Any negative polls are fake news, just like the CNN, ABC, NBC polls in the election. Sorry, people want border security and extreme vetting.
- tweet from @Realdonaldtrump
In our current partisan environment, polling data is becoming an increasingly controversial weapon in the policy fights. In this tweet from President Trump the other day, he states that negative polls are fake news, and thus are wrong, and should be discounted.
Let's step back for a moment and think about the implications of this statement from a purely scientific perspective.
In the past election cycle, we had a series of highly respected polling organizations which got the results of the presidential election wrong. As I have explained in prior columns, this is partly an error in the way the science was applied, and partly in the way the results were interpreted.
To be fair, pollsters aren't the most popular right now. President Trump is implying there is a bias in the polls against him, and therefore any poll that does not favor his administration is false.
Is President Trump correct that the polls are biased against him? As a data polling expert, here are at least two ways the polls may be misleading:
- The samples taken by the major polling organizations aren't truly representative of the overall population. What this implies is that the underlying set of people who pollsters are asking the questions of are more likely to be Democratic or left-leaning.
- The polling questions are worded in such a way to result in a biased outcome. For example, questions could be phrased in a manner that frames "extreme vetting" particularly negatively.
However, President Trump hasn't specifically stated either of these types of critiques as the basis for his claim that polls constitute fake news. A sound consumer of data has to be careful when interpreting polls. But, the answer doesn't lie in the assumption that any data which runs counter to our opinion is wrong, and data that supports our opinion is clearly correct. This is an extreme version of cherry-picking--when one only credits evidence that supports their opinion and ignores any evidence which is counter to their position.
Numbers and data are critical to shaping policy debates and political discourse. Questioning survey results on methodological grounds is a part of the scientific process and can result in healthy debate. But, attacking numbers because you don't agree with them is misleading, counter to sound scientific methods, and stifles opposing opinions.