Would not a “neutral”, unbiased polling operation be preferable - I mean wouldn’t a campaign just want results that are as accurate as possible? And why would a polling outfit want to cut their potential client base in half?
Polls serve more than one purpose. They don’t just provide the candidate with useful information – they also help to influence voters.
A poll can be phrased to produce specific results, which a politician can then publicize to support a position.
They can also do “push polls”, which not only give a skewed result but spread lies about the opposition in the guise of JAQing off: “If you found out that Clyde Omnibarrow introduced a bill to force teachers to cover goat felching in biology class, would you vote for him?”
Pollsters aren’t just statisticians that design and run polls. They are really political advisers that apply special skills. Pollsters don’t simply measure the electorate’s temperature, they also advise candidates on how to capitalize on their strengths, as well as exploit the weaknesses of their opponents.
It takes quite a bit of money to run a poll, so campaigns don’t want to run a lot of polls. Given that there’s a limit on how many questions most people will answer on the phone (would you really allow a pollster to ask you 50 questions?), the pollster has to craft questions carefully based on his knowledge of the candidates, taking into account both his and other campaign advisers’ views of how to win the race, to draw out the most relevant responses that will be used to further refine the campaign. Contrary to the OP’s implication, just because a pollster works for a particular side of a race, doesn’t mean that they want to feed their employer only good news. Accuracy is indeed valued.
Push polling does occur, but it isn’t on the short list of why political pollsters work for one party or another. The fact is that most of them have political views of their own, and they want to help elect their kind of people to office.
Missed further edit window: there’s also quite a bit of message testing that goes on in polls. For example, easy questions like, “What is the most important issue for you in this election?”
But there’s also things like, “If you were told that Frank Jones supports the assault weapons ban, would that make you more or less likely to vote for him? If you were told that Keith Smith wants to cut taxes on small buisnesses, would that make you more or less likely to vote for him?”
That sort of stuff.
There’s also a security advantage to having a pollster where the company & most of its’ workers support your political party – lower chance that the results will be leaked to someone from the other party.
First of all, no one wants to vote for a loser, second if you can convince the populous the other guy is losing they won’t come out and bother to vote at all.
There’s more to it than that.
Much of the science involved in polling is making sure the sample accurately represents the population. Pollsters spend a lot of effort on breaking down the sample into various categories and making sure it doesn’t have too many or too few of this or that subgroup, and if it’s off, they adjust them.
And here’s where the political lean is. Everyone is guessing to some extent about what is a true representative sample, so there’s some leeway here. And in interpreting this, people are influenced by their own political leanings. A Republican has 16 reasons to think that the electorate skews right, and a Democrat has just as many reasons to think it skews left. (You can see this in discussions on this MB.) So Republican pollsters tend to have results that skew in favor of Republicans and vice versa.
Pollsters also want to work in the next campaign, and the one after that, and the one after that, etc. It’s in their best interests to develop a reputation as reliable, accurate and like-minded for future candidates of the same political leanings.
As it happens, I’m now reading Theodore Sorensen’s Kennedy, which extensively covers the 1960 campaign. I was amazed to learn that the Kennedy and Nixon campaigns had an informal agreement to share some polling data. I don’t expect that happens much nowadays.
One legitimate reason for pollsters who lean one way or the other is that the different parties and their candidates want to raise the profile of different issues in the public debate. One way to do that is poll questions related to those issues, and try to make news out of the results.
Polls, and thus pollsters, aren’t always used to win elections. Sometimes their purpose to get a temperament of the electorate so a Senator knows how his constituents want him to vote. If I work for Hon. William E. Coyote, D-NM, then I’m a Democratic pollster. My family’s got to eat in the odd years, too, y’know.
Do you have a cite for members of Congress paying for polls to determine how to vote on a bill or something? Polls are not cheap, they take time to design, run, and analyze, and they must be done using non-government funds. Other than DNC or RNC sponsored polls, I’ve never heard of congressional members using their own funds for internal polling on how to vote on an issue.
In any case, internal polling for campaigns begins early. Pollsters have no problems paying bills in off-years.
Well, part of the reason is that, this early on, the parties are voting amongst themselves. Sure, some places allow people from the other party to vote, but that still leaves it predominantly people from one party.
I doubt. for instance, that Sarah Palin, were she to run, would care at all about what Democrats thought of her.