Also if you look at literally any study of them TAs for harder courses get harsher ratings, they get harsher ratings from weaker students.
I also spent a number of years as a TA I can say anecdotally that is my exact first hand experience - good reviews in easy courses like graphics, atrocious scores in design of concurrent systems. In fact for concurrent systems it was well known that the TA for the course would literally always get the lowest review rating in the department, and that was independent of their rating in any other courses that they TAd.
TAs also get punished if the actual instructor is bad - because students blame everyone in the course.
I mean the American system has its own slew of problems - lecturers run the tests themselves, they by design don’t provide prior exams (hint: if knowing the prior years exam questions tells you enough to be considered your exams are bad and you should feel bad)
Adjunct professors are not tenure track, so there is really no expectation that they would get tenure. Though I have seen the numbers that say the number of adjuncts is increasing.
One of them just plays videos of himself in his lectures. He doesn't actually go to them. He is actually around during less than half of his scheduled office hours. He complains to (and at least once has shouted at) the TAs about how students keep asking him questions. He has tenure.
You have to attend every lecture (I did anyway), redirect / fix during discussion / lab, help them learn what they need to know to pass the exams, and if you have time, what they should remember when the term is over for their future academic / work careers.
Only bad ratings I recall were from students who only attended the last discussion section; it's hard to teach 10 weeks of theoretical CS in 50 minutes.
You think that's bad, try being instructor of record for a remedial math course as a TA.
Do they reuse past questions forever? That is a bad practice. Each term's exams need to be equivalent but different.
My interpretation of the way US lecturers work is that they may reuse entire exams. Especially as a lot just get exams from book publishers.
In NZ (at Canterbury at least) all exams are archived and available in the libraries, and most also as PDFs. All of them. Shelf after shelf stretching back decades. At least back when I was there - maybe it’s different now? Publishers selling exam material as part of their text books seems to be moderately recent
From the perspective of most students, the ideal professor has low grading standards, “good enough” teaching, and most importantly, a “fun” class (primarily by means of humor). Most students are there to get a degree, to get a job. The best teacher according to students is the one that best fits that model (which is generally at odds with the goals of classical academia).
Undergrad feedback I wouldn’t trust at all regardless of major, Masters is major-dependent (ce, cs are particularly screwed) and phd candidates are probably fine. Simply by looking at the why the general population is even in the major (eg ask an undergrad and the reason is probably not actual interest in the subject: parents, jobs, money, dropout-major, couldn’t decide, best grades in hs, etc). Their feedback will naturally reflect the misaligned incentives
If your interest in feedback is to make sure the teachers are actually good at teaching, anyways.
Just set a threshold - such as “scores an average of < 4/10”. If the score is below that threshold, invest a little more time and effort into getting more detailed evaluations and figuring out why the score is low and whether it’s a genuine problem with the professor vs something like their having high standards and lazy students. Then train or fire as appropriate.
And by "you", you mean "no one ever". Teacher evaluations have immediate and universal impact. There is essentially no filter and no interpretation on this. It's like tweets. Once a social signal is out-there, public, everyone is acting on the implications regardless of hypothetical mitigating factors.
A lot of people talk about this stuff in the abstract, as some hypothetical, like we'll do this and we'll have safeguards in place. The news is, this is how things have been for college teachers for a while. You get an evaluation and it has an impact and there's no mitigation and no contextualizing.
Schools by and large reward research, not teaching (https://jakeseliger.com/2010/09/26/how-universities-work-or-...), and, even if they do reward teaching, it's not clear that students on evals do more than reward grades.
Evaluations are one of the extremely few levers students actually have to pull when dealing with a terrible teacher. I'd expect that given an entire class' worth of evaluations you'd be able to strip outliers and get some genuine useful responses.
Though the worst professor I had at RPI held our grades hostage until the evaluations were in, so they're not perfect.