Source
Levine, Brabander, Moore, Holding and Koestner (2021), LIT-1974, Motivation and Emotion 46(1); read in NOTE-1510.
What was actually shown
Study 1, a Canadian university, September, December and April 2018–19, N = 379 (304 at T3), 85% women. Study 2, the same university, April and late May 2020, N = 235. Need frustration is the mean of nine negatively worded BMPN items (“I feel lonely”, “I feel pressured in my life”); negative affect five adjectives; depression the CESD-R 10. Fully cross-lagged observed-variable path models, standardized coefficients.
| path | Study 1, T1→T2 | Study 1, T2→T3 | Study 2 |
|---|---|---|---|
| frustration → depression | .13, p = .01 | .17, p = .02 | .18, p = .010 |
| frustration → negative affect | .18, p = .002 | .16, p = .02 | .17, p = .018 |
| negative affect → depression | −.01, p = .85 | .10, p = .25 | .07, p = .287 |
| depression → frustration | .20, p = .007 | .01, p = .88 | −.01, p = .987 |
| negative affect → frustration | −.07, p = .27 | .17, p = .04 | .09, p = .287 |
In Study 2, self-reported frustration predicted family and peers’ reports of depression at β = .14, p = .074, controlling for their earlier reports.
This could have come out otherwise: had negative affect carried the prediction, its paths to later depression would have been the significant ones, as cross-sectional work had suggested.
What this does not say
- That frustration causes depression, or comes first. Without random intercepts (tried by the authors, reported only on OSF), lagged paths in a cross-lagged panel include stable between-person differences. The reverse path at Study 1’s first lag (.20) exceeds the forward one (.13).
- That what predicts is frustration rather than low satisfaction. The BMPN’s satisfaction half is not modelled, and the items are negatively worded, like the outcomes (THEORY-1311’s objection in another instrument).
- That frustration is transdiagnostic. The authors say so; the outcomes are one symptom domain and a mood state.
- Anything about the descriptives of Study 2. Its Table 3 repeats Study 1’s means and SDs digit for digit, so one of them is a copying error (NOTE-1510, Limitations). The path coefficients are reported separately and are not affected on their face.