Internal Polls--what are they, and how are they used?

Politics & Elections
1

In most elections, we hear a lot about each party’s “internal polling”, and how that affects their strategy.

My question is : What’s the difference between an internal poll, and a regular poll? (by,say, the Gallup company).

Gallup and all the many companies which do polling are usually accurate. They have vast infrastructure in place to make the thousands of phone calls. And, more importantly, they have the infrastructure and experience to find the right people to call, to select an accurate representative sample of the population.

For internal polling, I suppose the Dem and Repub parties could invest enough money to match the professionals at Gallup and all the other companies.
But what advantage does it give them? Why not just hire a regular company like Gallup? The could put a non-disclosure clause in the contract if they want to keep the results secret.

Why do the parties think that their privately organized poll is better than the professional ones?

Let’s take as an example: the last election when Hillary lost, despite all the polls saying that it was unlikely.
One small issue, now forgotten, raised a lot of curiosity at the time: the fireworks display planned for Hillary’s victory celebration, At the last minute, just a day or so before the election, she cancelled the fireworks,( which were a complex operation and had been very carefully planned and coordinated not to interfere with the nearby airports and shipping lanes, etc.)
Then suddenly–the fireworks are gone, cancelled with no explanation. Some people wondered if the Hillary campaign had internal polls showing that she would lose.

My question is: even if she did have an internal poll showing she would lose…there were dozens of other polls showing she would win, Why would she put so much faith in just one poll? Are the internal polls conducted in some way that they are more accurate? And if so, wouldn’t the commercial polling companies learn quickly and adopt the same techniques?

2

Just a guess but the internal polls are probably pretty good. Campaign officials want accurate data as much as and maybe more than anyone else. Perhaps they’re just targeted in ways that gives the campaign clues in how to persuade more on the fence voters. “If President Trump promised to send all of the country’s Muslims back to China, would that make you more or less apt to vote for him?”

3

I remember some talk about how Trump’s internal polling told him that the upper midwest was in play and they needed to push hard up there, while others were arguing a “blue wall.”

I also know that polls aren’t just pure data. They include algorithms on how to apply that data, because you’re not going to get a perfect sampling of everyone. You use demographic data from the you did poll and extrapolate that to how many people in that demographic exist, for example. So it could be less that the polling is different and more that the assumptions are different. I think Trump had different “likely voters” metrics, for example.

4

I think it’s interesting to see FiveThirtyEight’s take:

5

And this one …

6

There’s several issues. First of all, the campaigns don’t do the polling themselves, they contract out to polling companies to do the work for them.

Second, there’s a big difference between internal polls conducted for the campaign’s own use, and internal polls released to the public. The former are used to decide on strategy. The latter to create an impression for the public. Typically, a campaign runs many internal polls, but only releases the ones it thinks are helpful.

7

I don’t want to nitpick here but that 538 piece is about internal polls which get released. And those are meant to persuade – someone. I assumed we were talking about internal polling that only the campaign sees, perhaps an incorrect assumption on my part. If so, apologies to the thread.

There was a story out earlier this year analyzing how the number of internal polls released closely correlates with the results of presidential elections. Simply, the party releasing the fewest number of internal polls generally lost the election.

8

Perfectly fair point.

I think the second 538 article I posted takes that approach.

I’m cooking, so … I was way too brief. To add a bit …

Where I used to work, we did a lot of ‘internal polling.’ The simple use of that data was:

  • anything favorable became a press release, while
  • anything unfavorable went to the relevant department in charge of whatever didn’t look too good … with a very implicit “fix this.”

I suspect campaigns use this info similarly, and have very specific reasons for asking very specific questions – questions far more specific than the major pollsters are probably generally asking.

9

Jimmy Carter’s internal pollster Pat Caddell had told him in the Oval Office a few weeks before the 1980 election that it was all but over for his re-election. Even getting the hostages out right before election day was unlikely to get him four more years.

10

I wonder if the 1980 election was the biggest miss for pollsters ever. Or at least since Dewey defeated Truman.

11

Depends on your definition of unlikely. 538’s final weighted poll gave Trump a 29% chance of winning, which is distinctly non-zero.

As for internal polling, i’ve seen a third type in which former political consultants build a brand new polling group, and then develop polling results highly favorable to a preferred candidate. Typically they have indeed taken some kind of poll, either Internet or phone, but then used suspicious weightings or undersampled.

12

Internal polling isn’t just “will you vote for Candidate X” A lot of it is message testing, figuring out how to allocate resources, and so on.

In some cases, internal polling might be done strictly with a candidate’s supporters, to estimate how strong or weak their support is.

13

I think the most likely explanation here is that the fireworks cancellation had nothing to do with polling.