After the debate, Donald Trump and his campaign have claimed that the Republican nominee won— according to all the polls. One new press release from Trump’s campaign says he “leads post-debate surveys.” It’s not true. CNN and YouGov gave the win to Hillary Clinton, while the Drudge Report poll, among several others, had Trump winning handily. Which of these to trust?
The Drudge Report poll is not a scientific poll, and therefore its results do not tell us anything about what America as a whole thought. The same is true of other polls like those conducted by Time, CNBC, and the Washington Times. These polls are of no value for gauging public opinion.
A “scientific poll” is typically one that uses a random sample of the population. There are different ways to create such a sample, but the important point is that it is not self-selecting. It is in this way that polls of just 1,000 people can accurately capture the sentiments of a hundred-million voters. Online pollsters, such as YouGov, do not utilize random samples, but they employ an intricate weighting procedure to create a panel that is representative of the entire population. It’s still debatable whether online polls are as good as telephone surveys, but they are definitely getting better. Because they cost so much less than telephone surveys, they are probably the future of political polling—once the experts further refine their methods.
The Drudge Report poll is like the YouGov poll, in that it is opt-in; however, there is no weight given to the results. Everybody’s vote is counted equally, which means that Drudge’s sample is not representative of the nation, and therefore does not tell us anything about public opinion. It can only tell us about what his readers thought.
Even proper post-debate polls are hardly my favorite type of poll. They have a lot of methodological problems that are very difficult to overcome. We should take them with a grain of salt. But online polls like Matt Drudge’s tell us nothing about public opinion.
