Wikipedia removed RCP from their aggregate pollster list because they did not properly weigh their polling averages?

However, once again RCP was way closer to the actual results than 538, NYT, or Nate Silver. So Wikipedia quietly added them back after the election ended.

Now here is the thing. Wikipedia did it because they believed that RCP was not using the correct polls, not weighing the averages as they should, and making it falsely appear that Donald Trump was winning in order to make Democrats feel like they were losing. Wow. Isn't that special?

The reality is that Real Clear Politics ignores many of the garbage polls that 538 and Nate Silver routinely add to their polling averages. Pollsters such as Big Village and Outward Intelligence that had Harris up by six or seven points were included in the 538 and Silver averages, but not in the RCP averages. I would argue that RCP was correct to ignore these pollsters. In fact, Big Village (who provided a nice set of cross-tabs) was such an outlier that I actually removed them from my spreadsheet (which was, of course, the correct call in retrospect). That is something I have never done before, but simply had to in this case in all good conscious.

Moreover, RCP has the gall to simply have an average that takes the polling results and divides them by the number of polls to provide... well the mathematical definition of the term average (or more technically the mean average). There is no such mathematical definition that fits what 538 or Nate Silver is doing. It is basically putting a thumb on the scale using what appears to be mostly subjective reasons to make one poll worth more than another poll. That is not an average folks. For instance, it was well documented that Silver decided to substantially lower his weight on Rasmussen (what does he have against them) because there was a "rumor" that they had shared polling information with the Trump campaign. What on god's green earth does that have to do with accuracy? The correct answer, of course, is nothing.

What happens is that Silver and others have certain ideas (possibly outdated and old fashioned) about what makes a gold standard poll. Does it matter whether or not these gold standard polls are accurate? Not really, they just have to include long drawn out telephone conversations with potential voters. As long as they adhere to these rule, then they are then weighted much higher than other types of polls, such as a poll that only asks for demographics and who you are voting for. The more questions you ask the person, the better the poll? Not sure how that makes any sense, and quite frankly those have been some of the worst polls we see.

Nate Silver will no doubt put out an updated list of which are his highest graded pollsters. The previous list showed NYT/Siena at number one, even as they had been rated as one of the "least" accurate pollster in 2020 by several outlets who rely on just math and results. Pollster like Marist and Quinnipiac are also highly graded, while being routinely inaccurate. At the end of the day it would seem more like 538 and Nate Silver (who both changed their forecast from Trump to Harris quite literally in the 11th hour) were the ones putting their thumb on the scale to make Harris voters feel better and to make Trump voters feel worse. Perhaps Wikipedia should have removed them instead?