Dear Saidul

It depends on what kind of variable, is your reading test:

If it was a percentage of correct answers, you could transform it by mean of the function Arc sin(square root(percentage/100)) to obtain a variable more simetric and with variance homogeneus and apt to be analysed with a parametric statistic method.

So for each pupil you could transform his/her two tests, substract the first to the second and so you may obtain the diferences of the transformed variables. Then you could use and ANOVA with the factor cohort on the variable diference of transformed reading tests.

There are other options as using IRT(item response theory) to analyze the information or a non parametric method.

I would use the ANCOVA, i.e. a regression:

post = pre + groups (in that order)

ANCOVA in R:

https://www.datanovia.com/en/lessons/ancova-in-r/#check-assumptions-1

ANCOVA in SPSS:

https://www.spss-tutorials.com/spss-ancova-analysis-of-covariance/

Your pre-test and post-test data can be analyzed with repeated measures ANOVA.