The answer was deleted as not an answer.
The question: Is the death penalty effective?
There are a LOT of factors that go into crime rates that confound this issue, so if you want to compare something you have to find two places that are almost the same in all other aspects, but only different in the one you want to compare. In this case you'd want two (preferably a lot more than two) places with similar population density, ethnographic breakdown, poverty rates, police force sizes, other laws, etc. where one enacted the death penalty and the other did not.
You can see how places that match in all of those criteria will be hard to find. That's why you have not found a definitive answer. There probably is not a definitive answer at this time, but people who want the answer to be one way or another use poor statistics to back give credence to emotional arguments to sway people for/against.
This just scratches the surface really. There are the other issues of data collection you mentioned, such as the different ways crimes are reported and tried in court. Nor any comparative analysis of the death penalty vs. life in prison. Nor...
While I find it interesting, I don't think it's an answer to the question. It's not a matter of references (which, of course, are also important and were missing at the time). But it's just that it doesn't really address the issue. I don't think that answering "I don't know how this could be addressed therefore there's no answer" would be acceptable in any other SE site.
Now, I did leave this comment:
I think this question should actually be answered with real data. This (old) answer is more of a comment. Dont get hooked up on the wording of the question... It's certainly possible to find valid statistics on the effectiveness of death penalty. I deleted your question as not an answer.
Which says the same thing.
Ultimately, the question was standing out as a really bad example: only one answer, by a high rep user, which was not good and up voted.
By the way, there are five deleted answers on that question. This is why I took it upon myself to provide a good answer, since basically nobody else did it and it was about time someone took care of it. Also, it turns out there are 27 papers addressing the issue, so your answer was also incorrect.
In conclusion, there was a problem and I fixed it.