Posted on 10/26/2012 12:17:01 PM PDT by GOPFlack
Part II – Media Polls – What Are They Good For?
Let me start by drawing a quick distinction that I think is important – no one should confuse a public pollster who publishes results like Gallup or Rasmussen with a media sponsored poll that is, of course, made public. So when I’m speaking of media polls here I’m talking about polls paid for by “news” outlets. The first and most salient point is that a news outlet could get private polling data for its own edification about a race – either from a campaign or on their own. But that would assume two things: 1) that they actually care about the truth, and 2) that they would consume and or pay for information not for dissemination. If any of us believe that after all these years we can stop here.
Back to my point about campaign polling pros in Part I - they will virtually never do media polls. I have seen them do polling for groups who hire them under the assumption that the poll will be released if the data is favorable to the client. But on balance, they will not allow their work to be sponsored and or reported on by a news outlet. Why? Because the media client has only one interest in the poll – use it to make news and increase viewers, web hits, or readership. Campaign pros use polls in the Summer or Spring to test messages, probe a candidate’s or potential opponent’s weaknesses and strengths across a broad array of metrics. If you have ever seen an actual field poll book, it usually a three to four inch binder with reams of information. Usually, in addition to what are commonly called crosstabs and the actual polling info question by question, there is some sort of topline summary provided by the pollster where he picks out the salient facts that he believes are advantages and or problems for the client, where the opponent is strong or weak (by issue, geography, demography, etc) and any macro trends that might be developing (enthusiasm, issue importance, etc.).
The point I’m making is that the campaign pro has a totally different objective than a media outlet – he isn’t polling until very late in the campaign to see if the client is truly winning – he is polling as a means of recon and strategic planning for the campaign and then to see if the battle plan is working and adjust to the opponent’s counter moves. No one “wins” until the votes are cast. But you can lose before then if you don’t poll accurately and deeply to find your vulnerabilities or the opponent’s in order to have a battle plan. The media, in contrast wants to excitedly declare who is “winning.” Even if it is July or August! Any self respecting pollster would never compromise themselves this way - trying to make what is in essence a prediction 4 months before election day. It’s like declaring the war over before the first shot is fired. The pros are like are intel officers, analysts and forward observers. The best are also strategic planners as well. So who are the so called “pollsters” who do media polls?
First and foremost they are people who want or need to get paid and usually aren’t well thought of enough in the private polling world to get campaign work. Hence they are generally universities, marketing research firms (who usually sell soap to focus groups but want to pick up easy money in an election year), or local guys in states who will do work for a local TV station or paper. The explosion of media polls has also led to an explosion in the number of people who claim they are political pollsters. As with any boom, quality often suffers during the rush to an easy dollar. A real campaign pollster has an interpretive eye toward his data and an experience base to pick up anomalies or trends. He is more than just a recorder and reporter of raw data. That’s why the media polls are so bad on balance – they use bad pollsters who are out of their depth in campaigns. In some cases they are just flat out corrupt because they won’t bite the hand that feeds them or have an ideological agenda.
Let’s dissect this a bit more. My favorite whipping boys in the media poll circle are university polls. My first experience with a university pollster came in my first political race – a statewide race in 1986. I was a college volunteer (although there were only 6 of us, the campaign manager and a press person) in a long shot race. The state’s largest newspaper was sponsoring regular polls from a professor in the political science department where I was a student. He wasn’t a bad guy, but my impression was that he, like all my professors, had no clue how politics and campaigns actually worked. Plus, he, like most professors back then, was poor as a church mouse. He showed us consistently behind by large margins all the way through the end even though we squeaked by with a narrow margin in the final tally. Why? I think two factors – one he had no clue what was happening on the ground the last 4 weeks of the campaign because he had no ability to properly discern changes in his data that were real. Second, his sponsor had it out bad for the candidate I was working for and was working overtime to defeat him on their news pages.
Put yourself in his place for a second. You are the only pollster in the state being paid by a paper with deep pockets and you are a poor associate professor. Do you really want to give them data that doesn’t bolster their agenda and risk that extra income? It’s like the global warming racket funded with government grant money today. What professor is going to change their mind and go against the consensus and risk that money? Again, I don’t think this professor was abjectly corrupt and partisan (unlike the hack who runs Marist – see recent Breitbart piece), but he was absolutely cooking his data. I was told by folks in the department that one night late in the race he threw an entire night’s sampling in the trash because he declared it an outlier. It didn’t fit the narrative. Now if this was going on in 1986, what do you think is happening now when networks are paying small liberal arts universities millions to produce this stuff? It’s worse because of bigger money, publicity, and the over the top ideological agenda of the institutions doing the work and their sponsors. They are so to speak “all in.”
In addition, they want to show who is “winning.” As I said that is impossible because that isn’t how elections work from the inside out. So, when one of the university boys gets a sample set back that shows Romney actually pulling bigger numbers than Obama in early August, they just adjust the poll back to fit the narrative by reweighting it based upon their guess, or in most cases hope, of the partisan breakdown of the electorate. Their lack of experience in campaigns, the needs of their liberal sponsors, money, and basic ideology make it inconceivable to them that Obama’s magic ground game and super smart campaign team won’t get those turnout numbers where they want them by election day. Hence, the mystical reweighting magic is applied. This is not polling – it is prognosticating. And political scientists are generally the worst prognosticators in the world.
The other polling firms doing media work aren’t much better. They are usually commercial pollsters who use the election branding of media work to sell additional research in the commercial or policy space. In short, there are too many people who are calling themselves political pollsters who don’t have the expertise or experience, and are providing a product that is virtually useless except for its intended purpose – drive news coverage and satisfy the agenda of the corporate client.
What can I say about these polls that is good? Well, as is often pointed out, they can detect trends. I don’t put much stock in RCP averages because I think the polls they use and the staleness of them is an unsound methodology for predicting final outcomes. But it can show trend lines in the aggregation as a race enters the last month. That is the best I can say about why we should look at them.
One other item in passing, these pollsters don’t account for campaign resources and other intangibles that all go into how a candidate will manage and message his or her race. Just because John Doe is losing by 20 in July, it might not account for the fact that he is sitting $25 million to unload on a weak incumbent opponent the last 12 weeks. The race hasn’t even been engaged at that point. No media poll can account for these factors.
This ran a bit longer than I intended so I will get to robo polls – their utility and limits in the next piece including why PPP is not just a joke, but a warmed over propaganda outfit for the SEIU. And, why I believe Rasmussen has the most integrity of all of them.
Finally, I appreciate all the kind comments and hope this helps. I will be on the softball field all weekend with my daughter’s travel team unless Sandy pre-empts us so I probably won’t get around to the other stuff until Sunday evening if you still want it.
thanks for the ping on part ii; please ping me again on part iii.
Thank you GOPflack. I enjoy inside baseball.
PS Keep me on your ping list.
Please ping me on part III, this is great stuff.
Please ping me on part III, this is great stuff.
Thanks!
Poll Ping.
Thanks!
Poll Ping.
Thanks very much for this; informative and comforting.
Please also include me on your ping list, if you would.
This is excellent. Have done Part 3 yet?
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.