Predicting daily-life antisocial behaviour in institutionalized adolescents with Transgression-related Implicit Association Tests

Abstract

This study investigated whether implicit attitude (IA) and implicit self-concept (ISC) towards transgression predicted daily-life antisocial behaviour (ASB) in institutionalized adolescents. Eighty-seven adolescents completed two Implicit Association Tests (IAT) assessing IA and ISC towards transgression, and reported four times a day during eight days the intensity of their ASB. Staff members concurrently reported the intensity of each adolescent’s ASB. Adolescents filled in the Inventory for Callous-Unemotional Traits and answered a semi-structured interview assessing conduct disorder. Multilevel regression analyses confirmed that adolescents’ ISC towards transgression significantly predicted their self - and staff-reported ASB over and above conduct disorder and callous-unemotional traits. However, adolescents’ IA towards transgression did not predict their ASB. Results indicate that ISC towards transgression is a reliable predictor of daily-life ASB in institutionalized adolescents. These results suggest that transgression-related ISC represents a promising target for ASB prevention in institutionalized adolescents.

Keywords:

• Institutionalized adolescents
• antisocial behaviour
• explicit and implicit measures
• Implicit Association Test
• Ecological Momentary Assessment

Notes

1. This new scoring procedure enables (1) to reduce the correlations between IAT effects and individual subjects’ average response latencies, (2) to minimize the effect of the order of the IAT blocks and (3) to decrease the effect of previously completing one or more IATs on IAT scores, while retaining strong internal consistency. In addition, it has also been found to maximize the correlation between implicit and explicit measures (Lane et al., 2007). It implies that participants with more than 10% responses lower than 300 ms are eliminated from the analyses (such extreme responses being considered as random), as well as participants with more than 20% of error responses. Moreover, for each participant, response latencies greater than 10000 ms are replaced by the mean latency of the block to which 600 ms were added.