Measuring the negative experience of prison

So it has come to my attention that some fairly rigorous and pretty well thought out academic research believes that prison reduces crime -- that incapacitation overwhelms the school of crime and job-market displacement arguments. I would be curious to know whether the crime rate stats they use include crime in prison, but I suspect that they would still find a net decrease.
I don't think this is a complete argument for expansion of prisons or for keeping them at their current levels for two reasons -- first, it is likely that similar crime reductions could be achieved by substituting police force expansion. Second, I think it is necessary to count the subjective experience of prison against its net utility -- for philosophical reasons beyond the scope of this post, I don't think criminal activity forfeits the right to have your experience considered by policy-makers.
The question then becomes how to calibrate the experience of prison against the experience of crime victimization. I doubt that most people would trade even a day in prison for the ability to avoid most property crime, residential burglaries (maybe) and large-scale-life-saving-wiping fraud excepted. Nor, likely, would they trade the absence of most petty-assaults for more than a few days of prison, maiming assaults excepted. Sexual assaults, maiming assaults, some quantity of drug offenses equivalent to a long-term enslvaing addiction, and of course homicide are all likely to be worth prison.
But how much? In principle, it doesn't seem to difficult to test that question -- simply ask people how much time they would be willing to do to avoid each of these crimes. Three design problems occur to me though.
First, much of the negative subjective experience of prison probably stems from a fear of victimization. This creates the problem of deciding whether to subtract the actual number of prison assaults from the total crime reduction that occurs as a consequence of incapacitation. If you do subtract it, you run the risk of double counting them -- once by subtracting it from the overall crime reduction, and once because people will consider it as part of the negative subjective experience of prison. You could tell them to assume that they won't be victimized, but then you may be neutralizing not only the experience of being victimized in prison, but the fear thereof, which is probably even exceeds the actual experience of being victimized as the reason it sucks to be in prison. You could tell people to assume that they won't be victimized, but also to assume that they will be afraid of it, but that is not only complicated but might also be distorting the results by highlighting one side of the equation. But if you don't subtract the total rate of victimization in prison from the overall crime reduction, you run the risk that people will not appraise this risk accurately. You could tell them the actual rate of victimization, but much of the negative subjective experience of prison might arise from an overestimation of the risk of being victimized. On the other, they might not accurately guage the negative psychological experience of being victimized outside of prison either, if it has never happened to them.
Second, there is the question of whether the person taking the survey will assume their own innocence when answering the survey. I have no doubt that the experience of being imprisoned is worse when you are innocent, but if you tell people to assume their own guilt, they might tell you that they see no downside to imprisonment because they deserve it, which I doubt replicates the experience of the typical prisoner.
One approach would be to do the study twice, resolving all the design issues in favor of a pro-prison bias one time and in favor an anti-prison bias the second time, and then see if the results ultimately counsel for or against expansion of prison either way.

So, go ahead and get right on that.

h/t commenter JasonR on Ezra Klein.