I have some thoughts on this.
Let's say the 70's movie had 4 voilent scenes, then the sample size is 1 movie with the obsevation being 4.
Same goes for the 00's movie, sample size of one. These can't be t-tested b/c you can't have a standard deviation when n=1.
Considering your null hypothesis from above you simply compare the number of violent scenes in each film, are they the same number? This seems a bit simplistic, this is b/c your null hypothesis is in terms of the sample and should be in terms of the population. i.e. the difference in violence in 70s films are the same as 00's films.
I'm not sure how to proceed from here.