Gallup's mysteriously missing 4.8 million gun owners
OK, but here is the big puzzle. Gallup has apparently redone its survey data on gun ownership rates, shifting down gun ownership rates. The shift hasn't just occurred in recent years, but goes all the way back to at least 1991 and the drop seems to be by two percentage points in each survey. This data shows that gun ownership is about the same as it was in 1983 and 1996.
The old numbers showed 47 percent owning guns in 2011. Now the new numbers is supposedly 45 percent. For 2010, it used to be 41 percent. Now it is 39 percent. The 2009 number used to be 42 percent. Now it is 40 percent. This goes all the way back. The 1991 number used to be 50 percent. Now it is claimed to be 48 percent. The dates match up exactly and the survey questions are the same, so it isn't just that these results are from different surveys.
This rewriting of past survey data is not trivial. In 2011, that change implies 4.8 million fewer adult gun owners (the census information is from here).
The puzzle gets even more bizarre when one looks at other survey data. If there was a re-weighting that Gallup thought was more appropriate, it should have impacted other surveys that they did. Yet that isn't the case. Given that the October 2011 and earlier gun survey data was changed, I decided to look at the approval rating data for the 2011.
For example, here are the survey data on whether people are generally satisfied with the total cost of healthcare in the US. The most recently reported data is available here.
As another example, take the weekly survey data for the presidential approval rates from January to November 2011. It is identical whether one looks at the recent reports of that time period or the data as it was reported in 2011.
The most recently reported data is shown here.