Thanks to some helpful comments on my previous blog post (@BostonTea84 and matthewcarberryblog), I have done a bit more hunting and pecking on the question of gun ownership statistics.
In this post I want to look at a study matthewcarberryblog pointed me to, published in 1995 in Public Health Reports.
The study attempted to assess the validity of a household gun ownership survey question by calling up a sample of households in Ingham County, Michigan who were presumed to have a gun because they either purchased a hunting license or registered a handgun in 1990/1991. For comparison the researchers also conducted a telephone survey with a random sample of the entire county population. Surveys were conducted in November 1991 and January 1992.
Interesting findings of note, all of which point to the extent of systematic under-reporting of gun ownership in surveys, are:
(1) Systematic Non-Participation in Surveys by Gun Owners: The proportion of people who refused to be interviewed at all differed across these three samples:
- General population: 18.2%
- Handgun registrants: 23.7%
- Hunting licensees: 26.9%
So, households that were presumed to own guns were more likely than the general population to not participate in the survey at all. The authors argue that because potential respondents were not told there would be gun questions on the survey, refusal to take the survey at all cannot be taken as signaling unwillingness to answer questions about guns.
But an alternative hypothesis would be that gun owners are more distrustful of government institutions (in this case the Michigan Department of Public Health) or other outsiders asking them any questions at all.
(2) Systematic Non-Response to Gun Ownership Questions by Gun Owners: The survey asked, “Some people keep guns in their household. Are guns of any kind kept in your household?” The proportion of people who refused to answer the question about household gun ownership also differed across these three samples:
- General population: 5.7%
- Handgun registrants: 9.2%
- Hunting licensees: 7.6%
Again, households that were presumed to have guns – especially those with handguns – are more likely than the general population to not say whether they have a gun or not.
(3) Systematic Under-Reporting of Gun Ownership by Gun Owners: With the interview non-respondents and question non-respondents out of the way, there is the final question of how many people agree to be interviewed and agree to answer the gun ownership question, but do not answer the question truthfully (“false negatives”).
The authors of the study find that 89.7% of hunting licensee households answered “Yes” to the gun ownership question, and 87.3% of handgun registrant households answered “Yes.” As footnote 1 to Table 2 of the paper indicates, “Respondents who refused to answer the question were not included in this analysis.”
(*) A good beginning estimate of the under-reporting of gun ownership in this particular study, therefore, would add the “false negatives” to those who refused to answer the question, to yield the following:
- Handgun registrants: 21.9%
- Hunting licensees: 17.9%
- Combined: 19.7%
This is also a minimum estimate because it does not include the 5-9% difference between presumed gun owning households and the general population in the rate of refusal to take the survey in the first place.
A random sample of the population of Ingham County, Michigan in late 1991 and early 1992 found that 34.7% of households (+/-8%) said they had guns. That is, somewhere from 26.7% to 42.7% (at a 95% confidence interval.)
Correcting for known and probable under-reporting of gun ownership (due to refusals to take the survey, refusals to answer the gun ownership question, and “false negatives”), the likely proportion of Ingham County residence who have guns in their households is at least 20% more than the survey estimate.
So, more like 50+ percent, and possibly 60+ percent, than 35 percent.
And this was in 1991-92, when a Republican was in the White House and prior to the Brady Bill and the Assault Weapons Ban. If systematic refusals to respondent to surveys, refusals to answer particular questions, and willingness to lie on surveys has increased since then – as we have good reason to think they have – then this 20% correction may itself need to be corrected upwards.
@BostonTea84 commented on my original post:
Last but not least: the peer reviewed studies constantly show that indirect or veiled questions give a more accurate result. We already have a veiled or indirect question: Pew, Gallup and Rasmussen Reports find +60% of Americans say having a gun in the home makes the household safer from crimes and that seems to be the real gun ownership rate in 21st century America.
My initial response was that 60% is too high, but looking at this study is making me rethink that assessment.