I have been reading this paper on how to analyze linear relationships using a repeated measure ANOVA: Article Estimating linear effects in ANOVA designs: The easy way
.I was wondering though once you establish a linear relationship across your categorical variables (A, B, C, D) how can you check if the difference across conditions A vs. B vs. C vs. D is also significant?
I have been using pairwise t-tests (A vs. B; B vs. C; C vs. D), but is there a better test to look at this?
Just for completeness, I have been using "ols" from "statsmodels" to check for the linear relationship, and "pairwise_ttests" from "pingouin" to run post-hoc tests in Python.
There are contrasts (like Helmert, Deviation, Difference etc), which test several different cotrast. Maybe "Repeated" is what you are looking for, it tests A vs B, B vs C, and C vs D and nothing else. Just look it up.
Hi Giorgio,
One way of looking at the nuances of your ANOVA is with a Tukey test. Assuming you have not already installed the command in Stata, type ”findit tukeyhsd” into the ”Command“ window. Scroll down to the bottom of the search page, and the only package result will appear in blue. Click the blue link, then click the blue “click here to install” link. If you do not already have them installed, you will need to do install sg101 via the same process (findit sg101). Once you have installed sg101, you can also ”findit tkcomp” (to find and install the Tukey-Kramer test package). Once installed, first run your ANOVA and then the command is tukeyhsd [variable]. Additionally, you can set a p-value via the level ( ) option, or you can look at levels of one variable, restricted to one level of the other independent variable by the ”if” suffix. For example:
tukeyhsd condition if selfident==1
if you are familiar with SAS, obtaining the post hoc pairwise comparisons using the Tukey Kramer procedure would be so much easier and you could use the SAS code example below:
proc glm data = your dataset;
class Groupcat;
model Primcohort=Groupcat;
lsmeans Goupcat /pdiff=all cl adjust=tukey;
run; quit;
Hope this is helpful. Good luck!
Rainer Duesing I've been wondering about this question so I finally looked up the paper. This is not about contrasts of GLM estimates which is what we all thought . Article Estimating linear effects in ANOVA designs: The easy way
iIt is about cognitive tests of how people estimate distances on a mental number line not linear functions of estimates from a GLM but rather how people think about distances in 1-D. This is not statistical estimation. They did use ANOVA models to interpret their data but that's it. unless you want to know how people think about distances along a horizontal line this paper is of no interest . Apparently everybody thought in terms of the GLM when this really a cognition paper. There is no easy new way to contrast ANOVA estimates. Do it the way you suggested for whatever contrasts you have and no more. Everyone was fooled by the title. Best to all, David Booth
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA