respondents may not know about household possession or use. In addition, critics of survey approaches have raised concerns about how survey data might be used to establish what would be close to a national registry of firearm possessors.

The committee is not aware of any research assessing the magnitude or impact of response errors in surveys of firearms ownership and use. Similar concerns have been expressed about other sensitive behaviors for which research evidence on misreporting may be relevant. Surveys on victimization, such as the NCVS, and on the prevalence of drug use, such as Monitoring the Future and the National Household Survey of Drug Abuse, have undergone continuing and careful research efforts to identify the sources of response error and to correct for them (see National Research Council, 2001, 2003; Harrison and Hughes, 1997). The large literature assessing the magnitude of misreporting self-reported drug use surveys, for example, reveals consistent evidence that some respondents misreport their drug use behavior and that misreporting depends on the social desirability of the drug (see National Research Council, 2001, and Harrison and Hughes, 1997, for reviews of this literature).8 Moreover, the validity rates can be affected by the data collection methodology. Surveys that can effectively ensure confidentiality and anonymity and that are conducted in noncoercive settings are thought to have relatively low misreporting rates. Despite this large body of research, very little information exists on the magnitude or trends in invalid reporting in illicit drug use surveys (National Research Council, 2001).

While there is some information on reporting errors in surveys on other sensitive topics, the relevance of this literature for understanding invalid reporting of firearms ownership and use is uncertain. In many ways, the controversy over firearms appears exceptional. There is, as noted in the introduction, hardly a more contentious issue, with the public highly polarized over the legal and research foundations for competing policy options. Furthermore, the durable nature of firearms may arguably lead some respondents to provide invalid reports because of fears about future events (e.g., a ban on certain types of guns) even if they have no concerns about the legality of past events.

Nonresponse creates a similar problem. Response rates in the GSS are between 75 and 80 percent (Smith, 1995), less than 65 percent in the Police Foundation Survey, and even lower in some of the defensive gun use

8  

These studies have been conducted largely on samples of persons who have much higher rates of drug use than the general population (e.g., arrestees). A few studies have attempted to evaluate misreporting in broad-based representative samples, but these lack direct evidence and instead make strong, unverifiable assumptions to infer validity rates (National Research Council, 2001).



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement