I’ve seen a good deal said about this survey, and in particular about the view of 95.7% of the 15,000 respondents that “a federal ban on manufacture and sale of ammunition magazines that hold more than 10 rounds would [not] reduce violent crime.” But while I’m inclined to agree with the policy views of the respondents, I don’t think one can rely on the survey as an accurate gauge of police officer sentiment.
The survey states,
More than 15,000 officers completed the survey, which was promoted by PoliceOne exclusively to its 400,000 registered members, comprised of verified law enforcement professionals. Only current, former or retired law enforcement personnel were eligible to participate in the survey. The survey sample size was broadly distributed by geography and rank in proportion to the U.S. law enforcement community at large.
Thus, the survey measures the view of the 4% of PoliceOne members who chose to answer the survey. There’s no reason to think that these views reflect the views of all PoliceOne members. If the survey selected a random sample of members, then one could infer from that random sample (even a sample as small as 1,000 respondents) to the views of the group at large, subject to a statistical margin of error. If the survey selected a random sample of members and, say, 60% of them responded, one could still infer pretty well what the group as a whole thinks.
But when only 4% of the people who are asked respond, that gives you the views of those 4%, and in a way that tells you little about the views of the remaining 96%. (Indeed, as I understand it, social scientists are generally extremely skeptical of any surveys with that low a response rate.) Even if the 4% are representative of PoliceOne members geographically and by rank, there’s no reason to think that they are representative of the members by viewpoint.
It might be, for instance, that pro-gun-rights police officers are more energized and are more interested in responding. And in the Internet age, it’s possible that there were online “get out the vote” campaigns (I’ve often seen plenty of those, for instance, when a newspaper runs an online poll of readers on gun-related topics). If when the survey was sent out, people posted about it on blogs and discussion lists frequented by pro-gun-rights police officers, that would make it likely that the response would be disproportionately pro-gun-rights. We don’t know, of course — but we certainly have no reason to think that the 4% who chose to respond have the same views as the 96% who didn’t.
As I mentioned, I have no quarrel with the views of those PoliceOne members who responded. I just see no basis for assuming that police officers generally, or even PoliceOne members generally, share those views.