Because, in actual fact, they did not. Kleck & Gertz applied three criteria for counting a claimed DGU as such, to wit:
- the respondent had to have actually seen the assailant;
- the respondent had to be able to articulate what offense the assailant was attempting to commit; and
- the respondent needed, at the very least, to have displayed a firearm or made verbal reference to the fact he had it on him.
By way of example, there were two responses in the survey in which the respondent claimed to have had a DGU.
In the first, the respondent heard a scratching/rummaging noise at the rec room window in the night. He shouted "I've got a gun" and the noise stopped.
In the second, a motorist pulled up to a stop sign and stated a pedestrian started to walk up to his car, whereupon he drew a handgun and pointed it at the pedestrian, who fled.
These examples have been used to supposedly show the low credibility of DGU claims accepted into the study, but what such claimants curiously fail to mention is these were examples of
claimed DGUs that were discarded by K&G because they failed to meet the criteria. In the first, the respondent didn't actually see any assailant, and in the second, the respondent failed to describe what criminal offense the pedestrian was supposedly trying to commit. If anything, these examples demonstrate the soundness of the K&G study in that the respondents claimed DGUs without realizing their descriptions did not meet the criteria to be counted as valid DGUs.
So what you're saying is that you believe that K&G's survey was accurate, +/- ONE MILLION or so. Inneresting.
Not quite. Any survey based on a sample of the population is going to have a margin of error, and K&G acknowledge theirs must have, just like any other. But by that same token,
so did Cook & Ludwig's, and because their survey used a smaller sample than Kleck & Gertz's, the margin of error is concomitantly wider, to the extent that the upper end of their estimated range comfortably overlaps with the lower end of Kleck & Gertz's, at around 2.1 million.
But it's revealing how you assume that
only K&G could be inaccurate.
(At this juncture, I will reiterate a caveat I've expressed before on this forum, which is that both the K&G and the C&L studies were performed in the early to mid-1990s, when violent crime levels were about double what they are now, and it is accordingly prudent to assume that DGU numbers will have dropped concomitantly in the interim.)