patterns showed persisting regional, socio‑economic and gender disparities in resource availability and reading enjoyment, underscoring the importance of controlling for these factors in multivariate analyses. Associations with reading achievement Bayesian multilevel LASSO models were fitted to each of the five plausible values (PVs) of reading achievement (ASRREA01–ASRREA05) from the PIRLS 2021 Türkiye dataset, incorporating a random intercept for schools (IDSCHOOL) to account for hierarchical clustering. Laplace priors were applied to regularize fixed effects, promoting shrinkage of less influential predictors toward zero. Model diagnostics, including trace plots and R-hat values (< 1.1), confirmed convergence and good mixing across all models. Posterior summaries for fixed effects from individual PV models are presented in Table 2 , focusing on non-reference categories for factor variables (reference levels were set to estimate = 0 with no credible intervals or standard errors). To account for uncertainty in the PVs, posterior draws from each model were pooled by concatenating them across PVs, yielding combined posterior distributions for each predictor. Summaries of these pooled posteriors, including median estimates, 95% highest posterior density intervals (HPDIs), probability of direction (PD), and posterior standard deviations, are provided in Table 1 . This pooling approach incorporates both within-model and between-PV variability, serving as a Bayesian analog to Rubin's rules for multiple imputations. Predictors were deemed credible if PD > .95 and HPDIs excluded zero. Table 2 Pooled Bayesian LASSO coefficients for predictors of reading achievement Predictor term Median estimate 95% CI PD Posterior SD Gender (Boy vs. Girl) –14.03 [–18.38, –9.38] 1.00 2.31 Economically Disadvantaged (11–25% vs. 0–10%) –3.78 [–18.65, 11.19] 0.69 7.57 Economically Disadvantaged (26–50% vs. 0–10%) –6.38 [–23.37, 10.74] 0.77 8.75 Economically Disadvantaged (More than 50% vs. 0–10%) –20.46 [–38.38, –2.63] 0.99 9.07 Economically Affluent (11–25% vs. 0–10%) 5.77 [–6.42, 18.22] 0.82 6.32 Economically Affluent (26–50% vs. 0–10%) 11.23 [–2.28, 25.01] 0.95 6.94 Economically Affluent (More than 50% vs. 0–10%) 11.56 [–5.69, 29.40] 0.90 8.98 Home Resources (Some vs. Many) 8.24 [0.10, 16.50] 0.98 4.22 Home Resources (Few vs. Many) 21.89 [8.62, 35.14] 1 .00 6 .75 SES (Middle vs. Upper) 1.50 [–6.06, 8.96] 0.65 3.82 SES (Lower vs. Upper) –1.00 [–15.34, 13.23] 0.55 7.26 Parents Like Reading (Somewhat vs. Very Much) –9.75 [–14.65, –4.89] 1.00 2.48 Parents Like Reading (Do Not Like vs. Very Much) –19.15 [–26.09, –12.18] 1.00 3.54 Students Like Reading (Somewhat vs. Very Much) –11.92 [–16.49, –7.28] 1.00 2.36 Students Like Reading (Do Not Like vs. Very Much) –17.29 [–27.57, –7.99] 1.00 4.99 Parent Education Level (Post‑secondary, not university vs. university or higher) –13.21 [–22.86, –4.52] 0.999 4.68 Parent Education Level (Upper secondary vs. university or higher) –22.44 [–30.48, –14.42] 1.00 4.10 Parent Education Level (Lower secondary vs. university or higher) –38.92 [–49.12, –29.16] 1.00 5.13 Parent Education Level (Some primary/lower secondary or no school vs. university or higher) –46.22 [–56.32, –36.21] 1.00 5.14 Numbers of Books at Home (0–10 vs. >200) –8.77 [–20.81, 3.12] 0.93 6.07 Numbers of Books at Home (11–25 vs. >200) –2.02 [–12.04, 7.85] 0.66 5.07 Numbers of Books at Home (26–100 vs. >200) 0.74 [–8.29, 9.85] 0.56 4.63 Numbers of Books at Home (101–200 vs. >200) 8.54 [–0.88, 17.98] 0.96 4.80 Numbers of Children’s Books (0–10 vs. >200) –32.58 [–44.08, –21.26] 1.00 5.79 Numbers of Children’s Books (11–25 vs. >200) –28.73 [–38.32, –18.95] 1.00 4.94 Numbers of Children’s Books (26–50 vs. >200) –14.69 [–23.07, –6.38] 1.00 4.27 Numbers of Children’s Books (51–100 vs. >200) –12.34 [–20.07, –4.77] 0.999 3.93 Homework Reading Time (≤ 15 min vs. >60 min) –10.48 [–32.04, 10.13] 0.83 1 0.80 Homework Reading Time (16–30 min vs. >60 min) –9.09 [–22.63, 3.82] 0.91 6.72 Homework Reading Time (31–60 min vs. >60 min) –5.89 [–19.68, 7.36] 0.81 6.90 Study Supports (Moderate vs. Low) 16.71 [7.95, 25.28] 1.00 4.43 Study Supports (High vs. Low) 17.74 [8.76, 26.52] 1.00 4.52 Note. Reference levels for each categorical predictor are as follows: Gender = Girl; Economically Disadvantaged = 0–10%; Economically Affluent = 0–10%; Home Resources = Many Resources; SES = Upper; Parents Like Reading = Very Much Like; Students Like Reading = Very Much Like Reading; Parent Edu Level = University or Higher; Numbers of Books at Home = More than 200; Numbers of Children’s Books = More than 200; Homework Reading Time = More than 60 minutes; Study Supports = Low Support. Key findings from the pooled analysis highlight several critical predictors (see Table 2 ). Student gender emerged as a consistent negative predictor, with boys scoring approximately 14 points lower than girls (median estimate = -14.03, 95% CI [-18.38, -9.38], PD = 1.00). Parental education levels showed strong negative associations relative to the reference category (university or higher), particularly for lower secondary education (median estimate = -38.92, 95% CI [-49.12, -29.16], PD = 1.00) and some primary/lower secondary or no school (median estimate = -46.22, 95% CI [-56.32, -36.21], PD = 1.00). Students' attitudes toward reading were also significant, with those who "somewhat like" reading scoring lower (median estimate = -9.75, 95% CI [-14.65, -4.89], PD = 1.00) and those who "do not like" reading scoring even lower (median estimate = -19.15, 95% CI [-26.09, -12.18], PD = 1.00) compared to those who "very much like" reading. Home resources for learning (ASDGHRL) displayed counterintuitive positive associations for lower resource levels relative to "many resources," with "some resources" (median estimate = 8.24, 95% CI [0.10, 16.50], PD = 0.98) and "few resources" (median estimate = 21.89, 95% CI [8.62, 35.14], PD = 1.00) linked to higher scores, potentially indicating compensatory effects or context-specific dynamics in the Turkish sample. School composition variables (ACBG03A and ACBG03B) showed mixed effects, with higher proportions of economically disadvantaged students (ACBG03AMore than 50%) negatively associated with achievement (median estimate = -20.46, 95% CI [-38.38, -2.63], PD = 0.99). Home possessions of books (ASBH13) were strongly negative for lower categories, such as 0–10 books (median estimate = -32.58, 95% CI [-44.08, -21.26], PD = 1.00). The region of practical equivalence (ROPE) analysis (see Table 3 ), using a range of [-0.1, 0.1], identified several predictors with low practical significance (high % in ROPE), such as SES Lower (1.16%) and ACBG03A11–25% (0.99%), suggesting these may have negligible effects despite their estimates. Figure 1 illustrates pooled posterior medians and 95% HPDIs, ordered by magnitude. Additional diagnostics (e.g., PD plots; Fig. 2 ) confirmed directional consistency for key predictors. Figure 2 shows the PD for each pooled coefficient, with points colored by the median estimate (red for negative, green for positive). The dashed line at PD = 0.95 indicates strong evidence for direction; most socioeconomic (e.g., parental education, home books) and attitudinal factors exceed this threshold with negative effects, while lower home resources show positive directionality, highlighting their credible influence on reading achievement. Table 3 Pooled ROPE analysis for Bayesian LASSO coefficients Predictor term Median estimate % of posterior in ROPE Gender (Boy vs. Girl) –13.34 0.00 Economically Disadvantaged (11–25% vs. 0–10%) –4.17 1.28 Economically Disadvantaged (26–50% vs. 0–10%) –6.15 0.90 Economically Disadvantaged (More than 50% vs. 0–10%) –18.95 0.00 Economically Affluent (11–25% vs. 0–10%) 0.98 2.45 Economically Affluent (26–50% vs. 0–10%) 9.29 0.00 Economically Affluent (More than 50% vs. 0–10%) 5.32 0.84 Home Resources (Some vs. Many) 11.71 0.00 Home Resources (Few vs. Many) 26.27 0.00 SES (Middle vs. Upper) 0.04 1.99 SES (Lower vs. Upper) –1.41 1.11 Parents Like Reading (Somewhat vs. Very Much) –10.00 0.00 Parents Like Reading (Do Not Like vs. Very Much) –24.63 0.00 Students Like Reading (Somewhat vs. Very Much) –12.43 0.00 Students Like Reading (Do Not Like vs. Very Much) –19.22 0.00 Parent Education Level (Post‑secondary, not university vs. university or higher) –13.46 0.00 Parent Education Level (Upper secondary vs. university or higher) –23.93 0.00 Parent Education Level (Lower secondary vs. university or higher) –41.59 0.00 Parent Education Level (Some primary/lower secondary or no school vs. university or higher) –50.95 0.00 Numbers of Books at Home (0–10 vs. >200) –9.51 0.45 Numbers of Books at Home (11–25 vs. >200) 1.05 1.66 Numbers of Books at Home (26–100 vs. >200) 3.36 1.44 Numbers of Books at Home (101–200 vs. >200) 9.24 0.36 Numbers of Children’s Books (0–10 vs. >200) –39.77 0.00 Numbers of Children’s Books (11–25 vs. >200) –33.56 0.00 Numbers of Children’s Books (26–50 vs. >200) –17.28 0.00 Numbers of Children’s Books (51–100 vs. >200) –11.74 0.00 Homework Reading Time (≤ 15 min vs. >60 min) –12.92 0.36 Homework Reading Time (16–30 min vs. >60 min) –10.11 0.23 Homework Reading Time (31–60 min vs. >60 min) –8.43 0.44 Study Supports (Moderate vs. Low) 16.71 0.00 Study Supports (High vs. Low) 17.74 0.00 Predictors such as parental education levels (e.g., ASDHEDUPSome primary/lower secondary or no school: 0%) and home books (e.g., ASBH130–10: 0%) have near-zero percentages in ROPE, confirming their practical significance, whereas non-children’s books at home (e.g. ASBH1211–25) (1.49%) and household SES (e.g. ASDHSESLower) (1.16%) have higher percentages, suggesting effects that are practically equivalent to zero despite non-zero medians (see Fig. 3 ). The comparison of coefficient estimates and 95% CIs from Bayesian LASSO versus non-regularized multilevel models for a representative PV, demonstrating the shrinkage effect of LASSO are provided in Fig. 4 : coefficients are pulled toward zero, with reduced variance (narrower CIs) for less influential predictors (e.g., homework time categories shrink more than parental education). This highlights LASSO's role in mitigating overfitting, as non-regularized estimates show wider variability and larger magnitudes for socioeconomic factors like low parental education (-60 to -40 range in non-regularized vs. -50 to -30 in LASSO). We re‑examined moderation patterns under the multilevel specification. Qualitative patterns mirror the single‑level exploratory checks (e.g., benefits of children’s books and study supports may be larger for students with less‑educated parents), but credible intervals largely overlap and most PD values do not exceed 0.95. We therefore treat moderation evidence as suggestive rather than definitive. Random‑intercept multilevel estimation (schools) produced an average ICC of 0.113 (range 0.107–0.116). Key signals (children’s books, parental education, gender gap) were robust across plausible‑value pooling and prior choices; direction and magnitude were similar in a non‑regularised multilevel baseline. Posterior predictive checks indicated reasonable fit. Results using ROPE show that many smaller coefficients carry substantial mass within the ± 0.1 SD region, aligning with the wide CIs reported in the pooled table. Sensitivity analyses and robustness checks To assess the robustness of our findings, we conducted the following analyses: 1. Alternative shrinkage priors . We replaced the Laplace prior with a horseshoe prior. The horseshoe prior is less aggressive in shrinking small coefficients and allows for a few large coefficients (Piironen & Vehtari, 2017 ). Results were broadly similar: book access, parental education and study supports remained the strongest predictors. A few additional variables (e.g., parental communication) had nonzero medians, but PD values remained below 0.75, indicating weak evidence. 2. Different ROPE widths . We varied the ROPE width from ± 0.05 to ± 0.2. Narrower ROPEs increased the number of predictors deemed practically significant, while wider ROPEs reduced it. However, the relative ordering of effects remained unchanged (Kruschke, 2018). 3. Including additional schoollevel covariates . We tested models including schoollevel variables such as average class size, school resources and teacher qualifications. These variables had small effects and did not alter the coefficients of home variables, suggesting that home influences operate independently of these school characteristics (Mullis et al., 2023 ). 4. Unweighted models . Running unweighted models yielded slightly different estimates: effects of book access and parental education were attenuated, while those of SES and school composition became marginally larger. This underscores the importance of weights for population inference (Rutkowski et al., 2010 ). 5. Excluding outliers. We excluded students with extremely low or high reading scores (2.5% in each tail). Results were essentially unchanged, indicating that extreme scores were not driving the associations (Barnett & Lewis, 1994 ). Discussion Synthesis of findings This study presents a comprehensive examination of home, parental and socioeconomic predictors of reading achievement among Turkish fourthgraders. By applying Bayesian LASSO regression, we handle multicollinearity among predictors and obtain interpretable posterior distributions (Hans, 2009 ). We summarize the direction and magnitude of associations using PD and ROPE, providing a nuanced picture of which factors show clear, suggestive, or indeterminate relationships with reading (Makowski et al., 2019 ). Clear associations. Three sets of predictors stand out: (1) the number of children’s books, (2) parental education and (3) study supports. These variables exhibit PD values ≥ 0.95 and negligible posterior mass in the ROPE, indicating strong evidence of positive associations (Makowski et al., 2019 ). Effect sizes are substantial: having very few children’s books is associated with a 0.3–0.5 SD deficit and moving from universityeducated parents to parents with no schooling is associated with a 0.5 SD deficit. High study supports correspond to a + 0.18 SD gain (Bozkuş, 2025 ). From a policy perspective, these factors represent promising levers: increasing access to children’s books and providing study supports could yield meaningful improvements in reading achievement, particularly for children of less educated parents (Bozkuş, 2025 ). Suggestive associations. Variables such as moderate study supports, the “somewhat like reading” categories for parents and students, and a few interaction terms exhibit PD values between 0.60 and 0.90 and low ROPE percentages. These effects may be real but require further evidence (Makowski et al., 2019 ). For example, the incremental benefit of moving from few to some home resources is modest, and the effect of reading enjoyment appears stronger when dislike is contrasted with strong liking rather than moderate liking (Bozkuş, 2025 ). Indeterminate associations. Many variables—including SES categories, school economic composition and parental engagement behaviours—display PD values close to 0.50–0.70 and substantial posterior mass in the ROPE. These findings imply that, after controlling for more proximal homeliteracy resources and parental education, these variables provide little additional predictive power (Bozkuş, 2025 ). This does not mean that socioeconomic factors or engagement are unimportant; rather, their influence may be captured by more proximal variables, or the measures used may be insufficiently sensitive (Aikens & Barbarin, 2008 ). Summarising posterior distributions using PD and ROPE emphasises effect sizes and the degree of uncertainty rather than binary significance, consistent with a growing emphasis on estimation in statistical practice (Cumming, 2014 ). Our findings also suggest that microlevel interventions such as providing books and study supports can mitigate socioeconomic disadvantages, but we recognise that our measures may not capture all dimensions of home literacy and that qualitative work is needed to understand underlying mechanisms (Buckingham et al., 2013 ). Implications for theory Our findings support the Family Literacy Model and cultural capital theory, which highlight the importance of tangible literacy resources and parental education for children’s literacy development (Ihmeideh & Al-Maadadi, 2020 ; Bourdieu, 1986 ). The diminishing returns observed beyond 50 children’s books are consistent with the notion that a threshold level of print exposure is necessary for literacy growth, after which the quality and diversity of materials may matter more than quantity. The strong association of parental education underscores human capital theory: parents with higher education likely possess greater literacy skills and knowledge about educational systems, enabling them to support their children more effectively (Fischer & Lipovská, 2013 ). The weak associations for socioeconomic status and parental engagement may appear to contradict ecological systems theory, which emphasizes the interplay of multiple contextual influences (Şengönül, 2022 ). However, these results may reflect measurement limitations. For example, the SES composite may not capture important dimensions such as parental income stability, wealth or employment conditions (Dickinson & Adelson, 2014 ). Parental engagement variables may measure frequency rather than quality; supportive involvement (e.g., reading together) could differ markedly from controlling involvement (e.g., monitoring homework), but our measures do not distinguish these nuances (Şengönül, 2022 ). Future research should develop more nuanced measures to better test ecological and cultural capital theories. Literature Support for Discussion Points Several large-scale studies report that girls outperform boys in reading. For example, OECD ( 2019 ) found that across OECD countries in PISA 2018, girls scored on average about 30 points higher than boys in reading. Lundberg ( 2020 ) likewise notes a persistent gap favoring girls: in an OECD survey of 15-year-olds, “boys are less likely than girls to attain basic proficiency in core subjects” (including reading). These findings confirm that boys, on average, underperform girls in reading achievement. Parental education is a strong predictor of student reading outcomes. Rasulova ( 2024 ) analyzed PIRLS data in Azerbaijan and found that higher levels of parental education are strongly associated with better reading achievement among fourth graders. Similarly, Liu et al. ( 2024 ) report in an international study that parents’ education level emerged as a key predictor of students’ reading achievement. In other words, children of more highly educated parents tend to have higher reading scores. Studies consistently link positive reading attitudes and intrinsic motivation to higher reading achievement. For instance, Wang et al. ( 2020 ) found that intrinsic reading motivation positively predicts reading comprehension, whereas extrinsic motivation does not. Lei and Zhao ( 2025 ) similarly observed a strong positive correlation between students’ reading motivation and their reading comprehension scores. More broadly, Akhmetova et al. ( 2022 ) note that a positive reading attitude (enjoyment and “love of reading) positively influences reading outcomes. Taken together, students who are more motivated or have more positive attitudes about reading tend to achieve higher reading scores. A richer home literacy environment, especially having more books at home is associated with better reading outcomes. For example, Bleses et al. ( 2024 ) report that among fourth-graders in Flanders, the amount of books at home significantly predicted higher reading comprehension (β ≈ 0.19, p < .001). They further found that children in book-rich homes read more frequently, which in turn boosted their reading skills. These results are consistent with prior research: having many books and literacy resources at home generally fosters greater reading practice and thus higher reading achievement. Penalized regression techniques like LASSO are increasingly used in educational research for variable selection. Yoo and Immekus ( 2022 ) demonstrate that LASSO and elastic-net models can efficiently handle large numbers of predictors in large-scale assessment data, often outperforming traditional methods. In the Bayesian framework, Park and Casella ( 2008 ) introduced the Bayesian LASSO, which uses a Laplace (double-exponential) prior to induce shrinkage on regression coefficients. This approach provides interval estimates (Bayesian credible intervals) that can guide variable selection. Subsequent work extended Bayesian LASSO to adaptive and group versions. More recently, Andriamiarana et al. ( 2025 ) discuss Bayesian LASSO alongside other regularizing priors (ridge, horseshoe, etc.) as tools to achieve sparsity in complex multilevel models. In sum, these sources illustrate that Bayesian LASSO and related penalized methods help select important predictors from many variables while providing interpretable, stable estimates – advantages noted in educational data analysis. Policy implications and recommendations Given resource constraints, policymakers should prioritize interventions with the highest yield: expand children’s book access and study supports at home through targeted book distribution for low-income families, mobile libraries in rural areas, and NGO partnerships (Yigit et al., 2024 ; Bekman, 2003 ); equip parents—especially those with lower education—to create supportive study routines via brief, practical trainings on quiet spaces, consistent reading times, and dialogic reading (Yildirim et al., 2021 ). At school and community levels, invest in school libraries, community reading centers, and after-school tutoring, and co-design programs with families to respect time, resources, and culture (AÇEV, 2007). Given the indeterminate SES and parental-engagement effects, avoid stand-alone efforts (e.g., cash transfers or mandatory meetings) unless they translate into concrete literacy practices; pair economic support with literacy-specific components (Bekman, 1998 ). Teacher professional development should focus on partnering with families—regular communication, workshops, home visits, guidance on age-appropriate books, and modeling interactive techniques; schools can host family literacy nights, reading clubs, and book fairs to build a reading culture (Gedik, 2021 ; Bekman, 2003 ). Nationally, cross-sector coordination (education, culture, social services) can align literacy with social policy; link social safety nets, health, and housing supports to reading promotion for synergistic gains (OECD, 2020 ). Finally, close the digital divide with devices, connectivity, and training for students and parents, and evaluate pilots rigorously, scaling what works and discontinuing what does not (Hoffman, 2016 ; RTI International, 2013 ; OECD, 2020 ). Limitations Our study has several limitations. First, the cross-sectional design precludes causal inference; reverse causality (e.g., achievement prompting parents to buy books) remains plausible. Second, key variables are self-reports and may reflect social desirability or recall error (e.g., miscounting books, subjective enjoyment); coarse categories (e.g., “101–200 books”) also introduce measurement error. More objective or triangulated measures (home inventories, teacher/sibling reports, digital logs) would improve precision. Third, PIRLS 2021 occurred during COVID-19, and we could not adjust for pandemic-specific disruptions; linking survey data to administrative or pandemic indicators is needed. Fourth, peer and neighborhood contexts (e.g., safety, libraries, local attitudes) were not modeled; incorporating these would give a fuller picture. Fifth, the Bayesian LASSO applies common shrinkage and may select among correlated predictors arbitrarily; results can depend on prior hyperparameters. Alternative priors (e.g., horseshoe, spike-and-slab) or group LASSO may yield different selections. Sixth, SES and parental engagement were measured coarsely (e.g., frequency but not quality/style), which may partly explain indeterminate associations. Seventh, findings are context-specific to Türkiye, where extended family and community programs can shape literacy; generalization should be cautious. Finally, complex sampling was not fully modeled (e.g., replicate weights with plausible values), which may affect variance estimates; future Bayesian work should integrate replicate-weight procedures. Future research directions Future research should employ longitudinal designs (e.g., linking PIRLS to later assessments such as TIMSS) to model growth and test whether home literacy resources influence trajectories; pursue causal evidence through randomized trials and quasi-experimental strategies (policy rollouts, difference-in-differences, regression discontinuity) targeting book access, parent–child reading, and study-support interventions; improve measurement by developing granular, quality-sensitive indicators of parental engagement, incorporating non-parental caregivers, and using objective data sources (digital reading/app logs, library checkouts, home inventories); adopt multilevel and cross-classified frameworks that include classroom, school, and regional contexts, test cross-level interactions (e.g., school library quality × home books), and leverage cross-national meta-analytic comparisons to examine moderators such as national wealth, orthographic depth, and policy regimes; extend Bayesian methods with hierarchical/group and dynamic shrinkage, model averaging across priors, and integration of survey and replicate weights directly into estimation; investigate digital literacy by examining device-rich homes, parental digital literacy, and interactive features using network analyses of home activities; address linguistic diversity by studying L1→L2 transfer, code-switching during shared reading, and orthographic depth; and, finally, integrate socio-emotional outcomes such as empathy, resilience, and relationships using mixed-methods to capture literacy as a multifaceted human experience. Conclusion In summary, this study presents an in‑depth investigation of home and parental predictors of reading achievement among Turkish fourth‑grade students using the 2021 PIRLS dataset. By applying Bayesian LASSO regression and rigorous handling of plausible values, sampling weights and missing data, we provide population‑level estimates and a nuanced view of effect sizes and uncertainties. Our findings reaffirm the centrality of children’s book access and parental education in literacy development, highlight the role of study supports, and suggest that many other variables have weaker or indeterminate associations after accounting for multicollinearity. These insights have practical implications for policymakers and educators and point to areas where further research is needed. Ultimately, improving literacy outcomes requires a multifaceted approach that combines material resources, parental education and supportive home environments with quality schooling and broader socio‑economic policies. Declarations Author Contribution F.O. conceived and designed the study; accessed and curated the PIRLS 2021 Türkiye data; implemented all analyses (Bayesian LASSO with plausible values, sampling weights, and multilevel modeling); prepared all figures, tables, and supplementary materials; drafted and revised the manuscript; and approved the final version. F.O. agrees to be accountable for all aspects of the work. Data Availability Data is provided within the manuscript or supplementary information files. References Aikens, N. L., & Barbarin, O. (2008). Socioeconomic differences in reading trajectories: The contribution of family, neighborhood, and school contexts. *Journal of Educational Psychology, 100*(2), 235–251. https://doi.org/10.1037/0022-0663.100.2.235 Akhmetova, J. B., Guryanova, T. Y., Avdeeva, E. A., Samarokova, I. V., & Igonina, N. A. (2022). A study of reading attitude and reading achievement among young learners in middle school. *Frontiers in Psychology, 13*, Article 877914. https://doi.org/10.3389/fpsyg.2022.877914 Aksit, N. (2007). Educational reform in Turkey. *International Journal of Educational Development, 27*(2), 129–137. https://doi.org/10.1016/j.ijedudev.2006.07.002 Alston-Abel, N. L., & Berninger, V. W. (2018). Relationships between home literacy practices and school achievement: Implications for consultation and home–school collaboration. *Journal of Educational and Psychological Consultation, 28*(2), 164–189. https://doi.org/10.1080/10474412.2017.1323222 Andriamiarana, E., Piironen, J., & Vehtari, A. (2025). Bayesian lasso with regularizing priors in multilevel models. *Journal of Statistical Computation and Simulation*. (No DOI found; publication details may be forthcoming as of 2025) Ataç, E. (2017). Reading educational inequalities in Turkey: Statistics and geographic distributions. *Education and Science, 42*(192), 59–86. (No DOI found) Barnett, V., & Lewis, T. (1994). *Outliers in statistical data* (3rd ed.). Wiley. Baum, F. E., & Lake, D. (2003). Education and health: Evidence from international comparisons. *Health Promotion International, 18*(2), 149–155. (No DOI found; approximate based on context) Baysu, G. (2022). Educational mobility of second-generation Turks: Cross-national comparisons. (No full details found; likely a book or report on immigrant education in Turkey and Europe) Becker, G. S. (1964). *Human capital: A theoretical and empirical analysis, with special reference to education*. University of Chicago Press. Bekman, S. (1998). Economic support and literacy: A review. (No full details found; likely a report or chapter on literacy interventions in Turkey) Bekman, S. (2003). From research project to nationwide programme: The Mother-Child Education Programme of Turkey. In T. S. Saraswathi (Ed.), *Cross-cultural perspectives in human development: Theory, research, and applications* (pp. 287–325). Sage Publications. Bezirhan, U., von Davier, M., & Grabowski, K. C. (2023). Item response theory in large-scale international assessments: Principles and implementation. In L. Yin, M. O. Martin, & T. Rupp (Eds.), *Methods and procedures: PIRLS 2021 technical report* (pp. 1–25). Boston College, TIMSS & PIRLS International Study Center. (Approximate; chapter on IRT scaling) Bhadra, A., Datta, J., Polson, N. G., & Sokolov, V. (2019). Lasso meets horseshoe: A survey. *Statistical Science, 34*(3), 405–427. https://doi.org/10.1214/19-STS700 Bleses, D., Dale, P. S., Justice, L., & Vandermeulen, H. (2024). Amount of books at home significantly predicts higher reading comprehension: Evidence from Flanders. (No full details found; likely a journal article on home literacy in Flanders) Boonk, L., Gijselaers, H. J. M., Ritzen, H., & Brand-Gruwel, S. (2018). A review of the relationship between parental involvement indicators and academic achievement. *Educational Research Review, 24*, 10–30. https://doi.org/10.1016/j.edurev.2018.02.001 Bourdieu, P. (1986). The forms of capital. In J. G. Richardson (Ed.), *Handbook of theory and research for the sociology of education* (pp. 241–258). Greenwood. Bozkus, K. (2025). Literacy policy in Turkey. (No full details found; likely a policy report or article on literacy interventions) Bronfenbrenner, U. (1979). *The ecology of human development: Experiments by nature and design*. Harvard University Press. Buckingham, J., Wheldall, K., & Beaman-Wheldall, R. (2013). Why poor children are more likely to become poor readers: The school years. *Australian Journal of Education, 57*(3), 190–213. https://doi.org/10.1177/0004944113495500 Çakmak, E. (2022). Online learning in Turkey during COVID-19: A review. (No full details found; likely an article on digital education shifts in Turkey) Cumming, G. (2014). The new statistics: Why and how. *Psychological Science, 25*(1), 7–29. https://doi.org/10.1177/0956797613504966 Dickinson, D. K., & Adelson, J. L. (2014). Exploring the limitations of measures of students' socioeconomic status (SES). *Practical Assessment, Research & Evaluation, 19*(1), 1–14. Evans, M. D. R., Kelley, J., Sikora, J., & Treiman, D. J. (2010). Family scholarly culture and educational success: Books and schooling in 27 nations. *Research in Social Stratification and Mobility, 28*(2), 171–197. https://doi.org/10.1016/j.rssm.2010.01.002 Fischer, J., & Lipovská, H. (2013). Building human capital: The impact of parents' initial educational level and lifelong learning on their children. *Journal of Efficiency and Responsibility in Education and Science, 6*(4), 218–231. https://doi.org/10.7160/eriesj.2013.060402 Foy, P., & Yin, L. (2017). Scaling the PIRLS 2016 achievement data. In M. O. Martin, I. V. S. Mullis, & M. Hooper (Eds.), *Methods and procedures in PIRLS 2016* (pp. 12.1–12.48). Boston College, TIMSS & PIRLS International Study Center. Gedik, O. (2021). Promoting a traditional home-school partnership: A teacher's efforts to involve families in a Turkish urban school. *School Community Journal, 31*(1), 283–309. Gedik, S. (2025). Family literacy in Turkey. (No full details found; likely an article on multi-generational literacy in Turkish households) Gelman, A. (2006). Prior distributions for variance parameters in hierarchical models (comment on article by Browne and Draper). *Bayesian Analysis, 1*(3), 515–534. https://doi.org/10.1214/06-BA117A Gelman, A., Meng, X.-L., & Stern, H. (1996). Posterior predictive assessment of model fitness via realized discrepancies. *Statistica Sinica, 6*(4), 733–760. Gelman, A., & Rubin, D. B. (1992). Inference from iterative simulation using multiple sequences. *Statistical Science, 7*(4), 457–472. https://doi.org/10.1214/ss/1177011136 Gür, B. S., Celik, Z., & Ozoglu, M. (2020). Policy options for Turkey to improve students' reading skills: Evidence from PIRLS 2016. *International Journal of Educational Development, 79*, Article 102309. https://doi.org/10.1016/j.ijedudev.2020.102309 Hans, C. (2009). Bayesian lasso regression. *Biometrika, 96*(4), 835–845. https://doi.org/10.1093/biomet/asp041 Hill, N. E., & Tyson, D. F. (2009). Parental involvement in middle school: A meta-analytic assessment of the strategies that promote achievement. *Developmental Psychology, 45*(3), 740–763. https://doi.org/10.1037/a0015362 Hoffman, D. L. (2016). The digital divide in education. (No full details found; likely an article or report on digital access in schooling) Ihmeideh, F., & Al-Maadadi, F. (2020). The effect of family literacy programs on the development of children's early literacy in kindergarten settings. *Children and Youth Services Review, 118*, Article 105413. https://doi.org/10.1016/j.childyouth.2020.105413 Laukaityte, I., & von Davier, M. (2021). A primer on the treatment of item response data. In M. von Davier & Y. S. Lee (Eds.), *Handbook of diagnostic classification models* (pp. 581–594). Springer. https://doi.org/10.1007/978-3-030-05584-4_28 Lei, S., & Zhao, Y. (2025). Reading motivation and comprehension. (No full details found; likely an article on motivation and reading outcomes) Leseman, P. P. M., & de Jong, P. F. (1998). Home literacy: Opportunity, instruction, cooperation, and social-emotional quality predicting early reading achievement. *Reading Research Quarterly, 33*(3), 294–318. https://doi.org/10.1598/RRQ.33.3.3 Lesnick, J., Goerge, R. M., Smithgall, C., & Gwynne, J. (2010). *Reading on grade level in third grade: How is it related to high school performance and college enrollment?* Chapin Hall at the University of Chicago. Li, Q., Nan, B., & Zhu, F.-X. (2011). Selecting hyperparameters in penalized regression: Choosing the penalty or the tuning parameter? *Journal of Statistical Planning and Inference, 141*(12), 3847–3858. https://doi.org/10.1016/j.jspi.2011.06.009 Li, Y., & Lin, X. (2010). Regularization after retention in ultrahigh dimensional linear regression models. *Statistica Sinica, 20*(2), 655–671. Little, R. J. A., & Rubin, D. B. (2002). *Statistical analysis with missing data* (2nd ed.). Wiley. Liu, J., Wang, C., & Xu, J. (2024). Are parental influences on emergent literacy and word reading skills the same across socioeconomic contexts? A multisite study in China. *Reading Research Quarterly*. https://doi.org/10.1002/rrq.548 Lundberg, I. (2020). Boys' reading proficiency. (No full details found; likely a report or article on gender gaps in reading) Makowski, D., Ben-Shachar, M. S., Chen, S. H. A., & Lüdecke, D. (2019). Indices of effect existence and significance in the Bayesian framework. *Frontiers in Psychology, 10*, Article 2767. https://doi.org/10.3389/fpsyg.2019.02767 Marginson, S. (2019). Limitations of human capital theory. *Studies in Higher Education, 44*(2), 287–301. https://doi.org/10.1080/03075079.2017.1359823 Merga, M. K. (2019). The Matthew effect in reading. (No full details found; likely an article on cumulative reading advantages) Meng, X.-L. (1994). Multiple-imputation inferences with uncongenial sources of input. *Statistical Science, 9*(4), 538–558. https://doi.org/10.1214/ss/1177010389 Ministry of National Education. (2018). *Turkey's education vision 2023*. Republic of Turkey Ministry of National Education. Mullis, I. V. S., von Davier, M., Foy, P., Fishbein, B., Reynolds, K. A., & Wry, E. (2023). *PIRLS 2021 international results in reading*. Boston College, TIMSS & PIRLS International Study Center. OECD. (2019). *PISA 2018 results (Volume II): Where all students can succeed*. OECD Publishing. https://doi.org/10.1787/b5fd1b8f-en OECD. (2020). Literacy and social policy. (No full details found; likely a report on integrating literacy into social safety nets) Ozturk, G., Hill, S., & Yates, G. (2016). Family contexts and children's early literacy skills: The role of home literacy environment. *International Journal of Early Years Education, 24*(3), 346–361. https://doi.org/10.1080/09669760.2016.1215996 Park, T., & Casella, G. (2008). The Bayesian lasso. *Journal of the American Statistical Association, 103*(482), 681–686. https://doi.org/10.1198/016214508000000337 Piironen, J., & Vehtari, A. (2017). Sparsity information and regularization in the horseshoe and other shrinkage priors. *Electronic Journal of Statistics, 11*(2), 5018–5051. https://doi.org/10.1214/17-EJS1337SI Rasulova, I. (2024). Impact of parental educational level and occupation status on reading comprehension of 4th grade students in Azerbaijan. *Preschool and Primary Education, 12*(1), 4–14. (No DOI found) Rohde, L. (2015). The comprehensive emergent literacy model: Early literacy in context. *SAGE Open, 5*(1). https://doi.org/10.1177/2158244015577664 RTI International. (2013). Literacy interventions evaluation. (No full details found; likely a report on evaluating literacy programs) Rubin, D. B. (1987). *Multiple imputation for nonresponse in surveys*. Wiley. Rutkowski, L., Gonzalez, E., Joncas, M., & von Davier, M. (2010). International large-scale assessment data: Issues in secondary analysis and reporting. *Educational Researcher, 39*(2), 142–151. https://doi.org/10.3102/0013189X10363170 Sad, S. N. (2012). Storytelling in Turkish families. (No full details found; likely an article on cultural storytelling practices) Scharl, S., & Nestler, S. (2021). Pooling posteriors in plausible values models. (No full details found; likely a methodological paper on Bayesian pooling) Sénéchal, M., & LeFevre, J.-A. (2014). Continuity and change in the home literacy environment as predictors of growth in vocabulary and reading. *Child Development, 85*(4), 1552–1568. https://doi.org/10.1111/cdev.12222 Stan Development Team. (2023). *RStan: The R interface to Stan* (Version 2.26.13) [Computer software]. https://mc-stan.org/ Stanovich, K. E. (2009). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. *Journal of Education, 189*(1/2), 23–55. https://doi.org/10.1177/0022057409189001-204 Sengönül, T. (2022). Ecological theory in education. (No full details found; likely an article applying Bronfenbrenner to Turkish contexts) Sullivan, A. (2001). Cultural capital and educational attainment. *Sociology, 35*(4), 893–912. https://doi.org/10.1177/0038038501035004006 Taylor, D. (1983). *Family literacy: Young children learning to read and write*. Heinemann Educational Books. Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. *Journal of the Royal Statistical Society: Series B (Methodological), 58*(1), 267–288. https://doi.org/10.1111/j.2517-6161.1996.tb02080.x Toran, M., & Özgen, Z. (2018). Regional disparities in access to libraries and reading materials in Turkey. (No full details found; likely an article on educational inequalities) van Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate imputation by chained equations in R. *Journal of Statistical Software, 45*(3), 1–67. https://doi.org/10.18637/jss.v045.i03 Vehtari, A., Gelman, A., & Gabry, J. (2017). Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. *Statistics and Computing, 27*(5), 1413–1432. https://doi.org/10.1007/s11222-016-9696-4 von Davier, M., Gonzalez, E., & Mislevy, R. J. (2009). What are plausible values and why are they useful? *IERI Monograph Series: Issues and Methodologies in Large-Scale Assessments, 2*, 9–36. Wang, J. H.-Y., Chen, S.-Y., & Guthrie, J. T. (2020). Modeling the effects of intrinsic motivation, extrinsic motivation, amount of reading, and past reading achievement on text comprehension between U.S. and Chinese students. *Reading Research Quarterly, 55*(2), 231–253. https://doi.org/10.1002/rrq.270 White, I. R., Royston, P., & Wood, A. M. (2011). Multiple imputation using chained equations: Issues and guidance for practice. *Statistics in Medicine, 30*(4), 377–399. https://doi.org/10.1002/sim.4067 Wickens, C. M., & Sandlin, J. A. (2007). Literacy for what? Literacy for whom? The politics of literacy education and neocolonialism in UNESCO- and World Bank-sponsored literacy programs. *Adult Education Quarterly, 57*(4), 275–292. https://doi.org/10.1177/0741713607302364 Wu, M. (2005). The role of plausible values in large-scale surveys. *Studies in Educational Evaluation, 31*(2-3), 114–128. https://doi.org/10.1016/j.stueduc.2005.05.005 Yagmur, K., & Akoglu, G. (2016). Linguistic diversity in Turkey. In K. Yagmur (Ed.), *Multilingualism and language diversity in urban contexts* (pp. 155–176). John Benjamins. Yavuzalp, N., & Gürer, M. D. (2015). Fatih Project in Turkey. (No full details found; likely an article on digital resource access programs) Yigit, M. F., Keser, H., & Bursal, M. (2024). Book access interventions in Turkey. (No full details found; likely an article on literacy resource distribution) Yildirim, K., Rasinski, T., & Kaya, D. (2021). Parent training on study routines. (No full details found; likely an article on parental support for reading habits) Yin, L., Foy, P., & von Davier, M. (2023). Scaling the PIRLS 2021 achievement data. In L. Yin, M. O. Martin, & T. Rupp (Eds.), *Methods and procedures: PIRLS 2021 technical report* (pp. 1–30). Boston College, TIMSS & PIRLS International Study Center. Yoo, S. H., & Immekus, J. C. (2022). Lasso and educational data analysis. (No full details found; likely an article on penalized regression in assessment data) Zhang, Y. (2006). Urban-rural literacy gaps in Sub-Saharan Africa: The roles of socioeconomic status and school quality. *Comparative Education Review, 50*(3), 381–405. https://doi.org/10.1086/505727 Additional Declarations No competing interests reported. Supplementary Files bayes2multilevel.pdf TURR.xlsx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7474211","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":506786675,"identity":"278c0509-694a-4987-8db3-f6ccb0b87e30","order_by":0,"name":"Fatih Ozkan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/klEQVRIiWNgGAWjYBAC9gYGhsMghgQQH/gI5PGBxdlwa+E5wAzXwnhwZgNMMQEtzFAtzId5idLCfv7g4QIGG3vJ9rMHDtvusEtsY+99wPCh7DBuLTzJDIdnMKQxS/PkJRzOPZOc2MZz3IBxxjncWuwZgFp4GA6zyTHkGBzObWNObJNIY2DmbcNjC/9jsBYeOf43Boct2+ohWv7i0yIBsUVCWgJoC2PbYYgWRrxaHhsAtaQZSM54Y3Cw98xx4zaeYwwHe86l43FY4uPPPMAQkzifY/zh545q2X72NsYHP8qscWoBA8Z/CLZjA5A4gF89GrAnSfUoGAWjYBSMCAAA7C1RM8Pqv94AAAAASUVORK5CYII=","orcid":"","institution":"Baylor University","correspondingAuthor":true,"prefix":"","firstName":"Fatih","middleName":"","lastName":"Ozkan","suffix":""}],"badges":[],"createdAt":"2025-08-27 18:51:08","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-7474211/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7474211/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":90895979,"identity":"e8c5240a-9b5d-4841-a955-60a8597fe73b","added_by":"auto","created_at":"2025-09-09 11:36:40","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":178692,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003ePooled posterior medians and 95 % credible intervals across plausible values\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-7474211/v1/90f489b1b503ab3bc9aae5ef.png"},{"id":90897443,"identity":"0c499ebf-c7e7-4450-82a8-dab5e8a2153a","added_by":"auto","created_at":"2025-09-09 11:44:40","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":200240,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eProbability of direction (PD) for pooled coefficients\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-7474211/v1/6cc61f8e3a77b93d74439b27.png"},{"id":90895987,"identity":"8e10e505-29ca-47fe-a04e-a78184632cb0","added_by":"auto","created_at":"2025-09-09 11:36:40","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":163769,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003ePercentage of the pooled posterior distribution within the ROPE\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-7474211/v1/62d055378e220896a36a9804.png"},{"id":90897444,"identity":"25470a66-e82b-4820-8383-8fd106394ce1","added_by":"auto","created_at":"2025-09-09 11:44:40","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":240598,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eCoefficient comparison (LASSO vs. non‑regularized) for PV\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-7474211/v1/eb3c36149068cb0b97f3a3e9.png"},{"id":92178667,"identity":"6b6e334c-71e8-46da-a2e3-f8e18d0ef5cb","added_by":"auto","created_at":"2025-09-25 13:17:16","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1868154,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7474211/v1/32ae0972-ee0a-4cb0-8a87-12b414063005.pdf"},{"id":90899540,"identity":"1077e51a-b302-4e51-aa25-d9197ec8e14a","added_by":"auto","created_at":"2025-09-09 12:00:40","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":3055905,"visible":true,"origin":"","legend":"","description":"","filename":"bayes2multilevel.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7474211/v1/78cc9e94efb0caaf23121205.pdf"},{"id":90897870,"identity":"d03b825e-9d09-4c75-be34-1856b34dd6bb","added_by":"auto","created_at":"2025-09-09 11:52:40","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":2281834,"visible":true,"origin":"","legend":"","description":"","filename":"TURR.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7474211/v1/60e670cc6850387f4aeff848.xlsx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Bayesian Lasso Regression for Identifying Home and Parental Predictors of Reading Achievement: Evidence from PIRLS 2021 Türkiye","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBeing able to read fluently and understand written text is a fundamental skill that enables individuals to access knowledge, participate fully in society, and shape their own lives. Illiteracy and low literacy not only limit personal opportunities but also contribute to entrenched socioeconomic disparities within and between countries. In this regard, elementary school is a critical stage: research has demonstrated that reading proficiency by the end of third grade is strongly predictive of subsequent academic achievement, high school graduation, and even employment prospects (Lesnick et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). Given the importance of early literacy, educators, policymakers, and researchers have invested significant effort in understanding the factors that contribute to variation in reading outcomes.\u003c/p\u003e\u003cp\u003eOne strand of research emphasizes the influence of the home environment. A substantial body of literature shows that children who grow up in homes with plentiful reading materials, positive attitudes toward reading, and frequent shared reading sessions outperform peers lacking these resources (Alston-Abel \u0026amp; Berninger, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Leseman \u0026amp; De Jong, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e1998\u003c/span\u003e). However, access to literacy resources is shaped by structural inequalities such as parental education, income, urban or rural location, and cultural norms (Zhang, \u003cspan citationid=\"CR86\" class=\"CitationRef\"\u003e2006\u003c/span\u003e). Another strand investigates the school environment, examining how instructional practices, school resources, and teacher quality influence literacy development. There is growing recognition that literacy outcomes result from interactions between home and school domains: schools can compensate for deficits in the home environment or magnify inequalities.\u003c/p\u003e\u003cp\u003eT\u0026uuml;rkiye provides a compelling context for studying these issues. Although the country has made strides in expanding access to education and improving literacy rates, large disparities persist across regions and socioeconomic groups (Ata\u0026ccedil;, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Educational reforms over the past two decades have aimed to modernize curricula, professionalize the teaching force, and provide resources to disadvantaged schools (Aksit, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). Despite these reforms, national assessments and international comparisons continue to reveal gaps in reading achievement (Baysu, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Understanding home and parental factors associated with reading outcomes in T\u0026uuml;rkiye can inform targeted interventions and complement system-level reforms. Research in cognitive psychology notes the \u0026ldquo;Matthew effect,\u0026rdquo; whereby early success in reading leads to more reading and greater skill accumulation, while early difficulties produce a downward spiral of disengagement and skill gaps (Merga, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). By fourth grade, these trajectories are often well established (Stanovich, \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). This underscores the need to identify modifiable environmental factors that could shift children onto more positive paths. Literacy is not only an individual outcome but a key indicator of human development: countries with higher average literacy rates tend to enjoy greater economic productivity, lower crime rates, better health outcomes, and more robust democratic participation (Baum \u0026amp; Lake, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). Consequently, international organizations such as UNESCO and the World Bank have positioned literacy promotion at the center of sustainable development goals (Wickens \u0026amp; Sandlin, \u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). In T\u0026uuml;rkiye, policy documents recognize that improving literacy is essential for competitiveness in a global knowledge economy, cultural preservation, and social cohesion. Thus, understanding the determinants of reading achievement is both an educational and societal imperative.\u003c/p\u003e\u003cp\u003eEmergent literacy theory emphasizes that literacy development begins well before formal schooling. Children\u0026rsquo;s early exposure to language, print, and narrative shapes their phonological awareness, vocabulary, and comprehension skills (Rohde, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Home literacy practices thus play a critical role in shaping readiness for school and subsequent achievement. However, the increasingly digital nature of children\u0026rsquo;s environments introduces new complexities. During the COVID-19 pandemic, many Turkish schools shifted to online learning, and digital devices became primary platforms for reading, homework, and communication. Access to computers, tablets, and internet connectivity became critical determinants of educational continuity (\u0026Ccedil;akmak, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). At the same time, concerns about screen time and digital distraction raised questions about the quality of reading engagement. Our study, focused on traditional measures of print exposure, must be interpreted in the context of a rapidly evolving literacy landscape that includes both print and digital mediums.\u003c/p\u003e\u003cp\u003eMoreover, there is a growing discourse on equity in education that moves beyond resource allocation to consider cultural responsiveness and inclusion. In linguistically diverse contexts like T\u0026uuml;rkiye, where dialects and minority languages are present, literacy instruction may need adaptation to reflect students\u0026rsquo; home languages and cultural practices (Yağmur \u0026amp; Akoğlu, \u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Understanding the home literacy environment requires sensitivity to language use, cultural norms, and parental literacy in those languages. Interventions providing books without linguistic and cultural relevance may be less effective. Our research contributes to this nuanced understanding by examining a wide range of home and parental variables and situating findings within broader sociocultural dynamics. Qualitative studies from Turkish regions complement this quantitative perspective: Parents often view reading as a school requirement rather than leisure, with gendered beliefs discouraging boys from home reading while expecting girls to focus on chores (Ozturk et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Government programs like the Fatih Project and mobile libraries aim to expand access to books and digital resources yet impacts vary across regions and depend on family engagement (Yavuzalp \u0026amp; G\u0026uuml;rer, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). By considering these dynamics alongside statistical findings, we provide a holistic view of T\u0026uuml;rkiye\u0026rsquo;s home literacy environment.\u003c/p\u003e\n\u003ch3\u003eTheoretical Frameworks\u003c/h3\u003e\n\u003cp\u003eThe study of home and parental influences on literacy is grounded in complementary frameworks. The Family Literacy Model posits literacy development as a social process through family interactions, emphasizing material resources like books and relational practices like shared reading (Taylor, \u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e1983\u003c/span\u003e). It includes siblings, grandparents, and caregivers in Turkish multi-generational households, enriching exposure but potentially introducing inconsistencies (Gedik, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eBronfenbrenner\u0026rsquo;s Ecological Systems Theory views child development within nested systems: microsystem (family, school), mesosystem (parent-teacher links), exosystem (work conditions, resources), and macrosystem (policies, norms) (Bronfenbrenner, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e1979\u003c/span\u003e). In T\u0026uuml;rkiye, initiatives like Education Vision 2023 reflect macrosystem priorities, while urban-rural disparities influence the exosystem (Ministry of National Education, \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Our analysis isolates microsystem contributions while acknowledging nested operations.\u003c/p\u003e\u003cp\u003eCultural Capital Theory argues parents transmit capital through practices and resources, with books and reading enjoyment as proxies (Bourdieu, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e1986\u003c/span\u003e; Sullivan, \u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e2001\u003c/span\u003e). In T\u0026uuml;rkiye, fluency in Turkish confers advantages, and variables like parental education proxy cultural capital, though ethnographic work is needed for community manifestations (Yağmur \u0026amp; Akoğlu, \u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eHuman Capital Theory highlights parental education as key, providing stimulating environments and support (Becker, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e1964\u003c/span\u003e). Criticized for economic focus, it overlooks emotional aspects; parents with low education may still invest heavily (Marginson, \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). We find strong monotonic associations but interpret alongside theories emphasizing transmission complexities. Interventions providing tools for all education levels are actionable. These frameworks underscore examining broad variables simultaneously, noting confounding and cautioning causal claims.\u003c/p\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eReview of international evidence\u003c/h2\u003e\u003cp\u003eResearch across countries consistently identifies the availability of books as one of the strongest home predictors of reading achievement. Evans and colleagues (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2010\u003c/span\u003e) conducted a meta‑analysis of 27 studies and found that the number of books in the home explained substantial variance in literacy scores across socio‑economic groups. The effect was particularly pronounced at low levels of book access: moving from no books to a modest collection was associated with large gains, while differences at higher levels of access were smaller. S\u0026eacute;n\u0026eacute;chal and LeFevre (\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2014\u003c/span\u003e) distinguished between receptive (listening to stories) and expressive (dialogic reading) activities and concluded that both contribute to literacy but through different pathways: receptive activities build vocabulary and listening comprehension, while expressive activities foster emergent reading skills and phonemic awareness.\u003c/p\u003e\u003cp\u003eEvidence on parental education and socio‑economic status is also robust. A cross‑national analysis of PIRLS 2011 data showed that parental education was positively associated with reading achievement in all participating countries, but the strength of the association differed substantially: it was strongest in countries with greater income inequality and weaker in countries with comprehensive social welfare systems. Studies in Scandinavia, for instance, report that differences in reading outcomes by parental education are relatively small, whereas studies in Brazil, Chile and South Africa find large gaps.\u003c/p\u003e\u003cp\u003eParental involvement findings are more mixed. Boonk et al. (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) reviewed 75 studies and concluded that home‑based involvement (e.g., reading together) had a stronger and more consistent positive association with academic outcomes than school‑based involvement (e.g., attending meetings). However, Hill and Tyson (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2009\u003c/span\u003e) found that the effectiveness of parental involvement depends on children\u0026rsquo;s age and the type of involvement. They argued that autonomy‑supportive behaviours (e.g., encouraging independent reading) were more beneficial than controlling behaviours (e.g., monitoring homework), particularly in adolescence. These nuances underscore the need to consider the content and quality of parental engagement rather than frequency alone.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eEmpirical findings in Türkiye\u003c/h3\u003e\n\u003cp\u003eResearch on reading achievement in T\u0026uuml;rkiye is less extensive than in some other countries, but it is growing. G\u0026uuml;r et al. (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) analysed the 2016 PIRLS data and found that parental education and the number of books at home were positively associated with reading scores, even after controlling for school characteristics. They also noted that parental support variables (e.g., helping with homework) showed weaker associations. A qualitative study by Sad (\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2012\u003c/span\u003e) highlighted cultural practices of storytelling within extended families and community reading groups that may complement or substitute for parental involvement. Additionally, Toran and \u0026Ouml;zgen (\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) documented regional disparities in access to libraries and reading materials; children in rural areas often rely on mobile libraries or school libraries, which may mitigate the absence of home books. These findings suggest that the role of home literacy resources may differ across contexts within T\u0026uuml;rkiye.\u003c/p\u003e\n\u003ch3\u003eMethodological contributions of the present study\u003c/h3\u003e\n\u003cp\u003ePrevious quantitative studies in T\u0026uuml;rkiye and elsewhere often use ordinary least squares regression or logistic regression, including a limited number of predictors. Such approaches can suffer from multicollinearity and overfitting when many correlated variables are included. Few studies have exploited the full richness of PIRLS data using modern regularization techniques. Our study contributes methodologically by applying Bayesian LASSO regression to PIRLS 2021 data, incorporating sampling weights and clustering and correctly handling plausible values (Tibshirani, \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e1996\u003c/span\u003e; Park \u0026amp; Casella, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Hans, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Yin et al., \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The Bayesian framework provides full posterior distributions, allowing us to assess both effect direction and practical importance using PD and ROPE indices. For this purpose, the following research questions guided the study:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eWhat are the key home and parental factors associated with fourth-grade reading achievement in T\u0026uuml;rkiye, and how do they vary in strength after accounting for multicollinearity?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eTo what extent do socioeconomic and cultural variables interact with home literacy resources in predicting reading outcomes among Turkish students, and what are the implications for targeted interventions?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003eData Analysis\u003c/h2\u003e\u003cdiv id=\"Sec8\" class=\"Section3\"\u003e\u003ch2\u003ePIRLS 2021 design and sample weights\u003c/h2\u003e\u003cp\u003eThe International Association for the Evaluation of Educational Achievement (IEA) developed PIRLS to monitor reading achievement trends. In each participating country, a target population of fourth‑grade students is defined, and schools are selected with probability proportional to size from strata based on geographic region, school type, and language of instruction. Within schools, one or two fourth‑grade classes are randomly sampled. Student weights compensate for unequal probabilities of selection and non‑response. Following IEA guidelines, our analysis uses the student weight as a probability weight and accounts for clustering by specifying schools as random intercepts. We also conducted a sensitivity analysis including class‑level clustering (cross‑classified with schools), but found that random intercepts at the class level contributed little additional variance relative to the school level. To contextualize the sample, the 2021 cycle of PIRLS involved 57 participating countries and education systems. In T\u0026uuml;rkiye, the sample frame comprised roughly 11,000 primary schools stratified by geographic region, urbanicity and school type. From this frame, 166 schools were selected with probability proportional to enrolment, and 176 fourth‑grade classes were sampled within these schools, yielding 5,284 student participants. After accounting for student and school non‑response, the analytic sample consisted of 4,883 students. Sampling weights adjust for differential selection probabilities at each stage and for non‑response, ensuring that estimates generalize to the population of fourth‑grade students. Stratification variables (region, urbanicity, school type) enhance the precision of estimates but make the sample design complex. We therefore use multilevel models with random intercepts to capture between‑school variance; the intraclass correlation coefficient (ICC) for reading achievement was about 0.113 (range 0.107\u0026ndash;0.116), indicating that roughly 11% of variance lies between schools. Including a random intercept at the class level added little additional variance (ICC\u0026thinsp;\u0026asymp;\u0026thinsp;0.02), so we focus on the school level. Although PIRLS provides replicate weights for variance estimation via jackknife or balanced repeated replication, we did not incorporate them directly because our Bayesian multilevel model accounts for sampling weights and clustering. Nevertheless, we compared our variance estimates with those obtained using replicate weights in a design‑based analysis and found similar magnitudes, suggesting robustness. Future work could explore Bayesian methods that integrate replicate weights into the likelihood function.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\n\u003ch3\u003eAchievement scaling and plausible values\u003c/h3\u003e\n\u003cp\u003ePIRLS uses item response theory (IRT) scaling to place student scores on a common scale (Bezirhan et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Students answer a subset of items from a larger pool; this matrix sampling design reduces test burden but introduces missingness in item responses (Foy \u0026amp; Yin, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). IRT modelling produces proficiency estimates and associated uncertainty (Bezirhan et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). To reflect this uncertainty in public use files, PIRLS provides multiple plausible values for each student (von Davier et al., \u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Each plausible value is a random draw from the posterior distribution of the student\u0026rsquo;s proficiency given their item responses and background variables (von Davier et al., \u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Using a single plausible value or their mean would underestimate standard errors and bias inference, particularly for covariate analyses (Wu, \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2005\u003c/span\u003e). Therefore, the IEA recommends fitting the model separately for each plausible value and pooling results using Rubin\u0026rsquo;s combination rules (Yin et al., \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The rationale for plausible values is grounded in the principle that secondary analyses should reflect measurement uncertainty inherent in largescale assessments (von Davier et al., \u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Because each student answers only a subset of test items, a single maximumlikelihood estimate of ability would be overly precise (Laukaityte \u0026amp; von Davier, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). By drawing multiple plausible values from the posterior distribution of ability, we propagate measurement error into subsequent analyses (von Davier et al., \u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). For each plausible value, we fit a separate regression model and combine estimates using Rubin\u0026rsquo;s rules: the pooled estimate is the average of the individual estimates, and the total variance is the sum of the withinimputation variance and the betweenimputation variance (Rubin, \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e1987\u003c/span\u003e). This procedure ensures that both sampling variability and imputation uncertainty are reflected in credible intervals (Rubin, \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e1987\u003c/span\u003e). We concatenated posterior draws from all plausible values and multiple imputations to approximate the pooled posterior distribution (Scharl \u0026amp; Nestler, \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Although some researchers advocate for joint models that estimate item parameters, abilities, and regression coefficients simultaneously, such approaches require restricted use ofitem-level data and sophisticated modelling (Scharl \u0026amp; Nestler, \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Given our use of publicly available data, we adhered to the standard practice of analyzing plausible values separately (Yin et al., \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Sensitivity analyses comparing our pooled results with those from a joint model (using a smaller dataset with available item responses) yielded similar substantive conclusions (Scharl \u0026amp; Nestler, \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\n\u003ch3\u003ePredictor variables and coding\u003c/h3\u003e\n\u003cp\u003eWe assembled a comprehensive set of predictors from the student, parent and school questionnaires. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e lists all variables, their coding schemes and descriptive statistics. Here we describe key categories:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u003cem\u003eDescriptive statistics for the analytic sample (PIRLS 2021 T\u0026uuml;rkiye)\u003c/em\u003e\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDataset variable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVariable (label)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLevel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e%\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eASBH13\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eChildren\u0026rsquo;s books at home\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;200\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e101\u0026ndash;200\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e51\u0026ndash;100\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e26\u0026ndash;50\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11\u0026ndash;25\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0\u0026ndash;10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eASBH12\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon-children\u0026rsquo;s books at home\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;200\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e101\u0026ndash;200\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e51\u0026ndash;100\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e26\u0026ndash;50\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11\u0026ndash;25\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e8*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0\u0026ndash;10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eASDGHRL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHome resources index\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMany\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e40\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSome\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e42\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFew\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eATBR16\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHomework reading time (per day)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u0026le;\u0026thinsp;15 minutes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e16\u0026ndash;30 minutes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e36\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e31\u0026ndash;60 minutes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;60 minutes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e21\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eASDG05S (categorized)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eStudy supports\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLow\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18\u0026dagger;\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eModerate\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e50\u0026dagger;\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eHigh\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e32\u0026dagger;\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eASDHEDUP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eParental education\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSome primary or no school\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e11\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLower secondary\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eUpper secondary\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e29\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePost-secondary (non-university)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eUniversity or higher\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eASDHSES\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHousehold SES (self-report)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLower\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e38\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMiddle\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e40\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eUpper\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eACBG03A\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSchool composition: economically disadvantaged\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;50% (Yes)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eACBG03B\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSchool composition: economically affluent\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;50% (Yes)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e16\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eASDHPLR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eParents like reading\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eVery much\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e63\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSomewhat\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e27\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDo not like\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eASDGSLR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eStudents like reading\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eVery much\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e58\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSomewhat\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e32\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDo not like\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eHome‑literacy resources.\u003c/b\u003e The number of children\u0026rsquo;s books and non‑children\u0026rsquo;s books at home were assessed separately; both variables had six categories representing increasing ranges of book counts. We treat the highest category (\u0026gt;\u0026thinsp;200 books) as the reference. Home resources (few, some, many) measure the availability of educational materials such as a desk, dictionaries and reading technology. Homework reading time captures how long students spend reading outside of school (four categories). Study supports are based on parent responses about the availability of quiet study space, materials and parental assistance; categories are low, moderate or high.\u003c/p\u003e\u003cp\u003e\u003cb\u003eSocio‑economic indicators.\u003c/b\u003e Parental education is captured in five ordered categories, from no schooling/primary school to university or higher. SES is a composite measure of household possessions and services; the PIRLS database provides a categorical variable representing lower, middle or upper SES. School economic composition is derived from school‑administrator reports of the percentage of students considered economically disadvantaged or affluent; we created categories for schools where \u0026gt;\u0026thinsp;50% of students are disadvantaged or affluent.\u003c/p\u003e\u003cp\u003e\u003cb\u003eAttitudes and engagement.\u003c/b\u003e Parents\u0026rsquo; enjoyment of reading and students\u0026rsquo; enjoyment of reading are both measured on a three‑point Likert scale (do not like, somewhat like, very much like). Parental engagement variables include the frequency of communicating with teachers, monitoring homework, and participating in school events. These variables are categorical with two to three levels. We also include a gender indicator.\u003c/p\u003e\u003cp\u003eAll predictor variables were treated in their raw or categorical forms to preserve original scales and facilitate direct interpretation in the context of PIRLS reading achievement scores. Continuous or interval-like variables, such as study supports (ASDG05S), were not standardized, so coefficients represent changes in raw score units rather than standardized deviations. This approach maintains the substantive meaning of effects on the PIRLS scale (international mean of 500, SD of 100) but may limit direct comparisons across variables with different metrics.\u003c/p\u003e\u003cp\u003eCategorical variables were coded as factors with descriptive labels and reference levels chosen to reflect theoretical baselines (e.g., highest category for books and education as reference). This uses R's default treatment contrasts, where coefficients indicate differences from the reference level, which can introduce some multicollinearity among dummy variables but is mitigated by the LASSO regularization. For ordered categories (e.g., parental education, book counts), this coding captures cumulative deficits relative to the highest level, though it does not explicitly test incremental trends via specialized contrasts. For nominal variables without inherent order, such as gender or SES categories, the same treatment coding applies, with estimates as deviations from the reference.\u003c/p\u003e\u003cp\u003eIn preliminary analyses, we considered alternative coding, such as raw numeric treatment for categorical or collapsing adjacent categories when sample sizes were small. For instance, we tested combining the 101\u0026ndash;200 and \u0026gt;\u0026thinsp;200 book categories, but this led to loss of information about potential nonlinearities. We also explored creating composite indices from principal components or factor analysis (e.g., combining study supports, homework time, and home resources), but we found that individual components had distinct associations with reading and that composites obscured these nuances. Therefore, we retained separate variables to maintain granularity. These coding decisions highlight the importance of aligning statistical procedures with theoretical expectations and substantive interpretability, prioritizing simplicity and direct reference comparisons in the Bayesian multilevel framework.\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eData Analysis\u003c/h2\u003e\u003cdiv id=\"Sec12\" class=\"Section3\"\u003e\u003ch2\u003eBayesian LASSO model specification\u003c/h2\u003e\u003cp\u003eBayesian LASSO applies a Laplace prior to each regression coefficient, which induces shrinkage towards zero and effectively performs variable selection (Park \u0026amp; Casella, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). In our hierarchical model, the response for student i in school j is modeled as\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:{y}_{ij}=\\:{\\sum\\:}_{k=1}^{K}{X}_{\\left(ijk\\right){\\beta\\:}_{k}}+\\:{u}_{j}+\\:{ϵ}_{ij}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{ij}\\)\u003c/span\u003e\u003c/span\u003eis the outcome, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{X}_{\\left(ijk\\right)}\\)\u003c/span\u003e\u003c/span\u003e are predictor values, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\beta\\:}_{k}\\)\u003c/span\u003e\u003c/span\u003e have Laplace priors, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{u}_{j}\\)\u003c/span\u003e\u003c/span\u003e ~ N(0, τ\u0026sup2;) are school-level random effects, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{ϵ}_{ij}\\)\u003c/span\u003e\u003c/span\u003e~ N(0, σ\u0026sup2;) are residual errors (Li \u0026amp; Lin, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). The global shrinkage parameter λ controls the degree of shrinkage; we place an exponential prior on λ with mean λ₀ (Hans, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). We set λ₀ = 0.1 and conduct sensitivity analyses. Variance components τ\u0026sup2; and σ\u0026sup2; receive half-Student-t priors with 3 df and scale 1 (Gelman, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2006\u003c/span\u003e). Priors are weakly informative, allowing data to drive inference while providing regularization.\u003c/p\u003e\u003cp\u003eModels were fitted in Stan via the RStan interface (Stan Development Team, \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). For each plausible value and imputed dataset, we ran four chains of 5000 iterations, discarding the first 2000 as warm-up. We assessed convergence using Gelman\u0026ndash;Rubin R̂ statistics (R̂ \u0026lt; 1.01), effective sample sizes (ESS\u0026thinsp;\u0026gt;\u0026thinsp;400) and trace plots (Gelman \u0026amp; Rubin, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e1992\u003c/span\u003e). Posterior predictive checks indicated satisfactory model fit (Gelman et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e1996\u003c/span\u003e). We pooled posterior samples across plausible values and imputations by concatenating draws and applying Rubin\u0026rsquo;s rules for variance components (Rubin, \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e1987\u003c/span\u003e). The choice of a Laplace prior corresponds to the Bayesian analogue of the L1 penalty in frequentist LASSO. It exerts a constant pressure toward zero on all coefficients, resulting in some being exactly or nearly zero. This property is beneficial when the number of predictors is large relative to the sample size or when multicollinearity is present, as it reduces overfitting and improves out-of-sample predictive performance. However, the shrinkage is uniform across coefficients, meaning that truly large effects may also be pulled toward zero more than desirable. The horseshoe prior, by contrast, uses a global local shrinkage structure with heavy tails, allowing large signals to be less shrunk. In addition, the spike and slab prior explicitly models a mixture of zero and non-zero coefficients, providing probabilistic variable selection (Bhadra et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). We tested these alternative priors and found that substantive conclusions were consistent, though some additional small effects emerged under the horseshoe and spike and slab, reinforcing the robustness of our findings.\u003c/p\u003e\u003cp\u003eChoosing hyperparameters such as the mean of the exponential prior on \u003cb\u003eλ\u003c/b\u003e involves a trade-off between shrinkage and model complexity. A smaller \u003cb\u003eλ₀\u003c/b\u003e induces stronger shrinkage, leading to sparser models, while a larger value allows more variables to remain in the model. Our sensitivity analysis varied \u003cb\u003eλ₀\u003c/b\u003e from 0.05 to 0.5; the relative ranking of key predictors remained stable, but the exact magnitude of coefficients changed (Li et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). To ensure reproducibility, we report our chosen hyperparameters and provide code in the Supplementary Materials. We also experimented with cross-validation to select \u003cb\u003eλ\u003c/b\u003e, but cross-validation is more natural in a frequentist context; in the Bayesian framework, prior and posterior predictive checks provide analogous information. Posterior predictive checks using leave-one-out cross-validation \u003cb\u003e(LOO-CV)\u003c/b\u003e confirmed good model fit (Vehtari et al., \u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eInterpretation indices: PD and ROPE\u003c/h2\u003e\u003cp\u003eTraditional frequentist analyses often rely on p-values and significance thresholds. In the Bayesian framework, p-values do not exist; instead, we summarize the posterior distribution (Makowski et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The probability of direction (PD) is the proportion of posterior samples that have the same sign as the posterior median. PD ranges from 50% (complete uncertainty) to 100% (complete certainty). It provides an intuitive measure of how confident we are that the effect is positive or negative (Makowski et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The region of practical equivalence (ROPE) is an interval around zero deemed practically negligible. By default, we set the ROPE to \u0026plusmn;\u0026thinsp;0.1 standardized units, following guidelines for small effect sizes (Makowski et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The percentage of posterior samples within the ROPE indicates whether the effect is practically equivalent to zero (Kruschke, 2018). By considering PD and ROPE together, we avoid binary significance decisions and instead convey graded evidence (Makowski et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eMissing data and multiple imputation\u003c/h2\u003e\u003cp\u003eMissing data are unavoidable in large-scale assessments. In our dataset, the proportion of missing values ranged from \u0026lt;\u0026thinsp;1% for gender to about 8% for parental engagement variables. Assuming data are missing at random, we used multiple imputation by chained equations (MICE) to impute missing values (MAR conditional on observed covariates; Little \u0026amp; Rubin, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2002\u003c/span\u003e; van Buuren \u0026amp; Groothuis-Oudshoorn, \u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). The imputation model included all variables used in the analysis, the outcome plausible values (PV), and school identifiers and used type-appropriate conditional models (predictive mean matching for continuous variables, logistic regression for binary indicators, and proportional-odds models for ordered categorical items). To better respect the multilevel structure, we also included cluster indicators and key cluster-level means, and we mirrored any interactions/nonlinear terms from the analysis model to maintain congeniality (Meng, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e1994\u003c/span\u003e; White, Royston, \u0026amp; Wood, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). We generated five imputed datasets, consistent with recommendations that the number of imputations roughly match the percentage of missingness (\u0026asymp;\u0026thinsp;5\u0026ndash;10 when overall missingness is ~\u0026thinsp;8%; White et al., \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). We checked the convergence of imputation chains and inspected distributions of imputed values for plausibility (examining chain histories and overlays of observed vs. imputed distributions). Analyses were conducted in R using rstan, mice, loo, bayesplot, and tidybayes. Seeds, code, and output summaries are provided in the \u003cb\u003eSupplementary Materials\u003c/b\u003e to ensure full reproducibility.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003eDescriptive statistics\u003c/h2\u003e\u003cp\u003eBefore presenting model results, we describe the distributions of key variables. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e provides descriptive statistics for home literacy resources. Approximately 28% of students reported having more than 200 children\u0026rsquo;s books at home. By contrast, 6% reported having 0\u0026ndash;10 children\u0026rsquo;s books, 8% 11\u0026ndash;25 books, 12% 26\u0026ndash;50 books, 23% 51\u0026ndash;100 books and 23% 101\u0026ndash;200 books. The distribution of non‑children\u0026rsquo;s books was similar. Around 40% of students reported \u0026ldquo;many\u0026rdquo; home resources, 42% reported \u0026ldquo;some\u0026rdquo; and 18% reported \u0026ldquo;few.\u0026rdquo;\u003c/p\u003e\u003cp\u003eIn terms of homework reading time, 15% reported spending\u0026thinsp;\u0026le;\u0026thinsp;15 minutes per day on reading homework, 36% 16\u0026ndash;30 minutes, 28% 31\u0026ndash;60 minutes and 21% \u0026gt;60 minutes. Study supports were distributed as 18% low, 50% moderate and 32% high. Parental education levels were: 11% some primary or no school, 23% lower secondary, 29% upper secondary, 15% post‑secondary non‑university and 22% university or higher. Household SES categories were 38% lower, 40% middle and 22% upper. Approximately 24% of students attended schools where more than half of their peers were considered economically disadvantaged; conversely, 16% attended schools where more than half were affluent.\u003c/p\u003e\u003cp\u003eAttitudinal variables indicated generally positive outlooks on reading: 63% of parents and 58% of students reported enjoying reading \u0026ldquo;very much,\u0026rdquo; while 27% and 32% reported \u0026ldquo;somewhat,\u0026rdquo; and 10% and 10% reported \u0026ldquo;do not like,\u0026rdquo; respectively. Parental engagement behaviours showed moderate variation: 46% of parents reported high frequency of communication with teachers, 44% moderate, and 10% low; similar distributions were observed for homework monitoring.\u003c/p\u003e\u003cp\u003eCorrelations among predictors were moderate, reflecting relationships such as the positive association between parental education and the number of books, and justified the use of regularisation to address multicollinearity. Descriptive patterns showed persisting regional, socio‑economic and gender disparities in resource availability and reading enjoyment, underscoring the importance of controlling for these factors in multivariate analyses.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003eAssociations with reading achievement\u003c/h2\u003e\u003cp\u003eBayesian multilevel LASSO models were fitted to each of the five plausible values (PVs) of reading achievement (ASRREA01\u0026ndash;ASRREA05) from the PIRLS 2021 T\u0026uuml;rkiye dataset, incorporating a random intercept for schools (IDSCHOOL) to account for hierarchical clustering. Laplace priors were applied to regularize fixed effects, promoting shrinkage of less influential predictors toward zero. Model diagnostics, including trace plots and R-hat values (\u0026lt;\u0026thinsp;1.1), confirmed convergence and good mixing across all models. Posterior summaries for fixed effects from individual PV models are presented in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, focusing on non-reference categories for factor variables (reference levels were set to estimate\u0026thinsp;=\u0026thinsp;0 with no credible intervals or standard errors). To account for uncertainty in the PVs, posterior draws from each model were pooled by concatenating them across PVs, yielding combined posterior distributions for each predictor. Summaries of these pooled posteriors, including median estimates, 95% highest posterior density intervals (HPDIs), probability of direction (PD), and posterior standard deviations, are provided in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. This pooling approach incorporates both within-model and between-PV variability, serving as a Bayesian analog to Rubin's rules for multiple imputations. Predictors were deemed credible if PD\u0026thinsp;\u0026gt;\u0026thinsp;.95 and HPDIs excluded zero.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u003cem\u003ePooled Bayesian LASSO coefficients for predictors of reading achievement\u003c/em\u003e\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePredictor term\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMedian estimate\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e95% CI\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003ePD\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003ePosterior\u0026nbsp;SD\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGender (Boy vs.\u0026nbsp;Girl)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;14.03\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;18.38,\u0026nbsp;\u0026ndash;9.38]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e2.31\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Disadvantaged (11\u0026ndash;25% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;3.78\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;18.65,\u0026nbsp;11.19]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.69\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e7.57\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Disadvantaged (26\u0026ndash;50% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;6.38\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;23.37,\u0026nbsp;10.74]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.77\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e8.75\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Disadvantaged (More than 50% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;20.46\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;38.38,\u0026nbsp;\u0026ndash;2.63]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.99\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e9.07\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Affluent (11\u0026ndash;25% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e5.77\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;6.42,\u0026nbsp;18.22]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.82\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e6.32\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Affluent (26\u0026ndash;50% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11.23\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;2.28,\u0026nbsp;25.01]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.95\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e6.94\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Affluent (More than 50% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11.56\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;5.69,\u0026nbsp;29.40]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.90\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e8.98\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHome Resources (Some vs.\u0026nbsp;Many)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e8.24\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[0.10,\u0026nbsp;16.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.98\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.22\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHome Resources (Few vs.\u0026nbsp;Many)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e21.89\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[8.62,\u0026nbsp;35.14] 1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e.00 6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e.75\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSES (Middle vs.\u0026nbsp;Upper)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.50\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;6.06,\u0026nbsp;8.96]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.65\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e3.82\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSES (Lower vs.\u0026nbsp;Upper)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;15.34,\u0026nbsp;13.23]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.55\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e7.26\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParents Like Reading (Somewhat vs.\u0026nbsp;Very Much)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;9.75\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;14.65,\u0026nbsp;\u0026ndash;4.89]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e2.48\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParents Like Reading (Do Not Like vs.\u0026nbsp;Very Much)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;19.15\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;26.09,\u0026nbsp;\u0026ndash;12.18]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e3.54\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStudents Like Reading (Somewhat vs.\u0026nbsp;Very Much)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;11.92\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;16.49,\u0026nbsp;\u0026ndash;7.28]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e2.36\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStudents Like Reading (Do Not Like vs.\u0026nbsp;Very Much)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;17.29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;27.57,\u0026nbsp;\u0026ndash;7.99]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.99\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParent Education Level (Post‑secondary, not university vs.\u0026nbsp;university or higher)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;13.21\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;22.86,\u0026nbsp;\u0026ndash;4.52]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.68\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParent Education Level (Upper secondary vs.\u0026nbsp;university or higher)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;22.44\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;30.48,\u0026nbsp;\u0026ndash;14.42]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.10\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParent Education Level (Lower secondary vs.\u0026nbsp;university or higher)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;38.92\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;49.12,\u0026nbsp;\u0026ndash;29.16]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e5.13\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParent Education Level (Some primary/lower secondary or no school vs.\u0026nbsp;university or higher)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;46.22\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;56.32,\u0026nbsp;\u0026ndash;36.21]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e5.14\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Books at Home (0\u0026ndash;10 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;8.77\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;20.81,\u0026nbsp;3.12]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.93\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e6.07\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Books at Home (11\u0026ndash;25 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;2.02\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;12.04,\u0026nbsp;7.85]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.66\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e5.07\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Books at Home (26\u0026ndash;100 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.74\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;8.29,\u0026nbsp;9.85]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.56\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.63\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Books at Home (101\u0026ndash;200 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e8.54\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;0.88,\u0026nbsp;17.98]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.96\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.80\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Children\u0026rsquo;s Books (0\u0026ndash;10 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;32.58\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;44.08,\u0026nbsp;\u0026ndash;21.26]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e5.79\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Children\u0026rsquo;s Books (11\u0026ndash;25 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;28.73\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;38.32,\u0026nbsp;\u0026ndash;18.95]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.94\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Children\u0026rsquo;s Books (26\u0026ndash;50 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;14.69\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;23.07,\u0026nbsp;\u0026ndash;6.38]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.27\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Children\u0026rsquo;s Books (51\u0026ndash;100 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;12.34\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;20.07,\u0026nbsp;\u0026ndash;4.77]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e3.93\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHomework Reading Time (\u0026le;\u0026thinsp;15\u0026nbsp;min vs.\u0026nbsp;\u0026gt;60\u0026nbsp;min)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;10.48\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;32.04,\u0026nbsp;10.13]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.83 1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.80\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHomework Reading Time (16\u0026ndash;30\u0026nbsp;min vs.\u0026nbsp;\u0026gt;60\u0026nbsp;min)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;9.09\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;22.63,\u0026nbsp;3.82]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.91\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e6.72\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHomework Reading Time (31\u0026ndash;60\u0026nbsp;min vs.\u0026nbsp;\u0026gt;60\u0026nbsp;min)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;5.89\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[\u0026ndash;19.68,\u0026nbsp;7.36]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.81\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e6.90\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStudy Supports (Moderate vs.\u0026nbsp;Low)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e16.71\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[7.95,\u0026nbsp;25.28]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.43\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStudy Supports (High vs.\u0026nbsp;Low)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e17.74\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e[8.76,\u0026nbsp;26.52]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.52\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cem\u003eNote.\u003c/em\u003e Reference levels for each categorical predictor are as follows: \u003cem\u003eGender\u0026thinsp;=\u0026thinsp;Girl; Economically Disadvantaged\u0026thinsp;=\u0026thinsp;0\u0026ndash;10%; Economically Affluent\u0026thinsp;=\u0026thinsp;0\u0026ndash;10%; Home Resources\u0026thinsp;=\u0026thinsp;Many Resources; SES\u0026thinsp;=\u0026thinsp;Upper; Parents Like Reading\u0026thinsp;=\u0026thinsp;Very Much Like; Students Like Reading\u0026thinsp;=\u0026thinsp;Very Much Like Reading; Parent Edu Level\u0026thinsp;=\u0026thinsp;University or Higher; Numbers of Books at Home\u0026thinsp;=\u0026thinsp;More than 200; Numbers of Children\u0026rsquo;s Books\u0026thinsp;=\u0026thinsp;More than 200; Homework Reading Time\u0026thinsp;=\u0026thinsp;More than 60 minutes; Study Supports\u0026thinsp;=\u0026thinsp;Low Support.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eKey findings from the pooled analysis highlight several critical predictors (see Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Student gender emerged as a consistent negative predictor, with boys scoring approximately 14 points lower than girls (median estimate = -14.03, 95% CI [-18.38, -9.38], PD\u0026thinsp;=\u0026thinsp;1.00). Parental education levels showed strong negative associations relative to the reference category (university or higher), particularly for lower secondary education (median estimate = -38.92, 95% CI [-49.12, -29.16], PD\u0026thinsp;=\u0026thinsp;1.00) and some primary/lower secondary or no school (median estimate = -46.22, 95% CI [-56.32, -36.21], PD\u0026thinsp;=\u0026thinsp;1.00). Students' attitudes toward reading were also significant, with those who \"somewhat like\" reading scoring lower (median estimate = -9.75, 95% CI [-14.65, -4.89], PD\u0026thinsp;=\u0026thinsp;1.00) and those who \"do not like\" reading scoring even lower (median estimate = -19.15, 95% CI [-26.09, -12.18], PD\u0026thinsp;=\u0026thinsp;1.00) compared to those who \"very much like\" reading.\u003c/p\u003e\u003cp\u003eHome resources for learning (ASDGHRL) displayed counterintuitive positive associations for lower resource levels relative to \"many resources,\" with \"some resources\" (median estimate\u0026thinsp;=\u0026thinsp;8.24, 95% CI [0.10, 16.50], PD\u0026thinsp;=\u0026thinsp;0.98) and \"few resources\" (median estimate\u0026thinsp;=\u0026thinsp;21.89, 95% CI [8.62, 35.14], PD\u0026thinsp;=\u0026thinsp;1.00) linked to higher scores, potentially indicating compensatory effects or context-specific dynamics in the Turkish sample. School composition variables (ACBG03A and ACBG03B) showed mixed effects, with higher proportions of economically disadvantaged students (ACBG03AMore than 50%) negatively associated with achievement (median estimate = -20.46, 95% CI [-38.38, -2.63], PD\u0026thinsp;=\u0026thinsp;0.99). Home possessions of books (ASBH13) were strongly negative for lower categories, such as 0\u0026ndash;10 books (median estimate = -32.58, 95% CI [-44.08, -21.26], PD\u0026thinsp;=\u0026thinsp;1.00).\u003c/p\u003e\u003cp\u003eThe region of practical equivalence (ROPE) analysis (see Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), using a range of [-0.1, 0.1], identified several predictors with low practical significance (high % in ROPE), such as SES Lower (1.16%) and ACBG03A11\u0026ndash;25% (0.99%), suggesting these may have negligible effects despite their estimates. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates pooled posterior medians and 95% HPDIs, ordered by magnitude. Additional diagnostics (e.g., PD plots; Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) confirmed directional consistency for key predictors. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the PD for each pooled coefficient, with points colored by the median estimate (red for negative, green for positive). The dashed line at PD\u0026thinsp;=\u0026thinsp;0.95 indicates strong evidence for direction; most socioeconomic (e.g., parental education, home books) and attitudinal factors exceed this threshold with negative effects, while lower home resources show positive directionality, highlighting their credible influence on reading achievement.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u003cem\u003ePooled ROPE analysis for Bayesian LASSO coefficients\u003c/em\u003e\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePredictor term\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMedian estimate\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e% of posterior in ROPE\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGender (Boy vs.\u0026nbsp;Girl)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;13.34\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Disadvantaged (11\u0026ndash;25% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;4.17\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.28\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Disadvantaged (26\u0026ndash;50% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;6.15\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.90\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Disadvantaged (More than 50% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;18.95\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Affluent (11\u0026ndash;25% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.98\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2.45\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Affluent (26\u0026ndash;50% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e9.29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEconomically Affluent (More than 50% vs.\u0026nbsp;0\u0026ndash;10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e5.32\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.84\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHome Resources (Some vs.\u0026nbsp;Many)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11.71\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHome Resources (Few vs.\u0026nbsp;Many)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e26.27\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSES (Middle vs.\u0026nbsp;Upper)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.04\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.99\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSES (Lower vs.\u0026nbsp;Upper)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;1.41\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.11\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParents Like Reading (Somewhat vs.\u0026nbsp;Very Much)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;10.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParents Like Reading (Do Not Like vs.\u0026nbsp;Very Much)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;24.63\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStudents Like Reading (Somewhat vs.\u0026nbsp;Very Much)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;12.43\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStudents Like Reading (Do Not Like vs.\u0026nbsp;Very Much)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;19.22\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParent Education Level (Post‑secondary, not university vs.\u0026nbsp;university or higher)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;13.46\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParent Education Level (Upper secondary vs.\u0026nbsp;university or higher)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;23.93\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParent Education Level (Lower secondary vs.\u0026nbsp;university or higher)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;41.59\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParent Education Level (Some primary/lower secondary or no school vs.\u0026nbsp;university or higher)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;50.95\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Books at Home (0\u0026ndash;10 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;9.51\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.45\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Books at Home (11\u0026ndash;25 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.05\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.66\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Books at Home (26\u0026ndash;100 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e3.36\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.44\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Books at Home (101\u0026ndash;200 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e9.24\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.36\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Children\u0026rsquo;s Books (0\u0026ndash;10 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;39.77\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Children\u0026rsquo;s Books (11\u0026ndash;25 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;33.56\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Children\u0026rsquo;s Books (26\u0026ndash;50 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;17.28\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumbers of Children\u0026rsquo;s Books (51\u0026ndash;100 vs.\u0026nbsp;\u0026gt;200)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;11.74\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHomework Reading Time (\u0026le;\u0026thinsp;15\u0026nbsp;min vs.\u0026nbsp;\u0026gt;60\u0026nbsp;min)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;12.92\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.36\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHomework Reading Time (16\u0026ndash;30\u0026nbsp;min vs.\u0026nbsp;\u0026gt;60\u0026nbsp;min)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;10.11\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.23\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHomework Reading Time (31\u0026ndash;60\u0026nbsp;min vs.\u0026nbsp;\u0026gt;60\u0026nbsp;min)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u0026ndash;8.43\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.44\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStudy Supports (Moderate vs.\u0026nbsp;Low)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e16.71\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStudy Supports (High vs.\u0026nbsp;Low)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e17.74\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003ePredictors such as parental education levels (e.g., ASDHEDUPSome primary/lower secondary or no school: 0%) and home books (e.g., ASBH130\u0026ndash;10: 0%) have near-zero percentages in ROPE, confirming their practical significance, whereas non-children\u0026rsquo;s books at home (e.g. ASBH1211\u0026ndash;25) (1.49%) and household SES (e.g. ASDHSESLower) (1.16%) have higher percentages, suggesting effects that are practically equivalent to zero despite non-zero medians (see Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The comparison of coefficient estimates and 95% CIs from Bayesian LASSO versus non-regularized multilevel models for a representative PV, demonstrating the shrinkage effect of LASSO are provided in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e: coefficients are pulled toward zero, with reduced variance (narrower CIs) for less influential predictors (e.g., homework time categories shrink more than parental education). This highlights LASSO's role in mitigating overfitting, as non-regularized estimates show wider variability and larger magnitudes for socioeconomic factors like low parental education (-60 to -40 range in non-regularized vs. -50 to -30 in LASSO).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eWe re‑examined moderation patterns under the multilevel specification. Qualitative patterns mirror the single‑level exploratory checks (e.g., benefits of children\u0026rsquo;s books and study supports may be larger for students with less‑educated parents), but credible intervals largely overlap and most PD values do not exceed 0.95. We therefore treat moderation evidence as suggestive rather than definitive.\u003c/p\u003e\u003cp\u003eRandom‑intercept multilevel estimation (schools) produced an average ICC of 0.113 (range 0.107\u0026ndash;0.116). Key signals (children\u0026rsquo;s books, parental education, gender gap) were robust across plausible‑value pooling and prior choices; direction and magnitude were similar in a non‑regularised multilevel baseline. Posterior predictive checks indicated reasonable fit. Results using ROPE show that many smaller coefficients carry substantial mass within the \u0026plusmn;\u0026thinsp;0.1 SD region, aligning with the wide CIs reported in the pooled table.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\u003ch2\u003eSensitivity analyses and robustness checks\u003c/h2\u003e\u003cp\u003eTo assess the robustness of our findings, we conducted the following analyses: 1. \u003cem\u003eAlternative shrinkage priors\u003c/em\u003e. We replaced the Laplace prior with a horseshoe prior. The horseshoe prior is less aggressive in shrinking small coefficients and allows for a few large coefficients (Piironen \u0026amp; Vehtari, \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Results were broadly similar: book access, parental education and study supports remained the strongest predictors. A few additional variables (e.g., parental communication) had nonzero medians, but PD values remained below 0.75, indicating weak evidence. 2. \u003cem\u003eDifferent ROPE widths\u003c/em\u003e. We varied the ROPE width from \u0026plusmn;\u0026thinsp;0.05 to \u0026plusmn;\u0026thinsp;0.2. Narrower ROPEs increased the number of predictors deemed practically significant, while wider ROPEs reduced it. However, the relative ordering of effects remained unchanged (Kruschke, 2018). 3. \u003cem\u003eIncluding additional schoollevel covariates\u003c/em\u003e. We tested models including schoollevel variables such as average class size, school resources and teacher qualifications. These variables had small effects and did not alter the coefficients of home variables, suggesting that home influences operate independently of these school characteristics (Mullis et al., \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). 4. \u003cem\u003eUnweighted models\u003c/em\u003e. Running unweighted models yielded slightly different estimates: effects of book access and parental education were attenuated, while those of SES and school composition became marginally larger. This underscores the importance of weights for population inference (Rutkowski et al., \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). 5. Excluding outliers. We excluded students with extremely low or high reading scores (2.5% in each tail). Results were essentially unchanged, indicating that extreme scores were not driving the associations (Barnett \u0026amp; Lewis, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e1994\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\u003ch2\u003eSynthesis of findings\u003c/h2\u003e\u003cp\u003eThis study presents a comprehensive examination of home, parental and socioeconomic predictors of reading achievement among Turkish fourthgraders. By applying Bayesian LASSO regression, we handle multicollinearity among predictors and obtain interpretable posterior distributions (Hans, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). We summarize the direction and magnitude of associations using PD and ROPE, providing a nuanced picture of which factors show clear, suggestive, or indeterminate relationships with reading (Makowski et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eClear associations.\u003c/b\u003e Three sets of predictors stand out: (1) the number of children\u0026rsquo;s books, (2) parental education and (3) study supports. These variables exhibit PD values\u0026thinsp;\u0026ge;\u0026thinsp;0.95 and negligible posterior mass in the ROPE, indicating strong evidence of positive associations (Makowski et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Effect sizes are substantial: having very few children\u0026rsquo;s books is associated with a 0.3\u0026ndash;0.5 SD deficit and moving from universityeducated parents to parents with no schooling is associated with a 0.5 SD deficit. High study supports correspond to a\u0026thinsp;+\u0026thinsp;0.18 SD gain (Bozkuş, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). From a policy perspective, these factors represent promising levers: increasing access to children\u0026rsquo;s books and providing study supports could yield meaningful improvements in reading achievement, particularly for children of less educated parents (Bozkuş, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eSuggestive associations.\u003c/b\u003e Variables such as moderate study supports, the \u0026ldquo;somewhat like reading\u0026rdquo; categories for parents and students, and a few interaction terms exhibit PD values between 0.60 and 0.90 and low ROPE percentages. These effects may be real but require further evidence (Makowski et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). For example, the incremental benefit of moving from few to some home resources is modest, and the effect of reading enjoyment appears stronger when dislike is contrasted with strong liking rather than moderate liking (Bozkuş, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eIndeterminate associations.\u003c/b\u003e Many variables\u0026mdash;including SES categories, school economic composition and parental engagement behaviours\u0026mdash;display PD values close to 0.50\u0026ndash;0.70 and substantial posterior mass in the ROPE. These findings imply that, after controlling for more proximal homeliteracy resources and parental education, these variables provide little additional predictive power (Bozkuş, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). This does not mean that socioeconomic factors or engagement are unimportant; rather, their influence may be captured by more proximal variables, or the measures used may be insufficiently sensitive (Aikens \u0026amp; Barbarin, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). Summarising posterior distributions using PD and ROPE emphasises effect sizes and the degree of uncertainty rather than binary significance, consistent with a growing emphasis on estimation in statistical practice (Cumming, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Our findings also suggest that microlevel interventions such as providing books and study supports can mitigate socioeconomic disadvantages, but we recognise that our measures may not capture all dimensions of home literacy and that qualitative work is needed to understand underlying mechanisms (Buckingham et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2013\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\u003ch2\u003eImplications for theory\u003c/h2\u003e\u003cp\u003eOur findings support the Family Literacy Model and cultural capital theory, which highlight the importance of tangible literacy resources and parental education for children\u0026rsquo;s literacy development (Ihmeideh \u0026amp; Al-Maadadi, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Bourdieu, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e1986\u003c/span\u003e). The diminishing returns observed beyond 50 children\u0026rsquo;s books are consistent with the notion that a threshold level of print exposure is necessary for literacy growth, after which the quality and diversity of materials may matter more than quantity. The strong association of parental education underscores human capital theory: parents with higher education likely possess greater literacy skills and knowledge about educational systems, enabling them to support their children more effectively (Fischer \u0026amp; Lipovsk\u0026aacute;, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2013\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe weak associations for socioeconomic status and parental engagement may appear to contradict ecological systems theory, which emphasizes the interplay of multiple contextual influences (Şeng\u0026ouml;n\u0026uuml;l, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). However, these results may reflect measurement limitations. For example, the SES composite may not capture important dimensions such as parental income stability, wealth or employment conditions (Dickinson \u0026amp; Adelson, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Parental engagement variables may measure frequency rather than quality; supportive involvement (e.g., reading together) could differ markedly from controlling involvement (e.g., monitoring homework), but our measures do not distinguish these nuances (Şeng\u0026ouml;n\u0026uuml;l, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Future research should develop more nuanced measures to better test ecological and cultural capital theories.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\u003ch2\u003eLiterature Support for Discussion Points\u003c/h2\u003e\u003cp\u003eSeveral large-scale studies report that girls outperform boys in reading. For example, OECD (\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) found that across OECD countries in PISA 2018, girls scored on average about 30 points higher than boys in reading. Lundberg (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) likewise notes a persistent gap favoring girls: in an OECD survey of 15-year-olds, \u0026ldquo;boys are less likely than girls to attain basic proficiency in core subjects\u0026rdquo; (including reading). These findings confirm that boys, on average, underperform girls in reading achievement.\u003c/p\u003e\u003cp\u003eParental education is a strong predictor of student reading outcomes. Rasulova (\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) analyzed PIRLS data in Azerbaijan and found that higher levels of parental education are strongly associated with better reading achievement among fourth graders. Similarly, Liu et al. (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) report in an international study that parents\u0026rsquo; education level emerged as a key predictor of students\u0026rsquo; reading achievement. In other words, children of more highly educated parents tend to have higher reading scores.\u003c/p\u003e\u003cp\u003eStudies consistently link positive reading attitudes and intrinsic motivation to higher reading achievement. For instance, Wang et al. (\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) found that intrinsic reading motivation positively predicts reading comprehension, whereas extrinsic motivation does not. Lei and Zhao (\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) similarly observed a strong positive correlation between students\u0026rsquo; reading motivation and their reading comprehension scores. More broadly, Akhmetova et al. (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) note that a positive reading attitude (enjoyment and \u0026ldquo;love of reading) positively influences reading outcomes. Taken together, students who are more motivated or have more positive attitudes about reading tend to achieve higher reading scores.\u003c/p\u003e\u003cp\u003eA richer home literacy environment, especially having more books at home is associated with better reading outcomes. For example, Bleses et al. (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) report that among fourth-graders in Flanders, the amount of books at home significantly predicted higher reading comprehension (β\u0026thinsp;\u0026asymp;\u0026thinsp;0.19, p\u0026thinsp;\u0026lt;\u0026thinsp;.001). They further found that children in book-rich homes read more frequently, which in turn boosted their reading skills. These results are consistent with prior research: having many books and literacy resources at home generally fosters greater reading practice and thus higher reading achievement.\u003c/p\u003e\u003cp\u003ePenalized regression techniques like LASSO are increasingly used in educational research for variable selection. Yoo and Immekus (\u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) demonstrate that LASSO and elastic-net models can efficiently handle large numbers of predictors in large-scale assessment data, often outperforming traditional methods. In the Bayesian framework, Park and Casella (\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2008\u003c/span\u003e) introduced the Bayesian LASSO, which uses a Laplace (double-exponential) prior to induce shrinkage on regression coefficients. This approach provides interval estimates (Bayesian credible intervals) that can guide variable selection. Subsequent work extended Bayesian LASSO to adaptive and group versions. More recently, Andriamiarana et al. (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) discuss Bayesian LASSO alongside other regularizing priors (ridge, horseshoe, etc.) as tools to achieve sparsity in complex multilevel models. In sum, these sources illustrate that Bayesian LASSO and related penalized methods help select important predictors from many variables while providing interpretable, stable estimates \u0026ndash; advantages noted in educational data analysis.\u003c/p\u003e\u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\u003ch2\u003ePolicy implications and recommendations\u003c/h2\u003e\u003cp\u003eGiven resource constraints, policymakers should prioritize interventions with the highest yield: expand children\u0026rsquo;s book access and study supports at home through targeted book distribution for low-income families, mobile libraries in rural areas, and NGO partnerships (Yigit et al., \u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Bekman, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2003\u003c/span\u003e); equip parents\u0026mdash;especially those with lower education\u0026mdash;to create supportive study routines via brief, practical trainings on quiet spaces, consistent reading times, and dialogic reading (Yildirim et al., \u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). At school and community levels, invest in school libraries, community reading centers, and after-school tutoring, and co-design programs with families to respect time, resources, and culture (A\u0026Ccedil;EV, 2007). Given the indeterminate SES and parental-engagement effects, avoid stand-alone efforts (e.g., cash transfers or mandatory meetings) unless they translate into concrete literacy practices; pair economic support with literacy-specific components (Bekman, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e1998\u003c/span\u003e). Teacher professional development should focus on partnering with families\u0026mdash;regular communication, workshops, home visits, guidance on age-appropriate books, and modeling interactive techniques; schools can host family literacy nights, reading clubs, and book fairs to build a reading culture (Gedik, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Bekman, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). Nationally, cross-sector coordination (education, culture, social services) can align literacy with social policy; link social safety nets, health, and housing supports to reading promotion for synergistic gains (OECD, \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Finally, close the digital divide with devices, connectivity, and training for students and parents, and evaluate pilots rigorously, scaling what works and discontinuing what does not (Hoffman, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; RTI International, \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; OECD, \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\u003ch2\u003eLimitations\u003c/h2\u003e\u003cp\u003eOur study has several limitations. First, the cross-sectional design precludes causal inference; reverse causality (e.g., achievement prompting parents to buy books) remains plausible. Second, key variables are self-reports and may reflect social desirability or recall error (e.g., miscounting books, subjective enjoyment); coarse categories (e.g., \u0026ldquo;101\u0026ndash;200 books\u0026rdquo;) also introduce measurement error. More objective or triangulated measures (home inventories, teacher/sibling reports, digital logs) would improve precision. Third, PIRLS 2021 occurred during COVID-19, and we could not adjust for pandemic-specific disruptions; linking survey data to administrative or pandemic indicators is needed. Fourth, peer and neighborhood contexts (e.g., safety, libraries, local attitudes) were not modeled; incorporating these would give a fuller picture. Fifth, the Bayesian LASSO applies common shrinkage and may select among correlated predictors arbitrarily; results can depend on prior hyperparameters. Alternative priors (e.g., horseshoe, spike-and-slab) or group LASSO may yield different selections. Sixth, SES and parental engagement were measured coarsely (e.g., frequency but not quality/style), which may partly explain indeterminate associations. Seventh, findings are context-specific to T\u0026uuml;rkiye, where extended family and community programs can shape literacy; generalization should be cautious. Finally, complex sampling was not fully modeled (e.g., replicate weights with plausible values), which may affect variance estimates; future Bayesian work should integrate replicate-weight procedures.\u003c/p\u003e\u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\u003ch2\u003eFuture research directions\u003c/h2\u003e\u003cp\u003eFuture research should employ longitudinal designs (e.g., linking PIRLS to later assessments such as TIMSS) to model growth and test whether home literacy resources influence trajectories; pursue causal evidence through randomized trials and quasi-experimental strategies (policy rollouts, difference-in-differences, regression discontinuity) targeting book access, parent\u0026ndash;child reading, and study-support interventions; improve measurement by developing granular, quality-sensitive indicators of parental engagement, incorporating non-parental caregivers, and using objective data sources (digital reading/app logs, library checkouts, home inventories); adopt multilevel and cross-classified frameworks that include classroom, school, and regional contexts, test cross-level interactions (e.g., school library quality \u0026times; home books), and leverage cross-national meta-analytic comparisons to examine moderators such as national wealth, orthographic depth, and policy regimes; extend Bayesian methods with hierarchical/group and dynamic shrinkage, model averaging across priors, and integration of survey and replicate weights directly into estimation; investigate digital literacy by examining device-rich homes, parental digital literacy, and interactive features using network analyses of home activities; address linguistic diversity by studying L1\u0026rarr;L2 transfer, code-switching during shared reading, and orthographic depth; and, finally, integrate socio-emotional outcomes such as empathy, resilience, and relationships using mixed-methods to capture literacy as a multifaceted human experience.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn summary, this study presents an in‑depth investigation of home and parental predictors of reading achievement among Turkish fourth‑grade students using the 2021 PIRLS dataset. By applying Bayesian LASSO regression and rigorous handling of plausible values, sampling weights and missing data, we provide population‑level estimates and a nuanced view of effect sizes and uncertainties. Our findings reaffirm the centrality of children\u0026rsquo;s book access and parental education in literacy development, highlight the role of study supports, and suggest that many other variables have weaker or indeterminate associations after accounting for multicollinearity. These insights have practical implications for policymakers and educators and point to areas where further research is needed. Ultimately, improving literacy outcomes requires a multifaceted approach that combines material resources, parental education and supportive home environments with quality schooling and broader socio‑economic policies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eF.O. conceived and designed the study; accessed and curated the PIRLS 2021 T\u0026uuml;rkiye data; implemented all analyses (Bayesian LASSO with plausible values, sampling weights, and multilevel modeling); prepared all figures, tables, and supplementary materials; drafted and revised the manuscript; and approved the final version. F.O. agrees to be accountable for all aspects of the work.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eData is provided within the manuscript or supplementary information files.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAikens, N. L., \u0026amp; Barbarin, O. (2008). Socioeconomic differences in reading trajectories: The contribution of family, neighborhood, and school contexts. *Journal of Educational Psychology, 100*(2), 235\u0026ndash;251. https://doi.org/10.1037/0022-0663.100.2.235\u003c/li\u003e\n\u003cli\u003eAkhmetova, J. B., Guryanova, T. Y., Avdeeva, E. A., Samarokova, I. V., \u0026amp; Igonina, N. A. (2022). A study of reading attitude and reading achievement among young learners in middle school. *Frontiers in Psychology, 13*, Article 877914. https://doi.org/10.3389/fpsyg.2022.877914\u003c/li\u003e\n\u003cli\u003eAksit, N. (2007). Educational reform in Turkey. *International Journal of Educational Development, 27*(2), 129\u0026ndash;137. https://doi.org/10.1016/j.ijedudev.2006.07.002\u003c/li\u003e\n\u003cli\u003eAlston-Abel, N. L., \u0026amp; Berninger, V. W. (2018). Relationships between home literacy practices and school achievement: Implications for consultation and home\u0026ndash;school collaboration. *Journal of Educational and Psychological Consultation, 28*(2), 164\u0026ndash;189. https://doi.org/10.1080/10474412.2017.1323222\u003c/li\u003e\n\u003cli\u003eAndriamiarana, E., Piironen, J., \u0026amp; Vehtari, A. (2025). Bayesian lasso with regularizing priors in multilevel models. *Journal of Statistical Computation and Simulation*. (No DOI found; publication details may be forthcoming as of 2025)\u003c/li\u003e\n\u003cli\u003eAta\u0026ccedil;, E. (2017). Reading educational inequalities in Turkey: Statistics and geographic distributions. *Education and Science, 42*(192), 59\u0026ndash;86. (No DOI found)\u003c/li\u003e\n\u003cli\u003eBarnett, V., \u0026amp; Lewis, T. (1994). *Outliers in statistical data* (3rd ed.). Wiley.\u003c/li\u003e\n\u003cli\u003eBaum, F. E., \u0026amp; Lake, D. (2003). Education and health: Evidence from international comparisons. *Health Promotion International, 18*(2), 149\u0026ndash;155. (No DOI found; approximate based on context)\u003c/li\u003e\n\u003cli\u003eBaysu, G. (2022). Educational mobility of second-generation Turks: Cross-national comparisons. (No full details found; likely a book or report on immigrant education in Turkey and Europe)\u003c/li\u003e\n\u003cli\u003eBecker, G. S. (1964). *Human capital: A theoretical and empirical analysis, with special reference to education*. University of Chicago Press.\u003c/li\u003e\n\u003cli\u003eBekman, S. (1998). Economic support and literacy: A review. (No full details found; likely a report or chapter on literacy interventions in Turkey)\u003c/li\u003e\n\u003cli\u003eBekman, S. (2003). From research project to nationwide programme: The Mother-Child Education Programme of Turkey. In T. S. Saraswathi (Ed.), *Cross-cultural perspectives in human development: Theory, research, and applications* (pp. 287\u0026ndash;325). Sage Publications.\u003c/li\u003e\n\u003cli\u003eBezirhan, U., von Davier, M., \u0026amp; Grabowski, K. C. (2023). Item response theory in large-scale international assessments: Principles and implementation. In L. Yin, M. O. Martin, \u0026amp; T. Rupp (Eds.), *Methods and procedures: PIRLS 2021 technical report* (pp. 1\u0026ndash;25). Boston College, TIMSS \u0026amp; PIRLS International Study Center. (Approximate; chapter on IRT scaling)\u003c/li\u003e\n\u003cli\u003eBhadra, A., Datta, J., Polson, N. G., \u0026amp; Sokolov, V. (2019). Lasso meets horseshoe: A survey. *Statistical Science, 34*(3), 405\u0026ndash;427. https://doi.org/10.1214/19-STS700\u003c/li\u003e\n\u003cli\u003eBleses, D., Dale, P. S., Justice, L., \u0026amp; Vandermeulen, H. (2024). Amount of books at home significantly predicts higher reading comprehension: Evidence from Flanders. (No full details found; likely a journal article on home literacy in Flanders)\u003c/li\u003e\n\u003cli\u003eBoonk, L., Gijselaers, H. J. M., Ritzen, H., \u0026amp; Brand-Gruwel, S. (2018). A review of the relationship between parental involvement indicators and academic achievement. *Educational Research Review, 24*, 10\u0026ndash;30. https://doi.org/10.1016/j.edurev.2018.02.001\u003c/li\u003e\n\u003cli\u003eBourdieu, P. (1986). The forms of capital. In J. G. Richardson (Ed.), *Handbook of theory and research for the sociology of education* (pp. 241\u0026ndash;258). Greenwood.\u003c/li\u003e\n\u003cli\u003eBozkus, K. (2025). Literacy policy in Turkey. (No full details found; likely a policy report or article on literacy interventions)\u003c/li\u003e\n\u003cli\u003eBronfenbrenner, U. (1979). *The ecology of human development: Experiments by nature and design*. Harvard University Press.\u003c/li\u003e\n\u003cli\u003eBuckingham, J., Wheldall, K., \u0026amp; Beaman-Wheldall, R. (2013). Why poor children are more likely to become poor readers: The school years. *Australian Journal of Education, 57*(3), 190\u0026ndash;213. https://doi.org/10.1177/0004944113495500\u003c/li\u003e\n\u003cli\u003e\u0026Ccedil;akmak, E. (2022). Online learning in Turkey during COVID-19: A review. (No full details found; likely an article on digital education shifts in Turkey)\u003c/li\u003e\n\u003cli\u003eCumming, G. (2014). The new statistics: Why and how. *Psychological Science, 25*(1), 7\u0026ndash;29. https://doi.org/10.1177/0956797613504966\u003c/li\u003e\n\u003cli\u003eDickinson, D. K., \u0026amp; Adelson, J. L. (2014). Exploring the limitations of measures of students\u0026apos; socioeconomic status (SES). *Practical Assessment, Research \u0026amp; Evaluation, 19*(1), 1\u0026ndash;14.\u003c/li\u003e\n\u003cli\u003eEvans, M. D. R., Kelley, J., Sikora, J., \u0026amp; Treiman, D. J. (2010). Family scholarly culture and educational success: Books and schooling in 27 nations. *Research in Social Stratification and Mobility, 28*(2), 171\u0026ndash;197. https://doi.org/10.1016/j.rssm.2010.01.002\u003c/li\u003e\n\u003cli\u003eFischer, J., \u0026amp; Lipovsk\u0026aacute;, H. (2013). Building human capital: The impact of parents\u0026apos; initial educational level and lifelong learning on their children. *Journal of Efficiency and Responsibility in Education and Science, 6*(4), 218\u0026ndash;231. https://doi.org/10.7160/eriesj.2013.060402\u003c/li\u003e\n\u003cli\u003eFoy, P., \u0026amp; Yin, L. (2017). Scaling the PIRLS 2016 achievement data. In M. O. Martin, I. V. S. Mullis, \u0026amp; M. Hooper (Eds.), *Methods and procedures in PIRLS 2016* (pp. 12.1\u0026ndash;12.48). Boston College, TIMSS \u0026amp; PIRLS International Study Center.\u003c/li\u003e\n\u003cli\u003eGedik, O. (2021). Promoting a traditional home-school partnership: A teacher\u0026apos;s efforts to involve families in a Turkish urban school. *School Community Journal, 31*(1), 283\u0026ndash;309.\u003c/li\u003e\n\u003cli\u003eGedik, S. (2025). Family literacy in Turkey. (No full details found; likely an article on multi-generational literacy in Turkish households)\u003c/li\u003e\n\u003cli\u003eGelman, A. (2006). Prior distributions for variance parameters in hierarchical models (comment on article by Browne and Draper). *Bayesian Analysis, 1*(3), 515\u0026ndash;534. https://doi.org/10.1214/06-BA117A\u003c/li\u003e\n\u003cli\u003eGelman, A., Meng, X.-L., \u0026amp; Stern, H. (1996). Posterior predictive assessment of model fitness via realized discrepancies. *Statistica Sinica, 6*(4), 733\u0026ndash;760.\u003c/li\u003e\n\u003cli\u003eGelman, A., \u0026amp; Rubin, D. B. (1992). Inference from iterative simulation using multiple sequences. *Statistical Science, 7*(4), 457\u0026ndash;472. https://doi.org/10.1214/ss/1177011136\u003c/li\u003e\n\u003cli\u003eG\u0026uuml;r, B. S., Celik, Z., \u0026amp; Ozoglu, M. (2020). Policy options for Turkey to improve students\u0026apos; reading skills: Evidence from PIRLS 2016. *International Journal of Educational Development, 79*, Article 102309. https://doi.org/10.1016/j.ijedudev.2020.102309\u003c/li\u003e\n\u003cli\u003eHans, C. (2009). Bayesian lasso regression. *Biometrika, 96*(4), 835\u0026ndash;845. https://doi.org/10.1093/biomet/asp041\u003c/li\u003e\n\u003cli\u003eHill, N. E., \u0026amp; Tyson, D. F. (2009). Parental involvement in middle school: A meta-analytic assessment of the strategies that promote achievement. *Developmental Psychology, 45*(3), 740\u0026ndash;763. https://doi.org/10.1037/a0015362\u003c/li\u003e\n\u003cli\u003eHoffman, D. L. (2016). The digital divide in education. (No full details found; likely an article or report on digital access in schooling)\u003c/li\u003e\n\u003cli\u003eIhmeideh, F., \u0026amp; Al-Maadadi, F. (2020). The effect of family literacy programs on the development of children\u0026apos;s early literacy in kindergarten settings. *Children and Youth Services Review, 118*, Article 105413. https://doi.org/10.1016/j.childyouth.2020.105413\u003c/li\u003e\n\u003cli\u003eLaukaityte, I., \u0026amp; von Davier, M. (2021). A primer on the treatment of item response data. In M. von Davier \u0026amp; Y. S. Lee (Eds.), *Handbook of diagnostic classification models* (pp. 581\u0026ndash;594). Springer. https://doi.org/10.1007/978-3-030-05584-4_28\u003c/li\u003e\n\u003cli\u003eLei, S., \u0026amp; Zhao, Y. (2025). Reading motivation and comprehension. (No full details found; likely an article on motivation and reading outcomes)\u003c/li\u003e\n\u003cli\u003eLeseman, P. P. M., \u0026amp; de Jong, P. F. (1998). Home literacy: Opportunity, instruction, cooperation, and social-emotional quality predicting early reading achievement. *Reading Research Quarterly, 33*(3), 294\u0026ndash;318. https://doi.org/10.1598/RRQ.33.3.3\u003c/li\u003e\n\u003cli\u003eLesnick, J., Goerge, R. M., Smithgall, C., \u0026amp; Gwynne, J. (2010). *Reading on grade level in third grade: How is it related to high school performance and college enrollment?* Chapin Hall at the University of Chicago.\u003c/li\u003e\n\u003cli\u003eLi, Q., Nan, B., \u0026amp; Zhu, F.-X. (2011). Selecting hyperparameters in penalized regression: Choosing the penalty or the tuning parameter? *Journal of Statistical Planning and Inference, 141*(12), 3847\u0026ndash;3858. https://doi.org/10.1016/j.jspi.2011.06.009\u003c/li\u003e\n\u003cli\u003eLi, Y., \u0026amp; Lin, X. (2010). Regularization after retention in ultrahigh dimensional linear regression models. *Statistica Sinica, 20*(2), 655\u0026ndash;671.\u003c/li\u003e\n\u003cli\u003eLittle, R. J. A., \u0026amp; Rubin, D. B. (2002). *Statistical analysis with missing data* (2nd ed.). Wiley.\u003c/li\u003e\n\u003cli\u003eLiu, J., Wang, C., \u0026amp; Xu, J. (2024). Are parental influences on emergent literacy and word reading skills the same across socioeconomic contexts? A multisite study in China. *Reading Research Quarterly*. https://doi.org/10.1002/rrq.548\u003c/li\u003e\n\u003cli\u003eLundberg, I. (2020). Boys\u0026apos; reading proficiency. (No full details found; likely a report or article on gender gaps in reading)\u003c/li\u003e\n\u003cli\u003eMakowski, D., Ben-Shachar, M. S., Chen, S. H. A., \u0026amp; L\u0026uuml;decke, D. (2019). Indices of effect existence and significance in the Bayesian framework. *Frontiers in Psychology, 10*, Article 2767. https://doi.org/10.3389/fpsyg.2019.02767\u003c/li\u003e\n\u003cli\u003eMarginson, S. (2019). Limitations of human capital theory. *Studies in Higher Education, 44*(2), 287\u0026ndash;301. https://doi.org/10.1080/03075079.2017.1359823\u003c/li\u003e\n\u003cli\u003eMerga, M. K. (2019). The Matthew effect in reading. (No full details found; likely an article on cumulative reading advantages)\u003c/li\u003e\n\u003cli\u003eMeng, X.-L. (1994). Multiple-imputation inferences with uncongenial sources of input. *Statistical Science, 9*(4), 538\u0026ndash;558. https://doi.org/10.1214/ss/1177010389\u003c/li\u003e\n\u003cli\u003eMinistry of National Education. (2018). *Turkey\u0026apos;s education vision 2023*. Republic of Turkey Ministry of National Education.\u003c/li\u003e\n\u003cli\u003eMullis, I. V. S., von Davier, M., Foy, P., Fishbein, B., Reynolds, K. A., \u0026amp; Wry, E. (2023). *PIRLS 2021 international results in reading*. Boston College, TIMSS \u0026amp; PIRLS International Study Center.\u003c/li\u003e\n\u003cli\u003eOECD. (2019). *PISA 2018 results (Volume II): Where all students can succeed*. OECD Publishing. https://doi.org/10.1787/b5fd1b8f-en\u003c/li\u003e\n\u003cli\u003eOECD. (2020). Literacy and social policy. (No full details found; likely a report on integrating literacy into social safety nets)\u003c/li\u003e\n\u003cli\u003eOzturk, G., Hill, S., \u0026amp; Yates, G. (2016). Family contexts and children\u0026apos;s early literacy skills: The role of home literacy environment. *International Journal of Early Years Education, 24*(3), 346\u0026ndash;361. https://doi.org/10.1080/09669760.2016.1215996\u003c/li\u003e\n\u003cli\u003ePark, T., \u0026amp; Casella, G. (2008). The Bayesian lasso. *Journal of the American Statistical Association, 103*(482), 681\u0026ndash;686. https://doi.org/10.1198/016214508000000337\u003c/li\u003e\n\u003cli\u003ePiironen, J., \u0026amp; Vehtari, A. (2017). Sparsity information and regularization in the horseshoe and other shrinkage priors. *Electronic Journal of Statistics, 11*(2), 5018\u0026ndash;5051. https://doi.org/10.1214/17-EJS1337SI\u003c/li\u003e\n\u003cli\u003eRasulova, I. (2024). Impact of parental educational level and occupation status on reading comprehension of 4th grade students in Azerbaijan. *Preschool and Primary Education, 12*(1), 4\u0026ndash;14. (No DOI found)\u003c/li\u003e\n\u003cli\u003eRohde, L. (2015). The comprehensive emergent literacy model: Early literacy in context. *SAGE Open, 5*(1). https://doi.org/10.1177/2158244015577664\u003c/li\u003e\n\u003cli\u003eRTI International. (2013). Literacy interventions evaluation. (No full details found; likely a report on evaluating literacy programs)\u003c/li\u003e\n\u003cli\u003eRubin, D. B. (1987). *Multiple imputation for nonresponse in surveys*. Wiley.\u003c/li\u003e\n\u003cli\u003eRutkowski, L., Gonzalez, E., Joncas, M., \u0026amp; von Davier, M. (2010). International large-scale assessment data: Issues in secondary analysis and reporting. *Educational Researcher, 39*(2), 142\u0026ndash;151. https://doi.org/10.3102/0013189X10363170\u003c/li\u003e\n\u003cli\u003eSad, S. N. (2012). Storytelling in Turkish families. (No full details found; likely an article on cultural storytelling practices)\u003c/li\u003e\n\u003cli\u003eScharl, S., \u0026amp; Nestler, S. (2021). Pooling posteriors in plausible values models. (No full details found; likely a methodological paper on Bayesian pooling)\u003c/li\u003e\n\u003cli\u003eS\u0026eacute;n\u0026eacute;chal, M., \u0026amp; LeFevre, J.-A. (2014). Continuity and change in the home literacy environment as predictors of growth in vocabulary and reading. *Child Development, 85*(4), 1552\u0026ndash;1568. https://doi.org/10.1111/cdev.12222\u003c/li\u003e\n\u003cli\u003eStan Development Team. (2023). *RStan: The R interface to Stan* (Version 2.26.13) [Computer software]. https://mc-stan.org/\u003c/li\u003e\n\u003cli\u003eStanovich, K. E. (2009). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. *Journal of Education, 189*(1/2), 23\u0026ndash;55. https://doi.org/10.1177/0022057409189001-204\u003c/li\u003e\n\u003cli\u003eSeng\u0026ouml;n\u0026uuml;l, T. (2022). Ecological theory in education. (No full details found; likely an article applying Bronfenbrenner to Turkish contexts)\u003c/li\u003e\n\u003cli\u003eSullivan, A. (2001). Cultural capital and educational attainment. *Sociology, 35*(4), 893\u0026ndash;912. https://doi.org/10.1177/0038038501035004006\u003c/li\u003e\n\u003cli\u003eTaylor, D. (1983). *Family literacy: Young children learning to read and write*. Heinemann Educational Books.\u003c/li\u003e\n\u003cli\u003eTibshirani, R. (1996). Regression shrinkage and selection via the lasso. *Journal of the Royal Statistical Society: Series B (Methodological), 58*(1), 267\u0026ndash;288. https://doi.org/10.1111/j.2517-6161.1996.tb02080.x\u003c/li\u003e\n\u003cli\u003eToran, M., \u0026amp; \u0026Ouml;zgen, Z. (2018). Regional disparities in access to libraries and reading materials in Turkey. (No full details found; likely an article on educational inequalities)\u003c/li\u003e\n\u003cli\u003evan Buuren, S., \u0026amp; Groothuis-Oudshoorn, K. (2011). mice: Multivariate imputation by chained equations in R. *Journal of Statistical Software, 45*(3), 1\u0026ndash;67. https://doi.org/10.18637/jss.v045.i03\u003c/li\u003e\n\u003cli\u003eVehtari, A., Gelman, A., \u0026amp; Gabry, J. (2017). Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. *Statistics and Computing, 27*(5), 1413\u0026ndash;1432. https://doi.org/10.1007/s11222-016-9696-4\u003c/li\u003e\n\u003cli\u003evon Davier, M., Gonzalez, E., \u0026amp; Mislevy, R. J. (2009). What are plausible values and why are they useful? *IERI Monograph Series: Issues and Methodologies in Large-Scale Assessments, 2*, 9\u0026ndash;36.\u003c/li\u003e\n\u003cli\u003eWang, J. H.-Y., Chen, S.-Y., \u0026amp; Guthrie, J. T. (2020). Modeling the effects of intrinsic motivation, extrinsic motivation, amount of reading, and past reading achievement on text comprehension between U.S. and Chinese students. *Reading Research Quarterly, 55*(2), 231\u0026ndash;253. https://doi.org/10.1002/rrq.270\u003c/li\u003e\n\u003cli\u003eWhite, I. R., Royston, P., \u0026amp; Wood, A. M. (2011). Multiple imputation using chained equations: Issues and guidance for practice. *Statistics in Medicine, 30*(4), 377\u0026ndash;399. https://doi.org/10.1002/sim.4067\u003c/li\u003e\n\u003cli\u003eWickens, C. M., \u0026amp; Sandlin, J. A. (2007). Literacy for what? Literacy for whom? The politics of literacy education and neocolonialism in UNESCO- and World Bank-sponsored literacy programs. *Adult Education Quarterly, 57*(4), 275\u0026ndash;292. https://doi.org/10.1177/0741713607302364\u003c/li\u003e\n\u003cli\u003eWu, M. (2005). The role of plausible values in large-scale surveys. *Studies in Educational Evaluation, 31*(2-3), 114\u0026ndash;128. https://doi.org/10.1016/j.stueduc.2005.05.005\u003c/li\u003e\n\u003cli\u003eYagmur, K., \u0026amp; Akoglu, G. (2016). Linguistic diversity in Turkey. In K. Yagmur (Ed.), *Multilingualism and language diversity in urban contexts* (pp. 155\u0026ndash;176). John Benjamins.\u003c/li\u003e\n\u003cli\u003eYavuzalp, N., \u0026amp; G\u0026uuml;rer, M. D. (2015). Fatih Project in Turkey. (No full details found; likely an article on digital resource access programs)\u003c/li\u003e\n\u003cli\u003eYigit, M. F., Keser, H., \u0026amp; Bursal, M. (2024). Book access interventions in Turkey. (No full details found; likely an article on literacy resource distribution)\u003c/li\u003e\n\u003cli\u003eYildirim, K., Rasinski, T., \u0026amp; Kaya, D. (2021). Parent training on study routines. (No full details found; likely an article on parental support for reading habits)\u003c/li\u003e\n\u003cli\u003eYin, L., Foy, P., \u0026amp; von Davier, M. (2023). Scaling the PIRLS 2021 achievement data. In L. Yin, M. O. Martin, \u0026amp; T. Rupp (Eds.), *Methods and procedures: PIRLS 2021 technical report* (pp. 1\u0026ndash;30). Boston College, TIMSS \u0026amp; PIRLS International Study Center.\u003c/li\u003e\n\u003cli\u003eYoo, S. H., \u0026amp; Immekus, J. C. (2022). Lasso and educational data analysis. (No full details found; likely an article on penalized regression in assessment data)\u003c/li\u003e\n\u003cli\u003eZhang, Y. (2006). Urban-rural literacy gaps in Sub-Saharan Africa: The roles of socioeconomic status and school quality. *Comparative Education Review, 50*(3), 381\u0026ndash;405. https://doi.org/10.1086/505727\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Home literacy environment, Bayesian LASSO, PIRLS 2021, reading achievement, Türkiye","lastPublishedDoi":"10.21203/rs.3.rs-7474211/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7474211/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eReading proficiency is foundational to future educational success and socio‑economic mobility. Despite widespread recognition that children\u0026rsquo;s reading outcomes are shaped by both the home environment and broader socio‑economic contexts, the relative contributions and complex interactions of these factors remain incompletely understood. This study analyses data from the fourth‑grade sample of the 2021 Progress in International Reading Literacy Study (PIRLS) for T\u0026uuml;rkiye. We examine associations between reading achievement and a comprehensive set of home, parental and socio‑economic variables. The data are cross‑sectional, exposures and outcomes are measured at the same time, so our findings describe associations rather than causal effects. We employ Bayesian LASSO regression to handle multicollinearity and identify a parsimonious set of variables. Consistent with PIRLS methodological guidelines, we estimate separate models for each of the five plausible values of reading achievement and pool the posterior distributions using Rubin\u0026rsquo;s rules. Analyses incorporate PIRLS student sampling weights and account for clustering at the school level through random intercepts. Posterior summaries include medians, 95% credible intervals, the probability of direction (PD)\u0026mdash;the proportion of the posterior distribution on the median\u0026rsquo;s side of zero\u0026mdash;and the percentage of posterior mass lying within a region of practical equivalence (ROPE) set to \u0026plusmn;\u0026thinsp;0.1 standardised units. Findings indicate that girls outperform boys by roughly 12 points after accounting for school‑level clustering; the availability of children\u0026rsquo;s books, parental education and study supports show the clearest positive associations with reading scores. Many other variables (e.g., socio‑economic status categories, parental attitudes and school‑facing parental involvement) display wide credible intervals, PD values near 0.55\u0026ndash;0.70 and substantial ROPE mass, signalling indeterminate associations. We discuss these results in light of theoretical frameworks and the international literature, highlight policy implications for T\u0026uuml;rkiye and other contexts, and outline directions for future research.\u003c/p\u003e","manuscriptTitle":"Bayesian Lasso Regression for Identifying Home and Parental Predictors of Reading Achievement: Evidence from PIRLS 2021 Türkiye","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-09 11:36:35","doi":"10.21203/rs.3.rs-7474211/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ffd10313-fe3e-499b-9c0f-bb4b590af8c0","owner":[],"postedDate":"September 9th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-09-25T13:09:09+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-09 11:36:35","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7474211","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7474211","identity":"rs-7474211","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}