Page 163 - Emotions through the eyes of our closest living relatives- Exploring attentional and behavioral mechanisms

Congruency (Congruent first)
Task Version (V2) Task Version (V3)
(Intercept)
Congruency (Congruent first)
Task Version (V2) Task Version (V3)
.15 .04
-.01 .06 -.01 .06
.15 .06 -.21 .06
.12 .08 .19 .08
.07 – .23
-.12 – .11 -.12 – .10
.09 – 0.21 .04 – 0.17
-.19 – .02 -.06 – .10
Validation of the Pictorial Implicit Association Test
added, this effect remained present (D-score: .15, 95% CI [.09, .21], t(138) = 4.83, p < .001). We also found a significant effect of Congruency on D-score (F(1, 138) = 9.18, p = .003); children who received incongruent trials first responded significantly faster than children who received congruent trials first (incongruent: M = .25, SD = .34; congruent: M = .06, SD = .39, supplemental Figure S4). Additionally, there was a significant effect of Task Version on D-scores (F(2, 138) = 3.08, p = .05, Figure S4). When plotting the data, all task versions cause the IAT effect in the expected direction (i.e., an average D-score above 0), but task version 1 causes the lowest (M = .05, SD = .38) and task version 3 the highest effect (M = .18, SD = .34, task version 2: M = .17, SD = .40, S3). Indeed, post-hoc comparisons using the Tukey procedure revealed a significant difference between task version 1 and 3 (V1-3: p = .040, V1-2: p =.233, V2- 3: p = .627).
Finally, to assess the consistency of results across all items, we correlated D-scores based on only the first half of the trials within the critical blocks with D-scores based on only the second half and found a split-half reliability of r = .69, indicating acceptable internal consistency in the child PIAT.
Table 1. Model results for the adult and child PIAT
Adult PIAT
Child PIAT
Predictors Estimates
(Intercept) .24
D-score
SE 95% CI
.04 .16 – .32
t p
5.97 <.001 3.78 <.001
-.13 .900 7 -.21 .832
4.83 < .001 3.37 .001
-2.33 .02 .45 .65
Conclusion
The results show that both children and adults appeared to have a significantly negative implicit bias towards faces of men of Moroccan descent and a positive implicit bias towards faces of men of Dutch descent, but children appear to have a weaker implicit bias than do adults. The PIAT thus appears suitable for testing both
161