Measurement Invariance of the Parental Acceptance-Rejection Questionnaire: Across Ratings of Children and Mothers

• In Event: Poster Session 10
  In Poster Session: PS 10 Topic - Parenting & Parent-Child Relationships

Abstract

The short version of the Parental Acceptance-Rejection Questionnaire (PARQ/S; Rohner et al., 2012) has been widely used cross-culturally. Both the parent and child versions of the PARQ/S ask about reporters’ feelings about the parent-child relationship with respect to hostility, neglect, rejection, warmth, and behavioral control. Tests of measurement invariance have been done across countries and across different ethnic groups (e.g., Gomez & Suhaimi, 2015; Senese et al., 2016). These tests include within-reporter comparisons (e.g., test for measurement invariance in child reports across countries). In addition, focused on mean comparisons and correlations, researchers have compared mother’s and father’s self-reports to one another (e.g., Gomez & Suhaimi, 2015), and likewise children’s reports on mothers and fathers (e.g., Artemis & Touloumakos, 2016). However, to our knowledge, no studies have tested for measurement invariance across parent and child reports. Given that many studies using the PARQ/S incorporate reports from both members of the child-parent dyad, it is important to establish that it is valid to jointly examine and compare parent and child reports. Thus, the aim of the current study was to examine measurement invariance between mother’s and child’s reports of PARQ/S using both the parent and child versions.

Participants were 300 mother-child dyads (49% female) from the southeast of United States. Participants were diverse in terms of race and ethnicity, including 36% non-Hispanic White, 33% African American, and 31% Hispanic American. Data were collected when children were around nine years old (M = 8.6, SD = .66). Children and mothers reported maternal acceptance and rejection using PARQ/S (Rohner et al., 2012), which has five subscales: warmth, hostility/aggression, undifferentiated rejection, indifference/neglect, and behavioral control.

Confirmatory factor analyses (CFAs) were conducted separately for child’s and mother’s reports of PARQ/S to examine if configural variance existed between their reports (see Table 1 for factor loadings and see Table 2 for the five factors’ variances and correlations). Since the correlations among child’s reports of hostility, rejection, and neglect were .88-1.11, we combined the three factors into one factor. This more parsimonious model did not lead to a deterioration in model fit, ∆χ2(7) = 13.59, p = .06, suggesting the more parsimonious model is the optimal model. For the model of mother’s report, one loading in the neglect factor, one loading in the behavioral control factor, and two loadings in the rejection factor were not significant. Results suggested that the five-factor model may not fit either the model of mother’s report or the model of child’s report. We also combined hostility, rejection, and neglect into one factor for the model of mother’s report. This more parsimonious model led to a deterioration in model fit, ∆χ2(7) = 18.58, p = .01, suggesting that the three-factor model may not fit the model of mother’s report either. Thus, configural variance existed between models of child’s and mother’s reports. Where feasible, we will continue to examine other types of measurement invariance (i.e., metric and scalar) and use multiple-group CFAs to examine measurement invariance of both reports across different ethnic groups.

Authors

• Xiaoye Xu, University of Southern California
  Presenting Author

• Justin Jager, Arizona State University
  Non-Presenting Author