Data Denial: the fallacy of ‘unskewing’

Ian Palmer

In GCC, Polling, Strategy, US Posted

Over the pond, political commentators have spent the last couple of weeks arguing with themselves about the accuracy of the opinion polls which show Hillary Clinton taking a clear lead over her rival, with claims centering on the fact that Democrats are overly sampled in the groups polled.

Nothing new here. At the last election, Republicans became great advocates of the UnSkewed Polls website, which claimed to ‘fix’ badly sampled polls (i.e. any which didn’t show Romney winning). And even The Trump himself, tweeted this last election:

But why deny the data in front of you?

Of course, there’s a chance that what you’re stating is correct: that Democrats have been over-sampled, and that the polls are actually wrong. But pollsters make their money from being accurate – it’s a very short sighted pollster that knowingly mis-samples. And, as Harry Enten over at fivethirtyeight.com says:

Most pollsters don’t weight their results by party self-identification, which polls get by asking a question like “generally speaking, do you usually think of yourself as a….” Party identification is an attitude, not a demographic. There isn’t some national number from the government that tells us how many Democrats and Republicans there are in the country. Some states collect party registration data, but many states do not. Moreover, party registration is not the same thing as party identification. In a state like Kentucky, for example, there are a lot more registered Democrats than registered Republicans, but more voters identified as Republican in the 2014 election exit polls.

A lot more is to do with keeping up the appearance of winning. That your guy isn’t falling behind, and that it’s still all to play for. On this very blog, I’ve criticized the value of that approach. But with Trump falling away so significantly over the past couple of weeks, you can see why his team may be doing everything they possibly can to cling on to still looking like contenders.

chanceofwinning

Of course, this isn’t just a phenomenon in the States. In the GCC office, we often share war stories of candidates we’ve seen either explaining the perceived bias of external polls, or even forcing re-sampling and weighting of internal ones to favour them. Of course, this is sometimes about perception and PR-facing spin, but more often than not, it’s part of a journey of self-denial throughout the course of a campaign where candidates deny the truth of the situation they find themselves in.

The best campaigns make the best use of the best data available to them. They make informed, calculated decisions which are in line with an overall strategy, adapting to changes in the environment and context around them. That’s the way we do it at the Good Campaigns Company. But re-sampling polls, or denying what they say, is in direct opposition to this data-led approach. More often than not, it won’t end well. And if there’s one thing the polls stateside tell us right now – which we can trust for sure – Trump’s heading in that direction…