Ok... To use your question as an example: we've discussed and researched this particular threshold quite a bit over the years. However... Whether 50 is an effective threshold depends a lot on the base assumptions.
For example: if you consider noise-reduction a worthy goal, then limiting the ability of most participants to post comments consisting of "thanks" "me too" etc is worth some sort of threshold. If you don't consider that a worthy goal, then any threshold > 1 will be pointless.
Same applies to any of the other uses / abuses of comments
thus, these discussions tend to founder, because no amount of data will resolve a question if it is approached by participants with conflicting goals
Now... You'll note that none of that actually answers your question.
If I was to answer, I'd have to write something like that, because otherwise I would be dishonest in my response
But, if you're determined to get an answer to your literal question, you'd remain unsatisfied by such a response.
The literal answer is something like, "the threshold only matters once we've agreed on a set of valid uses for comments, at which point we can identify the reputation at which those uses completely dominate misuses"
And of course, I can present data for that... But now consider your clarification to the base question - the bit to which I said "this turns it into a loaded question"
both of those options are simultaneously true and untrue
or to put it bluntly: while neither are entirely false, both are inaccurate.
No one has bothered to change it (except on some sites where we have bothered to change it)
a study with data was conducted, but the decision wasn't made because we couldn't agree on a base set of goals
if this starts to sound like not a very useful answer... I concur.
As with most of the privilege thresholds on SO/SE, the actual numbers are fairly arbitrary; their validity is connected in how they relate to other numbers: the ability of new users to earn reputation, the values of other thresholds, etc.