Page 43 - Demo

Attentional Biases to Facial Emotions412to normal with a light right skew and mesokurtic shape. Internal consistency of the LSAS in our sample was excellent (%u03b1 = .91, 95% CI [.89, .94]).Autistic TraitsWe used the Autism-Spectrum Quotient (AQ; Baron-Cohen et al., 2001) to measure variations in traits associated with Autism Spectrum Conditions in our sample. The AQ consists of 50 items associated with five different domains in which alterations are typically observed: social skill, attention switching, attention to detail, communication, and imagination. Ten items belong to each domain respectively and build one subscale of the AQ. In this questionnaire, respondents rate the degree to which items apply to them on a 4-point Likert scale, from 1 = definitely agree to 4 = definitely disagree. Some items are reverse coded and all scores are eventually transformed to binary values (1/2 to 0 and 3/4 to 1). A higher sum score of all items, potentially ranging between 0 and 50, reflects higher autistic trait levels. Three participants did not complete the AQ at all and we imputed missing items for four participants who had incomplete data (see Online Resource 2). The AQ sum scores encompassed values between 2 %u2013 38 (M = 18.3, SD = 7.6) in our sample (N = 99; participants without task data excluded). With a skewness of 0.69 and a kurtosis of 3.33, the AQ score distribution was also close to normal, yet slightly right-skewed and platykurtic. The AQ in our sample showed a good internal consistency (%u03b1 = .84, 95% CI [.79, .88]).Data AnalysisBefore fitting the models, reaction times smaller than 250 ms were excluded given that they likely represent random responses (see van Berlo et al., 2020). Further, for each participant, trials exceeding their median reaction time + 2.5 median absolute deviations were excluded to filter out trials in which participants might have been distracted, thus resulting in relatively unusual high RTs. This lead to an exclusion of 9.47 % of all trials (angry: 8.83 %, happy: 9.24 %, sad: 10.25 %, fearful: 10.02 %, neutral: 8.59 %). All analyses were performed in R 3.6.3 (R Core Team, 2020; see Online Resource 2 for further information). Pre-registered Data AnalysisFor all hypotheses, we aimed to fit multiple linear mixed models on reaction times. Looking at the model diagnostics, we did not spot major divergences from assumptions. The selection of the basic fixed and random effect structure