Hi,
I'm trying to do a hierarchical multiple regression with level of depression as my DV and it appears to be showing a floor effect according to the scatterplot. People seem to generally score low on depression which appears to be causing this. Is there anything I need to do in this case and if there is, what would I need to do in order to proceed with the analysis?
Any help would be really appreciated.
Hi Robert
What about categorizing the depression scores into a binary variable and using binary logistic regression. If the scale has an established cut-off score to identify 'caseness' then you have a clear way to do the categorization. If no cut-off score is available then you could use a cut-off that identifies the 'high' participants (maybe those scoring more than 1.5 standard deviations above the mean). Categorizing continuous variables is generally not a good idea (follow the links in http://www.methodsappraisal.com/education-dichotomania/) but you may get dispensation in this situation.
Mark
Thank you for taking the time to respond. I'm not entirely sure how to do that as I've never heard of binary logistic regression before. I'm just a student and am working in SPSS if that means anything. Depression is measured by the DASS-21 and they do give some categories for severity. I was also wondering if seeing a floor effect is that unusual seeing as how most people in the general population which I sampled would not have much depression anyway.
Hello Robert,
The population from which your sample comes may, in fact, be one with relatively low rates of clinically identifiable depression. So, you simply acknowledge that fact and continue with your analysis. When all is said and done, however, it might be a good idea to note that the results you obtain might or might not generalize to a clinical population.
Good luck with your work!
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress