1 of 5 12/13/ :59 PM

Transcription

1 Make This My Home Page Search Advanced Search PRINT EDITION In This Issue Welcome MARK WATTS, Logout Subscriber Info Change Your Profile ---- Print Edition --- Features Inside Politics Home > Consultants' Corner > Article Print People & Organizations Consultant Sign Ups Movers and Shakers Consultants' Corner Vol 25, NO 10, December-January 2005 Voting Matters Opinion Online Campaigner Consultants' Corner International Politics Editor's Note Letters Campaign Doctor WEB EDITION On the Record Political Oddsmaker Events Political Trivia Books Off The Record RESOURCES Political Pages The Political Jobline Blog Directory C & E Polls ABOUT C&E On Staff Contact Us Customer Service Advertising Articles/Reprints Seminars Don't Hang Up: Use Quality Phone Banks to Get Out the Vote By :- By Janet Grenzke and Mark Watts It is not at all surprising that a new book called Get Out the Vote! would attract attention from campaign professionals this year. It may be that in the end, what made the difference was who did the better job at getting out the vote (GOTV). Yale University professors Donald Green and Alan Gerber s Get Out The Vote! tries to answer the question of what works and is most cost effective for GOTV. As former academics trained in research methods and published in academic journals, we are hungry for good empirical research that tests the relative effectiveness of different political campaign tools. To their credit, Green and Gerber are among the few in academia who pose questions relevant to practitioners and attempt to answer them in a real world setting. In this way, they clearly establish that interaction with individual voters is decisive in GOTV. That said, defects in Green and Gerber s methodology limit the usefulness of their findings on the ground in political campaigns. The central finding of Get Out The Vote! [reviewed in Campaigns & Elections, June 2004, p. 42] is that a personal approach to mobilizing voters is generally more effective than an impersonal approach. The authors conclude that door to door canvassing and unhurried phone calls are the most cost effective mobilization tactics. They emphasize that the quality of the phone call makes all the difference. However, Green and Gerber broadly, and in our view wrongly, generalize that high quality commercial phone banks are less cost effective than canvassing. They are on more solid ground in concluding that recorded calls and short, mechanical calls have almost no impact. Certainly, few in our field will disagree that face to face contact is better than phones, and a real person on the phones is better than an automated voice. The important question is whether a high quality commercial phone bank, with an affordable 30 to 1 of 5 12/13/ :59 PM

2 45 second message (about $1.00 per call) is cost effective compared with door to door canvassing. In order to usefully compare voter contact approaches, Gerber and Green need to establish the true relative costs and conduct experiments with sound research designs that replicate real campaigns. In these two respects, Green and Gerber fall short. Their conclusions betray basic misunderstandings about why and how GOTV efforts are conducted and to whom they are directed. Several of the experiments cited by Green and Gerber use reputable phone banks that deliver messages that run between 30 and 60 seconds. Based primarily on their most comprehensive experiment in 1998 in New Haven and West Haven ( The Effects of Canvassing, Direct Mail, and Telephone Contact on Voter Turnout: a Field Experiment, American Political Science Review, vol. 94. no. 3, 2000), and their follow up experiments in congressional elections in Iowa and Michigan in 2002, they find no indication whatsoever that telephone appeals raise turnout and that the impact of GOTV phone calls were not statistically distinguishable from zero. We reviewed those studies mentioned above, as well as more than a dozen other articles and unpublished manuscripts on phones referenced in their book. All of these studies are flawed in one or more of the following ways. The Importance of Whom You Call and What You Say Effective GOTV campaign messages are aimed at a partisan audience. A campaign wants to get its voters out to the polls. Once it reaches a likely supporter of the candidate or initiative (identified by party registration, polling or a previous phone id call), the campaign delivers a partisan message tailored to get that voter to the polls. Such scripts stress the importance of each vote and cite specific issues that appeal to the potential voter s self interest, such as abortion, civil rights, crime, funding of local schools, gay marriage, the war in Iraq, etc. On p. 78, Green and Gerber write the effectiveness of professional phone banks has little to do with the reasons for voting stressed in the scripts. However, the GOTV scripts cited by Green and Gerber are not written specifically for nor are they delivered to a partisan audience. Moreover, most of the scripts are fatally flawed because they do not appeal to the voter s self interest. In order to engage the voter, it is critical that the caller deliver a targeted, well crafted message that tells the voter how the election outcome could affect his or her financial security, health care, educational system or another matter of self interest. In much of Green and Gerber s research, the standard introduction identifies the caller from a nonpartisan group such as the League of Women Voters, or Vote 2002 and the script concludes with a variation of We hope you ll come and vote. The following non partisan messages constitute 2 of 5 12/13/ :59 PM

3 the core of these scripts. We want to encourage everyone to do their civic duty and exercise their right to vote. Democracy depends on the participation of our country s citizens. Each year, some election is decided by only a handful of votes. Who serves in important national, state and local offices depends on the outcome of the election, and your vote can make a difference on Election Day. Politicians sometimes ignore a neighborhood s problems if the people in that neighborhood don t vote. When politicians see a lot of people turning out to vote, they know they should pay attention to issues important to people who live around here. Green and Gerber acknowledge that in the only large scale experiment before their own 1998 research, William C. Adams and Dennis J. Smith wrote in a 1980 Public Opinion Quarterly issue that partisan phone calls were effective in mobilizing voters. More recent experiments by Green, Gerber, John E. McNulty, David Nickerson and their colleagues begin to assess the effectiveness of paid phones with issue based scripts tailored to a target audience. As summarized below, the results of these studies are mixed, and the results are muddied by the methodological problems inherent in experiments in campaigns. While these more recent experiments are more relevant to practitioners, we could not find any reference in Green and Gerber s book to a GOTV experiment with paid phones that asked the voter to vote for Mary Jane, because... The typical GOTV call promotes a candidate and works in conjunction with media messages. Since it appears that none of the experiments cited in their book assess GOTV calls that mention either a candidate or a party, they certainly don t form the basis for concluding that paid phones don t work. Problems In The Research Design Experimental research in campaigns is necessarily plagued by serious methodological problems. For example, Green worked with the NAACP National Voter Fund (NVF) to assess the impact of its massive phone and direct mail GOTV program into lists of likely black voters in 10 states during the 2000 election (5.2 million calls and 6.7 million mail pieces). Two of the phone scripts were partisan and powerful, with references to hate crimes. The experiment was designed so that one group of voters (the treatment group) would receive phone calls and mail, and a smaller group (the control group) would get neither mail nor phone calls. However, in the real world of politics, it is very difficult to coordinate such experiments. In four of the 10 states, about one half of the control group (between 45 and 52 percent) mistakenly received a GOTV call. In these same states, between 33 and 49 percent of the treatment group were mistakenly never called at all. Gerber and Green 3 of 5 12/13/ :59 PM

4 report that phones were marginally effective, but the errors are so great that even with regressions that attempt to factor in these errors the conclusion is far from definitive. Errors in the research design and statistical models used to analyze the data make a huge difference in the conclusions. For example, Imai (2004) corrects errors in Green and Gerber s 1998 New Haven study and finds that even with weak, nonpartisan scripts, phones increased turnout by 5 percent. In their article, Green and Gerber concluded that phones had a slight negative impact. The Cost of Apples and Oranges Green and Gerber favor person to person canvassing over paid phones based on their estimates of the cost effectiveness of each approach. Because Green, Gerber and their colleagues have no experience running GOTV firms or campaigns, their comparative estimates of phones versus canvassing are flat out wrong and grossly misleading. Gerber and Green conclude that top of the line paid phones would cost the campaign about $45 per vote. This is based on a variety of experiments that reveal tremendous differences in their estimates of the cost per vote, ranging from: $20 per vote to $37.50 per vote in their Youth Vote 2000 study; $89 per vote to $142 per vote in their NAACP study (depending on which statistical model they apply to the same data). These variations are too large to allow a conclusion about the true cost per vote of a paid phone program. Green and Gerber get into even more trouble when they compare the cost per vote of phones with canvassing. First, they report that labor costs of canvassing range between $12 and $20 per vote, and up to $24 per vote in Youth Vote One would expect a greater range of labor costs given the differences in canvassing rural and urban areas. More important, their cost comparison of canvassing versus paid phones is a comparison of apples and oranges, and therefore is totally invalid. In their cost estimates, phones include all of the overhead involved in running the consulting firm. However, the door to door canvassing excludes all of those overhead costs and includes only the hourly labor cost of going door to door. The overhead for canvassing for a campaign or consulting firm is huge compared to that for paid phones. Canvassing involves recruiting and training canvassers for each campaign, getting paper lists and maps into their hands, obtaining insurance coverage, transporting all these individuals to appropriate locations, and tracking the results from people scattered all over a large geographic area. Paid phones, on the other hand, can buy time for the week before the election from consumer marketing firms with callers who are already trained and located in one supervised space. The selection of target voters is another problem. For GOTV, 4 of 5 12/13/ :59 PM

5 you do not simply pick a street and canvass everyone on that street, as Green and Gerber have done in some of their experiments. Rather, you must target specific voters who are identified as highly likely to support your candidate. With the exception of some ethnically homogenous urban areas, there are few locations where you can just canvas a street and expect to find voters who are highly likely to vote for one candidate. If you are hitting only a few people in each block, the labor and transportation costs of door to door canvassing go up tremendously. On the other hand, paid callers can call targeted voters across the district or state with equal ease. In other words, GOTV campaigns can be cost effective when the targets are geographically condensed. That, however, is seldom the case. Don t Hang Up Those Phones A few poorly designed experiments reflecting mixed results cannot outweigh the experiences of thousands of campaign managers. They have struggled with scarce resources and made the strategic decision to hire phones for GOTV. They do it because their experience and post election analysis shows that paid phones make a difference. While we applaud Green and Gerber s efforts to move beyond anecdotal assessments, they have not yet produced the kind of solid research that can be successfully utilized by campaign professionals. Ultimately, research is the most useful to political groups and coalitions that fund their own research and integrate the findings into tried and true GOTV appeals based on self interest and partisanship. Unfortunately, political parties, interest groups, and 527s are so focused on the short term goal of winning the next election that they often lose sight of the long term benefits of good empirical research. Such research should help them win elections. Janet Grenzke, Ph.D. and Mark Watts, Ph.D. are partners at Abacus Associates ( a Democratic polling firm founded in 1990 and based in Hatfield, Mass. Prior to their work as political consultants, both were university faculty. Sign up for C&E Updates Write to Us Campaigns & Elections Magazine 1629 K Street, NW The Mezzanine Washington, DC Phone: , Toll Free: , Fax: Return Policy Privacy Statement Feedback Site Index Published by Votenet Solutions, Inc Campaigns & Elections Magazine. All Rights Reserved. 5 of 5 12/13/ :59 PM