1. Objectifs

Intentions :
« The aim of the study was to test several latent class models of inter-informant agreement. We examined agreement between adolescents and their mothers on aggressive and nonaggressive conduct disorder. » (p. 78)

2. Méthode

Échantillon/Matériau :
The sample was composed of 14 to 17-year-old adolescents (n=1165) and their mothers. « We collected diagnostic interview data and applied latent class analyses to estimate inter-informant agreement » (p. 75)

Instruments :
« We used a French translation (Bergeron et al., 1992) of the Diagnostic Interview Schedule for Children-2.25 (DISC-2.25; Shaffer et al., 1991). » (p. 80)

Type de traitement des données :
Analyse statistique

3. Résumé

« The preferred model for aggressive conduct disorder for both males and females was a one-latent-variable/two-class model specifying no inter-informant disagreement beyond chance expectations. This model estimated the prevalence of aggressive conduct disorder to be 13% for males and 0.4% for females. For nonaggressive conduct disorder, a one-latent-variable/three-class model specifying asymmetric agreement was preferred for both males and females. This model estimated the prevalence of nonaggressive conduct disorder in adolescents to be 18% according to males and 13% according to mothers. Prevalence estimates were 12% according to females and 7% according to mothers. Symptom sensitivity estimates for all models were poor whereas specificity estimates were near perfect to perfect. Males had higher rates of aggressive and nonaggressive conduct disorder across informants. There was a high level of adolescent-mother agreement on both types of conduct disorder. However, there were some differences, suggesting that aggressive and nonaggressive are two valid subtypes of conduct disorder with different prevalence estimates and agreement levels. » (p. 75)