The (in)accuracy of gun ownership surveys
It's been widely assumed that telephone surveys of gun owners understate the real numbers, because some proportion of them are reluctant to disclose ownership to a stranger. I recently found two pieces of research on the question.
Arthur L. Kellerman, et al., Validating Survey Responses to Questions About Gun Ownership Among Owners of Registered Handguns, 131 J. OF EPIDEMOLOGY 1084 (1990)... yes, that Kellermann. They surveyed 35 households in Seattle and Memphis (I wasn't aware that either had registration) who had a handgun registered to them. 31 said that they did have a gun, 1 denied ever having had one, and 3 said they'd had one, but didn't now (which the study counts as a valid answer, but I'd classify it as fishy, at least). So 3% gave an incorrect answer and 9% gave a fishy one.
Ann C. Rafferty, et al., Validity of a Household Gun Question in a Telephone Survey, 110 PUBLIC HEALTH REPORTS 282 (1995). A larger sample (around 190) of households that have a registered handgun, or had a hunting license. 13% of the first and 10% of the second denied owning any type of gun.
These suggest that telephone surveys result in numbers that are significantly too low, as by an eighth to a tenth, even when the ownership is completely legal.