washington, dc

The Democratic Strategist

Political Strategy for a Permanent Democratic Majority

BlogiStan (Greenberg)

by Scott Winship
I’ve been meaning to plug a number of pieces by organizations that one of my bosses (Stan Greenberg) heads up. First, those who are sick of me intimating that Democrats ought to moderate their positions should definitely check out “How Democrats Can Use Polling to Win Elections,” by Amy Gershkoff of Greenberg Quinlan Rosner Research. Amy argues that Democratic pollsters and polling audiences need to ape Republicans not on issue positions, but in the way they design and use polls.
Amy offers three recommendations:
Collect data on issue importance and then analyze the voting patterns of the electorate by issues rather than by demographics. No more, “Married blue-collar Catholic men vote for X.” Instead, consider how people for whom, say, Iraq is the most important issue allocate their votes. This approach filters out data on issue preferences that are only weakly held and focuses on the issues that really affect voting.
When attempting to woo Independents, emphasize only issues that are of critical importance to them. It turns out that these issues are more likely to be terrorism and red-button values issues rather than health care and jobs. Similarly, when trying to increase Democratic turnout, campaigns should also focus on their critical issues.
Focus on issues that are extremely important to lots of people (rather than just somewhat important) and where the Democratic position is the most popular. Attempt to increase the salience of these issues among voters for whom the issue is only somewhat important (e.g., Iraq in 2004).
This accords with my distrust of lots of demographic segmentation analyses. A pollster might find that married women favor pulling out of Iraq by a 60 to 40 margin and then recommend that Democrats emphasize a pull-out so that they can improve their performance among married women. But it may be that many of those pro-pull-out voters were going to vote Democratic anyway, and so if one really wanted to improve one’s performance among married women, one should emphasize staying the course (to woo some of the anti-pull-out women. That would only make sense if a stay-the-course position didn’t actually lose voters who otherwise would have supported the candidate, but without knowing anything about how many and who finds Iraq extremely important, it’s not obvious which strategy to take.
The second Greenberg-related piece is a Democracy Corps memo [pdf] written with James Carville, “Getting Heard: Points of Engagement for a Change Election.” Greenberg and Carville argue that to maximize chances of winning back control of Congress, Democrats need to advance “ideas and critiques that are newsworthy”, promote an agenda that portrays them as “agents of change”, and elevate the importance of economic issues in the campaign. In terms of specifics, they advocate linking Congressional pay raises to minimum wage increases, making consideration of hot-button values issues contingent on passing legislation to reduce economic insecurity, pushing for Congressional oversight of Iraq spending, and repealing corporate tax breaks. They also suggest a college tuition tax credit, changing energy policy to focus on alternative fuels, and allowing Medicare to negotiate prices with drug companies.
The call for repealing corporate tax cuts is a bit incongruent with their finding that the biggest fear voters have of Democrats is that they will raise taxes. On the other hand, Greenberg and Carville do call for middle-class tax cuts (the college tuition tax credit), and the proposed Iraq oversight would likely promote reduced spending.
Finally, Democracy Corps has also released an analysis of public polling, by Karl Agne, that provides nearly 50 pages of polling questions from various organizations on all of the hot political topics of the day. The basic conclusions are that things haven’t changed much for a few months and that that’s good new for Democrats, given how lousy things look for Bush and the GOP. Polling junkies go nuts. (Incidentally, you can google the Democracy Corps website to find these comprehensive summaries going back several years. Definitely a valuable resource if you’re interested in a particular topic that isn’t currently in the news or if you are looking for trend data.)
Time for more coffee. Did you know that Starbucks has the most caffeinated coffee on the market? Did you know that you can order a “short” coffee, latte, or cappuccino even though that option’s not mentioned anywhere by Starbucks? Can you guess where I do most of my blogging?

2 comments on “BlogiStan (Greenberg)

  1. Chris Glaze on

    Gershkoff’s piece is excellent. A cogent analysis of how the GOP has had an advantage on minority-supported issues. She also has an interesting prescription, that Dems should convert “somewhat important” voters to “very important”. A tall order but worth taking seriously, and at least a good starting point for thinking about implications.
    I am also skeptical about the current emphasis on demographic segmentation, especially the microtargeting that uses consumer information. These are brute-force methods with tons of dimensions, very unconstrained. Things like CHAID and neural nets can do great things, but if you don’t understand *why* they give the results they do, they may not be as useful to strategy over time.
    Gershkoff has shown a strong result and a critical component to variability in voting behavior that gets at some of the underlying causes.

  2. Sam on

    I have long questioned the methodology of democratic pollsters. Though I agreed with a lot of Ms. Gershkoff’s article, I still think that her way of looking at the data is clumsy. In the past, I have asked several pollsters why regression analysis is used so infrequently in polling. They have answered with two reasons: First, that analysts are often untrained in this type of sophisticated statistical technique, and second, that it is difficult for clients to understand. These should not be barriers to the use of a better technique. Regression analysis can allow us to see if a factor actually determines vote choice, WHILE ACCOUNTING for the impact of all the other significant variables. Tabular analysis still has a multitude of uses, but to find out what actually drives a voter’s vote decision, only regression analysis can do the job (especially when combined with predicted probability software like Gary King’s “Clarify”).


Leave a Reply

Your email address will not be published. Required fields are marked *

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.