Here’s the thing about that infamous new CNN poll

Dear Palmer Report readers, we all understand the difficult era we're heading into. Major media outlets are caving to Trump already. Even the internet itself and publishing platforms may be at risk. But Palmer Report is nonetheless going to lead the fight. We're funding our 2025 operating expenses now, so we can keep publishing no matter what happens. I'm asking you to contribute if you can, because the stakes are just so high. You can donate here.

Depending on which news outlet or pundits you’re listening to today, you’re either thrilled to learn that Joe Biden now has a twelve point lead in the new Washington Post poll, or disheartened to learn that Joe Biden now has a four point lead in the new CNN poll. Of course once you get away from punditry and back into the real world, you shouldn’t be looking at either poll individually.

Ask any polling analyst, or any statistician in general, or even anyone with a math degree, and they’ll tell you that the only valid way to look at the polls is to look at the average of the polls. Several polls have been released over the past week, and if you average them all together, you see that Biden currently has a roughly eight point lead. When you take these two new polls, Biden+12 and Biden+4, and you plug them into the average, it turns out Biden still has about an eight point lead.

This CNN in particular has some severe problems. In May, it had Biden+5. In June it had Biden+14. Now it has Biden+4. If you’ve been following the election, you know that there have been no real-world news events that could have caused Biden to jump nine points or drop ten points. It’s fairly clear that when CNN saw its original Biden+5 poll was way below the polling averages, it concluded its methodology must have been off, and ended up overcompensating. Then after CNN realized the Biden+14 poll was way above the averages, it shifted its methodology again, overcompensating in the other direction. This means the CNN poll is invalid. There’s no conspiracy here; sometimes statistical sampling just goes awry.

Fortunately, when you average the polls together, it compensates for any bad apples. In fact the polling averages tend to be pretty consistent from one week to the next, and pretty accurate when you look back at the final results. The averages are pretty boring. The way the media sees it, that’s the problem with the averages: they’re bad for ratings. No one stays tuned in for a race that keeps not changing.

Accordingly, some news outlets are promoting the Washington Post’a Biden+12 poll today as if it were the only new poll, because they’ve calculated that the best way to keep you tuned in this week is to give you a feel-good story about Biden soaring. A lot of other news outlets are promoting CNN’s Biden+4 poll today as if it were the only new poll, because they’ve calculated that you’re more likely to stay tuned in this week if you’re scared into doing so.

Unfortunately, this kind of misleading reporting leads a lot of people to conclude that the polls can’t be trusted and should be ignored. But there is a lot to be learned from the polls, if you look at them in a valid way. Ignoring the polls entirely is as much of a mistake as believing the TV pundits when they present one poll out of context at a time.