PREreview del Context matters: Need frustration predicts self-critical perfectionism within domains and over time

This is a copy of a review that I prepared for a journal in August 2023.

The current paper presents two studies examining associations between basic psychological need frustration and perfectionism. First, the good: The paper is well written and clear. I appreciated the authors’ attention to open science practices. I also appreciated the simple and clear tables with descriptive statistics. However, on the negative side, the paper draws too strong of conclusions from its results, particularly for Study 2. I outline specific critiques below.

1. Concerning the Study 2 results, I had a clarifying question, and a more fundamental concern. As to the clarifying question, I do not believe the values in Table 3 match the description in the text. Specifically, the coefficient for T3 need frustration is negative (indicating a decrease in self-critical perfectionism), and the coefficient for T3 need satisfaction is positive (indicating an increase in self-critical perfectionism), but the text states the opposite (top of p. 15). Do I have that right? I would have checked the results myself, but the data/syntax for Study 2 are missing from OSF.

2. Regarding my more fundamental concern about Study 2, models like the ones done here are prone to spurious results. Specifically, there is a strong relationship between T1 measures and follow-up measures. The variance that remains to be explained after accounting for that stable over time association is minimal, and laden with measurement error. See Westfall and Yarkoni (2016, PLOS) for more information. Also relevant is Hamaker et al. (2015, Psych Methods). Unfortunately, there is not much that can be done to rectify the problem when only two occasions of measurement exist. The only solution I can see is for either this study to be removed from the paper or for its conclusions to be made much more tentative, acknowledging the serious problems that the study’s design causes for conclusions. (Note, this flaw might also be a problem for Study 1, as variables are not modeled accounting for measurement error, and there are strong associations between some variables, such as a need frustration and need satisfaction).

3. Heavy use of causal language/conclusions, particularly in the discussion, is another example of overclaiming that ought to be toned down in revision.

4. Both types of perfectionism were correlated very strongly over time (.74 and .79, respectively) in Study 2. By any definition, this is very high rank-order stability. The paper’s conclusion that implies that perfectionism is not stable (not trait-like) is not supported by the data here. Likewise, the ICCs in Study 1 suggest reasonably high levels of variability are explainable by individual differences.

5. There are some missing details need to understand the design in Study 1. The text says that “participants were presented with four textboxes to indicate their chosen domains, which were later categorized by two sets of research assistants,” (p. 8), but (unless I’m missing it), the text does not explain how the responses were categorized (e.g., what categories, how many categories). I’m confused about whether or how domains are at play in the current analyses. Based on the OSF code, it appears that domain classifications are not at play here, but I was confused nonetheless.

6. Notes about study transparency: Thank you to the authors for including much of the information needed to assess study transparency. Please see the journal’s submission guidelines for directions about how and where to include such information (specifically, there should be a subsection of the method section where all relevant information is put in one place). Regarding power analyses, the authors write “We were unable to calculate any a priori power analyses” (p. 6). This is not a problem, but instead the authors should present sensitivity analyses. These analyses test what size effect a study with their sample size is 80% powered to detect. I examined the OSF page briefly and found it to be well organized. However, the folder for Study 2 “data files and syntax” appears to be empty.

7. Please clarify in the text (pp. 8-9) how the alpha reliabilities in Study 1 were obtained. Because the data are multilevel, the authors have a few options, and it isn’t clear what was done here. Reliability information is missing from Study 2.

I sign all of my reviews,

Katherine S. Corker

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.