Where are the polls really at?

You cannot just take the polls straight up, but rather you should be comparing to where they were at four and eight years ago.

A few formulas on a spreadsheet and bit of math tells us that right now Kamala Harris is running about four points behind a combination of 2000 and 2016 polling if you were to average it all out.

On a national scale, the logic is easy to follow. Biden was up 7.8% and Clinton was up 5.5% which is an average of 6.65%. Harris is running ahead 1.5% now, which is about 5.15 points behind the averages of 2020 and 2016.

If you did the same on the battleground states (I removed Ohio and Florida) and added everything together Harris is running 3.98% behind the averages of 2020 and 2016. The differences go from a high of 7.65% in Pennsylvania to smallish 0.95% in Georgia. In only one case (Georgia by 0.1% in 2020) is Harris running ahead of either Biden or Clinton in any of the individual state races. The polling has been so bad the past two elections that both Biden and Clinton led Trump well outside the margins of error in Florida and Ohio, States that Trump won by a total of 1.5 million voters over those two elections. Pollsters should be embarrassed, but so far are not.

Now do I believe that some of this might have been corrected since 2020? I don't know. Some of the numbers I am accumulating suggests that the top line results of many pollsters are closer to the demographically adjusted numbers I come up with than the past two elections. But as I have suggested just recently, many of the pollsters showing the best results for Harris (outliers) are not providing their crosstab information. So we may still see the averages being skewed, especially in places like 538 and the Silver Bulletin where they will use any poll from anyone that is released and toss them into their average. Well, at least any poll not called Rasmussen.

At the end of the day, my spreadsheet adjusted number is still considerably better for Trump than all of the aggregate polling places (538, Silver, Hill) show in their average. My spreadsheet shows Harris at around 1.7% up, whereas those other aggregates have her up by around 4% on average. RCP, which is only using major polling, still shows their average at 1.5% and 2% depending on the number of candidates provided. This clearly shows that lesser-known pollsters (who are not providing crosstabs) are the main driver for why these other aggregate pollsters are showing a larger lead for Harris. Some examples include polls by Bullfinch group showing Harris up eight, Fairleigh Dickinson University showing Harris up seven, and Outward Intelligence showing her up by six. If you are asking yourself, who are these pollsters, well then you are not alone.

Perhaps there will be other pollsters coming along with these sorts of results over the next few days, but I suspect that the JFK Jr news will upend some of that polling, given any poll including him is no longer valid. Pollsters will need to adjust for this, and they may or may not want to release anything now that still shows him in the mix. My own calculations will likely change as I also have used crosstabs specifically from the five-way race or anything that includes Kennedy. I think we need to hang tight for about a week or two and see how all of this pans out. Whatever "bump" Harris might have gotten from the DNC will be undermined by the lack of the Kennedy option moving forward (which will likely cost her a point or two).