Presidential “Going Bipartisan” and the Consequences for Institutional Approval Brandon Rottinghaus Senator Don Henderson Endowed Chair Associate Professor bjrottinghaus@uh.edu Kent L. Tedin Professor ktedin@uh.edu University of Houston Department of Political Science 447 Philip G. Hoffman Hall Houston, Texas Abstract Although scholars have described how legislative efforts to spur bipartisanship fare, we have little knowledge about how bipartisanship can affect political opinions with their rhetoric or the most impactful message for opponents to respond. Using President Obama’s bipartisan speech to the GOP House Issues Conference in 2010, we look at the effect of the one-sided message on President Obama’s favorability rating. We then pair this message with three competing messages of varying partisanship to determine the degree of change (if any). The results show that the President’s one-sided message is effective, but if met with a competitive bipartisan message from the opposition party, approval of the President by all partisan groups increases even more. However, if the President’s bipartisan message frame is met using a partisan message from the opposition party, the President’s approval declines among all partisans, and approval of the Republicans in Congress increases but only for Republican identifiers. Keywords: presidential leadership; bipartisanship; presidential approval As candidate and chief executive, Barack Obama promised the American public that his interaction with Congress would be one of accommodation and bipartisanship. Mr. Obama sought to “build a cordial relationship with Republicans by seeking guidance on policy proposals, asking for advice on appointments and hoping to avoid perceptions of political arrogance given the wide margins of his victory” (Zeleny 2008). Lamenting that “in the nation’s political debate, something is broken,” the President sought to foster an image that he and his Administration would be willing to listen to and work with Republicans in Congress (Fletcher 2010). His approach is not altogether surprising. First, the American public was tired of the partisan bickering and disappointed with the half-hearted efforts at bipartisanship (Nagourney and Thee-Brenan 2010). Second, “bipartisanship” is an “electoral strategy that some politicians believe will broaden their appeal, and secure the support of middle-of-the-road or swing voters” (Tubowitz and Mellow 2005, 433), a vote getting strategy for President Obama. Third, bipartisanship may help the president get his agenda passed into law, as was the case in the 2010 lame-duck session of Congress (Newton-Small 2010). How presidents and Members of Congress deal with partisan conflict matters a great deal for political and policy reasons. Scholars have also found that perceptions of bipartisanship matter for how the public perceives government and the individuals who run it. Simply put, the actions of the individuals involved matters for the public’s support of the actors and the institutions. For instance, Morris and Witting (2008) found that citizen exposure to Congressional partisan speeches increased perceptions of partisanship and lowered support for Congress, while exposure to bipartisan speeches accomplished the reverse. Harbridge and Malhotra (2009) similarly found that those exposed to bipartisan messages increased their approval of Congress. Related to this, Ramirez (2009) found perceptions of partisan conflict 2 decreased approval for Congress. Newman and Siegle (2010, 342) argue that the “depth and breadth of polarization may hold dire implications for presidential governance.” Among these implications are consistently negative approval ratings for government (Mutz and Reeves 2005), an inability to negotiate across the political divide (and thus pass legislation), policy obstruction and gridlock. Of course, there are two key actors on this stage: President Obama and the Republicans in Congress. Each has their own decisions to make about the most effective rhetoric to accomplish their goals. Yet, there is debate about how to accomplish this rhetorical bargaining. Among those advocating the President take a partisan position is Nobel Prize winning economist Paul Krugman (2010). He argues that Ronald Reagan consistently gave ground on policy issues (like tax increases) but never varied from his hard right political rhetoric, giving little cover to Democrats wanting an expansion of government services. Obama, in “going bipartisan,” according Krugman, has helped the Republicans by giving legitimacy to some of their ideas, such as the need to curb government spending and the symbolic freezes on federal wages. Alternatively, others suggest the need for bipartisan tactics. Matt Bai (2010), in another New York Times editorial, suggested that “President Obama and the newly empowered Republicans may actually find more to talk about in the next Congress than they did in the last” because the public demands action and the only way to accomplish legislative action is to build cross-party coalitions. Addressing this debate and the efficacy of these tactics, our questions in this paper are simple and direct: (1) does a one-sided “need for bipartisanship” message (which will serve as our baseline) from President Obama affect his favorability with the public, (2) what is the effect of competitive messages varying in degree of partisan messages, and (3) what is the interaction 3 between message framing and individual partisan predispositions? Employing a large scale experimental survey design, we expose respondents to one or more bipartisan and or partisan messages and examine the effect of each message on respondents’ views toward President Obama and the Republican Party. The contribution of this article is three fold. First, we can extend the literature on political polarization from Congress (where much of the work has taken place) to the presidency. This is important because the president often sets the tone for the interbranch relationship. Second, we can decipher whether or not President Obama’s general bipartisan overture to the Republican Party can be successful and whether or not specific Republican responses of varying partisan tones undermined, or failed to undermine, the favorability evaluation of the President. We can do the same for the Republicans in Congress: how does the partisan tone of the opposition’s message responding to the President affect their own favorability rating? Third, using an experimental design and real world treatments, we can add new insights into the effect of framing. The Psychology of Opinion Change Our theoretical approach is grounded in the research on message framing. Beyond the empirical and political implications of a partisan / bipartisan message exchange, our design allows us to test aspects of framing theory in a fashion that is more nuanced than the typical design. The literature on framing is now voluminous, and we will just touch briefly on the underlying theory. Framing is about the weight people give to considerations (Chong 1996; Druckman 2004). In a framing effect, attitudes (in theory) change not because underlying considerations change, but because the underlying weights given to considerations change (Zaller 1992, 83). Frames are defined only in relation to specific issues, events or actors. In an oft cited example, the willingness of people to support a rally permit for an unpopular group like 4 the Klu Klux Klan (KKK) differs significantly if their behavior is framed as free speech or framed as the need for public safety (Nelson, Clawson and Oxley 1997; Sniderman and Theriault 2004). But most real world frames are not so simple or describe the issue with such starkly divergent frames with no counter-frame (and often no control group). The key predictors of a framing effect are the strength of the frame and the motivation of the message recipient. Thus, for the KKK example, a safety frame that focused on possible violence would be stronger than one that focused on traffic hazards. Similarly, a credible source is more effective than one of less creditability (Druckman 2001). From the message recipient aspect, there exist mediating predispositions. For instance, those holding well defined core values may respond differentially to competing frames (Sniderman and Theriault 2004) or may resist certain message despite a strong frame. The literature on framing has been the source of considerable critical analysis, much having to do with empirical testing. As stated above, a major goal of our research is to test subject responses to a partisan versus bipartisan approach to politics. We have integrated these tests within the message framing literature. The early work on message framing focused on one-sided messages. In the research on the KKK cited above, one group only got the free speech frame, while the other group got only the public safety frame. It is now generally appreciated that one-sided messages usually change people’s preferences (Chong and Druckman, 2010). However, this approach has serious problems with external generalizability. In the real word, there is competition among messages. It is very unlikely that in the case of a petition for a KKK rally, citizens would hear only one issue frame. Rather, in reality, most interested parties would be exposed to competitive frames. Indeed, Chong and Druckman (2007, 113) argue: 5 . . .little is known about the dynamics of framing in competitive contexts. We need to study further whether competing frames cancel one another and reinforce existing values, push people in conflicting directions, or motivate a more careful evaluation of the applicability of competing alternatives. Framing and Partisan Resistance Expectations For this study we use four substantive treatment groups and a control group ― the design is summarized in Figure 1. We refer to the four treatments that are not the control treatment as “frames,” meaning they vary in the text, emotion, images, and the style of presentation.1 We test first for the effect of the one-sided Obama bipartisan speech (“presidential bipartisanship” treatment). This treatment will serve as our baseline. Once the baseline is established (the effect of the Obama speech calling for bipartisanship), our major theoretical interest is in the effect of the varying Republican competing messages on the original baseline treatment. For the baseline treatment, we expect a “mainstreaming” effect. These individuals watch only the single bipartisan treatment and are exposed to a single message (Zaller 1992). Simply stated, approval of the President should increase if the respondents see just the president’s bipartisan message.. The three remaining treatment groups see the same bipartisan Obama speech followed by varying video recordings of Republican Leader John Boehner responding to President Obama’s call for bipartisanship (again, see Figure 1). These three differing responses vary greatly in their partisan content, or their “loudness,” as described in Table 1 (and below).2 What is being framed We are using what is generally referred to as a “media frame” meaning the frames vary style and emotion. Media frames can be contrasted to a “frame of thought” or the subjects understanding of a message (Chong and Druckman 2007, 101). 1 2 Loudness is sometimes defined as how often a message is repeated, but it can also be conceived as a message frame most likely to change consideration weights. We assume a consistent, biting partisan message frame is more likely to change consideration weights is a message frame that is blander, less partisan and less consistent (Druckman and Chong 2007, 102-105). 6 is the response tone of the Republican counter message to the Obama proposal for bipartisan cooperation. Our general theory is that in the aggregate, ceteris paribus, the “louder” (more consistently partisan) the response to the initial bipartisan message, the stronger the Republican competing frame and the greater the chance that the response will undercut Obama’s appeal for bipartisanship in the minds of the respondents. This should manifest in reduced favorability ratings for the President but higher the favorability ratings for the Congressional Republicans for all respondents. The explanation is straightforward ― consistent partisan frames will strengthen partisan considerations. So, partisan responses to bipartisan overtures should increase an individual’s partisan considerations. As the message of the counter frames becomes more mixed (for instance as bipartisan messages from the president are met with bipartisan messages from the opposition), the ability of the message to alter the weight given to partisan considerations weakens. [Figure 1 Here] Of course, in the real world, all things are not equal, and we would expect an interaction between the frames and individual predispositions. Resistance or acceptance by loyal partisans may limit or enhance the effect of the “loudness” of partisan frames. Given this, we should see differences across partisan attachments after viewing partisan or bipartisan messages. For instance, Democrats and Republicans are likely to react differently to the same messages. In effect, as suggested in the preceding paragraph, particular frames strengthen partisan preferences by highlighting the partisan position of the speaker (President Obama or John Boehner) which then cues the respondent of the same party to follow that opinion. In the context of our experiment, we assume that in-party identifiers (Democrats) will react favorably to an uncontested message from their party’s president. Independents will also react favorably as they 7 hear a one sided message. Republicans will be pulled in one direction by the one-sided message and in another by their own partisan predisposition. 3 In treatments 2, 3 and 4 competing messages are introduced that vary in their “loudness.” We expect that increasingly “loud” partisan frames (progressively more partisan) will increase the strength of partisan considerations for members of each party. Further, our expectation is that once conflict is introduced, as inparty identifiers (Democrats) view progressively more partisan responses to the President’s bipartisan message, the favorability of the President will increase but the favorability of Republicans in Congress will decrease. Simply put, competing strong partisan frames will raise the importance of partisanship and increase approval for the in-group (because of the partisan challenge) while decreasing approval of the out-group (the purveyors of the partisan challenge). We expect the same for out-party identifiers (Republicans), except the approval valences will be reversed. Contrasting this, in terms of bipartisanship, consistent frames are most effective in overcoming core value resistance (e.g., Republicans seeing consistently bipartisan messages from the opposition party and their party will approve of bipartisanship more). When the frames are consistent (e.g. endorsing bipartisanship) we should see approval of both President Obama and the Republicans in Congress increase for both Democratic and Republican respondents because all the cues are in the same direction. But, as the competing partisan frame grows louder (the challenge from John Boehner becomes stronger), the partisan identifiers will split: for Republicans the approval of the President should decrease while approval for the Republicans in 3 Keep in mind we are analyzing the effect of the message and change between Wave 1 (pretreatment) and Wave 2 (post-treatment). 8 Congress should increase while for Democrats the approval of the President should increase while approval for the Republicans in Congress should decrease. Independents who view only the one-sided bipartisan message from President Obama should increase their favorability of the President since this is the only message to which they are exposed.4 The same is true for the “dual bipartisanship” frame, since the two frames are consistently bipartisan. Importantly, Independents lack the most important of recipient predispositions—party identification―which would allow them to reject (or embrace) a consistently framed message. We expect Independents will grow less favorable towards the Republicans in Congress as the counter message grows increasingly loud (partisan). First, we expect that Independents are more inclined to favor bipartisanship over partisan conflict (Tubowitz and Mellow 2005; Shear 2011). Second, it may be that independents are offended by what they this perceive as uncivil language, and perceived lack of comity is why they chose to avoid party labels in the first place (Mutz and Reeves 2005; Mutz 2009). As Mutz (2007, 623) notes, “clearly there is something about incivility that rubs Americans the wrong way,” and Independents may be more off put by uncivil language than partisans. Therefore, we should expect Independents’ favorability of the President to increase and the favorability of the Republicans in Congress to decrease as the treatments become louder, or, put another way, as the frames become increasingly partisan. Unfortunately, the number of possible predispositions that might change consideration weighting among Independents is very large, making it difficult to predict theoretically. The Treatments and Research Design The President’s own partisans already have high favorability ratings; the Republicans have partisan resistance. The reverse would hold for the favorability of Republicans in Congress. 4 9 In order to validate our perceptions of the “partisan” or “bipartisan” nature of each speech, we counted individual sentences that included particular valence statements as either partisan or bipartisan. These data for each of the four speeches are summarized in Table 1 (again, the research design is identified in Figure 1). If an individual sentence had elements of both partisanship and bipartisanship, the most prominent direction was coded. In practice, though, this tended to not be a problem because the statements were generally either partisan or bipartisan as defined by our coding rubric. Partisan statements were statements (by either party) that offered criticisms of current or past programs, personalities, personal references to individuals or the other party, statements that expressed disbelief, incredulity or criticism of statements by the opposition and statements explaining the advantages of the speaker’s approach to a policy or a situation. Bipartisan statements included those where the speaker indicated they would work with the opposition by looking for “common ground,” identifying political or structural (Constitutional) advantages to working together, noting any agreement on specific issues or estimates and the use of “bipartisan” as a word in the appropriate context. It may be that it is the ratio of partisan to bipartisan statements, rather than the absolute number of statements, that is driving the respondents’ reactions, but, since the respondents view the full speech, the effect is equivalent. [Table 1 Here] The first randomly assigned treatment group saw the “one side only” frame (the “presidential bipartisanship” frame). In this frame, subjects were exposed only to a video recording of President Obama’s speech to the GOP House Issues Conference. They did not see a 10 Republican response. 5 As explained above, we expect a significant overall increase in Obama’s favorability rating among all groups. Many studies have shown the efficacy of one-sided messages, and we expect the same for the “one side only” frame when compared to a control group.6 Obama’s speech was the President’s bipartisan effort to show voters that the White House would “engage Republicans rather than govern in a partisan manner” (Hulse and Zeleny 2010). To open the speech, President Obama claimed the importance of “bipartisanship, not for its own sake, but to solve problems, that’s what our constituents, the American people, need from us right now.”7 Stressing the need for vigorous debate in a democracy, the President noted, “I hope that the conversation we begin here doesn’t end here; that we can continue our dialogue in the days ahead. It’s important to me that we do so.”8 The New York Times (Baker and Hulse 2010) reported the day after the speech: At a moment when the country is as polarized as ever, Mr. Obama traveled to a House Republican retreat on Friday to try to break through the partisan logjam that has helped stall his legislative agenda. What ensued was a lively, robust debate between a president and the opposition party that rarely happens in the scripted world of American politics. The Washington Post also noted that the meeting, while often tense, produced “bipartisan comity over the stark policy differences that separate the two sides” (Kane and Bacon 2010). The data from Table 1 bear these observations out. The President’s speech captured 88 total “bipartisan” statements to 13 “partisan” statements according to our coding rubric. 5 To preserve the more purely bipartisan nature of the speech, respondents only viewed the President’s speech, not the question and answer session that originally followed. 6 The control group watched a 20 minute video on the Founding Fathers. Barack Obama, 2010, “Remarks by the President at the GOP House Issues Conference,” Baltimore, Maryland, January 29. 7 8 Ibid. 11 As identified in Figure 1, the subjects not assigned to the one-sided “presidential bipartisanship” treatment were randomly assigned to one of three other treatment groups. Each group viewed the Obama speech and one of three speeches by Republican Leader John Boehner. First, the least “partisan” (or most “bipartisan”) of the messages from John Boehner was his interview on CNN (we label this the “dual bipartisanship” treatment). The data in Table 1 show this interview was the most bipartisan of the three Boehner speeches (22 partisan statements, 22 bipartisan statements). For instance, Minority Leader Boehner noted he would “never, ever give up on trying to find common ground with the President” on a range of issues. Similarly, the tone of Leader Boehner’s message was polite, measured and reasonable as he answered questions from CNN host Anderson Cooper. Although we offer this interview as an example of a “bipartisan” message, the interview did contain 22 partisan statements. Thus, the message was not uniformly bipartisan, but, of those messages available, this was the most accommodating and “bipartisan” presentation. While not as bipartisan as President Obama’s speech, this message is most bipartisan message among the three Republican response frames. In the moderately partisan speech, a four minute video “Weekly Republican Address,” Leader Boehner stressed his desire to work with President Obama and Speaker of the House Nancy Pelosi, but he also noted that “where there are differences, it was our obligation as a party to explain to the American people how we [the Republicans] would do things better.” We label this treatment “conflicting ideology.” Leader Boehner indicated that his and the Republican Caucus’ ideas for governing and on specific public policies (the stimulus package, health care, tort reform, the budget) were better than those offered by the Obama White House and that in a bipartisan meeting with the President, Leader Boehner provided these ideas to the President in writing. In contrast to Leader Boehner’s CNN interview which was primarily bipartisan, Table 1 12 shows that this Weekly Republican Address is slightly more partisan than bipartisan (26 partisan statements to 22 bipartisan statements). Likewise, although his tone and approach to his statement was mild and direct, the language used was more partisan and confrontational than in the CNN interview, differentiating the two speeches. Also, as noted below, this speech was more personal (with respect to President Obama) than the “bipartisan” Boehner speech. The third speech by Leader Boehner was the most partisan of the three speeches, and this is the speech watched by the respondents in the “partisan attack” treatment group. This speech was given on the House floor on March 21, 2010 in the month before the health care reform legislation was debated and a month after the President delivered his bipartisan speech to the House GOP Issues Conference. In contrast to the other two speeches, in this speech Leader Boehner aggressively challenged the President and his agenda. In this speech, he loudly and angrily shouted “Hell no, you can’t” (which became a hit on YouTube) to a litany of criticisms of the Obama Administration and several policies (to the point where the parliamentary leader had to admonish those in attendance to calm down). Leader Boehner argued that the President has failed to listen to the voters, and that the White House’s legislative agenda does not “reflect the will of our constituents.” Table 1 clearly demonstrates that this speech is the most partisan of the three used in the experiment. Leader Boehner made 47 partisan statements to 1 bipartisan statement. This is the strongest (and loudest) partisan frame and is the most inconsistent with the original Obama bipartisan speech. The speeches are also distinguished from each other by the number of political and personal statements. Policy statements were statements that were substantially related to a policy, where the reference or point was about a policy or issue. For instance, John Boehner noted in the CNN interview (the moderate position) “In order to put their government-run health 13 care plan together, they have got to raise taxes in their budget. Already $634 billion over the next 10 years, and that’s only about half of what it is expected to cost.” Personal statements were statements where an individual was the reference and where no policy was mentioned. For instance, John Boehner (speaking of then Speaker Nancy Pelosi) said, “Lying to the Congress of the United States is a crime. And if the Speaker is accusing the CIA and other intelligence officials of lying or misleading the Congress, then she (Pelosi) should come forward with evidence and turn that over to the Justice Department so they can be prosecuted.” The results show that the more partisan the response, the more likely the speech is to have more personal statements and less likely to have policy statements. Aggregating the statements, Boehner’s response speech in the “dual bipartisan” treatment has 3 personal statements, “conflicting ideology” has 14 and “partisan attack” has 23. Likewise, the percent of policy issues is lower for those speeches that are more partisan: 52% of the statements in “partisan attack” were policybased, compared to 68% for “conflicting ideology” and 86% for “dual bipartisanship.” While we are unable to completely isolate to which effect the respondents are reacting, this additional level of analysis helps us to clarify the tone and substance of the individual treatments and establish that partisan speeches tend to be more conflictual. The Experiment and Respondents In order to experimentally manipulate the effects of exposure to individual matched speech formats, we conducted a panel survey of undergraduate students at a major university in a large state in the southwest. Three weeks into the spring semester of 2010, student subjects in multiple Introduction to American Government classes9 were surveyed on a range of factual, 9 Introduction to American Government is a required course at the institution making selfselection not an issue. Because all students are required to take these introductory classes, we are 14 policy, ideological, political issues and standard demographics. The second wave was conducted two months later, when the subjects randomly received one of four treatments or a control treatment. The surveys were designed using Qualtrics and conducted over the Internet. A total of 869 subjects completed both waves of the panel.10 The sample was highly diverse, with a significant representation of African-Americans, Asians, Hispanics and Whites.11 The sample was also very diverse politically.12 For the post-test portion of the experiment, the students were randomly assigned to one of five experimental treatment groups, again as shown in Figure 1 above. As with any experiment, there are advantages and disadvantages. The first advantage is that random assignment allows us to test competing hypotheses with a high degree of internal validity. We can investigate presidential leadership and the effects of varying from opposition response with confidence concerning cause and effect. The usual confounding factors in crosssectional designs are removed. On the other hand, the most important limitation to our approach is external validity. First, our survey participants were asked to watch all of a particular presentation, where in reality they might only watch part or highlights from the speeches or not watch at all. Second, the subjects watched the videos on the Internet in their homes or dorms instead of on television during the real time of the addresses. This method has some advantages not biasing the sample in favor of political science majors who may be attentive to or otherwise involved in politics. The students were given minor course credit for participating in the project. 10 Non-citizens were deleted as were those who expressed a party preference other than Democratic, Independent or Republican, yielding a total of 859 respondents who completed both waves of the study. 11 There were 173 blacks, 159 Asians, 184 Hispanics, 272 non-Hispanic whites and 71 “other.” 12 The question on partisanship asked them: Generally speaking, do you consider yourself a Democrat, a Republican, an Independent or what?” The sample was 49% Democratic, 22% Independent, and 31% Republican. 15 in terms of reality that would not be achieved in a conventional laboratory setting. The subjects were not under supervision and they could watch the videos at their convenience over a two week period. As a precaution against inattentive respondents, we gave each respondent a five item test on the video they had just watched. The results showed the mean number correct was 3.78, suggesting most respondents paid attention to the presentations.13 Third, using real world messages costs us a bit in terms of experimental control. It may be that some factor other than our proposed treatment (partisan/bipartisan messages) is affecting the results.14 Finally, although the sample was very diverse and the randomization worked well, care needs to taken with generalizations (Nelson, Clawson, and Oxley 1997; Scott, Matland, Michelbach, and Bornstein 2003; Chong and Druckman 2010).15 13 We found no statistically significant differences by treatment group for the attention to the treatment questions (using a χ2 test). Specifically, the correct right for control was 3.70; for Obama’s speech 3.75, for Boehner’s partisan attack 3.66; for the conflicting ideologies 4.06, and for the dual partisanship 3.66. 14 Order effects of the speeches may also be playing a part in the findings (showing the same speech first each time). In this article we are specifically interested in the president’s efforts to initiate bipartisanship (and the concurrent levels of challenge) rather than the changes in bipartisanship more broadly, making the current design the most appropriate. We found no statistically significant differences between any of the treatment groups using χ2 tests. Eyeballing the differences between the groups for partisanship or ideology, the percentages are all similar across the randomly distributed treatment groups. For instance, for partisanship, Democrats made up 49% of the control, 45% of the “presidential bipartisanship” treatment, 54% of the “partisan attack” treatment, 53% of the “conflicting ideology” treatment and 45% of the “dual bipartisanship” treatment. Republicans made up 21% of the control, 23% of the “presidential bipartisanship” treatment, 19% of the “partisan attack” treatment, 20% of the “conflicting ideology” treatment and 22% of the “dual bipartisanship” treatment. Independents made up 29% of the control, 31% of the “presidential bipartisanship” treatment, 27% of the “partisan attack” treatment, 27% of the “conflicting ideology” treatment and 32% of the “dual bipartisanship” treatment. For instance, for ideology, liberals made up 37% of the control, 40% of the “presidential bipartisanship” treatment, 35% of the “partisan attack” treatment, 35% of the “conflicting ideology” treatment and 32% of the “dual bipartisanship” treatment. Conservatives made up 20% of the control, 19% of the “presidential bipartisanship” treatment, 18% of the “partisan attack” treatment, 14% of the “conflicting ideology” treatment and 20% of the “dual 15 16 The Effects of Going Bipartisan and Opposition Messages The dependent variables for this analysis are changes in favorability. For instance, we asked respondents to indicate favorable or unfavorable opinions of President Barack Obama on a one-to-seven scale. We focus on changes in favorability because this is likely the most immediate and prominent effect of watching and interpreting political events. Wave 2 is then subtracted from Wave 1. The difference was contrasted to the control group difference and tested for statistical significance with a difference of means test. This design is stronger than a standard cross-sectional experiment. First, by using a control group we can account for the events (history) that occurred between waves. Second, we minimize order and priming effects. Partisanship and key demographics are measured at Wave 1.16 Since the substantive items on the two waves are identical, if there are priming or order effects, they are equal in each wave. The result is that they do not contaminate the measurement of the dependent variables. Looking first at Obama’s favorability change among all subjects between Wave 1 and Wave 2 (shown in Table 2) for the one side only treatment, we see the expected pattern. The question asked: “On a one to seven scale do you have a favorable or unfavorable opinion of President Barack Obama?”17 Positive numbers reflect increases in favorability. Change in the control group is virtually zero. Seeing only President Obama’s (one-sided) bipartisan bipartisanship” treatment. Moderates made up 42% of the control, 40% of the “presidential bipartisanship” treatment, 47% of the “partisan attack” treatment, 50% of the “conflicting ideology” treatment and 47% of the “dual bipartisanship” treatment. Similar (non-significant) effects were found for race and gender. Further details are available from the authors. The question querying partisan attachments was: “Generally speaking, do you consider yourself a Democrat, a Republican, and Independent or what?” The choices were: strong Democrat, Democrat, but not strong, Independent, Republican, but not strong, Strong Republican, other or none. Strong and weak partisans were grouped together. 16 17 One is marked “unfavorable,” four is marked “neutral” and seven is marked “favorable.” 17 presentation modestly increases his approval by .25 among all subjects when compared to change in the control group (p. < .05). However, seeing the Obama presentation followed by the John Boehner “partisan attack” speech decreases Obama’s change in approval essentially back to zero (.03; n.s.) when contrasted to the control. Next, for John Boehner’s moderate response (“conflicting ideology”) to the Obama presentation, we see a positive change in Obama’s approval of .22 points, similar to where it was for the one-sided Obama treatment (“presidential bipartisanship”) (p. <.10). Finally, in the Obama bipartisan presentation followed by the Boehner bipartisan interview (“dual bipartisanship”), we see an increase of .49 compared to the control group (p. < .001), the greatest increase in Obama’s favorability rating of any of the treatments. In fact, this improvement in Obama’s favorability exceeds the one sided frame. When it comes to President Obama’s favorability rating, the lesson for Republicans is clear: going bipartisan has the potential downside of boosting the President’s approval rate beyond what would exist if the Republicans simply did nothing. [Table 2 Here] We also asked respondents at Wave 1 and Wave 2 to indicate their favorable or unfavorable rating of the Republicans in Congress.18 The question asked: “On a one to seven scale do you have a favorable or unfavorable opinion of the Republicans in Congress?”19 Positive numbers reflect an increase in favorability. Following the 2008 election, John Boehner was the face of the Republican Party. While the question was generic to the Republicans in Congress, the specific referent was House Minority Leader John Boehner since he was the 18 For both survey waves the order of these Obama-House Republicans favorability items were randomly rotated. 19 One is marked “unfavorable,” four is marked “neutral” and seven is marked “favorable.” 18 featured Republican in the treatments. As can be seen in Table 2, the treatments had no effect in either increasing or decreasing the favorability of Congressional Republicans. The change for each treatment differs little from the control. When it comes to evaluating the Republicans in Congress, the bipartisan Obama speech at the Republican Conference (“presidential bipartisanship”) had no mean effect. In fact, all of the treatments resulted in an aggregate movement of essentially zero among all respondents. These results need to be put in context. At a minimum it important to note in the aggregate sample that Democrats outnumber Republicans by a margin of 1.6 to 1.0. Thus, we need to disaggregate by partisan predisposition. The Effect of Going Bipartisan Considering Partisan Predispositions The significant findings for the changes in approval in Table 2 could be masking important subgroup changes. For instance, in the case of the approval of the Republicans, if the Democrats move toward less favorability, the Republicans toward more favorability and the Independents split about evenly, that would account for the null results in Table 2. To examine this possibility, we next turn to the motivation of the message recipient. As we argued above, Democrats and Republicans are not likely to react in the same way to the message frames. Independents (who are largely without such predispositions) also react differently to different and competing partisan frames. For public, high-level political battles, almost everything is filtered through the lens of partisan predispositions providing a possible resistance agent to the power of the frame. To account for this possibility, the analysis from Table 2 is repeated for Democrats, Independents and Republicans in Figures 2 and 3 allowing us to more closely examine the interaction of each frame and partisans predispositions. 19 Figure 2 shows the change in Obama’s favorability by treatment multiplied by party using the same favorability question as above.20 Positive numbers reflect increases in favorability on the seven point scale. Generally, Democrats experienced meaningful treatment effects. When contrasted to the control, those exposed to the one sided “presidential bipartisanship” treatment give Obama a favorability boost of .25 (p. < .05). As expected, a one-sided message among ingroup partisans is effective. Yet, among Democrats, when Obama’s speech is followed by Boehner’s partisan speech on the House floor (“partisan conflict”), the favorability of President Obama decreases by .08 from the first wave and actually falls slightly below the control group. Boehner’s more balanced Republican video address (“conflicting ideology”) yields a .14 (p. < .10). increase in Obama’s favorability, while those exposed to the “dual bipartisanship” treatment raised Obama’s favorability by .40 units contrasted to control among Democrats (p. < .01).21 This latter effect is the largest point change in Obama’s favorability movement among Democratic respondents. The joint “frame” of bipartisanship is effective as respondents are more likely to follow the “mainstreamed” bipartisan approach, as expected, but louder competing frames tends to decrease approval of the President among even Democrats. [Figure 2 Here] The data for Republican identifiers reveal somewhat similar findings to the Democratic identifiers. The first point of consequence is that the one-side only treatment (“presidential bipartisanship”) does not move Republicans to more favorably rate Obama (as it does Democrats and Independents). The recipient predispositions outweigh the Obama one-sided frame, even 20 The total respondents for each were as follows: presidential bipartisanship (154), dual bipartisanship (178), conflicting ideology (188) and partisan attack (149). Both the treatment and party affiliation variables were statistically significant at .p. >.01. 21 Significance tests contrast treatment to control group. 20 when he is proposing cooperation with Republicans. For students of public opinion this is somewhat reassuring. Not all citizens are blown around by current information, rather these Republicans encounter internal partisan resistance (Zaller 1992). Interestingly, the Boehner partisan address (“partisan attack”) has essentially no effect. However, as hypothesized, when Boehner’s addresses become increasingly conciliatory (bipartisan), the favorability ratings of Obama among Republicans increase significantly. The “conflicting ideology” treatment moves Obama’s favorability by .50 (p < .05) and the “dual bipartisanship” treatment moves it by .84 units (p <.001), both significant differences from the control group. The pattern we saw among the Democrats is essentially the same among Republicans but with greater movement. In practical terms, if Republicans want to undermine a Democratic president’s favorability rating among their own partisans, it seems best to engage in conflicting partisanship in order to alter the effect of the one- sided frame. Finally, we examined Independents and the effect of each treatment on approval of Obama. We anticipated the one-sided argument to have its greatest effect on Independents (since they have no modifying predisposition), and we find that this hypothesis is supported. Self-identified Independents register positive movement (.38) in response to the solo Obama bipartisan speech (“presidential bipartisanship”) (p. <.05), a difference of .57 units when contrasted to the control. This result supports much previous work on framing—no recipient predispositions and a one-sided frame are very effective in moving public opinion. The “dual bipartisanship” frame showed remarkably similar results, also registering a .38 point movement in approval. The loudness hypothesis is supported in that the mainstream effect of the singlesided frame (“presidential bipartisanship”) and the two-sided consistent bipartisan frame (“dual bipartisanship”) are both equally successful in increasing approval of the president. However, 21 the Boehner challenge treatments showed fewer significant effects. Compared to the control group, the “conflicting ideology” treatment yielded a .26 unit movement and “partisan attack” yielded a .24 unit movement. For the Presidential favorability ratings, the message seems reasonably clear ― Independents’ approval of the President are slightly lower when counter messages have a partisan bite. Figure 3 identifies respondent attitudes towards the Republicans in Congress on the same seven point scale described above. Again, positive numbers reflect increases in approval.22 As expected, Democrats and Republicans have different favorability ratings for the Republicans in Congress. Among Democratic identifiers in the sample, they become more favorable to the Republicans in Congress following Obama’s bipartisan gesture (“presidential bipartisanship”) (.38 change), as predicted. Similarly, the Democrats were more likely to favor the Republicans in response to Boehner’s more moderate address, where the “conflicting ideology” treatment produces a .17 increase and the “dual bipartisanship” treatment produces a .37 increase (equal to the one-sided surge in “presidential bipartisanship”). However, Boehner’s aggressive speech (the “partisan attack” treatment) undermined the favorability of the Republicans in Congress among Democrats and fell below the control group (-.05). For the Democrats evaluating the Republicans in Congress, partisan message loudness is overwhelmed by partisan predisposition resistance. [Figure 3 Here] Interestingly, Republicans, after viewing the one-sided Obama message, became less favorable to the Republicans in Congress (-.52) between Wave 1 and Wave 2. Presidents do, it 22 The total respondents for each were as follows: presidential bipartisanship (169), dual bipartisanship (179), conflicting ideology (188) and partisan attack (151). 22 seems, find success in moving partisans (Tedin, Rottinghaus and Rodgers 2011). However, given some “red meat” (in the form of the Boehner partisan attack in the “partisan attack” treatment), Republican respondent’s favorability of Republicans in Congress jumped significantly between the two waves (.40 points contrasted to control). However, as the Boehner message frame moderated, Republican favorability of the Republicans in Congress declined, falling below the changes in approval held by the control group for the “dual bipartisanship” frame (-.23). In this political battle, Republican identifiers clearly desired some degree of challenge to the ideas of the opposition party, although bipartisanship produced less displeasure with the Republicans in Congress than the solo Obama message. The conflictual framing works to the advantage of the Republicans here, but the “mainstreaming” effect (of multiple bipartisan statements) did not work as expected to produce consistent positive approval change for the Republicans in Congress. The implication is that predispositions matter in limiting framing effects. Finally, for Independents’ approval of the Republicans in Congress, there was virtually no movement for the “presidential bipartisanship” or “dual bipartisanship” frames ― the effects of each were approximately the same (just under zero change) although we expected to see positive change. However, there was a significant negative reaction to John Boehner’s stridently partisan address (the “partisan conflict” treatment). Independents viewing the “partisan attack” treatment moved to a more negative approval of the Republicans in Congress (-.40), significantly below their approval changes for any of the other treatments. As suggested above, being primarily nonpartisan, perhaps they dislike uncivil discourse or perhaps they are more likely to favor bipartisan solution (consistent with national polling findings). Whatever the reason, we find that Independents are not likely to move their opinions on the Republican Party 23 very much positively or negatively, except when the partisan rancor increases wherein they are much less likely to approve of the Republicans in Congress. Conclusions Commentators suggest that the future of President Obama’s presidency may depend on how much he is able to foster bipartisan cooperation with the Republicans, especially after the 2010 midterm elections provided major gains for the opposition party and control of the House of Representatives (Stolberg 2010). Indeed, presidents facing an opposition Congress are often encouraged to reach out to the other side of the aisle. In many ways, this is a classic framing issue. The President and Congress must engage each other. But, what is the best strategy in terms of tone of the message? While we know a great deal about presidential efforts to bargain with Congress (Edwards 1989; Beckmann 2010), we know less about how presidential rhetoric stands up to message competition from the opposing party in a way that enhances our understanding of how the rhetoric over bipartisanship is best handled. This article has given us some insight into this topical question and an interesting approach to analyzing political framing. Although we explore the framing debate in a topical setting, there are some limitations to our experimental research design that limits our ability to capture the president’s effectiveness at going bipartisan. First, the politics of the situation changed since the experiment was conducted: Minority Leader Boehner in 2011 became Speaker Boehner. While we captured a range of reactions to the President’s speech, from accommodation to hostility, it may be the case that the public alters their opinions of the situation when the group challenging the President’s message control a branch of government and are therefore more relevant as political actors. Notably, does the strength the counter message interact with holding majority control of the House of Representatives? One might expect an even stronger reaction to partisan attack message if 24 Republicans are responsible for managing the House. Or, people may expect more civility from the House leader. Second, although we identify the relative number of partisan, policy, political and personal statements for each speech, we are unable to know concretely which precise condition caused the respondents to act. The tone of the speeches could have been the key characteristic rather than the bipartisan effect that we posit. This is an inherent disadvantage in using real life speeches where we cannot control precisely the content. Despite these limitations, the implications from our findings suggest that the study of bipartisanship is not an empty exercise. President Obama improved his own approval for all respondents when the message they saw was one sided, but that is hardly a surprise given the research on framing. Importantly, our data show that competing messages can sometimes undermine the one-sided message and in other instances enhance the effect of one-sided messages. Specifically, for all partisan groups, a consistent partisan frame actually seems to work against the Republicans by increasing support for the President and decreasing support the Republicans in Congress. But, for Republicans, the “mainstreaming” message from the bipartisanship frame does not have the same registered effect on approval of Republicans in Congress. The take away from this analysis is that bipartisan framing can break through partisan predispositions for all groups of partisan identifiers (such as when the bipartisan message is consistent across parties), even among groups who would otherwise disapprove of the bipartisan message by itself. Differences in ideology and their role in resistance to framing is an important question to address in the future. In practical terms, the findings show that the opposition party can hurt the approval of the sitting president if they reject bipartisan advances with a partisan response. In fact, Democrats and Independents who see the President making overtures and being rejected in a bitterly 25 partisan way are more likely to disapprove of President Obama. The response from the opposition matters as well. As the response messages from the House Minority Leader grow more partisan, ratings by Democratic and Independent respondents of the Republicans in Congress go from very small to negative, not a surprise given the predispositions of the message recipients. On the other hand, as the message frame from Representative Boehner grew more conflictual (more partisan but a less consistent bipartisan frame), Republicans became more likely to favor the Republican Party. Finally, although Independents are not especially persuaded by bipartisan messages from either or both parties, the strong partisan frame offered by John Boehner reduced their approval of all institutional actors. 26 References Ansolabhere, Stephen and Shanto Iyengar. 1995. Going Negative: How Political Advertisements Shrink and Polarize the Electorate. New York: Free Press. Baker, Peter and Carl Hulse. 2010. “Off Script, Obama and the GOP Vent Politely.” The New York Times, January 30. Bai, Matt. 2010. “A New Chance for Bipartisanship, All Posturing Aside.” The New York Times, December 23. Beckmann, Matthew N. 2010. Pushing the Agenda: Presidential Leadership in U.S. Lawmaking, 1953-2004. Cambridge: Cambridge University Press. Fleisher, Richard and Jon R. Bond. 2000. “Partisanship and the President’s Quest for Votes on the Floor of Congress.” In Polarized Politics: Congress and the President in a Partisan Era. Washington, DC: CQ Press. Canes-Wrone, Brandice. 2006. Who Leads Whom? Presidents, Policy and the Public. Chicago: University of Chicago Press. Chong, Dennis. 1996. “Creating Common Frames of Reference on Political Issues.” In Political Persuasion and Attitude Change, eds. Dinana C. Mutz, Paul M. Sniderman and Richard Brody. Ann Arbor: University of Michigan Press. Chong, Dennis and James N. Druckman. 2007. “A Theory of Framing and Opinion Formation in Elite Environments.” Journal of Communication 57: 99-118. Chong, Dennis and James N. Druckman. 2010. “Dynamic Public Opinion Communication Effects over Time.” American Political Science Review 104: 663-680. Druckman, James N. 2001. “On the Limits of Framing Effects.” Journal of Politics 63 (3): 1041-66. Druckman, James N. 2004. “Political Preference Formation.” American Political Science Review 98 (3): 671-86. Edwards III, George C. 2003. On Deaf Ears: The Limits of the Bully Pulpit. New Haven: Yale University Pres. Erikson, Robert S. and Kent L. Tedin. 2011. American Public Opinion (New York: Longman). Fletcher, Michael A. 2010. “Obama, at National Prayer Breakfast, Calls for Civil Political Debate.” The Washington Post, February 5. Harbridge, Laurel, and Neil Malhotra. 2011. “Electoral Incentives and Partisan Conflictin Congress: Evidence from Survey Experiments." American Journal of Political Science. 55(3): 494-510. Hulse, Carl and Jeff Zeleny. 2010. “Obama Continues Policy Outreach to Republicans.” New York Times, February 3. 27 Kane, Paul and Perry Bacon Jr. 2010. “Obama Goes to GOP’s House for a Wide Open Exchange.” The Washington Post, January 30. Kernell, Samuel. 2007. Going Public: New Strategies of Presidential Leadership. Washington: CQ Press. Lee, Frances E. 2008. “Divders, Not Uniters: Presidential Leadership and Senate Partisanship, 1981-2004.” Journal of Politics 70 (4): 914-928. Morris, Jonathan and Marie Witting. 2001. “Congressional Partisanship, Bipartisanship and Public Opinion: An Experimental Analysis.” Politics and Policy 29 (1): 47-67. Mutz, Diana C. 2006. Hearing the Other Side: Deliberative Versus Participatory Democracy. Cambridge: Cambridge University Press. Nagourney, Adam and Megan Thee-Brenan. 2010. “Poll Find Edge for Obama Over GOP Among the Public.” New York Times, February 12. Nelson, Thomas E., Zoe M. Oxley, and Rosalee A. Clawson. 1997. “Issue Framing in a Civil Liberties Conflict and its Effect on Tolerance.” American Political Science Review 91 (3): 567-83. Newman, Brian and Emerson Siegle. 2010. “The Polarized Presidency: Depth and Breadth of Public Partisanship.” Presidential Studies Quarterly 40 (2): 342-63. Newton-Small, Jay. 2010. “Obama’s Lame-Duck Comeback: Hello, Bipartisanship.” Time, December. Peterson, Mark A. 1990. Legislating Together: The White House and Capitol Hill from Eisenhower to Reagan. Cambridge: Harvard University Press. Pious, Richard M. 1999. “The Limits of Rational Choice: Bush and Clinton Budget Summitry.” Presidential Studies Quarterly 29 (3): 617-37. Scott, John T., and Richard E. Matland, Philip A. Michelbach and Brian H. Bornstein. 2003. “Doing Rawls Justice: An Experimental Study of Distributive Justice Norms.” American Journal of Political Science 47: 523-39. Sinclair, Barbara. 2001. “Bipartisan Governing: Possible, Yes; Likely, No.” PS: Political Science 34 (1): 81-83. Shear, Michael D. 2011. “Obama Strategy: Share Credit (and Blame?)” The New York Times, March 15. Sniderman, Paul and S.M. Theriault. 2004. “The Structure of Political Argument and the Logic of Issue Framing.” In W. E. Saris and P. M. Sniderman (Eds.) Studies in Public Opinion. Princeton, New Jersey, Princeton University Press. Stolberg, Sheryl Gay. 2010. “Obama’s Playbook After November 2.” The New York Times, October 24. Ramirez, Mark D. 2009. “The Dynamics of Partisan Conflict on Congressional Approval.” American Journal of Political Science 53 (3): 681-94. 28 Rottinghaus, Brandon. 2009. “Strategic Leaders: Identifying Successful Momentary Presidential Leadership of Public Opinion.” Political Communication 26 (3): 296-316. Rottinghaus, Brandon. 2010. The Provisional Pulpit: Modern Presidential Leadership of Public Opinion. College Station: Texas A&M University Press. Tedin Kent, Brandon Rottinghaus and Harrell Rodgers, 2011. “Revisiting Presidential Leadership: Widening the Dimensions of Opinion Change by Group and Communication Tactic.” Political Research Quarterly 64 (3): 506-19. Thomas, Scott J. and Bernard Grofman. 1992. “Determinants of Legislative Success in House Committees.” Public Choice 74 (2): 233-43. Trubowitz, Peter. 2005. “‘Going Bipartisan’: Politics by Other Means.” Political Science Quarterly 120 (3): 433-53. Zaller, John. 1992. The Nature and Origins of Mass Opinion. Cambridge, England: Cambridge University Pres. Zeleny, Jeff. 2008. “Initial Steps by Obama Suggest a Bipartisan Flair.” New York Times, November 24. 29 FIGURE 1 Experimental Design for Treatment Groups Control Group “Founding Fathers” Video Group 1: “ PRESIDENTIAL BIPARTISANSHIP” BIPARTISAN Message from President, no response from opposition Obama’s speech to GOP Conference Survey Group 2: “DUAL BIPARTISANSHIP” BIPARTISAN message from President and from Republican Leader Obama’s speech to GOP Conference Boehner’s Bipartisan Interview (CNN) Survey Group 3: “CONFLICTING IDEOLOGY” BIPARTISAN message from President and PARTISAN message from Republican Leader Obama’s speech to GOP Conference Boehner’s Weekly Republican Video Address Survey Group 4: “PARTISAN ATTACK” BIPARTISAN message from President and VERY PARTISAN message from Republican Leader Obama’s speech to GOP Conference Boehner’s Speech on House floor 30 Survey TABLE 1 Partisan and Bipartisan Statements in Sample Speeches “Presidential Bipartisanship” “Partisan Attack” “Dual Bipartisanship” “Conflicting Ideology” PARTISAN STATEMENTS (both parties) Criticism of programs 3 25 9 6 Criticism of personalities Criticism of past programs Criticisms of other party 0 0 4 19 0 3 9 0 0 8 0 0 Statements critical of their claims or statements 2 0 0 3 Explains advantages of their party’s approach (over the other party) 4 0 8 5 SUBTOTAL 13 47 26 22 Cooperative statements – implications the parties will together Statements about “looking for common ground” 11 0 9 0 17 0 2 12 Explains obligation to work together (because the public wants it or the Constitution demands it) Agreement on issues (the importance of issues or on the substance of issues) 27 0 1 0 20 0 1 1 Agreement on numbers or estimates for budgets 1 0 4 0 “Bipartisan” (as a word) 3 1 3 0 Acknowledging the other party/side/person is working hard or has a tough job. 4 0 0 5 Statements of friendship or camaraderie 5 0 0 4 SUBTOTAL 88 1 20 22 BIPARTISAN STATEMENTS (both parties) NOTE: Data compiled by the authors. 31 TABLE 2 Change in Favorability for Obama, Republicans and Perceptions of Likely Bipartisan Success WAVE 1-WAVE 2 Treatment Sig.* increase in Obama’s favorability Control -.01 n.s. Presidential Bipartisanship .25 .05 Dual Bipartisanship .49 .001 Conflicting Ideology .22 .10 Partisan Attack .03 .n.s. ANOVA .002 WAVE 1-WAVE 2 Sig. Treatment increase Republican’s favorability Control .00 n.s. Presidential Bipartisanship -.03 n.s. Dual Bipartisanship .04 n.s Conflicting Ideology .07 n.s Partisan Attack -.05 n.s ANOVA n.s. NOTE: *Contrast to the control for statistical tests for entire sample. 32 Figure 2: Change in President Obama's Favorability by Treatment Effect Mean Change Pretest - Posttest 0.8 0.7 0.6 0.5 0.4 Democrat 0.3 Independent 0.2 Republican 0.1 0 -0.1 -0.2 Control Presidential Bipartisanship Dual Bipartisan Conflicting Partisan Attack Ideology NOTE: The question asked: “One a one-to-seven scale (and you can use any number in between), do you have a favorable or unfavorable opinion of President Barack Obama?” Positive numbers note increases in approval. Control refers to the control group. Going from left to right, the standard errors for the Democrats are: .12, .15, .15, .14, .15 For the Independents they are: .18, .20, .21, .19, .19. For the Republicans they are: .14, .16, .20, .16 .20. 33 Figure 3: Change in Republicans in Congress Favorability by Treatment Effect Mean Change Pretest - Posttest 0.4 0.2 -1E-15 Democrat Independent -0.2 Republican -0.4 -0.6 Control Presidential Dual Bipartisanship Bipartisanship Conflicting Ideology Partisan Attack NOTE: The question asked: “One a one-to-seven scale (and you can use any number in between), do you have a favorable or unfavorable opinion of the Republicans in Congress?” Positive numbers note increases in approval. Control refers to the control group. Going from left to right, the standard errors for the Democrats are: .13, .16, .15, .12,.16. For the Independents they are: .18, .19, .21..20,.27. For the Republicans they are: .16, .17, .17. .15, 22. 34