Toward an empirically based structure of infant temperament: Psychometric properties and factor structure of the Infant Behavior Questionnaire-Revised

Abstract

This dissertation aims to examine the psychometric properties of the Infant Behavioral Questionnaire-Revised (IBQ-R; Gartstein & Rothbart, 2003) and to improve the instrument on the process. It focuses mainly on the dimensional and higher-order factor structure of the IBQ-R because there is a lack of empirical examinations and supports despite the significant role of the IBQ-R as the most widely used instrument and theoretical framework in the infant temperament field. Datasets employed came from 11 previous studies using the IBQ-R, shared by the original authors. The final dataset included a total of 3068 unique infants (1484 Females, age 3 to 13 months old). I first examined the psychometric properties of the original IBQ-R and conducted item-level exploratory factor analyses. The results demonstrated that many of the original scales were not unidimensional and homogeneous in item contents. The new scales created based on the EFA included fewer items and scales but demonstrated improved/maintained reliability and fit to the unidimensional model. The original and new scales were further examined for construct validity using other measures of psychological constructs. The new scales improved convergent and discriminant validity by making each scale more homogeneous while explaining similar amounts of variance in the outcome variables, despite the reduced number of items and scales. Finally, I examined the higher-order factor structure and longitudinal invariance of the factor structure using the new scales. The results demonstrated that there existed age-dependent invariance in the factor structure. As the first item-level analysis of the IBQ-R utilizing a large sample, this dissertation demonstrated that the IBQ-R might not provide an accurate reflection of the dimensions of infant temperament and suggested ways to improve it. The newly built scales from this study will be a good starting point to build better scales, which ultimately can help to further clarify the dimensions and structure of infant temperament.