I am a bit desperate because I am writing my Msc thesis and I am not sure how to organize my data for an SPSS usage.
My research examines the performance of mobile banner campaigns and how the 3 independent variables: Source(in-app or mobile web promotion), Style (animated or static banner) and Size (full screen or small banner) will affect the dependent variables: Click-Through Rate((total clicks/total impressions)*100) and Conversion Rate((total installs/total clicks)*100).
Initially I was considering ANOVA tests like the ones below:
Click-Through Rate = β0 + β1*Source + β2*Style + β3*Size + β4*(Source*Style) + β5*(Source*Size) + β6*(Style*Size) + β7*(Source*Style*Size) + error Conversion Rate = β0 + β1*Source + β2*Style + β3*Size + β4*(Source*Style) + β5*(Source*Size) + β6*(Style*Size) + β7*(Source*Style*Size) + error
But my supervisor advises that I can not use percentages withing ANOVA and I should use loglinear analyses.
Can you please advice me if I can somehow still use ANOVA for my tests or if I need to use loglinear analyses and how to structure my data for SPSS use. My supervisor is mentioning I need to organize it into counts, but doesn’t give me more details and I am a bit confused: what does that mean?