What are the statistical tests need to be done before backward multiple regression?

#1
I am doing a Final Year project on the topic Depression where i want to find the which factor has the most significance effect on depression . But before i carried out the multiple regression tests, do I need to perform other test such as pearson's coefficient of correlation and chi square test?
 

ondansetron

TS Contributor
#2
If you are looking for the importance of a variable on depression, a p-value won't tell you that, necessarily.
 
Last edited:

hlsmith

Not a robit
#4
You don't need to perform test before regression - those tests can be misleading (don't address confounders, collinearity, mediation) and descriptive in nature. You just run your regression given your content expertise in the area and knowledge of the relationships between terms.

However, given your question, it may be prudent for you to better describe your scenario so we can provide additional pointers or areas of concern. So what is your analytic question, how are data formatted, how many variables, and what is the sample size!
 
#5
i attached the questionnaire i going to use.
the main purpose is to find out which factors is significant to depression score ( which measure by BDI)
For now , I'm planned my methodology part like this : Pearson correlation coefficient , multiple regression and if the data is not normal then non-parametric test such as chi square test , mann whitney u test will be use.......
Then , today after meeting with my supervisor , she did mention about something like goodness of fit test ......
 

Attachments