I hope someone can help and that this is not a simple repeat of other questions on here - none seem to quite match what I'm looking for. (Or, I have become confused by terminology...)

THE PROBLEM I have conducted OLS regression for the yearly mean values of a measured variable vs time (year), from 13 different locations. So, I have 13 different regression equations, where the independent variable, year, is the same for each, but the dependent variable (mean of the measured variable) differs. The slope of some of the lines varies qualitatively, but I want to be able to support/reject this with a statistical test. I am ultimately trying to demonstrate that the rate of change over time of my measured variable depends on location (which can be explained by physical reasons). What test can/should I do?

I have looked into ANCOVA, but I'm not sure that this is what I want, as (unless I am confused) this seems to be a test for equality of y-intercept assuming the slopes of the lines are equal?

  • 1
    $\begingroup$ Hi, it seems to me that you could use just one multiple regression model with both variables as predictors. In such model if the interaction between time and location is significant then you can reject the hypothesis that the slope is identical in all locations. Don't know about python, but in R such model would correspond to the formula value ~ time * location. $\endgroup$ – matteo Jan 11 '18 at 14:33
  • $\begingroup$ Thanks for your input Matteo. I hadn't considered that location was actually a variable as well - I guess I do not fully understand the concept of multiple regression! I'll follow this route. Thanks again $\endgroup$ – Ian Ashpole Jan 11 '18 at 15:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.