Framing the Social Security Debate

Why Politicians Speak Out and How They Influence the Media

By David Crow

April 18, 2005

I. Introduction

At the start of the second George W. Bush administration, Social Security reform looks set to become a defining issue for both Republicans and Democrats. The president and some congressional Republicans feel that reforming the Social Security system is an important step in creating an “ownership society,” while Democrats feel that keeping the Social Security system largely intact in its present form is not just an important issue, but also an important battle to win against the president in order to show that the Democratic party is still strong and viable despite the loses in the 2004 election (Stolberg 2005). In order to convince the American people that their side is right, many senators on both sides of the issue have been speaking out publicly on this issue. However, many other senators have remained silent about their stance on Social Security, despite the issue’s clear importance to both parties. This paper seeks to answer the question of what factors cause certain senators to speak out about Social Security and what factors cause senators to remain silent, as well as asking which side is more effective at getting their views represented in the news media?

In order to answer these questions I have divided this stufy into two parts. The first part deals with the question of what factors cause politicians to speak out about Social Security reform or to remain silent? The second part deals with whether or not the news media is picking up one side’s argument more than another’s.

First, I will look at the relationship of the media to the politicians, since so much of public opinion is determined by how the media picks up what politicians are saying. The literature on these issues focuses largely on the idea of framing a particular story or point of view. Therefore, I will explore the idea of framing from both the politicians’ perspective and from the media’s perspective in order to determine what causes the media to pick up one side’s frame of an issue rather than an alternate frame of that issue. I will outline several existing studies that explore the issue of how framing and what factors determine how frames are created.

Next, I will introduce my own framework for studying this issue based on some of the existing questions still left unanswered by the existing literature. Specifically, I will look at how party leadership, an individual senator’s support for President Bush, the percentage of a senator’s state’s population that is at the retirement age, and whether or not a senator is a potential presidential candidate in 2008 affect how they chose to frame this debate. After going through the methodology behind how I chose these variables, I will present present the data that I found and analyze that data to explain why some of these factors were important in determining how senators frame this issue while other factors were not important. Finally, I will conclude by explaining how what implications this research may have for future studies of politicians and media framing.

II. Literature Review

            The existing body of political science research provides several theories concerning how politicians influence the news media. Particularly in the wake of 9/11, and the subsequent build up to the War in Iraq, political scientists have examined how effective the president is at framing news stories in the media. The George W. Bush administration successfully framed the War in Iraq as being in the best interest of the American people by focusing on the threat of weapons of mass destruction and the liberation of the Iraqi people. Unlike under previous administrations, the media’s coverage of the build up to the War in Iraq following 9/11 shows how the president has become much more powerful and influential over the media, and that the media rarely questions what they are told by the executive branch (Heldman and Carter 2004).

Similarly, by examining front page news stories appearing in the New York Times and the Washington Post during the years 1981, 1993, and 2001, it is revealed that there is much more press coverage of the executive branch than the legislative branch in the news media. Furthermore, what press coverage there is of the legislative branch tends to be more negative than coverage of the executive branch (Fransworth and Lichter 2004). Given the prominent role President Bush has given the Social Security issue this research suggests that the news media will follow the framing that the President presents and will largely ignore the message provided by the Democrats in the senate.

There are, however, several questions left unanswered here. First, the existing literature has so far focused on the ability of the president to influence the media regarding the questions of whether or not to go to war. Domestic issues such as Social Security are very different from international issues, and particularly different from issues relating directly to military action where the president, as the Commander in Chief, has traditionally been given more direct authority than he has on domestic issue.

In the past, senators have been able to influence issues through concentrated efforts at getting their message across in the news media and thereby influencing public opinion on certain issues (Sellers 2000). However, it is unclear whether or not the same power of the legislature will hold true on an issue as prominent as Social Security reform, where the president has used the vast majority of his political capital. It is also unclear from these previous studies just what influences individual senators to either join with their party’s message or not. I will seek to answer these questions by using a similar methodology established by the previous studies. I will examine framing words used by both parties to see how successful each party is at getting those frames in news articles.

In establishing how to come up with specific frames for each party it is important to understand the most successful ways politicians frame debates, as well as where politicians make mistakes in framing debates. Among the most significant errors politicians make in framing debates is to try and refute the claims made by the opposition. In doing so, politicians almost always end up accepting the frames of their opponents, and thereby make their own messages less effective (Lakoff 2004). Therefore, in looking at the frames offered by politicians I will look just as closely at number of times a senator uses one of their opponents frames as when they use one of their own party’s frames since using the other side’s frames, even if it is to refute a claim, tends to increase public awareness of the other side’s argument.

             Another factor that can determine how much press coverage an issue receives is based on who specifically is talking about that issue. Previous research on presidential candidates during the 1996 Republican presidential primaries shows that the content of a message is less important than who the message comes from. Those candidates who were leading the field in the 1996 primaries almost always recieved more attention from the press than less well known or less successful candidates, regardless of what the content of the message may have been (Flowers, Haynes and Crespin 2003). Clearly, who you are is more important than what you are saying during a presidential race. However, it remains unclear if this pattern holds true for senators when they are fighting over policy rather than fighting a primary battle. Therefore, I will examine whether or not a particular senator is a member of their party’s leadership determines how they frame the Social Security debate. Also, given the prominent attention paid to presidential candidates in the existing literature, I will also examine the field of senators who are potential presidential candidates in 2008 to see if their framing of this issue differs from that of other senators.

            A similar theory on how issues are framed in the media is known as the “consensus theory.” This theory argues that the less political consensus there is on an issue the more likely the press is to insert drama and bias into that issue (Niven 2004). Similarly, when there is either inter-party conflict on an issue, or when there is praise from the opposing side, an issue will get much more press coverage than when there is little conflict (Groeling 2004). Given the significant disagreement over Social Security reform, both between the two parties and among certain individuals within the parties, these theories would suggest a significant amount of press coverage would be given over to this debate. With the Republicans fairly divided on Social Security and the Democrats seemingly more united on the issue, it will be interesting to see if the Republicans receive more press coverage than the Democrats because the press likes to cover conflict? By examining this question in my study I hope to use the example of Social Security to illustrate these theories.

             Like the “consensus theory,” the “paradox of objectivity theory” deals with what factors influence the press in their coverage of political stories. This theory argues that media coverage of the senate tends to be balanced until a major political change takes place, such as one party sweeping into office. After a major political change, media coverage tends to be less balanced as the party that lost becomes afraid to speak out against the perceived popularity of the party that won, creating imbalance in news coverage in favor of the winning party (Kuklinski and Sigelman 1992). However, there is a problem in this study since it uses no measure of the minority party’s press releases or speeches that are released in order to influence the media. By looking at how the Democrats try to influence press coverage through the use of press releases and speeches on the senate floor I hope to fill in this missing peace of this theory and see if the large Republican victory in the 2004 election had the effects of a “major political change” by causing the Democrats to stay silent out of fear of offending those who voted for their opponents.

            Once it is established how politicians influence the media, it is important to also look at how accurate the picture painted by the media of events truly is. In 1992 the media presented a very negative image of the economic conditions, portraying the economy as much worse than it really was. Bill Clinton took full advantage of this inaccurate portrayal causing George H. W. Bush to lose reelection (Hetherington 1996). With so much debate over whether or not Social Security is in a crisis or not, it will be interesting to assess how the media presents the Social Security debate. While there is no clear consensus on just how much of a crisis Social Security is in, or if it is even in a crisis at all, it is possible to see if the media is portraying Social Security as in crisis by looking at whether or not the media picks up more of the Democratic frames which claim there is no crisis or more of the Republican frames which claim there is a crisis.

III. Hypotheses

            This study seeks to test what factors make certain senators speak out in the debate over Social Security and what factors make senators stay silent on the issue. The study also examines how successful senators are at framing the Social Security debate in the news media. Since Social Security is a topic that has risen to the very top of the issues agenda it is assumed that every senator would have something to say on this topic since many constituents would likely want to know just where their senator stands. However, not every senator is speaking out or even taking a clear position on this issue. I hope to understand what factors cause a senator to either speak out or remain silent on such an important issue.

            In order to pick what factors might influence a particular senator to either speak out or to stay silent in this debate I first looked to the top party leadership since traditionally it is the leadership that shapes the agenda and works to push legislation through. Given the leadership positions these individuals hold I believe that they would be more likely to speak out on the Social Security issue than would most senators since it is traditionally the party leadership’s role to set the message for their party and to be the leading speakers on important issues.

            The next factor I will look at to determine if it affects whether or not a particular senator speaks out about Social Security is their level of support for President Bush. Bush has clearly placed Social Security reform at the top of his priority list for his second term, and thus staked a great amount of political capital on getting it past. The prominent role Social Security reform played in his 2005 State of the Union address, as well as the dozens of speeches Bush has made on Social Security in his multi-state tour around the country following the State of the Union is evidence of just how much he cares about this issue. Therefore, I expect that those senators who support President Bush on most issues would be more likely to speak out in support of Social Security reform and those senators who do not support Bush on most issues would be more likely to speak out against Bush’s Social Security reform plans.

            Another important factor that tends to determine how politicians vote is the demographic makeup of their constituencies. In the debate over Social Security reform polls suggest that the public opinion on this issue is strongly correlated to age, with those citizens who are older being more likely to say that they think the system is not in a crisis and that they are opposed to privatizing Social Security. Specifically, a recent ABC News/Washington Post poll shows that 62% of Americans who are 65 or older say they are opposed to privatizing Social Security, compared to just 55% of all Americans who are opposed to privatization (Polling Report 2005). Furthermore, opposition to privatizing Social Security increases with each age group in the poll. This data suggests that the older one is the less likely they are to support privatizing Social Security. Since many politicians follow a delegate model of representation in which they vote largely based on how their constituents feel on a particular issue I believe that many senators will base their level of support for Social Security reform based on the demographic makeup of their state (Pitkin 1967). I expect that those senators whose states have a higher proportion of their population who are age 65 or older will be more likely to oppose privatizing Social Security while those senators whose states have a lower proportion of their population age 65 or older will be more likely to support privatizing Social Security.

            The final factor I will look at to determine whether or not it affects how a senator is speaking out about the Social Security debate will be whether or not that senator is a potential presidential candidate in 2008. Given the high profile nature of this issue it seems likely that potential presidential candidates would want to use the Social Security debate to gain support. This issue would seem like an ideal opportunity for senators to make their name better known to the general public and to take a firm stands on this issue. Therefore, I believe that those senators who are potential presidential candidates in 2008 would be more likely to speak out on this issue than other senators.

            In addition to examining what factors influence individual senators to either speak out or stay silent on this issue, this study also seeks to examine whether or not one party is more effective than the other party at getting their framing of the issue represented in the news media. By examining news articles dealing with Social Security reform in the Washington Post and the Washington Times and seeing whether or not the articles mention Republican or Democratic frames it will be possible to see which side is more effective at getting their views portrayed in the media and whether or not different media outlets are biased in favor of one side or the other. Past studies on this topic have shown that the liberal media outlets tend to pick up the frames offered by Democrats while conservative media outlets tend to pick up the frames offered by Republicans (Sellers 2000). Given the sharply divided nature of the Social Security debate I believe that similar biases will be present in the news media on this topic and that the more conservative Washinton Times will be more likely to pick up the Republican frames and more liberal Washington Post will be more likely to pick up the Democratic frames. In addition, given the seemingly more unified message offered by the Democrats I predict that the Democratic frames will be more likely than the Republican frames to be picked up by both conservative and liberal newspapers.

IV. Data

Dependent Variables:

            For the first part of this study that examines what influences senators to either speak out or stay silent on this issue, I first had to determine a specific set of words or phrases for each side that are used to frame the Social Security debate. In order to determine what those words or phrases are my group and I looked through the text and transcripts of press releases and floor statements made by various senators about Social Security. Those words and phrases that appeared over and over in press releases and speeches by many different senators from the same party were deemed the best codewords to use. These words were then compiled together into a list of ten words or phrases for each side. For the Republicans this list of codewords was, “crises,” “save,” “secure,” “voluntary personal accounts,” “modernize,” “bankruptcy,” “problem with Social Security,” “strengthen,” “future/next generations,” and “nest egg.” For the Democrats the list of codewords was, “gambling,” “unnecessary,” “risk,” “scheme,” “privatization,” “debt,” “deficit,” “cutting benefits,” “dismantle,” and “irresponsible.” While this is certainly not a complete list of all of the key words and phrases used in order to frame the Social Security debate, these lists do represent how each side is trying to frame the debate. The fact that we found so many of these words and phrases continuously reappearing in press releases and floor speeches shows how they represent the point of view each side was trying to get across.

            Once the list of framing words was determined, we then set out to collect all the press releases and floor statements given by all 100 senators between the period of January 1, 2005 and February 28, 2005. This two month time period covered the period both leading up to President Bush’s State of the Union address, in which Bush focused significant attention on Social Security reform, and the period following the State of the Union during which Bush traveled around to key states and gave speeches about the need for Social Security reform in order to increase public support for his initiatives. While this time period does not cover the entire battle over Social Security reform, and in fact debate continues to this day, it does provide for a fairly accurate sample of statements made by senators on this issue.

            We chose to look at both press releases and floor statements in order to come up with the most accurate picture of which senators were speaking out on this issue. We looked on every senator’s website in order to access their press releases. However, for a few senators their websites were either not updated to include the period from January 1 to February 28, or it was not possible to access press releases going back to the start of this period. Since many senators will make floor statements that coincide with press releases, we also looked at floor statements made by all 100 senators during this same period in order to compensate for any missing press releases and to get the most accurate picture of who was speaking out. We used the Lexis Nexis Congressional database in order to access the floor speeches for all senators during this period relating to Social Security. Once we had compiled the press releases and floor statements we conducted word searches to search for both the Democratic and the Republican frames in each speech and press release. In conducting these searches we also used variations on each frame in order to capture all of the frames used. For instance, we included both “gamble” and “gambling” as the same frame.

            After coding all of the press releases and press conferences for these words or phrases, our dependent variable was the number of mentioned of coded frame words or phrases in a press release or floor statement over the stated period. We also included anytime a chosen frame was used in the negative (such as someone saying “Social Security is not in a crisis”) since anytime one uses a frame, even if it is used in the negative, one is accepting that frame (Lakoff 2004).

            In total, we found 123 individual press releases and floor statements were issued by 52 of the 100 senators during this two month period. Of those 52 senators, many issued just one press release or made just floor statement about Social Security. However, several senators accounted for multiple statements. The most outspoken senators on this issue were senators Harry Reid and Charles Schumer, who each had 15 speeches/floor statements. Many of the press releases and floor statements only mentioned a few of the frames, but several senators did use the framing words over and over in the same release. For instance, Senator Barbara Boxer made one floor statement in which she used 60 Democratic frames and 44 Republican frames.

            For the second part of this study, examining how successful both the Democrats and the Republicans were at getting their particular frames in the news media, we used two separate media outlets, the Washington Post and the Washington Times. Since the Washington Post is generally regarded as more liberal, and the Washington Times is generally regarded as more conservative, these two newspapers were chosen in order to give the most balanced picture of the press coverage possible. This methodology is in line with previous similar studies (Sellers 2000).

            The dependent variables for this part of the study were the number articles mentioning Democratic frames and the number of articles mentioning Republican frames that appeared in the Washington Post and the Washington Times between January 1, 2005 and February 28, 2005. In order to find all of the articles that mentioned these frames we conducted a Lexis Nexis search for articles in these two newspapers that mentioned at least one of the frame words within 25 words of the phrase “Social Security.” We found in the Washington Post there were 154 articles that mentioned at least one Democratic frame and 96 articles that mentioned at least one Republican frame. In the Washington Times we found 92 articles that mentioned at least one Democratic frame and 59 articles that mentioned at least one Republican frame.

Independent Variables:

            The first independent variable I used was whether or not a senator was a member of the party leadership. On the Democratic side the leadership includes Minority Leader Harry Reid of Nevada, the Minority Whip Richard Durbin of Illinois, and the Chairman of the Democratic Policy Committee Byron Dorgan of North Dakota. On the Republican side the leadership includes Majority Leader Bill Frist of Tennessee, the Majority Whip Mitch McConnell of Kentucky, the President Pro Tempore of the senate Ted Stevens of Alaska, and the Chairman of the Republican Policy Committee Jon Kyl of Arizona. This list was compiled by information provided by looking at the websites of each party.

            The next independent variable I used was whether or not a senator’s level of support for President Bush affects how often they use the frames adopted by their party. To determine each senator’s level of support for the president I used Congressional Quarterly’s rankings of presidential support. These rankings are compiled annually and list the percentage of votes on which each senator voted in accordance with the president’s stand. These numbers were available for all senators except the nine new senators who had no previous senate voting records. Therefore, those nine new senators were not included as part of this variable.

            Of the 91 senators who did have a presidential support score the average level of support was 74.34 with a standard deviation of 16.545, meaning that on average senators voted with the president 74% of the time. Four senators, Saxby Chambliss, John Ensign, Jon Kyl and Rick Santorum, had perfect 100 presidential support scores. The lowest presidential support score was a 50, received by Senator John Kerry. Among all Democrats the average level of support for the president was 61.88, while among all Republicans the average level of support for Bush was 93.08.

            Another independent variable I examined was the percentage of each senator’s home state that had a population age 65 or over. I used data compiled by the U.S. Census Bureau in order to determine the percentage of each state’s population that was age 65 or older. For all states, the average percentage of the population age 65 or older was 12.488% with a standard deviation of 2.009. Democratic senators, on average, were from states with a slightly higher percentage of citizens age 65 or over, with an average of 12.858% of Democratic senator’s home states 65 or older compared to 12.185% for Republican senator’s home states.

            The last independent variable I looked at was whether or not a senator was a potential presidential candidate in 2008. Since no senator has officially declared themselves a candidate for the presidency in 2008 determining which senators are likely to run was difficult. In order to ensure the most accurate and unbiased list of potential candidates I used several sources in order to compile the list. I looked at a collection of several different polls conducted by major polling companies that asked respondents to list their support for potential 2008 presidential candidates. Those senators who were used in at least three separate polls by different polling companies were considered as potential 2008 presidential candidates in my study. The final list of potential candidates that I used consisted of Senators George Allen, Evan Bayh, Joe Biden, Hillary Clinton, Russell Feingold, Bill Frist, Chuck Hagel, John Kerry, John McCain and Rick Santorum.

            For the independent variables in the second part of the study dealing with Democratic and Republican frames used in the Washington Post and the Washington Times I used four different independent variables: Democratic frames mentioned by Republicans, Republican frames mentioned by Republicans, Democratic frames mentioned by Democrats and Republican frames mentioned by Democrats. The values for these variables were obtained from the same data as used for the Dependent variables in the first part of the study.

V. Results

Regressions 1 and 2:

            In order to test these hypotheses I ran a total of six regression models in order to determine whether or not any of the independent variables were significant. The first two regression models dealt with the first part of this study, examining what factors influence senators to speak out on Social Security reform or to stay silent on the issue. The first regression used the Democratic frames as the dependent variable and the other second regression used the Republican frames as the Dependent variable. Other than the difference in dependent variables both of these regression models were identical and used the same four independent variables: Whether or not a senator is a member of the party leadership, each senators support level for President Bush, the percentage of a senator’s state that is age 65 or older and whether or not a senator is a potential presidential candidate in 2008.

Table 1 presents the regression coefficients and statistical significance of each independent variable in the first regression. In this regression, the R 2 value was 0.146, showing that the overall model explains 14.6% of variation in the dependent variable.

Table 1 about here

We can also see from this regression that there is only one significant variable, a senator’s support level for President Bush. This variable had a P value of .002 and a coefficient of -.8588. These numbers show how as the support level for President Bush increases, the number of Democratic frames used by a senator decreases. This finding is in line with the expectations of my hypothesis.

The results of the second regression are shown in Table 2. This regression model had an even smaller R 2 value of .0939, showing that the overall model explains 9.39% of variation in the dependent variable.

Table 2 about here

Like in the first regression, the support level for President Bush is also the only significant independent variable. Here, this variable had a P value of .033 and a coefficient of       -1416. While it was expected that the level of presidential support would have a negative relationship to the number of Democratic frames mentioned, it was unexpected that this same variable would have a negative relationship to the number of Republican frames mentioned. In examining the predicted values of this variable in the first two regression models we can see the actual influence this variable has on the dependent variable.

Tables 3 and 4 about here

As level of support for President Bush increases from 62% to 95% the number of mentions of Democratic frames decreases from about 26 to essentially 0. Similarly, as the support for President Bush increases from 62% to 95% the number of mentions of Republican frames decreases from about 7 to around 3.

In explaining the results of the first regression model it is likely that the lower a senator’s support of President Bush, the more likely they are to be either a Democrat or the more likely they are to be a Republican who is opposed to Bush’s Social Security plan. It makes perfect sense that Democrats would tend to use more Democratic frames, and so as the level of support for Bush falls, the more likely a senator is to use more Democratic frames.

The results of the second regression model are harder to explain. According to this model, as support for Bush rises, the likelihood of using Republican frames falls. This result is exactly the opposite of what one might first expect, since one would think that the stronger a particular senator supports the president, the more likely they are to use the same frames Bush uses. However, just the opposite appears to be true. There are several possible explanations for this result. First, when we coded the senator’s press releases and floor statements for the frames we made the conscious decision to include instances where a senator mentions a particular frame in the negative. In other words, if a Democratic senator says that Social Security is “not in a crisis,” the word “crisis” would be coded as a Republican frame even though it was used in the negative. This decision was based on the idea that politicians who accept the opposing side’s frames in any way, even if they are used in the negative, are still giving in to the opposing side’s views (Lakoff 2004). Our data suggests that several Democratic senators with low levels of support for President Bush used a high number of Republican frames in their statements. For instance, Senator Barbara Boxer made one statement where she mentioned 44 Republican frames. That is more Republican frames in a single statement than any other senator, including Republicans. Boxer also had a fairly low level of support for bush, just 65%. This high number of mentions of Republican frames, coupled with low levels of support for Bush was apparent among several Democratic senators, and it could have contributed to this negative relationship.

Another possible explanation for this unexpected result is offered by Tim Groeling and Robert Lichter in their study about partisan support in the evening news. According to this study members of the president’s party in Congress vote in accordance with the president about 75% of the time, but in their appearances on the evening news those same members of Congress support the president only about 40% of the time. This discrepancy is explained by the strategic nature many members of congress use in both their voting and in their public statements. Members of Congress may vote one way, but they may try to say just the opposite in order to appeal to certain constituencies (Groeling and Lichter 2004). If the same findings hold true for press releases and floor statements as for statements made on the evening news, then this theory offers another possible explanation for why Republican frames decrease as support for Bush increases. Given the controversial nature of the social security debate, and given how recent polls show that a majority of Americans currently oppose privatizing Social Security, it is possible that some senators with high support levels for President Bush have been speaking out in public against Bush’s plan.

            All three of the other independent variables in the first two regression models were insignificant. The first of these variables was whether or not a particular senator is a member of the party’s leadership. In the first regression this variable had P value of .064 and a coefficient of 30.4267. In the second regression this variable had a P value of .063 and a coefficient of 7.4358. It was expected that members of the party leadership would be more likely than most senators to speak out and use their party’s frames. However, this variable was found not to have a significant relationship to the number of frames used. One possible explanation for this result is that the top party leadership has designated other senators to be the primary spokespersons on this issue. For instance, Democratic senator Charles Schumer is considered one of his party’s leading spokespersons on Social Security and he is often sent to make speeches and appear on news programs as the representative of his party on this issue (O’Rourke 2005). It is possible that instead of the top senate leaders making statements about Social Security, other senators have been appointed to make statements for the party.

            Another independent variable that was insignificant was the percent of each senator’s state that is age 65 or over. In the first regression this variable had P value of .647 and a coefficient of -1.0456. In the second regression this variable had a P value of .414 and a coefficient of -.4557. Given the particular attention that senior citizens pay to matters concerning Social Security, and given how polls suggest that the older someone is the less likely they are to support Bush’s privatization plan, I expected that those senators with a higher proportion of their constituencies age 65 or older would be less likely to speak out against changing the current system (Polling Report 2005). However, this variable proved insignificant. A possible explanation for this variable’s insignificance is that many senators might take a trustee approach to representation rather than a delegate approach. A trustee approach to representation means that politicians follow what they think is best for their constituents and for the country, even if their decision goes against the wishes of their constituents (Pitkin 1967). Given that the founding father’s established the senate to have more of a trustee role than the House of Representatives, and established longer terms of service in the senate to account for that difference, it is possible that many senators are not affected strongly by the wishes of particular constituencies, but instead senators may simply do what they feel is best for the country.

            The final independent variable in the first two regressions was whether or not a senator is a potential candidate for president in 2008. In the first regression this variable had P value of .357 and a coefficient of -12.7699. In the second regression this variable had a P value of .54 and a coefficient of 2.0665. It was thought that those senators who are potential candidates would be more outspoken on this issue since many potential presidential candidates could use this issue as an opportunity to make their name better known. However, this variable also proved insignificant. This insignificance could be due to the fact that while some potential presidential candidates, such as Rick Santorum, see this issue as an opportunity to make a name for themselves and have been rather outspoken on this issue, other potential candidates, particularly those who are better known such as Hillary Clinton and John Kerry, see this issue as too controversial for a potential national candidate and would prefer to remain as silent as possible on the issue.

            Another possible explanation for this variable’s insignificance is that the group of potential presidential candidates is quit diverse and includes senators with a broad range of ideologies and from very different states. It is possible that while these senators may share an aspiration for higher office, they are different enough in other ways that there is no one particular pattern of speaking out or not speaking out on this issue.

Regressions 3-6:

            The other four regression models concerned the second part of this study, dealing with how successful the Democrats and the Republicans were in getting their particular frames in the news media. These four regression models used four different dependent variables: Articles in the Washington Post mentioning Democratic frames, articles in the Washington Post mentioning Republican frames, articles in the Washington Times mentioning Democratic frames and articles in the Washington Times mentioning Republican frames. All four of these regressions shared the same independent variables: Democratic frames mentioned by Republicans, Republican frames mentioned by Republicans, Democratic frames mentioned by Democrats and Republican frames mentioned by Democrats. Since news articles tend to follow a day behind when a press release is issued or a speech is made, the independent variables for these models were lagged one day.

Tables 5-8 present the regression coefficients and statistical significance of each independent variable in the regressions 3-6. In these regressions, the R 2 value was 0.0779 in regression 3, 0.1003 in regression 4, 0.0626 in regression 5 and 0.0944 in regression 6.

Tables 5-8 about here

From this table we see that only one of the independent variables in regressions 3-6 is significant. That variable, the Democratic frames said by Democratic senators and used in articles in the Washington Times, had a P value of 0.042 and a coefficient of 0.0151, showing that there is a positive relationship between the number of Democratic frames said by Democrats and the number of articles in the Washington Times published the next day that mention Democratic frames. In examining the predicted values of this variable in regression 6 we can see the actual influence this variable has on the dependent variable.

Table 9 about here

The predicted values show us that as the number of Democratic frames increases from what is essentially 0 to around 59, the average number of articles in the Washington Times mentioning Democratic frames increases from .99 to 2.2. This result is somewhat surprising. Since the Washington Times is generally considered to be a conservative newspaper it was expected that it would be more likely to have a significant and positive relationship with Republican frames than with Democratic frames, while the Washington Post would be more likely to have significant and positive relationships with Democratic frames. However, given how more Democratic senators spoke out on this issue, and given how the Democrats were more unified in their message it is not entirely surprising that the Democratic frames were picked up and the Republican frames were not.

The unified message of the Democrats can be clearly seen in the sheer number of Democratic frames used in some individual press releases and floor statements. For instance, on one day, February 16, the total number of Democratic frames mentioned in all press releases and floor statements on that day was 182, while on that same day the Republicans did not mention a single Republican frame. This barrage of Democratic language meant that reporters reading those releases and statements would likely pick up much of the same language.

What is surprising, however, is that none of the other independent variables in regressions 3-6 proved significant, showing that there is no significant relationship between the number of mentions of frames by senators and whether or not those frames are used in newspaper articles, with the exception of the Democratic frames used in Washington Times articles. It is especially surprising that the Washington Times was influenced by the Democratic frames while the Washington Post, which is thought of as the more liberal newspaper, was not influenced by Democratic frames.

The insignificance of these variables can be explained by several factors. First, the relatively low number of Republican frames during the two month period shows that there was not nearly as unified a message coming from the Republicans as was coming from the Democrats. For instance, the largest number of Republican frames mentioned by Republicans in a single day was 37, while the Democrats mentioned over 100 Democratic frames on four separate days. This discrepancy likely accounts for the fact that none of the variables containing Republican frames were significant.  

It is less clear why the Democratic frames mentioned by Democrats were significant in the Washington Times but not in the Washington Post. One possible explanation relates to the number of articles on this topic each paper published. During this two-month period the Washington Post published a total of 154 articles mentioning Democratic frames and 96 articles mentioning Republican frames. During the same period the Washington Times published 92 articles mentioning Democratic frames and 59 articles mentioning Republican frames. That averages 2.6 articles mentioning Democratic frames and 1.6 articles mentioning Republican frames every day in the Washington Post, and 1.6 articles men mentioning Democratic frames and 1 article mentioning Republican frames every day in the Washington Times.

Given that the Washington Post published significantly more articles on this topic every day than did the Washington Times it is likely that there are more individual writers who write articles about the Social Security debate at the Washington Post than at the Washington Times. If many of the articles on Social Security that appeared in the Washington Times were written by a single writer, as is common at many newspapers, then it is possible that that writer was personally biased towards the Democratic point of view and therefore was more likely to use Democratic frames in his articles. It is also possible that a single writer writing a series of articles on Social Security simply read so many of the Democratic press releases containing Democratic frames that he or she subconsciously began to use the Democratic frames in the news articles.

            In contrast to the Washington Times, the Washington Post published multiple articles relating to Social Security reform nearly every day. Given that the Washington Post published so many more articles on this topic, it is likely they devoted several different writers to writing articles on Social Security. More writers writing about a topic decreases the chances of a single   personal bias appearing, and also decreases the chances of bias subconsciously appearing in articles due to the large number of frames appearing in press releases since multiple writers would have to be influenced in order for the bias to appear. This explanation suggests that larger newspapers that write multiple articles in a given day on a topic tend to be less susceptible to influence by politicians than are smaller newspapers that may have a single reporter writing nearly every article on a particular topic.

VI. Conclusion

            As the results of this study show, it can be very hard to predict exactly what factors influence politicians to say what they say, and equally hard for politicians to significantly influence how stories are framed in the news media. The existence of only one significant variable in the first part of this study shows that certain characteristics of politicians, such as their role in the party leadership, the demographic makeup of their home state, and their potential presidential aspirations, did not significantly influence these senators to either speak out or stay silent on the Social Security debate. In fact, even the one variable that was significant, a senator’s support rating for President Bush, did not completely reveal the expected results. While this variable showed an expected negative relationship between the support level for Bush and the number of Democratic frames mentioned, it surprisingly also showed a negative relationship between a senator’s support for Bush and the number of Republican frames mentioned. This finding goes against most logical assumptions about how senators who strongly support Bush would act, and it reveals a great deal that could be useful for future research.

            In particular, the negative relationship between a senator’s support for Bush and their use of Republican frames could be an example of several different theories. First, it is possible that this negative relationship is the result of many Democratic senators using Republican frames in the negative in order to try and debunk the Republican arguments. If this explanation is correct then it would be a perfect example of Karl Lakoff’s argument that politicians who accept their opponents frames in any way, even if it is to refute their claims, are giving in to their opponents explanations and therefore helping their opponents win the debate (Lakoff 2004). Given the high number of Republican frames used by several Democrats, this explanation seems likely to be at least partially responsible for this unexpected negative relationship. This finding shows how important it is for politicians to not only use their own frames, but also to be sure not to accept the frames of their opponents, even if those frames are being refuted. It seems more effective to simply ignore the other side’s arguments and concentrate solely on one’s own positions.     

            The other explanation for this negative relationship follows that theory of Tim Groeling and Robert Lichter that the president’s party’s support for his policies in the news media is weaker than their actual votes suggest could also be at least partially responsible for this negative relationship (Groeling and Lichter 2004). This explanation would be hard to test definitively until there is an actual vote in the senate on Social Security reform. When that vote does occur, it will be possible to see if the same relationship that Groeling and Lichter found exists for senators on this issue.

            The other surprising revelation from this study was the lack of influence that most the senators’ statements had on the news coverage of Social Security. With the exception of the Washington Times, whose coverage was significantly influenced by the Democratic frames offered by Democratic senators, there was no other significant relationship between the number of frames in statements by senators and the existence of those same frames in news articles. This finding has several implications on the existing study of political framing and the relationship between politicians and the news media. Many of the previous studies on this topic have shown the significant power that the president has at influencing the news coverage of a particular topic (Heldman and Carter 2004 and Farnsworth and Lichter 2004). This study ads to that existing research by showing that the power of the executive branch to influence media framing does not seem to extend to the legislative branch, at least not to the same extent. While these previous studies found the president could be very influential at framing issues in the media, the relative inability of senators to successfully get their party’s frames in newspapers could be representative of the weakness of the legislative branch relative to the executive branch in getting their message out into the public. However, given that this study examined only two media outlets, and the Democrats were successful at influencing one of those outlets, a more extensive study examining a greater range of newspapers and other news sources is needed to definitively prove this theory.

            One final observation concerning both significant variables is important to make. It is possible that the relative lack of influence the senators had in framing the news stories was at least partially related to the negative relationship between the level of support for President Bush and the number of Republican frames mentioned by senators. If the reason for this negative relationship was due to Democratic senators using Republican frames in the negative, then it is possible that the Democrats significantly weakened their use of Democratic frames by trying to refute so many of the Republican frames. This explanation could be at least partially responsible for the lack of influence the Democratic frames had in the Washington Post. However, to prove this reasoning it would be necessary to conduct a more extensive study on this topic that looks at many more news outlets I examined. What this study does show, however, is that while having a strongly united party with a clear message is important, it is not always enough to win influence over the news media.

Works Cited

Farnsworth, Stephen and S. Robert Lichter. “The Mediated Congress: Coverage of Capital Hill             in the New York Times and the Washington Post.” Paper prepared for the Annual         Meeting of the American Political Science Association, Sept. 2-5, 2004.

Flowers, Julianne, Audrey Haynes and Michael Crespin. “The Media, the Campaign, and the     Message.” American Journal of Political Science. Vol. 47, No. 2 (Apr. 2003), 259-273.

Groeling, Tim. “Circling the Firing Squad: An experimental Examination of Divided         Communication in Unified Government.” Paper prepared for the Annual Convention of      the American Political Science Association, Philadelphia, Pennsylvania, 2004.

Groeling, Tim and Robert Lichter. “Circling the Firing Squad: A Six Year Examination of           Partisan Support and Discord in the Evening News.” Annual Meeting of the American            Political Science Association, September 2-5, 2004.

Heldman, Caroline and Erinn Carter. “Weapons of Mass Persuasion: Print Media Coverage of the Second Gulf War.” Paper prepared for delivery at the 2004 Annual Meeting of the American Political Science Association, Chicago, Illinois, Sept. 205, 2004.

Hetherington, Marc. “The Media’s Role in Forming Voters’ National Economic Evaluations in   1992.” American Journal of Political Science. Vol. 40, No. 2 (May. 1996), 372-395.

Kuklinski, James and Lee Sigelman. “When Objectivity if Not Objective: Network Television    News Coverage of U.S. Senators and the “Paradox of Objectivity.” The Journal of                                Politics. Vol. 54, No. 3 (Aug. 1992), 810-833.

Lakoff, George. Don’t Think of an Elephant. White River Junction, VT: Chelsea Green          Publishing Company, 2004.

Niven, David. “Consensus Theory of Political Journalism.” Prepared for the Annual Meeting of the American Political Science Association, Chicago, Illinois, 2004.

O’Rourke, Lawrence. “GOP May Yield on Bush’s Plan.” The New Observer. April 9, 2005.   

Meckler, Laura. “Democrats Bash Bush Social Security Plan.” The Associated Press. January 29,        2005.

Pitknin, Hanna Fenichel. “Political Representation” The Concept of Representation. University             of California Press; Berkeley, CA. 1967.

Polling Report. “Social Security Polls.” http://www.pollingreport.com/social.htm. Accessed on   April 17, 2005.

Rosenbaum, David. “Senators Urge Bush to Sell Overhaul of Social Security.” The New York   Times. January 26, 3005.

Sellers, Patrick. “Manipulating the Message in the U.S. Congress.” Harvard International      Journal of Press and Politics , Vol. 5, No. 1 (2000), 22-31. 

Stolberg, Sheryl Gay. “For Democrats, Social Security Becomes a Defining Test.” The New York         Times. January 30, 2005.

Watkins, Thayer. “Kahneman and Tversky’s Prospect Theory.” San Jose State University         Economics Department. http://www2.sjsu.edu/faculty/watkins/prospect.htm. Accessed   on January 30, 2005.

Table 1. Influences on Senators’ Mentions of Democratic Frames

(standard errors)

Table 2. Influences on Senators’ Mentions of Republican Frames

(standard errors)

Table 3. Predicted Number of Democratic Frames

Table 4. Predicted Number of Republican Frames

Table 5. Influences on News Articles Mentioning Democratic Frames in the Washington Post

(standard errors)

Table 6. Influences on News Articles Mentioning Democratic Frames in the Washington Times

(standard errors)

Table 7. Influences on News Articles Mentioning Republican Frames in the Washington Post

(standard errors)

Table 8. Influences on News Articles Mentioning Republican Frames in the Washington Times

(standard errors)

Table 9. Predicted Number of Articles Mentioning Democratic Frames in the Washington Times

While the party leadership actually does include several more senators than are listed here, other leadership positions such as the Chair of the Campaign Committees and the Secretaries of the Conference were not included in this study so as to limit the leadership positions to each party’s top leadership positions.

Although our data shows a predicted value of -1.986, we consider this number to read 0. The abnormal distribution of this variable caused the negative predicted value, but the number is equivalent to 0.