|
Abstract
|
Available at: https://scholarship.law.duke.edu/lcp/vol83/iss3/12
Empirical research has struggled to reach consensus about the impact of firearms regulations on crime.1 Consider, for example, the recent research on Stand Your Ground (SYG) laws that allow a person to use lethal force in self- defense in places outside of the home without first attempting to retreat. Using repeated cross-sectional data on annual state crime rates, recent studies have examined the impact of these laws on murder and other violent crimes.2 Unfortunately, this research has been inconclusive, with some studies finding positive effects, others reporting negligible or insignificant effects, and still others concluding that SYG laws decrease violent crime.3 Lott, for example, concludes SYG laws reduce murder rates by nine percent and overall violent crime by eleven percent, while Cheng and Hoekstra find that these laws increase the murder rate by eight percent.4 As in many other areas of research on the impact of gun regulations, empirical results on SYG laws are highly variable and sensitive to minor variations in the data or the model.
The fundamental difficulty in drawing inferences on the effects of gun regulations is that the outcomes of counterfactual policies are unobservable. Data alone cannot reveal what the murder rate in a state with a SYG law would have been had the state not adopted the statute. To address this selection problem, observed crime data must be combined with assumptions to enable inferences on counterfactual outcomes. Yet, the assumptions needed to identify these counterfactual outcomes cannot be tested empirically, and different assumptions can yield different inferences.
In this setting, where the data alone cannot reveal the effect of firearms regulations on violent crime, it is tempting to impose assumptions strong enough to yield a definitive finding.5 When this happens, the effect of a firearms regulation is said to be point-identified. Researchers often recognize that these strong assumptions may have little foundation, but defend their strong assumptions as necessary to “provide answers.” However, strong assumptions may be inaccurate, yielding flawed and conflicting conclusions. We have seen this repeatedly in the empirical literature on the firearms regulations in general and SYG laws in particular.
To focus attention on the sensitivity of inferences to the underlying identifying assumptions, we make two simplifying restrictions here. First, we examine only the effects of adopting SYG laws in a single year rather than at any point in time. In particular, to simplify the analysis, we draw inferences on the effect of SYG laws on average violent crime rates from 2008–2010 for the thirteen states that adopted these statutes in 2006. By focusing on the impact of adopting a SYG law in 2006, we do not need to make assumptions about how the effect of the statute varies with time.6 Second, we do not provide measures of statistical precision (for example, standard errors or confidence intervals).7 Instead, we view the states as the population of interest, rather than as realizations from some sampling process. Thus, imprecision expressed through the width of the bounds only reflects the selection problem, not sampling variability. We do this to focus attention on the selection problem discussed above. However, even if we wanted to provide measures reflecting the ...
Language: en |