{"paper_id":"d3692cbb-540b-44a7-b9fa-396ad08e6d5b","body_text":"Parent versus Student Reporting Drives Contradictory Gender Patterns in Socioeconomic Gradients | 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 Parent versus Student Reporting Drives Contradictory Gender Patterns in Socioeconomic Gradients Kimmo Eriksson This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8743057/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 Research on how socioeconomic status (SES) and gender interact to affect student achievement has produced contradictory findings. While some studies suggest boys are more vulnerable to socioeconomic disadvantage, others find the opposite. This paper argues that these conflicting results stem from a critical methodological artifact: because of greater measurement error in boys' proxy reports, the source of SES data in large-scale assessments is critical. PISA 2006 and 2009 provides data on 151,269 students from 20 countries for whom both parent-reported and student-reported SES data were available. This within-subject design allowed for a direct comparison of the SES-gender interaction in mathematics achievement, isolating the effect of the data source. The direction of the SES-gender interaction systematically reverses depending on the data source. Aligning with the \"vulnerable boys\" hypothesis, the SES achievement gradient is typically steeper for boys when using reliable parent-reported data; this holds in 18 out of 31 country-waves. In 12 of those 18 cases, the results reverse to instead indicate “vulnerable girls” when using the more commonly available student-reported proxy data. The choice of SES data source is not a minor technical detail but a critical factor that can lead to flawed conclusions about educational equity. Greater measurement error in boys' proxy reports artificially attenuates their SES gradient. Researchers should prioritize parent-reported data for this type of analysis, and findings based on student-reported SES should be interpreted with caution. socioeconomic status gender differences measurement error large-scale assessments PISA Figures Figure 1 Introduction Understanding how socioeconomic status (SES) and gender interact to influence student achievement represents a critical question in educational equity research. The answer shapes our understanding of which students are most academically vulnerable and guides the design of interventions to address educational inequality. The theoretical \"vulnerable boys\" hypothesis suggests that boys are more academically susceptible to socioeconomic disadvantage than girls (Autor et al., 2019 ). This perspective is supported by several proposed mechanisms, such as differential parental investment patterns and gendered behavioral responses to adversity. Yet despite extensive research using large-scale data, the field presents a puzzling landscape of contradictory findings. For example, a study of PIRLS and TIMSS data from 36 countries found that SES gradients in achievement —that is, differences in academic performance associated with socioeconomic status—were steeper for boys in most countries (Eriksson & Lindvall, 2023 ). In contrast, other large-scale assessments have produced opposing results; a comprehensive meta-analysis of international assessments concluded that SES gradients in achievement are, on average, slightly steeper for girls (Liu et al., 2022 ). The use of large-scale assessment data to support two mutually exclusive hypotheses—either that boys are more vulnerable or that girls are—raises important questions about the validity of these conclusions. Therefore, identifying the underlying reasons for these divergent findings is essential. We propose that a key factor underlying these conflicting results is whether studies use SES data reported by parents or by the students themselves. Large-scale assessments of adolescents—such as PISA (Programme for International Student Assessment), TIMSS (Trends in International Mathematics and Science Study), ICCS (International Civic and Citizenship Education Study), and ICILS (International Computer and Information Literacy Study)—often rely on students to provide information about their parents' background (OECD, 2024 ; von Davier et al., 2024 ; Schultz et al., 2024; Fraillon et al., 2025 ). This practice, known as using student proxy reports, can introduce bias. Specifically, we argue that gender differences in the reliability of student-reported data make the choice of data source a critical factor in shaping study outcomes. From Attenuation to Sign Reversal When a predictor variable contains random measurement error, standard regression analysis produces attenuation bias, meaning that the estimated effect is biased toward zero. The extent of this bias depends on the reliability of the measure, which refers to how much of the observed variation reflects true differences rather than random noise. Put simply, less reliable (noisier) data make statistical relationships appear weaker than they actually are. This general statistical phenomenon has specific implications for research on SES–gender interactions. While data on student gender is highly reliable regardless of who reports it, it is well-established that student participants in large-scale assessments provide significantly less reliable information about their parents’ socioeconomic status than parents themselves do (Jerrim & Micklewright, 2014 ). In particular, students’ reports of their parents’ educational attainment, which is a primary component of SES indices, are especially prone to error (Ensminger et al., 2000 ; Jerrim & Micklewright, 2014 ; Lien et al., 2001 ; Looker, 1989 ). The critical insight for the present study is that this measurement error is not uniform across genders. Research demonstrates that adolescent boys consistently report their parents' background characteristics less accurately than girls do, in line with broader evidence of gender differences in school-related behaviors, including lower test-taking effort among boys (Soland, 2019 ) and higher levels of conscientiousness and attention among girls (Duckworth & Seligman, 2006 ). Prior analyses of the accuracy of students' reports of their parents' education in PISA data found that errors were on average several percentage points more common among boys than girls (Eriksson et al., 2026 ; see also Supplementary Table 1). Importantly, this gender difference in accuracy was consistent across all levels of parental education, indicating random measurement error rather than systematic directional bias (e.g., boys systematically over- or under-reporting education levels). For the SES-gender interaction, gender-specific attenuation can actually reverse conclusions about which gender is most vulnerable. In parent-reported data, the actual SES gradient is steeper for boys than for girls (on average, 𝛽 boys = 16.4 vs. 𝛽 girls = 15.9, where 𝛽 represents the increase in achievement per unit increase in parental education). However, when student-reported SES is used instead, the SES gradient is more severely attenuated for boys than for girls (by 27% vs. 17%). As a result, the attenuated gradient for boys may fall below that for girls (𝛽 boys = 12.0 vs. 𝛽 girls = 13.2). Thus, differential attenuation can create the misleading appearance that the SES gradient is steeper for girls, even when the true relationship is the opposite. The Present Study This study tests whether the source of SES data—parent-reported versus student-reported—systematically affects estimates of the SES-gender interaction in academic achievement. We leverage large-scale assessment data with the unique feature that reports of parental education are available both from students and their parents. This within-subject design allows us to isolate the effect of the reporting source while holding the sample and underlying construct constant. If boys report parental education less accurately than girls, as prior research suggests, then student-reported data should produce attenuated SES gradients for boys relative to girls. We therefore predict that (1) the SES-gender interaction will be more negative (i.e., less supportive of the \"vulnerable boys\" hypothesis) when estimated from student-reported data than from parent-reported data, and (2) this bias will be systematic across countries. Methods We utilize data from PISA 2006 and 2009. These specific waves offer a unique methodological opportunity: they are the only iterations of PISA that collected parental education data directly from parents (via the optional parent questionnaire) in addition to the standard student reports for the same households (OECD, 2009 , 2012 ). This design creates a powerful natural experiment, allowing for a direct within-subject comparison that isolates the effect of the reporting source while holding the sample and the underlying construct constant. Participants We restricted the analysis to students with complete parent- and student-reported data on parental education for both parents. The proportion of excluded cases varied considerably across countries, from just a few percent to almost half the sample, with a mean of 25.2% (SD = 14.7%; see Supplementary Table 1 for details by country). However, exclusions based on missing data are not a serious concern for the purpose of demonstrating how estimates of the SES-gender interaction depends on whether parent reports or student reports are used, as parent non-response is unlikely to vary systematically by student gender. Additionally, prior research using these same PISA waves found that restricting analysis to cases with complete parent and student reports does not substantially bias cross-country comparisons of SES gradients (Jerrim & Micklewright, 2014 ). The final analysis includes 151,269 students, from 20 countries participating in PISA 2006 and/or 2009, with complete parent- and student-reported data on parental education. Measures Parental Education. This analysis focuses on parental education because it is the sole socioeconomic indicator available from both data sources (parent and student). The PISA questionnaires ask for the level of education of each parent using national adaptations of the International Standard Classification of Education scale. The parent questionnaire primarily asks about higher qualifications, whereas the student questionnaire also distinguishes between lower levels of schooling (OECD, 2009 , 2012 ). We recoded the data from both questionnaires to an identical four-point scale: bachelor's degree or higher (coded 3), shorter tertiary education (2), upper secondary education or post-secondary non-tertiary education (1), and any lower level of schooling (0). We then summed the values for the mother's and father's education to create a parental education sum score ranging from 0 to 6. While taking the maximum of the two parents' educational levels is common, prior research indicates the sum score is a more efficient predictor (Korupp et al., 2002 ). Mathematics Achievement. We focus on mathematics achievement in this study. PISA also assesses achievement in science, and reading, but all subjects exhibit very similar gradients with respect to parents' education (Eriksson et al., 2021 ). We utilized the plausible values provided by PISA, which are calibrated to a global mean of approximately 500 points and a standard deviation of 100 points (OECD, 2009 , 2012 ). Statistical Analysis To examine whether students’ proxy reports of parental education contain more random measurement error for boys than girls, we compute correlations between student-reported and parent-reported parental education separately by gender. We estimate the parental-education gradient on mathematics achievement by regressing student math scores (plausible values) on a parental education score. These analyses were performed separately for each country-wave, gender, and SES data source (parents or students) using SPSS 30.0 with syntax created by the IDB Analyzer 5.0 which ensures the appropriate application of weights and robust standard errors (IEA, 2025 ). The SES-gender interaction is calculated as the difference between the gradient for boys and the gradient for girls. To test whether the direction of bias was systematic, we examine how many country-waves show more negative interaction estimates with student-reported versus parent-reported data using the sign test. Results We begin by examining whether boys' proxy reports of parental education are less reliable than girls'. Confirming that boys' reports contain more random measurement error, the correlation between student-reported and parent-reported parental education was higher for girls than for boys in 29 of 31 country-waves (see Supplementary Table 1). Estimates of the SES-gender interaction change depending on the data source as illustrated in Table 1. The table reveals a striking and systematic reversal of findings. When using reliable parent reports of parental education, the results generally support the vulnerable boys hypothesis: a positive SES-gender interaction (meaning the SES achievement gradient is steeper for boys than for girls) was observed in the majority of country-waves (18 of 31, or 58.1%). However, when the same analysis was conducted using student reports of parental education—the data type most commonly available in PISA—the direction of the interaction reversed. In most country-waves (25 of 31, or 80.6%), the SES-gender interaction was negative, indicating that the SES achievement gradient appeared steeper for girls than for boys. Table 1 Estimates of the SES-Gender Interaction by Country and Data Source 2006 Data Source 2009 Data Source Country Parents Student Parents Students Bulgaria 0.94 -2.01 - - Chile - - 0.50 -1.17 Colombia 0.65 -1.85 - - Croatia 2.00 -1.54 -0.86 -2.51 Denmark -3.12 -3.42 -1.47 -1.94 Germany -0.65 -0.97 2.01 -0.56 Hong Kong 0.81 -2.15 -0.39 -1.85 Hungary - - 3.70 2.78 Iceland -1.09 -0.21 - - Italy 0.26 -2.43 3.71 1.21 Korea -0.26 -1.61 1.98 1.36 Lithuania - - 0.90 -3.25 Luxembourg -1.03 -0.31 - - Macao 2.92 -0.95 -2.13 -0.81 New Zealand -0.01 -0.66 -0.50 -3.13 Panama - - -5.20 -6.29 Poland 2.05 -1.02 4.62 1.30 Portugal 0.65 0.07 -1.37 -2.87 Qatar 3.79 2.24 2.03 -2.25 Turkey 0.19 -1.23 - - Mean (SD) 0.51 (1.67) -1.13 (1.27) 0.50 (2.60) -1.33 (2.32) Note . The SES-gender interaction was calculated as the difference between the gradient for boys and the gradient for girls. Positive values support the \"vulnerable boys\" hypothesis; negative values support the opposite. A sign test confirmed that the pattern was systematic: in 14 of 16 country-waves in 2006, and 14 of 15 in 2009, the SES-gender interaction estimate was more negative when using student-reported data than parent-reported data (p = .004 and p = .001, respectively). For 12 country-waves there was a reversal in the sign of the interaction from positive (supporting the vulnerable boys hypothesis) to negative (contradicting the vulnerable boys hypothesis), see Figure 1. No reversal occurred in the opposite direction. Discussion This analysis explains the apparent contradiction between major research findings. Studies using parent-reported data (like Eriksson & Lindvall, 2023) tend to find support for the vulnerable boys hypothesis, while those using student-reported data (contributing to meta-analyses like Liu et al., 2022) tend to find the opposite. These conflicting findings can be explained by gender-specific attenuation biasing results for student-reported data. Addressing this critical issue requires coordinated action from researchers, data producers, and policymakers. Researchers should prioritize parent-reported data when studying SES-gender interactions, and when only student-reported data are available, they should conduct sensitivity analyses and explicitly discuss the likely direction and magnitude of any bias. It is also important to assess the measurement properties of variables across all relevant subgroups before modeling interactions. Data producers, such as the OECD, should expand the collection of parent-reported data where possible and provide clearer guidance on reliability issues in technical documentation, as well as develop and share statistical correction methods. Policymakers should demand transparency about data sources, consider methodological factors when interpreting research, and recognize that headline findings about achievement gaps can be highly sensitive to measurement choices. The integrity of evidence-based policy depends on valid evidence. When foundational metrics are based on biased data, the consequences extend beyond academic debate to resource misallocation. If research suggesting girls are more vulnerable to socioeconomic disadvantage is impacted by measurement bias, as indicated by the more reliable parent-reported data, policies designed to support them may be addressing a disparity that is smaller than estimated, or even reversed in direction. More critically, if boys are more sensitive to socioeconomic disadvantage—as suggested by the more reliable parent-reported data in most countries in our study—interventions targeting low-SES schools may need to be reconsidered so that resources are directed toward addressing the specific mechanisms that make disadvantaged boys vulnerable. Limitations Several limitations should be acknowledged. First, our analysis is limited to a single SES indicator (parental education) and to students aged 15 years. The proposed mechanism behind the reversal phenomenon we document is that boys' proxy reports of parental education are noisier than girls’. Thus, we expect similar reversals to occur with other SES indicators to the extent that boys also report those indicators less accurately than girls, including composite measures that incorporate parental education as one of several components. Note that parent reports could also contain inaccuracies. However, it is unlikely that they contribute to the observed reversal of SES-gender interaction as parents are expected to report their education with equal accuracy regardless of the gender of their child. Second, families completing the parent questionnaire may not be fully representative of the broader PISA sample. Non-response rates varied considerably across countries, from less than 10% to nearly 50%. However, prior research using these same PISA waves found that restricting analysis to cases with complete parent and student reports does not substantially bias cross-country comparisons of SES gradients (Jerrim & Micklewright, 2014). Moreover, because our design compares parent-reported and student-reported data for the same students, any selection bias should affect both estimates similarly, leaving the systematic reversal pattern—our key finding—intact. Finally, while our analysis focuses on mathematics achievement, the proposed mechanism—differential measurement error in boys' versus girls' proxy reports—is unrelated to the specific achievement domain assessed. Therefore, we expect the reversal pattern to extend to other subjects where SES gradients exist. Future research should investigate gender differences in reporting accuracy across different SES indicators and age groups to establish the scope of this measurement artifact. Conclusion The systematic reversal documented here serves as a cautionary tale: in large-scale assessments, the 'who' of data reporting can be just as critical as the 'what'. We demonstrate that the SES-gender interaction is not merely a reflection of sociological reality, but is inextricably linked to the reliability of the proxy reporter. While the magnitude of this effect may vary across academic subjects or specific national contexts, any study relying solely on student-reported SES must interpret gender-specific interactions with caution. Declarations Ethics approval and consent to participate Not applicable. This study involves the analysis of existing, publicly available, and anonymized data. Competing interests The author declares that they have no competing interests. Funding This study received no specific funding. Use of AI A large language model (Anthropic’s Claude Sonnet 4.0) was used to edit the text for readability and style. Data Availability Statement The PISA 2006 and 2009 data are publicly available from the OECD (www.oecd.org/pisa/data/). The derived data supporting the findings of this study are presented in Table 1 and Supplementary Table 1. References Autor, D., Figlio, D., Karbownik, K., Roth, J., & Wasserman, M. (2019). Family disadvantage and the gender gap in behavioral and educational outcomes. American Economic Journal: Applied Economics , 11 (3), 338–381. von Davier, M., Fishbein, B., & Kennedy, A. M. (Eds.). (2024). TIMSS 2023 Technical Report: Methods and procedures. TIMSS & PIRLS Int’l Study Center, Lynch School of Education, Boston College & International Association for the Evaluation of Educational Achievement. https://timss2023.org/methods/ [timss2023.org], [iea.nl]. Duckworth, A. L., & Seligman, M. E. P. (2006). Self-discipline gives girls the edge: Gender in self-discipline, grades, and achievement test scores. Journal of Educational Psychology , 98 (1), 198–208. Ensminger, M. E., Forrest, C. B., Riley, A. W., Kang, M., Green, B. F., Starfield, B., et al. (2000). The validity of measures of socioeconomic status of adolescents. Journal of Adolescent Research , 15 , 392–419. https://doi.org/10.1186/s12874-016-0148-9 Eriksson, K., Lindvall, J., & Lansford, J. E. (2026). Socialized into knowing: Gender differences in children's knowledge of parental backgrounds as early relational labor across 45 nations. Preprint available at Research Square. https://doi.org/10.21203/rs.3.rs-8541982/v1 Eriksson, K., Lindvall, J., Helenius, O., & Ryve, A. (2021). Socioeconomic status as a multidimensional predictor of student achievement in 77 societies. Frontiers in Education , 6 . https://doi.org/10.3389/feduc.2021.731634 Eriksson, K., & Lindvall, J. (2023). Cultural variation in the SES-gender interaction in student achievement. Frontiers in Psychology , 14 , 1120211. https://doi.org/10.3389/fpsyg.2023.1120211 Fraillon, J., Rožman, M., Meyer, S., Musu, L., Liaw, Y. L., Christiansen, A., & Tieck, S. (Eds.). (2025). ICILS 2023 Technical Report (Revised ed.). International Association for the Evaluation of Educational Achievement. https://www.iea.nl/sites/default/files/2025-08/ICILS_2023_Technical_Report%20Revised.pdf IEA (2025). Help Manual for the IEA IDB Analyzer (Version 5.0) . Hamburg, Germany. (Available from www.iea.nl). Jerrim, J., & Micklewright, J. (2014). Socio-economic gradients in children's cognitive skills: Are cross-country comparisons robust to who reports family background? European Sociological Review , 30 (6), 766–781. Korupp, S. E., Ganzeboom, H. B., & van der Lippe, T. (2002). Do mothers matter? The effect of parental education on the educational attainment of their children in the Netherlands. Journal of Family Issues , 23 (7), 896–915. Lien, N., Friestad, C., & Klepp, K. I. (2001). Adolescents' proxy reports of parents' socioeconomic status: How valid are they? Journal of Epidemiology & Community Health , 55 (10), 731–737. https://doi.org/10.1136/jech.55.10.731 Liu, J., Peng, P., Zhao, B., & Luo, L. (2022). Socioeconomic status and academic achievement in primary and secondary education: A meta-analytic review. Educational Psychology Review , 34 (4), 2867–2896. https://doi.org/10.1007/s10648-022-09689-y Looker, E. D. (1989). Accuracy of proxy reports of parental status characteristics. Sociology of Education , 62 (4), 257–276. https://doi.org/10.2307/2112830 Martin, M. O., Mullis, I. V. S., & Hooper, M. (Eds.). (2016). Methods and Procedures in TIMSS 2015 . TIMSS & PIRLS International Study Center. OECD (2009). PISA 2006 Technical Report . OECD Publishing, Paris. https://doi.org/10.1787/9789264048096-en OECD (2012). PISA 2009 Technical Report. OECD Publishing, Paris. https://doi.org/10.1787/9789264167872-en PISA 2022 Technical Report . OECD OECD, & Publishing (2024). https://doi.org/10.1787/01820d6d-en Schulz, W., Friedman, T., & Fraillon, J. (Eds.). (2024). ICCS 2022 technical report [Open Access]. International Association for the Evaluation of Educational Achievement. https://www.iea.nl/sites/default/files/2024-07/ICCS%202022%20Technical%20Report.pdf Soland, J. (2019). Are achievement gap estimates biased by differential student test effort? Teachers College Record , 120 (12). Additional Declarations No competing interests reported. Supplementary Files SupplementaryTable1.docx 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-8743057\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":true,\"archivedVersions\":[],\"articleType\":\"Research Article\",\"associatedPublications\":[],\"authors\":[{\"id\":587704724,\"identity\":\"117ccf42-389f-47a0-8ae0-edb2757cbe16\",\"order_by\":0,\"name\":\"Kimmo Eriksson\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4ElEQVRIiWNgGAWjYBACCWYog40HTNkwMBwgUUsaEVrgLIiWw4S1SLZzp274wGCTz8dz/PGHjzvOR/PdSGD+8AGPFmlm3m03ZzCkWbbx9phJzjxzO3fmjQQ2yRl4tMgBtdzmYThswMbPw8bM23Y7dwNQCzMPYS3/gVrYH3/mbTsH0sL8+Q8BhwG1HDBg420wkOZtOwDSwiCN1/vNIL8YJBuw8ZwB+qUtOXfmmYdtkj14tEicP7vtxocKOwP5nnRgiLXZ5fYdTz784Qc+a8DAAIXH2EBQwygYBaNgFIwC/AAARWFLw2FzP2QAAAAASUVORK5CYII=\",\"orcid\":\"\",\"institution\":\"Mälardalen University\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Kimmo\",\"middleName\":\"\",\"lastName\":\"Eriksson\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2026-01-30 16:08:32\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-8743057/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-8743057/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":102438634,\"identity\":\"104f3145-f670-4ba0-9cfd-d6d3caa9e96f\",\"added_by\":\"auto\",\"created_at\":\"2026-02-11 16:35:39\",\"extension\":\"png\",\"order_by\":1,\"title\":\"Figure 1\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":262908,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003eSystematic Reversal by Data Source in the SES-Gender Interaction\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eNote. \\u003c/em\\u003eEach line connects estimates for the same country-wave using parent-reported (left) versus student-reported (right) parental education data. Positive values indicate steeper SES gradients for boys; negative values indicate steeper gradients for girls. The graph shows all 12 cases that reversed from supporting \\\"vulnerable boys\\\" with parent data to falsely suggesting \\\"vulnerable girls\\\" with student data. Data for all 31 country-waves are presented in Table 1.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8743057/v1/7b2d6c58559e0af83b487506.png\"},{\"id\":102438635,\"identity\":\"cdac6000-48cf-4864-9de5-03364d4b8c47\",\"added_by\":\"auto\",\"created_at\":\"2026-02-11 16:35:44\",\"extension\":\"pdf\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"manuscript-pdf\",\"size\":734392,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"manuscript.pdf\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8743057/v1/bd295f54-50ca-469d-bb7b-670df1c57279.pdf\"},{\"id\":102438633,\"identity\":\"b295a866-2adc-4b82-a457-dfb00fece759\",\"added_by\":\"auto\",\"created_at\":\"2026-02-11 16:35:39\",\"extension\":\"docx\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":19185,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"SupplementaryTable1.docx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8743057/v1/b8ad27ac9eef5bfd5716138a.docx\"}],\"financialInterests\":\"No competing interests reported.\",\"formattedTitle\":\"Parent versus Student Reporting Drives Contradictory Gender Patterns in Socioeconomic Gradients\",\"fulltext\":[{\"header\":\"Introduction\",\"content\":\"\\u003cp\\u003eUnderstanding how socioeconomic status (SES) and gender interact to influence student achievement represents a critical question in educational equity research. The answer shapes our understanding of which students are most academically vulnerable and guides the design of interventions to address educational inequality. The theoretical \\\"vulnerable boys\\\" hypothesis suggests that boys are more academically susceptible to socioeconomic disadvantage than girls (Autor et al., \\u003cspan citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e2019\\u003c/span\\u003e). This perspective is supported by several proposed mechanisms, such as differential parental investment patterns and gendered behavioral responses to adversity. Yet despite extensive research using large-scale data, the field presents a puzzling landscape of contradictory findings.\\u003c/p\\u003e \\u003cp\\u003eFor example, a study of PIRLS and TIMSS data from 36 countries found that SES gradients in achievement \\u0026mdash;that is, differences in academic performance associated with socioeconomic status\\u0026mdash;were steeper for boys in most countries (Eriksson \\u0026amp; Lindvall, \\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e). In contrast, other large-scale assessments have produced opposing results; a comprehensive meta-analysis of international assessments concluded that SES gradients in achievement are, on average, slightly steeper for girls (Liu et al., \\u003cspan citationid=\\\"CR13\\\" class=\\\"CitationRef\\\"\\u003e2022\\u003c/span\\u003e).\\u003c/p\\u003e \\u003cp\\u003eThe use of large-scale assessment data to support two mutually exclusive hypotheses\\u0026mdash;either that boys are more vulnerable or that girls are\\u0026mdash;raises important questions about the validity of these conclusions. Therefore, identifying the underlying reasons for these divergent findings is essential. We propose that a key factor underlying these conflicting results is whether studies use SES data reported by parents or by the students themselves.\\u003c/p\\u003e \\u003cp\\u003eLarge-scale assessments of adolescents\\u0026mdash;such as PISA (Programme for International Student Assessment), TIMSS (Trends in International Mathematics and Science Study), ICCS (International Civic and Citizenship Education Study), and ICILS (International Computer and Information Literacy Study)\\u0026mdash;often rely on students to provide information about their parents' background (OECD, \\u003cspan citationid=\\\"CR18\\\" class=\\\"CitationRef\\\"\\u003e2024\\u003c/span\\u003e; von Davier et al., \\u003cspan citationid=\\\"CR2\\\" class=\\\"CitationRef\\\"\\u003e2024\\u003c/span\\u003e; Schultz et al., 2024; Fraillon et al., \\u003cspan citationid=\\\"CR8\\\" class=\\\"CitationRef\\\"\\u003e2025\\u003c/span\\u003e). This practice, known as using student proxy reports, can introduce bias. Specifically, we argue that gender differences in the reliability of student-reported data make the choice of data source a critical factor in shaping study outcomes.\\u003c/p\\u003e\\n\\u003ch3\\u003eFrom Attenuation to Sign Reversal\\u003c/h3\\u003e\\n\\u003cp\\u003eWhen a predictor variable contains random measurement error, standard regression analysis produces attenuation bias, meaning that the estimated effect is biased toward zero. The extent of this bias depends on the reliability of the measure, which refers to how much of the observed variation reflects true differences rather than random noise. Put simply, less reliable (noisier) data make statistical relationships appear weaker than they actually are.\\u003c/p\\u003e \\u003cp\\u003eThis general statistical phenomenon has specific implications for research on SES\\u0026ndash;gender interactions. While data on student gender is highly reliable regardless of who reports it, it is well-established that student participants in large-scale assessments provide significantly less reliable information about their parents\\u0026rsquo; socioeconomic status than parents themselves do (Jerrim \\u0026amp; Micklewright, \\u003cspan citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e2014\\u003c/span\\u003e). In particular, students\\u0026rsquo; reports of their parents\\u0026rsquo; educational attainment, which is a primary component of SES indices, are especially prone to error (Ensminger et al., \\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e2000\\u003c/span\\u003e; Jerrim \\u0026amp; Micklewright, \\u003cspan citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e2014\\u003c/span\\u003e; Lien et al., \\u003cspan citationid=\\\"CR12\\\" class=\\\"CitationRef\\\"\\u003e2001\\u003c/span\\u003e; Looker, \\u003cspan citationid=\\\"CR14\\\" class=\\\"CitationRef\\\"\\u003e1989\\u003c/span\\u003e).\\u003c/p\\u003e \\u003cp\\u003eThe critical insight for the present study is that \\u003cem\\u003ethis measurement error is not uniform across genders.\\u003c/em\\u003e Research demonstrates that adolescent boys consistently report their parents' background characteristics less accurately than girls do, in line with broader evidence of gender differences in school-related behaviors, including lower test-taking effort among boys (Soland, \\u003cspan citationid=\\\"CR20\\\" class=\\\"CitationRef\\\"\\u003e2019\\u003c/span\\u003e) and higher levels of conscientiousness and attention among girls (Duckworth \\u0026amp; Seligman, \\u003cspan citationid=\\\"CR3\\\" class=\\\"CitationRef\\\"\\u003e2006\\u003c/span\\u003e). Prior analyses of the accuracy of students' reports of their parents' education in PISA data found that errors were on average several percentage points more common among boys than girls (Eriksson et al., \\u003cspan citationid=\\\"CR5\\\" class=\\\"CitationRef\\\"\\u003e2026\\u003c/span\\u003e; see also Supplementary Table\\u0026nbsp;1). Importantly, this gender difference in accuracy was consistent across all levels of parental education, indicating random measurement error rather than systematic directional bias (e.g., boys systematically over- or under-reporting education levels).\\u003c/p\\u003e \\u003cp\\u003eFor the SES-gender interaction, gender-specific attenuation can actually reverse conclusions about which gender is most vulnerable. In parent-reported data, the actual SES gradient is steeper for boys than for girls (on average, \\u0026#120573;\\u003csub\\u003eboys\\u003c/sub\\u003e = 16.4 vs. \\u0026#120573;\\u003csub\\u003egirls\\u003c/sub\\u003e = 15.9, where \\u0026#120573; represents the increase in achievement per unit increase in parental education). However, when student-reported SES is used instead, the SES gradient is more severely attenuated for boys than for girls (by 27% vs. 17%). As a result, the attenuated gradient for boys may fall below that for girls (\\u0026#120573;\\u003csub\\u003eboys\\u003c/sub\\u003e\\u0026thinsp;=\\u0026thinsp;12.0 vs. \\u0026#120573;\\u003csub\\u003egirls\\u003c/sub\\u003e\\u0026thinsp;=\\u0026thinsp;13.2). Thus, differential attenuation can create the misleading appearance that the SES gradient is steeper for girls, even when the true relationship is the opposite.\\u003c/p\\u003e \\u003cdiv id=\\\"Sec3\\\" class=\\\"Section2\\\"\\u003e \\u003ch2\\u003eThe Present Study\\u003c/h2\\u003e \\u003cp\\u003eThis study tests whether the source of SES data\\u0026mdash;parent-reported versus student-reported\\u0026mdash;systematically affects estimates of the SES-gender interaction in academic achievement. We leverage large-scale assessment data with the unique feature that reports of parental education are available both from students and their parents. This within-subject design allows us to isolate the effect of the reporting source while holding the sample and underlying construct constant.\\u003c/p\\u003e \\u003cp\\u003eIf boys report parental education less accurately than girls, as prior research suggests, then student-reported data should produce attenuated SES gradients for boys relative to girls. We therefore predict that (1) the SES-gender interaction will be more negative (i.e., less supportive of the \\\"vulnerable boys\\\" hypothesis) when estimated from student-reported data than from parent-reported data, and (2) this bias will be systematic across countries.\\u003c/p\\u003e \\u003c/div\\u003e\"},{\"header\":\"Methods\",\"content\":\"\\u003cp\\u003eWe utilize data from PISA 2006 and 2009. These specific waves offer a unique methodological opportunity: they are the only iterations of PISA that collected parental education data directly from parents (via the optional parent questionnaire) in addition to the standard student reports for the same households (OECD, \\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e2009\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR17\\\" class=\\\"CitationRef\\\"\\u003e2012\\u003c/span\\u003e). This design creates a powerful natural experiment, allowing for a direct within-subject comparison that isolates the effect of the reporting source while holding the sample and the underlying construct constant.\\u003c/p\\u003e\\n\\u003ch3\\u003eParticipants\\u003c/h3\\u003e\\n\\u003cp\\u003eWe restricted the analysis to students with complete parent- and student-reported data on parental education for both parents. The proportion of excluded cases varied considerably across countries, from just a few percent to almost half the sample, with a mean of 25.2% (SD\\u0026thinsp;=\\u0026thinsp;14.7%; see Supplementary Table\\u0026nbsp;1 for details by country). However, exclusions based on missing data are not a serious concern for the purpose of demonstrating how estimates of the SES-gender interaction depends on whether parent reports or student reports are used, as parent non-response is unlikely to vary systematically by student gender. Additionally, prior research using these same PISA waves found that restricting analysis to cases with complete parent and student reports does not substantially bias cross-country comparisons of SES gradients (Jerrim \\u0026amp; Micklewright, \\u003cspan citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e2014\\u003c/span\\u003e).\\u003c/p\\u003e \\u003cp\\u003eThe final analysis includes 151,269 students, from 20 countries participating in PISA 2006 and/or 2009, with complete parent- and student-reported data on parental education.\\u003c/p\\u003e\\n\\u003ch3\\u003eMeasures\\u003c/h3\\u003e\\n\\u003cp\\u003e \\u003cb\\u003eParental Education.\\u003c/b\\u003e This analysis focuses on parental education because it is the sole socioeconomic indicator available from both data sources (parent and student). The PISA questionnaires ask for the level of education of each parent using national adaptations of the International Standard Classification of Education scale. The parent questionnaire primarily asks about higher qualifications, whereas the student questionnaire also distinguishes between lower levels of schooling (OECD, \\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e2009\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR17\\\" class=\\\"CitationRef\\\"\\u003e2012\\u003c/span\\u003e). We recoded the data from both questionnaires to an identical four-point scale: bachelor's degree or higher (coded 3), shorter tertiary education (2), upper secondary education or post-secondary non-tertiary education (1), and any lower level of schooling (0). We then summed the values for the mother's and father's education to create a parental education sum score ranging from 0 to 6. While taking the maximum of the two parents' educational levels is common, prior research indicates the sum score is a more efficient predictor (Korupp et al., \\u003cspan citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e2002\\u003c/span\\u003e).\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eMathematics Achievement.\\u003c/b\\u003e We focus on mathematics achievement in this study. PISA also assesses achievement in science, and reading, but all subjects exhibit very similar gradients with respect to parents' education (Eriksson et al., \\u003cspan citationid=\\\"CR6\\\" class=\\\"CitationRef\\\"\\u003e2021\\u003c/span\\u003e). We utilized the plausible values provided by PISA, which are calibrated to a global mean of approximately 500 points and a standard deviation of 100 points (OECD, \\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e2009\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR17\\\" class=\\\"CitationRef\\\"\\u003e2012\\u003c/span\\u003e).\\u003c/p\\u003e \\u003cdiv id=\\\"Sec7\\\" class=\\\"Section2\\\"\\u003e \\u003ch2\\u003eStatistical Analysis\\u003c/h2\\u003e \\u003cp\\u003eTo examine whether students\\u0026rsquo; proxy reports of parental education contain more random measurement error for boys than girls, we compute correlations between student-reported and parent-reported parental education separately by gender. We estimate the parental-education gradient on mathematics achievement by regressing student math scores (plausible values) on a parental education score. These analyses were performed separately for each country-wave, gender, and SES data source (parents or students) using SPSS 30.0 with syntax created by the IDB Analyzer 5.0 which ensures the appropriate application of weights and robust standard errors (IEA, \\u003cspan citationid=\\\"CR9\\\" class=\\\"CitationRef\\\"\\u003e2025\\u003c/span\\u003e).\\u003c/p\\u003e \\u003cp\\u003eThe SES-gender interaction is calculated as the difference between the gradient for boys and the gradient for girls. To test whether the direction of bias was systematic, we examine how many country-waves show more negative interaction estimates with student-reported versus parent-reported data using the sign test.\\u003c/p\\u003e \\u003c/div\\u003e\"},{\"header\":\"Results\",\"content\":\"\\u003cp\\u003eWe begin by examining whether boys\\u0026apos; proxy reports of parental education are less reliable than girls\\u0026apos;. Confirming that boys\\u0026apos; reports contain more random measurement error, the correlation between student-reported and parent-reported parental education was higher for girls than for boys in 29 of 31 country-waves (see Supplementary Table 1).\\u003c/p\\u003e\\n\\u003cp\\u003eEstimates of the SES-gender interaction change depending on the data source as illustrated in Table 1. The table reveals a striking and systematic reversal of findings. When using reliable parent reports of parental education, the results generally support the vulnerable boys hypothesis: a positive SES-gender interaction (meaning the SES achievement gradient is steeper for boys than for girls) was observed in the majority of country-waves (18 of 31, or 58.1%). However, when the same analysis was conducted using student reports of parental education\\u0026mdash;the data type most commonly available in PISA\\u0026mdash;the direction of the interaction reversed. In most country-waves (25 of 31, or 80.6%), the SES-gender interaction was negative, indicating that the SES achievement gradient appeared steeper for girls than for boys.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eTable 1\\u0026nbsp;\\u003c/strong\\u003e\\u003cem\\u003eEstimates of the SES-Gender Interaction by Country and Data Source\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003ctable border=\\\"1\\\" cellspacing=\\\"0\\\" cellpadding=\\\"0\\\" width=\\\"570\\\"\\u003e\\n \\u003ctbody\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003e\\u0026nbsp;\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd colspan=\\\"2\\\" valign=\\\"top\\\" style=\\\"width: 238px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e2006 Data Source\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd colspan=\\\"2\\\" valign=\\\"top\\\" style=\\\"width: 238px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e2009 Data Source\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eCountry\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eParents\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eStudent\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eParents\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eStudents\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eBulgaria\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e0.94\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-2.01\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eChile\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e0.50\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-1.17\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eColombia\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e0.65\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-1.85\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eCroatia\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e2.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-1.54\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-0.86\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-2.51\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eDenmark\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-3.12\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-3.42\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-1.47\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-1.94\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eGermany\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-0.65\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-0.97\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e2.01\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-0.56\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eHong Kong\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e0.81\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-2.15\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-0.39\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-1.85\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eHungary\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e3.70\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e2.78\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eIceland\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-1.09\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-0.21\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eItaly\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e0.26\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-2.43\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e3.71\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e1.21\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eKorea\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-0.26\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-1.61\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e1.98\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e1.36\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eLithuania\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e0.90\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-3.25\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eLuxembourg\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-1.03\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-0.31\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eMacao\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e2.92\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-0.95\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-2.13\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-0.81\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eNew Zealand\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-0.01\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-0.66\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-0.50\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-3.13\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003ePanama\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-5.20\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-6.29\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003ePoland\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e2.05\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-1.02\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e4.62\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e1.30\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003ePortugal\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e0.65\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e0.07\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-1.37\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-2.87\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eQatar\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e3.79\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e2.24\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e2.03\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-2.25\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003eTurkey\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e0.19\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-1.23\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e-\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 94px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eMean (SD)\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e0.51 (1.67)\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e-1.13 (1.27)\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 115px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e0.50 (2.60)\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 122px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e-1.33 (2.32)\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003c/tbody\\u003e\\n\\u003c/table\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eNote\\u003c/em\\u003e. The SES-gender interaction was calculated as the difference between the gradient for boys and the gradient for girls. Positive values support the \\u0026quot;vulnerable boys\\u0026quot; hypothesis; negative values support the opposite.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eA sign test confirmed that the pattern was systematic: in 14 of 16 country-waves in 2006, and 14 of 15 in 2009, the SES-gender interaction estimate was more negative when using student-reported data than parent-reported data (p = .004 and p = .001, respectively). For 12 country-waves there was a reversal in the sign of the interaction from positive (supporting the vulnerable boys hypothesis) to negative (contradicting the vulnerable boys hypothesis), see Figure 1. No reversal occurred in the opposite direction.\\u0026nbsp;\\u003c/p\\u003e\"},{\"header\":\"Discussion\",\"content\":\"\\u003cp\\u003eThis analysis explains the apparent contradiction between major research findings. Studies using parent-reported data (like Eriksson \\u0026amp; Lindvall, 2023) tend to find support for the vulnerable boys hypothesis, while those using student-reported data (contributing to meta-analyses like Liu et al., 2022) tend to find the opposite. These conflicting findings can be explained by gender-specific attenuation biasing results for student-reported data.\\u003c/p\\u003e\\n\\u003cp\\u003eAddressing this critical issue requires coordinated action from researchers, data producers, and policymakers. Researchers should prioritize parent-reported data when studying SES-gender interactions, and when only student-reported data are available, they should conduct sensitivity analyses and explicitly discuss the likely direction and magnitude of any bias. It is also important to assess the measurement properties of variables across all relevant subgroups before modeling interactions. Data producers, such as the OECD, should expand the collection of parent-reported data where possible and provide clearer guidance on reliability issues in technical documentation, as well as develop and share statistical correction methods. Policymakers should demand transparency about data sources, consider methodological factors when interpreting research, and recognize that headline findings about achievement gaps can be highly sensitive to measurement choices.\\u003c/p\\u003e\\n\\u003cp\\u003eThe integrity of evidence-based policy depends on valid evidence. When foundational metrics are based on biased data, the consequences extend beyond academic debate to resource misallocation. If research suggesting girls are more vulnerable to socioeconomic disadvantage is impacted by measurement bias, as indicated by the more reliable parent-reported data, policies designed to support them may be addressing a disparity that is smaller than estimated, or even reversed in direction. More critically, if boys are more sensitive to socioeconomic disadvantage—as suggested by the more reliable parent-reported data in most countries in our study—interventions targeting low-SES schools may need to be reconsidered so that resources are directed toward addressing the specific mechanisms that make disadvantaged boys vulnerable.\\u003c/p\\u003e\\n\\u003ch2\\u003eLimitations\\u003c/h2\\u003e\\n\\u003cp\\u003eSeveral limitations should be acknowledged. First, our analysis is limited to a single SES indicator (parental education) and to students aged 15 years. The proposed mechanism behind the reversal phenomenon we document is that boys' proxy reports of parental education are noisier than girls’. Thus, we expect similar reversals to occur with other SES indicators to the extent that boys also report those indicators less accurately than girls, including composite measures that incorporate parental education as one of several components. Note that parent reports could also contain inaccuracies. However, it is unlikely that they contribute to the observed reversal of SES-gender interaction as parents are expected to report their education with equal accuracy regardless of the gender of their child.\\u003c/p\\u003e\\n\\u003cp\\u003eSecond, families completing the parent questionnaire may not be fully representative of the broader PISA sample. Non-response rates varied considerably across countries, from less than 10% to nearly 50%. However, prior research using these same PISA waves found that restricting analysis to cases with complete parent and student reports does not substantially bias cross-country comparisons of SES gradients (Jerrim \\u0026amp; Micklewright, 2014). Moreover, because our design compares parent-reported and student-reported data for the same students, any selection bias should affect both estimates similarly, leaving the systematic reversal pattern—our key finding—intact.\\u003c/p\\u003e\\n\\u003cp\\u003eFinally, while our analysis focuses on mathematics achievement, the proposed mechanism—differential measurement error in boys' versus girls' proxy reports—is unrelated to the specific achievement domain assessed. Therefore, we expect the reversal pattern to extend to other subjects where SES gradients exist. Future research should investigate gender differences in reporting accuracy across different SES indicators and age groups to establish the scope of this measurement artifact.\\u003c/p\\u003e\\n\\n\"},{\"header\":\"Conclusion\",\"content\":\"\\u003cp\\u003eThe systematic reversal documented here serves as a cautionary tale: in large-scale assessments, the 'who' of data reporting can be just as critical as the 'what'. We demonstrate that the SES-gender interaction is not merely a reflection of sociological reality, but is inextricably linked to the reliability of the proxy reporter. While the magnitude of this effect may vary across academic subjects or specific national contexts, any study relying solely on student-reported SES must interpret gender-specific interactions with caution.\\u003c/p\\u003e\"},{\"header\":\"Declarations\",\"content\":\"\\u003cp\\u003e\\u003cstrong\\u003eEthics approval and consent to participate\\u003c/strong\\u003e\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eNot applicable. This study involves the analysis of existing, publicly available, and anonymized data.\\u003cstrong\\u003e\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eCompeting interests\\u003c/strong\\u003e\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eThe author declares that they have no competing interests.\\u003cstrong\\u003e\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eFunding\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThis study received no specific funding.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eUse of AI\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eA large language model (Anthropic\\u0026rsquo;s Claude Sonnet 4.0) was used to edit the text for readability and style.\\u003cstrong\\u003e\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eData Availability Statement\\u003c/strong\\u003e\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eThe PISA 2006 and 2009 data are publicly available from the OECD (www.oecd.org/pisa/data/). The derived data supporting the findings of this study are presented in Table 1 and Supplementary Table 1.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\u003cli\\u003e\\u003cspan\\u003eAutor, D., Figlio, D., Karbownik, K., Roth, J., \\u0026amp; Wasserman, M. (2019). Family disadvantage and the gender gap in behavioral and educational outcomes. \\u003cem\\u003eAmerican Economic Journal: Applied Economics\\u003c/em\\u003e, \\u003cem\\u003e11\\u003c/em\\u003e(3), 338\\u0026ndash;381.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003evon Davier, M., Fishbein, B., \\u0026amp; Kennedy, A. M. (Eds.). (2024). \\u003cem\\u003eTIMSS 2023 Technical Report: Methods and procedures.\\u003c/em\\u003e TIMSS \\u0026amp; PIRLS Int\\u0026rsquo;l Study Center, Lynch School of Education, Boston College \\u0026amp; International Association for the Evaluation of Educational Achievement. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://timss2023.org/methods/\\u003c/span\\u003e\\u003cspan address=\\\"https://timss2023.org/methods/\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e [timss2023.org], [iea.nl].\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eDuckworth, A. L., \\u0026amp; Seligman, M. E. P. (2006). Self-discipline gives girls the edge: Gender in self-discipline, grades, and achievement test scores. \\u003cem\\u003eJournal of Educational Psychology\\u003c/em\\u003e, \\u003cem\\u003e98\\u003c/em\\u003e(1), 198\\u0026ndash;208.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eEnsminger, M. E., Forrest, C. B., Riley, A. W., Kang, M., Green, B. F., Starfield, B., et al. (2000). The validity of measures of socioeconomic status of adolescents. \\u003cem\\u003eJournal of Adolescent Research\\u003c/em\\u003e, \\u003cem\\u003e15\\u003c/em\\u003e, 392\\u0026ndash;419. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.1186/s12874-016-0148-9\\u003c/span\\u003e\\u003cspan address=\\\"10.1186/s12874-016-0148-9\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eEriksson, K., Lindvall, J., \\u0026amp; Lansford, J. E. (2026). Socialized into knowing: Gender differences in children's knowledge of parental backgrounds as early relational labor across 45 nations. Preprint available at Research Square. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.21203/rs.3.rs-8541982/v1\\u003c/span\\u003e\\u003cspan address=\\\"10.21203/rs.3.rs-8541982/v1\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eEriksson, K., Lindvall, J., Helenius, O., \\u0026amp; Ryve, A. (2021). Socioeconomic status as a multidimensional predictor of student achievement in 77 societies. \\u003cem\\u003eFrontiers in Education\\u003c/em\\u003e, \\u003cem\\u003e6\\u003c/em\\u003e. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.3389/feduc.2021.731634\\u003c/span\\u003e\\u003cspan address=\\\"10.3389/feduc.2021.731634\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eEriksson, K., \\u0026amp; Lindvall, J. (2023). Cultural variation in the SES-gender interaction in student achievement. \\u003cem\\u003eFrontiers in Psychology\\u003c/em\\u003e, \\u003cem\\u003e14\\u003c/em\\u003e, 1120211. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.3389/fpsyg.2023.1120211\\u003c/span\\u003e\\u003cspan address=\\\"10.3389/fpsyg.2023.1120211\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eFraillon, J., Rožman, M., Meyer, S., Musu, L., Liaw, Y. L., Christiansen, A., \\u0026amp; Tieck, S. (Eds.). (2025). \\u003cem\\u003eICILS 2023 Technical Report\\u003c/em\\u003e (Revised ed.). International Association for the Evaluation of Educational Achievement. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://www.iea.nl/sites/default/files/2025-08/ICILS_2023_Technical_Report%20Revised.pdf\\u003c/span\\u003e\\u003cspan address=\\\"https://www.iea.nl/sites/default/files/2025-08/ICILS_2023_Technical_Report%20Revised.pdf\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eIEA (2025). \\u003cem\\u003eHelp Manual for the IEA IDB Analyzer (Version 5.0)\\u003c/em\\u003e. Hamburg, Germany. (Available from www.iea.nl).\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eJerrim, J., \\u0026amp; Micklewright, J. (2014). Socio-economic gradients in children's cognitive skills: Are cross-country comparisons robust to who reports family background? \\u003cem\\u003eEuropean Sociological Review\\u003c/em\\u003e, \\u003cem\\u003e30\\u003c/em\\u003e(6), 766\\u0026ndash;781.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eKorupp, S. E., Ganzeboom, H. B., \\u0026amp; van der Lippe, T. (2002). Do mothers matter? The effect of parental education on the educational attainment of their children in the Netherlands. \\u003cem\\u003eJournal of Family Issues\\u003c/em\\u003e, \\u003cem\\u003e23\\u003c/em\\u003e(7), 896\\u0026ndash;915.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLien, N., Friestad, C., \\u0026amp; Klepp, K. I. (2001). Adolescents' proxy reports of parents' socioeconomic status: How valid are they? \\u003cem\\u003eJournal of Epidemiology \\u0026amp; Community Health\\u003c/em\\u003e, \\u003cem\\u003e55\\u003c/em\\u003e(10), 731\\u0026ndash;737. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.1136/jech.55.10.731\\u003c/span\\u003e\\u003cspan address=\\\"10.1136/jech.55.10.731\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLiu, J., Peng, P., Zhao, B., \\u0026amp; Luo, L. (2022). Socioeconomic status and academic achievement in primary and secondary education: A meta-analytic review. \\u003cem\\u003eEducational Psychology Review\\u003c/em\\u003e, \\u003cem\\u003e34\\u003c/em\\u003e(4), 2867\\u0026ndash;2896. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.1007/s10648-022-09689-y\\u003c/span\\u003e\\u003cspan address=\\\"10.1007/s10648-022-09689-y\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLooker, E. D. (1989). Accuracy of proxy reports of parental status characteristics. \\u003cem\\u003eSociology of Education\\u003c/em\\u003e, \\u003cem\\u003e62\\u003c/em\\u003e(4), 257\\u0026ndash;276. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.2307/2112830\\u003c/span\\u003e\\u003cspan address=\\\"10.2307/2112830\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eMartin, M. O., Mullis, I. V. S., \\u0026amp; Hooper, M. (Eds.). (2016). \\u003cem\\u003eMethods and Procedures in TIMSS 2015\\u003c/em\\u003e. TIMSS \\u0026amp; PIRLS International Study Center.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eOECD (2009). \\u003cem\\u003ePISA 2006 Technical Report\\u003c/em\\u003e. OECD Publishing, Paris. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.1787/9789264048096-en\\u003c/span\\u003e\\u003cspan address=\\\"10.1787/9789264048096-en\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eOECD (2012). \\u003cem\\u003ePISA 2009 Technical Report.\\u003c/em\\u003e OECD Publishing, Paris. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.1787/9789264167872-en\\u003c/span\\u003e\\u003cspan address=\\\"10.1787/9789264167872-en\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003e\\u003cem\\u003ePISA 2022 Technical Report\\u003c/em\\u003e. OECD OECD, \\u0026amp; Publishing (2024). \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://doi.org/10.1787/01820d6d-en\\u003c/span\\u003e\\u003cspan address=\\\"10.1787/01820d6d-en\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eSchulz, W., Friedman, T., \\u0026amp; Fraillon, J. (Eds.). (2024). \\u003cem\\u003eICCS 2022 technical report\\u003c/em\\u003e [Open Access]. International Association for the Evaluation of Educational Achievement. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://www.iea.nl/sites/default/files/2024-07/ICCS%202022%20Technical%20Report.pdf\\u003c/span\\u003e\\u003cspan address=\\\"https://www.iea.nl/sites/default/files/2024-07/ICCS%202022%20Technical%20Report.pdf\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eSoland, J. (2019). Are achievement gap estimates biased by differential student test effort? \\u003cem\\u003eTeachers College Record\\u003c/em\\u003e, \\u003cem\\u003e120\\u003c/em\\u003e(12).\\u003c/span\\u003e\\u003c/li\\u003e\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":false,\"hideJournal\":true,\"highlight\":\"\",\"institution\":\"\",\"isAcceptedByJournal\":true,\"isAuthorSuppliedPdf\":false,\"isDeskRejected\":\"\",\"isHiddenFromSearch\":false,\"isInQc\":false,\"isInWorkflow\":false,\"isPdf\":false,\"isPdfUpToDate\":true,\"isWithdrawnOrRetracted\":false,\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"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\":\"socioeconomic status, gender differences, measurement error, large-scale assessments, PISA\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-8743057/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-8743057/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"\\u003cp\\u003eResearch on how socioeconomic status (SES) and gender interact to affect student achievement has produced contradictory findings. While some studies suggest boys are more vulnerable to socioeconomic disadvantage, others find the opposite. This paper argues that these conflicting results stem from a critical methodological artifact: because of greater measurement error in boys' proxy reports, the source of SES data in large-scale assessments is critical. PISA 2006 and 2009 provides data on 151,269 students from 20 countries for whom both parent-reported and student-reported SES data were available. This within-subject design allowed for a direct comparison of the SES-gender interaction in mathematics achievement, isolating the effect of the data source. The direction of the SES-gender interaction systematically reverses depending on the data source. Aligning with the \\\"vulnerable boys\\\" hypothesis, the SES achievement gradient is typically steeper for boys when using reliable parent-reported data; this holds in 18 out of 31 country-waves. In 12 of those 18 cases, the results reverse to instead indicate \\u0026ldquo;vulnerable girls\\u0026rdquo; when using the more commonly available student-reported proxy data. The choice of SES data source is not a minor technical detail but a critical factor that can lead to flawed conclusions about educational equity. Greater measurement error in boys' proxy reports artificially attenuates their SES gradient. Researchers should prioritize parent-reported data for this type of analysis, and findings based on student-reported SES should be interpreted with caution.\\u003c/p\\u003e\",\"manuscriptTitle\":\"Parent versus Student Reporting Drives Contradictory Gender Patterns in Socioeconomic Gradients\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2026-02-11 16:35:32\",\"doi\":\"10.21203/rs.3.rs-8743057/v1\",\"editorialEvents\":[{\"type\":\"communityComments\",\"content\":0}],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"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\":\"6fdd2a6b-fb43-47a0-9aae-74d7eeaecb8a\",\"owner\":[],\"postedDate\":\"February 11th, 2026\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"posted\",\"subjectAreas\":[],\"tags\":[],\"updatedAt\":\"2026-04-20T12:11:13+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2026-02-11 16:35:32\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-8743057\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-8743057\",\"identity\":\"rs-8743057\",\"version\":[\"v1\"]},\"buildId\":\"XKTyCvWXoU3ODBz1xrDgd\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}