Your critiques for Sirois, Sears, and Marhefka were again generally well done.
I'm trying, again, to list more specific (good) vs. more generic (less good) critique points here. The real goal here is to take specific things about the analysis in question (here, discriminant) and focus the critique -- as much as possible -- on this. Sample size, normality, etc. are always important...but they are less specific to discrim.
Paper failed to reprt dfs, or multicollinearity
May not have met discriminant function assumptions (unequal n) ... consider logistic
Paper misclassified 80% of donors
Failure to interpret counterintuitive direction of parent knowledge loading
Demographic differences not controlled/covaried
R-squared terms not reported for each model (i.e., canonical correlations)
Failed to focus on standardized canonical loadings; failed to use 0.3 cutoff
Do not report centroids for groups
Consider parent-child dependencies; should this have been a more repeated measures design?
Possible selection bias: Research volunteer may also be more prone to volunteering organs; other sample bias issues
Did not validate that organ donation card actually signed--should have examined social desirability
So many analyses--paper is a fishing expedition
High missing data rate biases sample
No data screening offered
Psychometrics of parent-child questionnaire not reported