Measuring Racial Group Invariance of the Financial Well-Being Scale

Background and Purpose: The Consumer Financial Protection Bureau’s (CFPB) Financial Well-Being (FWB) Scale is widely used in research related to the Grand Challenge for Social Work—Build Financial Capability and Assets for All; yet, few measurement analyses have been conducted and none have explored cross-cultural validity concerning racial/ethnic group identity. The scale has been used in multiple countries, suggesting cross-cultural application; in the U.S., score differences by racial and income groups have been observed. FWB may be influenced by cultural factors and/or racial and ethnic economic marginalization. To better understand score differences and FWB across race and ethnicity, the aim of the study is to measure overall scale group invariance, as items may also be quantifying general patterns of optimism and pessimism, which vary by race. We hypothesize that a two-factor model that measures a global sense of optimism and pessimism will fit the data better than a single factor model, and that the scale is variant across racial/ethnic groups.

Methods: The study used data from CFPB’s 2017 National Financial Well-Being Survey (N = 6,394). Mean FWB score was 56.08 (SD = 14.07; scale range 14 to 95, with higher as better FWB); 29.7% of respondents identified with a minority racial/ethnic identity. Using confirmatory factor analysis (CFA), the as-designed single-factor model was compared to a two-factor model—distinguishing between optimism and pessimism in positively and negatively worded scale items. Measurement invariance of the two-factor model across racial groups was then assessed sequentially using configural, weak (factor loadings set to equality), and strong (intercepts and loadings set to equality) models. The project was carried out with the lavaan package for R, using oblimin rotation and the WLSMV estimator.

Results: Preliminary factor adequacy was ascertained (KMO = 0.94 overall, range 0.92 to 0.96). Initial assessment of the single-factor FWB model yielded modest, though poor overall, fit results (SRMR = 0.056; CFI = 0.989; TLI = 0.986; RMSEA = 0.107, 90% CI [0.104, 0.110]). The two-factor CFA model had good fit (SRMR = 0.036; CFI = 0.996; TLI = 0.995; RMSEA = 0.066, [0.063, 0.069]). Factor loadings ranged between 0.667 and 0.873. The test of invariance was not supported by the data; the change in 𝝌² from the configural model was significant (p = .00295). Thus, the model measuring FWB differs by racial/ethnic group identification. To confirm findings, a partial invariance test was conducted; only one item (I am just getting by financially, p = .05177) was modestly invariant.

Conclusions and Implications: The FWB Scale may be concluded as being variant across racial groups (such that different groups may construe statements differently). Results from its use in cross-racial and ethnic group analysis should be interpreted with caution. Grouping the scale items into two factors (optimism and pessimism) can help researchers better interpret the moderating role of racial/ethnic group identification in explaining differences in FWB. Further psychometric and qualitative research is needed to recenter well-being narratives on marginalized populations whose perspectives are often excluded from the discourse on FWB and its measures.