Talk Elections

how expensive is it to poll a state?  i cant understand why any firm would waste time and resources polling states like indiana or mississippi, or utah.

I can't give you a dollar figure, but I can tell you why these states are polled.  Usually it is a local TV station, University or other body in the state that wants some local information to reort on.

To really, really do it right with a really first class firm is quite expensive actually.

A 1000 sample size poll will typically come in at about $20-25,000 or so.

A pretty decent "rule of thumb" is $25 per interview.  That should buy you a survey where they didn't cut any corners.

If a firm has done the state very recently, the second time around will be quite a bit less, as the sample generation models, demographic data, etc can be "recycled"

The recent 1500 sample size in Ohio by Mason/Dixon was likely in the 35K (or so range) if they do it again, the second go-round would be $20K or so.

As always, you can always spend less, the result may or may not be any good.

I am sure "Fat Eddy's Tatoo Parlor, Telemarketing & Collection Agency" could do it for a lot less than 20K. :)

It also depends on how long the questionnaire is.  Most media polls don't ask too many questions, but candidates that are polling for information about potential campaign strategies often employ longer questionaires (sometimes lasting 30-45 minutes). These polls are more expensive, because they require more time per respondent and require more calls because fewer respondents are willing to get through them.

As the election gets closer, tracking polls become more frequent, and the questionnaires become much shorter, so each polling is much less expensive.

All good points.

You see very few really long surveys however.  Anything over maybe 10 minutes you are really starting to abuse your sample, and your validity goes way down.

If you are lucky to have a really, really good set of operators doing your calls you might get away with 15 minutes, but IMHO, that is really the top top top end of what is safe from a validity point of view

Pollsters do if for money.. :)

For enough money, you can get a 10,000 sample size poll in the District of Columbia and know for sure the Kerry will win.

I saw a 200 person "poll" in DC showing Kerry up only 67%

Given the small sample and huge MOE (+/- 7%)  we can only be 9999999999999991% certain Kerry is indeed ahead..

I think more polling is needed :)

But seriously, often a media outlet will "buy" a front page story by doing a poll.

X leads Y by 10% !! - easy story.

Being very cynical here, I think this is also why media polling is often so bad - the media wants news and "bad" polls are often great news stories

X surging, trailed by 12%, now leads Y by 3% !

Big changes and huge shifts make for "better" news and sells more copies of the paper or generates more viewers.  Put bluntly the low ball cheap poll from a bad firm is more likely to produce a "newsworthy" result than an accurate poll from a good firm.

From the media point of view, it's more newsworthy and it is cheaper.. :(

how much would a polling firm charge a media outlet, such as a newspaper, to poll a state?  50k?

To really, really do it right with a really first class firm is quite expensive actually.

A 1000 sample size poll will typically come in at about $20-25,000 or so.

A pretty decent "rule of thumb" is $25 per interview.  That should buy you a survey where they didn't cut any corners.

If a firm has done the state very recently, the second time around will be quite a bit less, as the sample generation models, demographic data, etc can be "recycled"

The recent 1500 sample size in Ohio by Mason/Dixon was likely in the 35K (or so range) if they do it again, the second go-round would be $20K or so.

As always, you can always spend less, the result may or may not be any good.

I am sure "Fat Eddy's Tatoo Parlor, Telemarketing & Collection Agency" could do it for a lot less than 20K. :)

yes, but are those numbers what it cost the polling firm to do the poll, or is it what they charged their 'client'.  im assuming they charge a newspaper or tv station more than it costs in order to make a profit.

isnt there an ethical problem with newspapers and news stations paying for polls?  they are essentially 'buying' stories, which is a big journalism no-no.

What I have quoted you are "retail" prices - that is about what a paper or TV station would pay.

Different clients get different prices however, depending on the number of polls, the PR value, what you can "piggy back" onto another poll, etc..

Is it "buying" a story - yes.

They typically argue that it is just fee for service, and hence not unethical.