In Which We Begin to Dabble in Political Polling

While the soul-less capitalists in our company focus on mundane but profitable consumer research, the data wonks at CIVICSCIENCE can’t help but test our technology in the sexier political sphere. If reputable pollsters can publish “scientific” results based only on people who answer their land-line telephones, then clearly there is room for innovation.

Consider that an estimated 25% of US households no longer own a land-line telephone. Of the remaining 75%, many if not most use some form of CallerID or Do-Not-Call services to ward off unwanted calls. So, let’s just say that, conservatively, 50% of US households are accessible by land-line. We know that these households are disproportionately older, less formally educated, and middle-income. This is what opinion researchers refer to as a bias.

Now, traditional political pollsters defend their phone-based methodology because they believe it is the only way to truly randomize which people are included in their research. However, with such clear biases among the shrinking land-line population, how “random” is it, really? If you cast a fishing line into a Koi pond, you will catch one at random. But you’re still only fishing for Koi.

At CIVICSCIENCE, we’re fishing in Lake Internet. While Internet usage is not yet ubiquitous, it is growing every day among every demographic and geographic segment of the US. When we deliver questions to web users across a multitude of websites and social networks, our overall population is disproportionately thin among people over 65, certain minorities, and the least-formally educated. Internet adoption among these groups is growing — Land-line telephone usage among the young, the educated, the affluent, and EVERYBODY ELSE is not. Moreover, because we can reach huge groups of respondents (as many as 50,000 per survey in some cases) we can perform far less-egregious weighting than a phone researcher with their 500-1000-person samples.

Is our approach to opinion research “radical,” compared to traditional methods? Absolutely. Will you see the Washington Post publish our poll results anytime soon? Probably not. But while everyone else is trying to squeeze the last few drops of juice out of the phone-polling lemon, technology, demographics, and human behavior will continue to trend in our direction. In the meantime, we’ll start sharing some of our political data, soon, so you can be the judge. Purists be damned!