Poster #202 - A Comparison of Person-Centered Measures of Child Temperament Based on Mother, Experimenter, and Coder Ratings

• In Event: 1-115 - Poster Session 03
  In Poster Session: PS 03 Section - Social, Emotional, Personality

Integrative Statement

Temperament researchers consistently report small or non-existent associations between parent reports and observed measures of child temperament (Stifter et al., 2008). Despite the increased popularity of person-centered approaches to study temperament via parent reports and observations of temperament, the resulting groups or classes have not been examined previously for consistency across methods of assessment. The purpose of the present analysis was to compare temperament group classifications derived from maternal reports, experimenter ratings, and observed behaviors at 18 months of age.

Children (N = 134, 52% male, 91% non-Hispanic white) completed a laboratory visit at 18 months of age and participated in tasks intended to assess child temperament and early regulatory abilities. Following the visit, trained experimenters provided multiple ratings about child temperamental qualities using the Infant Behavior Record (Stifter et al., 2008). Child temperamental affect and approach-withdrawal responses were coded from a Risk Room task at the start of the visit. Mothers reported on child temperament before the visit using the Early Childhood Behavior Questionnaire (Putnam et al., 2006).

Person-oriented temperament groups were formed separately from each set of data using latent profile analysis. All final models converged to a proper solutions with acceptable fit statistics (entropy values > .85). Mother ratings yielded average, exuberant, regulated, and negative groups. Experimenter ratings yielded average, non-compliant, social, negative, and positive/regulated groups. Coded behavior during the Risk Room task yielded average, exuberant, and inhibited groups. Children were classified to observed groups based on highest posterior probabilities of group membership.

Chi-square tests were used to compare the temperament group classifications across methods of assessment. Significant relations among group classifications emerged between the experimenter ratings and parent reports, χ2(12, N = 134) = 30.31, p < .01, as well as between the experimenter ratings and observed temperament, χ2(8, N = 134) = 38.20, p < .01. Specifically, toddlers in the experimenter-rated negative group were more likely than expected to also be classified in the mother-rated negative group (standardized residual = 3.0), as well as in the observed inhibited group (std. res. = 4.0). Toddlers in the mother-rated regulated group were more likely than expected to be also classified in the experimenter-rated positive/regulated group (std. res. = 2.2) and less likely than expected to be classified in the experimenter-rated negative group (std. res. = -2.2). There were no significant relations among the mother-rated and observed temperament group classifications.

Our results demonstrate some consistency in temperament group classifications derived from maternal reports, experimenter ratings, and observed behaviors at 18 months of age. In particular, experimenter ratings of behavior across a laboratory visit were more consistent with maternal ratings than were observations of behavior during one condition. Furthermore, experimenter ratings more closely aligned with maternal ratings on certain temperamental qualities, such as negativity and regulation, compared to others, such as surgency or exuberance. Our study results provide support for a components-of-variance approach to studying temperament which suggests that each type of assessment provides valuable, and in most cases distinct, information about child temperament.

Authors

• Kameron J. Moding, University of Colorado Anschutz Medical Campus
  Presenting Author

• Mairin Augustine, UNC Chapel Hill
  Non-Presenting Author

• Cindy Stifter, Pennsylvania State University
  Non-Presenting Author