rev 4/22/09


This seems like a question for which the answer wouldn't matter much.  Why should we care?  But, sometimes things turn out to be more important than we can imagine.  As will be explained later on this topic and in others at the site and to be added to it, this matter of determining how many of us own guns turns out to be one that is vitally important, not so much because of the number itself but because of the important facts that the quest brings out.


Nobody knows precisely how many guns there are in the U.S.  The only way to know with accuracy would be for the government to perform a surprise raid on every household simultaneously and to also search all buildings and likely hiding places at the same time.

Why not just keep track of how many are sold?

The U.S. Bureau of Alcohol, Tobacco and Firearms (BATF) does keep track of guns manufactured and sold by established companies for nonmilitary sales.  It even knows how many are legally imported and exported each year.  It doesn't know how many are made by individuals each year, but the number of these would be very small compared to the annual domestic production and net legal importation.

The BATF doesn't know how many are imported or exported illegally each year or how many were brought into the country by soldiers returning from wars.  It is improbable that many guns would be illegally imported each year since it would only be guns that are illegal to sell or possess in the U.S. (since legal types are readily obtained from domestic production).  Export enforcement and the records that have to be kept on U.S. production and sales limit illegal exports to small amounts.

For the years 1980 through 1993, 16.2 million civilian guns were imported, and 4.6 million were exported, according to the BATF.  The net for import and export, then, is about 11.6 million imported.  This compares to 57.0 million production for civilian use over the same period.  Hence, the net of civilian guns made available for use is about 68.6 million guns over a 14 year period, for an average of 4.9 million guns per year.

Nobody knows how many guns wear out and are discarded/destroyed each year, mostly by police agencies, but it is a very small number.  Unless a gun is shot a great deal, like in regular practice or competition, it will last 50 years or more with decent care.

It also is not known how many guns are destroyed by authorities after being seized.  In 1995 there were, according to the FBI Uniform Crime Report (UCR), about one-half million crimes committed by people with guns.  Criminals typically do several crimes for each time they are caught.  Someone was arrested for closely 45 percent of the 1995 violent crimes according to the UCR.  When the criminals are caught, the weapons in their possession, and sometimes some that aren't, are seized.  A huge portion of seized guns are destroyed by the authorities.  So, maybe 100,000 guns per year are destroyed as a result of criminals being apprehended.

And, the BATF received about 85,000 requests for gun traces in 1994, which was up from preceding years because of increased nationwide commitment to having all seized guns traced.  Yet, there are probably a large portion of gun crimes for which no traces are requested.  So, the preceding estimate of 100,000 guns destroyed per year by authorities is consistent with the number of gun trace requests per year.

The number of guns destroyed by authorities each year is therefore also small compared to the 4.9 million guns per year, and would reduce the number of guns that enter the population to about 4.8 million per year.  The result is not much affected even if the number destroyed per year is assumed to be as high as one million.

If we knew how many guns were in the population at a given time, we could use the number per year to determine the approximate number at any other time.  Unfortunately, knowing the number in the population at a given time doesn't tell us how many people own guns at the time.  This is because nobody knows the average number of guns possessed by the people who have guns.


So, why not just ask people how many guns they have, or ask them if they own a gun?

Someone does ask people something like whether or not they own a gun.  Several organizations have done so for several years, with varying results.  One of the main ones is the Gallup organization.  Its survey question is not really about ownership or possession.  They ask whether or not the person has a gun in their home.  It is likely that some people who own guns would honestly answer "no" simply because they don't keep their gun(s) in the "home," but the number of responses should be relatively small since few people keep their guns in places other than their homes (and because many people would think they were being asked about ownership even if they weren't, strictly speaking).

The Bureau of Justice Statistics Sourcebook of Criminal Justice Statistics lists results of 14 Gallup surveys from 1959 through 1997.  The results were as follows, including the '95 entry from another source.  The sampling error of these surveys is generally just 1 or 2 percent (not percent of the percentages).

                                                  93  93
YEAR  59  65  68  72  75  80  83  85  89  90  91 MAR OCT  95  96  97
%YES  49  48  50  43  44  45  40  44  47  47  46  48  51  35  38  42

Note that, at several points (years), the percentage of the people saying they had a gun in the home abruptly drops several percentage points then gradually rises until the next abrupt drop.  Exactly where the abrupt drops occur cannot be seen from the data because the data does not exist for every year.  So, we can't tell if the drop apparent at 1972 occurred in 1969, 1970, 1971 or 1972.  But, it is apparent that an abrupt drop occurred sometime in that period, and that others occurred between 1980 and 1983, and between 1993 and 1995.

Why would such abrupt drops occur in view of the fact that the net insertion of guns into the population is relatively stable from year to year?  If one (1)takes the numbers of guns actually inserted into the population each year, (2)assumes that these result in some plausible proportion of new owners, and (3)uses the actual adult populations over the years, the percentage of adults found to possess guns does not exhibit such abrupt drops.  The plausible means for increasing and decreasing the supply of guns cannot explain the year-to-year changes found by the surveys!

In fact, assuming that the contribution of new guns results in one fourth to one third of that population of new gun owners (i.e., 3 to 4 new guns for each new owner) results in a percentage that remains relatively constant.  Note that the 1994 National Survey on Private Ownership (and use) of Firearms (NSPOF) conducted by Chilton Research Services (ultimately for the U.S. Department of Justice (DOJ) National Institute of Justice (NIJ)) found that there was (reported) very closely four guns per gun owner.

If the 1968 survey result of 50 percent in the preceding table is assumed to be correct, the percentage remains very near 50 percent for the remainder of the years of the table with an assumed 3 to 4 guns per owner.  This is a much more plausible long-term variation than the one given us by the Gallup (and other) surveys.


What could have caused the abrupt drops in the survey results?  In 1968 Robert Kennedy and Martin Luther King, Jr. were killed with guns and the federal government passed the gun control act of 1968.  In 1980 John Lennon was killed with a gun and, in 1981, President Reagan was shot.  In 1991 there was a massacre in a Killeen, Texas Luby's Restaurant and, in late 1993, there was an armed rampage on the Long Island Railroad.  These and similar highly publicized events caused the federal government to pass a high visibility law that, among other things, banned so-called "semiautomatic assault weapons," (guns that look like true military weapons) in 1994.

What do these have to do with the abrupt drops?  We suggest that many more people than normal, in times when gun owners were being subjected to public opinion assault, gave false answers when asked about their having guns.  The positive response rates would go up gradually after the initial reactions as the events that caused the reactions faded in the public memory.

It is interesting to note that surveys of gun ownership in Canada showed an abrupt 26 percent drop in 1991, the year their major gun control law C-17 passed.  Over the next two years the percent of households admitting gun ownership returned to the 30 percent vicinity where it stayed for four years until ('95) it again abruptly dropped 23 percent when their gun registration law (C-68) passed.  Upon passing this last law, the government immediately spoke of going for something even more extreme.  The admitted ownership rate didn't return to normal.  In '97 the admitted ownership rate dropped another 30 percent.  Canada's true gun possession rates have not actually declined anything like these admitted rates would mean.

One reason for the abrupt drops would be embarrassment.  Another would be gun owners' suspicions that some government agency might some day show up at their doors to arrest gun owners or confiscate guns.  Even though the survey might be performed by a private organization, who is to say that such organization might not be doing their work for the government, or might voluntarily provide information to a government agency, especially if the organization is doing the survey for the government or with government funding?  The fact that the survey is done by telephone does not mean that someone could not determine the address of the respondent.

If the abrupt drops result from people giving false answers in times of stress, how many false answers might there be all the time?  Some people would be embarrassed or suspicious all the time rather than only after triggering events.

In their report on their 1993 survey about defensive gun uses (DGUs), Gary Kleck and Marc Gertz point out something significant about accuracy of surveys.  Their interviewers reported that, after asking if the interviewed person had ever used a gun to defend against another person, there was sometimes a long pause followed by a suspicious sounding response like "Who wants to know?" or "Why do you want to know?" then an answer of "no."  They also found that an interviewed person would seldom admit to a DGU by a member in their household other than themselves, even though later interviewing of the other members of the household resulted in the person actually involved admitting it.

The point of these observations is that people asked a question in a survey may not answer truthfully, especially if the answer is about possibly illegal or embarrassing acts, especially if the question is posed by or for a government agency, and especially if the acts are acts of others the person cares about.  And, the bias is not necessarily always in the same direction.  For example, people asked about having a gun in the home and locking it or storing it might purposely respond in a way to make it appear that large numbers of people keep guns in a condition that would increase child deaths/injuries (in case the surveyor might be a criminal "casing" the household).

What can we know about gun prevalence in the U.S. if the surveys say it varies between 38 and 50 percent of adults although it doesn't really vary nearly this much?  This amount of variation, which appears to be event driven rather than relatively continuous, is indicative of 20% under-reporting (lying) by survey respondents.  There is no telling how much additional under-reporting occurs always.  It is very likely that the continuing error could be as much as the event-driven errors.  Estimates made at the time of a 20% under-reporting would have to be increased by 25% to compensate.

To be sure, we have not absolutely proven that a large portion of people deny their gun ownership.  Such proof would be impossible without gestapo searches of survey respondent's homes.  But the logic of the explanation and the lack of any other reasonable explanation should be proof enough for people with open minds and some sense.


The final conclusion, then, as to what portion of U.S. adults own guns is that nobody knows—although it is probable that the proportion is about one half and remains fairly stable over time.  More importantly, however, is the fact that surveys cannot be trusted as being accurate, particularly when the survey is about possibly illegal or questionable acts or some topic that is controversial or could have controversial public policy ramifications.  In other words, if the answer really matters, any answer obtained through survey is highly suspect.

Those who do surveys seem oblivious to the certainty that there will be false answers to surveys about guns or any other controversial topic.  In the 1994 NSPOF study, for example, the principals/authors concluded (based on a telephone survey for a government agency) that only 25% of adults (35 +/- 1.3 percent of households) own guns, that some other recent surveys have had similar low results, and that "conventional wisdom" (about 50% ownership) therefore "appears out of date."

Medical people who have started doing research in attempts to discredit guns are just as stupid about the concept.  The ones who recently did the "studies" that supposedly prove that a gun in the home increases the risk of homicide by 3 times and the risk of suicide by 5 times, in an attempt to claim that their survey method was valid, said "a pilot study of homes listed as the addresses of owners of registered handguns confirmed that respondents' answers to questions about gun ownership were generally valid."

In that pilot study, only 41.3% of the registered owners actually admitted to gun ownership.  The reason that the figure was so low is that a 26.7% of the owners had given "erroneous" addresses (probably to confound the bureaucrats who wanted the handguns registered) and another equal percentage refused to cooperate with the survey.

Of those who did admit to owning a handgun, there was hardly anything else they could do since it was already a matter of official record that they did own a handgun!  The difference between this situation and one in which random people are asked about gun ownership or possession would be obvious to anyone but a gun control advocate blinded by emotion.

People who talk about the surveys but don't actually do surveys are even worse.  For example, surveys in which kids say they have guns or carry guns or can get guns easily are referenced by people who say that the surveys prove that kids do have guns, carry guns or can get them.  The idea that what is actually known is only what people said is totally lost.