Multiverse Interpersonal Physiological Synchrony: quantifying the impact of parameters selection on Windowed Cross-Correlation results

Interpersonal Physiological Synchrony (IPS), the alignment of physiological signals during social interactions requires advanced statistical methods to be quantified. Such methods rely on the application of researcher-defined parameters, which can significantly affect outcomes. Multiverse analysis, applying multiple analytical strategies to the same dataset, reveals how parameter choices influence results, thus mitigating possible bias and enhancing transparency and reliability of results. In this study, we used a multiverse approach to reanalyse - through Windowed Cross-Correlation (WCC) - a dyadic ECG dataset from participants involved in a Joint Action task as well as simulated data generated via Kuramoto models (mathematical models used in describing synchronyzation) . We explored how the combination of i) window size, ii) window overlap, iii) prewhitening and iv) use of Peak-Picking (PP) algorithm affected several indices of IPS strength and stability, as well as their ability to discriminate between real and surrogate dyads. The original results of the stufy were overall replicated, with all IPS indices successfully discriminating genuine from surrogate synchrony regardless of parameter selection. By contrast, the selection of parameters had a greater influence when distinguishing between task and baseline IPS. Moreover, the PP algorithm tended to overestimate synchrony strength, thereby hindering discrimination. Importantly, measures of synchrony stability proved effective in distinguishing genuine from surrogate synchrony. Based on our multiverse results, we provide recommendations for researchers applying WCC analysis to study cardiac IPS synchrony.