In the aftermath of a certain incident involving a 9-year-old and an Uzi, the usual gun control extremist suspects have been shopping around the idea that firearms and children are a bad combination. As we know from our own polling, over 50% of TTAG readers started shooting before the age of 11 years old. Huffington Post seems to think that’s a terrible idea, and so produced an equally terrible poll to promote their agenda. Why is it a terrible poll? Mainly because it breaks the one golden rule about how to conduct a survey in a statistically valid manner . . .
Here’s Huffpo’s conclusion:
According to the new poll, only 26 percent of Americans think it’s appropriate for a child under 10 years old to learn how to fire any kind of gun under supervision at a shooting range, while 63 percent said that is inappropriate. Only 7 percent said that it’s acceptable for a child under 10 to fire submachine gun like the Uzi, while 86 percent said it’s inappropriate.
In fact, an earlier HuffPost/YouGov poll, also conducted after the shooting, found that the median age at which Americans think it’s appropriate for someone to learn how to fire a gun is 16. That poll found variation along party lines on the median age listed as most appropriate — Republicans said 14, independents said 15, and Democrats said 18.
That would be an interesting result — assuming that you ignore the source of the polling information.
The HuffPost/YouGov polls were conducted Aug. 28-Sept. 2 and Sept. 4-8 among 1,000 U.S. adults using a sample selected from YouGov’s opt-in online panel to match the demographics and other characteristics of the adult U.S. population. Factors considered include age, race, gender, education, employment, income, marital status, number of children, voter registration, time and location of Internet access, interest in politics, religion and church attendance.
The issue I have with their methodology is that it breaks the golden rule about conducting a survey. A good survey requires a randomized sample of people from the population in question. There was nothing randomized about their survey.
There’s an inherent bias that comes with using a self-selecting group of people in any survey. The point of the exercise is to get a randomized group of people to respond to a question, and the most effective way to do that is by cold-calling people over the phone. It’s still not ideal, but it’s better than anything else out there at the moment. By using a website and its members for responses instead of a randomized group, you no longer get an accurate sample of the population. Rather than sampling the American public, all you’re really sampling are the users of that website.
Websites tend to become an echo chamber. People will gravitate toward sites that line up with their viewpoints, and it makes sense that a poll pulled from the Huffington Post’s readership will attract users who agree with their self proclaimed crusade to eliminate American gun ownership. So while I fully believe that most of HuffPo’s users really do think kids and guns don’t mix, I do not believe for a minute that their sample is a representative slice of the American population.
According to HuffPo’s methodology, it would be equally valid in saying 100% of Americans are against an “assault weapons” ban or magazine capacity restrictions based on our own survey, especially since our sample size was four times greater. But I know better, since I passed high school statistics.