Source
Campbell, Boone, Vansteenkiste and Soenens (2018), LIT-1964, Journal of Clinical Psychology 74(10); read in NOTE-1533.
What was actually shown
Study 1 (cross-sectional, N = 284, mean age 14.1). Latent variables from parcels, adjusted for age, gender and BMI. Depression (CDI) and eating pathology (EDI-II composite) correlate .50. Self-critical perfectionism (Frost CM and DA) predicts both (β .45, .34) and lowers the correlation to .42. With need frustration as full mediator (partial mediation no better, Δχ²(2) = 3.08), indirect effects are .55 [.41, .69] on depression and .28 [.18, .39] on eating pathology, and the residual symptom correlation is .18, not significant (pp. 6–7).
Study 2 (two waves six months apart, N = 608, 477 at both, mean age 16.2). Latent change models. Correlated change of the symptoms .43; with perfectionism change .39; with frustration change as partial mediator (better than full, Δχ²(2) = 17.34) .24, p < .01. Indirect effects of perfectionism change via frustration change: .26 [.17, .36] on depression change, .12 [.06, .19] on eating change (pp. 10–11).
The test could have failed: had the symptom association stayed near .50 with frustration in the model, frustration would not explain the comorbidity. In Study 1 it did not stay.
What this does not say
- That frustration precedes or causes the symptoms. Study 2’s mediator and outcomes are changes over one interval. The authors: “Due to the correlational nature of these findings, future experimental research is needed” (p. 12). LIT-561’s “predicted” is statistical.
- That it holds beyond internalizing symptoms, in clinical samples or in adults; the authors list these as untested.
- That it holds for need frustration as SDT defines it. The measure is the BPNSFS frustration items, which on THEORY-1311 may be keyed low satisfaction, and whose content (failure, pressure) is close to the perfectionism and depression measures.