Repeated Measures ANOVA - getting very small p-value when there is no difference between conditions

Question

Prabhjot Dhami 2022-2-3

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1641765-repeated-measures-anova-getting-very-small-p-value-when-there-is-no-difference-between-conditions

编辑： Prabhjot 2023-10-5

rmanovasmallpvalue.mat

Greetings,

I am having trouble with conducting a repeated measures ANOVA in matlab. I have one group of 31 participants, who underwent 6 different testing conditions. Please find the data table attached.

% fitting anova
withinDesign = table([1 2 3 4 5 6]','VariableNames',{'SiteOfStimulation'});
withinDesign.SiteOfStimulation = categorical(withinDesign.SiteOfStimulation);
rm = fitrm(t, 'curlpfc_timei,curlmc_timei,curlipl_timei,currpfc_timei,currmc_timei,curripl_timei ~ 1', 'WithinDesign', withinDesign); 
ranovatbl = ranova(rm, 'WithinModel','SiteOfStimulation'); 
tbl = multcompare(rm, 'SiteOfStimulation', 'ComparisonType', 'bonferroni');

Now, there are clearly no differences in the condition by plotting them, as well as looking at the multcompare results. However, when fitting the ranova, I am getting extremely large F-values and small p-values, when I was expecting the opposite.

Any insight into what I am doing wrong would be greatly appreciated.

Thank you.

2 个评论
显示无隐藏无

Maria 2023-9-29

Same issue here - for one of my variables it is the exact same for all 3 time points and yet it's giving me a very significant p value

Prabhjot 2023-10-5

编辑：Prabhjot 2023-10-5

Out of curiousity after seeing your post, I downloaded my old .mat above, ran the same code, and the ranovatbl is producing very small F-values and a p-value of 1 (as was expected originally). Unfortunately I am not sure what changed, outside of me currently using Matlab version 2022b. It may be worth checking your results with a different version of Matlab to see if the issue persists.

请先登录，再进行评论。

请先登录，再回答此问题。