Voter polling sometimes gets it wrong ... sometimes dramatically so ... witness Brexit. Why might this be so? I have previously discussed "thumb on the scale" polling ... see: Polling. But let us assume that a pollster is honestly trying the read the tea leaves ... what could go wrong? One slip could be a disconnect between the polling samples and the actual voter populations. Here are some examples:
- Since, surprisingly, many voters don't speak English these days, it is doubtful that these people, even though included in the polling samples, may be able to answer the pollsters' questions.
- Obviously, last hour events can have the ability to change minds inside the voting booth.
- Election Day turnout can be affected by the weather ... particularly between the younger and older voters.
- It is quite unlikely that voter fraud can be captured in polling results ... graves or phony addresses usually don't have phones
- Polls try to discern voter enthusiasm to determine if polling opinion translates into voter action. But "likely voter" is a binary choice, whereas enthusiasm is a wide continuum. And, of course, forward- leaning turn-out operations can clearly have an impact on more passive voters.
- Honest pollsters try to stratify their political party allocations based on past results ... but these past results can be historically stale ... particularly insofar as the impact of third-party candidates.
Are the polls right or wrong that show Hellary Clinton winning tomorrow? I'll be damned if I know...
