Heavy turnout favored GOP. I recall we had people who were poll junkies. What happened?

tequila4kapp said:

---

IIRC Rasmussen was pretty accurate.

The normal answer is Trump voters hide. There's truth in that but I also think pollster's natural bias comes into play. For example, determining what % to include from assorted groups. Also, many of the polls are customers of one side or another. That has to influence independence and accuracy.

---

This is exactly right. There's definitely a ton of wishcasting in many polls.

In addition to Trump voters hiding, I think many just lie due to a "Tom Bradley impact" or just hate for the pollsters/media.

And for pollsters, there's an incentive to show close polls so they won't be "too wrong" on election day.

What we've learned is that it is prevalent enough to impact the real clear politics average and other models that rely on external poll data (such as Nate Silver's) to distort those results as well.

An interesting question is whether this a dynamic is unique to Trump or if it will be a factor going forward (i.e., post 2028).

There is one group of related categories that polls/pollsters ignore …

going4roses said:

---

There is one group of related categories that polls/pollsters ignore …

---

Common decency non-democrats? Just joking folks.

BearGoggles said:

---

An interesting question is whether this a dynamic is unique to Trump or if it will be a factor going forward (i.e., post 2028).

---

it has been very common for polls to overstate the D in Pres elections even before Trump. Exit polls were especially bad.

Cal88 said:

---

sycasey said:

---

The polls didn't fail. If you looked at any aggregate they had it as a roughly tied race. A candidate winning by between 1-2 points on either side is a perfectly reasonable result, based on the polling we saw. Asking polls to be more accurate than 1 or 2 percentage points is simply unreasonable for what has always been a somewhat inexact science. This was actually a fairly good year for the polls.

---

All the errors were in the same direction, these are not just random errors.

---

No, it's fairly common to have polls miss in the same direction. In the Obama years they tended to undercount his support across the board. A 2-point miss is nothing. They missed by much more in 2016 and 2020, which indicates the pollsters have actually corrected a lot of the issues with polling Trump voters (though again, it will never be perfect).