I want to do a Repeated Measure Analysis for data for three time points. But I can't do One Way RM ANOVA as the data is non-parametric, as is elucidated by Shapiro WIlk's Test for Normaility and Skewness/Kurtosis. My question is can we do a RM analysis for a non-parametric data like this. For a regular single time point, I could have considered a Non-parametric test like Kruzkal Wallice test. I don't know how to approach this analysis. Please suggeste me a way out.
This is possible by creating three variables composed of daily data for 3 months?
I suggest looking at ANCOVA in regression form. To see how this is done see the ANCOVA chater in Montgomery, D C, Design and Analysis of Experiments, 8th ed. This gives you all of the usual tools for dealing with assumption difficulties, see:
Kutner et al: https://www.google.com/search?q=applied+linear+statistical+models&rlz=1C1CHBF_enUS915US915&oq=Applied+Linear&aqs=chrome.0.0j46j0l2j69i57j0l3.13914j0j15&sourceid=chrome&ie=UTF-8 NB: Data is never non parametric Only statistical methods are.
Best, D. Booth
The exact nonparametric equivalent of a simple one way RM ANOVA is the Friedman's ANOVA, available in SPSS.
If your variables are in seperate columns, just go to "analyze" button, scroll down to "Nonparametric" select "K related variables-analysis", migrate all the variables to be tested into the box. Select Friedman's ANOVA and click "run".
If you have a significant result, you can go further to click on "pairwise comparison" as a post hoc test to determine which pairs have a significant comparison.
I wish you good luck.
Ogochukwu Kelechi Onyesco@ thanks a lot. I got a significant value with the Friedman's ANOVA.Can you please elaborate about the post hoc analysis after that. How should I approach the problem.
Anmol Banerjee@ After the analysis you will have an output table, the column/box containing the significant p-value is coloured yellow? Just double click the table, another viewer window will open.
At the bottom right ribbon on that nee window, you will see about 3 drop-down buttons, click on the one at the extreme, it will give options containing "pairwise comparison" click on it and the comparison for all the pairs will show in a table.
It will color all significant ones in yellow. Very beautiful output, will also show you a web plot of the interactions. The test statistics, standard error, adjusted and unadjusted p-values based on Bonferroni will all appear.
You can mail me on [email protected], if you want a zoom meeting for practical explanation.
The non-parameter used in this regard (a) Test Runs b), (Test Points Turning c), (Test Sign) d), Test 'Daniels' test (E), Kendel's Test s' Kendall W. considered a strongest of these tests Test 'Daniels'
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA