Naming, Premature Optimization, Logic and Set Theory
While others point to Big-O and statistics and you see these featured prominently in interviews, this misses the point that they are frequently not that relevant to many QA automation and testing activities and work outside of performance testing.
Many organizations look to Big-O and statistics because they (application programmers) are considered first class citizens in application development and they consider these to be most important. Hence we frequently have QA interview tests on sort routines which is even further from the actual skills needed to write good automation and perform most testing in the QA landscape. Even for application developers a more modern mindset is to focus on naming and avoiding premature optimizations for essentially the same reasons.
Naming: I have spent my entire career working on good names. It is really hard
Premature Optimization: Code is for humans and when made harder to read for them in order to be more performant and efficient for the computer (at the expense of the humans) there are many issues.
Program logic (for reading application code) and set theory (helps with databases) is usually more useful in my experience.
Big-O over-focus: I've worked on code at several companies where the essential hiring test was Big-O. The codebases at these companies were a nightmare to work on.