reports confusion in the predicted results for the Senate race between incumbent Ted Cruz and Representative Beto O’Rourke. On Tuesday, Cruz boasted a nine-point lead over his opponent in a poll overseen by Quinnipiac University. But another poll on the same day showed a tie between the two, and on Wednesday, a poll from Ipsos Online found that O’Rourke had a 47-45 lead—the first poll to identify him as the frontrunner.
Polling is an extremely difficult process, and, as it turns out, there’s no one way to do it. That’s exactly what these three polls are showing. Both polls looked at what are referred to as likely voters, rather than registered voters. That method takes a step beyond just looking at who can vote, and considers who will vote. Conceptually, it’s a great idea. According to Dallas News, this gives a truer representation of who will actually show up on Election Day. In execution, though, there are problems.
Determining who a likely voter will be is abstract, and each of the aforementioned polls used different methods in doing so. Quinnipiac’s sample determined that 35 percent of voters would be Republicans, with 26 percent Democrat and 33 percent Independent. Ipsos, on the other hand, took 47 percent Republican, 43 percent Democrat and nine percent Independent. With those numbers alone, it’s obvious the polls will differ, but even the interview methods were different. While Quinnipiac talked directly to their sample over the phone, Ipsos conducted their polling online.
Texas’ Senate race is seeing different results because their polls are totally different to begin with. There’s no real way to determine exactly who will show up to vote, though these polls do represent real possibilities. They declare who the winner will be, given that the right people show up to vote. It’s not a guarantee; it’s a reminder.