Two polls came out today about President Trump's job approval; they were the Gallup poll and the Rasmussen poll. According to Gallup the voters disapprove of the President's performance in office by a vote of 57 to 39 %. According to Rasmussen, the voters are split 50% to 50% on Trump's job performance.
What does this tell us? The answer is simple: the polls are meaningless. During the 2016 election, Gallup did not poll the race between the President and Hillary Clinton. Rasmussen was the second most accurate of the polls in that election. Nevertheless, that does not mean that the Rasmussen poll is correct while Gallup is not. And, to be sure, this is not just a quirk in the polling. The statistical analysis of these results show that the difference between a 19% margin for disapproval and a tie in the two polls is much farther apart than any mere error in the sampling.
There is one difference that might help explain some of the difference. Rasmussen adjusts its results to have the sample group conform to the levels of voter turnout in the 2016 election. In other words, if that election had x% of Republican voters and y% of Democrats, then Rasmussen adjusts its figures to meet that percentage. Gallup does no such adjustment. That means that if Democrats answer the pollsters more often than Repubicans, then the anti-Trump numbers will be pumped up. The opposite is also true.
Right now, there's no way to know who will turn out in the next election, let alone who will be running for office. This makes the adjustments or lack of adjustments critical to the outcome. In fact, it may be that some people who support Trump won't tell the pollster because they don't want that information known.
If we put all this together, we can generalize about the results as follows: based upon scientific polling, President Trump either has very low approval numbers or he has rather good approval numbers or somewhere in between
Why do they bother doing these polls anyway?
I realize that perhaps I should also ask why do I bother writing about them?
What does this tell us? The answer is simple: the polls are meaningless. During the 2016 election, Gallup did not poll the race between the President and Hillary Clinton. Rasmussen was the second most accurate of the polls in that election. Nevertheless, that does not mean that the Rasmussen poll is correct while Gallup is not. And, to be sure, this is not just a quirk in the polling. The statistical analysis of these results show that the difference between a 19% margin for disapproval and a tie in the two polls is much farther apart than any mere error in the sampling.
There is one difference that might help explain some of the difference. Rasmussen adjusts its results to have the sample group conform to the levels of voter turnout in the 2016 election. In other words, if that election had x% of Republican voters and y% of Democrats, then Rasmussen adjusts its figures to meet that percentage. Gallup does no such adjustment. That means that if Democrats answer the pollsters more often than Repubicans, then the anti-Trump numbers will be pumped up. The opposite is also true.
Right now, there's no way to know who will turn out in the next election, let alone who will be running for office. This makes the adjustments or lack of adjustments critical to the outcome. In fact, it may be that some people who support Trump won't tell the pollster because they don't want that information known.
If we put all this together, we can generalize about the results as follows: based upon scientific polling, President Trump either has very low approval numbers or he has rather good approval numbers or somewhere in between
Why do they bother doing these polls anyway?
I realize that perhaps I should also ask why do I bother writing about them?
No comments:
Post a Comment