(PS8-44) Examining Measurement Invariance across Levels of Internalizing Psychopathology: A Test of the Ruminative Responses Scale

Abstract Text: Establishing measurement invariance is critical for accurately assessing group differences and changes in mental health symptoms over time. However, measurement invariance is typically examined only across sociodemographic factors, overlooking how internalizing psychopathology may alter response patterns and item interpretations due to cognitive biases.

We tested the invariance of the Ruminative Responses Scale (RRS) across levels of depression, anxiety, and suicidal ideation severity among 1,275 participants combined across four samples with varying levels of psychopathology and suicide risk (Mage = 27.65 years; 63.5% cisgender women, 68% White). Participants were grouped by levels of depression on the Beck Depression Inventory – II (Minimal: n = 605; Mild: n = 174, Moderate: n = 194, Severe: n = 281), levels of anxiety on the Beck Anxiety Inventory (Minimal: n = 377 ; Mild: n = 260; Moderate: n = 256; Severe: n = 343), and levels of suicidal ideation on the Beck Scale for Suicidal Ideation (No Risk [BSS = 0]: n = 805; Low Risk [BSS = 1-10]: n = 236; High Risk [BSS = 11+]: n = 216).

The two-factor (brooding and reflection) RRS model showed good fit (χ²[32] = 219.82, p < .001, CFI = .97, TLI = .96, RMSEA = .07, SRMR = .04). Across levels of depression severity, measurement invariance testing supported full metric invariance (Δχ2[24] = 36.19, p = .053), as well as partial scalar invariance after freeing one intercept (Δχ2[21] = 31.83, p = .061). Full uniqueness was not met, even after freeing over 25% of residual variances (Δχ2[21] = 123.75, p < .001). Across anxiety, partial metric (Δχ2[21] = 29.54, p = .102) and scalar (Δχ2[18] = 20.09, p = .328) invariance was supported after freeing one factor loading and two intercepts, respectively; full uniqueness was not met (Δχ2[21] = 123.04, p < .001). Finally, across absent, mild, and severe suicidal ideation, there was support for full metric (Δχ2[16] = 20.62, p = .194) and partial scalar invariance (Δχ2[12] = 19.50, p = .077), but not full uniqueness (Δχ2[14] = 99.35, p < .001).

Given that full measurement invariance was not achieved, the results highlight the importance of evaluating invariance specifically across severity levels of internalizing psychopathology (depression, anxiety, and suicidal ideation in the present study). Cognitive biases associated with internalizing disorders may influence response patterns, potentially affecting the interpretation of scores and the validity of group comparisons. Nonetheless, at least partial scalar invariance was obtained in all cases, demonstrating that the RRS can be used to assess ruminative tendencies across varying levels of internalizing psychopathology. Limitations, implications, and future directions will be discussed.