Just prior to Colorado’s recall election, the liberal-leaning pollsters at Public Policy Polling Company conducted a poll of likely voters in Angela Giron’s district (Senate District 3). The poll showed Giron losing by 12 points, a finding that we now know was spot-on but seemed bizarre at the time … so bizarre, in fact, that PPP chose not to publish it.
Now PPP is receiving fire not only from some conservatives but also from FiveThirtyEight editor Nate Silver for its decision to withhold those results from the public.
@ThePlumLineGS @ppppolls: But I'm especially skeptical when a pollster puts its finger on the scale in a way that matches its partisan views
— Nate Silver (@NateSilver538) September 11, 2013
PPP acknowledges that it routinely withholds results from the public, especially if it there are red flags in the results:
We are a private polling company. 90% of the polling we do is never released to the public, just like other private polling companies…
— PublicPolicyPolling (@ppppolls) September 11, 2013
The number of polls we've done over the years we didn't release where a Democrat was winning far exceeds ones where a Republican was…
— PublicPolicyPolling (@ppppolls) September 11, 2013
We were concerned when our Colorado recall poll showed 33% of Democrats supporting the recall that there may have been confusion about the q
— PublicPolicyPolling (@ppppolls) September 11, 2013
And decided to release poll today because of interesting disconnect between support for specific gun policies/election result
— PublicPolicyPolling (@ppppolls) September 11, 2013
We take a vote on our website every week on the polls that we are going to conduct that are intended for public consumption…
— PublicPolicyPolling (@ppppolls) September 11, 2013
And have released every single poll we've ever done that was conducted for the purpose of releasing it to the public
— PublicPolicyPolling (@ppppolls) September 11, 2013
@jennagiesta we do a post every week announcing what polls we're doing that weekend for public consumption and this wasn't one of those
— PublicPolicyPolling (@ppppolls) September 11, 2013
@jennagiesta Because of unique circumstances- first legislative recall ever in CO- did not want to commit to doing public poll
— PublicPolicyPolling (@ppppolls) September 11, 2013
@jennagiesta first recall ever, there's been lots of voter confusion, 33% of Dems supporting recall is red flag your data might be off
— PublicPolicyPolling (@ppppolls) September 11, 2013
@jennagiesta We made a decision not to release the poll because I didn't trust that 33% of Democrats supporting the recall was correct…
— PublicPolicyPolling (@ppppolls) September 11, 2013
@jennagiesta and that there may have been confusion about the question wording. Obviously that wasn't the case
— PublicPolicyPolling (@ppppolls) September 11, 2013
Here's an explanation of our public polling prodecures and why we handled our Colorado recall poll the way we did: http://t.co/U5Z0cQN4lX
— PublicPolicyPolling (@ppppolls) September 11, 2013
PPP specifically addressed Silver’s complaint:
@fivethirtyeight Nate I'm sorry but that is absurd. You're saying you would put out a model if you had serious concerns that it was wrong?
— PublicPolicyPolling (@ppppolls) September 11, 2013
@fivethirtyeight We were polling a race where there was no public data- found a counter intuitive result- 1/3rd of Dems supporting recall
— PublicPolicyPolling (@ppppolls) September 11, 2013
@fivethirtyeight So we made a decision not to release data on a race that no other pollster released data on either
— PublicPolicyPolling (@ppppolls) September 11, 2013
@fivethirtyeight Is it your belief that any pollster that has ever released a poll publicly should then release all polls it ever conducts?
— PublicPolicyPolling (@ppppolls) September 11, 2013
@fivethirtyeight @ThePlumLineGS Not releasing a poll we never announced we were conducting is not 'putting our finger on the scale'
— PublicPolicyPolling (@ppppolls) September 11, 2013
Silver isn’t backing down:
@ThePlumLineGS @ppppolls: If you suppress "outliers" you don't like but tolerate those you do, you wind up with a very biased average
— Nate Silver (@NateSilver538) September 11, 2013
.@ppppolls : We design our models to be robust in the first place. Then we publish ALL results from them, yes.
— Nate Silver (@NateSilver538) September 11, 2013
@ppppolls: Yes, so express your cautions in your write-up of the poll. Don't suppress the data.
— Nate Silver (@NateSilver538) September 11, 2013
Update:
Washington Examiner reporter Rebecca Berg has only one question:
Who will poll FiveThirtyEight vs. PPP?
— Rebecca Berg (@rebeccagberg) September 11, 2013
* * *
Update:
PPP has apologized for defending itself in public in response to Silver’s (very public) attacks:
Apologies to @fivethirtyeight – I think his criticism of us is unjustified but no need for a big fight in public #MovingOn
— PublicPolicyPolling (@ppppolls) September 11, 2013
And this guy makes a good point:
Lesson of the day: Poll aggregators would like survey firms to share all of their data. Preferably for free. @ppppolls @fivethirtyeight
— Drew Linzer (@DrewLinzer) September 11, 2013
* * *
Related:
Bombshell: Tancredo polling even with Hickenlooper in Pueblo, Colo.
