Polling Versus Voting In The Era of Modern Voter Suppression

[TomCADem]

Lets say that we take a poll of 10 registered voters the day before the election. Six say they will vote for Biden and four say they will vote for Trump.

The next day is the election and we stop 1 minority voter from voting because they do not have the proper ID. We make another voter who lives in a large city with small children wait for hours in line before they give up because they have to pick up their kids in the evening from family). Finally, a younger voter in the group procrastinates a bit and drops their ballot in the mail on Sunday evening before the election, but it is not counted because it does not arrive at the election official's offices until Wednesday after Election Day.

We tally the votes on Election Day and the results are 4 votes for Trump and 3 votes for Biden. Does this mean that the poll was not accurate?

[LakeArenal]

1. Polls are just an estimated guesses.

[TomCADem]

2. In that scenario, were voters guessing about who they intended to vote for?

It is only a guess if you believe a poll a prediction of an election outcome. If a poll is a measure of voters intended desire to vote for particular candidate at a particular point in time, then it is not really a guess.

[LakeArenal]

4. Huh? A poll is a guess only if you think it's right

Polls are limited. The can be pushed, pulled, or manipulated.

I personally have never been polled in the last four elections.

It’s an estimate based on the people polled.

[TomCADem]

9. Did you or are you going to vote for Biden/Harris?

Is that a guess?

I think the guess portion that you are referring to are the inferences that people draw from the polls. The point of my post is that polls can be totally accurate in measure the preferences of likely voters, but there could be significant impediments to voting that prevent people from voting or having their votes count.

[LakeArenal]

10. Basically it is a guess. I'm guessing Joe is right.

Time tells.

[Hugin]

3. They need to add a category to the polls in addition to likely voters.

Attempted voters.

That would provide a gauge on suppression.

I have a family member who came within a hair's breadth of being such. But, I recently checked and that vote is in the custody of the County Clerk's office. Not much more we can do.

[Hermit-The-Prog]

5. Voting should be easy, fast, convenient, and verifiable. We can make it so.

[Statistical]

6. Pollster DO account for suppression (I guess that is both sad and good)

They use likely voter models. These models are based on votes actually cast and counted. On average Black Americans vote less than White Americans. Some of that may be apathy, some of that may be suppression. From a polling standpoint it doesn't really matter why only that they do.

The effect of voter suppression is compensated for in the polling. Now the good news is that voter suppression is at most 2%. It can help you steal a close election. It can't help you steal a landslide. If you hate voter suppression (and any patriot should) then bury those bastards in your votes. Drown them in votes. Make the election one for the record books which annihilates the GOP across the map.

[TomCADem]

7. But if the Pollster Is Basing Its Analysis on Prior Data...

...then it is not going to pick up recent developments such as court decisions to bar the count of delayed VBM ballots. Likewise, if voter turnout increases, then the effect of long lines in urban areas may become more pronounced.

[Statistical]

8. True but voter turnout is a constant battle with supression worsening but activism improving.

It isn't going to like in the OP turn a 20 pt Biden win into a 10 pt Biden loss. Voter suppression is far more insidious and subtle. It robs voters on the margin. Even to WI "late" ballot crap will probably affect less than 3% of voters. It isn't like suddenly 30% of voters will be suppressed that weren't the prior year.

So if supression goes up in a given year the polls might marginally overestimate Democratic support but the effect will be small likely smaller than other forms of error. If supression goes down (not away but down) in a given year it might marginally underestimate Democratic support.