EXAMPLE:
You have 2 groups, each with 100 participants. Each group receives 10 questions. For each of the 10 questions, you have 2 measures of performance. The same 2 measures of performance are used for each of these 10 questions.
For each of these 10 questions, you can calculate the Pearson-r correlation between the 2 performance indicators. This will produce 10 correlations for each of the 2 groups, each with a df of N-2 = 98 and their own p value.
How would you determine if the average correlation in each condition was significant, and if the average correlations differed between conditions?
You can average the correlations following a Fisher-z transformation. But what is the proper way to determine the significance of this averaged correlation, and whether it differs between condition? And does this averaged correlation have the same degrees of freedom as the individual, un-averaged correlations (98)?
For clarification- there is a theoretically motivated reason I am interested in the average of the correlations rather than the correlation of the averages (i.e. the correlation between the average of the two performance indicators across the 10 questions)
https://psycnet.apa.org/record/1987-14534-001
Article Averaging Correlations: Expected Values and Bias in Combined...
https://www.psychometrica.de/correlation.html
https://methods.sagepub.com/reference/encyclopedia-of-measurement-and-statistics/n175.xml
Thanks for the links, Cristian. They were helpful for understanding averaged correlations generally, but unfortunately they did not address the core issue I was concerned with.
Fortunately, I believe I have found a suitable solution to my problem:
Article Repeated Measures Correlation
Edit: Though I am still a little unsure in this case what the appropriate test to compare these correlations between group would be. Perhaps the Eid, Gollwitzer, & Schmidt, 2011 method would suffice, adjusting for the df for the r(repeated measures).
Like any other averaged correlations is what Cristian Ramos-Vera . is saying.The Z transform just means you did the first part correctly. D. Booth
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication