Source
Chen et al. (2015), LIT-2000, pp. 223–227 and Table 3; read in NOTE-1527.
What was actually shown
The item pool. Of 42 drafted items, “half of the items were concerned with satisfaction of each need and the other half with frustration” (p. 224). The 24 retained (Table 3) are four per subscale. The twelve satisfaction items each assert a positive feeling (“I feel a sense of choice and freedom in the things I undertake”; “I feel capable at what I do”; “I experience a warm feeling with the people I spend time with”). The twelve frustration items each assert an aversive one (“I feel forced to do many things I wouldn’t choose to do”; “I feel insecure about my abilities”; “I feel excluded from the group I want to belong to”). No satisfaction item is reverse-worded, and no frustration item is.
The evidence for two dimensions per need. Within-need EFAs gave two factors, satisfaction items on one and frustration items on the other (p. 226). In CFA the six-factor model fit better than the three-factor one: CFI .93 against .87, ΔSBSχ²(12) = 450.79, p < .01. The final six-factor model was metric-invariant across Belgium, China, the USA and Peru (ΔSBSχ²(54) = 69.57, p > .05).
Why this is not decisive. In a scale where content and wording direction coincide, a method factor for positively against negatively valenced statements separates the same items into the same two groups. The fit comparison could have favoured three factors, and it did not; but the alternative it rules out is “one bipolar factor per need with no method variance”, not “one factor per need plus a keying factor”. The paper fits no model with method factors and no balanced-keyed version.
The check here is a reading of the items and the reported model comparison, not a reanalysis.
What this does not say
- That need frustration is not a distinct dimension. It may be; the instrument’s own evidence is compatible with both. Murphy et al. 2023 (LIT-1978) argue the artefact case and are unread here.
- That the frustration–ill-being correlations are spurious. Keying is one alternative; shared content with depressive items, raised in THEORY-052, is another, and the reconstructed covariance of −.84 between the latent satisfaction and frustration composites (Fig. 1) is a third reason for caution. None is tested in the paper.
- Anything about other frustration measures, such as Bartholomew et al.’s thwarting scale (LIT-1954).