Poster #126 - Chinese-American Mothers’ Depressive Symptoms: A Longitudinal Examination of the Bi-factor Structure of the BDI-II

• In Event: 1-169 - Poster Session 04
  In Poster Session: PS 04 Section - Methods, History, Theory

Integrative Statement

Due to acculturation stress (Sue et al.,2012), including linguistic difficulties, discrimination, lack of social support, and challenges in establishing new social relationships (Hwang et al., 2008), and parenting stress, Chinese-American mothers are particularly vulnerable to depression, which is significantly related to children’s increased internalizing and externalizing problems (Chang et al., 2004). The Beck Depression Inventory II (BDI-II) is widely used to assess depression (Wang et al., 2013), but the total score of depression created by summing all items of the BDI-II may fail to capture the unique effects of specific factors (Chen et al., 2012). The bi-factor model can resolve this disadvantage by specifying a general factor of depression and including the specific factors in one model. However, the bi-factor model has not been examined among Chinese-American mothers. Therefore, we examined the bi-factor structure of the BDI-II, and its measurement invariance across two waves to distinguish true change in depression from Chinese-American mothers’ response shifts over time (Fokkema et al., 2013).

Chinese-American mothers (N=233, Mage=37.80, SD=4.45) with preschool-aged children (Mage=4.56, SD=1.13) reported on depression (Beck et al., 1996) and psychological well-being (PWB, Ryff et al., 1995) at Wave 1(W1) and Wave 2 (W2), 6 months apart. Three specific depression factors included Negative Attitude (NA), Performance Difficulty (PD), and Somatic Symptoms (SS).

The bi-factor model demonstrated adequate fit at both W1 and W2 (Figure 1). After controlling for the specific factors, the General Factor of Depression accounted for 89% and 86% of the unique variance (ωh=0.89 in T1; ωh=0.86 in T2), which was higher than the recommended critical value of 75%. The unique variances accounted by specific factors ranged from 0.3% to 26%, lower than the recommended critical value of 50%. The General Factor and PWB were negatively correlated at W1, r=-.39, p<.01, and W2, r=-.24, p<.01, suggesting convergent validity of the general factor. For the three specific factors, only the correlations between NA and PWB were significant at W1, r=-.33, p<.01, and W2, r=-.22, p<.01. The relation between the factor of SS and PWB at W2 was significant (r=.24, p<.05). Finally, scalar measurement invariance was established over time (Table 1).

The bi-factor model reflected the optimal representation of the factor structure of the BDI-II in Chinese-American mothers. The general factor of depression demonstrated high reliability and convergent validity beyond subgroup factors. However, after controlling for the general factor, the specific factors did not achieve adequate reliability and convergent validity. These findings were consistent with previous research (Brouwer et al., 2013), which shows that the general factor reflected a reliable single construct, whereas the use of subscale scores may lead to misleading interpretation. Moreover, the measurement invariance in the factor structure, item loadings, and item thresholds over time indicated that the longitudinal changes can be attributed to true changes over time rather than Chinese-American mothers’ response shifts. Our focus of the BDI-II’s factorial validity and longitudinal measurement invariance facilitates a better understanding of the measurement of depression in Chinese-American mothers with young children, an understudied sample.

Authors

• Xiaofang Xue, University of Maryland, Baltimore County
  Presenting Author

• Jing Yu, NICHD
  Non-Presenting Author

• Kathy Vu, UMBC Department of Psychology
  Non-Presenting Author

• Charissa S. L. Cheah, University of Maryland, Baltimore County
  Non-Presenting Author