When you look at political polling, there are the so-called “independent” polls - Gallup, Rasmussen etc. (independent in the sense of neither party ordering the particular poll). And there are “internal” polls - where the candidate orders the poll.
Every time I have seen the results of “internal polls” released, they are always heavily skewed to the side of the one that ordered them, with no exceptions - compared to the “independent” polls. And, of course, as we saw with the Walker/Barrett election, as well as others I have seen, the independent polls do much better compared to the final results than the internal polls.
So - the question is, when the internal polls are released to the press, are those “real” numbers that the campaign got from the pollster that they ordered the poll from? Or does the campaign just substitute better numbers before releasing?
If it’s the first - then it is pretty weird that the campaigns keep paying for the polls that are so consistently inaccurate and skewed their way. And if it’s the second - that is just blatant lying, and if I was the pollster whose name was on that poll I would not work for that campaign, since the inaccurate results would undermine my reputation.
His argument is that it isn’t that internal polling is itself worse than independent polling, but that campaigns only release the results of polls that are in their favor to the public. It’s not that they’re changing the poll numbers, but that they’re cherrypicking the polls they release.
Yup, that’s it. If I commission daily tracking polls from some “reputable” company it’s not that hard to only release the best numbers. This will make the company look bad, but they get paid so what do they care? Considering at least one independent pollster has been accused of blatantly making up their numbers it’s not like it’s a profession ripe with repute anyways.
On the other hand, the candidates (if they’re sane, which is not necessarily a given) do actually want their internal polls to be as accurate as possible, because they use that to guide their campaign. If you launch a new line of advertising, or embrace a particular issue, or (for the incumbent) pursue a particular platform, and the polls following that improve for you, then you want to keep on doing that. If you do something and the polls following that get worse for you, you might want to consider a different tack. This is very valuable information, and if you cause your polls to be skewed, you’ll deny yourself that information.
This is why they’re called “internal” polls, since their primary use is just for informing the people working within the campaign. If you can occasionally release one publicly to fluff up your candidate, that’s just icing on the cake.
Exactly. An “internal” poll may be bought to explore the vulnerabilities of the opponent. These polls may ask about issues with the opponent: “John Smith had an 80% voting attendance record in the last session of the legislature. Do you believe his attendance record is a matter of great concern, some concern, or no concern?” The answers then may be used to determine whether Bob Brown is going to make an issue of Smith’s attendance.
So, if a poll contains quite a few questions about an opponent’s vulnerabilities, it is going to have an effect on the bias of the poll, especially if the poll intentionally oversamples some people (like if the intent is to get more responses from geographical areas that have higher proportions of one party or another, or whatever).