The first being shrink-wrapped software.

How would I run such a company? I would want to deliver to my customers different things, depending on who they were. To the elite media outlets, I would give them polls that showed the ways in which Democrats might win, and that Republicans had serious weaknesses. That fits the story line they are looking for, and there’s plenty of data for it. To the kool-aid drinkers on both (or all) sides, I’d dish up polls full of kool-aid: another analogy for this segment is the cosmetics business — we’re selling hope.

During the summer, I would particularly hype the chances of the Democrats. Kaus:

Guess It Really Was A Nagourney Problem: Richard Stevenson appears instead of Adam “Caterpillar” Nagourney in the NYT writing credits for the latest Times/CBS Poll. Have Nagourney’s superiors realized how awful his last effort was? … Whatever the reason for his absence, it’s not the same without him. There’s no cocooning pro-Democratic spin! Just a sensible account of a poll with some good news and not-so-good news for the Dems. Even the headline (“No Poll Boost From Edwards”) is unspun–at least not in a pro-Dem direction. This can’t go on. …It didn’t. The Web headline has now been changed to “Public Likes Edwards, But Race is Still Close.” The hed in the print edition is the even more morale-boosting “Public Warms to Edwards; Race Still Close.” Shift change at the copy desk? Or orders from Moscow? You make the call.

Rasmussen has it 44/39 for the Dems in the generic congressional poll now. This has always struck me odd, since no serious observer believes that the Democrats have a prayer of taking congress. When asked about this, Scott responded with the following email:

Let me start by addressing the Generic Congressional Ballot. Dems always (or almost always) poll better on this question before Labor Day. So, I expect that gap to decline. Second, given the small number of open seats, even the current data would not lead to a Democratic victory.

As for our Likely Voter polling, we control geographic distribution of the calls through our calling program. We ask a series of screening questions to determine who is a likely voter and weight the data by age, race, gender, and political party. Periodically, we test these assumptions and measure the impact of a tighter or looser screen.

I am confident that Scott Rasmussen is an excellent pollster. Let me tell you, however, what I would do if I were running his business. After Labor Day, it is time to wean the true believers from their kool-aid. Polls need to begin to be realistic, ever so gradually, so that the pollster has a company and a job in two years. Hence, our sample of “likely voters” needs to become a little bit tighter so that it actually mirrors, um, likely voters. And, mirabile dictu, slowly the generic congressional poll begins to move toward the GOP.

This, by the way, is among my key indicators in assessing the Bush/Kerry horse-race polls. If a poll shows Kerry winning among likely voters (forget adults and registered — those tell you zero), then how much of a lead do generic congressional Democrats have? If Kerry is ahead by two and the generic congressional democrat is ahead by seven, Kerry is mega-screwed.

You have to get real by the end, but never forget to sell kool-aid to the kool-aid drinkers during the hot season. What a great business!

This entry was posted on Thursday, August 5th, 2004 at 7:11 am and is filed under General. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.