I gave an educational intervention to Spanish speaking and English speaking students. I gave a pre and post test to these students. Due to certain issues I was not able to identify students with their scores. Below is my descriptive data. Each test had 16 questions possible. E.g., median score for Spanish speakers pre was 8 questions correct.
I would like to compare (spanish pre vs. spanish post), (english pre vs. english post), (spanish pre vs. english pre), and (spanish post vs. english post).
The data is not normal using Shapiro Wilk test but n > 30. The homogeneity of variance assumption is met as I used sdtest in Stata and all group combinations don't reject null hypothesis of homogeneity of variance.
I would like to use a Two-Factor Analysis of Variance for Independent Measures. But don't know if my sample size differences make this improper!!??