Netflix Intern Data Scientist Interview Experience — A 75-Minute Causal Inference OA That Left Me Guessing

New here, and internships are so hard to find. This is the second OA I've gotten after countless mass applications, and long story short, it's been two weeks with no response so I'm assuming it's dead.

The whole thing was about 75 minutes and had to be completed within a week. First there were two coding questions, and you could use Python or R — both coding questions were under the same setting. Note that both questions required loading a file, and at first I wasn't familiar with the platform, so when I couldn't print head(5) of the dataset or df.info(), I thought the file hadn't loaded correctly. Later I realized it seemed like it just couldn't print that specific kind of output — other things like a single numerical value such as an average printed fine. Not sure if it was a system bug, or if it was just that my internet in China was really laggy at the time, because every time something printed, it showed the result from my previous edit instead of the current one.

The first question asked you to estimate the ATE (of personalization on minutes streamed), and they gave you some irrelevant variables mixed in. The second question asked for ITT and TOT, and then there was a multiple-select question about IV/LATE and compliers. Note: the CodeSignal platform doesn't support Python's linearmodels IV2SLS, so don't waste time trying to import that package.

After the coding section it was all probability and statistics multiple-select questions — one on Bayes' rule, one on why you add controls in a regression, one on the CLT and the mean/variance of a uniform distribution, one on hypothesis testing where you had to compute the t-stat, and one on effect size compared to MDE. The questions themselves were all pretty basic, it's just that both the answer choices and the question stems were really long.

I'm not totally sure why I failed the OA — maybe I just made careless mistakes on some of the questions, ugh. Would love to hear from anyone who made it through successfully about what I might have missed.