[Q] How to combine multiple p-values into one smaller p-value?
I want to compare an intervention and control group on knowledge. They did three different knowledge tests with different question on the same topic, with different max scores. The intervention performed better on each of these knowledge tests, but comparing groups using t-test on each knowledge test produces insignificant p-values (0.391, 0.585 and 0.193). I want to test whether intervention had a significant effect on knowledge overall, but if I just combine the three knowledge test scores into one variable, artificial variance is introduced as they had different difficulty levels and max scores. I'm using SPSS, and have both wide and long format datasets.
Chat GPT suggested combining the three p-values using the Fischer method. Is this actually a suitable method? They also suggested using a mixed linear model, but I don't understand how to operate that analysis method in SPSS, and ChatGPT hallucinates when asked for a step-by-step guide. Can anyone help me?