Posted on

NUS Statistics Module Review: ST3232 Design and Analysis of Experiments

This module generally taught me how to come up with a experimental designs given some constraint conditions and also analyse the experimental results. There wasn’t much math involve, however it did get tedious in other ways towards the end of the module. Due to the nature of this module, there aren’t theorems and lemmas that one could use, most concepts are taught by examples, so when a general conceptual statement appears in the exam, it was a bit hard to judge whether it is true or false.

I took this module after ST4233 and ST3131 which helped a lot. The basics of ST3131 was covered in chapter 1, which might not have been enough for someone without strong intuition with regression models.

There weren’t tutorials in this module, the 5 sets of homework were essentially tutorials and prof always goes through the majority of it before the submission deadline, so this 20% was a giveaway.

The midterm took most people aback I suppose, the questions were not trivial for those without a strong intuition, so even though it is open-book, the mean score was 18/30. I couldn’t really finish the paper either, but I made some guesses and got 25/30. I wasn’t well-verse in calculating the estimators and ANOVA table by hand for two-way layout experiments so I ran out of time. I was then expecting the finals to be somewhat easier due to COVID-19 and the poor midterm performance, but prof Yu made some questions difficult to do by hand and was expecting us to do some programming since we now have access to a laptop. For example there was a 2^{k-p} design (can’t rmb what’s k and p exactly but quite large), the generators were given and we were expected to compute the defining contrast subgroup etc, not difficult but tedious af.

Head over to our Shop for more module content!