Individual Submission Summary
Share...

Direct link:

Poster #118 - Correlated Trait–Correlated Method Minus One Analysis of the Three SCBE-30 Traits Across Multiple Raters

Sat, March 23, 2:30 to 3:45pm, Baltimore Convention Center, Floor: Level 1, Exhibit Hall B

Integrative Statement

We examine the factorial structure of the SCBE-30 questionnaire (LaFrenière & Dumas, 1996), analyzing convergent and discriminant validity of the three SCBE-30 traits (Social Competence, Anger-Aggression, Anxiety-Withdrawal) and explored measurement invariance across multiple raters (mother, father and teacher). Each rater has a unique perspective and access to a partially overlapping information about the same child (e.g., parents have limited information concerning their child’s behavior at school and with peers, whereas teachers may not have access to a child’s behavior at home or with the community) and their specific perspective as raters is source of interest. Since the same child was evaluated by multiple raters on multiple traits, we used Correlated Trait-Correlated Method Minus One [CT-C(M-1)] model to compare and contrast the different raters against each other. We kept all the 30 items and considered three indicators per trait-method unit. We used radial parceling procedure, a multivariate parceling approach, that considers relations among items and all the constructs in the model, based on double-factoring (Cattell & Burdsal, 1975; Little et al., 2013). Participants were 369 Portuguese children (173 boys and 196 girls) ages range 32 to 78 months (M= 55.85, SD= 11.54), were attending public preschools in Portugal. Most of the parents were either married or cohabiting (95. 1%).We performed preliminaries analyses using Mplus 7 software, robust maximum likelihood was used for estimation. First, we used mothers as reference model, the fit indices for the CT-C(M-1) model were good CFI=.94, RMSEA=.06, SRMS=.05, when the reference model were the fathers the fit indices were also good CFI=.95, RMSEA=.05, SRMS=.04. As we were expecting, when we use teacher as reference model the fit indices were lower, CFI=.88, RMSEA=.08, SRMS=.05. Generally, there was more convergence between mother–father ratings than mother–teacher or father–teacher ratings. We explore modification indices, the variance components and latent correlations to generate a model that can be used with all raters.

Authors