0
$\begingroup$

I have a data set with 41 individuals, male and female, and 3 different measurements of a variable for each individual. The data is not normally distributed, nor does it have equal variances between the sexes, and simple transformations are not effective.

I want to determine if the means of the variable differ between the sexes. Is there a non-parametric test I could run in r that would be similar to a repeated measures ANOVA? I don't think Friedman would work since it is not an unreplicated block design. I know there are similar questions to this but I haven't found any answers that match my problem.

I've added a photo of an example of my data, with Sex as the first column, individual ID as the second, and the variable score as the third. I've also attached an image of the qqplot.

example of data

enter image description here

Improve this question
$\endgroup$
5
  • $\begingroup$ Please , what is r? If you mean R, that is CAPITALIZED. $\endgroup$
    – kjetil b halvorsen ♦
    Commented Nov 24, 2020 at 20:08
  • $\begingroup$ An option is repeated measures ANOVA with Brunner-Langer's approach. It is implemented in R-package nparLD. $\endgroup$
    – Michael M
    Commented Nov 24, 2020 at 20:12
  • $\begingroup$ What are the data? Is this some number correct out of some number of possible tries? $\endgroup$
    – gung - Reinstate Monica
    Commented Nov 24, 2020 at 20:19
  • $\begingroup$ It is a syllable diversity score for bird songs, so # of unique syllables/total number of syllables, scored for three songs per individual. $\endgroup$
    – michellemoyah
    Commented Nov 24, 2020 at 20:20
  • $\begingroup$ You have short tails, not long tails, maybe non-normality is not much of an issue? Maybe try a linear mixed model with random intercepts, and then bootstrap ... $\endgroup$
    – kjetil b halvorsen ♦
    Commented Nov 25, 2020 at 17:29

0

Reset to default