Hi all,
I am interested in squaring the Spearman rho correlation coefficients of my data, as a paper I read did the same thing and I felt that it offered a nice verbal interpretation.
I was wondering... is it simply a matter of squaring the correlation coefficient? E.g., if you have a correlation of rs= 0.47 between variable x and variable y, would this then equal 22.09% once squaring and multiplying by 100?
If so, does this mean that 22.09% of the variance in x can be explained by y? That is how the research paper I read phrased it.
Many thanks in advance!
@Evelina The link given below might help you out
https://stats.stackexchange.com/questions/44268/reporting-coefficient-of-determination-using-spearmans-rho#:~:text=Spearman's%20rho%2C%20for%20example%2C%20represents,the%20proportion%20of%20shared%20variance.
Hello Evelina,
Yes, your suggested interpretation is exactly what the squared rho means. The only caveat is that it is shared/explained variance of ranked scores rather than original scores.
The Spearman rho is just a Pearson correlation of the two sets of ranked scores, so a squared rho would be interpreted just as a squared Pearson r would be interpreted.
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation