How to preform Anova + Tukey for within-subject design

Question

Itay Coifman 2022-2-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1645300-how-to-preform-anova-tukey-for-within-subject-design

编辑： Scott MacKenzie 2022-2-9

Hey all, I am trying to write a code to analyze my experiment results:

My experiment is a within-subject design,each subject (1-5) has done the test at 8 different levels (NL_0,NL_1,TH_450...SH_1350).

The "df.csv" attached describe my results, after transforming the data for within-subject analysis I got "testDf.csv" (might not be needed at all...).

First I want to perform an ANOVA test, and afterward, I want to perform a Tukey test to compare all the levels combinations.

If it was a between-subject design i would use:

df = readtable('df.csv')
df = 40×3 table
subjecttrialNamey________________________

       1       {'NL_0'   }      13.3
       1       {'NL_1'   }    14.543
       1       {'SH_1350'}    15.326
       1       {'SH_450' }    13.926
       1       {'SH_900' }    14.999
       1       {'TH_1350'}    14.677
       1       {'TH_450' }    14.565
       1       {'TH_900' }    14.628
       2       {'NL_0'   }    13.503
       2       {'NL_1'   }    13.655
       2       {'SH_1350'}    15.159
       2       {'SH_450' }    14.014
       2       {'SH_900' }    13.967
       2       {'TH_1350'}    14.142
       2       {'TH_450' }    13.376
       2       {'TH_900' }    13.813
[p,tbl,stats] = anova1(df.y,df.trialName)
confidanceLevel = 0.05;
[c,m,h,gnames] = multcompare(stats,'Alpha',confidanceLevel);

After reading some questions at the forum, I figured that I need to use something more like this one:

testDf = readtable('testDf.csv')
testDf = 5×9 table
    Subject     NL_0      NL_1     SH_1350    SH_450    SH_900    TH_1350    TH_450    TH_900
    _______    ______    ______    _______    ______    ______    _______    ______    ______

       1         13.3    14.543    15.326     13.926    14.999    14.677     14.565    14.628
       2       13.503    13.655    15.159     14.014    13.967    14.142     13.376    13.813
       3       16.356    16.029    16.695     16.005    16.727    16.772     16.504    16.805
       4        13.98    13.563    15.107     14.232    14.627    14.188     14.381     14.09
       5        14.41    14.344     16.66     15.733    16.126    15.602     15.495    15.146
measurements = table([1 2 3 4 5 6 7 8]','VariableNames',{'Measurements'});
rm = fitrm(testDf,"NL_0,NL_1,SH_1350,SH_450,SH_900,TH_1350,TH_450,TH_900~Subject"...
    ,'WithinDesign',measurements);
ranovatbl = ranova(rm)
tbl = multcompare(rm,'NL_0')