I'd like to know whether there is a way to test the slopes for two or multiple independent datasets (populations) are equal in SPSS. I've found the link below showing methodology comparing the slopes for two independent samples using excel. However, this method is not quite tolerate for missing values within the data. Any other suggestions or sources I could look for will be very appreciated. Thank you very much.
http://www.real-statistics.com/regression/hypothesis-testing-significance-regression-line-slope/comparing-slopes-two-independent-samples/
This is an example of testing an interaction effect in a regression. You need to merge the two subsamples into one data, along with an identifier for which set of observations come from which sample -- it will be easiest if you make that indicator a dummy variable (0/1). You can then test the difference in means (intercepts) by entering that dummy variable into the regression model.
To test the difference in slope, create a new variable that multiples the dummy variable by the independent variable of interest. The resulting regression will give you the slope of the subgroup coded 0 as the b value for the independent variable, and the interaction effect will be the difference between that group and the other one. So, if the interaction effect is significant, it indicates a difference in slopes, and adding that difference in slopes to the b value will give you the slope for the group coded as 1 in the dummy variable.
I did a google search on this topic and here is the site I liked best;
http://socserv.socsci.mcmaster.ca/jfox/Courses/SPIDA/dummy-regression-notes.pdf
Hi there. I second Davids advice. Beyond that, there is an easy formula how you can test the differences in the slopes for significance. See
Paternoster, R., Brame, R., Mazerolle, P., & Piquero, A. (1998). Using the correct statistical test for the equality of regression coefficients. Criminology, 36(4), 859–866.
Both approaches should give the same results.
Of course, both approaches require that the IV and DV are exactly the same and measured on the same scale/metric.
Best,
Holger
Hi David,
Thank you for your note. I recalled the same procedure that I've learnt from my statistic courses four years ago :-)
I used to apply Z transform before to solve the problem, even though I still couldn't figure out the difference between Z transform and the dummy regression you suggested. I guess I will apply different procedure to an exemplar dataset and then compare the difference. I will read through the pdf file you attached. Thank you very much.
I investigated this today and found an unbelievably simple answer and excellent Excel tutorial on YouTube: https://youtu.be/myL_qzuLHTQ
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS