It's hard to imagine a political campaign without a good pollster in the room. It's easy to get basic data on what people think, but, for example, if I were a candidate pushing health care, I would want to know everything there is to know about public attitudes on health care and policy, and who trusted messengers are and what language works, and what constituencies I'm reaching, and all that. Good pollsters provide that context, and if they're very good, they also help you understand which views are deeply held and which can be vulnerable to persuasion.Quite so. But what I was really trying to get at was the idea of the "superstar pollster," someone like Mark Penn who not only polls but also acts as a key strategist for the campaign. I wanted to suggest that this is bad for two reasons.
First, sophisticated polling of the kind that Penn does is far more common than it was 30 years ago. There are lots of people in the consumer marketing biz (and elsewhere) who know how to do it, and this means that when you go looking for a pollster you're not limited to ten people in the whole world who truly know how this stuff works. These days, you can hire a mid-level staffer at a mid-level salary to do your polling if you want to.
Second, polling strikes me as an area where you can't afford to kid yourself. An adept number cruncher can twist big datasets to tell almost any story at all, and this means that a smart candidate will prefer someone who at least tries to do their best to look at the numbers honestly and tell you what they say. With a superstar like Penn you get exactly the opposite: he's almost dead certain to massage the numbers in order to produce results he likes. You'd be a lot better off hiring some nerdy statistics guru who's going to give you the numbers straight, and then hiring a completely separate stable of consultants who spend their time arguing over what the numbers mean.
As it happens, this is more or less what Mark also says in the rest of his post. It's worth reading.