Study the decision, not the calculation
The recurring exam question is: here is a situation, what do you do? That is a classification problem — what kind of data, how many groups, paired or independent, what distribution — and it is trained by seeing many different situations, mixed up, which is precisely interleaving.
Revising chapter by chapter, where every question in the t-test chapter is a t-test, trains none of it. The chapter told you the answer, so you never made the decision that the exam is entirely about.
Assumptions are where the marks live
- What does this test assume? Normality, independence, equal variance, a specific measurement scale.
- How would I know if the assumption failed? This is the question examiners love and students never prepare for.
- What do I do instead? The non-parametric alternative, and why it costs you power.
And the interpretation, which everyone gets wrong
A p-value is not the probability the hypothesis is true. "Not significant" is not "no effect". A significant result from a huge sample can be trivially small and useless. These misinterpretations are so common that examiners write questions specifically to catch them — and they catch almost everyone, because students learn to compute and never learn to *say what it means*.