Jon Delano from KDKA is a good man. He spent a lot of time over the past several months learning about our technology, scrutinizing our methods, and measuring our predictions against outcomes. So when he called us for data about Mayor Ravenstahl’s popularity, we offered them without hesitation.
Unfortunately, Jon’s story could only say so much about our methodology, as it’s not easily explained in sound bytes or “elevator pitches.“ So here is our best stab at tackling a few common questions asked about CivicScience, our data regarding Luke Ravenstahl’s popularity, and the confidence we have in our results.
#1: First and foremost, we have nothing against Mayor Ravenstahl. On the contrary, as a young company trying to recruit talent to the City, while withstanding investors trying to relocate us, he conveys a youthful and proud image to the outside world. That counts for a lot.
#2: Our data contain extensive profile characteristics about our response group, including age, gender, race, education, household income, party registration, voter likelihood, parental and marital status, community of residence, and dozens of other factors. Women and younger respondents, for example, supported the Mayor less than other groups. And, since our overall sample was disproportionately male and older, we performed some marginal weighting to reach our final numbers. With such a large sample, these were easy calculations to make.
#3: CivicScience responses are collected across a multitude of diverse online and digital channels. Respondents cannot predict which questions they will see when viewing our polls, nor can they vote more than once on a given question (without us knowing). They cannot “stuff the ballot box” or drive friends to the same poll to inflate results.
#4: As per points 2 and 3, any comparison of our technique to a listener’s poll on a radio website is misinformed. A single site, with a known ideological bias, where no demographic traits of the respondents are known, where people can mobilize their supporters to vote, as many times as they want…well, we don’t even know what else to say.
#5: Why 27,000 respondents? The way our software engine works, we deploy hundreds of questions across our network of polling applications on a sort of timer. The question regarding Mayor Ravenstahl’s popularity was calibrated to collect approximately 1,000 responses every week for 6 months (26 weeks and change). We can take slices of those responses at any point in time to identify trends or mine them for correlations in aggregate. As Jon Delano’s story pointed out, the Mayor’s numbers were remarkably consistent during the 6-month period.
#6: 78% of the US population uses the Internet. 74% have land line telephones. Those numbers are moving in clearly different directions.
#7: Yes, our CEO is an arrogant ass. But he believes very deeply in what we do and in the team that does it.
Hopefully that clears up some of the more common questions, though we suspect there will be many more. Please just ask us. Better yet, come visit us in the Strip District sometime and we’ll show you under the hood.