Hi, I am trying to do a project, which tries to answer: Is there a difference between the power that is necessary to pull the object from the wall with different surfaces and different types of glue. There are 2 different types of glue and 4 different types of wall surfaces. There are 20 observations for each type of surface and 10 of them is glue 1 and the other 10 is glue 2.

So I thought Two-way ANOVA will get the job done. However, the assumptions were not satisfied. I transformed the data(I tried ln transformation because of positive skewness). This time, although assumptions looked better it still not good enough yet.

Can I double transform data such as sqrt(ln()) (I couldn't find anything on literature yet, so I don't have high hopes for this) or should I use a different model entirely? If so what do you think I should use. I'm confused.