Page 59 - Demo

57How attractiveness preferences influence attention3the computation of ICC. In line with recommendations from McGraw andWong (1996) we used the ICC(A, 1) to test for absolute agreement betweenrates. We report the ICC estimate and the 95% confidence interval.Furthermore, we used the R package correlation(Makowski, BenShachar, Patil, & Lüdecke, 2020) to test the relationship between pre-dateattractiveness ratings, post-date attractiveness ratings, and date outcome.The correlation package allows for computation of a wide variety ofcorrelations, such as Bayesian multilevel correlations. In our case, we usedBayesian multilevel Spearman correlations to investigate the associationbetween pre-date and post-date attractiveness ratings. To test the relationships between date outcome and pre-date and post-date attractivenessratings, respectively, we used Bayesian point-biserial correlations. Theseanalyses were based on a dataset that consisted of only complete cases forall three variables of interest. In total, this concerned 482 datapoints of 58participants.For our main analyses, we used Bayesian mixed models. Bayesian analyses have gained in popularity over the past few years because they offera number of benefits compared to frequentist analyses (Kruschke et al.,2012; Makowski et al., 2019). While frequentist methods (e.g., p-value nullhypothesis testing Wagenmakers, 2007) inform us about the credibility of thedata given a hypothesis, Bayesian methods inform us about the credibilityof our parameter values given the data that we observed. This is reflected inthe different interpretation of frequentist and Bayesian confidence intervals:The first is a range of values that contains the estimate in the long run, whilethe latter tells which parameter values are most credible based on the data(Kruschke et al., 2012; McElreath, 2018). Furthermore, Bayesian methodsallow for the inclusion of prior expectations in the model, are less prone toType I errors, and are more robust in small and noisy samples (Makowski etal., 2019). Altogether, these reasons make Bayesian methods a useful toolfor data analysis.All models were created in the Stan computational framework and accessed using the brms package (Bürkner, 2017, 2018), version 2.17.0. Allmodels were run with 4 chains and 5000 iterations, of which 1000 werewarmup iterations. We checked model convergence by inspecting the traceplots, histograms of the posteriors, Gelman-Rubin diagnostics, and autocorrelation between iterations (Depaoli & van de Schoot, 2017). We found nodivergences or excessive autocorrelation in any model.Dot-probeTo analyze the dot-probe data, we used Bayesian mixed models with a Gaussian distribution. First, to study the association between attractiveness andimmediate attention, we modeled Reaction time (mean-centered by subject)as a function of Attractiveness rating of probe picture and AttractivenessIliana Samara 17x24.indd 57 08-04-2024 16:35