The present study is a replication of Pitiruț and colleagues (2022) [Attitudes towards global warming: The role of anticipated guilt and the Dark Triad traits. Personality and Individual Differences, 185, 111285.], who found that anticipated guilt positively predicted attitudes toward global warming while Dark Triad traits predicted them negatively in Romanian undergraduates. Using a more geographically diverse community sample (N = 202), we partially replicated the original findings. Anticipated guilt remained a significant positive predictor of global warming attitudes. However neither the Dark Triad total score nor any of its subscales predicted attitudes towards global warming. Discrepancies are discussed in terms of ceiling effects, sample characteristics, and measurement differences between instruments measuring traits related to the Dark Triad.
This study is a partial replication of Experiment 1 of Rensink, O’Regan, and Clark (1997), which introduced the flicker paradigm to investigate change blindness. Change blindness refers to the failure to detect large visual changes when a brief perceptual disruption eliminates the local motion signal that normally captures attention. Although widely cited as a foundational demonstration of attentional limits in visual perception, the original findings have not been subject to systematic independent replication. The present study addressed this gap using a larger sample (N = 69) and an online administration format via jsPsych. Participants were allocated to a flicker condition or a no-blank condition, with recruiters blind to condition assignment. Two hypotheses were tested: that the flicker manipulation would increase change detection reaction times, and that marginal interest changes would be detected more slowly than central interest changes, with this difference amplified under flicker conditions. Both hypotheses were supported. The flicker condition produced substantially longer detection times (Mann-Whitney U = 150.00, Z = -5.318, p < .001, Cohen’s d = 1.36). Marginal interest changes were harder to detect than central interest changes across all participants (Z = -6.62, p < .001), with the location effect substantially amplified under flicker conditions. Overall detection accuracy did not differ between conditions, ruling out a speed-accuracy tradeoff. These findings replicate the core results of Rensink et al. (1997), in a larger online sample, supporting the robustness of the flicker effect beyond controlled laboratory settings..
Health concentration curves are essential tools for quantifying inequalities in health outcomes across socioeconomic or biological gradients. In this study, we propose a novel copula-based piecewise health concentration curve constructed with the gluing copula technique and use it to reveal important insights into how stress-related biomarkers relate to inequalities in concentration performance. We employ stress-related concentration performance data for modeling localized changes in the dependence structure between concentration performance and stress-related variables, including cortisol, alpha-amylase responses, state anxiety, and cognitive stress appraisal. Through the application of this technique to experimental data from stress and unstress groups, we demonstrate how health concentration curves diverge from the line of equality and highlight stress-related biomarker ranges where concentration differences in performance are most noticeable. Our results show significant variations in concentration inequalities across stress biomarkers, emphasizing the significance of taking nonlinear dependence into account when evaluating how stress affects cognitive function. This copula-based approach offers an effective instrument for measuring and comparing health disparities in situations where the ability to focus under pressure is crucial, providing information for focused interventions aimed at minimizing performance gaps.
Meta-analysis is a powerful statistical tool that combines results from multiple studies to produce a pooled estimate, providing a clearer, more accurate picture of the research question. In behavioral research, meta-analysis of single proportions, such as the prevalence of specific behaviors or psychological conditions, is particularly useful for quantifying behavioral patterns across diverse populations. This article offers a practical guide to conducting meta-analyses of single proportions in R. Using the R package ‘meta’, this guide explains how to conduct a meta-analysis using the Generalized Linear Mixed Model (GLMM) approach, perform subgroup, sensitivity and meta-regression analyses, and detect publication bias. The step-by-step R scripts provided are designed to make meta-analysis accessible to behavioral researchers with varying levels of statistical expertise, particularly in low-resource settings. By synthesizing data from multiple smaller studies, meta-analysis helps bridge critical knowledge gaps and supports the development of evidence-based behavioral research and intervention strategies. To support the reproducibility of the analyses, the R script and a sample dataset are available in a GitHub repository.
Researchers sometimes make two flexible choices during a study: they look at results before data collection is finished, and they try several defensible analyses before deciding what to report. We call the joint use of these practices Forking-Peeks. Using computer simulations of simple two-group studies, we show that either practice alone can increase false-positive findings, but their combination can do so dramatically. In one illustrative setting with 16 correlated analysis options and four interim looks at the accumulating data, a nominal 5% false-positive rate rose to about 70% when the analyst repeatedly selected the best-looking analysis. Even a seemingly more consistent strategy, choosing the best-looking analysis once and then keeping it fixed, still produced severe inflation because the initial choice was data-dependent. The same selection process also made statistically significant effects look larger than they really were. The manuscript provides reproducible code and visual risk maps to help readers understand how small, plausible research decisions can compound when data collection and analysis remain flexible. We conclude with practical recommendations: plan stopping rules in advance, use valid sequential designs when interim looks are needed, treat multiple analyses as multiplicity, and report effect sizes and intervals rather than relying on a single threshold.
