A Crisis of Language:

Framing of the Social Security Debate by Government and Media Actors

Taylor Ansley

POL 467: Political Manipulation of the News

Dr. Sellers

18 April 2005

I. Introduction

Congressman Bill Thomas (R-CA) captured the essence of framing by observing: “Semantics are very important. . . Don’t dismiss the use of a word” (Allen 3 2005). The language of politics--the frame within which debate of a given issue occurs--is the first battle in the war over Social Security reform. What was previously a White House plan for “private accounts” quickly became a push for “personal accounts” in light of revealing public opinion polls (Allen 3 2005). On the other side, Democrats face a struggle to downplay the idea of a Social Security “crisis:” terminology that Bill Clinton used in 1998 (Allen 3 2005). While it is too early to tell which side will win the policy debate, it is almost certain that the language battle will have far-ranging implications on both sides.

Following pressure by President Bush, the Republican leadership in both houses began the 109 th congress by presenting ten new pieces of legislation--starting with social security reform (Frist 2005). Senate Majority Leader Bill Frist (R-TN) described a Republican effort “to guarantee retirement security not only for this generation, but for those generations to come” (Frist 2005). That guarantee is part of the Republicans’ frame: social security is a program in danger. The doomsday attitude towards the current system stems from President Bush, who urgently insists “the crisis is here,” and that social security is “on the road to bankruptcy” (Allen 3 2005). The sense of immediacy surrounding the Republican agenda can be attributed to the political volatility of the Social Security issue. Facing opposition from high-profile groups such as the AARP, Republican leaders decided to pursue Social Security reform immediately so as to provide as much time as possible “to explain [new legislation] before facing voters in the midterm elections” (Allen 1 2005).

Democrats opened their own congressional agenda with Senate Minority Leader Harry Reid (D-NV) stating “We have leaders who love to create crises that don’t exist. Social Security isn’t a crisis. For more than 50 years, we’re going to be just fine” (Reid 2005). The notion of a manufactured crisis dominates the Democratic message. In an ironic role reversal, the Democrats are earning support, especially from seniors, by promising to protect the status quo. As Congressman Bob Menendez (D-NJ) explained: “It goes to the fundamental belief Democrats have that every American [should have] a baseline of support in this country. We’re not going to let the Republicans put that security in danger” ( Billings and Cillizza 2005).

Democrats seem to sense weakness in the Republican Party; one Democratic strategist called the Bush Social Security plan an “electoral loser for congressional Republicans” ( Billings and Cillizza 2005). Many Republicans appear hesitant to support President Bush’s proposed “profound but politically risky changes in Social Security” (Allen 2 2005). In an effort to build support within the GOP, Bush reportedly told House Republicans he would be “providing some cover on some of the most important but also the toughest political battles--Social Security being at the top of that list” (Allen 1 2005). Bush viewed the center-stage platform of his February 2 nd State of the Union address, and a subsequent tour of five states, as an opportunity to build popular support behind his proposals (Allen 2 2005).

Democratic strategists and polling experts agree that the minority party “ought to be playing offense” if they will stand a chance against President Bush’s bold agenda ( Billings and Cillizza 2005). Due to numerous mentions in President Bush’s State of the Union Address, and a multiple-state tour touting his plan, the idea of a “crisis” in Social Security is spreading. Cognitive linguist George Lakoff argues: “when [politicians] negate a frame, [they] evoke a frame” (Lakoff 2004, 3). That is to say, every time the Democrats argue “there is no crisis in Social Security,” they will merely be reinforcing the connotation of impending disaster in the minds of voters. For the Democrats to win the issue, they must frame the Social Security issue as a predicament caused by this administration’s tax cuts. For every time the Bush administration quotes Bill Clinton describing the “fiscal crisis in Social Security,” Democrats must answer with a quote from that same speech: “Before we spend a penny on new programs or tax cuts, we should save Social Security first” (Weisman 2005). That message, to “save Social Security first,” harkens back to Democratic campaign themes of a poorly-timed Bush tax cut and fiscal irresponsibility on the part of the current administration (Weisman 2005). To frame this issue successfully the Democrats must use the fear of change they are so often fighting and present Social Security as a troubled program that, if handled correctly, is remediable.

Despite the strong advantage of controlling majorities in both houses of Congress and the Executive Branch, Republicans face an uphill battle in trying to reform Social Security. To succeed with this issue, President Bush must clearly demonstrate to Republicans in Congress that their support of this proposal will not hurt their political futures. To ensure the safety of the GOP in the long run, Bush must do what Republicans have been so successful with in recent years: weave the issue of Social Security reform into the larger fabric of his message. This strategy was evident in Senator Frist’s press conference--the ten bills introduced by the Majority Leader framed a legislative agenda of “meaningful solutions” (Frist 2005). The Republicans gamble in some ways by tying the fate of their entire agenda to Social Security (the first of their ten bills); however, if their push for Social Security reform is successful, it might provide momentum for other larger policy goals. Whatever happens in the coming weeks and months, Social Security promises to be a front page issue, and realistically one of the major policy clashes of President Bush’s second term.

II. Literature Review

The overarching aim of this treatise is to examine the relationships that develop between reporters and the politicians they cover. We as citizens rely on an independent, even-handed treatment of the news by journalists; we count on an ambiguous truth that often cannot be quantified or proven. When journalists engage with their subjects, do they eliminate a healthy separation that ensures partiality? Is it reasonable to expect journalists to remain isolated from their stories? If reporters adopt the language and framing of their subject, is this a dangerous thing? What realistic expectations should we have regarding the relationships between the media and politicians, and how do those expectations measure up to reality?

In his action guide for progressives, Dr. George Lakoff briefly expounds on what he perceives to be the media’s obligation in political coverage. Lakoff says “it is the special duty of reporters to study framing and to learn to see through politically motivated frames, even if they have come to be accepted as everyday and commonplace” (Lakoff 2004, 51). Framing is the focus of a great deal of political science research, but most researchers only examine one aspect of framing: politicians’ choice of words to fit their issue positions into a larger frame. In the media-politician relationship--evident most explicitly in press conferences--the politician only represents half of the equation. That is to say, in our discussion of framing and the language of politics, we must recognize the implications of reporters and their acceptance or denial of issue framing. In investigating the extent to which reporters covering politicians adopt those politicians’ frames for their questions, I will first examine the existing literature describing the relationship between reporters and politicians.

While a variety of case studies--including the examination of press coverage during the Clinton presidency by Jacobs and Shapiro--exist linking public statements by politicians to press coverage, there is a void of study examining the language used by media actors beyond what is written in a newspaper or read on television (Jacobs and Shapiro 2000). Many political scientists examine correlations between the frames politicians use and the way their frames are reported in the news, but I feel that the language of reporters is in many ways as important as the language of their subjects. Lakoff argues that reporters should never “go along” with a given politician’s frame of an issue, saying “Reframing is everybody’s job. Especially reporters’” (Lakoff 2004, 50). Given Lakoff’s assertion that the questions asked by reporters often, as Cook says “overrepresent [sic] particular points of view,” it is necessary to examine the questions asked of politicians (Cook 1998, 93).

As part of an argument for “completeness and proportionality” within the media, Kovach and Rosenstiel make the following analogy: “Journalism is our modern cartography. It creates a map for citizens to navigate society” (2001, 164). In this way, do we as a society perhaps place undue trust in the media? Kovach and Rosenstiel argue that when reporters allow politicians to unilaterally dictate the issues they cover, journalists (to use the cartographer analogy) produce misleading maps of our society and current events.

Edwards and Wood support the idea that reporters focus almost exclusively on issues mentioned by politicians in a study that found few significant instances of media coverage “focusing the issue attention of Congress” (1999, 327). This assertion could potentially relate to the way reporters are assigned: reporters generally work the same “beat” (the White House or Pentagon, for example) for long periods of time. No existing study scrutinizes the effect of the inevitable relationships and sympathies acquired by news reporters who grow close to their politician subjects.

Politicians generally seek news coverage, and competition for air time is fierce. As Schaffner and Sellers explain: “While many members desire news coverage, few actually receive it. The national press corps must decide which of 535 legislators to use as sources for congressional stories” (1993, 3). The competitive nature of press-seeking politicians has implications on framing and the relationships between political and media actors. Budding relationships between reporters and the politicians they cover might encourage reliance on certain sources of information over others. Do members of congress who compete for news coverage avoid common frames of an issue so as to differentiate themselves from the other 534 legislators? In seeking a unique sound-clip or quote, do reporters avoid the more prevalent, consistent frames in favor of more unique perspectives on policy issues? Should reporters respond--by using the same language as politicians in their questions or stories--to issue frames consistent throughout a party or voting bloc, or should they investigate other views more completely? Alternatively, perhaps reporters should capture the overall state of Congress more simply by finding the common messages between members—thereby reporting the frames presented by politicians.

Given the relative detachment of broadcast media hosts who question high-profile political guests, it seems reasonable to assume that these media actors will be more likely to avoid frames. Parker supports this idea when she says of radio talk-show hosts: “they are probably more in touch with the American people and the everyday issues and problems they deal with than are those journalists who work and live inside the capital beltway” (Parker 1994, 164). Though broadcast hosts may hold more freedom to share their own opinions, they inevitably cater to ratings-boosting guests. When these guests appear, are we to believe that the host continues to fiercely advocate his or her own views, or will the host adopt their guest’s frame? Some authors describe the tendency of journalists to follow the story, even when that means holding their own views in check. Cook quotes Herbert Gans, who wrote of the journalist-source relationship: “although it takes two to tango . . . more often than not, sources do the leading” (Cook 1989, 30).

The questions I raise are significant because of the normative standards we hold for journalists. As Kovach and Rosenstiel explain, “A [journalist’s] commitment to citizens is more than professional egoism. It is the implied covenant with the public . . . that the coverage is not self-interested or slanted for friends” (Kovach and Rosenstiel 2001, 51). We expect reporters to detach themselves from their subjects and present an unbiased account that is beyond a given political frame. How accurate and realistic are these expectations? To investigate the validity of common journalistic standards we must uncover the extent to which the questions reporters ask are a product of the politicians they cover.

III. Statistical Tests

The routinely contested issue of partisan bias in the media applies significantly to this study. A February 17-19, 2003 Gallup poll found that while 47.69% of respondents believed the media “favor[s] one political party over the other,” 45.57% believed the media “favor[s] neither political party” (Gallup 2003). Regardless of perceived bias (or lack thereof) in the American media, our expectations of a free press remain the same: “Every journalist . . . must have a personal sense of ethics and responsibility” (Kovach and Rosenstiel 2001). In examining the responsibility we ask of members of the press, I will investigate the factors that influence media actors’ willingness to accept political frames of the social security issue.

For the purpose of this study, I limit the scope of reporters’ active participation to the questions they ask of politicians in press conferences. Examining the words reporters choose for their questions should be an excellent indicator of their immediate acceptance (or rejection) of frames as presented by politicians. Each question or statement in congressional and presidential press conferences (for information on data collection, see Appendix A) between January 1 st, 2005 and March 1 st, 2005 is a unit of observation for this study. Since this data measures a variety of factors relating to the politician-reporter relationship, there are multiple dependent variables. In cases where I examine the effect of various possible factors in the mentions of either party’s frame, for instance, I utilize multiple regression models to capture the effect on first the Democratic and then the Republican frame.

Regression Model Group 1

Explanation of Variables

The dependent variable for the first two models is the number of mentions of the phrase “Social Security.” This variable captures the number of times “Social Security” is mentioned by reporters and politicians; the first model estimates the values for all statements and the second model estimates the values for all questions. The third and fourth models expand the dependent variable by including the mentions of a party’s frame (defined by a set of words and phrases detailed in Appendix B). The third model examines mentions of the Republican frame when politicians mention at least one of the frame words or phrases and “Social Security.” The fourth model examines mentions of the Democratic frame when politicians mention at least one of the frame words or phrases and “Social Security.” Both the third and fourth models are limited to statements made by politicians.

The set of independent variables for each of the four regression models of the first group are identical, and I will describe each in turn. The first independent variable, Speaker Number, captures the order of statements and questions within a press conference. Beginning with the politician’s opening statement (Speaker Number=1), a number is assigned to each statement or question in increasing order. The models also include four variables relating directly to the demographic make-up of the politicians participating in the press conference. One variable measures the number of participating Democratic members of congress. Another variable measures the number of participating Republican members of congress (President Bush and White House Press Secretary Scott McClellan are included in the count of participating Republicans). Additional dummy variables differentiate press conferences featuring President Bush and those featuring White House Press Secretary Scott McClellan from all other conferences.

Hypotheses

The first and second regression models estimate the number of mentions of “Social Security” in politicians’ statements and reporters’ questions, respectively, at press conferences. If either the number of Democrats or the number of Republicans has a significant effect on the independent variable, we will see how partisanship affects the likelihood that politicians or reporters will mention the issue. Given the current political context, in which the Republican Senate leadership introduced a White House-backed bill, one might expect the Republicans to be more likely to mention Social Security. The nature of the Social Security debate, however, might alter these expectations. Since the introduction of the Senate bill, many Congressional Republicans have approached the topic with caution--unwilling to ignore concerns raised by powerful interest groups like the AARP (VandeHei 2005). The schism on this issue between the Republican leadership and other Republican Members of Congress seems to be uniting Democrats. While I predict that press conferences featuring the President and Scott McClellan will increase mentions of Social Security (given the President’s active campaign for Social Security reform), on the congressional side I expect Democrats to raise the issue more frequently. I expect reporters to mention Social Security in their questions, increasing mentions as the number of Democrats or Republicans increases. Since there are a wide variety of stances on this issue within each party, I do not expect the reporters to confine the discussion along party lines.

Because the issue of Social Security reform is spearheaded by the Republican Senate leadership, I predict that days the Senate is in session will have a positive effect on mentions of Social Security and partisan frames. Recent emphasis by House Republicans on the Social Security issue leads me to expect a significant relationship for days the House is in session and mentions of Social Security and partisan frames (Hallow 2005). It seems logical to assume that mentions of Social Security in the question immediately preceding a politician’s statement would encourage mentions of Social Security by the politician--some might argue that politicians never answer questions, but it seems as though they generally address the topic. For the reporters, I would expect that if the statement immediately preceding their question addresses Social Security, they might be more inclined to ask follow-up questions on the same topic--thus having a positive effect on mentions of Social Security. What will be most interesting to see, however, is which (if either) frame politicians favor when they choose to respond to a question. That is to say, when a reporter uses the language of a given party’s frame, does that frame make politicians more or less likely to respond and talk about Social Security? Without evidence supporting positive or negative correlation, I suspect--based on observation of the political actors involved--that reporters will be more likely to utilize the Republican frame in their questions, and politicians will be equally likely to respond to either frame. Without further evidence, I cannot predict what effect speaker number might have on mentions of Social Security.

I expect the third and fourth regression models, in their estimation of partisan frames, to produce significant coefficients for the number of participants from each party--a greater number of Democrats producing a greater number of mentions of the Democratic frame, for example (for the same reasons as outlined above). I also expect the variables for President Bush and Scott McClellan to produce significant positive coefficients with regards to mentions of the Republican frame (since the White House seemingly sets many terms of the Republican frame) and negative coefficients with regard to the Democratic frame. For the independent variables capturing mentions of the partisan frames and “Social Security” in the question immediately preceding a politician’s statement, I predict that the regression model will yield significant coefficients. I expect, for the dependent variable of Democratic frame mentions, to find positive correlation with mentions of both the Democratic frame and “Social Security” in the previous question. This supports my theory that politicians, when answering questions, adapt whatever frame reporters use. In this way, I envision mentions of the Republican frame in the preceding question to have a negative effect on mentions of the Democratic frame. For the dependent variable of Republican frame mentions, I expect to find similar relationships--positive coefficients for previous mentions of the Republican frame and “Social Security,” alongside a negative coefficient for previous mentions of the Democratic frame.

Data and Results

Regression 1.1

Of the ten independent variables incorporated into the first regression model, two have a significant effect on the dependent variable (mentions of Social Security in politicians’ statements). The first significant variable is the number of Democrats participating in the press conference. The predicted value for number of mentions of the Democratic frame leaps by more than two mentions (from approximately .15 to 2.5) as the number of Democrats changes from zero to seven (the maximum number in any of the press conferences constituting my sample).

(Table 1 about here)

Mentions of Social Security in the previous question also had a significant effect on mentions of Social Security by politicians. As mentions of Social Security in the previous question changes from 0 to 3, the mentions of Social Security by politicians grows by nearly two—from .29 to 2.18 mentions.

(Table 2 about here)

The existence of two highly significant independent variables like these has implications on broader questions, which will be discussed in detail later in the treatise.

Regression 1.2

The single significant independent variable in the second regression model is the number of mentions, in the statement immediately preceding the reporter’s question, of the Republican frame and Social Security. Though not as significant as the coefficients in regression 1.1, the predicted values calculated by varying this variable illustrate a noteworthy effect. When mentions, in the preceding statement, of the Republican frame increase from 0 to 23, the corresponding predicted value for mentions of the Republican frame in the reporters’ questions increases nearly eight times (from .11 to .87).

(Table 3 about here)

Regression 1.3

The third regression model, capturing mentions of the Republican frame by politicians, yields a single significant variable: participation by White House Press Secretary Scott McClellan. Press conferences involving McClellan included far more mentions of the Republican frame than other conferences. For McClellan’s press conferences, the predicted value for mentions of the Republican frame was nearly twice that of all other press conferences (1.6 mentions versus .79 mentions).

(Table 4 about here)

Regression 1.4

None of the ten independent variables included in the fourth regression model are significant.

Regression Model Group 2

Explanation of Variables

The second group consists of two models. The dependent variable for the first model captures mentions of the Democratic frame and “Social Security” by politicians in their opening statements. The second model captures all mentions of the Republican frame and “Social Security” in politicians’ opening statements. Both models utilize the same independent variables, which essentially consist of a shortened list of those used in the first group of models: number of participating Democrats, number of participating Republicans, President Bush, Scott McClellan, Senate in session, and House of Representatives in session. The two models of this group limit their focus to include only opening statements.

Hypotheses

For the first model, which estimates the mentions of the Democratic frame and “Social Security” in politicians’ opening statements, I expect the number of Democrats to have a significant positive effect on the dependent variable. As the number of Democrats speaking at a given press conference increases, we should expect to see what one recent article called the “unified front against the White House proposal on Social Security” in action (Haddix 2005). For the same reason, I expect to find that the number of participating Republicans has a negative effect on mentions of the Democratic frame. If the Democrats indeed have a “unified front” on the issue, I expect to find that on days when either body of congress is in session—that is to say, days when Democrats are faced with the legislative agenda of the majority Republicans—the number of mentions of the Democratic frame in opening statements will increase. Given the White House’s increasing “use of pointed language,” I suspect that the President and Scott McClellan will stick to the Republican frame, and that their participation in a press conference will have a negative effect on mentions of the Democratic frame (Loven 2005).

In the second regression model of this group, which estimates mentions of the Republican frame and “Social Security” in politicians’ opening statements, I expect to find a much more divided Republican party—fueled by concern over an “intense level of opposition to restructuring Social Security” (VandeHei 2005). Given division within the GOP, I predict that increasing numbers of participating Republicans will in fact have a negative effect on mentions of President Bush’s own themes and language. Participation by President Bush or Scott McClellan, on the other hand, will most likely have a highly positive effect on mentions of the Republican frame. The White House began the recent debate over Social Security reform, therefore I expect the messages of the chief spokesmen of the executive branch to echo their own themes and language. I expect the involvement of Democrats to have a negative effect on mentions of the Republican theme, as it is in the Democrats interest to, as Lakoff writes: “always reframe [the issue] to fit your values and your frames” (Lakoff 2004, 116).

For both models of this group, I expect activity in the Senate and House (that is to say, days when either body is in session) to have a positive effect on the dependent variable. When the parties clash in congress—as most certainly is the case with Social Security reform—members of congress seek media attention to further their side of an argument. The Republican leadership introduced a Senate bill in their first press conference of the new congress, and Representatives like Steve King (R-IA) recently emphasized urgency for “the House leadership to move a bill to the floor” on Social Security reform (Hallow 2005).

Data and Results

Regression 2.1

The single significant independent variable in the first regression model of this group is the number of Democrats participating in a given press conference. When the number of Democrats increases from 0 to 7, the corresponding predicted value for mentions of the Democratic frame in politicians’ opening statements increases by a whopping 17 mentions. Seven participating Democrats represent the maximum number I encountered in a single press conference. Even a more modest increase in the number of Democrats, from 0 to about 3 (one standard deviation above the mean) still increases the mentions of the Democratic frame by more than 7.

(Table 5 about here)

Regression 2.2

The number of participating Democrats and participating Republicans are both significant variables in this regression model, which measures factors influencing politicians’ choices of frames in their opening statements. A press conference with three Democratic speakers yields 4 more predicted mentions of the Republican frame (from .8 to 5) than a press conference with no Democrats. When the number of Republicans participating in a press conference moves from 0 to 2, the number of mentions of the Republican frame increases by only .6 (from 2.2 to 2.8 mentions). The contrast between parties is curious and will be explored in a more in depth way later in this paper.

(Table 6 about here)

Regression Model Group 3

Explanation of Variables

The final group of regression models includes two models estimating mentions of partisan frames in reporters’ questions. The dependent variable in the first model is mentions of the Democratic frame and “Social Security.” In the second model, the dependent variable is mentions of the Republican frame and “Social Security.” Both models are limited to reporters’ questions when Social Security is mentioned.

The models in this group incorporate many of the independent variables used in previous estimations. The numbers of politicians from both parties who participate in the press conference are taken into account, as is the involvement of President Bush or Scott McClellan. Whether or not each body of congress is in session on the day of a given press conference is included in the model. Like many of the other models, this group includes mentions of both partisan frames and social security in the previous statement, as well as in the politicians’ opening statements.

The models of this group, limiting the scope of inquiry to questions containing mentions of Social Security, nonetheless represent the broadest group of independent variables in my study. The first group of regression models included the independent variable of speaker number. That variable is not included here, though a new variant of speaker number is exclusive to this group: speaker number squared. The inclusion of a variable calculated by squaring the speaker number value allows for the estimation of speaker number values in a non-linear way.

Hypotheses

The regression models included in this group, in their estimation of partisan frames in reporters’ questions, will shed light on the willingness of journalists to adopt one frame or another, or to refuse both frames. I expect both models to follow party lines—that is to say, I expect an increase in the number of participating politicians from either party to increase the reporters’ mentions of their party’s frame. This expectation is based on the assumption that as members of a party increase, mentions of that party’s frame also increase (as investigated in the first group of regression models). If politicians do indeed use their party’s frame on a consistent basis, it could follow logically that reporters would adopt the language they hear most frequently within the context of a single press conference. However, compelling logic speaks for an alternative effect. If reporters, instead of adopting frames as presented by politicians, reject what McQuail calls “the unthinking deployment of wider interpretive frames and schemes,” we can expect to see a neutral or possibly negative correlation between the frames politicians use and the frames reporters, in turn, utilize (McQuail, 191).

Predicting the effects of independent variables capturing mentions of a given party’s frame (or “Social Security”) in politicians’ opening statements and statements immediately preceding a question relies on the same logic as above. If the number of mentions of a given frame increases, we might expect reporters to mention that frame more frequently. If we, on the other hand, believe that journalists consciously avoid the “unwitting bias” McQuail describes, we might predict that mentions (by politicians) of a given frame has a neutral or negative effect on reporters’ mentions of those frames.

The effect of “Social Security” mentions in opening statements, and statements preceding the reporters’ questions, will provide further insight regarding the propensity of reporters to accept or reject frames. It is important to emphasize that these regression models are limited to questions containing the phrase “Social Security.” The effect of this limitation is that any statement included in the variable for Social Security mentions in the preceding statement will be followed by a question also mentioning Social Security. By observing when reporters follow a statement on Social Security with a question on Social Security, we can observe what frames reporters respond to, and if their response contains a frame. I predict that reporters will respond equally to the partisan frames, and that mentions of Social Security in the statement preceding a reporter’s question will increase the likelihood of both frames appearing in the question.

The final independent variable in these models is the value of speaker number squared. This variable arises out of observation of the press conferences included in my data set. Based on cursory examination of the text, there appeared to be a trend of increasing mentions of social security as the press conferences wore on. The variable tests this observed effect, and measures the significance of an exponential correlation between speaker number and frame mentions. I predict, based on my own previous observations, that the variable will be highly significant, exerting a positive effect on mentions of both frames in reporters’ questions.

Data and Results

Regression 3.1

The first model of the third group, measuring the influence of a variety of factors on mentions of the Democratic frame by reporters, yields two significant variables. Mentions of the Democratic frame in the politicians’ opening statement has a significant positive relationship with mentions of the same frame in reporters’ questions. When mentions of the Democratic frame in the opening statement increases from 0 to 9, the reporters also increase mentions of the Democratic frame in their questions, more than quadrupling the value when no Democratic frame is mentioned (from .087 to .37). The number of mentions of the Republican frame in politicians’ opening statements has a significant and negative effect on journalists’ mentions of the Democratic frame. When politicians’ mentions of the Republican frame increased from 0 to 11, mentions of the Democratic frame by reporters immediately plummeted to zero.

Regression 3.2

The single significant variable of this regression model, examining the influences on journalists’ mentions of the Republican frame, is the value of speaker number squared. Speaker number squared exerts a heavily positive influence on mentions of the Republican frame in reporters’ questions. As speaker number increases, the growth in journalists’ use of Republican frames is exponential—exploding from zero to almost 259.

IV. Interpretation of Data

In regression model 1.1, I observed that the number of Democrats participating in a press conference has a positive effect on the number of mentions of Social Security by politicians. No such effect is significant on the Republican side. This leads me to conclude that Democrats as a group are more apt to discuss Social Security. This seems to be an issue Democrats feel they can win, and as a result they speak about Social Security as much as possible. Representative Dennis Kucinich (D-OH) summed the Democratic confidence, saying: “If the President pushes Social Security and privatization, Democrats will retake the House, the Senate and in a few years the presidency” (Haddix 2005).

The number of Democrats participating in a given press conference is also a significant influence on mentions of the Republican and Democratic frames in politicians’ opening statements. On the one hand, logic tells us that more participating Democrats will lead to more mentions of the Democratic frame in opening statements. The positive correlation between number of Democrats and mentions of the Republican frame, however, is curious and has many potential explanations. One explanation lies in what Lakoff describes as negating, and thus evoking and confirming, the opposition’s frames (Lakoff 2004). When Senate Minority Leader Harry Reid (D-NV) says “Social Security isn’t a crisis,” he immediately accepts the terms of the Republican side of the debate (Reid 2005). The other explanation lies in what DNC Chairman Howard Dean recently called a lack of “message discipline” in the Democratic Party (Finnegan 2005). For the Republicans’ part, the number of participating Republicans was a crucial factor in mentions of the Republican frame in opening statements. The Republicans do no have perfect message discipline—a negative coefficient for mentions of the Democratic frame as Republican participants increases would be more desirable for the Republican leadership—but they are certainly more exclusive to their party’s frames than the Democrats.

Though I expected to find a significant influence, the participation of President Bush in a press conference did not factor in any of the dependent variables. This could be because of the relatively few Bush press conferences included in the sample (four in all, none of which focused specifically on Social Security), and the dispersion of the White House message between the president and Scott McClellan. In contrast to Bush, McClellan proved to be a reliable voice for the Republican frame. Press conferences featuring the White House Press Secretary included significantly more mentions of the Republican frame, showing that McClellan is doing his job well and sticking to the Republican frame.

The variables for days congress was in session held no significance in any regression model. This speaks to the prevalence of the Social Security debate—politicians and reporters alike are addressing the issue on a daily basis. As the issue has yet to come before either house of congress in the form of a vote, the debate is not confined to Capitol Hill.

Mentions of both parties’ frames in politicians’ opening statements influenced mentions of the Democratic frames by reporters. While increasing mentions of the Democratic frames by politicians has a strong positive effect on mentions (by journalists) of that frame, the negative effect of mentions of the Republican frame is equally significant with a stronger coefficient. This evidence posits a preference in journalists for the Democratic frame. When presented with the Democratic frame, reporters respond on the same terms. The negative effect of Republican frame mentions might lead us to believe that Reporters treat each frame equally. As regression model 3.2 demonstrates, however, reporters faced with the Republican frame do not echo that frame . . . they simply ignore both frames.

The value of speaker number squared is a highly significant positive influence on journalists’ mentions of the Republican frame. This confirms what I observed while skimming briefly through selected press conferences: a trend of Social Security appearing toward the end of question and answer sessions. Why reporters late in a press conference adopt the Republican frame so consistently is a mystery, and seemingly contradicts journalists’ preference for the Democratic frame as shown earlier.

In the model estimating influences on Social Security mentions by politicians (1.1), mentions of Social Security in the preceding question was significant. This result was to be expected, since we generally expect politicians answering media questions to address the topic. In the following model (1.2), the mentions of Republican frames in the preceding statement has a significant effect on mentions of Social Security in journalists’ questions. This result can be interpreted in a variety of ways. Do the Republican frames simply inspire further questioning? Perhaps, though the meaning of this result is very ambiguous given the conflicting results of other models.

Many other independent variables in my regression models were insignificant. In lieu of further detail (I have explained those insignificant variables I found most surprising and important), I offer potential problems with the data set, my data collection, and the models themselves that could have encouraged the insignificant variables. My data set was limited by time (two months, in the context of an important debate like Social Security reform, is merely a glimpse) and relatively few relevant press conferences. A number of different variations on the models I constructed—limiting one factor or another—might yield more significant results, but limited resources forced me to select the models that appear here.

V. Conclusions

The need for constant vigilance of the relationship between government and media actors cannot be over-emphasized. We cannot simply hope or assume that all journalists strive to, in the words of Kovach and Rosenstiel, “provide citizens with the information they need to be free and self governing” (2001, 17). In fact, our responsibility as citizens in a free society is to constantly question leaders and journalists. In the models presented here, I observed startling trends that--if expanded in a more complete context—could produce frightening truths about the media-government relationship. If reporters are more likely, in the context of the Social Security debate, to echo the Democratic frame (as my study demonstrates), that trend has far ranging implications on politics and the way journalists report political discourse.

To ensure that much of this analysis is not descriptive of the Social Security debate specifically, more study is needed. Such study could expand the models used here, using data collected over the course of a year or more. More comprehensive research could examine a breadth of issues—gay marriage, partial birth abortion, gun control, and judicial appointments to name a few—so as to produce a more complete picture of the media-government relationship. Improvements on this work could include the addition of new variables such as press conference participation by party leaders, or presidential support scores, seniority and re-election hopes of participants.

This research, though raw, is relevant and crucial to understanding the reality of political journalism and public policy. If journalists adopt partisan frames, and politicians are aware of it, we face a normatively terrible situation as citizens. Politicians who know with confidence that their frames will resonate in newspapers and television broadcasts will not be challenged to express their view in new ways or to address the opposition’s concerns. The type of dialogue such an awareness could create would be one of polarized views and would undoubtedly hurt public awareness and discourse of policy issues.

Thus we face a “crisis of language;” a collision of truth and spin that threatens the objectivity and validity of our news. In seeking more balanced, truthful, and objective coverage of divisive issues, we have the duty to question the methods and tendencies of journalists. As Kovach and Rosenstiel proclaim: “We as citizens have an obligation to approach the news with open minds, willing to accept new facts and examine new points of view as they are presented . . . the elements of journalism are principles that apply to citizens as well . . . and citizens have to do their part to sustain the relationship” (2001, 195).

Appendix A:

Detailed description of Data Collection

I collected the press conferences used as data for this study using the Lexis-Nexis search engine. Searching all FDCH Political Transcripts (from January 1, 2005 through March 1, 2005), my search terms were as follows: ((media w/5 availability) or (news w/5 conference)) and (Senator or Representative or Bush or McClellan) and (Social w/3 Security). This search produced a list of matches, which I screened for erroneous entries. I eliminated false matches (a Ralph Nader press conference, for example) and combined the remaining press conferences into a Microsoft Word document. Using text search utilities, I constructed a list of relevant speakers (Members of Congress, President Bush, and Scott McClellan) and began, with the help of Dr. Patrick Sellers, the construction of a program to separate the press conference files by question and statement. This program, in turn, became the avenue through which I created my variables and designed the regression models for this study.

Appendix B:

Detailed description of Independent Variables

Variable FrameDem, measuring mentions of the Democratic frame:

• This variable measures the number of mentions, in a given press conference (or, when defined more specifically, question or statement), of certain words and phrases defined to be the Democratic Party’s frame of the Social Security debate. These words and phrases, selected by careful consideration of the Democratic National Committee’s published platform, press releases from Democrats in Congress, and confirmed by textual searches of the press conferences used as data for this study, are as follows: "gambling," "gamble," "unnecessary," "risk," "scheme," "privatization," "private accounts," "debt," "deficit," "cutting benefits," "benefit cuts," "loss of benefits," "dismantle," "irresponsible," and "destroy social security."

Variable FrameRep, measuring mentions of the Republican frame:

• This variable measures the number of mentions, in a given press conference (or, when defined more specifically, question or statement), of certain words and phrases defined to be the Republican Party’s frame of the Social Security debate. These words and phrases, selected by careful consideration of the Republican National Committee’s published platform, press releases from Republicans in Congress, President Bush, and the White House Press Secretary (and confirmed by textual searches of the press conferences used as data for this study), are as follows: "personal accounts," "personal retirement accounts," "voluntary," "crisis," "future generation," "next generation," "future generations," "save, " "secure," "modernize," "modernization," "strengthen," "bankruptcy," "bankrupt," "problem with social security," "problems with social security," "collapse," "urgency," "urgent," and "nest egg."

Variable FrameSocDem, measuring mentions of the Democratic frame and Social Security

• This variable measures the number of mentions, in a given press conference (or, when defined more specifically, question or statement), of the Democratic Frame (defined above), limiting the sample to cases where “Social Security” appears in the same interval. This variable represents a measure of the mentions of the Democratic frame that is less susceptible to the flaw of unrelated terms (“gamble” used to describe something other than the Bush plan for Social Security, for example).

Variable FrameSocRep, measuring mentions of the Republican frame and Social Security:

• This variable measures the number of mentions, in a given press conference (or, when defined more specifically, question or statement), of the Republican Frame (defined above), limiting the sample to cases where “Social Security” appears in the same interval. This variable represents a measure of the mentions of the Republican frame that is less susceptible to the flaw of unrelated terms (“secure” used to describe something other than the Bush plan for Social Security, for example).

Variable FrameSocDemO, measuring value of the FrameSocDem variable in a politician’s opening statement:

• This variable represents the value of FrameSocDem in reference to the politician’s opening statement. Since the opening statement represents the issues a politician chooses to address—and the language he or she chooses to use—it is important to examine what factors might influence their likelihood to discuss Social Security and, when they address the issue, their likelihood to use the Democratic frame.

Variable FrameSocRepO, measuring the value of the FrameSocRep variable in a politician’s opening statement:

• This variable represents the value of FrameSocRep in reference to the politician’s opening statement. Since the opening statement represents the issues a politician chooses to address—and the language he or she chooses to use—it is important to examine what factors might influence their likelihood to discuss Social Security and, when they address the issue, their likelihood to use the Republican frame.

Appendix C:

Tables and Predicted Values

Regression Model 1.1: Influences on Mentions of Social Security by Politicians

Table 1:

Table 2:

Regression Model 1.2: Influences on Mentions of Social Security by Reporters

Table 3:

Regression Model 1.3: Influences on Mentions of the Republican Frame by Politicians

Table 4:

Regression Model 1.4: Influences on Mentions of the Democratic Frame by Politicians

Regression Model 2.1: Influences on Mentions of Democratic Frame in Politicians’ Opening Statements

Table 5:

Regression Model 2.2: Influences on Mentions of Republican Frame in Politicians’ Opening Statements

Table 6:

Regression Model 3.1: Influences on Mentions of the Democratic Frame by Reporters

Table 7:

Regression Model 3.2: Influences on Mentions of the Republican Frame by Reporters

Table 8:

Works Cited:

Allen 1, Mike. “Bush Pledges to Take the Lead on Social Security Changes.” The

Washington Post. 29 Jan., 2005. A08.

Allen 2, Mike. “Bush Seeks to Allay GOP Doubts; Lawmakers Urged to Be Patient on

Social Security Changes.” The Washington Post. 26 Jan., 2005. A04.

Allen 3, Mike. “Semantics Shape Social Security Debate; Democrats Assail ‘Crisis’

While GOP Gives ‘Privatization’ a ‘Personal’ Twist.” The Washington Post. 23 Jan., 2005. A04.

Billings, Erin P. and Chris Cillizza. “Democrats to Talk Up Social Security.” Roll Call.

27 Jan. 2005.

Cappella, Joseph N. and Kathleen Hall Jamieson. “News Frames, Political Cynicism,

and Media Cynicism.” Annals of the American Academy of Political and Social

Science. Vol. 546: Jul. 1996, 71-84.

Cook, Timothy E. Governing With The News: The News Media As A Political

Institution. Chicago: University of Chicago Press, 1998.

Cook, Timothy E. Making Laws & Making News. Washington, DC: The Brookings

Institution, 1989.

Edwards, George C. III and B. Dan Wood. “Who Influences Whom? The President,

Congress, and the Media.” The American Political Science Review. Vol. 93, No.

2: Jun. 1999, 327-344.

Finnegan, Michael. “Dean Says Democrats Will Make Schiavo Case an Election

Issue…” Los Angeles Times. 16 Apr., 2005.

Frist, Bill: U.S. Senator. “News Conference on the Republicans’ Legislative Agenda.”

24 Jan., 2005.

Gallup Poll. February 17-19, 2003. Sample Size: 1,002. Accessed 13 Feb. 2005.

<http://institution.gallup.com/documents/question.aspx?question=143681&save=1&ref=http://institution.gallup.com/search/results.aspx?SearchTypeAll=political%20media|SearchConType=1>.

Haddix, Dar. “Congress Jeers and Cheers Bush Econ Plans.” UPI. 3 Feb., 2005.

Washington Edition.

Hallow, Ralph Z. “House Urged to Take Lead on Social Security.” The Washington

Times. 14 Apr., 2005.

Jacobs, Lawrence R. and Robert Y. Shapiro. Politicians Don’t Pander: Political

Manipulation and the Loss of Democratic Responsiveness. Chicago: University

of Chicago Press, 2000.

Kovach, Bill and Tom Rosenstiel. The Elements of Journalism: What Newspeople

Should Know and the Public Should Expect. New York: Three Rivers, 2001.

Lakoff, George. Don’t Think of an Elephant! Know Your Values and Frame the Debate.

White River Junction, VT: Chelsea Green, 2004.

Loven, Jennifer. “Bush Chides Lawmakers Over Social Security.” Washington Dateline.

31 Mar., 2005.

McQuail, Denis. POL 467 Required Reading.

Parker, Kimberly Coursen. “How the Press Views Congress.” Congress, the Press, and

the Public. Ed: Thomas E. Mann and Norman J. Ornstein. Washington, DC:

American Enterprise Institute for Public Policy Research and The Brookings Institution, 1994. 157-170.

Reid, Harry: U.S. Senate Minority Leader. “News Conference on the Democrats’

Legislative Agenda.” 24 Jan., 2005.

Schaffner, Brian F. and Patrick J. Sellers. “Crafting Coverage in the U.S. Senate.”

February 23, 1993.

VandeHei, Jim. “Social Security Plan Meets Doubt in Iowa; GOP Lawmakers: Approach

Isn’t Selling.” The Washington Post. 31, Mar., 2005.

Weisman, Jonathan. “New Strategy on Social Security; With Some Risk, Bush Officials

Invoke Clinton, Moynihan.” The Washington Post. 24 Jan., 2005. A03