A suitably large sample size is needed, but after a certain amount of samples, any added one will not provide much more security about the statistic. I think usually to be considered empirically important you try to get 100-300 samples. The rule of thumb for accuracy of a given statistic is 1/n, where n is the amount of samples.
I think the weird thing is to judge all media along a one dimensional scale. All possible opinions about a movie - including how hilarious it is to give a wrong answer - compressed down to one score is frankly not the same ballgame movie critics should even be playing.
You want to know general sentiment from audiences, sure. That tells you if the movie has broad appeal (or at least cultural cachet). But maybe you're also interested in whether it attempted to do anything interesting artistically, how well it succeeded, how it compares to others with a similar vision, what the movie managed to say in the conversation of film as an art form - or at least to whom it owed its inspiration.
That's what I go to critics for. I couldn't give a shit what number they give it.
My wife and I sat down to watch Morbius on Netflix. I thought it was a show/series, but when it seemed to be going on a long time, asked if it was the pilot and how many episodes there were. My wife told me it was a movie. The funny thing is, when I thought it was a show, I thought it was pretty decent. When I found out it was a movie, my opinion plummeted. I might have had a similar shift for a made-for-Netflix movie vs a theatrical release.
Long way of saying, maybe it has more to do with audience expectations. I bet most people expected more from a movie in a major blockbuster series than from Morbius.
Any modern movie probably has a bot problem on Rotten Tomatoes with the audience score. In the case of Morbius, there's probably a lot of meme reviews too.
Based on 250,000 compared to less than 1,000 audience votes?
When I compare ratings, I always take note of how many people gave ratings.
It’s my understanding that sample size isn’t important to compare statistics.
A suitably large sample size is needed, but after a certain amount of samples, any added one will not provide much more security about the statistic. I think usually to be considered empirically important you try to get 100-300 samples. The rule of thumb for accuracy of a given statistic is 1/n, where n is the amount of samples.
If 500 bots are joining you better care