A recent study: https://psyarxiv.com/wsdu8/
This has the average Big Five scores of those who "like" 846 popular movies on IMDB.com, then predictions for a further 9067 movies.
The actual data for those movies is in the Supplemental Materials: 1) the 846 movies are in the Data.csv file within the We are What we Watch Data & Code.zip file, which includes a lot of other data describing the movie (for example, the gender balance of the ratings, the age restriction (on a 1-4 scale), the runtime, and the IMDB ratings), 2) the further 9067 movies are in the New_predictions .csv file.
I've uploaded a composite of all these scores (for a total of 9913 movies) - it'll only be there for about 30 days, though. (I intend to put a summary here sometime). I scored each movie to give them a rank for crudely approximated MBTI types.
https://ufile.io/o2wqgrlq
Columns VX to WM has the relative rank each film has in terms of being (disproportionately?) "liked" by a type (it isn't a measure of how well regarded each film actually is (terrible films can be disproportionately liked by anybody). (Columns VH to VW are the raw scores, in case you actually care about an exact ordering).
WP is "Emotional Stability", which you may or may not find useful.
WR to WU (Title, Gender (F/M), agerestrictnum, runtime, IMDB rating) just includes data far from the left that I found most interesting "Gender (F/M), agerestrictnum, runtime, IMDB rating" is only available for those 846 movies.
The rankings I'm doing in this thread are made by multiplying four columns together, which may be quite crude, but basically works with the idea that movies further away from a personality type are disproportionately penalized.