Printer Friendly

The polls: searching for determinism: a comparative assessment of first term approval volatility, buoyancy, and polarization.

The search for a recurring trend in presidents' approvals has received much attention by presidency scholars--and for good reason. The identification of a deterministic seasonal pattern in approval has implications for the debate between the presidency- and president-centered approaches. Evidence of an approval trend would appear to lend support to the idea of a president constrained by the institution, whereas an absence of pattern may suggest that approval is purely an ad-hoc reflection of the predilections of the individual.

There are two theory-driven expectations for approval trends, both plot either an increase or decrease in presidential approval relative to an earlier time in an administration. The first, the cumulative decision explanation, argues that a president's broad coalition of support, built while campaigning, will steadily erode in response to the accumulation of decisions that must necessarily follow an election. Many of these decisions may not fully resonate with coalition members' preferences, the consequence being a predictable, steady decline in approval. The second, the psychological response explanation, argues that presidents come to office with inflated approval due to unrealistic expectations. These expectations, when not met, lead to disillusionment and subsequent approval decay. Trend lines are often looked to for confirmation of these approaches. However, evidence of a pattern has been fleeting. This article argues that there is sound theory to suggest recurring patterns in approval, but to illustrate these patterns we must broaden our attention beyond the confining focus of individual-presidency trend lines constructed independent of each other.

Through a comparative assessment of first term approval, this article examines evidence of deterministic patterns in approval when approval is evaluated along the dimensions of volatility, buoyancy, and polarization. In addition, borrowing from both the cumulative decision and psychological response explanations, the patterns evidenced are consistent with the agenda articulation approach, which reasons presidential approval reflects the president's ability to define an agenda by staking positions on salient issues early in the administration.

Volatility, buoyancy, and polarization are useful because they allow us to make comparative assessments of the modern presidents at similar points during their terms. Simply, utilizing a comparative approach and looking beyond single-presidency trend lines, we are able to identify larger deterministic patterns. This comparative approach is inherent in the way we conceptualize and measure these dimensions of approval. Approval volatility measures a president's short-term fluctuations in approval and controls for larger trends, such as either downward or upward drift in approval. Approval buoyancy taps the ability of the president to earn and sustain approval relative to his colleagues at similar points in the term. Approval polarization identifies the range between in- and out-party approval for each president relative to his colleagues throughout the term.

Agenda Articulation

The nature of majoritarian election rules compels presidential candidates to cobble together disparate groups in an effort to build a broad coalition of support. Efforts to build this support often result in presidential campaigns that are marked by rhetoric, promises, and the embracement of valance issues. But once elected, the early period of a president's term is marked by a flurry of policy-defining activity as presidents immediately begin the process of defining and acting on an agenda (Burke 2000, 5). Policy-defining activities are those activities that signal both the direction and means of a president's agenda. Indications of the public policy direction of a new administration, and the means by which it hopes to accomplish its goals, may be explicit in official policy statements and speeches released by the White House or simply culled from the president's comments endorsing proposed legislation. The direction and means of an administration's agenda also may be implicit in those selected for cabinet-level appointments and agency heads. Early in the term, through these activities, an administration articulates its positions on any number of high-profile public policy issues that are to define the core of its agenda. The first year of an administration may be a settling-out period where primary policy positions are staked, lines are drawn, and winners and losers are identified. Through "symbolic" and "substantive" activities (Cohen 1997, 93-94), most key policy positions are staked relatively early in an administration. As policy directions are revealed, it is expected that appreciative supporters and disapproving detractors will respond with marked shifts in approval.

Whereas the tone or general direction of a president's agenda is largely set within roughly the first twelve months of the term, many later minor adjustments can be expected. Activities well into a president's term may be viewed as refinements on previously staked policy positions. These later activities do not signal a wholesale shift in policy, but rather confirmation, clarification, or tuning of existing positions. These refinement activities are less likely to receive news coverage and critical commentary relative to the coverage of activities that initially defined the agenda. The public's awareness of issues "is closely related to the amount and duration of attention these affairs receive in the mass media" (Edwards et al. 1995, 112) as well as the tone, either positive or negative, of the message (Grossman and Kumar 1981). Less coverage, in conjunction with most winners and losers already identified earlier in the term during the settling-out phase, means that the public's response in the form of short-term change in presidential approval will be less remarkable following the initial months of an administration.

The cumulative decision explanation, in comparison to this agenda articulation approach, argues that although presidents may enumerate several public policy preferences during the campaign season, policy specifics are largely left unattended. With few presidential decisions made early in the administration and few details formulated concerning president-sponsored legislation, the public withholds judgment and grants the president an opportunity to exercise leadership and establish a track record (Peterson 1990, 120-21)--thus, the classic "honeymoon." But soon, as Mueller (1970) argues, the number of decisions the chief executive must make following inauguration cumulates, resulting in a general decline in presidential approval. Simply, during the campaign a candidate must build a broad coalition of support and, once elected, the president will begin to act upon his agenda and will begin making decisions necessarily offensive to some minority coalition members' preferences. Over time, the number of decisions cumulate and approval declines.

Similarly, Stimson's (1976) psychological explanation for the assumed general decline in presidential approval suggests that presidents come to office with the naive admiration of most Americans. Weak supporters who are willing to accept the candidate's rhetoric may inflate approval during the initial period. After a campaign of promises, people approve of new presidents because they are generally uniformed and have unrealistic expectations which, when not met, will lead inevitably to disillusionment (see also Mueller 1973, 206-208). Pfiffner (1988, 170) states that the presidential election produces a "dynamic of inflated expectations and over promising" and Shienbaum and Spragens (1988, 591) cite a "burst of initial optimism." According to Brace and Hinckley (1992, 24), a "peak occurs with the unrealistic expectations aroused from the electoral period through the inaugural celebration." A decay in approval is assumed until, near the end of the term, the president may receive a respite "because as he is retiring, his motives are less open to cynical interpretation" (Thomas and Pika 1997, 107; see also Edwards and Wayne 1985, 109-110; Ostrom and Simon 1989; Ragsdale 1996, 182; Cronin and Genovese 1998, 75-85; Eisenstein and Witting 2000).

The agenda articulation approach, while similar on many points, does exhibit marked differences relative to both the cumulative decision and psychological response explanations. The cumulative decision and psychological response explanations suggest that there is a non-fractious honeymoon and that this stability will be short lived as decisions cumulate and disillusionment takes root. Certainly as the president's term progresses, the raw number of decisions made does indeed cumulate, but we argue that this line of reasoning does not account for the weight individual decisions carry in defining the general direction of an administration's agenda. Later-term decisions are expected, in general, not to signal a fundamental shift in the direction of an administration and, therefore, are less likely to illicit conversions from approval to disapproval, or vice versa. From this we extract the hypothesis that during the initial period of an administration approval ought to be marked by high volatility followed by declining volatility in subsequent periods.

Approval Volatility

The measurement of approval volatility over time is especially useful because it may be an indicator of a president's cohesion or clarity in public policy over the term, providing some insight into the ability of an administration to articulate its agenda. Whether positions are popular or not, or whether the president in general is popular or not, clarity in an administration's agenda over time ought to manifest in less approval volatility over time. Stability in the public agenda suggests stability in public approval. This is a different approach than the examination of either upward or downward trends in presidential approval over the term. The concept of approval volatility recognizes that the individual four-year trend lines are a product of many smaller bumps and dips. However, efforts to examine these short-term changes in approval have either focused on explanations for, or measurement of, individual events (e.g., Simon and Ostrom 1989; Baum 2002) or simply treated these fluctuations as non-systematic, random white noise along the more general trend line (e.g., Ragsdale 1993, 1943-48; Cronin and Genovese 1998, 75-85). Approval volatility, in contrast, is calculated by smoothing the larger trend lines and by computing the residuals between the original approval score and the smooth-fitted line, thus measuring the volatility of approval while controlling for the larger drift, either upward or downward, in approval.

The initial trends in presidential approval, presented in Figure 1 and disaggregated by in- and out-party approval, are derived from a unique data set drawn from seven polling organizations and contain a 14-day unit of analysis over the modern presidents' first terms, Eisenhower through Clinton (see Appendix for treatment of polls and data). By smoothing the data presented in Figure 1, we are able to capture volatility on a short-term basis. Because the measure of volatility is a measure of short-term volatility, it is not confounded by the larger direction in a president's approval. That is, whether a president is popular or not is distinct from the degree of volatility expressed in the president's approval. Eisenhower's average in-party volatility, for example, is 1.066 and out-party volatility is 1.609, meaning that the average two-week fluctuation in approval (i.e., peak above or valley below the smoothed trend line) is 1.066 and 1.609 percentage points, respectively. That is, Eisenhower's average two-week fluctuation in out-party approval is greater than the fluctuation in in-party approval by more than one-half a percentage point.


Some evidence is found to support the hypothesis of a fractious initial period following inauguration. The first quarter is marked by a fair amount of volatility, suggesting absent is the stable honeymoon period. In most cases, volatility decreases between the first and second quarters of the terms. In addition, most of the lowest volatility scores appear during the fourth quarter, and the latter half of the fourth quarter in particular is often marked by a decline and leveling-off of volatility, thus suggesting an end-of-the-term presidential respite. Both in- and out-party fourth quarter volatility is lower than first quarter volatility for presidents Eisenhower, Nixon, Carter, Reagan, and Clinton. For example, the out-party volatility for Clinton is 1.1 for the first year in office, 0.8 for the second and third years, and 0.5 for the fourth year. Clinton's in-party volatility is 2.5 in the first and second years, 2.2 for the third, and 1.4 for the fourth year. Johnson's out-party volatility declines for each of the first three terms and then peaks in the fourth year, while his in-party volatility increases between the first and fourth years. Bush is unusual in that both his in- and out-party volatility is lowest in the first quarter and highest in the fourth quarter. These data are illustrated in Figures 2 and 3 below.


In addition, the presidents may be ranked based on their average volatility. Both Clinton and Carter have the highest average in-party approval volatility for the entire term, and Kennedy, Bush, and Eisenhower have the lowest average in-party volatility. Johnson, Nixon, and Reagan have, relative to their colleagues, mid-range in-party volatility. On the other hand, Carter and Eisenhower have the highest average out-party volatility, and Clinton, Reagan, and Kennedy have the lowest average out-party volatility. Nixon, Johnson, and Bush have, relative to their colleagues, mid-range out-party volatility. These rankings are summarized in Table 1 below.

It has been argued that in-party approval is more staid and less volatile relative to out-party approval (e.g., Edwards 1983; Hurwitz et al. 1989, 365). That is, in-party identifiers are more consistent in their support and less likely to make radical changes in approval in response to events. However, although there is much variation among the individual presidents' in-party volatility, and likewise variation among the individual presidents' out-party volatility, Table 1 indicates that, over the course of the modern presidents, the average volatility of in-party identifiers and the average volatility of out-party identifiers is, remarkably, a nearly identical 1.314 and 1.333, respectively.

There are also some interesting differences between the approval volatility of Democrats and Republicans. The three presidents with the highest in-party volatility are Democrats. The average Democrat in-party approval volatility (1.487) is noticeably higher than the Republican in-party approval volatility (1.142). This suggests that Republican identifiers are steadier in the approval of their party's leaders relative to the Democratic identifiers' approval of their party's leaders. However, the average out-party approvals of Republican presidents and Democratic presidents are quite similar, 1.338 and 1.328, respectively. In addition, it is notable that Carter's approval volatility is quite high, having both the highest in-party volatility and the second highest out-party volatility. As noted by Kentleton (2002, 249), Carter's shifting agenda provided an "inevitable mishmash [that] pleased nobody" (see also Greenstein 2000, 136-137). Remarkably, Clinton's in-party approval volatility is nearly two and one half times as great as his out-party approval volatility--he not only has the highest in-party volatility but also the lowest out-party volatility.

Note also that the measure of volatility operates within the parameters of the floor and ceiling effect inherent in presidential approval measures. That is, presidential approval scores have a real-world range of zero to 100 percent approval. A president with high in-party approval is likely to exhibit less volatility relative to a president with moderate approval, ceteris paribus. This is due to the fact that there are few in-party member disapprovers available for conversion to approval and, thus, few available to constitute a surge in response to some presidential activity. In the words of Baum (2002, 266), "the more approvers there are among identifiers . . . the fewer remain to upgrade their evaluations of the president." In likewise fashion, extremely low out-party approval mitigates large dips in approval due to the relatively small pool of out-party members available for conversion. This pattern holds true in our examination of modern presidents: the lowest volatility scores correspond to the presidents that have either remarkably high or low approval. Those with high in-party approval and low volatility include Eisenhower during the third and fourth years, Kennedy during his first year and a half, and Bush during the middle of his term. Low out-party approval also corresponds with low volatility with Clinton and Reagan, especially when approval dips at or below 20 percent.

Approval Buoyancy

Although the course of a president's approval may be fraught with myriad short-term surges and declines, conceptually different is a president's ability to sustain buoyancy in approval over time relative to his colleagues. Buoyancy, more specifically, addresses the question: which presidents have been most successful in gaining and holding in-party support relative to their colleagues? Likewise, which presidents relative to their colleagues have been successful at earning and sustaining approval from out-party identifiers?

For example, both Mueller and Stimson's underlying theories suggest that presidents begin a term under a probationary period and an a priori determined inflated approval. Simply, presidents are given the benefit of the doubt at the beginning of their terms and are later judged on their actions. If this holds true, then we ought to see at the early stages of the presidents' terms a fairly tight range of approval scores and, later, a loosening of the range of approval scores. In other words, we hypothesize that we ought to see similarities in buoyancy at the beginning of the terms and increasing differentiation as the terms progress.

To answer these questions, the difference between the approval score of a president and the average approval score of the seven remaining presidents is plotted at each series point during the first term. In addition, for each of the points in the series, the standard deviation (above and below the average) for the approval scores of the remaining presidents is calculated and smoothed. Plotting the standard deviation allows us to account for any cyclical or seasonal pattern in the spread of the presidents' approval scores over time. The in- and out-party approval paths for each president, relative to the average of the other remaining presidents, are presented in Figures 4 and 5 below.


Illustrated in Figure 4, the range between the bands demarcating the standard deviation for in-party approval increases as the four-year term progresses. During the first several months of the term the range is fairly small, approximately 10 percentage points, indicating a standard deviation of roughly 5 percent above and below the average. As the term progresses, the bands steadily widen with the final several months having a large range, approximately 30 percentage points, indicating a standard deviation of roughly 15 percent above and below the average. The fairly tight spread of the presidents' in-party approval scores at the beginning suggests that, despite differences in the presidents' styles, party affiliations, erstwhile campaigns, and policy articulation, they largely enter office with similarly determined in-party popularity. However as the term progresses more differentiation is apparent, with the range increasing above 30 percent.

Note that for out-party approval for each of the eight presidents, illustrated in Figure 5, the range between the bands demarcating the standard deviation is approximately 20 percentage points at the beginning of the term and, within the first year, the range increases to 30 percent, fluctuating there for the remainder of the term. In contrast to the ranges for in-party identifiers, for out-party identifiers the evaluation of the president during the first months of a term can vary widely, thus suggesting that there is not an out-party probationary period. Out-party members, unlike in-party members, may be more inclined to evaluate presidents immediately, perhaps using as cues recent concessions made on the campaign trail or policy promises made during the transition period. The fragility of the in- and out-party approval coalition during the initial period of an administration largely stems from out-party identifiers.

These findings are consistent with our underlying theory. In-party approval appears to be within a fairly narrow range despite policy idiosyncrasies. The approval garnered from in-party identifiers tends to be determined, but as the term progresses, in-party approval becomes increasingly subject to the vagaries of the political process. In the broadest sense, this suggests that particular presidents may be better able than others, either by their own skillful volition or at the fateful hands of events, to retain support among in-party identifiers after an initial period and that the presidents' abilities to garner approval among out-party identifiers may be judged immediately.

What is more, however, is the comparison of the different abilities of presidents to win and maintain approval of their in-party identifiers. According to Figure 4, Eisenhower, relative to the other presidents, demonstrates high in-party buoyancy consistently across the entire term (with the exception of a brief dip within the first six weeks of his term). In fact, during the final year of his first term, Eisenhower exhibits unusually high in-party buoyancy. Kennedy is also popular with in-party identifiers during the first two years of his term, but then approaches average in the third year. Johnson, in stark contrast to Eisenhower and Kennedy, has consistently low in-party approval dropping beyond the standard deviation for the latter half of his term. Nixon, remarkably, has very average in-party approval, slightly above his colleagues and always within the standard deviation bands.

Whereas the first several months of Carter's term are quite average, very soon his in-party approval slides drastically, beyond the lower standard deviation band. In the later weeks of his third year, Carter's in-party approval climbs sharply approaching the lower band, but is quick to fall again. Reagan's in-party approval initially surges briefly above the standard range, but falls again and stays close to average before gradually approaching the upper band for the remainder of the term. For the first two and one half years, Bush's in-party approval remains largely above average. Bush's in-party approval is unique in the magnitude of its rapid descent in his third year. With the advent of the Gulf War, approval surges only to decline soon after, yet it remains noticeably within the standard deviation bands. Clinton's initial path of in-party approval shows similarities with Carter. After beginning with relatively low in-party approval, it slips well below the lower band. What is of interest is the steady increase in in-party approval over his term. Clinton, relative to his colleagues, becomes increasingly popular within his own party over his first term.

Which presidents, then, are more successful in securing and sustaining approval of out-party identifiers? According to Figure 5, Eisenhower, relative to the remaining presidents, demonstrates high out-party buoyancy over the four-year term, especially in the last two years when his approval is well above the standard deviation band. During the first two years Kennedy, too, maintains high out-party approval, but his approval is rather average for the remainder of his time in office. During the first year of his presidency, Johnson is fairly average in out-party approval, but for the remaining three years his performance is less than par. Note, however, that Johnson is better able to secure out-party approval relative to in-party approval. Once again, Nixon may be said to have the most average out-party approval of his colleagues, staying within the standard deviation bands for nearly the entire term. This average is in contrast to popular perceptions of Nixon's poor image among opposition identifiers (e.g., Kessler 1988). Carter enters office with high out-party approval. Although his approval declines to below average within the first year, it stays within the standard deviation bands until the last six months of his term when his support plummets. This is consistent with notions of the final year of the Carter presidency being marked by malaise. Whereas Carter's showing among out-party identifiers is not stellar, it is interesting to note it is markedly better than his showing among in-party identifiers. Carter's unpopularity stems largely from a failure to win approval from within his own party.

Reagan's out-party approval remains well below average while his in-party approval remains above average for the majority of his term. Bush begins the first four months of his term with relatively low out-party approval, but for nearly two years in the middle of his term Bush demonstrates exceptional buoyancy. Leading into the election near the end of his term, Bush makes a noticeable recovery in out-party (and in-party) approval. On the other hand, Clinton's ability to win out-party approval is dismal. He begins his term far below the lower standard deviation band and this quickly deteriorates even further. No other president has failed so sharply to win out-party approval. Clinton manages a modest recovery in the latter half of his term, but still remains well below average.

Approval Polarization

In-party identifiers, not surprisingly, are likely to exhibit greater support for their president relative to out-party identifiers, as demonstrated earlier in Figure 1. Approval polarization is the sharpness of the division between the Democratic and Republican approvals of a president. It has been suggested that the gap between Democratic and Republican evaluations of the president has increased in the 1980s and 1990s (Bond and Fleisher 2001). In addition, Edwards (1983, 218) suggests there to be a "depolarization of politics following an election." Our measure will allow us to examine the polarization following elections, identify either the increase or decrease of polarization since the 1950s, and make comparative assessments of the modern presidents.

Approval polarization is measured as the difference between the in- and out-party approvals at each of the points in our series. Figure 6 plots as a percentage the distance between a particular president's polarization and the average polarization of the remaining presidents. Also, the standard deviations (above and below the average) of these remaining presidents' polarization scores are plotted as smoothed lines. A point on the trend line scoring higher than zero (i.e., average polarization) indicates the percent distance above the average and, likewise, a point on the trend line scoring lower than zero indicates the distance below the average.


President Eisenhower not only shows high levels of approval for both in- and out-party members (demonstrated earlier), but his approval polarization is less than average and nearly always within the standard deviation bands. Kennedy, for the first year and one half of his tenure, has very low polarization, but by the middle of the second year it climbs above average. Johnson's polarization remains below average and, in the fourth year, is quite low, dropping below the standard deviation band. Nixon, once again, evidences much average polarization throughout the first term. Carter, in concert with his over-all dismal popularity among in- and out-party identifiers, evidences the least polarization. Bush's presidency is initially marked by above average polarization, but by the middle of the first year his polarization subsides and remains fairly low until the third year when it climbs again.

Bond and Fleisher's assertion of increased polarization in the 1980s and 1990s is only partly borne out. Presidents Reagan (1980-84) and Clinton (1992-96) are the most polarizing of the modern presidents: through the entirety of their first terms the distance between in- and out-party approval is substantial, often exceeding the standard deviation bands. Yet, most of Bush's term (1988-92) evidences less than average polarization. And there is little evidence for the suggestion of depolarization following an election: approval polarization increases between the back-to-back transitions of Johnson to Nixon, Carter to Reagan, and Bush to Clinton. Along the polarization dimension of approval there appears little determinism.


A persistent assumption concerning first term presidential approval is that it follows a deterministic path, usually that of predicable decay and rebound. As such, presidents will enter office with high levels of approval, but within several months presidential popularity begins a general descent signaling the end of the initial honeymoon period. This erosion may continue until close to the end of the president's term when there tends to be a slight leveling-off or rebound. Authors have suggested that this trend "may be almost wholly independent of the president's behavior in office" (Stimson 1976, 1). This trajectory is presupposed despite the striking diversity in personalities, leadership styles, and issues that have faced our modern presidents.

Our cursory examination of the simple trend lines (i.e., Figure 1), however, has not revealed a recurring pattern. While in selected instances the honeymoon, decay, and rebound may be isolated, there is little apparent consistency across presidents. Simply, there remains remarkable variation in the paths of presidential approval. The failure to identify a consistent pattern has led some to suggest that assertions of deterministic trends may be "folklore" (Cook and Ragsdale 1998; see also Ostrom and Smith 1992, 127-28). This article, however, suggests that it may be premature to discount completely notions of deterministic patterns in presidential approval, as well as discounting theory linking approval to a president's efforts to act on an agenda and direct public policy.

This analysis goes beyond reliance on the elementary paths of approval and presents a comparative assessment based on the three related, yet conceptually distinct, approval dimensions of volatility, buoyancy, and polarization. We have theorized that there may yet be a degree of determinism in approval among our dimensions of approval. By examining the volatility of the minor bumps and dips that form the general trend lines, we see some evidence of a seasonal pattern in approval. The early stages of a presidency are witness to volatility, notably in contravention to the idea of a non-fractious honeymoon period. In our comparative assessment of approval buoyancy there is evidence of a fairly narrow range of approval scores during the initial stages of the administrations. For in-party approval scores, the range at the beginning of the term is fairly small relative to the variation evidenced later in the term, suggesting that the deterministic constraints on approval early in the term lessen as the term progresses. The initial range in out-party approval is nearly double that of the in-party range, suggesting that out-party approval scores are relatively more subject to vagaries.

Hager and Sullivan state that institutional "responsibilities pressure presidents for attention, regardless of personal predilections" (1994, 1083). This foray into a comparative assessment of modern presidents' approvals underscores the continuing debate of whether a president's approval is dictated by a priori conditions or is a reflection of the president as an individual, his attributes and personal qualities directing his management of public policy. The presidency-centered perspective gives credence that the demands and responsibilities of the office are substantial institutional constraints within which all presidents operate. One's activities, while reflecting individual choice, are nevertheless embedded within the parameters of the institution. On the other hand, it is well argued that the character of a president shapes his management of the environment (e.g., George 1974; Barber 1992; George and George 1998). Thus, the context of the institution gives way to the person, as some may be better suited than others due to intelligence, political acumen, and personality to parlay circumstances to popular approval. As aptly noted by Greenstein (2000, 200), within presidents "human imperfection is inevitable, but some imperfections are more disabling than others." Evidence of deterministic seasonal patterns in approval would appear to lend support to the idea of a president constrained by the office, while an absence of pattern suggests that approval is dependent upon the unique context of the individual. This research indicates that during the early phases of the term presidents, on average, are less able to control public approval relative to later in the term, thus suggesting some determinism in approval. As the term progresses, however, seasonal assignment of approval wanes, indicating a loosening of constraints.


Treatment of Polls and Data

Unit of analysis. Each month of each four-year term is split into the first half and latter half of the month, providing a rough 14-day unit of analysis. This provides a maximum of 96 cases for each president's term. In constructing the approval trend lines, approval scores were obtained from seven polling organizations. When several polls are conducted within a single time unit, as often is the case, the average of the polls is assigned. Missing cases are interpolated. Because presidential approval is fluid and can change rapidly, measuring presidential approval on a yearly, semi-annual, quarterly, or even a monthly period is not theoretically sound. Approval may readily fluctuate and aggregation may mask short-term change. For example, the timing of polls may cluster around events such as an election, state of the union address, or national crisis. When presidential approval is measured as an average over an entire quarter, then the aggregate measure may be disproportionately weighted toward the time of the event. Generally, a smaller unit is more suited for accurately tracking the fluidity of presidential approval.

Survey wording. Surveys and wording are as follows: Princeton Survey Research Associates, "Do you approve or disapprove of the way [first & last name] is handling his job as President?"; Washington Post, "Do you approve or disapprove of the way [first & last name] is handling his job as President?"; Yankelovich Partners, "In general, do you approve or disapprove of the way President [last and/or first name] is handling his job as President?"; Gallup, "Do you approve or disapprove of the way [first & last name] is handling his job as President?"; Los Angeles Times, "Do you approve or disapprove of the way [first & last name] is handling his job as President?"; Mitofsky International, "How do you feel about the way [first & last name] is handling his job as President?"; New York Times, "Do you approve or disapprove of the way [first & last name] is handling his job as President?"

Wording differences. Although question wording is quite similar, concerns may arise about the preface "in general" (appearing in the Yankelovich Partners poll) providing more stable responses and cueing respondents about long-term, rather than short-term, influences on presidents' evaluations. Within the roughly two-week unit of analysis often there are several polls averaged for our summary statistic; wording differences potentially could systematically inflate (or deflate) our approval statistic. A comparison has been made of the approval responses from the Yankelovich poll and, in each case, its nearest temporal-wise neighbor poll. The Yankelovich poll is fairly consistent in being within 3 points of the nearest neighbor polls. Additionally, the Yankelovich poll is about evenly split in the number of times it overestimates and underestimates its nearest neighbor. By this measure, we feel confident that the cueing effect due to the question preface is not consequential.

Response rate. On average, 12.6 percent of those polled fall within the "don't know" response category (and have a standard deviation of 3.7 percentage points). The don't know response rate is an average of 7.13 percent for in-party identifiers with a standard deviation of 3.5 percentage points. The don't know response rate is an average of 14.8 percent for out-party identifiers with a standard deviation of 5.2 percentage points. There is minor variation in the don't know response rate among the polls. However, there is no pattern to the mix of polls averaged within our two-week units of analyses, and we feel confident that this mitigates any systematic effect, if any at all, variation in don't know responses may have.

Measure of approval volatility. Approval is the percentage of individuals surveyed responding in the affirmative. Plotted as a trend line are the approval scores at 96 points across the term. This trend line is smoothed using T4253H smoothing (standard SPSS for windows). At any given point, the distance in percentage points between the smoothed line and the actual approval trend line provides the residual. The residual is plotted for each unit of analysis and appears as spikes in Figures 2 and 3. The average volatility for any given quarter is calculated as the average of these residuals. For illustrative purposes, a smoothed line through these residuals is placed in the Figures.

Measure of approval buoyancy. Approval buoyancy for any given president is the difference between that president's approval score and the average approval score of the seven remaining presidents. This difference is calculated for each president at 96 points across the term. Thus, A[B.sub.t1...t96] = AS1 - {([Sigma] AS2 ... AS8) / 7}, where A[B.sub.t1...t96] is approval buoyancy at any of the 96 points in the term, AS1 is the approval score for the first of the modern presidents, and AS8 is the approval score for the last of the modern presidents. AB may be calculated using either in- or out-party approval scores.

Measure of approval polarization. Approval polarization for any given president is the absolute value of the approval difference between the in- and out-party approval relative to the average approval difference for the seven remaining presidents. This difference is calculated for each president at 96 points across the term. Thus, AD[1.sub.t1...t96] = [absolute value of IPA1 - OPA1], where AD[1.sub.t1...t96] is approval difference for the first of the modern presidents at any of the 96 points in the term, IPA1 is the in-party approval for this president, and OPA1 is the out-party approval for this president. And thus, AP[1.sub.t1...t96] = AD1 - {([Sigma] AD2 ... AD8) / 7}, where AP[1.sub.t1...t96] is approval polarization for the first of the modern presidents and AD2 ... AD8 are the approval difference scores for the remaining seven presidents.

Rank of Presidents by First Term Approval Volatility and Partisanship

 In-Party Out-Party

High range (D) Clinton 2.185 (D) Carter 2.056
 (D) Carter 1.633 (R) Eisenhower 1.609
 (D) Johnson 1.402 (R) Bush 1.460
Middle range (R) Nixon (D) Johnson 1.399
 (R) Reagan 1.264 (R) Nixon 1.222
 (R) Eisenhower 1.066 (D) Kennedy 1.047
Low range (R) Bush 0.931 (R) Reagan 1.022
 (D) Kennedy 0.730 (D) Clinton 0.850

All average 1.314 1.333
Democrat average 1.488 1.338
Republican average 1.142 1.328


Barber, James David. 1992. The presidential character: Predicting performance in the White House. Englewood Cliffs, NJ: Prentice Hall.

Baum, Mathew A. 2002. The constituent foundations of the rally-round-the-flag phenomenon. International Studies Quarterly 46 (Summer): 263-98.

Bond, Jon R., and Richard Fleisher. 2001. The polls: Partisanship and presidential performance evaluation. Presidential Studies Quarterly 31 (September): 529-40.

Brace, Paul, and Barbara Hinckley. 1992. Follow the leader: Opinion polls and the modern presidency. New York: HarperCollins.

Burke, John P. 2000. Presidential transitions: From politics to practice. Boulder, CO: Lynne Rienner Publisher.

Cohen, Jeffrey E. 1997. Presidential responsiveness and public policy-making. Ann Arbor, MI: University of Michigan Press.

Cook, Timothy E., and Lyn Ragsdale. 1998. The president and the press: Negotiating newsworthiness at the White House. In The presidency and the political system, 5th ed., edited by Michael Nelson. Washington, DC: Congressional Quarterly Press, pp. 323-57.

Cronin, Thomas E., and Michael A. Genovese. 1998. The paradoxes of the American presidency. Oxford: Oxford University Press.

Edwards, George C. 1983. The public presidency: The pursuit of popular support. New York: St. Martin's Press.

Edwards, George C., and Stephen J. Wayne. 1985. Presidential leadership: Policy and policy making. New York: St. Martin's Press.

Edwards, George C., William Mitchell, and Reed Welch. 1995. Explaining presidential approval: The significance of issue salience. American Journal of Political Science 39 (February): 108-34.

Eisenstein, Maurice M., and Marie A. Whitting. 2000. Time and the life cycle of presidential approval: A research note. The Social Science Journal 37 (January): 27-43.

George, Alexander L. 1974. Assessing presidential character. World Politics 26 (January): 234-82.

George, Alexander L., and Juliette L. George. 1998. Presidential personality and performance. Boulder, CO: Westview Press.

Greenstein, Fred I. 2000. The presidential difference: Leadership style from FDR to Clinton. New York: The Free Press.

Grossman, Michael B., and Martha Kumar. 1981. Portraying the president. Baltimore: Johns Hopkins University Press.

Hager, Gregory L., and Terry Sullivan. 1994. President-centered and presidency-centered explanations of presidential public activity. American Journal of Political Science 38 (November): 1079-103.

Hurwitz, Jon, Mark Peffley, and Paul Raymond. 1989. Presidential support during the Iran-Contra affair. American Politics Quarterly 17 (October): 359-85.

Kentleton, John. 2002. President and nation: The making of modern America. New York: Palgrave.

Kessler, Frank. 1988. Richard M. Nixon. In Popular images of American presidents, edited by William C. Spragens. New York: Greenwood Press, pp. 499-522.

Mueller, John E. 1970. Presidential popularity from Truman through Johnson. American Political Science Review 21 (March): 18-34.

--. 1973. War, Presidents, and Public Opinion; New York: John Wiley and Sons.

Ostrom, Charles W., Jr., and Dennis M. Simon. 1989. The man in the Teflon suit? The environmental connection, political drama, and popular support in the Reagan presidency. Public Opinion Quarterly 53 (Fall): 353-87.

Ostrom, Charles W., Jr., and Renee M. Smith. 1992. Error correction, attitude persistence, and executive rewards and punishments: A behavioral theory of presidential approval. Political Analysis 4 (annual): 127-83.

Peterson, Mark. 1990. Legislating together: The White House and Capitol Hill from Eisenhower to Reagan. Cambridge: Howard University Press.

Pfiffner, James P. 1988. The strategic presidency: Hitting the ground running. Chicago: The Dorsey Press.

Ragsdale, Lyn. 1993. Vital statistics on the presidency. Washington, DC: Congressional Quarterly Press.

--. 1996. Presidential politics. Boston: Houghton Mifflin Company.

Shienbaum, Kim Ezra, and William C. Spragens. 1988. Cycles in the public perceptions of the president. In Popular images of American presidents, edited by William C. Spragens. New York: Greenwood Press, pp. 587-606.

Simon, Dennis M., and Charles W. Ostrom, Jr. 1989. The impact of televised speeches and foreign travel on presidential approval. Public Opinion Quarterly 53 (Spring): 58-82.

Stimson, James A. 1976. Public support for American presidents: A cyclical model. Public Opinion Quarterly 40 (Spring): 1-21.

Thomas, Norman C., and Joseph A. Pika. 1997. The politics of the presidency, 4th ed. Washington, DC: Congressional Quarterly Press.

Joshua G. Behr is Professor of Political Science at Old Dominion University. His most recent work includes Race, Ethnicity, and the Politics of City Redistricting published by the State University of New York Press, 2004.
Old Dominion University
COPYRIGHT 2003 Center for the Study of the Presidency
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Behr, Joshua G.
Publication:Presidential Studies Quarterly
Geographic Code:1USA
Date:Sep 1, 2003
Previous Article:The law: Walker v. Cheney: legal insulation of the vice president from GAO investigations.
Next Article:Source material: the White House and the press: news organizations as a presidential resource and as a source of pressure.

Related Articles
Missing evidence.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |