Poll Jam
2014-10-16 - by
Francis DeLuca at Civitas took issue with my blog about their polls (“Cooking Numbers at Civitas.”) He tweeted (@fxdeluca): “@jgaryp thanks for the follow! But don’t appreciate your comments about Civitas polling! #youknowbetter.”
Here’s what I know. When the late Jack Hawke was at Civitas, he worked hard to give their polls credibility. But that credibility was stretched thin when Civitas put out a “flash poll” saying Tom Bradshaw (who I’m working for) led John Alexander by 52-36, then hours later sent out “a corrected version of today’s flash poll,” saying the lead actually was 10 points. No explanation was given for the correction.
Sorry, Francis; it smelled fishy.
This is just one example of why the plethora of polls this time in a campaign is more confusing than enlightening. There are so many polls in the news. Sometimes they’re consistent; sometimes they vary widely. How do you sort through them?
I’ve been looking at polls for 40 years now, and here’s my advice. First ask whether a poll has an institutional attachment (or big funder) that might influence the findings – or how the findings are reported. Then look at whether it used live callers or automated calls (IVR). Robo-polls can’t call cell phones, and 20-25 percent or more of voters may use only a cell phone.
Finally, ask yourself: Was the sponsor motivated to do the poll to get good information – or just get publicity for themselves? Would they spend the extra money it takes to do a good poll, or do it on the cheap to get a headline?
And a word to candidates and their campaigns. If you’re spending the money it takes to get good polls – and you should be – take public polls with a huge helping of salt. Take this advice from Senator Michael Bennet of Colorado, who won in 2010 after public polls showed him losing, though his own pollster had shown him ahead: “Trust your pollster. If you don’t, you’ll go crazy.”
Posted in General
Poll Jam
2014-10-16/
Francis DeLuca at Civitas took issue with my blog about their polls (“Cooking Numbers at Civitas.”) He tweeted (@fxdeluca): “@jgaryp thanks for the follow! But don’t appreciate your comments about Civitas polling! #youknowbetter.”
Here’s what I know. When the late Jack Hawke was at Civitas, he worked hard to give their polls credibility. But that credibility was stretched thin when Civitas put out a “flash poll” saying Tom Bradshaw (who I’m working for) led John Alexander by 52-36, then hours later sent out “a corrected version of today’s flash poll,” saying the lead actually was 10 points. No explanation was given for the correction.
Sorry, Francis; it smelled fishy.
This is just one example of why the plethora of polls this time in a campaign is more confusing than enlightening. There are so many polls in the news. Sometimes they’re consistent; sometimes they vary widely. How do you sort through them?
I’ve been looking at polls for 40 years now, and here’s my advice. First ask whether a poll has an institutional attachment (or big funder) that might influence the findings – or how the findings are reported. Then look at whether it used live callers or automated calls (IVR). Robo-polls can’t call cell phones, and 20-25 percent or more of voters may use only a cell phone.
Finally, ask yourself: Was the sponsor motivated to do the poll to get good information – or just get publicity for themselves? Would they spend the extra money it takes to do a good poll, or do it on the cheap to get a headline?
And a word to candidates and their campaigns. If you’re spending the money it takes to get good polls – and you should be – take public polls with a huge helping of salt. Take this advice from Senator Michael Bennet of Colorado, who won in 2010 after public polls showed him losing, though his own pollster had shown him ahead: “Trust your pollster. If you don’t, you’ll go crazy.”
Posted in General