Hello everyone! I am trying to correlate three types of assessments, two are continous and one is ordinal. Since they are different, I have different numbers of variables. Except Pearson correlation what other statistical method/test can I chose! I have now 32 subjects, 2 scales have each 19 variables and the other one has 6. Any idea?
thank you!
You can perform biserial correlation for ordinal data categories, also have to take into account the distribution of normality of the instrument to know whether to perform spearman or peak correlation
Hello Roman,
Spearman (rank-order) correlation could be used for any pairing that involved an ordinal variable. You could use it for the "continuous" variables as well, as it really is nothing but a Pearson correlation on the ranked values; you'll find that the results (Spearman vs. Pearson on interval/ratio data) tend to be pretty close.
Good luck with your work.
David Morse gave a pretty good answer to your question. For more details, you might see this link: https://www.google.com/search?q=correlation+coefficients+which+to+use&rlz=1C1CHBF_enUS874US874&oq=correlation+coefficients+which+to+use&aqs=chrome..69i57.38621j0j9&sourceid=chrome&ie=UTF-8 Best, D. Booth
- Badges
- Science method