I want to preform a regression analysis with one independent variable (questionnaire score) and one dependent variable (range from 0-100). The dependent score is the problematic one- most participants receive either the score of 100 or 0, but some participants (not many) receive a score in between (e.g., 67). Therefore, the data is not binary. Can someone recommend a statistical method which fits this type of data?
Asked
Active
Viewed 19 times
1
Nur Givon-Benjio
- 11
- 1
-
If you rescale the outcome to $[0,1]$ (on your graph you already have) then logistic regression might still work. But you should only use it if the implicit model is theoretically justified – Henry Oct 19 '21 at 12:11
-
Look at fractional response models, for instance https://stats.stackexchange.com/questions/172557/proportion-dependent-variable-0-1-range-in-an-unbalanced-non-dynamic-panel and search this site. A better link: https://stats.stackexchange.com/questions/216122/what-is-the-difference-between-logistic-regression-and-fractional-response-regre – kjetil b halvorsen Oct 19 '21 at 12:55