I need to look at the relationship between the number of counselling self-referrals among students from an institution and the total number of students in the same institution, for each year between 2000 and 2015. I have the number of self-referrals per year and the number of total students in that institution per year. What tests do I use to analyse the relationship between the two? Perhaps linear regression?
I'd be really grateful for your answers!
I'm thinking that a Chi-Square test might also be appropriate, given that I have information on both number of referrals and total number of students per year across all 15 years, according to gender, instrument and programme (undergraduate or postgraduate). Am I wrong in considering a Chi-Square test?
Hi Raluca,
As the number of students increases, the number of self-referrals should also increase simply because of population size. I think you would also be interested in whether as the population increases the closer proximity to other students also has the effect of increasing self-referrals.
This results in a parabolic looking regression curve. Contrary to some misperceptions linear regressions are capable of modeling this kind of curve. As long as the variables can be expressed in linear form, e.g y=a + bx +bx^2, one can add quadratic and cubic terms to the regression to model non-Linear relationships.
I would recommend starting with OLS after testing for heteroskedasticity and the other assumptions of linear regression. If this fails to generate significant results you could try more complex techniques like non-linear regression or Generalized Linear methods.
One caveat: if one student making a self-referral increases the chances that another student makes a self-referral you will need to model that with Generalized Estimating Equations. If they are independent linear regression will be sufficient.
Best,
Aran
Thank you so much, Aran and Raid. Really helpful - at least I know where to start from now!
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression