Talk Elections

Just out of curiousity:

Can anyone with any experience in polling give any ballpark figures of how much would it cost to commission a telephone poll of 625 likely voters?

Good question.  I've never seen any numbers on that.  Perhaps Vorlon could give us a ballpark figure.

Depends a lot on bot-or-not.

Whichever is the cheapest (I don't know why people dislike Robo-polls because SUSA and Rasmussen have a very good reputation).

think the Atlas should commission a poll, and spend a month weighing it?

Anyone interested can call SUSA and try to see how much it costs to commission them (SUSA only does polls for hire): 1-800-786-8000

I've done them before.  $8,000-$15,00 for live pollers, $4000 for a bot.

If we all pool our money together, we can pay for a poll. To make things easier, one of us should collect all the money. I guess I could be that person....

I might be willing to chip in a little for a poll....but only if we ban BRTD from helping to formulate the poll question.  I don't want "Nader", "suburbs", or "strip clubs" to be mentioned in the poll.

Actually, we should keep gporter away as well.  Otherwise, we'd end up with poll questions that tell the respondant "Please give a realistic answer, not a Democratic hack answer."

Answer to a person responding they'd vote Dem

"You are a Democratic hack"

Answer to a person responding they'd vote Nader

"You're a fascist scumbag"

Using offshore outbound centers, cost can be reduced by 40%.  If the center is only making calls and asking questions the cost can be considerably less than $4,000.  Here are some of the costs that go into polling by phone:

1.  Development of questionairre (client can provide this)
2.  List providing.  This is very important.  No poll is better than that data called and the outline of the respondants.  In other words, there needs to be work done to make sure the responders are a statistically representative sampling of the voter pool.  For instance, calling in NC to "likely Democratic voters" will give you a far different result than insuring that the responders are statically representations of subgroups such as age, income, race, education, rural/urban etc.  The more accurate the responders mirror the target voter pool, the more accurate the poll.  The more complex, the more calls need to be made to get the required number of "completes."
3.  Number of attempts, completions and length of call.  The call center ultimately has to reduce the project to number of hours dialed.  Domestic outbound is about $15 per hour.  Offshore about $10 per hour. (minimums and set up charges may apply)
4.  Development of data capture.  If the client already has a portal or questionairre software this expense can be avoided. 
5.  Compilation of results.   

625 is a pretty large sample, so I'd concur with the $4k ballpark.  I've commissioned live operator, accent-neutral polls with a smaller sample for around $2k.

Yeah! Let's actually do that for $2k.

I won't be able to help, of course.

Or you could do a poll for nothing by e-mailing the same questions to 1,000 random YAHOO acounts (using the YAHOO search mechanism) and see how many replies you get?

It would be really interesting if we could commission our own SUSA polls, but $4000 (or even $2000) is a lot of money to raise from at most thirty or so people who'd be willing to chip in. And that's just for one poll; we'd need more if we wanted to, say, get good state polling.