Regression analysis : with ranking data


New Member
In my survey

I have independent variables of age , sex and length of membership

I have then have a question where I ask participants to rank 5 statements in order of importance . with 5 being the most important and 1 being the least important .

Is regression a valid tool to analyse the relationship between these the independent variables and ranking

e.g did the older members choose statement 1 as the most important

did younger members choose statement 1 as the most important

I,e to asses the relationship between the independent variables and the choice in terms of ranking
Hey Epy,

You probably can do some sort of regression, but not a standardised multiple regression, or I wouldn't recommend it as you have clearly five defined values in your dependent variable. Thus the results from an OLS-regression wouldn't be intuitive to interpret.

However, you can check out an ordinal logistic regression. I think Long and Freese (2014) has some very decent chapters on it. It basically allows you to calculate either the odds ratio/relative risk or coefficients for one profile (young) or another (old) to select either 1-5 on the scale of your dependent variable. Keep in mind that logistic regression tends to require a larger sample size.

Good luck!
Last edited: