Hello everyone,
I have 2 samples distributions (approx 500 values per condition) that I would like to compare using the Kolmogorov smirnov test on SPSS.
I have been able to import my table in SPSS, but unfortunately since I'm not confident with SPSS I haven't been able to perform the KS test.
Does someone have any advice on how I could perform this test ?
Many thanks
Laura
This attached page gives a pretty straightforward path to using the Kolmogorov-Smirnov and Shapiro-Wilk tests for normality within the SPSS program. Since I would be just rehashing what is written on the site, I will just attach the link.
https://statistics.laerd.com/spss-tutorials/testing-for-normality-using-spss-statistics.php
If this proves to be problematic please respond back and I will take a closer look at any specific issue you may be experiencing.
Logan Netzer
Hi Logan,
Many thanks, unfortunately I have already read this article but I'm looking to compare 2 samples distributions with each other instead of testing their normality.
I did found many tutorials on how using the KS test for 1 sample but I'm still looking for a procedure that could allow me to use the Kolmogorov smirnov test on 2 samples. I'm usually using this online KS test (http://www.physics.csbsju.edu/stats/KS-test.n.plot_form.html), but my 2 datasets are exceeding the maximum number of values per condition to be handled online.
Alright, so you have two samples, did you try to combine thing into one sample with a flag variable that separates them? What I am saying is make both samples one variable, then create a second variable coded as 0 for the first sample and then 1 for the second sample and then add the two distributions combined into one variable and have them grouped by the flag variable. I don't have SPSS anymore so I would just do it for you. I have altered your excel file to show what I mean.
So for the explore option, you would put the populations variable (the combination of the two variables in the dependent list and then use the flag variable (coded as 0 or 1) in the factor list. Have you tried this?
If this is still a problem you could use R if you are familiar. And if not I would be willing to run it for you.
Logan Netzer
so using the shapiro-wilks test for normality the first population could be considered normally distributed but the second population would not.
Dear Logan,
Thanks a lot for your help ! That is working perfectly fine for me now, I knew that the distributions were different I was just missing a p value. But with your help manage to run the test on SPSS.
Many thanks !
Laura
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS