I have 20 subjects, all subjects received 2 interventions, and i measured outcome 5 times for each intervention, thus, each subject i will have 10 measurements.
Based on qqplot and shapiro-wilk test, i don't think parametric test would be a safe option. I am not so sure which non-parametric test is a good choice, so i decide to go for most general one: linear regression.
In this case, what are my predictor/input variable, and output variable? I assume all measurements for intervention 1 is the output, and all my measurements for intervention 2 is one of the input variable, what is the other input variable? index for each measurement? In addition, what is the relation between two input variables? confounder or interaction?
The Friedman test is a non-parametric statistical test developed by Milton Friedman. Similar to the parametric repeated measures ANOVA, it is used to detect differences in treatments across multiple test attempts.
https://www.researchgate.net/post/Is_there_a_non-parametric_equivalent_of_Repeated_Measures_ANOVA
here you can look at our papers with repeated measure if they give you any idea:
Hello,
Your data should be analysed using repeated-measure ANOVA or Friedman ANOVA (in case it does not fit the assumptions of repeated-measure ANOVA including: normality of distribution, test of sphericity, and the absence of statistical outliers).
Binnan, it would be helpful to know the context of your study. Are your interventions independent of each other? I think that make a lot of impact on what analytical methods you should consider.
thanks everyone. I think i figure out the solution. So i put all the measurement in one column, other two columns are intervention and measure #, both are index (categorical variable). i use linear regression with log transformation of outcome variable. i use R, so my model is: E (log[outcome] | intervention + measure #)
- Badges
- Science topic
- Similar topics
- Computer Science
- Research Topics