Newsgroups: talk.politics.guns
From: [d--a--p] at [lsid.hp.com] (Dean Payne)
Subject: NEJM: Locked Guns Reduce Homicide Risk?
Date: Tue, 12 Oct 1993 05:56:32 GMT

Do Locked or Unloaded Guns in the Home Reduce Homicide Risk?

The local library finally set out the latest NEJM issue, containing
Kellerman's "Gun Ownership as a Risk Factor for Homicide in the Home"
(NEJM 1993;329:1084-91).  Here is my first comment.

Considering the attention that this and an earlier report gave to
unlocked and/or loaded guns in the home, it is surprising that no
mention is made about whether or not locked or unloaded guns are
associated with less risk.  However, some clues are available in Table 3.

388 matched pairs responded to almost all questions in an analysis of
many hypothesized risks.  The figures below are univariate, meaning that
other risk factors have not yet been factored out.

174 (45.4%) of the homicides cases, and 139 (35.8%) non-homicide control
subjects, had gun(s) in the home.  This gives a crude odds raw of 1.6,
with a 95% Confidence Interval (CI) of 1.2-2.2.  The CI does not overlap
1.0, meaning that homicide had a statistically significant association
with having a gun in the home.

Under "Any gun kept unloaded," 105 (29.6%) cases and 69 (17.8%) controls
answered Yes, for an odds ratio of 2.1 (1.4-3.0).  Under "Any gun kept
unloaded," 93 (26.7%) cases and 48 (12.5%) controls answered Yes, for an
odds ratio of 2.7 (1.8-4.0).

In short, unlocked and loaded guns were more strongly associated with
homicide than were 'all guns'.  The obvious question:  how much risk is
associated with locked or unloaded guns?  The study is silent on this,
so I had to do my own figuring.

For unlocked guns, 69 cases did not answer Yes, and percentages
indicated that 28 did not answer at all, leaving 41 who answered No.  No
controls failed to respond, so all 70 who did not answer Yes must have
said No.

For unloaded guns, 81 cases did not answer Yes, and 35 did not answer,
leaving 41 who answered No.  91 controls did not answer yes, and only 4
did not answer, so 87 answered No.

I interpret this to mean that, at worst, locked guns are associated with
no increased homicide risk whatsoever (69 homicide cases versus 70
non-homicide controls).  At best (41 cases vs 70 controls), locked guns
are associated with a much decreased homicide risk.

Similarly, unloaded guns in the home are, at worst, associated with an
slight risk reduction (81 vs 91), and at best, a substantial risk
reduction (41 vs 87).

Unfortunately, I don't know how to compute the odds ratios.  Comparing
to other responses, the "worst cases" would seem to show no significant
difference, while the "best cases" would seem to show significant and
substantial odds in favor of having locked or unloaded guns in the home.

In summary:  This study, though purporting to show an increased homicide
risk when guns are kept in the home, also seems shows that absolutely
all of this risk is associated homes containing unlocked and loaded
guns.  Homes containing locked or unloaded guns are associated with no
increased homicide risk whatsoever, and may even be associated with a
reduced homicide risk, compared to homes without guns.

I hope that someone with a better statistical background can take a
better look at this.

More comments later.

Dean