I am planning to use Spearman, but can I still use Spearman Correlation even though my data is not ordinal?
Or do I convert my data to ranks first?
You can use the Spearman Correlation if your data are measured at the ordinal, interval, or ration levels. The good news is that there are no distributional assumptions. However, if your data are measured at the nominal level meaning you cannot attribute quantity to them, you cannot use Spearman. If your data are measured at the interval or ratio levels, you will lose information and precision by using Spearman. In those cases, you should consider Pearson, but Pearson does require distributional assumptions.
Dear Cady,
Real life exist before statistics, which is just a tool to expalin the world.
Then you have to know what you intend to analyse befeore saying that you want to use Ro...I think you have to "think" in the other way...
Have a nice time
Helena
A Spearman correlation is simply a Pearson correlation following a rank transformation. It doesn't require ordinal data (as noted above). If the problem is distributional you can also bootstrap CIs for a Pearson correlation (as Pearson's r doesn't assume normality except for inference). There are also other options such as various forms of robust correlation and regression.
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix