weil-hemenway '92 GUN "STORAGE STUDY"

rev. 11/10/2k







This study was reported in the June 10, 1992 issue of the journal of the American medical association (jAma).  The title of the article was "Loaded Guns in the Home:  Analysis of a National Random Survey of Gun Owners."  The authors (w-h) directed the study, which was a statistical analysis of data from a national (except Alaska and Hawaii, as usual) telephone survey that had been done in December, 1989."

Presenting background supporting the need for their study, w-h introduced the idea that a substantial portion of firearm deaths might be prevented through "appropriate storage" of guns and ammunition.  They noted that a 1991 Gallup national survey found that 46% of households [said they] had a firearm, one quarter of the households [said they] had a handgun, AND "over half of the gun owners surveyed said that their guns were currently [at the time they were asked] loaded."  They further said that this was consistent with results of a smaller study "in which 55% of gun owners questioned claimed they always kept their weapons loaded" and that "10% said that their guns were [at the time they were asked] 'loaded, unlocked and within reach of children.'"  Now, did this mean the guns were in locations where kids could reach them if they were there, or did it mean the guns were less than 8 feet above the floor?

A person told that two surveys were "consistent" would normally think that this meant that one survey acquired some data equivalent to what the other acquired, and that the equivalent information agreed between the two.  But the data w-h mentioned about the two surveys only include one common element.  The statement about one survey was about "households had firearm," "households had handgun," and "guns now loaded."  The other was about "individuals always kept guns loaded," and "guns now loaded, unlocked AND within reach of children."  Of all this, the only data elements that are comparable are the ones about owners having guns loaded.

One of the surveys reportedly found that over half (and presumably a little over half or it would have been said differently) of owners had their guns loaded at the time.  A much smaller portion (much less than 50%) of the owners would keep their guns loaded all the time.  But the other study reportedly found that 55% (again, a little over half) of the owners kept their guns loaded all the time.  So, the only things mentioned about the two studies that were comparable were not consistent with each other.

After the incorrect assertion that the results of two studies were consistent, w-h noted that virtually all experts say the safe handling of firearms requires that such firearms be stored in a locked area separate from the ammunition.


The survey that acquired the data w-h analyzed started off contacting people randomly by telephone.  But much of the randomness was lost by the fact that approximately 1/3 of the people contacted said they had no guns or refused to participate when asked the initial screening question which, although not specifically stated in the report, was apparently a question as to whether or not the individual or household had any firearms.  And, the surveyors only asked questions of gun owners over 17 years old.

One of the questions asked was, according to the article, "do you sometimes keep your gun loaded, always keep it loaded, or never keep it loaded?".  The entire study was to relate the answers of various other questions to the answers to this one.  From the standpoint of acquiring meaningful, unambiguous information, the word "keep" is a particularly bad word to use in a survey.  The word has a host of meanings, so use of it essentially guarantees that the respondents will not all understand the question in a consistent way, or that the surveyors or anyone else looking at the data from the survey will have a consistent understanding of the significance of the questions or the respondents' answers.  See the explanation of the problem of using the various forms of the word "keep."

So the true meanings of whatever information was collected regarding this question, and of any supposed relationships found by w-h between the answers to this question and the other questions is unfathomable.  But, it's instructive to examine the rest of their study.

Various other kinds of data had been collected in the survey, then included in the analysis by w-h.  The surveyors asked (1)if the person owns a handgun, (2)why the person owns the handgun, (3)if there were children in the home, (4)the source of any firearms training the owner received, (5)their race and age group, (6)if the gun was locked up at the time (even though there may have been more than one), (7)number of guns owned, (8)where the gun was stored (even though there may have been more than one), and (9)the person's education level.


w-h reported on relationships between several of these variables and a variable about the "kept loaded."  The ownership purpose variable was treated as "for crime protection" vs. "other."  The training variable was treated as "military or class" vs. "other or none."  The age variable was treated as "under 35" vs. "35 and over."  The number of guns variable was treated as "under 3" vs. "3 and over."  Apparently, they dealt with storage practice as "lock all the guns up when not in use" vs. "not" (they didn't actually state the question asked the respondents).  Storage place was "bedroom" vs. "other."  Education was "highschool or less" vs. "over highschool."

Now in the results w-h reported, they did not relate these variables to "always," "sometimes," and "never" keep your gun loaded.  Instead, they combined the "sometimes" and "always" loaded so that the results were about "never kept loaded" vs. everything else.  Had the survey respondents actually understood the question about "keep loaded" and answered it correctly, there wouldn't have been any answers in that "never keep loaded" category, so the relationships w-h "found" and reported are worthless.

But, examination of their results is instructive.  According to w-h:

The surveyors had collected data from 605 respondents (Rs) who owned one or more guns and agreed to participate in the survey.  75% of them were men and only 12% said they were nonwhite.  77% owned more than one gun, 72% at least one rifle, and 61% at least one handgun.

Per w-h again, 2/3 of the Rs never kept their guns loaded, 25% did all the time, and only 12% did some of the time.  Any gun owners reading this right now may be surprised that so many gun owners have those guns just for display since they never load them.  Obviously, that proportion of gun owners do not always keep all their guns unloaded.  The w-h claim just illustrates the sloppiness with which the surveyors designed their questions, the Rs answered them, w-h interpreted or represented them, or all of these.  No doubt, most of the Rs assumed that the question was about storage even if it wasn't so stated.  The question is, what proportion of them guessed right?

Now remember that one of the two studies w-h mentioned as background indicated that much less than 50% of gun owners kept their guns loaded all the time and the other found that 55% did.  So the first could be considered consistent with w-h's finding, but the second is about 2.2 times as large as w-h found.  Such large disagreement means that results of at least one of the two studies has to be way wrong.

w-h reported that their analysis indicated that those Rs who always or sometimes "kept" their gun loaded tended to be those who owned a handgun, those who owned a gun for protection, and those who had no children in the home.  So, they found that most handgun owners are being good, although w-h certainly did not focus on this in their article.  There was little correlation with claimed training received.

Note, however, that w-h reported about those who did this or that although the info they had was actually about what people said they did.  It is possible that some of the people represented in the data didn't even actually own a gun.  There is good reason for people to falsely claim gun possession and that the gun is loaded and accessible.


In the remainder of this explanation, we put the findings in terms of what the respondents said even though what w-h reported in their article was invariable not in such terms.

"Other things being equal":  (1)handgun owners were 2.17 times as likely as others to keep a gun loaded at least some of the time; (2)those who owned guns for protection were 1.65 times as likely as others to do so; (3)those with no children in the home were 1.43 times as likely as others to do so; (4)those in "the south" were 1.61 times as likely as others (Alaska and Hawaii weren't covered by the survey) to do so; and, (5)nonwhites tended a bit more than whites to do so.

There were only very weak correlations with:  not locking for storage (when not in use); being in one's thirties; owning more than three guns; and having more than highschool education.  There was virtually no correlation with any other age group, sex, where guns were stored, having more than a bachelors degree, OR having had any kind of gun safety training.

w-h found that including income level in the analysis didn't affect anything, and that totally dropping the data for those who said that they sometimes kept their guns loaded also did not affect anything.

People who know something about guns, their use, and gun safety know that people who have children and use guns for home defense might tend to tighten up the control on their guns at times when a child is around, and loosen it when the children are all out of the house.  That is, they would tend to have a gun loaded and unlocked when no child was around but then unload it, lock it, or both when a child comes home or even when a child is expected.  This means that there should be a correlation between having children and answering "I sometimes keep a gun loaded.

But w-h didn't report anything about this possibility.  This might be excusable on grounds that they maybe just didn't think about it or didn't know enough to think about it.  However, in statistical analysis, it is good practice and usual for the analyst(s) to check all their variables for correlations with each other and, if they observe any significant correlation, to investigate further and look for explanations.  So, it is possible that someone noticed something but just didn't want to let anyone know something that would not support their agenda.  On the other hand, it's likely that there was no pattern discernible in the data because of the question ambiguity caused by using the word "keep."  From the info w-h provided, it is not possible to separate out the "sometimes loaded" numbers for child-present cases, so it isn't possible to check that correlation ourselves.


A great deal of w-h's paper was devoted to "comment."  First, they noted that although "protection" is the most common reason given for owning a handgun, the number of people killed through home defense is only a small part of firearm deaths in the home (as though killing were the point of self defense) and had the usual citation of Kellermann & Reay's 43:1 "study."  They then noted that their study indicates that handgun owners and those who own guns for defense are more likely to "keep" their guns loaded, whatever that means.

They then noted that, although those owners with children were less likely to keep their guns loaded, "it appears that many families with children do keep loaded guns in the home (30%) even though experts, such as the American Academy of Pediatrics, warn of the danger." [pediatricians "experts" on gun safety?]

Notice again that the survey collected no information about people having guns loaded or unlocked around the home under circumstances in which the people did not have control over access to the guns or in which a child would be able to gain access to such guns.  Yet, w-h are taking the leap that they have found something dangerous.  And, one must wonder if the gun safety "expert" pediatricians have the sense to understand the problem with talking about sometimes "keeping" guns loaded or unlocked or the difference between keeping guns and storing them or keeping them some way when the guns are not under their personal control.  It's a safe bet that the answer is, "no."

W-h then noted that a '91 GAO study concluded that there was a need for "proper education in the use and handling of firearms."  They said then that "the results of this [their own] analysis indicate that training,.... is not associated with safe gun- handling procedures" although their analysis didn't even look as safe handling procedures.  It's obvious that w-h wouldn't recognize safe gun handling if they saw it (or much of the unsafe gun handling).

They wonder how this can be, and suggest that maybe "important safety information is omitted from the training," or the "information is insufficiently emphasized."

An honest person not overly confident of his own knowledge or understanding, faced with seemingly strange information, would think, "maybe there's something wrong with my understanding."  Not w-h.  They think only that there must be something wrong with someone else.

W-h concluded that "Our findings show that a significant proportion of gun owners disregard basic safety procedures," although they had not even looked at safety procedures or even at storage in a reliable way.  They then reported the proportions of gun owners that kept guns loaded or kept them unlocked and that owners of guns "for protection" were the most likely to keep guns loaded.  Note that these conclusions (which is all that many people look at in a paper) did not mention that the "kept" and "keep" were not restricted to circumstances in which the gun was not in use or under complete control of the owner.  They also did not mention that the statements were about claims, not necessarily about actual practices, or that the "kept loaded" was actually "sometimes or always kept loaded under unspecified circumstances."

W-h then mentioned that, "a significant number of gun owners with children reported that their gun(s) were neither unloaded nor locked up" (as though they should have been).  This is the only place in their paper, outside of the study method description, where w-h said that what they were talking about was what someone reported, even though even this statement didn't acknowledge that they didn't even know that the "gun owners" really were gun owners.

They finally concluded "instruction in the proper handling of firarms was not associated with whether or not a gun was kept loaded when not in use."  Again, the part that is the only part some people read said, "when not in use" although neither the study nor the basis survey addressed whether or not at any time a gun was or was not in use!

Finally, the fine print:  The study was paid for by.... one guess.  You and I via the doctors or clerks of the Federal Centers for Disease Control (CDC) even though the researchers don't have to provide their data to anyone, even the government.


So, what did the study actually prove?  Some things about medical gun control advocates.  About having or leaving guns loaded and unlocked so that children could have unsupervised access to them? .... Absolutely nothing, since the basis survey didn't ask anything about such things.  Value of what we paid for with no choice over the matter?... ZIP.