Northeastern University College of Social Sciences and Humanities
PROPOSAL
Submitted in response to
Village of Port Chester #2017-02
Alternative Governance Options for Consideration in Future Village Trustee Elections Statement of Qualifications
COSTAS PANAGOPOULOS,Ph.D.
Dr. Costas Panagopoulos is Professor of Political Science and Director of Big Data, Quantitative Methods and Networks Initiatives at Northeastern University. He was previously Professor of Political Science and Founder of the Center for Electoral Politics and Democracy and the graduate program in Elections and Campaign Management at Fordham University. A leading expert on campaigns and elections, voting behavior, political psychology, campaign finance, and experimental research. Dr. Panagopoulos has been part of the Decision Desk team at NBC News since the 2006 election cycle. He is also editor of American Politics Research, a peer-reviewed journal published by Sage.
In 2015-2016, Dr. Panagopoulos was Visiting Professor of Political Science and Resident Fellow at the Institution for Social and Policy Studies at Yale University. He was selected by the American Political Science Association as a Congressional Fellow during 2004-2005, and he served in the office of Senator Hillary Rodham Clinton (D-NY). Since 2012, he has also taught graduate courses in the Quantitative Methods in the Social Sciences program at Columbia University.
Panagopoulos is the author of over 60 scholarly articles published in outlets including: American Journal of Political Science, Journal of Politics, Political Research Quarterly, Political Behavior, Public Opinion Quarterly, Political Psychology, Presidential Studies Quarterly, the Journal of Elections, Public Opinion and Parties, American Politics Research, PS: Political Science and Politics, Women & Politics, the Journal ofSocial and Political Psychology, Social Influence, Basic and Applied Social Psychology, and the Journal of Political Marketing. He is also author of Political Campaigns: Concepts, Contexts, and Consequences (Oxford University Press) and coauthor of A Citizen's Guide to U.S. Elections: Empowering Democracy in America (Routledge). He is also editor of Rewiring Politics: Presidential Nominating Conventions in the Media Age (LSU Press), Politicking Online: The Transformation of Election Campaign Communications (Rutgers University Press), and Public Financing in American Elections (Temple University Press) and coauthor (with Joshua Schank) of All Roads Lead to Congress: The $300 Billion Fight Over Highway Funding(CQ Press).
Panagopoulos has received over $1 million in grant support for his research as well numerous awards including the Robert H. Durr Award for "Best Paper presented at the Midwest Political Science Association that applies quantitative methods to a substantive problem" in 2014, the "Best Paper published in Political Research Quarterly in 2014," and the Miller Prize for "Best Paper published in Political Analysis in 2013."
Previously, Dr. Panagopoulos was a research associate at the Institution for Social and Policy Studies at Yale University, where he completed a postdoctoral fellowship in 2006. He previously founded and directed the Master's Program in Political Campaign Management in the Department of Politics at New York University. Panagopoulos has also been a Visiting Scholar at the Institute for Social and Economic Research and Policy at Columbia University and Scholar-in-Residence in the Department of Government at American University. He has also been a Visiting Research Scholar at the Institute for Politics, the Internet and Democracy at the Graduate School for Political Management at George Washington University and is a Research Fellow at the Center for Congressional and Presidential Studies at American University, both in Washington, DC.
Dr. Panagopoulos has provided extensive analysis and commentary for print and broadcast media outlets including: CNN, CBS Evening News, NBC Nightly News, Fox News, BBC, The New York Times and the Los Angeles Times.
Panagopoulos has provided strategic advice to numerous candidates running for federal, state and local offices or in international elections. He has also served as a consultant to provide academic expertise, analysis and expert witness testimony in cases and public policy debates surrounding issues related to campaign finance, public policy compliance and implementation. His clients have included the Brennan Center for Justice and the State of Vermont. The former Editor-in-Chief of Campaigns & Elections magazine, Panagopoulos has also served on the Board of Directors of the American Association of Political Consultants and has also worked as a consultant in the Electoral Assistance Division at the United Nations.
A native of Lowell, Massachusetts, and the son of immigrants from Greece, Dr. Panagopoulos received his undergraduate degree in Government, magna cum laude, from Harvard University. He earned his doctoral degree in Politics from New York University. In 1992, while a student at Harvard University, Dr. Panagopoulos was a candidate for the Massachusetts State Legislature. References
Professor Donald P. Green Columbia University Department of Political Science Telephone:(646) 260-5719 Email: [email protected] Project: Experiments in Campaign Finance (2011-2017)
Eve Jacobs-Camahan Assistant Attorney General Vermont Attorney General's Office 109 State St. Montpelier, VT 05609-1001 Telephone:(802) 828-0370 ejacobs@atg. state. vt. us Project: Expert witness analysis (campaign finance)(2012-2013) COSTAS PANAGOPOULOS
Northeastern University Personal: Department of Political Science Renaissance Park 927 Boston, MA 0211 5 (917) 405-9069 (617) 373-4412 [email protected]
ACADEMIC EXPERIENCE
Northeastern University
Professor, Department of Political Science (201 7-present) Director, Big Data, Quantitative Methods and Networks Initiatives (201 7-present)
Yale University
Visiting Professor, Department of Political Science (201 5-201 6) Fellow, Center for the Study of American Politics, ISPS (201 5-2016) Postdoctoral Fellow, Institution for Social and Policy Studies (2005-2006) Research Associate, Institution for Social and Policy Studies (2006-2008)
Fordham University
Professor, Department of Political Science (2014-201 7) Associate Professor (with tenure). Department of Political Science (2011-2014) Assistant Professor, Department of Political Science (2007-201 1) Director, Graduate Program in Elections and Campaign Management (2005-201 7) Director and Founder, Center for Electoral Politics and Democracy (2008-201 7) Visiting Assistant Professor, Department of Political Science (2005-2007)
Columbia University
Professor (Adjunct), Quantitative Methods in the Social Sciences (201 1-201 7) New York University
■ Instructor, Department of Politics (2004; PCM Program, 2001-2005) ■ Executive Director, M.A. Program in Political Campaign Management (2000-2005)
APSA Congressional Fellow
■ Office of Senator Hillary Rodham Clinton (D-NY)(2004-2005)
EDUCATION
■ Ph.D., Department of Politics, New York University, 2005. «> M.A., Department of Politics, New York University, 2001. ■ A.B., Magna Cum Laude, Department of Government, Harvard University, 1994.
OTHER EMPLOYMENT/POSITIONS
2006-present Elections Analyst and Decision Desk, NBC News 2010-201 1 Editor-in-Chief, Campaigns & Elections 2002-2003 Consultant, Electoral Assistance Division, United Nations 2000-present Principal, XVOTES, International Political Consulting Firm 1995-1 996 Features Editor/Politics Editor, Adweek Magazine 1995 Associate, Public Affairs, Burson-Marsteller/New York 1993 Researcher, Ministry of Social Affairs, Paris, France 1993-1994 Research Assistant, Professor Stanley Hoffmann, Harvard University 1992 Candidate, House of Representatives, Massachusetts State Legislature
RESEARCH INTERESTS
American Politics (Campaigns & Elections, Public Opinion, Voting Behavior, Campaign Finance, Congress) Methodology (Econometrics, Experiments) Applied Politics (Campaign Management and Strategy, Political Communications and Media) PUBLICATIONS
BOOKS
■ Rewiring Politics: Presidential Nominating Conventions in the Media Age (editor), Louisiana State University Press (2007).
■ AH Roads Lead to Congress: The $300 Biiiion Fight Over Highway Funding (with Joshua Schank), CQ Press (2007).
■ Politicking Online: The Transformation of Election Campaign Communications (editor), Rutgers University Press (2009).
■ Public Financing in American Elections (editor). Temple University Press (2011).
■ Strategy, Money and Technology in the 2008 Elections (editor), Taylor & Francis (2012).
■ A Citizen's Guide to Elections (with Aaron Weinschenk), Routledge {2016).
JOURNAL ARTICLES
■ "Partisan Social Pressure and Voter Mobilization." (with Meghan Condon and Christopher Larimer) American Politics Research (forthcoming).
■ "A Closer Look at Eyespot Effects on Voter Turnout: Reply to Matland and Murray." Political Psychology{^ovXhcommq).
■ "Do Public Matching Funds and Tax Credits Encourage Political Contributions? Evidence from Three Field Experiments Using Nonpartisan Messages." (with Michael Schwam-Baird, Donald Green and Jonathan Krasno). Election Law Journal(f o rt h co m i n g).
■ "All About That Base: Changing Campaign Strategies in U.S. Presidential Elections." Party Politics 22 (2): 179-190.(March 2016). "Billboards and Turnout: A Randomized Field Experiment." (with Shang Ha). Journal of Political Marketing 14 (4): 391 -404.(December 201 5).
"The Enduring Relevance of National Presidential Nominating Conventions." (with Kyle Endres). The Forum. 1 3 (4): 559-576.(December 201 5).
"Encouraging Small Donor Contributions: A Field Experiment Testing the Effects of Nonpartisan Messages." (with D. Green, J. Krasno, M. Schwam-Baird and B. Farrer). Journal ofExperimental Political Science 2 (2): 1 83-191. (Winter 201 5). DOI: http://dx.doi.orq/l 0.101 7/XPS.201 5.1.
"Be All That You Can Be: The Electoral Impact of Military Service as an Information Cue." (with Monika McDermott). Political Research Quarterly (2): 293-305.(May 2015).
"Preference Dynamics in the 2014 Congressional Midterm Elections." Forum 1 2 (4): 729-737.(December 2014).
"Raising Hope: Hope Inducement and Voter Turnout." Basic and Applied Social Psychology i" S (6) 494-501.(October 2014).
"Watchful Eyes: Implicit Observability Cues and Voting." Evolution and Human Behavior (4): 279-284. Quly 2014).
"Personality, Negativity and Political Participation." (with Aaron Weinschenk). Journai ofSociaiand PoiiticaiPsychoiogy 2 (1): 164-1 82. Quly 2014).
"Preelection Poll Accuracy and Bias in the 201 2 Elections." (with Benjamin Farrer). PresidentiaiStudies QuarterlyAA (2): 352-363 Gune 2014).
"I've Got My Eyes on You: Implicit Social Pressure Cues and Prosocial Motivation." Political Psychology. (Published September 30, 2013: DOI: 10.1111/pops.12074) ■ "Deus ex Machina: Candidate Web Presence and the Presidential Nomination Campaign."(with Dino Christenson and Corwin Smidt). Political Research Quarterly. (Published July 15, 2013: DOI: 10.1 1 77/106591 291 349401 7).
" "Social Pressure, Descriptive Norms and Voter Mobilization." (with Christopher Larimer and Meghan Condon." Political Behavior. (Published June 12, 2103: DOI: 10.1007/sl 1 109-01 3-9234-4).
■ "Big Five Personality Traits and Responses to Persuasive Appeals: Results from Voter Turnout Experiments." (with Aian Cerber, Gregory Huber, Conor Dowling and David Doherty). Political Behavior i" S (4): 687-728.
■ "Life-Cycie Effects and Social Pressure to Vote." (with Marisa Abrajano). Electoral Studies. (Published July 24, 2013: DOI: 10.1016/J.elecstud.201 3.07.019).
■ "Mobilizing Latino Voters: The Impact of Language and Co-Ethnic Policy Leadership." (with Thad Kousser, Michael Binder and Vladimir Kogan). American Politics Research.(Published September 13, 201 3: DOi: 10.1 1 77/1 532673X1 3502848).
■ "Who Participates in Exit ?o\\sT Journal of Elections, Public Opinion and Parties 23 (4): 444-455 Quly 2013).
■ "Public Awareness and Attitudes about Redistricting Institutions."7oiy/r7a/o/" Politics and Law ^ (3): 45-54 Guly 201 3).
■ "The Sociai Brain Paradigm and Social Norm Puzzles." (with Marcus Holmes). Journal of Theoretical Politics. (Published October 1 5, 201 3: DOI: 10.1 177/0951629813502710).
■ "Extrinsic Rewards, Intrinsic Motivation, and Voting." 201 '^.Journal of Politics 75 (1): 266-280. Ganuary 201 3).
■ "Campaign Duration and Eiection Outcomes." Politics and Governance 1 (1): 66- 73.(May 2013). ■ "Positive Social Pressure and Prosocial Motivation: Evidence from a Large-Scale Field Experiment on Voter Mobilization." Political Psychology ZA (2): 265-275. (April 2013).
■ "Reasoning about Interference between Units: A General Framework." (with Jake Bowers and Mark Fredrickson). Political Analysis 2^ (1): 97-124.(Winter 2013).
■ "Campaign Effects and Dynamics in the 2012 Election." Forum 10 (4): 36-39. (February 201 3).
■ "Does Public Financing Chill Political Speech? Exploiting a Court Injunction as a Natural Experiment." (with C. Dowling, R. Enos and A. Fowler). Election Law Journai 1 1 (3): 302-31 5.(September 2012)
■ "Ex-Presidential Approval: Retrospective Evaluations of Presidential Performance." PresidentiaiStudies QuarterlyA2 (4): 719-729.(December 2012).
■ "Campaign Context and Preference Dynamics in U.S. Presidential Elections." Journal of Elections, Public Opinion and Parties 22 (2): 123-127.(September 2012).
■ "Thank You for Voting: Gratitude Expression and Voter UobW'xzdiWon." Journal of Politics 73 (3): 707-71 7. Ouly 2011).
■ "Social Pressure, Surveillance and Community Size: Evidence from Field Experiments on Voter Turnout." Electoral Studies 30 (2): 353-357. (June 2011).
■ "Does Language Matter? The Impact of Spanish vs. English-language COTV Efforts on Latino Turnout." (with Marisa Abrajano). American Politics Research 39 Ouly): 643-663. Ouly 2011).
■ "Spanish-Language Radio Advertisements and Latino Voter Turnout in the 2006 Congressional Elections: Field Experimental Evidence." (with Donald P. Green). Political Research Quarterly bA (3): 588-599.(September 2011). ■ "Timing is Everything? Primacy and Recency Effects in Voter Mobilization Campaigns." Political Behavior 33'. 79-93.(March 2011).
■ "Polls and Elections: Firing Back: Out-Party Responses to the President's State of the Union Addresses, 1966-2006." Presidential Studies Quarterly AA (3): 604- 61 7 (September 2011).
■ "Voter Turnout in the 2010 Congressional Midterm Elections." PS: Political Science & Politics 44 (2): 31 7-319. (April 2011).
■ "Filled Coffers: Campaign Contributions and Contributors in the 2008 Electioons." (with Keena Ups\Xz). Journal of Political Marketing 10 (1-2): 43-57. (June 2011).
■ "Introduction: Money and Technology in the 2008 E\e(:X\ox\s." Journal ofPolitical Marketing 10 (1 -2): 1-3. Oune 2011).
■ "The Dynamics of Campaign Preferences in the 2010 Congressional Midterm Elections." The Forum 8 (4): Article 9.(December 2010).
■ "Affect, Social Pressure and Prosocial Motivation: Experimental Evidence of the Mobilizing Effects of Pride, Shame and Publicizing Voting Behavior." Political Behavior32 (3): 369-386.(September 2010)
■ "The Enduring Effects of Social Pressure: Tracking Campaign Experiments over a Series of Elections." (with Tiffany Davenport, Alan Gerber, Donald Green, Christopher Larimer and Christopher Mann). Political Behavior 32 (3): 423-430. (September 2010).
■ "Are Caucuses Bad for Democracy?" Political Science Quarterly 125 (3): 425- 442.(September 2010)
■ "The Crystallization of Voter Preferences during the 2008 Presidential Campaign."(with Robert Erikson and Christopher WIezien). Presidential Studies Quarterly AO (3): 482-496. ■ "Polls and Elections—Preelection Poll Accuracy in the 2008 Election Cycle." Presidential Studies Quarterly ^9 (4): 896-907.(December 2009).
■ "Grassroots Mobilization in the 2008 Presidential Election." (with Peter Francia). Journal of Political Marketing 8 (4): 31 5-333.(Dece m be r 2009).
■ "Street Fight: The Impact of a Street Sign Campaign on Voter Turnout." Electoral Studies 2% (2): 309-313. Oune 2009).
■ "Campaign Dynamics in Battleground and Nonbattleground States." Public Opinion Quarterly 73 (1):1 19-1 30. (Spring 2009).
■ "Partisan and Nonpartisan Message Content and Voter Mobilization: Field Experimental Ew'idence."Political Research Quarterly E>2 (1): 70-77.(March 2009).
■ "The Calculus of Voting in Compulsory Voting Systems." Political Behavior 30 (4): 455-467.(December 2008).
■ "Polls and Elections—The Ground War 2000-2004: Strategic Mobilization in Presidential Campaigns." (with Peter Weilhouwer) Presidential Studies Quarterly 38 (3): 347-362 (June 2008).
» "Lady Luck? Female Political Consultants in American Campaigns."(with David Dulio and Sarah Ere\Nex). Journal of Political Marketing
■ "Poll Trends—Labor Unions in the United States." (with Peter Francia). Public Opinion Quarterly 72 (1): 134-160.(March 2008).
■ "Field Experiments Testing the Impact of Radio Advertisements on Electoral Competition." (with Donald Green). American Journal of Political Science 52(1): 1 56-168.(January 2008). "Presidential Nominating Conventions: Past, Present and Future." The Forum 5 (4): 1-12.
"Online Fundraising and Contributors in the 2004 Presidential Election." (with Daniel Bergan). Social Science Computer Review 2S (4): (December 2007).
"Size Doesn't Always Matter: White House Staff and Presidential Success in Congress." White House Studies 7 (1): 1 3-34. (June 2007). Reprinted in The Presidency of the United States: New Issues and Developments, Anthony J. Eksterowicz and Glenn P. Hastedt, eds. Nova Publishers (2009).
"The Polls—Cabinet Member and Presidential Approval." Presidential Studies Quarterly {\)\ 1 53-163.(March 2007).
"The Polls—Trends: Arab and Muslim Americans and Islam in the Aftermath of 9/11." Public Opinion Quarterly 70 (4): 608-624.(Winter 2006).
"Contributions and Contributors in the 2004 Presidential Election Cycle." (with Daniel Bergan) Presidential Studies Quarterly 10 (2): 155-172.0une 2006).
"The Polls—Public Opinion and Secretary of Defense Donald Rumsfeld." Presidential Studies Quarterly 10 (1): 11 7-1 27.(March 2006).
"The Polls—Trends: Obesity." Public Opinion Quarterly 70 (2): 249-268. (Summer 2006).
"Vested Interests: Interest Group Resource Allocation in Presidential CampdAqns" Journal ofPolitical Marketing S (1 /2): 59-78. Gune 2006). Reprinted in Campaigns and Poiiticai Marketing, Wayne Steger, Sean Kelly and J. Mark Wrighton, eds., Binghamton, NY: Hayworth Press. (2006).
"Political Consultants, Campaign Professionalization and Media Attention." PS: Political Science 39 (4): 867-870.(October 2006). ■ "Grassroots Mobilization and Voter Turnout in 2004."(with Donald Green, Alan Gerber and Daniel Bergan) Public Opinion Quarterly (5): 760-778.(2005).
■ "The Parliamentary Election in Greece, March 2004." (with Nikos Maratzidis) Electoral Studies 2S (2): 404-409.(June 2006).
■ "The Frontrunners 2008." PS: Political Science 38 (4): 803-808.(October 2005).
■ "Follow the Money: Assessing the Allocation of E-Rate Funds." Social Science Computer Review 2"^ (4): 502. (Winter 2005).
■ "Likely (and Unlikely) Voters and the Measurement of Campaign Dynamics." (with Robert S. Erikson and Christopher WIezien) Public Opinion QuarterlY6S (4): 588-602.(Winter 2004).
■ "The Polls—Trends: Electoral Reform." Public Opinion Quarterly (4): 623- 641.(Winter 2004).
■ "Boy Talk/Girl Talk: Gender Differences in Political Campaign Communications." Women & Politics 26 (3/4): 131-.(2004).
CONTRIBUTIONS: BOOK CHAPTERS, ENCYCLOPEDIAS
■ "Fundraising for Social Change in the U.S.: Interest Group Advocacy in Contemporary U.S. Elections." In Public Opinion and interest Croup Politics: South Africa's Missing Links?, Heather Thuynsma, ed. Africa Institute of South Africa. (201 2). ■ "Big Government and Public Opinion." (with Robert Shapiro). In The Oxford Handbook ofPublic Opinion and the Media, Robert Y. Shapiro and Lawrence R. Jacobs, eds. Oxford: Oxford University Press.(2011). ■ "Flush with Cash: Individual Campaign Contributions in the 2008 Elections." (with John C. Green). In The State of the Parties, 6^^ ed., John C. Green and Daniel Coffey, eds., Rowman & Littlefield. ■ "Presidential Leadership and Public Opinion in an Age of Polarization." (with Jeffrey Cohen). In The Obama Presidency: Change and Continuity, Andrew Dowdle, Dirk Van Raemdonck and Robert Maranto, eds. New York: Routledge. (2011). "The 2010 Elections in New York." In Key States, High Stakes: Sarah Paiin, the Tea Party and the 2010 Elections, Charles Bullock, ed. Rowman & Littlefield. (2012). "Opinion and Election Polls." (with Kathy Frankovic and Robert Shapiro). In Handbook ofStatistics: Sample Surveys: inference and Analysis, Danny Pfeffermann and C.R. Rao, eds. Amsterdam: Elsevier. (September 2009). "The Politics of Economic Policy in a Polarized Era: The Case of George W. Bush." (with Jeffrey Cohen). In Judging Bush, Robert Maranto, Tom Lansford and Jeremy Johnson, eds. Stanford University Press.(September 2009). "Technology and the Modern Political Campaign." In Politicking Online: The Transformation of Election Campaign Communications, Costas Panagopoulos, ed., Rutgers University Press (2009). "Clicking for Cash: Campaigns, Donors, and the Emergence of Online Fund- Raising."ln Politicking Online: The Transformation of Election Campaign Communications, Costas Panagopoulos, ed., Rutgers University Press (2009). "Special Interests and Political Campaigns," "Radio and Politics," "Talk Radio," "Presidential Radio Address," "Spin Doctors," "News Management," "Media Consultants," In Encyclopedia of Media and Politics, Todd Shaefer and Thomas Birkland, eds. Washington, DC: CQ Press. (October 2006). "Ballot Access," "Campaign Commercials," "Campaign Expenditures," "Hard Money," "Help America Vote Act," "Pre-primary Endorsement," "Run-off Primaries," "Targeting," "Swing Ratio," "Unit Rule," "Voter Registration," "Voting—Pocketbook Issues," In Encyclopedia ofAmerican Parties and Elections, LarryJ. Sabato and Howard R. Ernst, eds. Facts on File Inc. (January 2005) "Consequences of the Cyberstate: The Political Implications of Digital Government in International Context." In Digital Government: Principles and Best Practices, G. David Carson, ed. Idea Group Publishers.(2003) "The Political Implications of Digital Government"(with Paul M.A. Baker). In Digital Government: Principles and Best Practices, G. David Garson, ed. Idea Group Publishers.(2003)
BOOK REVIEWS ■ Allan LIchtman. The Keys to the White House, 2008 edition. PoiiticaiScience Quarterly {forthcoming) ■ Michael Malbin, ed. The Election After Reform: Money, Politics, and the Bipartisan Campaign Reform Act. Political Science Quarterly (Spring 2006) ■ George C. Edwards, On Deaf Ears: The Limits of the Buiiy Pulpit and Douglas A. Lathrop, The Campaign Continues: How Political Consultants and Campaign Tactics Affect Public Policy. Public Opinion Quarterly {Spx'xnq 2005). ■ William G. Mayer and Andrew E. Busch. The Front-Loading Problem in Presidential Nominations. Political Science Quarterly{Summer 2004). ■ Walter Shapiro, One-Car Caravan: On the Road with the 2004 Democrats Before America Tunes in. Political Science Quarterly {S^r'mg 2004). ■ William G. Mayer. The Making of the Presidential Candidates 2004. Campaigns & Elections(May 2004).
IN PROGRESS:
■ "The Psychology of Choice in Low Salience Elections." (with Donald Green). (Revise and Resubmit at American Journal ofPolitical Science). ■ "Fisher's Randomization Mode of Statistical Inference: Then and Now."(with Jake Bowers). (Revise and Resubmit at American Journal of Political Science.) ■ "Do Newspaper Ads Raise Voter Turnout? Evidence from a Randomized Field Experiment." (with Jake Bowers). (Revise and Resubmit at American Politics Research.) ■ "Negative Affectivity, Political Contention and Turnout: A Genopolitics Field Experiment." (with Jamie Settle, Christopher Dawes and Peter Loewen).(Under Review).
CONSUMER PUBLICATIONS, OPINION
■ "Obama Supporter Oprah Takes a Big Dive." Politico. April 7, 2008. ■ "Strategy Session: Could Bloomberg Enter in the Middle?" Politico. December 11, 2007. ■ "Ford Won the Public's Affection." Baltimore Sun. December 29, 2006. ■ "Polls: Democrats in Striking Distance for Senate." (with Sheldon Gawiser and Tiffany Turner." msnbc.com. October 2, 2006. "Polls: Democrats Run Strong in Senate Races." (with Sheldon Gawiser and Tiffany Turner." msnbc.com. October 1, 2006. "Calendar Games: Timing and Sequence in Presidential Selection." Campaigns & Elections. December 2005. "The Neglected Majority: Resource Allocation in Presidential Campaigns." Campaigns & Elections, August 2004. "Behind the Balloons: Political Consultants and the National Nominating Conventions." Campaigns & Elections, July 2004. "Politics after Pinochet: Chilean Election 2005." Campaigns & Elections, \u\\/ 2004 (with Raquel Villapando) "South Africa Decides 2004." Campaigns & Elections, June 2004. (with Christopher Cullen). "Don't Believe the Hype." (Political Impact of "Fahrenheit 9/11") Aclweek,\u\y 19, 2004. "Part-E Politics: Investigating the Online Presence of American Political Parties." Campaigns & Elections. September, 2003. "Do Image-makers Need a Makeover? (Or, Why Do Americans Hate Political Consultants?) Assessing Public Perceptions about Political Consultants." (with James Thurber). Campaigns & Elections. October/November, 2003. "Primaries are Meaningless and Irrelevant for Many." Op-Ed, Los Angeles Times, February 1 5, 2004. "How Political Scientists Have Failed New Yorkers." Op-ed, New York Sun, October 28, 2003. "Voters in the Dark." Letters to the Editor. Newsday, October 21, 2003. "Buchanan: The No-Poll Pol." Brandweek. March 4, 1996.
CONFERENCE PAPERS
■ "Friends-and-Neighbors Voting: Field Experimental Evidence." (with J. Leighley and B. Hamel). Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 201 5. ■ "Please Recuse Yourself: A Field Experiment on Campaign-Finance Related Conflicts of Interest and Judicial Behavior." (with Donald Green, Jonathan Krasno, Michael Schwam-Baird and Dane Thorley). Paper presented at the Annual Meeting of the American Poljtical Science Association, Washington, DC, August 2014, and at the Annual Meeting of the Southern Political Science Association, New Orleans, l_A, January 201 5. "Partisan Social Pressure." (with Christopher Larimer and Meghan Condon). Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 201 3. "Genes, Negative Affectivity and Turnout: A Field Experiment." (with Jamie Settle, Christopher Dawes and Peter Loewen). Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 2011. "A Game Changer? Understanding Candidate Web Presence in the 2008 Invisible Primary." (with Dino Christenson and Corwin Smidt) Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 2011. "Turning Out, Cashing In: Extrinsic Rewards, Intrinsic Motivation and Voting." Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 2009. "Thank You for Voting" Gratitude Expression and Voter Mobilization." Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 2009. "Emotions, Motivation, and Prosocial Behavior: Field Experimental Tests of the Effects of Alternative Civic Duty Appeals on Voter Turnout." Paper presented at the ISPS AO'^h Anniversary Conference, Yale University. November 2008. "Name Recognition and Incumbency Advantage: Evidence from a Mass Media Field Experiment Targeting Latino Voters" (with Donald P. Green). Paper presented at the Annual Meeting, American Political Science Association. Boston, MA. September 2008. "Spanish-Language Radio Advertisements and Latino Voter Turnout: Field Experimental Evidence" (with Donald P. Green). Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 2008. "Billboard Advertisements and Voter Turnout: Evidence from a Field Experiment" (with Shang Ha). Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 2008. "Street Signs: Experimental Evidence on the Impact of a Street Sign Campaign." Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 2007. "Leveling the Playing Field: A Field Experiment Testing the Impact of Radio Advertisements on Electoral Competition." (with Donald P. Green). Paper presented at the Annual Meeting, American Political Science Association. Philadelphia, PA, September 2006. "The Impact of Newspaper Advertising on Voter Turnout: Evidence from a Field Experiment." Paper presented at the Annual Meeting, American Political Science Association. Philadelphia, PA, September 2006. "The Impact of Radio Advertisements on Voter Turnout: Evidence from a Field Experiment." Paper Presented at the Midwest Political Science Association Annual Meeting, Chicago, IL, April 2006. "The Ground War 2000-2004: Strategic Targeting in Grassroots Campaigns." (with Peter Weilhouwer). Paper Presented at the Midwest Political Science Association Annual Meeting, Chicago, IL, April 2006. "Campaign Dynamics in Battleground and Non-battleground States." Paper Presented at the Midwest Political Science Association Annual Meeting, Chicago, IL, April 2006. "Talk Soft and Carry a Big Stick? Candidate Quality and Campaign Communication Strategies." Paper Presented at the Midwest Political Science Association Annual Meeting, Chicago, IL, April 2005. "The Calculus of Voting in Compulsory Voting Systems." Paper presented at the Public Choice Society Annual Meeting, New Orleans, LA: April 10-12, 2005. "Lady Luck? Female Political Consultants in American Campaigns." (with David Dulio and Sarah Brewer). Paper presented at the Southern Political Science Association Annual Meeting, New Orleans, LA: January 2005. "Spotting the Running Men (and Women): Political Communication in the 2003 California Gubernatorial Recall Election." Paper presented at the "Faith, Fun and Futurama" APSA Political Communication Pre-conference, Chicago, IL. September 1, 2004. "States of Ambition: Challenger Quality in the American States." Poster presented at the American Political Science Association Annual Meeting, Chicago, IL. September 1, 2004. "The Kingmakers: Assessing Americans' Attitudes Toward Political Consultants." Paper presented at the American Association of Public Opinion Researchers Annual Meeting, Phoenix, AZ. May 2004. "More Nobodies: Challenger Quality in U.S. Senate Elections, 1972-2000." Paper Presented at the Midwest Political Science Association Annual Meeting, Chicago, IL, April 2004. "Movin' On Up: Term Limits and Candidate Quality in U.S. House Elections, 1972-2000." Paper presented at the Joint Conference on Term Limits and State and Local Government, Akron, OH. April 2004. "Gender Differences in Challenger Quality in U.S. House Elections, 1 972-2000." Paper presented at the Women and American Political Campaigns Research Symposium, New York University, March 3, 2004. "The 3/2's Rule at Thirty: Resource Allocation in Presidential Campaigns." Paper presented at the Northeast Political Science Association Annual Meeting, Philadelphia, PA. November 6-8, 2003. "Vested Interests: Interest Group Resource Allocation in Presidential Campaigns." Paper presented at the Northeast Political Science Association Annual Meeting, Philadelphia, PA. November 6-8, 2003. "Likely (and Unlikely) Voters and the Measurement of Campaign Dynamics"(with Robert S. Erikson and Christopher WIezien). Paper presented at the World Association of Public Opinion Researchers Annual Meeting. Prague, Czech Republic. September 19, 2003. "Great Expectations: The Promise of Digital Government in the American States." Poster presented at the American Political Science Association Annual Meeting, Philadelphia, PA. August, 2003. "Virtual Parties: Investigating the Online Presence of American Political Parties." Paper presented at the New York Political Science Association Annual Meeting. Wagner College: Staten Island, New York. April 1 1, 2003. "Not on the Run: Assessing Aggregate Challenger Quality in U.S. House Elections, 1972-2000." Paper presented at the Midwest Association of Political Science Annual Meeting. Chicago, IL. April, 2003. "Boy Talk/Girl Talk: Gender Differences in Political Campaign Communications." Paper presented at the Midwest Association of Political Science Annual Meeting. Chicago, IL. April, 2003. "Marching to the Beat of a Different Drum?: Examining the Political Behavior and Attitudes of Young Americans." Poster presented at the Midwest Association of Political Science Annual Meeting. Chicago, IL. April, 2003.
DISSERTATION: Quality Control: Candidate Quality in U.S. Congressional Elections, 1972-2000. Committee: Jonathan Nagler (Chair), Robert Erikson (Columbia), Anna Harvey, Richard Fleisher (Fordham)
Invited Presentations:
■ "Negative Affectivity, Political Contention and Turnout: A Genopolitics Field Experiment." Department of Political Science, Stony Brook University. September 2014. ■ "Bases Loaded: The Causes and Consequences of Shifting Campaign Strategies in U.S. Elections." Workshop on Context and Voter Mobilization, University of Manchester (UK). November 201 3. ■ "I've Got My Eyes on You: Implicit Social Pressure Cues and Prosocial Motivation." Manhsip School of Communication and Department of Political Science, Louisiana State University. January 201 3. ■ "Negative Affectivity, Political Contention and Turnout: A Genopolitics Field Experiment." Department of Political Science, Columbia University. February 2012. ■ "Big Five Personality Traits and Responses to Persuasive Appeals: Results from Voter Turnout Experiments." Department of Psychology, Fordham University, December 2011. ■ "Turning Out, Cashing In: Extrinsic Rewards, Intrinsic Motivation and Voting." Department of Political Science, Temple University, October 2011. ■ "The Psychology of Choice in Low-Salience Elections." Department of Political Science, University of California at San Diego, January 2009. ■ "Denver Showdown? The 2008 National Nominating Conventions." Dole Institute. University of Kansas, April 2008. ■ "Presidential Nominating Conventions." American University, April 2008. ■ "U.S. Presidential Elections." China University of Political Science and Law. Beijing, China. May, 2007. ■ "Women and U.S. Elections." Womens' Campaign Fund, New York, May 2004. ■ "State Parties and the Internet: An Assessment" Democratic National Committee, Washington, DC, July, 2004.
Grants ■ "Field Experiments in Campaign Finance." (with Donald Green and Jonathan Krasno). Open Society Foundation and Democracy Fund/Omidyar Network. 2013-201 5. $900,000. ■ "2012 New York Redistricting Project." Alfred P. Sloan Foundation. 2012. $104,000. ■ "Public Financing Project." Open Society Foundation. $10,000.
TEACHING
Undergraduate:
Fordham University (Department of Political Science) Municipai Elections Presidential Nominations Presidential Elections Parties and Elections Money and Politics Campaigns & Elections Introduction to Politics
New York University (Department of Politics) ■ American Public Opinion, Department of Politics ■ Power and Politics in America, Department of Politics
Other (Adjunct) U.S. Elections, Yeshiva University American Government and Political Institutions, Pace University Applied Research Methods for Political Science, Pace University Media and Politics, Hunter College State and Local Politics, Pace University Comparative Politics: Political Structures and Institutions, Yeshiva University Introduction to World Affairs; SUNY/FIT Introduction to Comparative Politics, SUNY/FIT ■ Contemporary Western European Politics, SUNY/FIT
Graduate: Theories and Methodologies in the Social Sciences, Columbia University (QMSS) Experimentation in the Social Sciences, Columbia University (QMSS) American Political Behavior, Fordham University Campaign Finance and Ethics, Fordham University Special Topics in Campaigns and Elections, Fordham University Political Campaigns in Emerging Democracies (co-taught with Richard Dresner), NYU New Media and Political Campaigns, NYU Fundamentals of Political Campaign Management, NYU Crisis and Issues Management, NYU
Other: ■ ICPSR Summer Program, "Field Experiments: Design, Analysis and Interpretation," (co-instructor with Donald Green). Summer 2013; Summer 2014. ■ Visiting Lecturer, Program on Elections and Representation, Universite de la Sorbonne (Assas II), Paris, France, Summer 2002
HONORS, ACTIVITIES, AND AWARDS
201 5 "Best Paper published in Political Research Quarterly\x\ 2014."(with Dino Christenson and Corwin Smidt).
2014 Robert H. Durr Award for "Best Paper presented at the Midwest Political Science Association that applies quantitative methods to a substantive problem." (with Donald Green, Jonathan Krasno, Michael Schwam-Baird and Benjamin Farrer).
Miller Prize for "Best Paper published in Political Analysis" (with J. Bowers, M. Friedrickson)
201 3 Executive Council, APSA Elections, Public Opinion and Voting Behavior Section (201 1-2013) Member, Doris Craber Book Prize selection committee
2012 Chair, Nominations Committee, APSA Experimental Politics Section
2011 Elected Secretary, APSA Experimental Politics Section
2009 Elected Associate Program Chair, NYAAPOR
2008 Elected At-Large Member of Executive Committee, APSA Political Communication Section
2004 APSA Congressional Fellow (2004-2005) Visiting Research Scholar, Institute for Politics and Democracy and the Internet, Graduate School of Political Management, George Washington University Selection Committee, "Pollie" Awards (Political Advertising) AAPC Research Fellow, Center for Congressional and Presidential Studies, American Universil
2003 Visiting Scholar, Institute for Social and Economic Research and Policy, Columbia University. Visiting Research Fellow (Summer), Center for Congressional and Presidential Studies, American University, Washington, DC. Member, Academic Committee, American Association of Political Consultants Member, Board of Directors, SAFEOnline (Standards Association For Elections Online) President, American Association of Political Consultants, Mid-Atlantic Chapter Who's Who in America
2002 Dirksen Congresional Center Research Fellowship. Steering Committee, Academic Conference, AAPC, American University Selection Committee, "Pollie" Awards (Political Advertising) AAPC
2001 Nominated, Outstanding Teaching Award (TA), NYU Member, Curriculum Committee, Department of Politics, NYU
1994 Pforzheimer Scholar, Harvard University 1993 Undergraduate Research Fellow, Center for European Studies, Harvard University Research Grant, Ford Foundation for Undergraduate Research, Harvard University
1992 Research Grant, Ford Foundation for Undergraduate Research, Harvard University
1990- John Harvard Scholar, Harvard University; Representative, Harvard Foundation for 1994 Cultural and Race Relations; Representative, Undergraduate Council
1990 National Winner, NASSP Century III Leadership Program
CONFERENCE PARTICIPATION, SPECIAL PROJECTS
2014 Annual Meeting, Midwest Political Science Association, Chicago, IL. Annual Meeting, American Political Science Association, Washington, DC
2013 Annual Meeting, Midwest Political Science Association, Chicago, IL.
2012 Annual Meeting, Midwest Political Science Association, Chicago, IL.
2011 Guest Editor, "The 2010 Congressional Midterm Elections" Symposium, PS: Political Science & Politics. Annual Meeting, Midwest Political Science Association, Chicago, IL. Annual Meeting, American Political Science Association, Seattle, WA.
2010 Annual Meeting, Midwest Political Science Association, Chicago, IL. Annual Meeting, American Political Science Association, Washington, DC.
2009 Guest editor, "Strategy and Choice in the 2008 Elections" Special \ssue Journai ofPoiitical Marketing Annual Meeting, Midwest Political Science Association, Chicago, IL. Annual Meeting, American Political Science Association, Toronto, Canada. Annual Meeting, Political Methodology section. New Haven, CT. 2008 Invited participant, ISPS 40^^^ Anniversary Conference, Yale University, November 14-15. Conference co-chair, "Citizen-Owned Elections" Public Financing Past, Present and Future," Fordham University. February. Northeast Political Methodology Conference, New York Univeristy Annual Meeting, Midwest Political Science Association, Chicago, IL. Annual Meeting, American Political Science Association, Chicago, IL. Annual Meeting, Political Methodology section, Ann Arbor, Ml.
2007 Guest editor, "Technology and Political Campaigns" Special Issue, Social Science Computer Review Annual Meeting, American Political Science Association, Chicago, IL. Annual Meeting, Midwest Political Science Association, Chicago, IL.
2006 Annual Meeting, American Political Science Association, Philadelphia, PA. Annual Meeting, Midwest Political Science Association, Chicago, IL.
2005 Panelist, "Voting in America." New School University. September 13, 2005. Annual Meeting, American Political Science Association, Washington, DC. Annual Meeting, Midwest Political Science Association, Chicago, IL. Annual Meeting of the Public Choice Society, New Orleans, LA. April 10-12.
2004 Panelist, "Unprecendented: Why Voting Matters." New School University, October
1. Invited participant (presenter). Faith, Fun and Futurama: APSA Political Communication Pre-conference, Chicago, IL, Sept. 1 Annual Meeting, American Political Science Association, Chicago, IL, September 1-4. Observer, Republican National Convention, New York, NY, August 29- September 1. Observer, Democratic National Convention, Boston, MA. July 26-29 Invited Participant, Political Consultants and American Campaigns Conference. Center for Congressional and Presidential Studies, American University. June 1 Campaigns & Elections Jxdi\x\\x\q Seminar. (Speaker: International Campaigns) Washington, DC, June 1, 2004. Annual Meeting, American Association of Public Opinion Researchers, Phoenix, AZ, May Guest Lecturer, Women's Campaign Fund, New York, NY. May 11. Title: "Women and American Campaigns." Invited Participant, Joint Conference on Term Limits and State and Local Government, University of Akron, Akron, OH. April 29-May 1. Moderator, Technology and Politics Symposium, New York University. May. Midwest Political Science Association Annual Meeting. Chicago, IL. April 1. Politics Online Conference, George Washington University, Washington, DC. Invited Panelist: Constituent Mobilization. March 19. Chair, Women and American Political Campaigns Research Symposium, NYU, March 3. Campaigns & Elections Jra\n\nq Seminar. (Speaker: Crisis Communication in Campaigns) Washington, DC, February 14. Invited Participant, Campaign Finance Reform Conference, Hinkley Institute of Politics, University of Utah. January 18-19.
2003 Annual Meeting, Northeast Political Science Association, Philadelphia, PA, November 6-8. Academic Conference, American Association of Political Consultants. University of Akron, Akron, Ohio. Panelist: "Bush v. Mystery Democrat: Prospects for the 2004 Campaign." Oct. 2-4. Annual Conference, World Association of Public Opinion Researchers, Prague, Czech Republic. September 1 7-19. Annual Meeting, American Association of Political Consultants, Philadelphia, PA. August. Annual Meeting, Midwest Political Science Association. Chicago, IL. April. 2002 Conference Chair, "Back and Forth: Looking Back Elections 2002 and Ahead to 2004," Department of Politics, NYU, December 10. Annual Conference, International Association of Political Consultants, Rio de Janeiro, Brazil, November 1 5-20. Invited Participant, Electoral Politics Conference, Columbia University, October 18-19. Political Methodology Conference, New York University. Annual Meeting, American Political Science Association, Boston, MA. August. Speaker, Campaigns & Elections seminar on political electioneering, Washington, DC. Panels: Earned Media, Crisis Management.
2001 Annual Meeting, American Political Science Association, San Francisco. August. Seminar on Experimental Methods in the Social Sciences, Center for Basic Research in the Social Sciences, Harvard University. June. Seminar Program, Campaigns & Elections, Washington, D.C. Academic Conference, American Association of Political Consultants, Institute of Politics, Kennedy School of Government, Harvard University
2000 Organizer/Chair: MHieniais Rising: The Poiiticai Behavior of Young Americans, New School University Organizer/Chair: Gen X Voters and the 2000 Presidential Election, New School University
COMMENTARY, POLITICAL ANALYSIS, MEDIA APPEARANCES (SELECTED)
2014 Election 2014 analysis/commentary: FOX News, CNN, NBC News, ABC News
2012 Election 2012 analysis/commentary: FOX News, CNN, NBC News, ABC News
2008 Election 2008 analysis/commentary: FOX News, CNN, New York Times, USA Today.
2006 CBS Nightly News ("In the Minute")(Topic: Iraq and Elections 2006). FOX News (Topic: Iraq and Elections).
2004 NBC Nightly News with Tom Brokaw (2004 Presidential TV Ad Strategy); CNN ("Inside Politics")(Topic: Minority Voters 2004 Election); New York One, Eyewitness News (WABC), Fox News (Reagan Legacy). Los Angeles Times, Seattle Post-Intelligencer, New York Sun, CuarcHan, Village Voice, Newsday, BBC Television, NBC Nightly News with Tom Brokaw, BusinessWeek TV. (Topic: 2004 Democratic Presidential Primary)
2003 CNN ("Inside Politics")(Topic: 2004 Democratic Presidential Primary) BBC Worldwide (Topic: Howard Dean and the 2004 Presidential Primary) Fox News (Topic: Nonpartisan Municipal Elections)
2000 New York Times (Topic: 2000 Presidential Election)
PROFESSIONAL SERVICE
Editor American Politics Research (2016-present)
Associ ate Ed ito r Journal of Experimental Political Science (201 1-2015)
Editor "Polls and Elections" feature, Presidential Studies Quarterly{2007
Senior Editor Journal ofPolitical Marketing
Editorial Board American Journal ofPolitical Science (201 3-present), Journal of Experimental Political Science (2016-present), internationalJournal of Public Opinion Research (201 3-201 A), Social Science Computer Review, Open Political Science Journal, Politics and Governance.
Contributor Campaigns & Elections Magazine (2001 -2006); Politico (2008-present)
Refe ree American Political Science Review, American Journal of Political Science, Journal of Politics, Political Research Quarterly, Public Opinion Quarterly, Political Behavior,Journal of Elections, Public Opinion and Parties, Gender and Politics, Electoral Studies, Political Psychology, British Journal ofPolitical Science, Comparative Poiiticai Studies, Party Poiitics, Poiiticai Communication, American Poiitics Research.
Editorial Assistant Poiiticai Science Quarteriy{200\-200S)
Discussant American Political Science Association, Annual Meeting, 2004, 2006-2014 Midwest Political Science Association, Annual Meeting, 2004-2006, 2010- 2014 Southern Political Science Association, Annual Meeting. 2006. New York Political Science Association Annual Meeting, 2003.
Chair Midwest Political Science Association, Annual Meetings: 2009, 2010. NPSA Annual Meeting 2003 (Panel: Off the Air and Back to Ground: Television Advertising versus Grassroots Mobilization in Political Campaigns).
MEMBERSHIPS AND PROFESSIONAL ASSOCIATIONS
American Political Science Association (Sections: Eiections, Pubiic Opinion and Voting Behavior; Poiiticai Communication, Poiiticai Organizations/Parties; Poiiticai Methodoiogy; Poiiticai Psychoiogy, Experimentai Poiitics) Midwest Political Science Association Southern Political Science Association International Society of Political Psychology American Association of Public Opinion Researchers World Association of Public Opinion Researchers American Association of Political Consultants -(President 2003-2004; Board of Directors Member, Mid-Atlantic Chapter 2002-2003) Academic Programs in Applied Politics Association (Board of Directors 2003-present) SAFEOnline (Standards for Election Activities Online)(Board of Directors 2003-present) ELJ-2011-0143-ver9-Dowlin^1 P.3d 07/19/12 3:17pm Page 1
EU-2011 -0143-ver9-Dowling_1 P Type: research-article
FXECTION LAW JOURNAL Volume 11, Number 3, 2012 Original Article (0 Marv Ann Licbcrt, Inc. DOl: 10.1089/e|j.201L0143
Does Public Financing Chill Political Speech? Exploiting a Court Injunction as a Natural Experiment
Conor M. Dowling, Ryan D. Enos, Anthony Fowler, and Costas Panagopoulos
ABSTRACT
In 2011, the Supreme Court struck down the matching provisions in Arizona's campaign finance law on the grounds that they violate free speech by chilling private spending. In this article, we explicitly test the effects of Arizona's matching provisions in two ways. First, we find that privately funded state legis lative candidates do not strategically cluster their spending below the threshold that would trigger money to their opponents. Second, we exploit a 2010 Court injunction as a natural experiment. When Arizona's matching provisions were removed, private spending did not increase relative to other states. Contrary to the view of the Court, we find no empirical evidence that campaign finance laws chill private political speech. More generally, our analysis demonstrates the value of exploiting couit injunctions as natural experiments to assess the causal effects of laws.
According to supporters, campaign finance (Gora 2011). In McComish v. Bennett 564 U.S. laws can help to reduce corruption, increase 664 (2011), the Supreme Court weighed these com electoral competition, and thus improve political peting considerations. Arizona's campaign finance representation.' In recent years, reforms designed law allowed candidates to opt into or out of a public to improve the performance of campaign finance financing program. The state would provide public policies have been enacted federally and in many funds for any participating candidate to match, up to states. One class of reforms focusing on public a certain point, the spending of any opposing nonpar- financing of elections has been adopted in several ticipating candidates or independent groups. On one jurisdictions, but these policies have been contro hand, the law protects equal access to resources and versial (Panagopoulos 2011). Critics argue that the enables political speech of participating candidates. public financing of campaigns may crowd out the Conversely, the law may create a disincentive for non- political spending of private groups and individuals, participating candidates and independent groups to thus inhibiting certain forms of political speech spend, thus chilling political speech. As a result of this latter concern, a 5-4 majority struck down the matching funds provision. Because similar cases are Conor M. Dowling is an assistant professor in the Department likely to arise in the near future, the arguments and of Political Science at tlie Univensity of Mississippi in Univer empirical evidence deseiwe closer examination. sity, MS. Ryan D. Enos is an assistant professor in the Dcpan- ment of Government at Harvard University in Cambridge, MA. Anthony Fowler is a graduate student in tltc Department of Government at Harvard University. Costas Panagopoulos is an assistant professor in the Department of Political Science 'whether such systems achieve these goals is still up for debate at Fordham University in New York, NY. We thank Deanne (see, e.g., Dowling 2011; GAO 2010; Kraus 2006; Malhotra Ccvasco. Alexis Coll-Very. Clmstopher Judge. George Morris, 2008; Mayer, Werner, and Williams 2006; Mayer and Wood Serena Orloff. and Michelle Woodhouse who assisted with the 1995; Miller 2011b; Milyo. Primo, and Jacobsmeier 2011; collection of data and the drafting of an amici brief where this Primo and Milyo 2006). See Miller (201 la) for a recent review analysis originally appeared. Thanks to Raul Campilio and Paul of the current state of knowledge concerning the effects of pub Collins for thoughtful comments. lic election funding.
1 ELJ-2011-0143-ver9-Dowling_1 P.3d 07/19/12 3:17pm Page 2
DOVVLING ET AL.
The majority's decision relies on the empirical and interest groups are often reviewed by the courts, claim that Arizona's matching provisions decreased in many cases the Supreme Court, to ensure that they campaign spending by nonparticipating candidates do not abridge individuals' or groups' First Amend and independent groups. We do not address the ment right to freedom of speech. As such, campaign Court's logic on whether this would constitute a vio finance laws are subject to strict scrutiny, requiring lation of free speech. Instead, we focus on the claim the state to prove that a law furthers a compelling gov noted in the majority opinion that the empirical ernment interest and is narrowly tailored to achieve that question is untestable, but nevertheless "evident interest (f//7/7(?c( Slates i'. Carotene Products, 304 U.S. and inherent." Despite the assumption that the 144, 153 n.4 (1938)). The Court has applied strict scru claim is untestable. we argue that such causal effects tiny standards to a number of campaign fiiuuice laws can and should be tested. We conduct several tests over the years (for a more detailed review than what of this claim and find no evidence that matching is presented here, see, for example Esenberg 2011; funds inhibit political spending. Our methods and Gora 2011: Lowenstein, Hasen, and Tokaji 2008). analysis could serve as a model for future studies In Buckley V. Vh/eo,424 U.S. 1 (I91b)(per curiain), interested in using natural experiments to measure the Court upheld the contribution limits in the Federal judicial and policy effects. Election Campaign Act(FECA) of 1971 (because the The article proceeds as follows. First, we provide a act of contributing, not the amount of a contribution, summary of McComish and campaign finance law. was deemed the expression of support that needed to Next, we discuss the Court's flawed scientific reason be protected) but suaick down FECA's expenditure lim ing in this particular case. Then, we discuss the use of its (because it placed undue restrictions on the quantity empirical data in campaign finance cases. Causal of political speech). Other campaign finance laws questions are always difficult to answer, but thresh have also been invalidated by the Supreme Court olds. injunctions, and other natural experiments can including those that restrict independent expendi provide answers to questions previously deemed tures by express advocacy groups (Federal Election untestable. Finally, we provide two explicit tests of Conini'n v. Massachusetts Citizens for Life, Inc., the claim that matching provisions inhibit speech. 479 U.S. 238 (1986)), those that limit uncoordinated We find that the spending of privately funded candi political ptirty expenditures (Colorado Republican dates exhibits no signs of clustering below the trigger Federal Campaign Comm. v. Federal Election ing threshold (that is, there was not a disproportionate Comm'n, 518 U.S. 604 (1996) (Colorado /)). and number of privately funded candidates just below the those that bai" unions and corporations from making tlireshold that would trigger matching funds to a pub independent expenditures(Citizens United v. Federal licly funded opponent, as we would expect if such Election Comm'n, 558 U.S. 50 (2010)). In addition, provisions inhibit speech), suggesting that candidates the Court has upheld limits on coordinated party did not strategically restrict their spending in response expenditures (FederalElection Comm'n v. Colorado to the law. Furthermore, the spending behavior of pri Republican Federal Campaign Comm., 533 U.S. 431 vately funded candidates and independent groups (2001) (Colorado //)), limits on requirements that did not change relative to other states after the 2010 political donors disclose their identities (Citizens Supreme Court injunction that deactivated the match United v. Federal Election Comm'n, 558 U.S 50 ing provisions. If the matching funds provisions were (2010)), and limits on the presidential public financ inhibiting the spending (speech) of privately funded ing program reviewed in Buckley. candidates, we would expect the spending behavior Since Buckley, many states have instituted volun of these candidates to have changed after the Court's tary public financing systems. Nearly half (twenty- decision. Both sources of evidence suggest that four) of the fifty states and sixteen local jurisdictions matching provisions do not inhibit political speech. currently offer some fomi of public funding for cam paigns (Stern 2011). In Arizona, a system of public financing of elections (passed by initiative in 1998) CAMPAIGN FINANCE LAW AND provides all candidates for state office—for example, MCCOMISH V. BENNETT: A BRIEF REVIEW candidates for govemor. secretary of slate, attomey general, and the state legislature (both the House Campaign finance laws governing contributions and Senate)—funding for their primary and general to and spending by candidates, political parties. election campaigns if they are (1) able to collect a ELJ-2011-0143-ver9-Dowling_1P.3d 07/19/12 3:17pm Page 3
DOES FDBl.IC FINANCING CHILL POUTICAI. SPEECH?
certain number of S5 contributions from registered eral Election Coinm'n, 554 U.S. 724 (2008), the /\jizona voters and (2) agree to certain campaign Court reversed the decision of tlic Ninth Circuit restrictions and obligations (limiting the expenditure Court of Appeals, concluding that '"Arizona's matching of personal funds to $500, participating in at least one funds scheme substantially burdens political speech debate, and returning all unspent money to the State). and is not sulBcicntly justified by a compelling interest Arizona's public financing system, similar in many- to survive First Amendment scnrliny" at 667.^ ways to those of Connecticut and Maine, was put in place in an effort to reduce corruption and improve political representation. More specifically, according FLAWED REASONING to a recent report by the Government Accountability IN THE MAJORITY OPINION Office(GAG), the goal of Arizona's full public fund ing systems was to increase electoral competition, After reviewing the evidence for the petitioners' voter choice, and voter participation, while reducing claim that the presence of matching funds caused the influence of interest groups and curbing increases in the cost of campaigns(GAG 2010, 12-14). One way Arizona attempted to achieve this goal "More specific;)IIy. Uic Court summarized llic Arizona matching was through the use of matching funds. In addition funds provision as follows: to an initial lump sum of public funding to those Matching funds arc available in both primary and general candidates that chose to participate, the state provi elections. In a primary, matching funds are triggered when a prixatcly financed candidate's c.xpcnditurcs. combined des public funds to any participating candidate to with the expenditures of independent groups made in sup match, roughly dollar for dollar up to a certain port of the privately financed candidate or in opposition to point, the funds of any opposing privately financed a puhlicly financed candidate, exceed the primary election allotment of state funds to the publicly financed candidate. candidates or independent groups that spent money 'J§I6-9.S2(A). (C). During the general election, matching against the publicly funded candidate (see Ariz. funds are iriggered when the amount of money a privately Rev. Stat. Ann. § 16-952).^ The state argued that financed candidate receives in contributions, combined with such a matching provision was crucial, perhaps the expenditures of independent groups made in support of the privately financed candidate or in opposition to a publicly even necessary, to fulfill the goals of public financ financed candidate, exceed the general election allotment of ing .systems outlined above. Certain (privately state funds to the publicly financed candidate. §16- financed) candidates and independent expenditure 952(B)...Once matching funds are triggered, each additional dollar that a privately financed candidate spends during the groups, however, challenged the constitutionality primary results in one dollar in additional state funding to of the matching provisions on the grounds that his publicly financed opponent (less a 6% reduction meant matching funds inhibit their freedom of speech by to account for fundraising expenses). §I6-952(A). During a general election, every dollar that a candidate receives in con placing an undue burden on their ability to exercise tributions—which includes any money of his own that a can their First Amendment rights (sec, for example, the didate spends on his campaign—results in roughly one dollar quotes from candidates in Miller 2008). In particu in additional state funding to his publicly financed oppo nent...Once the public financing cap is exceeded, additional lar, the petitioners argued that the matching funds expenditures by independent groups can result in dollar-for- provision created a disincentive for nonparlicipating dollar matching funds as well. .Spending by independent candidates and independent groups to spend money groups on behalf of a privately funded candidate, or in oppo during the campaign because they did not want to sition to a publicly funded candidate, results in matching funds. §I6-952(C). trigger matching funds that would be directed to 564 U.S. 664. 672-673 (2011). their opponents. Based on this argument, "The Dis trict Court entered a permanent injunction against ^Davis invalidated the "Millionaire's Amendment" of the Bipartisan Campaign Refonn Act (BCRA) of 2002, 2 U.S.C. the enforcement of the matching funds provision. § 441a-l(a). The "Millionaire's Amendment" permitted a can The Ninth Circuit reversed, concluding that the pro didate for the U.S. House of Representatives to receive individ vision imposed only a minimal burden and that the ual contributions of up to S6.900 per contributor (llirce times the regular contribution limit of S2,300) if the candidate's oppo burden was justified by Arizona's interest in reduc nent spent more than 5350,000 of his or her own funds, while ing quid pro quo political corruption." McComish v. the opponent remained .subject to the regular contribution Bennen, 564 U.S. 664 (2011) at 667. limit. 'The Court concluded that the Millionaire's Amendment was unconstitutional because it forced a candidate "to choose The Supreme Court heard oral arguments on March between the First Amendment right to engage in unfettered 28. 2011 and rendered its decision on June 27, 2011. political speech and subjection to discriminatory fundraising Relying in large part on precedent set in Davis v. Fed limitations." Davis at 739. ELJ-2011-0143-ver9-Dowling_1 P.3d 07/19/12 3:17pm Page 4
DOVVLING ET AL.
a reduction in spending. Chief Justice John Roberts, mon sense and scientific thinking, the claim that it writing for the majority, eoncedes that "it is never is "never easy to prove a negative" has appeared easy to prove a negative." Despite this admission with explicit reference to Elkin.s in at least three of a lack of evidence, he declares it "evident and opinions from the Supreme Court since the Elkins inherent" that the law imposes a burden, writing decision.^ Since tliis notion has legal traction, we that it is "clear...that a candidate or independent believe it is worth further discussion.'' group mifiht not spend money" (italics added). Perhaps negatives are more difficult to demon The majority argues that because there is the poten strate in the social world where a researcher deals tial that the law might change behavior, it should with human behavior, rather than the medical world be assumed that it does so. Moreover, the majority where Aspirin or smoking is under consideration. argues that the lack of evidence to support this But the claim is also not valid here either: negatives claim is not consequential because the nature of the can be demonstrated in the social world too. To see claim means that it is difficult to test. Here we demon why negatives are al.so readily demonstrated in the strate that Chief Justice Roberts' claim is not valid in social world, consider the opinion of Justice Stewart this particular case or as a matter of general principle. in Elkins v. United States, from which the claim orig Causal claims about the effects of a law can be tested, inates. Stewart is considering whether laws excluding even when the claim implies a negative effect. improperly obtained evidence deter law enforcement The claim tbat "it is never easy to prove a nega from conducting improper scarcbes: tive" appears to have originated in Elkins v. United States..364 U.S. 206(1960). From a scientific stand Empirical statistics are not available to show point. this is a strange proposition. To accept the that the inhabitants of states which follow claim is to ignore the principles of modern science the exclusionary rule suffer less from lawless and to discount the countless scientific studies searches and seizures than do those of states that have "proven" a negative effect.' Of course, which admit evidence unlawfully obtained. in practice, testing a claim, positive or negative, is Since, as a practical matter, it is never easy difficult in the complex social world. However, to prove a negative, it is hardly likely that con here we argue that, not only can we show very com clusive factual data could ever be assembled. pelling evidence for the negative—but that, unlike For much the .same reason, it cannot positively Roberts, we also do not need to make assumptions be demonstrated that enforcement of the crim about the effect of the law. Ample evidence is avail inal law is either more or less effective under able for scientific inquiry. In fact, the Court, by its cither rule. own actions, provides empiricists with tools that allow us to test for a negative or a positive. In June 2010, the Court issued an injunction that sus ''To be more precise, in scientific inference, the researcher will pended the distribution of matching funds during usually .set out to test a "Null Hypothesis" of no relationship between the variables of interest. If sufficient valid evidence the 2010 election cycle in Arizona. This interven e.xists that is not supportive of the Null Hypothesis, than the tion provides a unique opportunity to test for the .Null Hypothesis w ill be rejected in favor of evidence for a neg effects of matching provisions on overall campaign ative or positive effect. spending. This is a generally applicable tool that the ^Harrison v. United Slates, 392 U.S. 219 11968); United States V. Janis. 428 U.S. 433 (1976); Reno. Attorney General v. Boss Court should harness whenever it is useful. ier Parisli School Board et at. (95-1455). 520 U.S. 471 (1997). Consider the statement that a negative relationship To be cleiir, one sense of the phrase, "it is never easy to prove a is "never easy to prove." How could this be? Causal negative." is correct. Philosophers and logicians will remind us that we can never prove claims of non-existence. For example, relationships are always difficult to test, but to the no scientific analysis could ever disprove the existence of uni extent that we can ever answer causal questions, a corns. Santa Clans, or the Tooth Fairy. Certainly, the lack of negative effect is no more difficult to demonstrate empirical cs idcnee should tell us something about the likeli hood of the.se propositions, but any statement of non-existence than anything else. To see tliis, imagine where med (i.e.. a negative statement) camiot be proven—;ilthough this ar- ical science would be if we thought that we could gumeniunt ad ignorantiam is usually seen as a logical fallacy not test statements like "Aspirin is associated witli and previous courts have recognized it as an impossibly high burden of evidence (Hamling v. United Stales. 418 U.S. 87 a decrease in heart-attacks." Researchers demon (1974)). While negative propositions about existence cannot strate this sort of negative all the time—often through be proven, there is nothing special about causal claims that randomized, controlled experiments. Despite com have a mathematically negative sign. ELJ-2011-0143-ver9-Dowling_1P.3d 07/19/12 3:17pm Page 5
DOES PUBLIC FINANCING CHILL POLITICAL SPEECH?
The rise of modem, statistically based criminology limited speech. In many cases, causal inference— makes this statement outdated. Even in the 1960s, a that is, testing whether the action actually causes researcher could have addressed the question, given the outcome—is very difficult or impossible. Fortu the proper resources. Just as with Aspirin and heart nately, ample opportunities exist to lest for the attacks, a criminologist might approach this question effects of the law in question. by collecting data on searches and seizures in states Consider the empirical question at hand: did Ari with and without the exclusionary rule and then com zona's matching provisions inliibit speech? Even paring them. The researcher could then decide, based with easy access to data, we cannot easily demon on accepted standards of evidence, whether or not the strate this causal effect. Suppose that a researcher negative relationship had been demonstrated. If a rela decided to compare spending in states with match tionship between the exclusionary rule and searches ing provisions (Arizona, Connecticut, and Maine) and seizures existed, it should be revealed. However, to those without matching provisions to see whether at least two objections might be raised. First, that data states with matching provisions see less spending. separating "lawless searches and seizures" from other This would not answer the question at hand, searches and seizures might not exist. This is a valid because the states that adopted these provisions concern that can apply to any scientific domain: if were probably different to begin with. Those differ there is no data, empirical investigation is not possi ences between states might explain any differences ble. Fortunately, in the case at hand, evidence is read in campaign finance law and campaign spending. ily available: campaign spending and contributions Moreover, campaign spending probably influences are a matter of public record. Second, and more cm- campaign laws just as much as laws influence cially, a sophisticated critic might point out that, spending, so a strong correlative relationship even if the data did exist, a reseai'cher could not estab would not reveal the direction of causation. For lish a causal relationship between the exclusionary example, maybe these states felt free to adopt the mle and the lawless searches and seiziu'es. There laws because candidate spending was already might be a relationship betu'een the exclusionary low. Now suppose a second researcher proposes mle and lawless searches and seizures, but the rela another approach; she will test for changes in cam tionship does not mean that the exclusionaiy rule paign spending immediately after the adoption of causes fewer lawless searches and seizures. The num a new law. This strategy is an improvement, but ber of lawless searches and seizures might be corre we would still worry that any changes in behavior lated with exclusionary mles for reasons that are not were explained by other changes around that time. accounted for. This is the cmeial and more difficult For example, a state might only adopt a new law point. In Arizona, we can easily determine whether when the amount of political speech is expected spending went up or down after the implementation to change in the future. of the law. However, it is more difficult to assess Randomized experiments are the ideal method whether the law caused spending to change. This is for assessing causal effects. In an ideal world, we where the power of the court to issue injunctions could test for the effect of matching provisions becomes particularly useful. by randomly assigning the law to be in place in As we show in the following sections, the empir some states and not others. This study can serve ical claim surrounding McComish can and should be as a model for future assessment of the causal tested. Our tests provide the opportunity for a nega effects of the law. For most legal questions, we tive effect to manifest itself. However, the tests cannot design the ideal, randomized experiment. reveal that matching provisions have no detectable In the case of Elkins, we cannot randomly assign negative effect on political speech. some states to use the exclusionary rule and other states to not, just like we cannot randomly assign campaign finance laws. For this reason, social sci HOW CAN WE ASSESS THE CAUSAL entists must search for opportunities where the EFFECTS OF LAWS? conditions of an experiment have been imple mented without interference by the researcher or While the Court has applied strict scrutiny to by those being studied, often referred to as "natu campaign laws on the basis that they mighi impede ral experiments" (see, for example, Robin.son. speech, the underlying logic is that the laws cause McNulty, and Krasno 2009). ELJ-2011-0143-ver9-Dowling_1 P.3d 07/19/12 3:17pm Page 6
DOVVLING ET AL.
Occasionally, institutional features of a law or a this analysis as a model for future investigation natural event closely mimic the ideal experiment. regaiding the effects of campaign finance law. We In this case, we identify two such opportunities. now to turn to our tests as a demonstration of First, matching funds are only triggered when a pri how empirical data can be harnessed to address vately funded candidate spends above a specific causal claim.s—even negative ones—in questions threshold. Therefore, if matching funds truly inhibit of law. spending, candidates should cluster their spending just below the threshold. A strategic candidate that would otherwise spend above the threshold might NONPARTICIPATING CANDIDATE hold back their spending to avoid triggering the SPENDING DID NOT CLUSTER BELOW matching funds for their opponent.' This threshold THE THRESHOLD is a unique feature of Arizona's law that we can exploit for the purpose at hand. Second, the June If nonparticipating candidates in Arizona con 2010 Supreme Court injunction serves as a natural sciously hold back their campaign spending as a experiment where Arizona's law was suddenly result of the matching provisions, we would expect deactivated. When the court issues injunctions, some clustering just below the triggering threshold. they often unintentionally act as social experiment As noted in the majority opinion, in order to maxi ers, giving researchers the opportunity to examine mize the available funds, candidates can. of cour.se, the effects of the law in a way that makes clear simply participate in the public financing option. causal inferences. In this sense, the court acts as However, not all candidates did participate, so both investigator and judge, providing the evidence among those that are privately financed and want for their own answer to the question. As such, the to win their elections, if the law was a burden, we Court itself allows the opportunity to avoid reliance should see this reflected in their spending behavior. on proclamations like that of Chief Justice Roberts We can assume that every candidate should spend as that the effect of the law is self-evident. much as they have to spend to win. However, if can The legal world generally provides a promis didates react strategically to the expectation that ing place to look for natural experiments in socio their opponent will receive public funding, as peti political processes, because the courts are isolated tioners have argued, then we would expect some from immediate socio-political influences. In an candidates to decide that the advantage they gain experiment, the researcher intervenes into the rela from any amount spent above the threshold will tionship between tlie subject and the outcome of be a net disadvantage because of the spending it interest, perhaps by administering a drug or some would trigger for their opponents. These candidates other treatment. Importantly, the researcher inter\'cnes should spend every dollar up to ilie threshold (be- for reasons that arc unrelated to the characteristics of cau.sc their opponent has this amount anyway) and the subject. After the intervention, the researcher can then slop their spending. If candidates think this measure its effect. In the social world, where we want way. as the petitioners assert, then we should .see a to understand the effects of laws, the courts often significant clustering of candidates near the thresh intervene into the relationship between the law and old amount. If we see candidates clustered near the the law's effects—for example by issuing an injunc threshold, this would be key evidence of a burden. tion. Although this injunction has not been randomly To test for clustering, we collected general elec assigned, it is a sudden intervention that allows tion spending data for every nonparticipating state researchers to assess its effects. By examining behav legislative candidate in Arizona for 2006, 2008, ior before and after the injunction and by comparing and 2010 that ran against at least one participating Arizona to other states, we can understand the effect of the law. Here we examine the effect of matching provisions ^Alternatively, a strategic catididate might hold back tlieir on election spending in Arizona, using the triggering spending until right before Election Day to avoid triggering threshold and tlie Court's injunction as evidence. We matching funds with sufficient time for the publicly funded can find no evidence that the law affected the spending didate to respoitd. If such behavior was common, we would observe a cluster of spending just above the triggering thresh of non-participating candidates. Although it is diffi old. As we show below, we find no evidence for this .sort of clus cult to assess the causal effects of laws, we offer tering either. ELJ-2011-0143-ver9-Dowiing_1P.3d 07/19/12 3:17pm Page 7
DOKS PUBLIC FINANCING CHILL POLITICAL SPEECH?
.5 1 1.5 2 0 .5 1 1.5 2
General Spending I Threshold
FIG. I. Distribution of spending by privately funded (nonparticipating) state legislative candidates facing publicly funded oppo nents in Arizona (General Elections. 2006-2010). Note: The figure shows the distribution of camp.tign spending by privately funded Slate legislative candidates in Arizona who faced a publicly funded opponent. .Spending is re-scaled so that the triggering threshold is 1. Under the matching provisions in 2006 and 2008. wc see no evidence that candidates clustered their spending below tJie triggering threshold. In 2010 after the matching provisions were out-of-effect, the distribution looks almost identical.
candidate. For each nonparticipating candidate we We see no evidence that nonparticipating can divided his or her expenditures by the applicable didates intentionally spend below the threshold. If non- statutory triggering threshold (517,918 for 2006, participating candidates were intentionally spending 519,382 for 2008, and 521,479 for 2010). This below the threshold, wc might expect a cluster of mcan.s that a candidate that spent exactly as much spending just to the left of 1 (the vertical line). How as the triggering threshold (for example, 517,918 ever, there arc just as many nonparticipating candi in 2006) receives a value of 1. dates spending barely above the ilircshold as there Figure 1 shows the distribution of spending for arc nonparticipating candidates spending just below. privately funded (nonparticipating) candidates who Moreover, in no year is there a large amount of candi faced a publicly funded opponent in each of the dates around the threshold (just below or above it). three election cycles we analyze.^ The horizontal A more detailed examination of(he data reveals a axis displays the converted spending Figures (non- pattern that is strikingly unsupporiive of the asser participating candidate spending in the general elec tion that candidates were withholding their spend tion divided by the triggering thre.shold), while ing. For 2006, 2008, and 2010 there were 153 the vertical axis displays the number of nonpartici nonparticipating candidates for state legislature run pating candidates that fall into each bin.'"' The trig ning against a participating opponent. Of those, 91 gering threshold is denoted by the solid vertical line. Most nonparticipating candidates spend well below the threshold. The most obvious feature of this figure is the clustering near zero dollars in For visual purposes, we do not show the eandidales (less than 10'^' of the sample) who spent more ilum two times the tlircshold. spending. Many of these candidates were probably 'Each bin is etiuivalem to one-tentli of the spending threshold, cither likely to win by a large margin, uncompcli- such thai the bin to the left of the vertical line includes candi- tive, or unable to qualify for public funding. The date.s who spent .somewhere between nijie-ienih.s of the spend ing limit and c.xaclly the spending limit. The size of the bins is area of interest in these figures is near the vertical immaterial to the conclusions we draw (resull.s available upon line. Do candidates cluster around the line? request). ELJ-2011-0143-ver9-Dowling_1P.3d 07/19/12 3:17pm Page 8
DOWLING ET AL.
nonparlicipating candidates spent less than $10,000, candidates to behave diftcrentiy in 2010 than 2006 which is well below the triggering threshold. These or 2008. Yet, Figure 1 suggests that tlie distribution candidates were not detencd hy matching funds of spending is similar for all three years, and there because they did not raise or spend enough funds is no evidence of clustering below the triggering for the threshold to be a relevant consideration. threshold in any year. This analysis would not have For example, in 2008, these candidates could have been possible without the court's injunction, which spent an extra $9,382 without triggering matching mimics an experimental intervention. funds. Of the remaining 62 candidates, 23 spent As an additional test, we conduct a parallel anal more than $30,000, well above the triggering ysis using general election campaign spending data threshold. These candidates also do not appear to for privately funded (nonptu-licipating) stale legisla be deterred by matching provisions. That leaves tive candidates who faced publicly funded oppo less than one-fourth of nonparticipating candidates nents in Maine between 2000 and 2010. Maine has whose spending could have been plausibly deterred a similar, although not identical, public financing by the matching provisions. Among these candi program with a matching funds provision for state dates, we see no evidence that they strategically legislative candidates. With the inclusion of Maine stop their spending below the threshold. in our analysis, we are moving towards a more Tliis analysis alone does not establish the causal robust test of the relationship between the law and effect of the law, While the distribution of spending candidate spending. When wc only examined Ari is suongly inconsistent with the claim of chilled zona. the pattern expected by the petitioners might spending, we cannot possibly account for all of the not emerge because of some oddity of Arizona many factors that contribute to the shape of the distri that we cannot account for. By examining another bution of spending. However, the Court's injunction state with a similar law, we are, essentially, giving in 2010 gives us the first opportunity to explicitly the petitioners two opportunities to prove their case. test the causal effect of the law. If the law had an Figure 2 presents the spending distributions effect on candidate spending, then we would expect among these candidates in Maine for each election
.5 1 1.5 2 0 1 1.5 2 0 .5 1 General Spending / Threshold
FIG. 2. Dislribulion of spending by privaicly funded (nonpaiticipaiing) state legislative candidates facing publicly funded oppo nents in Maine (General Elections. 2(}0()-2()lU). Note: The figure show.s the distribution of campaign spending tiy privately funded slate legislative candidates in Maine who faced a publicly funded opponent. Spending is re-scaled so that the triggering threshold is 1. Matching provisions were in effect for all six elections shown. We see no evidence that candidates clustered their spending below the triggering threshold in any of the elections. ELJ-2011-0143-ver9-Dovvling_1 P.3d 07/19/12 3:17pm Page 9
DOES PUBLIC FINANCING CHILL POLITICAL SPEECH?
cycle. A.s can be seen by the decrea.sing frequency thresholds. These empirical results lead us to con of candidates in each graph, moving from 2000 to clude matching provisions do not chill spending." 2010, the number of nonparticipating candidates That this threshold was set by an agent other than has decreased over time, presumably because candidates themselves, provides for an opportunity more and more eandidates have seen the program to assess the effect of the law. If this law created a as beneficial and chosen to take part. As in Arizona, distribution of spending tbat would not otherwise we can look for visual evidence of clustering below exist, it should be in relation to the threshold set by the corresponding triggering thresholds—with can the law. In this sense, the law itself provides an didate spending spread around the distribution, no opportunity to test for the effect of the law, much clear visual evidence emerges. However, we do like an experimenter intervening in a process and not have to rely on visual evidence alone. Below seeing if the results are different than if there was we proceed to a more rigorous statistical analysis no intervention. However, for reasons already dis that allows us to test for the presence of patterns cussed, there are limitations to what we can leam that may not be apparent to the naked eye. about the causal effects of this law from this The most rigorous statistical test for clustering, approach. We need something that more closely proposed by McCrary (2008), tests for a discontinu approximates an experiment. Next we tum more ity of the density function at the threshold by con fully to tlie natural experiment provided by the Court. ducting local linear regression on both sides of the threshold.'^ This procedure builds on previous work applying the principles of regression to the THE 2010 INJUNCTION DID NOT ALTER estimation of a density function at a boundary or CANDIDATE OR INDEPENDENT threshold (Cheng, Fan, Marron 1993). Appendix GROUP SPENDING Table 1 presents statistical estimates for the extent of clustering around the threshold for the analysis As discussed, the Court's 2010 injunction serves of candidate spending in Arizona. The coefficients as a natural experiment, allowing us to test for the we report can be interpreted as the difference causal effect of matching provisions. The Court between the number of nonparticipating candidates only intervened in one state, making Arizona a spending just below the threshold and the number of treatment group, while the other states serve as a nonparticipating candidates .spending just above the control. The petitioners claimed that the law caused threshold. Positive values would indicate clustering less spending in Arizona than would have occurred below the threshold and negative values would indi without the law. If the law did influence spending, cate clustering above the threshold. This analysis then we should see a change in Arizona, relative reveals no statistically significant evidence of clus to other states, after the Court's intervention. If tering in Arizona or Maine. Positive coefficients matching provisions had previously had a cfiilling would serve as evidence of clustering. The cluster effect on spending by nonparticipating candidates, ing coefficients are all subslantively small and sta we should see their spending increase dramatically tistically indistinguishable from zero (the smallest in Arizona relative to other states in 2010. p-value is .33). We return to spending data for nonparticipating We can again look at 2010 as an experiment that stale legislative candidates to test this hypothesis. allows us insight into the causal effect of the law. In 2010, candidates in Arizona decided whether or The coefficients for 2006-2(X)8 and 2010 are verx' not to participate in public funding before the similar. This indicates that, even with the law not injunction occurred. This means that their choice in effect in 2010, in terms of spending the candidates to participate in the election could not have been behaved in a very similar manner to the 2006-2008 influenced by the injunction, suggesting that the period when the law was in effect. This is evidence that the law had no discernible effect causing candi dates to hold back their spending. '"This method requires arbitrary decisions to be made about the In short, both the visual and statistical analyses bin si/.e and the window of analysis. However, the results we for clustering around the triggering thresholds in pre.scnt here are in.sensiiivc to changes in these specifications. 'This conclusion is consislciil with those made by Gicrzynski Arizona and Maine do not reveal evidence that can (2011). in his own analysis ot the effect of .Maine's public fund didate spending is strategic around the triggering ing program on fundraising and spending. ELJ-2011-0143-ver9-Dowling_1 P.3d 07/19/12 3:17pm Page 10
10 DOVVLING ET AL.
types of candidates participating should he similar for 2006. 2008, and 2010. As such, this allows for a valid comparison of the spending of candidates across those years and the test approximates a ran domized experiment in which subjects would be assigned to the treatment and control conditions by the researcher to ensure the validity of comparison. At first glance, 2010 was not an extraordinary yeai- in terms of campaign spending. Among pri vately funded state legislative candidates who ran r-^ against a publicly funded opponent. 2010 spending Arizona was similar to that in 2008 ($15,223 per candidate versus $16,172). This initial comparison suggests 2006 that the injunction did not have a significant effect Yaw on the spending behavior of state legislative candi FIG. 3. Spending ircnds for privaicly funded (nonparticipat dates. The magnitude of spending was no different ing) candidates in Arizona and Maine (2(K)2-201()). Nore: Ttie figure .shows ihe level of campaign spending by privately funded under a system of matching provisions than it was state legislative candidates in .Arizona and Maine between 2(K)2 when the provisions were removed. Tn fact, spend and 2010. The 2010 Court injunction took matching provisions ing was even lower in 2010 than it was in 2008. out-of-cffect in Arizona in 2010. If these provisions inhibit spending, we should see spending increase dramatically in This is evidence that matching funds did not chill 2010 in Arizona relative to Maine over the same time period. the spending of nonparticipating candidates; if However, we find no such effect. matching funds had chilled spending, we should observe more spending in 2010 when the injunction had been issued. campaign spending for all state legislative and state However, there are many factors that affect wide candidates from 2002 to 2010 who opted out spending year to year—for example, a lack of pub of public funding. Both states exhibit a similar lic interest in the campaign that year—so perhaps trend in spending from 2002 to 2008. If matching spending in Arizona would have been exception provisions chill political spending, we should ally low in 2010, but the injunction caused spend expect that the removal of these provisions in Ari ing to return to normal levels. We can try to zona in 2010 would cause spending in Arizona to account for this by comparing spending in Arizona increase more precipitously relative to spending to other states over time, with these states .serving in Maine. However, the evidence shows that per as a "eontrol" group. Of course, every election capita spending in Maine actually increased more year is different, and the magnitude of spending so than in Arizona between 2008 and 2010. Over will vary with the political climate of the lime. this period, nonparticipating candidates in Maine For example, spending in stales holding elections increased their spending by 7.4 dollars per person, for high-profile, statewide offices, like governor, while Arizona's nonparticipating candidates only will likely be higher than in elections in slates with increased spending by 1.3 dollars per person. This out statewide contests. However, states should is evidence that the injunction, that is the treatment exhibit similar trends in campaign spending over of removing the law, did not have any effect on time. Therefore, we can compare changes in spend spending, indicating that the law itself was not ing in Arizona in 2010 to the spending changes in affecting spending. other states to obtain a better estimate of the effect Of coiu:sc, a control group of only one state is of the injunction. problematic. There may be characteristics of As we note above, Maine's campaign finance Maine, which we do not account for. that cau.sed program is similar to Arizona's, and both states their spending to rise. For a stronger test, we col have similar levels of per capita campaign spending. lected data on general election campaign spending The crucial difference is that tlie Court did not issue between 2000 and 2010 for as many slates as possi an injunction in Maine, making Maine a natural ble. For each state-election, we obtained two control state to compare with Arizona for an initial figures: (1) the total amount of spending by state F3^ cross-state analysis. Figure 3 shows the trends in wide and state legislative candidates and (2) the ELJ-2011-0143-ver9-Dowrmg_1 P.3d 07/19/12 3:17pm Page 11
DOES PUBLIC FINANCING CHILL POLITICAL SPEECH? 11
total amount of spending by PACs and interest groups.'^ Appendix Table 2 shows the slate elec tions for which this data has been collected. All of these states hold their gubernatorial and statewide & races at the same time, so for the purposes of our Svrthetic ComtcJ' I analysis, we should expect that these states exhibit S similar trends in campaign spending over time. f 3 / /^fizooa Since we are using many states, rather than just Maine as our only control state, we can employ a statistical technique called synthetic control (see Abadie, Diamond, and Hainmueller 2010). This technique creates a composite of the states to serve as a baseline against which to compare spending 2006 Year changes in Arizona. This approach allows us to find the weighted average of other states that FIG. 4. Candidate spending in Arizona and synthetic control mo.st closely mirrors Arizona in terms of campaign group (2(K)6-2010). Note: The figure shows the level of cam paign spending by privately fimdcd state legislative candidates spending, population, and population growth, and in Arizona and a synthetic control unit between 2006 and 2010. other demographic or political variables. This pres The synthetic control unit is tlie optimal weighted average of ents us with a state that is well balanced as a control other stales that most closely mirrors Arizona along numerous political and demographic characteristics. The 2010 Court group to test against Arizona, which is what we injunction took matching provisions out-ol-ettcct in Arizona would want to obtain in a setting where we could in 2010. If tliese provisions inhibit spending, we should .see use a taie randomized experiment to assign states spending increase dramatically in 2010 in Arizona relative to the synthetic control unit over the same time period. However, to treatment and control. It also allows us to account we find no such effect. for objections voiced by Justice Kennedy during oral arguments that the change in population in a state had to test whether spending in Arizona in 2010 in- to be accounted for when considering changes in spending.'^ We do that directly here by including crca.sed more than would have been expected in the absence of the injunction. A difference-in population change in our regression models. differences method simply means we look at the Tlirough multiple tests and specifications, we differences in changes in spending between sets find no evidence that campaign spending increased of states. If the injunction exerted an effect, we in Arizona in 2010 relative to the synthetic control would expect the difference between Arizona in of other comparable states. If anything, the increase 2010 and previous years to be greater than the dif in spending in Arizona between 2008 and 2010 was ference between 2010 and previous years for other smaller than the increases in other states. states. This is a rigorous way to control for any pre Figure 4 provides one example of such a synthetic existing differences in .spending between states. At control analysis. For all states where we could obtain this point wo are coming close to approximating a candidate spending data for 2006, 2008, and 2010 laboratory experiment—one in which an experi (CA. ME, MI, and TN). we estimated the control menter would take identical groups, measure group that best matches Arizona in terms of per cap their condition before administering a treatment, ita candidate spending in 2(X)6 and 2008 and popula and then measure their condition after administer tion growth. This particular synthetic control group is ing a treatment. a weighted average of Tennessee (weight =.93) and In Appendix Table 3. we present the results of six California (weight =.07). Between 2008 and 2010, regression analyses designed to systematically test spending in the synthetic control group increased by 4.0 dollars per person, while spending in Arizona increased by only 1.8 dollars per person. This tells us '^Each state has different reporting requirements ami differing that even when looking at very similar states, we find degrees of data accessibility. However, for the purposes of hardly any evidence that the injunction changed our analysis, it is most important that the data is consistent spending by candidates. within each state. This is because the difference-in-differences design u.sed below compares differences within states, which As a final test of the effect of the 2010 injunction, effectively controls for any other differences between states. we can employ difference-in-differences methods '^Oral Arguments. Transcript page 32. line 16. ELJ-2011-0143-ver9-Dowling_1P.3d 07/19/12 3:17pm Page 12
12 DOVVLING KT AL.
for such an effect. The regression models test for the there is no systematic, empirical evidence of such effect of the 2010 injunction in Arizona while con a chilling effect. Moreover, our estimates are not trolling for different average levels of spending in plagued by omitted variables or other problems each state and for different spending levels in each because we exploit a natural experiment in which year (captured through the use of state and year the law was exogcnously manipulated. This is a fixed effects variables in the models). The table strong assessment of the causal effect of matching reports the estimated effect of the injunction on funds in Arizona (and Maine). We have shown, three types of spending separately: (1) total candi despite the claims of the Chief Justice, that a nega date spending, (2) total spending of privately tive effect could be shown, with a high degree of sci- funded, nonparticipating candidates, and (3) total entific rigor. spending by PACs and independent groups. The We emphasize that the tests we conducted could estimates are reported as both units of dollars have yielded a different answer. We could have spent per person living in the state and as logs of found that the Arizona law did significantly decrease those spending levels. The table also presents the candidate and outside group spending. The interpre standard errors associated with each estimate, indi tation of this fact would have been left to the Court. cating whether the estimates are statistically differ By conducting rigorous and methodologically sound ent from zero. tests, we were open to the po.ssibility of finding evi If matching provisions chilled campaign spend dence that supported either side of this far-reaching ing before 2010, we would expect a large, positive case. The ai-gument used by the majority, that their effect of the injunction on spending. For all three expectation is simply "evident and inherent" types of .spending, we fail to estimate a statistically could, of course, be used by either side of any argu significant or substantively meaningful effect of ment. However, there would be no objective way to the injunction. In fact, the models indicate that adjudicate between competing claims of inherent candidate spending actually decrea.sed after the truths. injunction in 2010. This is the opposite of what Prior studies have demonstrated that the Coiu-t should happen if spending was chilled. Focusing often has difficulty interpreting and harnessing on the analysis of private and independent spending, social scientific research (Rustad and Koenig we estimate that nonparticipating candidates spent 1993), so, perhaps, the majority's dismis.sal of about 3 dollars per person less than expected and empirical evidence should have been expected. independent groups spent 25 cents per person However, even if the majority had chosen to engage more than expected on average. Neither result, the evidence weighing against the petitioners' however, is statistically significant or substantively claim, the essential empirical question, that of a meaningful. In sum, we fail to find any evidence causal effect of the law on candidate speech, may that the injunction increased spending or that still have been difficult to answer. Establishing matching provisions inhibited spending.'"* The causal claims are among the most vexing problems lack of effect on independent spending is notewor in all of scientific inquiry, but we have shown that, thy, considering the attention devoted to it by the in this case, the Court was distinctly able to address Court. On average, Arizona candidates appear to this question of causality. This can serve as a model have spent Just as much in the presence of matching for the potential investigatory power of judicial provisions as they would have in the absence of injunctions. When the Court chooses to entertain such a policy. empirical evidence, its own actions can create a uniquely powerful social experiment.
CONCLUSION
Wc conducted more than ten statistical analyses For an even simpler test of tlie liypoilicsis of a ehitl, wc can to determine whether there is empirical support e.siimate these .same regression models without controlling for for the claim that public financing matching provi different average levels of spending in each state (state fixed sions chill spending, and thereby speech, in elec effects) and for different spending levels in each year (year fi.xed effects). These .specifications also fail to detect any statis tions. The evidence we present consi.stently points tical or substantive evidence for increased spending after the in the same direction and leads us to conclude Court's injunction. ELJ-2011-0143-ver9-Dovvling_1P.3d 07/19/12 3:17pm Page 13
DOES PUBLIC FINANCING CHILL POLITICAL SPEECH? 13
REFERENCES from Wisconsin. 1964-1990." Ixtgislative Studies Quar terly 20i I): 69-88.
Abadic. Alberto, Alexis Diamond, and Jens Hainmueller. 2010. McCrary, Justin. 2008. "Manipulation of the Running Variable "Synthetic Control Methods for Comparative Case Studies: in the Regression Discontinuity Design: A Density Test." Estimating the Effect of California's Tobacco Control Pro Journal of Econometrics 142(2): 698-714. gram." Journal of the American Statistical Association Miller. Michael G. 2008. "Gaming Arizona: Public Money and 105(490): 493-505. Shifting Candidate Strategies." PS: Political Science «£ Cheng..Ming-Yen. Jianqing Kan, and J.S. Matron. 1993. "Mini- Politics 4\0y. 527-532. max Efficiency of Local Polynomial Fit Estimators at Miller. Michael G. 201 la. "After the GAO Report: What Do We Boundaries." Unpublished Manuscript Series #2098. Insti Know .About Public Election Funding?" Election Law tute of Statistics, University of North Carolina. Journal 10(3): 273-290. Dowling. Conor. 2011. "Public Financing and Candidate Par Miller. Michael G. 201 lb. "Public Money. Candidate Time,and ticipation in Gubernatorial Elections." In C. Panagopoulos, Electoral Outcomes in State Legislative Elections." In C. ed. Public Financing in American Elections. Philadelphia: Panagopoulos, cd. Public Financing in American Elections. Temple University Press. Philadelphia: Temple University Press. Esenberg. Richard. 2011. "Playing Out the String: Will Public Milyo, Jeffrey. David M.Primo. and Matthew I.. Jacobsmeier. 2011. Financing of Elections Surt ive McComish i. Bennett!" "Does Public Funding of .State Election Campaigns Increa.sc Election Law Journal 10(2): 165-173. Voter Tiunoiit?" In C. Panagopoulos, cd. Public Financing in Government Accountability Office. 2010. Campaign Finance American Elections. Pliiladelphia: Temple University Press. Reform: Experiences of Two Stales That Offered Full Pub Panagopoulos, Costas. 2011. "Introduction." In C. Panagopou lic Funding for Political Candidates. G AO-10-390. Wash los, ed. Public Financing in American Elections. Philadel ington, D.C.: United States Government. phia: Temple University Press. Gier/.ynski, Anthony. 2011. "Do Maine's Public Funding Pro Primo. David M.,and Jeffrey Milyo. 2006. "Campaign Finance gram's Trigger Provisions Have a Chilling Effect on Fund Laws and Political Efficacy: Evidence from tlie States." Raising." Typescript, University of Vermont. Election Law Journal 5(1): 23-39. Gora. Joel M. 2011. "Don't Feed the Alligators: Government Robinson, Gregory, John E. McNulty. and Jonathan S. Krasno. Funding of Political Speech and the Unyielding Vigilance 2009. "Observing the Counterfactual? The Search for of the First Amendment." Brooklyn Law School. Research Political Experiments in Nature." Political Analysis 17(4): Paper No. 249. September. 341-.357. Kraus, Jeffrey. 2006. "Campaign Finance Reform Reconsid Rustad, Michael, and Thomas Koenig. 1993. "The Supreme ered: New York City's Public Finance Program After Fif Court and Junk Social Science: Selective Distortion in Ami- teen Years." The Forum 3(4): 1-27. cus Briefs." North Carolina Law Review 72(1): 91—162. Lowenstein. Daniel Hays. Richard L. Hasen, and Daniel P. Stem, Robert. 201 1. "Public Financing in the States and Tokaji. 2008. Election Law: Cases and Materials. 4th cd. Municipalities." In C. Panagopoulos. ed. Public Financ Durham, NC: Carolina Academic Pre.ss. ing in American Elections. Philadelphia: Temple Univer Malhotra. Neil. 2008. "The Impact of Public Financing on sity Press. Electoral Competition: Evidence from .Arizona and Maine." State Politics & Policy Quarterly 8(3): 263-281. Adiiress correspondence to: Mayer, Kenneth R.. Titnothy Werner, and .Amanda Williams. 2006. "Do Public Funding Programs Enhance Electoial Anthony Fowler Competition?" In M. P. McDonald and J. Samples, eds. Department of Government The Marketplace of Democracy: Electoral Competition Harvard University and American Politics. Washington, D.C.: Brookings Insti 1737 Cambridge St tution Press. Cambridge. MA 02138 Mayer. Kenneth R.. and John M. Wood. 1995. "The Impact of Public Financing on Electoral Competitiveness: Evidence E-mail: fowler®'fas.harvard.edu
(Appendix follows —►) ELJ-2011-0143-ver9-Dowling_1 P.3d 07/19/12 3:17pm Page 14
14 DOVVLING ET AL
APPENDIX
Api'enuix Table 1. Statistical Tests for Clusterinc Appendix Table 2. Available Campaign I-inance Data Around the Triggering Threshold BY State and Year
Coefficient Staiulard Error p-vahie 2000 2002 2004 2006 200H 2010
Arizona 2(K)6-2(K)8 -0.30 0.94 0.75 AK Arizona 2010 -0.42 0.43 0.33 AZ Maine 2002-2010 -2.90 7.10 .69 CA CO Note: Following the recommendations of McCrarv' (20081, the table provides explicit statistical tests for clustering around the triggering CT threshold in three sets of elections. The coefficients can be interpreted FL as the number of candidates clustering just below the threshold. In all ME three cases, the negative coefficients indicate clustering above the MI threshold, exactly opposite what we would sec if candidates strategi OR cally chilled their spending in respon.se to the matching provisions. TN The coefHcients ;ue statistically and substanlively indistinguishable from zero, suggesting that cantiidates do not strategically chill their Note: A value of 2 indicates that both candidate and independent spend spending as a result of the campaign tinance law. ing data is available for this election. A value ol 1 indicates that only candidate spending is available. No entry indicates that no data was available.
Appendix Table 3. The Ei fecf of the 2010 Injunction in Arizona on Campaign Spending
Candidate Spendini' Private Spending Independent Spending
per capita log per capita log per capita log
Injunctiun -5.45(3.90) -0.43 (0.38) -2.62 (2.22) -0.06 (0.33) 0.25 (1.81) 0.63 (0.47) Constant 8.21 (1.64)" 14.95 (0.22)* 10.12 (1.25)* 15..39 (0.22)" 7.63 (0.81)* 14.70 (0.23)* Observations 40 40 37 37 30 30 R-st|uarcd 0.64 0.90 0.76 0.93 0.86 0.86 SHR 4.04 0.50 2.54 0.50 1.95 0,86 Note: Ordinary least squares (OLS) regression coefficients with state-clustered standard errors in parentheses. All models include state and year fixed effects.'Denotes statistical significance at the p < .05 level. The coefficient on "Injunction" represents tlie effect of the 2(110 Court injunction in Arizona on the dependent variable. In all six models, we find no substantively or statistically significant evidence that the injunction (i.e.. the removal of matching funds) altered the spending behaviors of independent groups and privately funded candidates. Political Research Quarterly XX{X) 1-12 Spanish-Language Radio © 2010 University of Utah Reprints and permission: Advertisements and Latino sagepub.com/journalsPermissions.nay DOI: 10.1 177/1065912910367494 Voter Turnout in the 2006 http://prq.sagepub.com ®SAGE Congressional Elections: Field Experimental Evidence
Costas Panagopoulos' and Donald P. Green^
Abstract Because Hispanic voters are seldom targeted for campaign communication and because they listen to radio at higher rates than non-Hispanics, Spanish-language radio represents an attractive venue for testing whether nonpartisan mass media messages can mobilize voters. We conducted a large-scale, national field experiment testing the impact of nonpartisan Spanish-language radio advertisements on Latino voter turnout in the 2006 congressional elections. The experiment, encompassing 206 congressional districts, indicates that nonpartisan radio ads represent an effective and cost-efficient means of raising Latino turnout in federal elections.
Keywords field experiment, voter mobilization, radio advertising. Latino voting behavior, electoral campaigns, congressional elections
Hispanic' participation in elections lags behind that of rates of citizenship (Michelson 2005) and low socioeco non-Hispanic groups. In 2008, 54 percent of voting- nomic status are said to account for part of the discrep eligible Hispanic citizens were registered to vote in the ancy in participation (Garcia and Sanchez 2008, 139). 2006 midterm elections, compared to 71 percent of non- Other scholars point to low levels of political information Hispanic whites and 61 percent of African Americans and interest in public affairs (Garcia and Sanchez 2008, (U.S. Census Bureau, File 2008). Among eligible voters, 140). Although surveys reveal relatively high levels ofcivic 32 percent of Hispanics voted in 2006, compared to 41 duty, political efficacy, and patriotism—attitudes typically percent of African Americans and 52 percent of non- associated with higher tumout levels—Hispanics vote with Hispanic whites. Even among registered voters, Hispan lower frequency than other ethnic groups (dc la Garza, ics voted at rates lower than non-Hispanic whites and Falcon, and Garcia 1996). This remains a puzzle, given African Americans in 2006: 60 percent compared to 72 that many of the structural and institutional barriers— and 67 percent, respectively (File 2008). Although racial- including onerous registration requirements, English- ized politics (de la Garza, Menchaca, and DeSipio 1994) language-only ballots, and literacy tests—that inhibited or elections with competitive Latino candidates on the Hispanic participation historically have been dismantled. ballot (Barreto 2007), for example, may energize Hispanic Another set of theories focuses on mobilization- voters, most studies conclude that in comparison to other centered explanations to account for relatively low levels ethnic groups, Hispanic voters are, on average, less likely of Hispanic electoral participation. Limited resources to engage politically. Lower levels of Hispanic participa tion are seen across a broad range of political activities 'Deparcment of Political Science. Fordham University, Bronx. NY, USA such as contributing to candidates, volunteering in cam ^Department of Political Science. Yale University. Westport. CT. USA paigns, and contacting elected officials (Verba, Schloz- Corresponding Author man, and Brady 1995; Garcia and Sanchez 2008). Costas Panagopoulos. Department of Political Science, Fordham Several theories have been advanced to explain His University. 441 E. Fordham Rd.. Faber 667. Bronx, NY 10458, USA panics' relatively low levels of electoral engagement. Low Email: [email protected] Political Research Quarterly XX(X) compel campaigns to target their mobilization efforts to radio evidently played a key role in marshalling Latino segments of the population they perceive to be most recep support and participation for the massive immigration tive, often at the expense of Hispanies. who tend to have demonstrations against House Resolution 4437 during low propensities to vote (Panagopoulos and Wielhouwcr the spring of 2006 (Ramirez 2007). 2008; Gcrshtcnson 2003; Roscnstonc and Hanson 1993).^ Spanish-language appeals to Hispanic radio listeners Although some argue that campaigns arc increasingly may be effective not simply because they overcome lan courting Hispanic voters (Segal 2002), several studies guage barriers but also because they may activate ethnic reveal Hispanies are routinely neglected by campaigns' identification. While Latinos in the United States typi mobilization efforts relative to other ethnic and racial cally pursue acculturation (learning and adapting sel groups (Leighley 2001; Verba, Schlozman, and Brady ected cultural traits and patterns of the dominant society), 1995; Ramirez 2005; Hero et al. 2000). In the 2000 presi many simultaneously espouse pluralism (selectively dential election, for example, Democrats and Republicans keeping and expressing eultural and other traits of their aired more than 275,000 advertising spots on television Latino heritage). Moreover, Spanish-language outreach in the nation's top seventy-five media markets; but only may serve an important symbolic function, revealing to 3,900 of these, or 1.4 percent, were Spanish-language or Hispanic voters that campaigns recognize their impor bilingual advertisements (Obcrfield and Segal 2008). tance in the American electorate (Abrajano 2010). Notwithstanding anecdotal evidence that both major par Subervi-Velez (2008a, 53) contends that "even highly ties courted Latino voters aggressively in the 2004 presi acculturated Latinos would not be immune to the candi dential election (Segal 2004),the 2004 American National dacy and messages of a Latino candidate, nor to the Election Studies (weighted) indicates that only 16 per appeals that non-Latino candidates make to them as Lati cent of Hispanies overall were contacted by the political nos." Similarly, Guernica (1982. 5) argues that "U.S. parties in that election cycle, compared to 26 percent of Hispanies are most receptive to media content in the blacks and 49 percent of whites. When the sample is Spanish language. Spanish programming elicits an emo restricted to battleground states in 2004 to account for tional response from the Hispanic audience that is miss strategic targeting in the most competitive states, the dis ing in English-language media." crepancy persists: 26 percent of Hispanies in battleground Our research examines whether Spanish-language states were contacted in 2004, compared to 49 percent of media can be used to raise Hispanic voter turnout. In an blacks and 64 percent of whites in these states. attempt to reach the Hispanic audience that is ordinarily Another possibility is that pervasive Spanish raonolin- ignored by political campaigns, we broadcasted nonparti- gualism in the Latino community inhibits participation san radio advertisements in 206 uncompetitive congres (Ramirez 2007; Uhlaner, Cain, and Kiewiet 1989). Cen sional districts shortly before the November 2006 elections. sus data indicate that 28.1 million Americans five years Our focus on mass media represents something of a of age and older spoke Spanish at home in 2000. Almost departure from recent work on Hispanic voter mobiliza half of these people—13.8 million—did not speak Eng tion. A scries of experimental studies demonstrate that lish well or at all. Just 28 percent of respondents to the grassroots mobilization efforts targeting Hispanic voters 2006 Latino National Survey indicated they could carry can raise tumout, while more impersonal tactics such as on a conversation "very" or "pretty" well in English direct mail seem ineffective (Michelson 2003, 2005, (Latino National Survey 2006). 2006; Ramirez 2005). But grassroots campaigns arc dif Given these figures, it is not surprising that Hispanies ficult to organize, and political campaigns have rarely rely heavily on Spanish-language media. In 2004, the directed large-scale canvassing efforts at the Hispanic Pew Hispanic Media Study found that about two-thirds community. In light of these constraints, the question is of the adult Hispanic population in the United States got whether alternative forms of voter mobilization, such as at least some news from Spanish-language media and mass media campaigns, generate Hispanic votes in a that Latinos regard Spanish-language media more favor cost-effective manner. Recent experimental studies that ably than English-language media (Suro 2004). The 2006 examine the impact of mass media advertising on politi Latino National Survey found that 45.9 percent of respon cal behavior find positive effects (Panagopoulos and dents reported they relied more heavily on Spanish- Green 2008; Panagopoulos 2006; Green and Vavreck language television,radio, and newspapers for information 2008), but these statistical results are equivocal, and they about public affairs and politics, compared to 29.3 per have not targeted ethnic audiences. This study builds cent who relied more on English-language sources (23.8 upon that work to deploy field experimental techniques percent indicated they relied equally on bolh).^ Spanish- to examine the impact of delivering political messages language media may provide a erucial connection to poli via radio advertisements that target Hispanic voters. tics. Reminiscent of black radio during the civil rights This article is organized as follows. First, we describe movement in the 196()s(Barlow 1998), Spanish-language the growth of Spanish-language media and provide an Panagopoulos and Green
overview of the literature on mass media's effects on question of whether radio advertising affeets outeomes voter participation. Next, we describe the key features of or merely represents a marker for better-funded and our experimental design: the procedure by which the more professionalized campaigns. Descriptive studies experimental sample was created, the way in which find that candidates make widespread use of radio at observations were randomly assigned to treatment and both the federal and municipal levels (Herrnson 2000; control groups, and the content and timing of the radio Strachan 2003). Using survey data from two states with campaign. We then explain the statistical models used to competitive Senate races in 2002, Overby and Barth test the hypothesis that these radio ads enhance turnout. (2006) argue that political advertisements broadcast on We conclude by commenting on the theoretical and pol radio exert greater influence than do television ads icy implications of our results and by suggesting direc beeause radio ads impart more information as a result of tions for futiue research. greater repetition and length. But again, this nonexperi- mental evidence may not show convincingly that radio advertising works best. Hispanics and Radio The first field experiments to test the impact of radio Changing demographics have fueled new trends in the advertisements on voting behavior were condueted in the mass media environment. Latinos' annual purchasing context of municipal elections in 2005 and 2006. Pana power of more than $960 billion has attracted the atten gopoulos and Green (2008) find that radio ads have posi tion of advertisers. Spanish-language radio has experi tive but statistically insignificant effects on turnout. enced tremendous growth—a tenfold increase in the Applying a similar experimental approach to a different number of Spanish-language radio stations in the United electoral context and target audience, this study takes up States since 1980, to 730 in 2007 (Castaneda Paredes the question of whether radio mobilized Hispanic voters 2003). Whereas radio generally has struggled to maintain in the 2006 congressional elections. Given how seldom listeners in recent years, Spanish-language radio audi Hispanic voters are targeted by campaign communica ences are booming (Arbitron 2007). Exposure to radio is tions, we antieipate detecting effects that are as large as, currently higher among Hispanics than non-Hispanic or larger than, previous studies. whites, African Americans, or Asians (Multicultural Mar keting Equation Study 2007). According to the 2004 Pew Experimental Design Hispanic Media Study cited above, only television sur passes radio as the main source of news amongst Hispan Political scientists have increasingly turned to field e.xp- ics (Suro 2004). Of special appeal to advertisers is the eriments to isolate the impact of various activities on vot fact that Spanish-language radio listeners regard radio as ing behavior (Gerber and Green 2000; Green and Gerber a reliable source of information. For those interested in 2008). Experimentation is a researeh method in which votes rather than sales, the unique features of Spanish- units of observation arc assigned randomly to treatment language radio give it enormous mobilization potential. It and control groups. The method corrects many of the reaches voters who listen intently and who receive few deficiencies of obscr\'ational approaches, which often rely competing political messages(de la Garza, Brishetto, and on self-reported media exposure, and permits researchers Vaughan 1983. 25-26; Subervi-Velez 2008b, 370). to draw valid inferences about causal effects. Field exper Despite its potential importance, radio's effects on iments, as distinct from laboratory experiments, study the Latino political behavior has received little attention effects of an intervention within a naturalistic setting and from researchers (Ramirez 2007; Subcrvi-Vclez 2008b). strive to use unobtrusive outcome measures. In this case, The dearth of research on Spanish-language radio reflects the units of observation are congressional districts, the a lack of scholarly attention to the political effects of radio intervention is a radio campaign, and the dependent vari more generally. able is voter turnout as recorded in public records. This Although the mass media and radio in particular have section describes the experimental design. for decades attracted scholarly attention (e.g., Bcrclson, Sample construction. The overall sample included the Lazarsfeld, and McPhec 1954), the number of studies population of congressional races with contested incum that systematically evaluate the effeets of radio on vot bents seeking reelection in November 2006. To isolate the ing behavior is surprisingly small. Geer and Geer(2003, effect of the intervention by avoiding competing commu 70) focus on voter reactions to positive versus negative nications, we excluded competitive races."* We also exc radio advertising and acknowledge that radio's effects luded districts in states for which data about prior history on voting behavior remain largely unaddressed. Nonex- of Hispanic voters was unavailable from commercial data perimental studies find that candidates who broadcast aggregators. Due to cost constraints, we also excluded ads on radio tend to do better than those who do not congressional districts that were mainly located in the (McCleneghan 1987), but these studies leave open the Los Angeles or New York City media markets.^ Political Research Quarterly XX(X)
The remaining sample included a total of 206 con Table i.Treacment Districts and Random Assignments of gressional districts in twenty-eight states. As detailed Gross Ratings Points (GRPs) below, a total of 36 districts were randomly selected to State District GRP buy be treated, and the remaining districts became the con trol group. Appendix 1, available at http;//prq.sagcpub California 1 50 California .com/supplemental/, provides a complete list of treat 3 75 California 13 50 ment and control districts. California 18 100 Radio treatment. Districts in the treatment group were California 21 75 exposed to 60-sccond radio advertisements that presented California 41 50 a nonpartisan get-out-the-vote message to listeners. The California 44 50 size of the media buy was varied in each district so that California 49 50 districts were exposed to 50, 75, or 100 gross ratings California 52 50 points(GRPs) of radio advertising.^ To conserve money, California 53 50 purchase of 75 or 100 GRPs was restricted to less expen Florida 10 75 Florida 24 50 sive media markets.' Districts randomly assigned to the Georgia 6 75 treatment group whose cost per point was less than $50 Georgia 7 75 were treated with 100 GRPs. Districts randomly assigned Georgia 1 1 75 to the treatment group whose cost per point was greater Maryland 8 50 than $59, but less than $80, were treated with 75 GRPs. Missouri 5 100 All other treatment districts received 50 GRPs. This Nevada 1 75 design implies that our statistical analysis must control New Jersey 2 50 for the cost per point of radio ads because interdistrict New jersey 3 50 New Jersey 4 50 variation in the volume of GRPs is random within, but New Mexico 2 50 not across, strata. (In effect, we will be analyzing three New Mexico 3 100 distinct randomized experiments, each occurring within North Carolina 2 100 populations with different advertising rates.) Total media North Carolina 4 100 expenditures to conduct the experiment in the thirty-six North Carolina 9 100 treatment districts amounted to $156,650. Pennsylvania 13 50 Radio advertisements were broadcasted from Oetober Rhode Island 1 100 31 to November 6, 2006. Details about the size of the Rhode Island 2 100 radio media buy in eaeh district included in the treatment Texas 1 100 Texas 3 50 group are presented in Table 1. Advertisements were pro Texas 7 50 fessionally recorded and produced by a partnering politi Texas 12 50 cal consulting and media firm. A media consulting firm Texas 14 50 selected the radio stations within each market, favoring Utah 3 50 stations that reached a broad Spanish-speaking audience. Virginia 1 1 50 California's 3rd district, located mainly in the Sacra mento radio market, is a typical example. A total of 75 GRPs were purchased to reach Hispanic voters in this Hispanics in the targeted congressional districts resided district. Ads were broadcast to capitalize on peak audi in the radio markets treated. In the analysis below, wc ence times during the workweek, early morning traffic note how the failure to treat all Hispanics in targeted dis (6-10 a.m.) and afternoon rush hour(3-7 p.m.) as well as tricts leads us to slightly underestimate the average effects throughout the day (10 a.m.-3 p.m.). Additional ads were of the radio treatment on those who live within the broad aired during the weekend, in this district, a total of 36 cast range of Spanish-language stations. GRPs were aired on KRCX-FM (99.9), and 23 GRPs Voters in each district were urged to vote on Election were aired on KTTA-FM (97.9), both popular stations Day, and the ads included the names, incumbency sta- with a regional Mexican format. Seventeen GRPs were Uis, and party affiliations of the major-party candidates aired on KXSE-FM (104.3), a popular station with a in each race. It should be stressed that the intervention Spanish contemporary music hits format.'' was strictly nonpartisan in nature. Using issues that Recognizing that radio market propagation zones and were believed to be of interest to Hispanic listeners, the congressional district boundaries do not overlap perfectly, radio scripts were designed to pique voters' interest in we pursued media buys in each district to maximize the contest and provide the names of the candidates, but exposure to Hispanic voters. On average, 75.4 percent of scripts made no evaluative remarks.^ For example, a Panagopoulos and Green
Spanish-language translation of the following sample attributes of the observations in the experimental groups script was used for Florida's 24th congressional district should be similarly distributed. This expectation is easily (see Appendix 2, at httpi/Zprq-sagepub-Com/supplemental/', evaluated using regression. The dependent variable is the for Spanish-language version): assigned level of GRPs, as described in Table 1; and the independent variables are the assignment strata, the lev Many people don't realize how important the els of prior tumout among Hispanics in November 2002, upcoming congressional election is. But think about Hi.spanic tumout in November 2004, the partisan affilia it. Our representative in Congress deals with the big tion of the incumbent, the proportion of registered voters gest issues confronting our country: immigration, who are Hispanic in each district, and whether there taxes, education, war—it's all part of what makes was a statewide race in November 2006. We also apply, Congress so important. as an analytic weight, the proportion of registered voters Here's where you come in: voting. If you're a who are Hispanic in each district(see below). This regres registered voter, you have an opportunity to shape sion permits an F-test of the significance of these covari- the direction of your coimtry by electing your ates, which, as expected, is insignificant at conventional member of Congress. On Tuesday, November 7th levels, F(5,198)= 1.13,/? = .35. Repeating the same exer people in Florida's 24th congressional district will cise using non-Hispanic turnout in past elections and vote to decide whether to re-elect Republican Con weighting by non-Hispanic population also produces the gressman Tom Feeney or to support his opponent expected nonsignificant result, F(5,198) = 0.76,/? = .58. Democrat Clint Curtis. Even so. there appears to be a slight tendency for districts Take part in shaping your country's futiue. Be with higher voting rates to appear in the treatment group, sure to vote on November 7th. and so the results below present specifications with and Paid for by the Institution for Social and Policy without covariates. Having confirmed that random assign Studies, a nonpartisan organization that encourages ment produced the expected degree of balance between citizens to participate in public affairs. treatment and control groups, we now estimate the effects of the experimental ad campaign on Hispanic tumout. The ingredients of this intervention were drawn from a The dependent variable in our analyses is the ehange variety of different theoretical strands in the voter in Hispanic tumout between the November 2002 and mobilization literature in an effort to maximize the November 2006 midterm elections for each congressio message's effect on voter turnout. The message reminds nal district. We partnered with professional list vendors to listeners of the candidates' names and the issues at stake, identify registered Hispanic voters using sumame evalu consistent with the theory that voters must overcome ation software. We then obtained verified voting history information costs and feel a sense of policy benefit for Hispanics in each district. The vendor aggregated its (Downs 1957). The message also provides notice of the individual-level data and provided us with district-level upcoming election,consistent witli the theoiy that registered turnout percentages for current and past elections. Focus voters are predisposed to participate and simply need a ing on Hispanics who were registered in the 206 congres timely reminder to nudge them to the polls (Dale and sional districts included in our study, on average 34.5 Strauss 2009). The message is read in Spanish to provide percent voted in the November 2006 elections. These an inclusive message to Hispanic voters and increase same voters turned out for the 2002 and 2004 elections at their sense of importance and efficacy (Morton 1991; rates of 30.0 and 55.8 percent, respectively. Putnam 2001). Finally, the fact that the listener encounters To estimate the effects of tlie radio buys(as measured in a paid ad aired on mass media may signal the importance GRPs), linear regression was applied to two nested models. of the election (Cox and Munger 1989). In sum, our The first model (equation 1) includes radio GRPs and the approach was to assess whether a maximal treatment two strata dummies as regressors to account for the fact that produced a detectable effect before launching experiments random assignment was conducted within price strata. to determine which specific elements(language, medium, issues) are responsible for this effect. HispaincTiirnoutj2QQ(,- HispanicTurnoutj2QQ2 - Py + ^^RadioGRPSj+ ^2^oderateCoslStratumj + ^yHighCostStraium- + Uj. (1) Statistical Models and Results Random assignment ensures that, in advance of the exp Equation (1) expresses the change in Hispanic tumout erimental intervention, the treatment and control groups share from 2002 to 2006 as a linear function of the have the same expected levels of voter turnout. One by treatment, covariates, and a disturbance term (Uj). Because product of random assignment is that the background the quantity ofradio GRPs is randomly assigned conditional Political Research Quarterly XX(X) on cost strata, it is statistically independent ofthe disturbance, One check on the validity of this claim is to analyze the which satisfies the key assumption necessary for unbiased behavior of non-Hispanic voters. Market research stud causal inference. ies show that very few non-Ilispanics listen to Spanish- Equation (2) is expanded to include Hispanic turnout language radio (Arbitron 2007); we therefore expect no in 2004, the party of the incumbent, whether there was a treatment effects among non-Hispanics. This robustness statewide race, and the proportion of registered voters in check involves modifying equations (1) and (2) to include each district who are Hispanics as additional covariates." the relevant current and past voting rates for non-Hispanic The latter variable was measured based on voter records voters in each district, this time weighting by the propor that appeared on the voter file as of 2006. tion of non-Hispanic voters in each district. The results of the analysis, presented in the third and fourth columns of HispanicTunwutI 2qq(, - HispanicTurnoutj 20^2 = Table 2, demonstrate that Spanish-language radio had no Po + ^^RadioGRPsj -t P2A/oc/e/ at
Hispanic voters Non-Hispanic voters independent variables Strata only Strata anai d covariates Strata only Strata and covariates Gross ratings points of radio ads 0.053 (0.019) 0.043(0.017) -0.006 (0.025) 0.004 (0.025) (0-100) Turnout in 2004 0.056 (0.027) 0.041 (0.057) Statewide race 2006 9.747 (3.223) 8.990 (2.346) Incumbent Republican -1.167 (0.986) -0.287 (1.378) Percentage registered voters. -0.170 (0.034) Hispanic Percentage registered voters, non- 0.067 (0.092) Hispanic Moderate cost per point stratum 0.300 (1.888) -0.323 (1.730) -0.741 (1.683) -1.107 (1.646) dummy High cost per point stratum dummy -1.607 (1.528) -0.183 (1.499) 0.625 (1.603) -0.493 (1.756) N 206 206 206 206 Root mean square of the residual 7.369 6.705 9.647 9.381 (RMSE) R} The dependent variable in the analyses is the change in percentage of registered Hispanic/non-Hispanic voters In each district who voted in November 2006 compared to November 2002. For the first two columns, data are weighted by the percentage of Hispanic registered voters in each congressional district. For the third and fourth columns, data are weighted by the percentage of non-Hispanic registered voters in each district.
0 50 Reslduallzed GRP Buy
90% Confidence Interval Fitted va ues Change in Turnout Among Hispanics
Figure I. Graphical representation of weighted least squares (WLS) regression results (full model for Hispanic voters, as shown in the second column of Table 2) Sizes of the circles are proportional to the size of the Hispanic registered voter population in each congressional district. The results, which illus trate the regression analysis reported in the second column of Table 2. indicate that iOO gross ratings points (GRPs) translate into 4.3 percentage points' higher turnout. template: it first targets a congressional district and then inefficient way to target Hispanic voters; a better approach deploys ads on Spanish-language radio stations. This would be to select just those districts in which Spanish- procedure is scientifically sound but an economically language radio stations provide good coverage of the Political Research Quarterly XX(X)
2-1
o>
■o
-50 100 Residualized GRP Buy
90% Confidence Interval Fitted values Change in Turnout Among Non-Hlspanics
Figure 2. Graphical representation of weighted least squares (WLS) regression results (full model for non-Hispanic surname voters, as shown in the fourth column of Table 2) Sizes of the circles are proportional to the size of the non-Hispanic surname registered voter population in each congressional district. The results, which illustrate the regression analysis reported in the fourth column of Table 2. indicate that 100 gross ratings points (GRPs) translate into 0.4 percentage points' higher turnout.
Hispanic population of the district. In three of the dis- can be mobilized cost-effectively by voter outreach cam- tricts that we randomly assigned to the treatment group, paigns such as ours. Our results underscore the important we discovered that the available Spanish-language sta- distinction between the rate at which a group typically tions reached only a fraction of home addresses of His- votes and the extent to which it can be mobilized by tar- panic voters (although some may have heard the ads gctcd appeals. anyway cn route to work, at work, or while away from The current study reflects a limited exploitation of the home for other reasons). If anything, this kind of cover- full power of radio as a medium. Budgetary constraints, age problem causes us to understate the impact of our ads, for example, restricted us to broadcasting a maximum of because the analysis above focuses solely on the causal 100 GRPs in select localities. Future work could expand effect of assignment to trealmcnl, regardless of whether the reach and frequency of radio messages. Relative ease people lived within range of the radio stations airing the in radio ad production also makes it possible to vary mes- ads. Correcting for this coverage issue, using methods sage content in future experiments. Voters in the current that assume that those living outside the broadcast areas study, for example, were exposed exclusively to nonparti- were unaffected by the ads (Angrist, Imbcns, and Rubin san get-oul-lhe-vote messages. Additional research would 1996), makes the estimates in Table 2 larger. allow us to investigate how the results may (or may not) change if the appeal were partisan in nature or involved not simply a direct encouragement to vote but rather a Discussion and Conclusions vignette that dramatizes the significance of the upcoming Although ethnic media represent an important source of election (Green and Vavreck 2008). political information for many ethnic groups (Ramirez From a theoretical standpoint, the results presented 2007), they have seldom been the subject ofexperimental here have two important implications. First, they challenge inquiry. The results of this study suggest that Spanish- the interpretive template that has been applied to field exp- languagc radio advertisements can have a profound effect erimcnls on voter mobilization. Scholars such as Gerber on Latino voter turnout. Hispanics may be dismissed by and Green (2000) have argued that personal forms of campaigns as low-propensity voters, but evidently they voter mobilization such as door-to-door canvassing are Panagopoufos and Green more effective than impersonal forms such as direct century ago remains important today: under what condi mail. Based on this line of reasoning, radio ads might be tions does mass communication spur political action? expected to be impersonal and therefore ineffective. The findings presented here present an anomaly for this Acknowledgments hypothesis and invite a closer evaluation of the relevant An earlier version of this article was presented at the 2008 differences among different forms of impersonal mobili annual meeting of the Midwest Political Science Association. zation tactics: direct mail, e-mail, recorded phone calls, The authors are grateful to the Institution for Social and Policy and messages conveyed by radio and television. One Studies at Yale University, which funded this research but bears plausible explanation is that listeners are responsive to no responsibility for the conclusions we draw. We also thank radio because they perceive it to be a more personal Bart Robbett, Kevin Bemey, and Beau Picotte for invaluable medium (Overby and Earth 2006). Second, and related, assistance. with the notable exception of messages conveyed via text messaging (Dale and Strauss 2009), reminders to Declaration of Conflicting Interest vote—whether delivered via phone calls, leaflets, or The author(s) declared no conflicts of interest with respect to e-mail—seem to have negligible effects on voter tumout the authorship and/or publication of this article. (Green and Gerber 2008). Providing information on the upcoming election in the form of mailed voter guides Funding also seems ineffective as a means of increasing turnout, The author(s) disclosed receipt of the following financial sup even when they are mailed to ethnic subgroups in trans port for the research and/or authorship ofthis article; the Institu lated form (Garcia Bedolla and Michelson 2009). Thus, tion for Social and Policy Studies at Yale University funded this the question arises as to why our advertisements, which research. combine a reminder and information, succeeded when similar content proved ineffective when communicated Notes via other channels. It is conceivable that radio, or ethnic 1. Consistent with much of the literature in the field (Garcia radio, in particular, makes these messages especially and Sanchez 2008; Barreto 2007), we use the terms "His effective. panic" and "Latino" interchangeably to describe persons The findings presented here invite several lines of fur living permanently in the United States whose ancestry can ther research. One question is the relative effectiveness be traced to any ofthe Spanish-speaking countries of North, of Spanish- versus English-language appeals. This rather Central, or South America. This classification excludes basic empirical question has received relatively little persons of(European) Spanish and Brazilian decent. attention from experimental researchers. A second issue 2. Although experimental research suggests that those with concerns the question of which listeners are most affected midrange propensities to vote are the most responsive to by radio appeals. One might suppose that low-propensity mobilization (Arceneaux and Nickerson 2009), some non- voters might benefit most from a reminder and encour experimental studies suggest low-propensity voters may agement to vote, but it is also possible that those with actually be most responsive to mobilization(Stevens 2008). higher vote propensities are most receptive to messages 3. Question wording: "For infomiation about public affairs of this kind. A related question concerns the electoral and politics, would you say you rely more heavily on context. We focused our attention on low-salience con Spanish-language television, radio, and newspapers, or on gressional races, and the question remains whether the English-language TV, radio, and newspapers?" effects would be attenuated in a higher-salience election, 4. Specifically, we excluded open seat contests and races clas where more is known about the candidates or whether, sified as "toss-ups" by the Cook Political Report on October instead, the effects remain strong in areas where Hispan- 18, 2006. ics are receiving relatively few Spanish-language campaign 5. These sample restrictions do not affect the intcmal valid advertisements. Finally, our ads, while professionally ity of our results but potentially limit extemal validity. produced, made little attempt to harness the creative We suspect that our intervention would be less effective potential of radio. It remains to be seen whether dramati in competitive districts, where competing campaign mes zations of the sort that Green and Vavreck (2008) studied sages abound. When we estimate the interaction between are more effective than simple narrative messages. A the treatment effect (sec below) and the actual vote margin long list of questions remains for testing in lab, survey, in 2006, we find the expected positive coefficient (the big and field settings, and the payoff is much larger than sim ger the victory, the bigger the radio effect) that falls well ply understanding the conditions under which radio short of statistical significance. mobilizes ethnic voters. One of the key questions that 6. There are 286 Arbitron Metro Markets in the United States set in motion the first studies of radio more than a half corresponding to cities (or urban regions) of various sizes. 10 Polidcal Research Quarterly XXfX}
Reaching the audience in each market is measured in rat uses all possible random assignments to form a sampling ings points, with 1 ratings point being equal to 1 percent distribution under the null hypothesis of no effect, produces of the number of total listeners living in the market. When more conservative p-values. Without covariates, p = .063; running an advertising campaign, summing the number of and with covariates. p = .092. These p-values fall to .020 points for all ads aired over the duration of the campaign and .047 when the number of GRPs is replaced by a dummy determines the gross ratings points (GRPs) that have been variable marking whether a district was assigned to the achieved. Theoretically, 100 GRPs means 100 percent of treatment group. the market was exposed to the ad. But as some people may hear the spot multiple times and others not at all, adver References tisers have to measure other factors: reach (the percentage Abrajano, Marisa. 2010. Campaigning to the new American of the market tliat has heard the spot one or more times) electorate: Advertising to Latino voters. Stanford, CA: Stan andfrequency (the number of times the audience has heard ford University Press. the spot). Thus, 100 GRPs can also mean 50 percent of the Angrist, Joshua D., Guido W. Imbcns, and Donald B. Rubin. audience heard the spot an average of2 times, or 25 percent 1996. Identification of causal effects using instrumental heard it 4 times, or any other combination that equals 100. variables. Journal of the American Statistical Association Advertisers also have the option to target advertising buys 91 (June): 444-55. to specific populations. In this study, we target Ilispanics. Arbitron. 2007. Hispanic radio today: How America listens The radio advertisements we purchased reflect the share to radio. Retrieved on January 8, 2010 from http://www of the overall Hispanic audiences in the respective radio .arbitron.com/downloads/hispanicradiotoday07.pdf markets and not the general audience. Thus. 100 GRPs .Arceneaux, Kevin, and David Nickerson. 2009. Who is mobi implies 100 percent of Hispanics in the radio market were lized to vote? A re-analysis of 11 field experiments. Ameri exposed to the ad once, or 50 percent were exposed twice, can Journal ofPolitical Science 53:1-16 and so forth. Because radio markets do not coincide with Barlow, William. 1998. Voice over: The making ofblack radio. congressional district boundaries, there are a few instances Philadelphia: Temple University Press. where our nominal Hispanic GRPs exceeded our actual Barreto, Matt, 2007. Si Se Puede! Latino candidates and the coverage. mobilization of Latino voters. American Political Science 7. The cost of broadcast Spanish-language radio advertise Review 101 (3): 425-42. ments targeting Hispanics in the various districts ranged Berelson, Bernard R.. Paul Lazarsfeld, and William McPhee. from $20 per point in the Tyler-Longview market in Texas 1954. Voting. Chicago: University of Chicago Press. to $300 per point in the Miami, Florida, market. The aver Cardy, Emily A. 2005. An experimental field study of the age cost per point in the sample of districts was $88. GOTV and persuasion effects of partisan direct mail and 8. Details about the ads purchased in each market are avail phone calls. Annals of the American Academy of Political able from the authors upon request. and Social Science 601:28-40. 9. The 2006 Latino National Survey found that the four most Castaneda Paredes, Mari. 2003. The transformation of Spanish- important issues for Hispanics in the 2006 election cycle language radio in the U.S. Journal ofRadio Studies 10:5-16. were (in order): the Iraq War,the economy (including jobs, Cox, Gary, and Michael Munger. 1989. Closeness, expenditures, unemployment, and, presumably, ta.\es), immigration, and and turnout in the 1982 U.S. House elections. American education. Political Science Review 83 (1): 217-31. 10. We lack access to the individual-level data that would enable Dale, Allison, and Aaron Strauss. 2009. Don't forget to vote: us to examine how subgroups of Hispanics of varying base Text message reminders as a mobilization tool. American line propensities to vote responded to oiu intervention. Journal ofPolitical Science 54 (4): 787-804. 11. The dummy variable marking statewide races is meant to de la Garza, Rodolfo, R. Brichetto, and D. Vaughan. 1983. The take into account the mobilizing efforts of statewide cam Mexican-American electorate: Information sources and pot- paigns or elevated interest in the election due to a statewide icy orientations. The Mexican-American Electorate Series. race. At the suggestion of a reader of an earlier version of Occasional Paper 2. Austin: University of Texas at Austin, this article, we included as additional covariates whether Center for Mexican American Studies, the challenger or incumbent was Hispanic. These covari de la Garza, Rodolfo, Angelo Falcon, and F. Chris Garcia. ates had no appreciable effect on our results. These results, 1996. Will the real Americans please stand up: Anglo and along with other robustness checks and a replication data set, Mexican American support for core American political val will be available at http://vote.research.yale.edu/replication ues. American Journal ofPolitical Science 40:335-51. .html. de la Garza, Rodolfo, Martha Menchaca, and Louis DeSipio, 12. As a robustness check, we analyze our experimental data eds. 1994. Barrio ballots: Latino politics in the 1990 elec using randomization inference. This procedure, which tions. Boulder, CO: Westview. Panagopoulos and Green I I
Downs, Anthony. 1957. Art economic theoiy of democracy. Michelson, Melissa. 2006. Mobilizing the Latino youth vote: New York: Harper & Row. Some experimental results. Social Science Quarterly 87(5): File, Thorn. 2008. Votirtg and registration in the election of 1 188-1206. November 2006. U.S. Census Bureau Population Report Morton, Rebecca. 1991. Groups in rational turnout models. P20-557. Washington, DC; U.S. Census Bureau. American Journal ofPolitical Science 3:758-76. Garcia, P. Chris, and Gabriel R. Sanchez. 2008. Hispanics Multicultural Marketing Equation Study. 2007. Old and new and the U.S. political system: Moving iitto the mainstream. media use. Florida State University/DMS Research Multi Upper Saddle River, NJ: Pearson Prentice Hall. cultural Research Program, http://hmc.cotnm.fsu.edu/01d Garcia Bedolla, Lisa, and Melissa R. Michelson. 2009. What do andNewMcdia091307.pdf(accessed June 8,2008). voters need to know? Testing the role of cognitive informa Oberfield, Zachary, and Adam Segal. 2008. Pluralism exam tion in Asian American voter mobilization. American Poli ined: Party television expenditwes focused on the Latino tics Research 37 (2): 254-74. vote in presidential elections. In The mass media and Latino Geer, John G., and James H. Geer. 2003. Remembering attack politics: Studies of U.S. media content, campaign strategies ads: An experimental investigation of radio. Political Behav and survey research: 1984-2004. ed. Federieo Subervi- ior 25 {\)\ 69-95. Vclcz. New York: Taylor and Francis. Gerber, Alan S., and Donald P. Green. 2000. The effects of Overby, L. Mar\'in, and Jay Barth. 2006. Radio advertising in canvassing, direct mail, and telephone contact on voter turn American political campaigns: The persistence, importance out: A field experiment. American Political Science Review and effects of narrowcasting. American Politics Research 34 94:653-63. (4): 451-78. Gcrshtenson, Joseph. 2003. Mobilization strategies of the Panagopoulos, Costas. 2006. The impact of newspaper adver Democrats and Republicans, 1956-2000. Political Research tisements on voter turnout: Field experimental evidence. Quarterly 55-.(iZl-l\A. Paper presented at the annual meeting of the American Polit Green, Donald P., and Alan S. Gerber. 2008. Get out the vote: ical Science Association, Philadelphia, September 1. How to increase voter turnout, 2nd cd. Washington, DC: Panagopoulos, Costas, and Donald Green. 2008. Field experiments Brookings Institution Press. testing tlie impact ofradio advertisements on electoral competi Green, Donald P., and Lynn Vavrcek. 2008. Analysis ofcluster- tion. American Journal ofPolitical Science 52(1): 156-68. randomized experiments: A comparison of alternative esti Panagopoulos, Costas, and Peter Wielhouwer. 2008. Polls and mation approaches. Political Analysis 16 (2): 138-52. elections: The ground war 2000-2004: Strategic targeting Gucmica, A. 1982. Reaching the Hispanic market ejfectively. in grassroots campaigns. Presidential Studies Quarterly 38 New York: McGraw-Hill. (2): 347-62. Hero, Rodney, F. Chris Garcia, John Garcia, and Harry Pachon. Putnam, Robert. 2001. Bowling alone: The collapse and revival 2000. Latino participation, partisanship and office-holding. ofAmerican community. New York: Simon & Schuster. PS: Political Science & Politics 33:529-34. Ramirez, Ricardo. 2005. Giving voice to Latino voters: A field Hermson, Paul S. 2000. Congressional elections: Campaign c.xperiment on the effectiveness of a national nonpartisan ing at home and in Washington. 3rd ed. Washington, DC: mobilization effort. Annals of the American Academy of CQ Press. Political and Social Science 601:66-84. Latino National Survey. 2006. Topline results. http://dcpts Ramirez, Ricardo. 2007. Mobilization en Espanol: Spanish- .washington.edu/uwiser/documents/LNS_toplincs_FIP language radio and the activation of political identities. _Dec6.pdf(accessed June 10, 2008). Paper presented at the Immigration Protests of 2006 Confer Leighley, Jan. 2001. Strength in numbers? The political mobili ence, University of California, Berkeley, Institute for Indus zation ofracial and ethnic minorities. Princeton, NJ; Princ trial Relations, Berkeley, April 20. eton University Press. Rosenstone, Steven J., and John M. Hansen. 1993. Mobiliza Maxwell, Scott E., and Harold D. Delancy. 2004. Designing tion. participation and democracy in America. New York: experiments and analyzing data: A model comparison per Macmillan. spective. Mahwah, NJ: Lawrence Erlbaum. Segal. Adam. 2002. Records broken: Spanish-language televi McCleneghan, J. Sean. 1987. Impact of radio ads on New Mex sion advertising in the 2002 election. Johns Hopkins Journal ico mayoral races. Journalism Quarterly 64:590-93. of American Politics online, http://www.jhu.edu/advanced/ Michelson, Melissa. 2003. Getting out the Latino vote: How govemmen t/li vp/H VPFinalReport2002.pdf. door-to-door canvassing influences voter turnout in niral Segal, Adam. 2004. Bikini politics: The 2004 presidential cam central California. Political Behavior 25 (3): 247-63. paigns' Hispanic media efforts cover only the essential parts Michelson, Melissa. 2005. Meeting the challenge of Latino of the body politic: A select group of voters in a few battle voter mobilization. Annals of the American Academy of ground states. Report issued by the Hispanic Voter Project, Political and Social Science 601:85-101. Johns Hopkins University, Baltimore. 12 Political Research Quarterly XX(X)
Stevens, Daniel. 2008. Measuring exposure to political adver Subervi-Velcz. Federico A. 2008b. Summary, conclusions tising in sur\'eys. Political Behavior 30 (1): 47-72. and recommendations. In The mass media and Latino poli Strachan, Cherie J. 2003. High-tech grass roots: The profes- tics: Studies of U.S. media content, campaign strategies sionalization of local elections. Lanham, MD: Rowman & and survey research: 1984-2004, ed. Federico Subervi- Littlefield. Velez. New York: Taylor and Francis. Subervi-Velez, Fcderico A. 1992. Republican and Democratic Suro. Roberto. 2004. Changing channels and cri.sscro.ssing cid- mass communication strategies: Targeting the Latino vote. tures: A survey ofLatinos on the news media. Pew Hispanic In From rhetoric to reality: Latino politics in the 198S Media Study. Washington. DC: Pew Hispanic Center. elections, ed. R. O. de la Garza and L. DeSipio. Boulder, Uhlaner. Carole, Bnice Cain, and D. Roderick Kiewiet. 1989. CO: Westview. Political participation ofetlmic minorities in the 1980s. Polit Suber\'i-Velez. Federico A. 2008a. Introduction. In The ma.ss ical Behavior 11:195-221. media and Latino politics: Studies of U.S. media content, Vcrba, Sidney. Kay Schlozman, and Henry Brady 1995. Voice campaign strategies and .survey research: 1984-2004, cd. and equality: Civic volunteerism in American politics. Federico Subervi-Velez. New York: Taylor and Francis. Cambridge, MA: Harvard University Press.