I want to test the effect of a suicide prevention film on help-seeking. I have help-seeking intentions as an outcome variable and measured this variable at three time points (pre, post, follow-up) in 50 participants. To analyse the data I would conduct a one-way repeated measures ANOVA and compare mean differences between the three time points. However, in the pre condition I also assessed depressionscore, indentification with male gender (cses) and identification with the protagonist. Now I would like to consider these variables in the comparison of the means. Since my sample is small I don't want to use dichotomized factor variables and conduct a mixed ANOVA. So I would prefer to include the three variables as covariates. However, I have searched a lot of sources, and did not find valid information on Wether this can be done. Since usual ANCOVA models are applied in situations with additional treatment factors, which I don't have. So checking for assumptions is based on comparing treatment groups. Moreover I have doubts on what the analysis would tell me?
I used the r rstatix package in r. Here is the code:
res.aov <- anova_test(
data = data_csv, dv = help_seeking_privat, wid = ID,
within = time, covariate = c(suicidalityscore, cses, identification)
)
get_anova_table(res.aov)
The results I get are the following:
ANOVA Table (type II tests)
Effect DFn DFd F p p<.05 ges
1 suicidalityscore 1.00 35.00 3.950e-06 0.998 9.36e-08
2 cses 1.00 35.00 5.470e-01 0.464 1.30e-02
3 identification 1.00 35.00 2.806e+00 0.103 6.20e-02
4 time 1.56 54.55 4.351e+00 0.026 * 2.10e-02
5 suicidalityscore:time 1.56 54.55 5.480e-01 0.538 3.00e-03
6 cses:time 1.56 54.55 2.820e-01 0.700 1.00e-03
7 identification:time 1.56 54.55 3.200e-02 0.940 1.56e-04
I would conclude that there is a significant effect of time and proceede with pairwise t.tests for post hoc analysis. However the covariates did not have significant effects. did I include them properly? how do I interpret the results of the covariates? And how do I check assumptions for covariates in this model? I checked anova assumptions, but nothing else, as if the covariates did not exist in my model.
I would appreciate your help a lot!