If you watch the headlines today, you will see the big "news" that a CNN poll shows that every Democrat other than Elizabeth Warren is beating President Trump in head to head races. The Democrat doing best is Beto O'Rourke (even though Beto is sinking lower and lower in most preference polls among Democrats only.) What you won't see is that this poll is rigged. Let me explain:
1. Pollsters want their polls to be representative of the electorate as a whole in order to assure maximum accuracy. That makes sense. If one wants to know who will carry Pennsylvania in 2020, it won't work if the only people you poll are in Philadelphia. That city is 15% of the state population, but it is also, by far, the strongest Democrat area in the state. No matter who you list as the Democrat candidate, he or she will beat Trump in a poll. Likewise, if you interview only Republicans, there's no Democrat who could beat Trump.
2. Despite this, CNN's poll did not come up with a representative sample. In 2016, Pew found that the electorate had 4% more Democrats than Republicans. The CNN poll showed that there were 7% more Democrats in their sample.
3. CNN's polling further rigged the head to head presidential races. For some reason, the people doing the poll didn't ask all of those questioned who they preferred in head to head matchups. The pollster asked only about 40 to 45% of the total group for their preferences. This strange practice means that detailed breakdowns as to the nature of the group polled is not available with one exception. CNN did publish numbers that indicate that somewhere between 25 and 28% of those polled in the head to head races were non-white. Since African Americans tend to vote Democrat by more than 90%, this is the best possible way to rig the result to make sure that Trump would lose. Pew found that in 2016, African Americans and Hispanics comprised 21% of the electorate. That means that CNN put a figure of between 4 and 7 percent too many minorities in the mix if one uses the 2016 baseline (which most pollsters do.) But it's really worse. Many Hispanics would self-identify as "white". Think of people like Ted Cruz or Marco Rubio. Because CNN didn't disclose their identification questions, we can't tell if this group is counted as "white" or "Hispanic" or both. Putting these distortions together, the head to head figures would provide about an additional 2-3% for Trump and 2-3% lower result for the Democrat. And that assumes that these small groups chosen by CNN for polling on head to head matchups were not further skewed.
4. If the figures get adjusted by the 2-3% discussed above, then every one of the races fall within the margin of error. Trump leads in some, is tied in others, and is behind only O'Rourke. The problem with that result for CNN, however, is that it isn't the outcome that they want.
It's bad enough that CNN pushes Fake News. Now that they are rigging polls, it is truly disgusting.
One more thing: The first part of the polling data asks people how they rate President Trump on specific issues and also on his overall job performance. This section was asked of the entire sample. These results show Trump with his highest numbers in many months. Just under 60% approve of his handling of the economy. Total job approval is 45% which is a big jump from the last poll. Strangely, in the head to head matches, Trump doesn't even get to 45% in half of the races. This shows the distortion in the results. Normally, presidents outpoll their job approval numbers in head to head matchups.
1. Pollsters want their polls to be representative of the electorate as a whole in order to assure maximum accuracy. That makes sense. If one wants to know who will carry Pennsylvania in 2020, it won't work if the only people you poll are in Philadelphia. That city is 15% of the state population, but it is also, by far, the strongest Democrat area in the state. No matter who you list as the Democrat candidate, he or she will beat Trump in a poll. Likewise, if you interview only Republicans, there's no Democrat who could beat Trump.
2. Despite this, CNN's poll did not come up with a representative sample. In 2016, Pew found that the electorate had 4% more Democrats than Republicans. The CNN poll showed that there were 7% more Democrats in their sample.
3. CNN's polling further rigged the head to head presidential races. For some reason, the people doing the poll didn't ask all of those questioned who they preferred in head to head matchups. The pollster asked only about 40 to 45% of the total group for their preferences. This strange practice means that detailed breakdowns as to the nature of the group polled is not available with one exception. CNN did publish numbers that indicate that somewhere between 25 and 28% of those polled in the head to head races were non-white. Since African Americans tend to vote Democrat by more than 90%, this is the best possible way to rig the result to make sure that Trump would lose. Pew found that in 2016, African Americans and Hispanics comprised 21% of the electorate. That means that CNN put a figure of between 4 and 7 percent too many minorities in the mix if one uses the 2016 baseline (which most pollsters do.) But it's really worse. Many Hispanics would self-identify as "white". Think of people like Ted Cruz or Marco Rubio. Because CNN didn't disclose their identification questions, we can't tell if this group is counted as "white" or "Hispanic" or both. Putting these distortions together, the head to head figures would provide about an additional 2-3% for Trump and 2-3% lower result for the Democrat. And that assumes that these small groups chosen by CNN for polling on head to head matchups were not further skewed.
4. If the figures get adjusted by the 2-3% discussed above, then every one of the races fall within the margin of error. Trump leads in some, is tied in others, and is behind only O'Rourke. The problem with that result for CNN, however, is that it isn't the outcome that they want.
It's bad enough that CNN pushes Fake News. Now that they are rigging polls, it is truly disgusting.
One more thing: The first part of the polling data asks people how they rate President Trump on specific issues and also on his overall job performance. This section was asked of the entire sample. These results show Trump with his highest numbers in many months. Just under 60% approve of his handling of the economy. Total job approval is 45% which is a big jump from the last poll. Strangely, in the head to head matches, Trump doesn't even get to 45% in half of the races. This shows the distortion in the results. Normally, presidents outpoll their job approval numbers in head to head matchups.
No comments:
Post a Comment