Creating two groups based on numeric variable
Putting aside the issues raised above, if you decide that you still want to split your predictor variable into high-low groups, the following are some options
- Use Statistical properties of your sample
- Median split
- Above or below the mean
- Take bottom 25% and top 25% and throw out the middle
- Take bottom third and top third and throw out the middle third
- Use accepted or externally validated cut-offs
- e.g., medical diagnoses are often based on certain cut-offs on a continuous scale
- Use your own understanding of the phenomena to define a cut-off
- Examine a histogram or density plot and look for a natural split in the data