« Weighting by Party | Main | No Mystery »

September 23, 2004

In medias res

Thank you, Mr. Kaus (and, for that matter, Mr. Reynolds as well).

Now that I have your attention, you may be wondering, who am I, and what is this blog about?

Starting a political polling blog right now is a bit like the opening of the first Star Wars movie. We are in medias res – in the middle of events – with bombs exploding and lasers blasting in all directions. Nonetheless, it is also a moment of peak interest in the workings of my chosen career. An unexpected confluence of events has provided the opportunity to act on an idea I’ve pondered for months. So ready or not, here we go.

I am a Democratic pollster, but my aim here is not to try to spin the latest poll results in any particular direction. There are many other voices in the blogosphere that perform that function admirably. My hope is to provide a bit of straight shooting – the same role I play for my clients – with respect to polling methodology and what the polls can and cannot tell us about the state of the campaign. I also hope provide a bit of criticism and fact checking on the good, bad and ugly of surveys in the public domain. That’s criticism in the formal sense, less about attacking and casting blame than about evaluating and providing an authoritative though, admittedly, subjective opinion.

I have spent most of the last 18 years as an apprentice analyst, senior analyst and ultimately a partner in firms that conduct surveys for Democratic candidates, working with some of the most brilliant pollsters and political strategists in America. Over the years, I have had the opportunity to design and analyze literally hundreds – perhaps thousands – of polls designed to plot strategy and track campaigns as they progress toward Election Day.

With one foot firmly planted in the world of applied polling, I have also dabbled in the academic side of survey research. I got exposed to the gods of polling at the Institute for Social Research (ISR) while a Political Science Undergrad at the University of Michigan. I completed work toward a yet unfinished Master’s Degree at Joint Program in Survey Methodology (JPSM) at the University of Maryland and have attended conferences of the American Association for Public Opinion Research (AAPOR). Perhaps more important, I have dog-eared my subscription to AAPOR’s academic journal, Public Opinion Quarterly throughout my continuing apprenticeship in political polling.

The moment seems ripe. Political surveys have never been more ubiquitous, yet never more challenged. On the Internet, the consumers of polling are swarming sites and blogs like realclearpolitics, ThePollingReport, kausfiles, donkeyrising, mydd, dailykos, pollkatz (to name just a few) to download the latest polling numbers and debate their meaning. Yet I am struck the pervasive lack of knowledge in the blogosphere of the most basic concepts of survey research. It has never been more important for those of us who poll for a living to do a better job of explaining what we do.

My business partner, David Petts, likes to say that campaign pollsters are the "plumbers” of survey research. He means that our role is distinct from the "architects,” the brilliant academic survey methodologists at major universities, government statistical agencies and private research institutions who advance the true science of survey research. Our work applies their designs to the political world the way plumbers install pipes in a new house designed by an architect. We sometimes apply our practical knowledge and compromise the design in ways the architects don’t like, but we also have a unique insight on the practical side of our jobs. Campaign pollsters also have a role distinct from the big media pollsters, who aim to inform the public rather than drive campaign strategy.

On a basic level, I hope to help political junkies of all stripes do a better job reading and evaluating polls. On a more lofty level, I’m hoping to foster greater communication between the architects, plumbers and consumers of survey research.

This blog will be, like most, a work in progress (I'm embarrased that the curtin is going up before I've had a chance to post a basic blogroll!) I'd welcome your comments, suggestions or criticisms via email. Do you have a question about polls that has confused you? Please, send it my way.

But enough for lofty goals. Back to the campaign!

UPDATE: Very well qualified reader WP writes to correct my Latin: "In medias res means 'INTO the middle of things.' It's the famous phrase because this is what Horace says in the Ars Poetica, that the storyteller should jump right into the middle of things, the way Homer does - rather than spending a lot of time with preliminaries." Ah...That's what I get for trusting Google. Thanks WP, I stand corrected.

Related Entries - MP Housekeeping, Polling & the Blogosphere

Posted by Mark Blumenthal on September 23, 2004 at 04:56 PM in MP Housekeeping, Polling & the Blogosphere | Permalink


The modifier is "Democratic," not "Democrat." The only people who say "Democrat candidate," "Democrat pollster," are wingnuts butchering the grammar to score some kind of political point that Democrats are not Democratic, but hacks who belong to a party. So why do you use it if you are indeed a Democrat? Maybe it's just a mistake, but spare the English language, spare the partisanship, and call yourself a "Democratic pollster."

Posted by: Democratic Hacker | Sep 23, 2004 5:15:13 PM

Yikes! The Hacker makes a good point. My apologies, typo corrected.

What's even more embarrassing is that both my DemocratIC wife and DemocratIC co-worker also missed it.


Posted by: MysteryPollster | Sep 23, 2004 5:40:06 PM

"I have spent most of the last 18 years as an apprentice analyst, senior analyst and ultimately a partner in firms that conduct surveys for Democratic candidates"

What an odd specialization.

I suppose this must be more common than not in your field but I never considered it before. I always assumed that polling firms were apolitical professionals hired by campaigns in the same manner that a corporation would hire a marketing research firm.

It seems to me very risky to hire an ideologically affiliated polling firm. Long term this could lead to unconscious biases creeping into the construction of the polls. How do you avoid this.

I say all this from the perspective of near pure perfect ignorance about how the political polling business works.

Posted by: Shannon Love | Sep 23, 2004 6:09:54 PM

I've always wondered what it means to be "a [Democratic | Republicanic :-)] pollster." Does it mean you work only for candidates of that party, or that you skew the results to conform to that party's point of view, or what? What's different from being just "a pollster?"

Posted by: Hank Fenster | Sep 23, 2004 7:16:25 PM

As a confirmed anarchist pollster (well, I teach it sometimes, anyway) it's great to have someone willing to declare his interest as the Brits would say. I'm sure I'll enjoy reading your blog.

My wife is a more-or-less ideological disinterested sort who works for a polling firm. Makes for all sorts of interest discussions when I am around her colleagues.

Good pollsters conduct good polls. End of story. A willingness to disclose your interest is great.

I'll tell my students about your blog and see if I can drive at least a little traffic your way. (Not as much as Instapundit, of course.)

Good luck.

Posted by: JorgXMcKie | Sep 23, 2004 7:48:42 PM

Excellent. I expect to learn a lot, and either confirm or readjust what I think I already know about polls. Ignore the hack complaints, it just means that somebody's watching.

Posted by: Pete | Sep 23, 2004 10:26:47 PM

Welcome to the blogosphere, Mystery Pollster. You are right that there is a lot of spin, and a lot of misunderstanding, regarding polls. I try to cut through that as much as possible on my site. I hope that if you find me getting things wrong you call me on it pronto.

Posted by: Gerry | Sep 24, 2004 8:47:30 AM


The reason for specialization is trust. To design appropriate questions for surveys, campaigns can't help but reveal something of their strategies and tactics to pollsters. Given the possibility for damaging leaks, party consistency is at least one way to be a gain a bit more comfort -- though leaks still occur.

Posted by: Rob | Sep 24, 2004 11:22:06 AM


Thanks, that makes sense. I still think it is a dangerous practice but given the security tradeoff I suppose most campaign find it necessary.

Posted by: Shannon Love | Sep 24, 2004 11:32:46 AM


Great to see you----

Any blog recommended by Kaus is going to be a major stopping place for me.

Posted by: Catherine | Sep 24, 2004 1:54:56 PM

I appreciate hearing from another political pollster. I sometime goggle at the AAPOR list because of the strong prejudice against politics actually creeping into political polling.

As a Republican consultant, I can answer the questions about what it means to be a partisan pollster.

It does not mean that we attempt to skew results to favor our party or our candidate. Rather, it means the opposite. Our clients rely on us to give them valuable and accurate information by which they can develop their message and target their paid media. Political pollsters who consistently tell their clients what the clients want to hear usually have losing records.

But I do bring something else to the table as a professional who concentrates on a single party's candidates. I bring an institutional understanding of the primary electorate in my party and especially in my state. If you want to run a winning campaign, hire a seasoned political consultant who has a depth and breadth of knowledge about the electorate, rather than the University pollster who produces pretty charts and PowerPoint presentations.

I typically run 50 or more surveys per election cycle, most of them in the primaries. Most races in the South, at least, are decided in the primary.

A strong understanding of the electorate also helps me identify when polling results don't "feel right", where someone less specialized doesn't know the electorate intimately.

An additional asset I bring to the table is that I also buy media for clients, so I understand the targeting they will be doing and can translate polling results into concrete strategic terms and help them develop their message and the paid media targeting to help them win their elections.

I have seen time and again how brilliant pollsters fail their political clients by failing to translate their findings into concrete political terms that can be acted upon.

Finally, someone who polls exclusively in political races will often have a better time figuring out what are the right questions to ask to draw out useful data. If you've got a brilliantly-designed questionnaire and flawless fieldwork, you can still make the biggest mistake of not asking the most important questions. Immerding myself in politics helps me know what the right questions are in order to get the strategically-valuable information my clients pay me for.

Another reason for working specifically for a single party is that it helps minimize conflicts between clients. After each round of redistricting, our firm always ends up with a number of clients running against each other because their districts were combined or they are both seeking a higher office. If I worked for both parties' candidates, the conflicts would increase greatly.

Now, if I could only get my home state to have off-year elections so I could put food on the table every year.

Posted by: Todd Rehm | Sep 26, 2004 12:21:31 AM

The comments to this entry are closed.