Table 1.
Study design overview.
| Question | Hypothesis (if applicable) | Sampling Plan (e.g., power analysis) | Analysis Plan | Interpretation given to different outcomes |
|---|---|---|---|---|
| RQ1. Is the judgement of trustworthiness of singular facial parts (eyes/ mid-face /mouth) different from the judgement of trustworthiness of whole faces? |
H0: Trustworthiness judgements of singular facial parts are not different from trustworthiness judgements of the whole face H1: Trustworthiness judgements of singular facial parts are different from trustworthiness judgements of the whole face |
α = 5%, minimum power = 99%, two-sided; ICC = 0.30; 15% non-response and dropout rate), needed sample size is N = 2276 raters53 | Random-effects multilevel model: L1 (within-person) = ratings of different faces/facial parts; L2 (between-person) = raters | If trustworthiness judgements of one or more facial parts do not significantly differ (p ≥ 0.05) from the whole face (i.e., reference category), then this/these facial part/s is/are primarily responsible for the trustworthiness judgments of faces |
| RQ2A. Do humans judge the trustworthiness of faces/facial parts of the stimuli from their own ethnicity differently compared to stimuli from other ethnicities? |
H0: Humans do not judge the trustworthiness of faces/facial parts of the stimuli from their own ethnicity differently compared to stimuli from other ethnicities H1: Humans judge the trustworthiness of faces/facial parts of the stimuli from their own ethnicity differently compared to stimuli from other ethnicities |
α = 5%, minimum power = 99%, two-sided; ICC = 0.30; 15% non-response and dropout rate), needed sample size is N = 2276 raters53 | Random-effects multilevel model: L1 (within-person) = ratings of different faces/facial parts; L2 (between-person) = raters |
If one of the rater ethnicities is significant (p < 0.05), this means that raters judge the trustworthiness of whole faces/facial parts differently depending on whether target’s ethnicity matches/mismatches their own ethnicity If none of the rater ethnicities are significant (p ≥ 0.05), this means that raters judge the trustworthiness of whole faces/ facial parts independent of whether target’s ethnicity matches/mismatches their own ethnicity |
| RQ2B. Do humans judge the trustworthiness of faces/facial parts of stimuli from the dominant ethnicity of their social environment differently compared to stimuli from other ethnicities? |
H0: Humans do not judge the trustworthiness of faces/facial parts of stimuli from the dominant ethnicity of their social environment differently compared to stimuli from other ethnicities H1: Humans judge the trustworthiness of faces/facial parts of stimuli from the dominant ethnicity of their social environment differently compared to stimuli from other ethnicities |
α = 5%, minimum power = 99%, two-sided; ICC = 0.30; 15% non-response and dropout rate), needed sample size is N = 2276 raters53 | Random-effects multilevel model: L1 (within-person) = ratings of different faces/facial parts; L2 (between-person) = raters |
If the rater’s dominant ambient ethnicity is significant (p < 0.05), this means there is a difference between raters’s dominant ambient ethnicity and other ethnicities regarding the judgements of trustworthiness of whole faces and facial parts If the rater’s dominant ambient ethnicity is not significant (p ≥ 0.05), this means there is no difference between the rater’s dominant ambient ethnicity and other ethnicities regarding the judgements of trustworthiness of whole faces and facial parts |
| EA1: target sex | Not applicable | α = 5%, minimum power = 99%, two-sided; ICC = 0.30; 15% non-response and dropout rate), needed sample size is N = 2276 raters53 | Random-effects multilevel model: L1 (within-person) = ratings of different faces/facial parts; L2 (between-person) = raters | |
| EA2: rater sex, eye color, and hair color | Not applicable | α = 5%, minimum power = 99%, two-sided; ICC = 0.30; 15% non-response and dropout rate), needed sample size is N = 2,276 raters53 | Random-effects multilevel model: L1 (within-person) = ratings of different faces/facial parts; L2 (between-person) = raters | |
| EA3: difficulty of rating the stimuli of the full faces, eyes parts, mid-face parts, and mouth parts | Not applicable | α = 5%, minimum power = 99%, two-sided; ICC = 0.30; 15% non-response and dropout rate), needed sample size is N = 2,276 raters53 | Random-effects multilevel model: L1 (within-person) = ratings of different faces/facial parts; L2 (between-person) = raters |