Public Opinion Quarterly Advance Access originally published online on August 10, 2007
Public Opinion Quarterly 2007 71(3):349-366; doi:10.1093/poq/nfm020
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
Testing the Permanence of the Permanent Campaign
An Analysis of Presidential Polling Expenditures, 1977–2002
Address correspondence to James A. McCann; e-mail: mccannj{at}purdue.edu.
 |
Abstract |
|---|
 TOP Abstract IntroductionThe Relationship between Time...Research Design and FindingsConclusionReferences |
|---|
Many presidential observers argue that the modern White House is the site of more-or-less permanent campaigning. In a recent POQ piece, Murray and Howard (2002) [Public Opinion Quarterly 66:527–558] explore one indicator of the "permanent campaign," the extent to which Presidents Carter, Reagan, G.H.W. Bush, and Clinton commissioned independent opinion polls and focus groups to assist in policymaking and political maneuvering. Murray and Howard suggest that while a sophisticated polling operation has been institutionalized in the White House, there is substantial variation in how much a president uses this operation. In this article, we model presidential polling expenditures over time using monthly figures. We find that presidents do not vary significantly in the average amount spent per month on polls. There are, however, two recurring patterns of variation within presidential administrations: Presidents tend to spend significantly more on internal polling during the most intense months of a presidential reelection campaign; and polling expenditures increase over the course of each presidential term. These findings suggest that there are common forces (e.g., elections, natural decline in support) that have driven all presidents since Ford to poll.
|
Introduction |
|---|
|
TOPAbstractIntroductionThe Relationship between Time...Research Design and FindingsConclusionReferences |
|---|
From its humble roots in the eyes of the Founding Fathers, the U.S. presidency has grown into the central branch of government. To be effective, presidents must not only have the "power to persuade" policymakers within the Beltway, but they must actively cultivate a personal following across the country. For this reason, scholars have noted a marked tendency for presidents to "go public" – i.e., to bolster public support via nationally televised speeches, symbolic photo opportunities, and travel (Kernell 1997
). In recent decades, has this tendency expanded into a permanent campaign within the White House? Some political observers suggest that it has, and shudder at the prospects of unceasing mobilization and plebiscitary governance (e.g., Blumenthal 1982; Lowi 1986; Heclo 2000). While there is no question that the public responsibilities of the president have grown tremendously in an age of "24/7" news broadcasting and rising demands for strong executive leadership, researchers assessing the "permanence" of campaigning from the White House continue to wrestle with core issues of conceptualization, instrumentation, and model building.1 How do we know such campaigning when we see it, given that the familiar markers of campaign politics – conventions, televised debates, voter registration drives, and the like – are absent? What are the standards to judge whether campaigning, however defined, is indeed nonstop over the course of a presidential administration?
One intriguing approach to these questions appeared recently in this journal, Murray and Howard's (2002) "Variations in White House Polling Operations: Carter to Clinton." Murray and Howard measure campaigning in the White House by tabulating the amount a president spends per year on public opinion polling. On the face of it, this is a sensible measurement. Polling is an integral part of political campaigning, whether one is running for county assessor, the state legislature, or the U.S. presidency. Presidents have, at their disposal, a state-of-the-art polling apparatus, an operation that would be the envy of many university research departments. By law, the president's party must disclose to the Federal Election Commission (FEC) how much it spends on pollsters and survey research. These figures consequently offer a kind of running tally for White House campaigning. In short, when spending increases, we can infer more campaigning is taking place.
Murray and Howard argue that there is a significant amount of variation in polling across presidential administrations – so much variation that it would be difficult to conclude that anything like a "permanent" campaign exists in the executive branch. Rather, some presidents appear to take their public roles more seriously than others and poll accordingly. In two cases (Reagan and Clinton), a hefty amount of polling is done from the start of the administration; i.e., there appears to be a "permanent campaign" for public support. In two others (Carter and G.H.W. Bush), polling is rather light. The authors thus offer a novel and plausible view of White House campaigning: The ability of presidents to poll has become institutionalized, in that all executives are able to sound out the public at a moment's notice; but the use of this resource is largely a matter of leadership style.
In this piece, we reconsider this conclusion. While accepting the authors’ main premise that trends in presidential polling can offer valuable insights into how permanent the so-called "permanent campaign" is, we explore the timing of polling expenditures in far greater detail through the use of monthly expenditures rather than annual aggregate figures. On methodological grounds, this is a marked improvement, for it allows us: (a) to estimate far more precisely the tendencies of presidents to poll; (b) to conduct formal significance tests for differences in these tendencies across several administrations; and (c) to chart the contours of spending within presidential terms with much greater clarity.2
Our analysis points to somewhat different substantive conclusions. First, there appears to be a rather consistent disposition to poll among recent presidents, despite substantial differences in leadership styles. Furthermore, our findings confirm that the electoral calendar, a factor that Murray and Howard emphasize, can account for a great deal of variation in polling expenditures. However, this is not the entire story. The empirical results show that in each presidential term there was a significant month-to-month increase in spending on polls. Moreover, in a few months between 1977 and 2000, the president spent appreciably more on polls. Not all of these "outlier" periods can be directly linked to the timing of the general election (though some occurred during the presidential primary season), suggesting that there are nonelectoral events that take place within a president's term that drive presidents to call for more polling.
Our analysis unfolds in three sections. Below we discuss a hypothetical four-part typology to describe the relationship between White House spending on polls and the passage of time in a presidential term. We then review the empirical findings with a focus on this hypothetical typology. Following this, a concluding section briefly considers the implications of our models for the study of the public presidency.
|
The Relationship between Time and Presidential Polling |
|---|
|
TOPAbstractIntroductionThe Relationship between Time...Research Design and FindingsConclusionReferences |
|---|
How much do presidents poll, and when do they do it? Chief executives certainly have the potential to field extensive surveys whenever they wish. The modern White House staff has transformed itself from less than a dozen or so offices into a massive public relations apparatus capable of vast outreach and promotional efforts. Since President Nixon, the addition of the Offices of Political Affairs, Communications, Public Liaison, Intergovernmental Relations and a wide array of specialty offices (e.g., women's issues, environmental initiatives) have worked together to create a White House that relentlessly seeks to improve the president's public standing in myriad ways. Beginning with President Johnson, executives worked to gain control of the national party apparatus, such that the Democratic or Republican National Committee has become a virtual White House annex replete with presidential loyalists and resources to promote the president. Adding to these institutional resources, presidents since Nixon have hired external pollsters and political consultants to take the pulse of the electorate on an "as-needed" basis (Jacobs and Shapiro 1995
; Tenpas 2000; Eisinger 2003, 112–121). The question above can thus be rephrased: When do presidents need to poll?
One perspective focuses on the "electoral connection," arguing that presidents, first and foremost, are actors tied to the electoral calendar (cf., Mayhew 1974; Tenpas 1997; Tenpas and Dickinson 1997; Charnock et al. 2006). Late in the third year of the administration, on the eve of the primaries and with a looming general election contest, an executive's need for high quality public opinion data becomes integral to campaign strategy. Such data would help the president frame his political positions so as to put himself or his preferred successor in the best possible light as the election approaches. It may also inform the political calculation that determines where the president will spend his time and what issues he will emphasize. An increase in the frequency and length of surveys would likely result in a rise in polling expenditures beginning around, say, the 37th month of the administration, near the start of the nomination caucuses and primaries, and continuing through the general election. Before this time, there may be little need for extensive public opinion polling since policymaking would be largely focused on "inside the Beltway" considerations. Figure 1a formalizes this perspective. To critics of the permanent campaign (cf., Mann and Ornstein 2000), this is the optimal scenario for presidential leadership and public administration. There is a relatively clean break between periods of campaigning within the confines of the White House, and "presidential management" and "bargaining," where much less opinion polling is required (a situation reminiscent of Eisenhower's 1956 reelection campaign).
|
Diametrically opposed to this notion is the "constant campaign," which is characterized in figure 1b. Here we posit a time-invariant tendency. A president sets a relatively high level of spending for polls at the start of his term and sticks with it. The specific dollar amount dedicated each month to internal survey research may vary from president to president, due in part to leadership style (Murray and Howard 2002
, 537–545). The key point, however, is that a president invests a constant amount in polling throughout the administration.
Heclo (2000, 222) sees this perspective as the most debilitating for the executive branch, and ultimately for the country as a whole:
The permanent campaign is a school of democracy, and what it teaches is that nothing is what it seems, everything said is a ploy to sucker the listener, and truth is what one can be persuaded to believe ... Why should one care? Because our politics will become more hostile than needed, more foolhardy in disregarding the long-term and more benighted in mistaking persuasions for realities.Of course, Heclo is not the first to look askance at the possibility of presidential leadership-via-public-opinion. Fears of too much "democracy" in this vein are as old as the American presidency itself. If White House spending patterns conform to figure 1b, Heclo's sobering assessment may be all the more pressing.
On the other hand, Jacobs (1992) argues that presidents who engage in prodigious amounts of polling throughout their term are not necessarily playing voters for "suckers." Where Heclo sees manipulation, Jacobs sees more potential for accountability. Chief executives, after all, can learn what the public thinks through polling, and in turn take these judgments under advisement as they formulate policies. White House polling over the full course of the term consequently offers much useful feedback and guidance. Ultimately, democratic governance may be enhanced. "The evolution of the public opinion apparatus had a profound effect on policymakers’ perspective toward the public ... [I]t educated government officials to be aware of and sensitive to public opinion" (200).
We will not attempt to resolve these two very different views of polling in the White House; this would take us well beyond the scope of this article. Rather, we wish only to stress the much simpler point that works such as Heclo (2000) and Jacobs (1992) are based on the belief that public opinion polling within the White House is a nonstop activity. Up until now, this assumption has not received the rigorous empirical scrutiny it deserves.
For their part, presidents uniformly deny relying on pollsters and "plebiscitary" devices of any kind. President Clinton was dogged by reports describing him as consumed by the latest survey results and governing accordingly – a perception in large measure fueled by the pollsters’ unusually public and frequent access to the president. "The role of the consultants in the Clinton Administration was without precedent ... Clinton's consultants were omnipresent, involved in everything from personnel to policymaking to the President's schedule" (Drew 1994, 124; see also Jones 1996). Yet Clinton himself claimed to be "above it all." His successor, George W. Bush, has taken pains to distance himself entirely from the White House polling apparatus.3 This was manifested by the unusual presence of a "buffer" between President Bush and his pollster. Throughout the first term, a key White House aide, Karl Rove, was the sole contact for the president's primary pollster, Jan van Lohuizen (Tenpas 2003).
In figure 1c, we posit a model of "episodic campaigning." This hypothetical pattern combines elements from the first two figures. Presidents in this case are said to poll at a rather constant level, but this tendency is disrupted by one or a few significant episodes of greater spending. If these episodes occur in the later months of a presidential term, figure 1c will resemble figure 1a, and we would infer that there is a clear distinction between governing and campaigning within the White House. However, in figure 1c we leave open the possibility that these polling expenditure episodes can occur at any point in the term, perhaps in response to the emergence of new policy items on the public agenda or a crisis. The picture of executive leadership that emerges under this scenario stresses flexibility; some periods in a term call for more information about public attitudes, but these episodes are short lived.
The final model, figure 1d, posits a growing reliance on internal polling data. Early in a presidential term, approval ratings for the chief executive are likely to be relatively positive, and he may have more freedom to design and implement policies (see, e.g., Light 1991; Brace and Hinckley 1992). As terms progress, however, presidents face more significant challenges, and their approval ratings are liable to suffer. Well before the next presidential election, then, figure 1d suggests that a president's demand for polling data increases, and increases steadily from the start of the administration.4
Differentiating between election-based, constant, episodic, and steadily increasing polling involves a certain amount of judgment call, especially since we cannot know the substantive aims of the survey research purely from expenditures data. A series of "episodes" of active survey research over a four-year period might indicate more or less "constant" campaigning. Furthermore, as noted above, episodes of intense polling that happen to coincide with the formal start of the presidential campaign would likely be a reflection of election-based surveying more than a response to a specific event or issue. And it is certainly possible that elements of two or more hypothetical models may be present in the data. We strive to be sensitive to all of these nuances in our subsequent analysis.
|
Research Design and Findings |
|---|
|
TOPAbstractIntroductionThe Relationship between Time...Research Design and FindingsConclusionReferences |
|---|
Presidents and their advisors are generally wary of revealing any strategic maneuverings. Though President Nixon is credited with the innovation of hiring a pollster to conduct polls solely for the administration (Jacobs and Shapiro 1995
), it was not until the establishment of the FEC in 1975 that polling expenses were revealed to the public.5 Presidents and their staffs cannot commission polls themselves, but must work through the party organizations. Since federal law requires the two national party organizations (the Democratic and Republican National Committees, "DNC," "RNC") to report their expenditures on a monthly basis (it was once on a quarterly basis), one can ascertain how much money the parties spent on polling and approximately when. In addition to party committee data, presidents form reelection campaign committees well in advance of the campaign. These temporary committees also render payments to presidential pollsters. Taken together, the party and reelection committee disbursements provide the most comprehensive picture of spending on presidential polling. Despite the public accessibility, identifying polling expenditures is no small task. For one, there are literally thousands of checks issued by the party, and thus thousands of entries in each reporting period. In some cases, expenditures are not listed in alphabetical order, so one must examine each record, trolling for a polling expense. In addition, amendments to the report are often filed subsequently and must be examined to determine whether they contain additional polling expenses.
Perhaps the most vexing feature of the data is that while one can determine the amount spent on polling, there is no information about the substance of the poll. Nor can the date the check was issued tell us exactly when the polling data were actually collected. One professional pollster noted that there might be 30 to 60 day time lags between when the poll was fielded, when the pollster billed the national party, and when the party issued payment. Consequently, if one is trying to examine the role that events play in influencing polling, this approach is not entirely helpful. Nevertheless, the FEC records offer the only cumulative look at party spending on presidential polling from 1977 through 2002. From these reports, one can at least get a general sense of how frequently presidents (and their staffs) commissioned polls.6
Turning to the top of table 1, we examine whether presidents on the whole varied in the resources they dedicated to polling during their administration. To facilitate comparisons, we standardized dollar figures to 2005 values, and we compare only complete presidential terms (Carter through Clinton). On the surface, it appears that average spending varied a great deal from administration to administration, with fluctuations being understandable given differences in presidential styles and skills. The two presidents who were best known for their ability to communicate with the mass public, Ronald Reagan and Bill Clinton, were also the two highest spenders in their two terms. In Reagan's case, just over $200,000 was spent on average each month for polling services, an amount that would dwarf the budgets of many major academic survey research firms. Clinton's expenses were only slightly below this average, at $196,000 each month. Reagan and Clinton, for all their ideological differences, appear to have taken the responsibilities of the "public presidency" quite seriously. One manifestation of this was their seemingly greater attentiveness to voter opinions.
|
On the low side were George H.W. Bush and Jimmy Carter, both of whom have often been criticized for having a "tin ear" when communicating with the public (see for example Bonafede 1977
, 1978; Matalin et al. 1994). In Carter's case, approximately $186,000 on average was dedicated to polling each month, while George H.W. Bush devoted less than $100,000.7 It is important, however, not to overstate these personal "stylistic" differences. An F-test for spending in the Carter, Reagan, G.H.W. Bush, and Clinton White House indicates that the four executives did not significantly differ on average (p =.14), and just a tiny amount of variation in monthly spending is accounted for by differentiating among presidents. If we compare spending tendencies across distinct terms (bottom of table 1), statistically significant differences emerge, but these differences are not great. Both Reagan and Clinton spent less on average in their second terms, and in Clinton's case this decrease is statistically significant. Moreover, George H.W. Bush's spending was on average significantly less than what Ronald Reagan and Bill Clinton spent in their first terms. We find that for most mean comparisons, the "null" assumption of a shared tendency to spend at a particular level (say, $180,400 per month, which is the "grand mean") cannot be rejected.
Were these tendencies constant throughout the presidents’ terms? The standard deviations in table 1 suggest that there was a substantial amount of dispersion around the means, especially in the case of President Carter. We see in figure 2 that in no presidential term does the "No Clear Distinction" model from figure 1b appear to hold. Rather, each president paid handsomely for internal polls in the later months of the administration, especially as the next presidential election approached. In this respect, Carter clearly stands out. After spending rather modestly early in his administration, there is an enormous spike in the heat of the 1980 campaign. In payments standardized to 2005 values, President Carter spent more than two million dollars in these months! George H.W. Bush and Bill Clinton went on spending sprees at approximately the same periods in their administration, though each spent less than half of what Carter spent. For his part, Ronald Reagan also seems to have spent more on polling in the final months of both terms. On the whole, we have strong visual evidence to support the claim in figure 1a and reject the claim in figure 1b: Presidents generally commission more internal polling work in the later stages of their administration, though spending increases vary somewhat from term to term.
|
In table 2, we put this interpretation on a more secure footing through regression analysis. The first column in this table presents the results when presidential spending is regressed on presidential terms (dummy-coded, with the administration of G.H.W. Bush serving as the omitted category) plus a dummy for the fourth year of a presidential term, starting in Month 37 and ending in Month 46, the time of the general election. As we would expect given the findings from table 1, two of the presidential term dummies are significant, Reagan's and Clinton's first terms. After controlling for term-specific effects, there are clear indications of a boost in election-year polling (in the spirit of figure 1a). This first model suggests that chief executives spent an average of $280,900 each month in their fourth year leading up to the election, a highly significant effect (p <.01).8 However, the R2 indicates that a substantial amount of variation in White House spending has been left unexplained.
|
In the second specification, this fourth-year effect is allowed to vary across presidential terms.9 Carter is found to have spent nearly $500,000 more per month in the last year of his administration, while Reagan spent less overall at the end of his second term. An analysis of the residuals from this model points to several months in the series as being "outliers," when markedly more was spent than would be expected given the fitted values from the model. In figure 1c, we raised the possibility of such short-lived "episodes," and in the third regression model of table 2 we formally include them as dummy predictors.
Incorporating these predictors more than doubles the R2 for the model (from.31 to.82). In more substantive terms, we find much more evidence for campaign-related polling at the White House. Four out of the seven-outlier dummies overlap with the electoral calendar. Carter, Bush, and Clinton in both terms spent more on polls in Month 45 and/or 46. In both of the Reagan terms, significantly more was spent early in the fourth year, Month 38 (RR I) and Month 36 (RR II). While we do not know the content of the polls that Reagan commissioned, we believe it likely that issues related to the presidential election, both the primaries and the fall campaign, were salient. The one curious anomaly that defies easy explanation is the notable rise in spending for Month 16 of Reagan's second term. This was before the president became mired in the Iran-Contra scandal, the most highly charged and politically challenging period in this term, and we see no other major events that would have prompted Reagan to poll above and beyond his usual tendency.10 We suspect that the Reagan administration was sizing up American public opinion during this period in an effort to frame policies effectively for the second term, but we can only speculate about this.
If the first three columns in table 2 make a strong case for consistent election-centered polling across all presidential terms, and point to at least one instance of "episodic" polling, the final three regression specifications in the table highlight an underlying tendency for presidents to increase their polling gradually over the course of their term. Model 4 includes a month counter ranging from 1 to 48 as an additional predictor. The coefficient for this variable is marginally significant (p =.11), and implies that after controlling for any differences in presidential style, specific outlier months, and the general tendency for presidents to poll more in the fourth year, there is a discernable month-to-month average spending increase of $1,130.11 The equation in Model 5 permits this effect to vary across presidencies. Little variation surfaces, however, none of the "Month x President" interaction effects are significant, and the R2 shows no improvement.12 All presidents tended to increase their polling by approximately the same amount.
In the interest of parsimony, the sixth and final specification drops the "President x Fourth Year" terms that are clearly insignificant. These results plainly illustrate the tendency for election-year campaigning in the White House, rather than constant campaigning across all four years, and some support is given to the short-lived "episodic" and "gradual increase" hypotheses.
As the FEC releases RNC expenditures, it is possible to assess whether the current president, George W. Bush, stands out from his predecessors, as he himself has asserted. For the time being, we were able to access expenditure data for the first 24 months of Bush's first term (2001–2002). Table 3 presents the findings from a much simpler but suggestive regression. When monthly spending for public opinion data is regressed on time (Months 1–24), George W. Bush is found to have increased his spending by an average of $1,670 each month. This effect is significant at the.10 level and nearly matches the coefficient of $1,140 estimated in the sixth model of table 2. Our tentative inference, consequently, is that George W. Bush's use of internal polling was very much in line with historical practice.
|
|
Conclusion |
|---|
|
TOPAbstractIntroductionThe Relationship between Time...Research Design and FindingsConclusionReferences |
|---|
Scholars and observers of the modern presidency claim that old-fashioned leadership has been replaced by plebiscitary governance, where the executive has no choice but to remain in a permanent campaign mode. Presidential polling is admittedly just one manifestation of this phenomenon, but we believe it to be a critical element of the modern presidency. Following from the innovative work of Murray and Howard (2002)
, this analysis takes a first crack at applying statistical analyses to the study of presidential polling trends, seeking to advance our understanding of the "permanence" of polling. On that note, we were committed to taking temporal fluctuations seriously by breaking down aggregate polling expenditures into monthly increments and carefully considering both systematic and stochastic variation.
We conclude that presidents almost across the board spend on average about the same amount on monthly polling. This conclusion runs somewhat contrary to Murray and Howard (2002), which emphasized the distinctiveness of presidential polling styles, but offered no formal statistical tests and focused nearly exclusively on the first three years in office. Within all presidential terms, there is a steep, statistically significant rise in spending that coincides with electoral campaigning. For Reagan, this increase came in the form of highly significant spikes in spending; for Carter, the fourth year of the term featured a sizeable increase in average polling expenses each month, and astoundingly large figures immediately before the general election. Furthermore, all presidents from Carter through George W. Bush polled more diligently the longer they were in office.
Of course, it should go without saying that this topic is worthy of further research. What do these findings mean for the study of presidential leadership in the modern era? The tendency for presidents to conform to an underlying mean level of spending per month (table 1) implies a kind of "consistency" or "permanence" in public opinion intelligence gathering. That said, the electoral calendar and other pressing policy matters can dictate a level of polling well beyond what the institution normally provides, an important feature of all the administrations. This finding points to a good amount of governing flexibility and discretion on the part of chief executives. In future work, we hope to address not only the trajectory of spending, but also the substance of White House surveying. What do chief executives get for all of their spending? How is this information interpreted? Such questions would bear directly on any assessment of modern presidential leadership, governance, and representation.
|
Footnotes |
|---|
KATHRYN DUNN TENPAS is the Director of the University of Pennsylvania Washington Semester Program, 1608 Rhode Island Avenue, NW, Washington, DC 20036, USA. JAMES A. MCCANN is Professor of Political Science, Purdue University, West Lafayette, IN 47907, USA. An earlier version of this paper was presented at the 2004 meeting of the American Political Science Association, Chicago, IL. We thank Sarah Binder, Tom Mann, Anthony King, Brandon Rottinghaus and the POQ reviewers for helpful advice. In addition, we are indebted to Andrew Furlow, Sarah Taylor, and, especially, Emily Charnock for research assistance. Tenpas would like to thank the Weidenbaum Center at Washington University in St. Louis for providing support during the academic years 2004–2005. The authors are solely responsible for the analyses and interpretations presented here. Replication data files are available upon request.
1 The term "permanent" may have different meanings in different studies and settings. In our case, as described below we interpret "permanent" to mean consistency over the course of a presidential administration. 
2 Murray and Howard focus most closely on spending patterns during the first three years of the term. As discussed below, to determine whether presidents are engaging in "permanent campaigns," it is necessary to consider expenditures across all 48 months. In the cases of Reagan and Clinton, we also model second-term spending.
3 In fact, when responding to critics during the mobilization for war in Iraq, the president proudly declared that his foreign policy would not be based on polls or focus groups. "In his first public comments about the antiwar demonstrations by millions of people over the weekend ... Mr. Bush said his overriding goal was to protect the American people and that leadership sometimes involved bucking public opinion. ‘Size of protest—it's like deciding, well, I’m going to decide policy based upon a focus group. The role of a leader is to decide policy based upon the security, in this case, the security of the people.’" (Stevenson 2003) Since taking office, George W. Bush has spent a great deal on polling. However, more than any of his predecessors, he has fought the image of governing-by-polling.
4 For ease of presentation, we have sketched a simple linear relationship between time and White House polling expenses. Of course, if such a relationship exists, it need not be linear.
5 While our data set could have included the truncated term of President Gerald Ford, we chose to omit this administration because we were interested in determining trends over the course of four-year terms. The timing of the FEC's establishment in 1975 enables researchers to access RNC polling expenditures for the last 20 months of the Ford administration, thus limiting the utility of these records. In addition, the unusual circumstances surrounding President Ford's ascendancy to the presidency as well as the brevity of his term contributed to our decision to exclude him from this study.
6 Our data collection method involved obtaining all polling-related expenditures for presidential pollsters from the Federal Election Commission. These pollsters included: President Carter–Pollster Pat Caddell/Cambridge Survey Research; President Reagan–Pollster Richard Wirthlin/ Decision Making Information, The Wirthlin Group; President Bush–Pollsters Robert Teeter and Fred Steeper/Market Opinion Research, Coldwater Corporation and Market Strategies; President Clinton–Pollsters Stanley Greenberg, Mark Penn and Douglas Schoen; and President George W. Bush–Pollster Jan van Lohuizen/Voter Consumer Research. We searched for total polling expenditures incurred by the national party committees as well as those paid by the primary and general reelection committees. All expenditures are available on microfiche at the Federal Election Commission. After 1997, researchers were able to download the filings and access them on www.fec.gov (since the time of initial data collection in 2001, the FEC has made many more years accessible via the website). For these most recent years, we deliberately did not rely on the FEC search engine, which enables one to plug in pollsters’ names or companies, but is not comprehensive. Instead, researchers paged through each and every party/reelection committee expenditure. At a minimum, data were reviewed by two different researchers. For many years, the data has been triple-checked and, in the case of the Carter party data, we verified our findings on four occasions.
7 These variations are in keeping with Murray and Howard (2002), though spending that occurred after Month 36 was largely excluded from their analysis. If the goal is to assess variations in polling tendencies from one White House to the next, we see little reason to exclude these data.
8 The Durbin–Watson statistics hint at the presence of autocorrelation in the residuals, which is common when the data are temporally ordered. The results from a Prais–Winsten (AR1) model, where first-order autocorrelation is taken into account, are very much in keeping with the OLS findings.
9 Murray and Howard argue that Clinton and Reagan polled significantly more than Carter and G.H.W. Bush during the first three years of the term, when there was no presidential election taking place. This implies, according to the authors, that Clinton and Reagan had different leadership styles and were more attuned to the "public" dimension of the presidency. As shown in these regressions, however, the "Clinton/Reagan" versus "Carter/Bush" dichotomy does not cleanly surface. In Model 2, for example, the dummy predictors for the presidential terms (i.e., the first five coefficients) indicate that Bush's average monthly expenditures in the first three years were not significantly different from Carter's spending or Clinton's in the second term.
10 As noted at the bottom of table 2, none of the other months would be considered outliers based on the residual diagnostics for Models 3–6 (Bollen and Jackman 1990, 264).
11 Adding quadratic and/or cubic transformations into the equation [i.e., "(Month of Term)2," "(Month of Term)3"] would not significantly improve the fit.
12 The difference in F-ratios for Models 4 and 5 does not approach statistical significance (p =.66).
|
References |
|---|
|
TOPAbstractIntroductionThe Relationship between Time...Research Design and FindingsConclusionReferences |
|---|
Blumenthal Sidney. The Permanent Campaign (1982) Rev. New York: Simon and Schuster.
Bollen Kenneth A., Jackman Robert W. "Regression Diagnostics: An Expository Treatment of Outliers and Influential Cases." In: Modern Methods of Data Analysis—Fox John, Long J. Scott, eds. (1990) Newbury Park, CA: Sage Publications. 257–91.
Bonafede Dom. "The Education of a Public Man." National Journal (1977) 1237. July 28).
Bonafede Dom. "A Report Card on Carter—Lowered Expectations After a Year." National Journal (1978) 44–9. January 14.
Brace Paul, Hinckley Barbara. Follow the Leader: Opinion Polls and the Modern Presidents (1992) New York: Basic Books.
Charnock Emily Jane, McCann James A., Dunn Tenpas Kathryn. "Goin’ Mobile: Assessing First-Term Presidential Travel from Eisenhower to George W. Bush." (2006) Paper delivered at the annual meeting of the American Political Science Association, Philadelphia, PA, August 30–September 3.
Drew Elizabeth. On the Edge (1994) New York: Simon and Schuster.
Eisinger Robert. The Evolution of Presidential Polling (2003) New York: Cambridge University Press.
Heclo Hugh. "Campaigning and Governing: A Conspectus." In: The Permanent Campaign and Its Future—Mann Thomas, Ornstein Norman, eds. (2000) Washington, DC: The Brookings Institution and American Enterprise Institute. 1–37.
Jacobs Lawrence. "The Recoil Effect: Public Opinion and Policymaking in the U.S. and Britain." Comparative Politics (1992) 24(2):199–217. January.[CrossRef][Web of Science]
Jacobs Lawrence, Shapiro Robert Y. "The Rise of Presidential Polling: The Nixon White House in Historical Perspective." Public Opinion Quarterly (1995) 59(2):163–95. Summer.
Jacobs Lawrence, Shapiro Robert Y. "Presidential Manipulation of Polls and Public Opinion: The Nixon Administration and the Pollsters." Political Science Quarterly (1995–1996) 110(4):519–38. Winter.
Jones Charles O. "Campaigning to Govern: The Clinton Style." In: The Clinton Presidency: First Appraisals—Campbell Colin, Rockman Bert A., eds. (1996) Chatham, NJ: Chatham House Publishers.
Kernell Sam. Going Public: New Strategies of Presidential Leadership (1997) Washington, DC: Congressional Quarterly Press.
Light Paul. The President's Agenda: Domestic Policy Choice from Kennedy to Reagan (1991) Baltimore: Johns Hopkins University Press.
Lowi Theodore. The Personal President: Power Invested Promise Unfulfilled (1986) Ithaca, NY: Cornell University Press.
Mann Thomas, Ornstein Norman. The Permanent Campaign and Its Future (2000) Washington, DC: Brookings Institution and the American Enterprise Institute.
Matalin Mary, Carville James, Knobler Peter. All's Fair (1994) New York: Random House.
Mayhew David R. Congress: The Electoral Connection (1974) New Haven: Yale University Press.
Murray Shoon Kathleen, Howard Peter. "Variation in White House Polling Operations: Carter to Clinton." Public Opinion Quarterly (2002) 66:527–58.[Abstract]
Stevenson Richard W. "Threats and Responses: The White House; Anti-war Protests Fail to Sway Bush on Plans for Iraq." The New York Times (2003) A1. February 18.
Tenpas Kathryn Dunn. Presidents as Candidates: Inside the White House for the Presidential Campaign (1997) New York: Routledge.
Tenpas Kathryn Dunn. "The American Presidency: Surviving and Thriving Amidst the Permanent Campaign." In: The Permanent Campaign and Its Future—Mann Thomas, Ornstein Norman, eds. (2000) Washington, DC: The Brookings Institution and American Enterprise Institute. 108–33.
Tenpas Kathryn Dunn. "Words vs. Deeds: President George W. Bush and Polling." The Brookings Review (2003) Summer.
Tenpas Kathryn Dunn, Dickinson Matthew. "Governing, Campaigning, and Organizing the Presidency: An Electoral Connection?" Political Science Quarterly (1997) 51–66. Spring.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Conaghan and C. de la Torre The Permanent Campaign of Rafael Correa: Making Ecuador's Plebiscitary Presidency International Journal of Press/Politics, July 1, 2008; 13(3): 267 - 284. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||