Comparing two dependent, non-overlapping correlations

I have two questions regarding the comparison of two correlation coefficients from a single (paired / dependent) sample:
  1. Is it legitimate to use a test for comparing independent correlation coefficients instead? My understanding is that the only drawback is that this would compromise power (SOURCE). Since I'm running power simulations, this shouldn't be a problem?
  2. If this is not possible, I would like to know if this procedure is legitimate: Steigerts Test for two dependent, non-overlapping correlations [psych::r.test()] requires specifying three correlations (r12, r13 and r23). I want to compare the average reliability (Cronbach's alpha) across multiple scales for two questionnaires (this can be done by transformations by Feldt and Charter (2006). My question is: If r12 is the average transformed alpha from questionnaire 1 and r13 for questionnaire 2, then what value do I choose for r23? Would that be overall alpha across questionnaire 1 and 2?

Thank you very much