What Do Voters Need to Know? - CiteSeerX

4 downloads 0 Views 94KB Size Report
Sep 3, 2008 - The true bar- rier, Berinsky argued, was the .... sions and distribution of her organization's Easy Voter Guide to adult school and community .... from the Los Angeles County Registrar's list of registered voters as of April. 1, 2006.
American Politics Research OnlineFirst, published on September 3, 2008 as doi:10.1177/1532673X08320844

What Do Voters Need to Know?

American Politics Research Volume XX Number X Month XXXX xx-xx © 2008 Sage Publications 10.1177/1532673X08320844 http://apr.sagepub.com hosted at http://online.sagepub.com

Testing the Role of Cognitive Information in Asian American Voter Mobilization Lisa García Bedolla University of California, Berkeley

Melissa R. Michelson California State University, East Bay

Previous field experiments have found that indirect methods, particularly direct mail, are not effective in increasing voter turnout. Most of the mail used in these experiments provided procedural information regarding voting and a message encouraging the voter to turn out. Yet, in his review of efforts to increase voting, Berinsky (2005) concluded that it was the cognitive costs of voting—the effort needed to formulate political opinions—that constituted the greatest barrier to voter turnout. This suggests it may be important to consider the type of information being provided when examining direct mail efforts. To test this proposition, we present evidence from field experiments conducted by two Asian-American–serving community organizations during the June and November 2006 elections. We find that the type of information provided does not increase direct mail’s effectiveness. Yet unlike previous experiments, we find that phonebanking does have a statistically significant, positive effect on Asian American voter turnout. Keywords: Asian Americans; voter mobilization; voter turnout; field experiments; information; political participation

R

ecent years have seen a large and growing number of field experiments aimed at examining how best to mobilize individuals to vote, including door-to-door canvassing campaigns, live phonebank campaigns, and various types of indirect methods such as robocalls and direct mail. Research on the effectiveness of direct mail dates back to the 1920s when Harold Gosnell sent letters to residents of Chicago encouraging them to vote. His experiment found that direct mail was quite powerful increasing turnout by 1 percentage point in the presidential election of 1924 and 9 percentage 1

2

American Politics Research

points in the municipal election of 1925 (Gosnell, 1927). By contrast, more recent experiments using direct mail have found much smaller effects, perhaps due to the explosion of direct mail campaigns in the intervening 80 years (Green & Gerber, 2008). Much of the rationale behind all voter mobilization efforts, either using direct mail or other tactics, is that more individuals will participate if the costs of doing so can be sufficiently reduced (see, for example, Abramson & Aldrich, 1982; Aldrich, 1993; Rosenstone & Hansen, 1993; Rosenstone & Wolfinger, 1978). This has led to various efforts to reduce the institutional constraints on voting such as registration laws. Yet in their analysis of the effects of “motor voter” registration laws, Wolfinger and Hoffman (2001) found that changes in the “cost” of registration increased the number of individuals registered but did not have a significant effect on voter turnout. Similarly, Berinsky (2005), in his review of efforts to ease restrictions on voting, including voting-by-mail, early voting, the relaxing of absentee balloting rules, and Internet voting, found that these efforts have failed to expand the electorate. That leads Berinsky (2005) to conclude: “Political information and interest, not the high tangible costs of the act of voting, are the real barriers to a truly democratic voting public” (p. 473). The true barrier, Berinsky argued, was the cognitive cost of participation—the effort needed to formulate political opinions to be communicated at the ballot box. Mobilization campaigns often send mailings to voters, but the information provided in these mailers generally does not address these cognitive costs, instead focusing on procedural information such as polling place locations and times. The use of direct mail is appealing to campaigns because it is very inexpensive. The groups designing these mailers assume

Authors’ Note: This research was funded by a grant from The James Irvine Foundation as part of its California Votes Initiative, a multiyear effort to increase voting rates among infrequent voters—particularly those in low-income and ethnic communities—in California’s San Joaquin Valley and targeted areas in Southern California. The Initiative also aims to discern effective approaches by which to increase voter turnout and share those lessons with the civic engagement field. For more information about the Initiative, see http://www.irvine.org/ evaluation/program/cvi.shtml. The James Irvine Foundation bears no responsibility for the content of this report. The authors also thank the Asian Pacific American Legal Center of Southern California (APALC) and the Orange County Asian Pacific Islander Community Alliance, Inc. (OCAPICA) for their cooperation and for allowing us to evaluate their mobilization efforts. The authors are also indebted to Donald Green and to the anonymous reviewers for their helpful comments. Correspondence concerning this article should be addressed to Lisa García Bedolla, University of California, Berkeley, 5639 Tolman Hall, Berkeley, CA 94720-1670, [email protected].

García Bedolla, Michelson / What Do Voters Need to Know?

3

that individuals do not need assistance in developing their political opinions but rather procedural information meant to facilitate the voting process itself. A direct test of the power of procedural information is provided by Wolfinger, Highton, and Mullin (2005), who examined the impact of mailed polling place information and sample ballots on participation levels across states. They found turnout was 2.5 percentage points higher in states that mailed polling place information to registered voters and 2 percentage points higher in states that mailed sample ballots. The effects were even stronger among younger citizens and those with less education. Wolfinger et al. noted (2005): [P]roviding information matters more for people who are less likely to acquire it elsewhere. Thus, receiving sample ballots in the mail is most consequential for people with less access to information, the least-educated and young registrants, while its effect shrinks to insignificance for college graduates. In addition, these best practices are less valuable to young adults still living with their parents, who can acquire this information more easily from older, more experienced people. (p. 17)

However, as mentioned above, previous studies using field experiments have found mailings to be ineffective in increasing turnout even when they include a voter’s polling place information (Gerber & Green, 2000, 2001; Green & Gerber, 2008; Green, Gerber, & Nickerson, 2003). One possible explanation for this difference is simply methodological. Wolfinger et al. conducted a quasi-experiment, whereas the other studies employed randomized field experiments. Another possibility is that these studies did not take into sufficient consideration the type of mailer being used, that is, with procedural information, cognitive information, or both, or for the type of voter receiving the information. In today’s campaign climate, low-propensity voters are less likely than high-propensity voters to receive any electoral information. It is possible, within that context, that a single mailer could have more of an effect with low-propensity voters than with high-propensity voters. Or, if Berinsky’s conclusion is correct, it is also possible that a mailing providing information designed to reduce the cognitive costs of voting, rather than procedural information, could have a more positive impact on voter turnout. Field research by Elizabeth Addonizio and Susan Clark provides support for the idea that reducing voters’ cognitive information costs will increase their participation. They argue that the true cause of low voter turnout is anxiety about the process and a lack of information about what

4

American Politics Research

voters are being asked to decide. Mobilization activities must therefore include not only encouragements to vote but also the cognitive tools to do so, including discussions about issues and candidates. In an experiment spanning elections from November 2002 through the spring of 2004, Addonizio found that providing high school students with detailed information about how to vote and linking the election to issues relevant to young people (e.g., financial aid for college, the military draft, sales taxes) increased participation by 19 to 24 percentage points. Addonizio (2006) concluded: 18-year olds, who are mobilized to vote in a way that incorporates interpersonal, communal, purposive, convivial, and instrumental experiences, turn out to vote at a statistically significantly higher level than those who are not afforded this mobilization. Participating in the First-time Voter Program increases the probability that an 18-year-old will vote by 19 to 24 percentage points. (p. 39)

Similarly, in 1996 Clark combined voting workshops with issue discussions and distribution of her organization’s Easy Voter Guide to adult school and community college students aged 18 to 24 years. These efforts increased participants’ turnout rates from an expected 35% to 36% to more than 70%, although we should note that the sample sizes were too small to allow for rigorous statistical analysis (Clark, Wold, & Mayeri, 1997). Clark sees Easy Voter Guides as the answer to the relatively inaccessible official voter guides mailed by the California Secretary of State’s office to every registered voter. In contrast to those official guides, which are often long and written in obscure language, the Easy Voter Guides present shorter and simpler discussions of the major candidates and measures on the ballot (Clark et al., 1997): While originally developed for Californians with limited literacy skills, the Easy Reading Voter Guide has become popular with better-educated people who are too busy to study the official 100+ page Voter Information Guide supplied by the Secretary of State. Key aspects of its appeal: the “layman’s language,” the easy-to-skim layout with large type, color and photos, basic definitions and background information.

Although the small sample sizes and lack of rigorous statistical analysis limit the strength of the conclusions reached by Clark and her colleagues, we believe that the theories underlying that project, combined with the statistically significant results from Addonizio’s work, provide support for a

García Bedolla, Michelson / What Do Voters Need to Know?

5

hypothesis that providing voters with information that helps them to overcome cognitive barriers to voting can have a positive effect on turnout. The remaining question is whether it was the interactive discussions featured in both of these previous experiments that was crucial to helping move voters to the polls or whether simply providing more information to help voters make political decisions would be sufficient. To test this hypothesis, we conducted an experiment using a direct mailing that was specifically designed to address voters’ lack of cognitive information, the Easy Voter Guide. Easy Voter Guides are used by a wide variety of groups in California seeking to help educate registered voters about upcoming elections.1 They are available free of charge in five languages and can also be printed directly off of the Internet. Many groups use the guides as a form of direct mail. They include user-friendly information about ballot items, including pro and con arguments for the ballot propositions, and some information about the candidates for office. The guides are designed to provide information that is more accessible than what is normally provided to voters in the official sample ballot and that is designed to help voters decide which candidates and ballot measures to support or oppose. Thus the Easy Voter Guide provides the type of cognitive information that we hypothesize may be a more effective form of direct mailing compared with mailings that provide strictly procedural information. Given Wolfinger et al.’s finding that information was especially effective with low-propensity voters, we test our hypothesis on a low-propensity voting group in southern California: Asian-American voters.2 During the weeks prior to the November 2006 election, we worked with two Asian-American–serving organizations in Southern California who used Easy Voter Guides as part of a voter mobilization campaign aimed at low-propensity Asian-American voters. We also include for comparison a campaign conducted by one of these same organizations for the June 2006 election that used a more standard piece of direct mail. Scholars have noted with concern that although Asian Americans constitute a large and growing segment of the population, they are generally excluded from get-out-thevote (GOTV) campaigns, not only because of their low rates of participation, but also because such efforts would entail organizing a time-consuming and complex multiethnic, multilingual campaign (Barnes & Bennett, 2002; Jamieson, Shin, & Day, 2002; Wong, 2005, 2006). In short, compared to efforts in English, and perhaps Spanish, to target other potential voters, the expected payoff is simply too small in the face of the high cost of organizing a campaign that recognizes the many language backgrounds of Asian Americans living in the United States.

6

American Politics Research

Prior to the efforts described in this article, only three field experiments aimed at increasing Asian-American turnout had been conducted, some of which suggest the possibility that direct mail is effective with this population. For the November 2002 elections, one effort used live telephone calls and postcards to mobilize East Asian and South Asian registered voters living in high-density Asian-American neighborhoods of Los Angeles County. Although those contacted by phone or receiving a postcard were slightly more likely to vote, the differences were not statistically significant (Wong, 2005). Another field experiment aimed at South Asian voters (Hindu and Sikh Indian Americans) prior to the November 2004 elections used English-language postcards with varying messages designed to make ethnic or civic identities salient (Trivedi, 2005). Again the difference in turnout between those in the treatment group and those in the control group was not statistically significant. Gimpel, Cho, and Shaw (2005) used partisan direct mail (working in cooperation with the Republican incumbent) to try to increase Asian-American turnout in a high-profile state legislative election in Texas in 2004. They found that mailers did increase participation, particularly when focused on neighborhood interests, although the statistical power of the effect was limited due to a small control group. This work showed important national-origin differences among the different Asian-American groups that were studied. It also suggests there may be differences in the effectiveness of mobilization strategies for Asian Americans versus what has been found for non-Latino White voters (Green & Gerber, 2008). This is consistent with findings from other studies looking at racial or ethnic politics in the United States that have shown that the participation and mobilization models developed to explain turnout among non-Latino Whites do not always explain fully what drives turnout among members of different racial minority groups (Fraga et al., 2006; García Bedolla, 2005; Leighley, 2001; Lien, Conway, & Wong, 2004; Michelson, 2005; Tate, 1993). Thus, it is important to test findings across different groups before assuming their generalizability. Asian Americans constitute a growing voting bloc within the state of California— yet little is known about their voting behavior beyond the fact that they often vary by national origin and have a low propensity to turn out. Our study contributes to the knowledge base for this understudied yet important group within American politics. Our hypothesis, based on this past research, is that Asian-American voters will be more influenced by the provision of information about the election, particularly when that information is cognitively focused and made even more accessible due to its provision in their native language.

García Bedolla, Michelson / What Do Voters Need to Know?

7

The experiments described in this article were conducted as part of The James Irvine Foundation’s California Votes Initiative, a multiyear project designed to increase participation among low-propensity voters in Central and Southern California. The efforts were multilingual, based on surname lists of Asian national-origin groups, and used both direct mail and live phonebanks. However, these experiments differed from Wong’s in two important ways. Wong’s experiment focused on all Asian-American registered voters living in high-Asian density neighborhoods in Los Angeles County. Our experiments were instead limited to low-propensity Asian American voters—defined for the June 2006 election as individuals who voted in two or fewer of the last four statewide elections, were younger than age 25, or were newly registered. In Los Angeles County, this included South Asians, Cambodians, Chinese, Filipinos, Japanese, Koreans, and Vietnamese; in Orange County, the effort was limited to Chinese, Koreans, and Vietnamese. Second, whereas one group’s effort conducted in Los Angeles for the June 2006 election included a relatively simple piece of direct mail, both organizations used English and in-language versions of the informationrich Easy Voter Guide for their efforts for the November 2006 election. All of these GOTV efforts used the increasingly popular method of randomized field experiments to isolate the treatment effect (Green & Gerber, 2008). Once pools of targeted registered voters were determined by each community organization, individuals from that list were randomly divided into treatment and control groups through the use of a computerized random number generator. The mobilization efforts then targeted all individuals in the treatment groups, although for the telephone efforts not all individuals in the treatment group could be successfully contacted. After each election, turnout rates (using validated voter information from the relevant county registrars’ offices) for the treatment and control groups were compared. The different voting rates were compared, taking into account both contact rates (the number of individuals in the treatment groups successfully contacted) and prior voting history for each individual. Random assignment ensures that participants in the treatment and control groups have the same demographic makeup and the same underlying propensity to vote, making unnecessary the inclusion of other independent variables in the postelection analyses. For each campaign described here, we examine intent-to-treat effects, the differences in turnout between the treatment groups and the control groups, as well as treatment-on-treated effects, the effect of the treatment on those who actually received it. This distinction arises due to the failure-to-treat problem, whereby some people assigned to the treatment group are not successfully contacted. Because the failure to contact is not random (some

8

American Politics Research

individuals are simply easier to contact and therefore possibly easier to turn out), it must be controlled for. Gerber and Green (2000) describe how to correct for this problem using 2SLS analysis with “contact” as an explanatory variable and “assignment to the treatment group” as an instrumental variable; consistent with other subsequent research using field experiments to test GOTV campaign effectiveness, we adopt this approach in the analysis below.

The Asian Pacific American Legal Center of Southern California (APALC) APALC was founded in 1983 and has long worked in Southern California to advance Asian Pacific American civil rights, provide legal services and education, and work for a more equitable society. Although they are the nation’s largest organization to focus on the legal needs of the Asian Pacific Islander community, they had not conducted a voter mobilization campaign prior to the June 2006 election. APALC ran a campaign consisting of phone calls and direct mail aimed at a variety of Asian-origin groups. APALC’s phone banking campaign for the June 2006 primary in California was conducted from May 18 to June 5, from 5 p.m. to 9 p.m. on weekdays and 1 p.m. to 5 p.m. on weekends. Bilingual interviewers were used to allow for in-language mobilization of the included national-origin groups (South Asians, Chinese, Filipino, Japanese, Korean, and Vietnamese).3 The pool of registered voters included in the June experiment was culled from the Los Angeles County Registrar’s list of registered voters as of April 1, 2006. Asian individuals were identified based on place of birth and their full name. APALC used internal ethnic surname lists to determine the particular Asian ethnic background of each individual.4 The file was further culled to include only low-propensity voters, as defined above. The list of names was then narrowed to only include individuals residing in geographic areas with large numbers of Asian voters. Finally, the list of about 110,000 individuals was cleaned by a commercial vendor to include only those with a valid mailing address and phone number, resulting in a pool of 43,875 registered voters that was then randomly divided into treatment and control groups. Prior to the June 2006 primary election, APALC called approximately 9,000 registered voters and sent bilingual mailers to about 11,000 registered voters. Groups of registered voters on their list were selected randomly to receive only a phone call, only a mailer, or both. The phone calls were

García Bedolla, Michelson / What Do Voters Need to Know?

9

Table 1 Intent-to-Treat Effects and Contact Rates for Asian Pacific American Legal Center of Southern California Campaign, June 2006

South Asian (N = 1,520) Chinese (N = 18,061) Filipino (N = 5,328) Japanese (N = 2,774) Korean (N = 8,930) Vietnamese (N = 5,660)

Phone Call Contact Rate (%)

Control Group Turnout Rate (%)

Phone Call Treatment Group Turnout Rate (%)

Mailer Treatment Group Turnout Rate (%)

Phone Call and Mailer Treatment Group Turnout Rate (%)

29.8

6.3 (1,101) 8.0 (12,076) 9.2 (3,298) 12.4 (2,162) 7.6 (5,386) 8.2 (3,829)

7.4 (419) 8.4 (956) 10.5 (631) 13.6 (612) 8.1 (1,113) 7.0 (213)

n/a

n/a

8.2 (3,408) 8.3 (576) n/a

8.8 (1,621) 9.5 (823) n/a

7.9 (994) 7.8 (1,294)

8.2 (1,437) 6.2 (324)

19.3 20.2 33.8 27.6 13.6

Note: Number of observations in parentheses.

conducted in cooperation with eight other local Asian Pacific Islander organizations. The same GOTV message was used for all ethnic groups: “voting empowers our communities and is easy.” Mailers also included basic information about the races and measures on the ballot, and procedural information, including the voter’s polling place and the right of decline-tostate voters to request partisan ballots. Translation was provided by the relevant partner organizations, and the photos were changed for each mailer to be appropriate to each national-origin group. Contact rates for the phonebank ranged from 13.6% to 33.8% (see Table 1). As Table 1 indicates, APALC’s mailers had weak effects on voter turnout, but its phone banking efforts were more successful. With the exception of Vietnamese Americans, turnout was higher in the treatment groups than in the control groups. Table 2 reports the corresponding treatment-on-treated effects. Pooling all national-origin groups, 2SLS analysis reveals that turnout among those who received a phone call was 2.5 percentage points higher than the control group, which is slightly stronger than the results achieved by Wong (2005). Given the population’s base voting rate of 8.4%, this is actually a substantial effect. To raise a group’s turnout from 8.4% to 10.9% represents a 30% relative gain in votes.

10

American Politics Research

Table 2 Treatment-on-Treated Effects for Asian Pacific American Legal Center of Southern California Campaign, June 2006

N All

43,397

South Asians

1,520

Chinese

18,061

Filipino

5,328

Japanese

2,774

Korean

8,930

Vietnamese

5,660

Pooled

42,273

2SLS Phone No Covariates

2SLS Mail No Covariates

2SLS Phone With Covariates

2SLS Mail With Covariates

2.6 (1.6) 3.8 (4.9) 2.7 (3.2) 5.9 (4.8) 3.4 (4.6) 1.3 (2.5) –10.7 (8.9) 2.3* (1.6)

–0.05 (0.3) n/a

2.9* (1.6) 4.8 (4.9) 2.8 (3.2) 6.9 (4.8) 3.1 (4.5) 1.5 (2.4) –10.5 (8.7) 2.5* (1.5)

0.01 (0.3) n/a

0.3 (0.5) –0.9 (1.0) n/a 0.2 (0.7) –0.4 (0.8) 0.01 (0.3)

0.4 (0.5) –0.8 (1.0) n/a 0.2 (0.7) –0.3 (0.8) 0.09 (0.3)

Note: Standard errors in parentheses. Estimates obtained by regressing vote on phone contact and mail, using assignment to phone calls and mail as instrumental variables. Covariates are voter history for four previous statewide elections. *p < .10. Estimates for “All” include individuals for whom national-origin information was not available.

During the November 2006 campaign, APALC made contact attempts to approximately 18,700 voters, including call attempts to 12,000 voters and mailers to 10,900 voters. Phone bankers made bilingual calls in nine languages and dialects, including English, Khmer, Mandarin, Cantonese, Hindi, Tagalog, Japanese, Korean, and Vietnamese. In-language mailers were sent to targeted Chinese, Korean, and Vietnamese American voters. The GOTV message used in those contacts was developed using the results of a series of seven focus groups conducted by APALC and their partner organizations in the months prior to the campaign. The target population for the November campaign consisted of registered voters (a) whose surnames implied Asian ancestry, (b) who were deemed to be low-propensity voters (those who voted in fewer than three of the last five statewide elections, were younger than 25 years of age, or had recently registered), and (c) who resided in geographic areas with large

García Bedolla, Michelson / What Do Voters Need to Know?

11

numbers of Asian voters. The resulting subset of the voter registration list of 132,563 voters was sent to a commercial vendor to identify individuals with valid mailing address and phone number information. This procedure yielded 33,457 voters for assignment to treatment and control groups. The phonebank scripts were nonpartisan in content.5 Callers also informed interested voters about their polling location and gave them other basic Election Day information, such as polling place hours and the availability of translated materials. Scripts included ethnic-specific hotline numbers that voters could call for assistance. In addition, the scripts listed common reasons that voters may give for not voting (e.g., voting is inconvenient) followed by responses that callers could provide to address what the voter said. The phone bank ran almost daily from October 18 to November 6 at APALC’s office. The hours of phone banking were 5 p.m. to 9 p.m. on weekdays and 1 p.m. to 5 p.m. on weekends. As they had done in June, in November 2006 APALC also conducted an experiment using mailings sent to targeted voters. In contrast to the June campaign, and to test our hypotheses regarding the effects cognitive information provision may have on low-propensity voter mobilization, APALC’s November mailings consisted of Easy Voter Guides that APALC sent either in-language (for Chinese, Korean, and Vietnamese American voters) or in English. To determine which language materials to send to a voter, APALC used age and place of birth information to make a rough estimate of voters’ English proficiency, and sent English versions of the guide to voters who were either (a) U.S.-born or (b) foreign-born and 35 years or younger, and sent translated versions to Chinese, Korean, or Vietnamese voters who were foreign-born and older than 35 years old.6 APALC sent English versions of the guides to Cambodian, Filipino, Japanese, and South Asian voters. APALC and its partner organizations prepared bilingual labels that were attached to the front cover of the guides. The label identified the organizations participating in the project to let voters know who was sending them the guides, listed a phone number to call if the voters had questions, and provided a link to the Easy Voter Guide web site. The guides were mailed between October 16 and October 23, arriving between October 20 and 31, within two weeks of Election Day. In preparing to carry out the campaign, we randomly assigned 33,457 low-propensity Asian-American voters residing in targeted geographic areas with valid mailing address and phone number information to one of four groups: control, phone call only, mailer only, or both a phone call and a mailer. Contact rates for the November phonebank were higher than those from June, ranging from 26.9% to 39.5% (see Table 3).

12

American Politics Research

Table 3 Intent-to-Treat Effects and Contact Rates for Asian Pacific American Legal Center of Southern California Campaign, November 2006

South Asian (N = 1,588) Cambodian (N = 581) Chinese (N = 14,497) Filipino (N = 4,186) Japanese (N = 2,554) Korean (N = 5,984) Vietnamese (N = 3,894)

Phone Call Contact Rate (%)

Control Group Turnout Rate (%)

Phone Call Treatment Group Turnout Rate (%)

Mailer Treatment Group Turnout Rate (%)

Phone Call and Mailer Treatment Group Turnout Rate (%)

35.7

29.9 (271) 12.5 (40) 25.7 (7,618) 32.9 (1,520) 37.1 (1,138) 28.5 (1,999) 23.1 (1,947)

33.0 (542) 17.4 (242) 26.7 (2,889) 34.6 (1,457) 35.4 (804) 29.3 (1,521) 25.7 (342)

40.2 (276) 5.0 (40) 27.8 (2,741) 33.2 (596) 33.6 (387) 28.0 (1,338) 24.8 (1,349)

34.2 (486) 21.4 (257) 27.3 (1,137) 34.9 (585) 39.6 (316) 30.8 (1,069) 22.2 (225)

26.9 36.3 29.9 34.4 39.1 39.5

Note: Number of observations in parentheses.

Table 3 shows the breakdown of intent-to-treat results for each experimental group and target population, comparing turnout in the treatment and control groups. By and large, the effects are positive. Table 4 reports the corresponding treatment-on-treated effects. Pooling all national-origin groups, 2SLS analysis reveals that turnout among those who received a phone call was 3.7 percentage points higher than the control group, an improvement from the June experiment and the strongest effect found to date for a phonebank effort targeting Asian-American voters. However, despite the fact that the Easy Voter Guides provided qualitatively different information than the June mailing, they fail to have a discernable impact on voter turnout.7 Among Asian-American low-propensity voters in Los Angeles County, these experiments suggest that live phone calls do make a significant difference in terms of mobilizing Asian American voters. Contrary to our expectations, the Easy Voter Guides, despite their cognitive content and provision in-language, were not more effective in turning out voters than other forms of direct mail. The question then becomes whether the effects would be similar among Asian-American voters living in a different geographic context.

García Bedolla, Michelson / What Do Voters Need to Know?

13

Table 4 Treatment-on-Treated Effects for Asian Pacific American Legal Center of Southern California Campaign, November 2006

N All

33,204

South Asian

1,575

Cambodian

579

Chinese

14,385

Filipino

4,158

Japanese

2,545

Korean

5,927

Vietnamese

3,863

Pooled

33,059

2SLS Phone No Covariates

2SLS Mail No Covariates

2SLS Phone With Covariates

2SLS Mail With Covariates

2.6 (1.6) –3.9 (8.2) 39.6** (14.9) 1.4 (2.3) 5.7 (5.2) 1.4 (5.7) 4.1 (3.2) 1.5 (4.9) 2.7** (1.6)

1.2** (0.5) 4.2* (2.5) 3.8 (3.5) 1.6* (0.8) 0.4 (1.6) –0.1 (2.1) 0.5 (1.2) 0.9 (1.4) 1.2** (0.5)

3.7** (1.5) 1.2 (7.3) 43.6** (14.3) 2.6 (2.1) 7.3 (4.9) 1.9 (5.3) 4.2 (3.0) 2.6 (4.4) 3.7** (1.4)

0.6 (0.5) 4.7** (2.3) 4.3 (3.3) 0.8 (0.8) –0.05 (1.5) –1.2 (2.0) –0.02 (1.1) 0.7 (1.3) 0.7* (0.5)

Note: Standard errors in parentheses. Estimates obtained by regressing vote on phone contact and mail, using assignment to phone calls and mail as instrumental variables. Covariates are voter history for five previous statewide elections. *p < .10. **p < .05. Estimates for “All” include individuals for whom national-origin information was not available.

The Orange County Asian Pacific Islander Community Alliance, Inc. (OCAPICA) Wong’s (2005) study found important differences among Asian Americans not only by national origin but also in terms of the geographic context of the turnout effort. Those Asian-American voters residing in high-density Asian-American neighborhoods were more likely to be mobilized when contacted. In order to test the effects of geographic context and national origin, we also conducted a field experiment in Orange County, California, with the cooperation of OCAPICA. The Asian-American community in Orange County varies from that of Los Angeles County in that it is of smaller size and the national origin distribution is somewhat different. The

14

American Politics Research

most high-density Asian-American areas in Los Angeles County tend to be of Chinese or Korean origin. In Orange County, the most high density area is majority Vietnamese. Thus a comparison of the effectiveness of mobilization campaigns in the two areas allows us to explore the effects of both geographic context and national origin on Asian-American voter mobilization efforts. OCAPICA has worked for over a decade to improve opportunities and outcomes for low-income Asian and Pacific Islander Americans in Orange County, including programs covering youth development, education, community health, and economic development. Although OCAPICA has long been active in political issues of concern to the Asian-American community, they directly mobilized voters for the first time in the fall of 2006.8 In the weeks prior to the November 2006 election, OCAPICA conducted a phone-based voter mobilization campaign designed to encourage participation by members of the Asian-American community living in Orange County, California. Canvassers included 11 Vietnamese, four Chinese (Mandarin speakers), and five Koreans, all of whom were fully bilingual. One of the Korean volunteers was a prominent community leader whose voice was often recognized by contacted voters. All of the Vietnamese volunteers were older than age 40. Younger volunteers were generally assigned to call voters younger than age 35. The target communities were registered voters of Chinese, Korean, or Vietnamese ancestry. The outreach campaign consisted of mailed Easy Voter Guides and/or phone calls from the group’s local phone bank. Each ancestry group was divided according to whether a given household was likely to be predominantly English-speaking or not, using the same criteria as that used by APALC (described above). OCAPICA’s callers encouraged individuals to take an interest in the issues surrounding the November election, provided polling locations and times, and, as with APALC, attempted to reply to common reasons given for not voting. Contact rates for the phone-calling campaign varied by national origin and by language within each national origin group, ranging from 19.8% to 54.9%. Contact rates among all groups tended to be lower among English speakers (see Table 5). Table 5 shows the different intent-to-treat effects, comparing turnout among treatment and control groups. Table 6 displays the corresponding treatment-on-treated effects, pooling across all groups to account for different contact rates. This generates a treatment-on-treated effect of phone contact of 2.9 percentage points (SE = 2.5). Adding controls for voting in the previous five elections increases the estimate to 4.2 percentage points and decreases the standard error to 2.3 percentage points. Again, similar to

García Bedolla, Michelson / What Do Voters Need to Know?

15

Table 5 Intent-to-Treat and Contact Rates for the Orange County Asian Pacific Islander Community Alliance, Inc. Campaign, November 2006

Ethnicity

Language

Phone Call Contact Rate (%)

Chinese

Chinese

40.1

Chinese

English

19.8

Korean

Korean

54.9

Korean

English

38.0

Vietnamese

Vietnamese

40.8

Vietnamese

English

30.4

Control Group Turnout Rate (%)

Phone Call Treatment Group Turnout Rate (%)

Mailer Treatment Group Turnout Rate (%)

Phone Call and Mailer Treatment Turnout Rate (%)

n/a

35.0 (1,385) 21.3 (314) 35.4 (644) 14.1 (206) 36.4 (1,443) 20.9 (345)

32.9 (489) 26.5 (324) 33.7 (1,145) 14.8 (522) 38.5 (1,445) 26.3 (438)

32.7 (1,809) 26.5 (1,879) 32.3 (1,501) 17.7 (864) 36.1 (2,560) 24.5 (4,927)

n/a n/a 15.2 (33) n/a 29.8 (312)

what we saw with APALC, this constitutes a significant effect, the most significant found to date among Asian-American voters. This finding is especially significant given these are low-propensity Asian-American voters and given the low base rate of voting for this group of voters. Mail’s effects are weakly negative and statistically insignificant. As we saw with APALC, the Easy Voter Guides generated results similar to that of other direct mail efforts, whereas phone calls generated treatment-on-treated results close to or exceeding the 3 percentage-point level that is typical of this type of intervention. Similar to findings from experiments looking at the provision of procedural information to voters, the mailings were not effective in increasing turnout. Instead, as has been found in other experimental work, a live conversation was most effective in turning out these low-propensity Asian-American voters.

Conclusion Political scientists long have examined the role of information in raising or reducing the costs of voting. The results have been mixed, leading Berinsky

16

American Politics Research

Table 6 Treatment-on-Treated Effects for the Orange County Asian Pacific Islander Community Alliance, Inc. Campaign, November 2006

Ethnicity

Language of Contact

N

Chinese

Chinese

3,683

Chinese

English

2,517

Korean

Korean

3,290

Korean

English

1,625

Vietnamese

Vietnamese

9,971

Vietnamese

English

6.022

All groups pooled

27,108

2SLS 2SLS 2SLS 2SLS Phone No Mail No Phone With Mail With Covariates Covariates Covariates Covariates –5.2 (6.4) 26.4 (17.3) –3.1 (4.6) 0.3 (7.4) 5.1 (4.5) 18.0** (6.7) 2.9 (2.5)

2.3 (1.7) –5.2** (2.6) 3.1 (2.3) –3.1 (2.7) 0.2 (1.4) –3.8** (1.9) –0.6 (0.9)

0.2 (6.0) 22.9 (16.4) 1.3 (4.2) 2.8 (7.0) 3.8 (4.2) 13.7** (6.1) 4.2** (2.3)

1.1 (1.6) –4.1* (2.5) 0.7 (2.1) –3.1 (2.6) 0.3 (1.3) –3.1* (1.7) –0.8 (0.7)

Note: Standard errors in parentheses. Estimates obtained by regressing vote on phone contact and mail, using assignment to phone calls and mail as instrumental variables. Covariates are voter history for five previous statewide elections. *p < .10. **p < .05.

(2005) to emphasize the analytical distinction between procedural and cognitive information in the voting process. To exercise their franchise, voters need to know not only how to vote, but also they must develop opinions about what to vote for. It is intuitively logical that voters with fewer resources will be less likely to have either type of information readily available, but that overcoming the lack of cognitive information threshold may be especially important (Verba, Schlozman, & Brady, 1995). Our study tested this hypothesis by using a mailing directed at lowering the cognitive costs to voting for low-propensity voters. The Easy Voter Guide was designed specifically to address this issue, and is therefore a good instrument for this test; although it was not originally intended for use as direct mail, it is widely used as such by various organizations. The guides’ failure to have a discernible effect on turnout raises two possible conclusions: (a) cognitive information is important, but needs to be received in another format (i.e., in person) to be effective; or (b) there are other psychological factors low-propensity voters must overcome to be encouraged to vote and

García Bedolla, Michelson / What Do Voters Need to Know?

17

levels of information, procedural or cognitive, form only a small part of the barriers they experience with regard to voting. We will discuss each in turn. As discussed above, the study by Clark using the Easy Voter Guide also included face-to-face discussions of the content of the guides. Addonizio’s work, even though it did not use the guides, also included significant interpersonal interaction and discussion of political issues. This, combined with our findings, suggests that sending voters cognitively-focused information, even if of high quality and provided in-language, is simply not sufficient to overcome the discomfort low-propensity voters may feel vis-à-vis the political process. Thus, Berinsky (2005) may be correct in the sense that cognitive information is key. But, our work indicates that voters should be given that information in a format other than a mailing, such as within a context of political discussions, in order for it to be effective. Another possible lesson to take from this study, combining our findings with those from other mobilization campaigns, is that perhaps a lack of information is not the most important barrier to participation. Both APALC and OCAPICA, through their live phonebanks, were able to have statistically significant effects on voter turnout. As is true with most successful phonebanks, these were not long phone calls. Most phonebank scripts simply entail identifying the voter, letting them know about the election, and encouraging them to make their voice heard. Door-to-door canvassing efforts, which have been found to be the most effective tactic for turning people out, also do not often entail a deep and in-depth conversation with voters; in most cases, a canvasser will spend no more than five minutes at the door with the voter. This suggests that it is not necessarily the information conveyed during these contacts that is most important, but rather the invitation and reminder to participate. Most political science scholarship looking at political participation has focused on the role that resources, like information, play in voter turnout. Those with more resources, socioeconomic, organizational, or occupational, consistently have been found to be more likely to vote (Verba et al., 1995; Verba & Nie, 1972; Wolfinger & Rosenstone, 1980). Yet, the growing experimental findings on voter mobilization call into question some of the core assumptions underlying this literature. If, after the mobilization conversation, the voter still has the same low level of resources, why does that relatively short contact work to change their behavior? Our findings and those of other experimental scholars suggest that it is that personal invitation to participate that matters, rather than necessarily the information provided in that conversation, or the resource level of the particular voter.9 Scholars like Rosenstone and Hansen (1993) and Leighley (2001)

18

American Politics Research

emphasize the importance of mobilization in general, yet we do not have a well-defined theoretical understanding of why an in-person invitation would be so effective, or why it could counteract the negative effect of low voter resources. One possible explanation is that this kind of invitation can make the voter feel that the political process is more inclusive, or it can temporarily increase his or her feelings of personal efficacy. Perhaps being told that your vote matters, in person, is powerful because it happens so rarely within American politics. Untangling the factors underlying these effects needs to be the focus of future research. On the practical side, in every election cycle American political campaigns and interest groups spend millions of dollars on direct mail. Although it is possible that direct mail can influence candidate choice, many of these organizations also send mailings in the hope of turning their constituency out to vote. Direct mail is an appealing option for those candidates and groups because it allows them to contact large numbers of voters without expending the time and resources necessary to implement labor-intensive voter mobilization strategies such as door-to-door canvassing or live phonebanking. Until recently, however, there have been few rigorous tests as to whether or not the expenditure of those resources is achieving the expected goal. The few tests that had been done indicated that direct mail was not effective in increasing turnout (Green & Gerber, 2008). Yet, none of these studies had looked at low-propensity Asian-American voters, specifically, making it unclear as to whether or not these findings would apply in other ethnic communities. Our findings are consistent with previous research demonstrating that personal contact (either through door-to-door canvassing or conversational phonebanks) is much more effective at increasing turnout than are indirect methods such as direct mail or robocalls. In fact, our results provide the first field experiment evidence that phonebanks targeting Asian Americans can have a statistically significant impact on turnout, despite the challenges posed by a multiethnic, multilingual campaign. Groups interested in increasing turnout should focus their resources on providing potential voters with these kinds of personal contacts, or perhaps should use the Easy Voter Guides as an informational tool within a collective discussion as was done successfully by Clark. It seems that what deters individuals from voting is not only a lack of understanding of the candidates or issues, but the lack of a feeling of personal connection with the process (García Bedolla, 2005). Our findings, and those of previous work, suggest that no amount of direct mail, however informative, easy to understand or user-friendly, will be able to substitute for that personal touch.

García Bedolla, Michelson / What Do Voters Need to Know?

19

Notes 1. Information about the Easy Voter Project and copies of the guides themselves may be found at: http://www.easyvoter.org/. 2. We should note that this analysis is not meant to test whether or not the guides serve the purpose of educating voters and ensuring that those voters who do go to the polls are better informed about the process; it is entirely possible that the guides accomplish that goal quite successfully. Our purpose here is to look only at their effectiveness as a tool to increase voter turnout. 3. Although Asian Pacific American Legal Center of Southern California (APALC) also worked to mobilize Cambodians, the number of registered voters in this group was too small to allow for their inclusion in a field experiment in June 2006. In addition to the phonebanking campaign, APALC held a press conference prior to each election (on May 3, 2006 and October 4, 2006), to explain and publicize the get-out-the-vote (GOTV) campaign. These press conferences were well attended, particularly by the ethnic media. Prior to the November 2006 election, they also held seven focus groups to help refine their GOTV messages and strategies. 4. One of the important findings from the Wong experiment was that it showed the feasibility of multilanguage efforts, demonstrating that surname lists could be used successfully and cost-effectively to sort Asian registered voters into national-origin subgroups for languagespecific targeting. Wong also found that about one fourth of contacted voters preferred to speak a language other than English, emphasizing the importance of implementing a multilingual campaign when targeting Asian-American voters in the United States. These insights were applied when designing the APALC and Orange County Asian Pacific Islander Community Alliance, Inc. (OCAPICA) experiments. 5. Using feedback from their focus groups, APALC decided to compare the effects of two messages for each ethnic group (except Japanese and Vietnamese American voters, for whom they used only the universal message). APALC decided not to divide up the message for these two groups because the relatively small size of their lists for those national-origin groups made it unlikely that we would find a statistically discernable difference from the messaging. Call scripts featured either the universal message which emphasized that “voting empowers our community” or an alternate message that was specific to each ethnic group depending on the responses of its focus group participants. Voters in each ethnic group (except Japanese and Vietnamese) were divided randomly into either a universal treatment group or an alternate treatment group, and each group was called using the appropriate script. However, consistent with prior experiments comparing nonpartisan messages, no statistically significant effect was found between the two messages. 6. Information on nativity was missing for a significant portion of APALC’s pool of voters. Language of outreach to these voters was thus determined by age (i.e., assuming older voters were more likely to be foreign-born). This may have resulted in some voters being contacted in a language they were unable to understand or in which they lacked proficiency, thus reducing the effectiveness of the phonebank and/or mailers. However, since the phonebanks used the same guidelines in deciding in which language to contact targeted voters and were nevertheless able to significantly increase turnout, we would argue that the mailings’ lack of effectiveness cannot solely be attributed to the downward bias caused by language of outreach. 7. Pooled 2SLS is used because of the varying contact rates for different national-origin and language-use groups. 8. In addition to their phonebanking campaign, OCAPICA collaborated with the Orange County Registrar to engage the community in workshops and presentations that focused on how to register absentee ballots, and how to request in-language voting materials. They also worked with the registrar to discuss the recruitment of bilingual poll workers, the planning and

20

American Politics Research

organizing of town hall meetings, poll worker training, the display of in-language materials at polling places, name badges for bilingual poll workers, and polling place signage. During the period prior to the November 2006 election, OCAPICA also conducted a variety of community outreach activities. In the Chinese community, they distributed over 3,000 Easy Voter Guides at outreach events at Chinese churches, community organizations, and markets. They sent a community bulletin with voting information to 500 residents and worked to place five articles about voting in Chinese-language newspapers. In the Korean community, they distributed over 3,000 Easy Voter Guides at outreach events at Korean churches, community organizations, and markets, and another 2,000 at a local Korean Festival. In the Vietnamese community, they distributed over 2,000 Easy Voter Guides and other voting information at three local businesses, worked to place thirteen articles about voting in Vietnamese-language newspapers and participated in three talk shows on Vietnamese radio. They collaborated with the weather and traffic announcer from the local Vietnamese radio station to remind the community about voting every morning and also arranged for a voting song to be played on three different Vietnamese radio programs every day from September 17 through Election Day. 9. We should note that, thus far, experiments that have looked at the effects of different messaging have found that the message itself does not matter (Green & Gerber, 2008). That said, these experiments have looked at messages within the context of nonpartisan campaigns. Gimpel, Cho, and Shaw’s work (discussed above), however, suggests that messages may matter within a partisan context. Future research will be needed to answer this question.

References Abramson, P. R., & Aldrich, J. H. (1982). The decline of electoral participation in America. American Political Science Review, 76, 502-521. Addonizio, E. (2006). A social approach to voter mobilization and election day. Unpublished manuscript, Yale University Institution for Social and Policy Studies. Aldrich, J. H. (1993). Rational choice and turnout. American Journal of Political Science, 37, 246-278. Barnes, J. S., & Bennett, C. (2002). The Asian population: 2000. Census Brief (C2KBR/0116). Washington, DC: U.S. Department of Commerce, Bureau of the Census. Berinsky, A. J. (2005). The perverse consequence of electoral reform in the United States. American Politics Research, 33, 471-491. Clark, S. S., Wold, M., & Mayeri, H. (1997). The key to community voter involvement project: Fall 1996 election study. Retrieved April 16, 2008, from http://literacynet.org/slrc/vip/home.html Fraga, L. R., García, J. A., Hero, R. E., Jones-Correa, M., Ebers, V. M., & Segura, G. (2006). Su Casa es Nuestra Casa: Latino politics research and the development of American political science. American Political Science Review, 100, 515-521. García Bedolla, L. G. (2005). Fluid borders: Latino power, identity, and politics in Los Angeles. Berkeley: University of California Press. Gerber, A. S., & Green, D. P. (2000). The effects of canvassing, telephone calls, and direct mail on voter turnout: A field experiment. American Political Science Review, 94, 653-663. Gerber, A. S., & Green, D. P. (2001). Do phone calls increase turnout? A field experiment. Public Opinion Quarterly, 65, 75-85. Gimpel, J., Cho, W., & Shaw, D. (2005, September). Turning out the vote in Texas. Paper presented at the annual meeting of the American Political Science Association, Washington, DC. Gosnell, H. F. (1927). Getting-out-the-vote: An experiment in the stimulation of voting. Chicago: University of Chicago Press.

García Bedolla, Michelson / What Do Voters Need to Know?

21

Green, D. P., & Gerber, A. (2008). Get out the vote: How to increase voter turnout (2nd ed.). Washington, DC: Brookings Institution. Green, D. P., Gerber, A. S., & Nickerson, D. (2003). Getting out the vote in local elections: Results from six door-to-door canvassing experiments. Journal of Politics, 65, 1083-1096. Jamieson, A., Shin, H. B., & Day, J. (2002). Voting and registration in the election of November 2000. Current Population Reports (P20-542). Washington, DC: U.S. Department of Commerce, Bureau of the Census. Leighley, J. E. (2001). Strength in numbers? The political mobilization of racial and ethnic minorities. Princeton, NJ: Princeton University Press. Lien, P., Conway, M. M., & Wong, J. (2004). The politics of Asian Americans: Diversity and community. New York: Routledge. Michelson, M. R. (2005). Meeting the challenge of Latino voter mobilization. Annals of Political and Social Science, 601, 85-101. Rosenstone, S. J., & Hansen, M. (1993). Mobilization, participation, and democracy in America. New York: Macmillan. Rosenstone, S. J., & Wolfinger, R. E. (1978). The effect of registration laws on voter turnout. American Political Science Review, 72, 22-45. Tate, K. (1993). From protest to politics: The new black voters in American elections. Cambridge, MA: Harvard University Press. Trivedi, N. (2005). The effect of identity-based GOTV direct mail appeals on the turnout of Indian Americans. Annals of the American Academy of Political and Social Science, 601, 115-122. Verba, S., & Nie, N. H. (1972). Participation in America: Political democracy and social equality. Chicago: University of Chicago Press. Verba, S., Schlozman, K. L., & Brady, H. E. (1995). Voice and equality: Civic voluntarism in American politics. Cambridge, MA: Harvard University Press. Wolfinger, R., Highton, B., & Mullin, M. (2005). How postregistration laws affect the turnout of citizens registered to vote. State Politics and Policy Quarterly, 5, 1-23. Wolfinger, R., & Hoffman, J. (2001). Registering and voting with motor voter. PS: Political Science and Politics, 34, 85-92. Wolfinger, R. E., & Rosenstone, S. J. (1980). Who votes? New Haven, CT: Yale University Press. Wong, J. (2005). Mobilizing Asian American voters: A field experiment. Annals of the American Academy of Political and Social Science, 601, 102-114. Wong, J. (2006). Democracy’s promise: Immigrants and American civic institutions. Ann Arbor: University of Michigan Press. Lisa García Bedolla is an associate professor of social and cultural studies at the University of California, Berkeley. She is author of Fluid Borders: Latino Power, Identity, and Politics in Los Angeles (Berkeley: University of California Press, 2005), winner of the American Political Science Association’s Ralph Bunche Award. Her research focuses on the political incorporation of Latinos and other racial or ethnic groups in the United States with a particular emphasis on the intersection of race, class, and gender. Melissa R. Michelson is an associate professor of political science at California State University, East Bay. She is principal investigator for the evaluation of the James Irvine Foundation’s California Votes Initiative. Her research includes work on Latino, African American, Asian American, and youth voter mobilization as well as work on the political incorporation of Latino immigrants.