n-way anova produces the same p-value for all groups
조회 수: 2 (최근 30일)
이전 댓글 표시
I have 6 vectors of values. These are split into groups i and m. There are 3 i groups and 2 m groups, which yields a combination of 6 total groups. Supposedly, my data should indicate that the data I have are grouped better according to 'Hm' vs 'Rm' than 'Li' vs 'Ri' vs 'Hi'. When I run the ANOVA using anovan() I get the same P-value for the i grouping, the m grouping and i*m interaction. When I plot the distributions using a boxplot they are clearly more separable along the m dimension.
All necessary files are attached. I'm using R2021a
I am running:
mdacdlcdp=[cdlcdpLiRm;cdlcdpLiHm;cdlcdpRiRm;cdlcdpRiHm;cdlcdpHiRm;cdlcdpHiHm];
p=anovan(mdacdlcdp,{igroup,mgroup},'model','interaction','varnames',{'i','m'})
But the results shown below are not consistent with my expectations.
Source Sum Sq. d.f. Mean Sq. F Prob>F
-----------------------------------------------------------------
i 0.4867 2 0.24335 2.98 0.054
m 1.2762 1 1.27619 15.62 0.054
i*m 0.2255 2 0.11276 1.38 0.054
Error 11.7664 144 0.08171
Total 13.7548 149
댓글 수: 2
Megumi Fukuda
2021년 7월 1일
I ran these codes with my computer (WIndows, MATLAB 2021a)
load('anovadata.mat')
load('igroup.mat')
load('mgroup.mat')
mdacdlcdp=[cdlcdpLiRm;cdlcdpLiHm;cdlcdpRiRm;cdlcdpRiHm;cdlcdpHiRm;cdlcdpHiHm];
p=anovan(mdacdlcdp,{igroup,mgroup},'model','interaction','varnames',{'i','m'})
and got these results. Could you confirm you run your code correctly?
Source Sum Sq. d.f. Mean Sq. F Prob>F
-----------------------------------------------------
i 0.4867 2 0.24335 2.98 0.054
m 1.2762 1 1.27619 15.62 0.0001
i*m 0.2255 2 0.11276 1.38 0.2549
Error 11.7664 144 0.08171
Total 13.7548 149
답변 (1개)
참고 항목
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!