The Polls: Presidential Traits and Job Approval: Some Aggregate-Level Evidence
Newman, Brian, Presidential Studies Quarterly
In a previous article in this journal, Cohen (2001) introduced time series measures of public perceptions of Bill Clinton's personal characteristics. Here, I explore the political impact of these perceptions, asking whether they affect the public's evaluations of presidential job performance. I find that they do, adding aggregate-level support to existing individual-level evidence of the importance of character assessments. Finding a connection between character perceptions and job approval in the aggregate time series context helps answer questions previous studies leave unresolved, with significant implications for our understanding of presidential approval and presidential politics more generally.
Does the American public's view of the president as a person affect its evaluations of his job performance? Presidency and public opinion scholars have been exploring this question since the public's surprising reaction to the Lewinsky scandal and Bill Clinton's subsequent impeachment. Although raised in a particular and dramatic context, this is a general question and its answer holds implications not only for our understanding of the Clinton presidency but also for our study of the relationship between the public and the president more generally. Extant studies of this question find considerable evidence that assessments of the president's personal characteristics are related to evaluations of his job performance, but to date, these studies rely on individual-level data and are therefore subject to the limits of that framework. This study uses the time series measures of presidential traits that Cohen (2001) introduced to bolster these individual-level findings with aggregate-level evidence that perceptions of Bill Clinton's personal characteristics affected his job approval ratings.
Examining the connections between views of Clinton's character and his approval ratings speaks to a number of politically and theoretically significant issues. First, assessing these connections can enhance our understanding of presidential approval ratings, which occupy an important place in presidential politics and have been a subject of study among presidency scholars for over 30 years (see Gronke and Newman 2003 for a review). Most of the vast literature on presidential approval focuses on the impact of economic conditions and major international and domestic events. Consequently, our understanding of character's role in these evaluations remains limited.
Several recent works have argued for a broader view of approval ratings, one that pays attention to the role public perceptions of the president's personal traits play in these evaluations. Theoretically, Kinder and other political psychologists draw from social psychological models to argue that perceptions of personal traits are important because when relatively inattentive and uninformed individuals are asked to evaluate political leaders, they do so as they would evaluate ordinary people--evaluating leaders, in part, as people (e.g., Kinder 1986; Rahn et al. 1990; Sullivan et al. 1990). Individuals can use their views of the president as a person as hints about the aspects of the president's activities about which they are ignorant. If a person is convinced that the president is an intelligent and strong leader, she may conclude that the president must be handling his job well, even though she may not pay much attention to what the president has done. In addition to using character as an information shortcut, there is plenty of evidence that, all else equal, the public simply prefers competent and trustworthy leaders. For example, Edwards's (1983, 189-91) study of public expectations of the president found that the public "had lofty expectations for [the president's] personal behavior." More recently, as Clinton's impeachment was drawing near, the 1998 National Election Study found 67 percent agreeing with the statement that "people who run for high public office should display higher moral standards than does the average citizen." (1)
Studies of public views of the president's character have focused on traits deemed especially relevant to the presidency, especially competence and integrity, though scholars sometimes use different names for similar concepts. (2) Empirical analyses of these slippery concepts have consistently found that character perceptions are in fact linked to job approval evaluations. Gilens (1988) found that assessments of Reagan's competence and integrity were related to evaluations of his job performance in 1982, while Kinder (1986) demonstrated the same for evaluations of Reagan in 1984. Extending the period of study beyond the Reagan presidency, Greene (2001) found significant relationships between competence and integrity assessments and job approval at four time points from 1982 to 1994, and Newman (2003) found similar results over ten points from 1980 to 2000. Finding that character is consistently related to approval over 20 years, even in 1998, the year of the Lewinsky scandal, strongly suggests that these perceptions are stable and important elements in the public's evaluations of the president's performance.
Like all studies, however, these have their limits. All four employ individual-level data, showing that individuals with positive views of the president's character are more likely to approve of his job performance. However, they provide little sense of how these relationships affect aggregate approval ratings. Individual-level evidence of a link between character and approval does not necessarily imply a relationship at the aggregate level. Because approval ratings are most often reported and discussed in aggregate terms, individual-level relationships may have little political impact if they do not translate into aggregate effects. The possibility that individual-level relationships get lost in aggregation seems especially likely in the Clinton presidency. Despite Newman's (2003) finding that individuals' mostly negative perceptions of Clinton's integrity were significantly related to job evaluations in 1998, Clinton's aggregate approval ratings remained remarkably high through the Lewinsky scandal and impeachment. Thus, it is important to examine the aggregate influence of presidential character.
Exploring the role of character assessments in aggregate models of approval ratings may also enhance our understanding of the ways significant events affect approval. Presidential approval models virtually always include various events variables to capture the impact of so-called rally events. Few disagree that major events can have a dramatic effect on approval ratings, but there is no agreement about which events should matter or the process by which these events get translated into individuals' evaluations (see Brody 1991; Edwards and Swenson 1997 for a discussion of this process). In most instances, models include dummy variables indicating specific events or groups of similar events, tracking approval's response to these events. It seems possible that perceptions of the president's character may be one of the paths through which these events influence job performance evaluations. For example, a scandal may depress views of the president's integrity, thereby depressing the president's approval ratings. Although it is admittedly a small one, including direct measures of public views of the president's character may take us a step closer to understanding the effects of events on approval ratings.
In addition to speaking to questions about the public-president relationship in general, exploring the link between character and job approval during the Clinton presidency will help us better understand the surprising relationship between Clinton and the public. Few have captured the puzzle of Clinton's approval more starkly than when Renshon (2002, 169-70) asked,
how was it possible for a president who consistently lied to the public and to his own administration; who was found guilty of perjury for lying under oath while testifying in a civil suit and before a federal grand jury, and who in both cases was guilty of obstructing justice; who personally orchestrated the most massive stonewalling effort since Watergate to keep the truth of his inappropriate behavior from the public; who was believed by the public to have committed the offenses for which he was impeached; and whose behavior would not be tolerated in any CEO, professor, military commander, or anyone in a position of power and responsibility nonetheless manage to maintain high levels of public approval throughout his and our ordeal? (3)
Scholars have spilled considerable ink trying to answer this question (see Newman 2003 for several citations). It attracts attention largely because it touches on deeply important normative issues such as what constitutes good leadership and whether the public is using appropriate criteria to evaluate presidents. For example, Renshon argues that the public's reaction to the Lewinsky scandal raises significant questions about presidential accountability. After asking the question quoted above, Renshon (2002, 170) continues,
this is a serious question for a nation committed to giving its leaders wide discretion and relying on accountability to provide constraints. If citizens are now willing to grant the former without exercising the latter, the political culture that has historically provided the foundation for our democracy has surely shifted. A country with powerful executive institutions that declines to hold the occupants of those institutions accountable raises obvious and extremely troubling issues.
On the other hand, many argued that "we should judge presidents more by what they do than who they are" (Cronin and Genovese 1998, 896), reminding us that many presidents we deem great led flawed personal lives and some of the best men proved lackluster presidents (e.g., Glad 1998; Hargrove 1998; Pfiffner 1998). Although this empirical study certainly cannot answer these normative questions, it can inform normative discussion.
Aggregate-Level Character Assessments: Data and Method
As Cohen (2001) informed us, not long after Bill Clinton's inauguration, the Gallup Poll began asking Americans whether various traits describe Clinton. These assessments, asked frequently, but irregularly, constitute the best available time series of character perceptions (see Cohen's article for a detailed treatment of these measures). (4) Three of these traits, "can get things done," "honest and trustworthy," and "shares your values" were measured more often than any others and are especially relevant to the existing literature. (5) Assessments of whether Clinton "can get things done" seem quite close to the notion of competence, while opinions about whether he was "honest and trustworthy" relate well to the concept of integrity. In fact, these items are typically included in measures of competence and integrity in cross-sectional studies. The "shares your values" item is close to the integrity construct, but is different in some ways as well. Asking about values can raise moral or ethical considerations, bringing to mind values such as honesty or faithfulness to family, but may also raise other values, some of which may be more policy-oriented, such as environmentalism or protectionism. Presumably, asking about Clinton's values in the midst of various scandals and constant questioning of his personal integrity generally raises at least some consideration of his moral values. In fact, the "shares your values" and "honest and trustworthy" series correlate at .80. However, asking whether Clinton "shares your values" may also raise considerations of representation--how well Clinton represents my interests and values. In summary then, one item taps perceptions of competence, one taps integrity quite clearly, and the other taps integrity and some sense of representation.
Table 1 presents descriptive statistics for these series. As Cohen (2001) pointed out, views of Clinton as a person changed dramatically over his presidency. Each item was asked at least 20 times over the 1993-1999 period and the percentage of the public that agreed that the trait described Clinton changed by at least 18 points during this time. In fact, the percentage agreeing that Clinton was "honest and trustworthy" and "can get things done" changed by over 35 points during his term. Including these measures in a time series model of presidential approval provides a first step toward assessing the political impact of character perceptions.
Doing so requires some care, because the trait items were asked at irregular intervals and not every month, leaving many months without a measure. For instance, the "can get things done" item was asked eight times in 1994 and again in February of 1995, but not again until January of 1996, almost a year later. The Kalman smoothing technique, which Green and his colleagues describe and advocate, was used to deal with this irregular and infrequent measurement (Gerber and Green 1998; Green, Gerber, and De Boer 1999; see Baum and Kernell 2001 for an application to presidential approval).(6) This technique interpolates missing observations in a rigorous way and takes account of the uncertainty embedded in those interpolations.
Figure 1 portrays the smoothed trait series along with Gallup's presidential approval series. The public's view of Clinton as a person was clearly multidimensional, as perceptions of Clinton's ability to get things done diverged significantly from other perceptions, especially after 1996. The figure also demonstrates why so many journalists and scholars concluded that "the public makes a distinction between private and public character and between the personal and the presidential" (Jamieson and Aday 1998, 856). Especially in 1998, the year of the Lewinsky scandal, Clinton's approval ratings rose while the percentage of those who felt Clinton was "honest and trustworthy" and "shares your values" fell. The correlation between the approval and "honest and trustworthy" series is -.43, and between approval and the "shares your values" series is -.32, both significant at the .01 level.
[FIGURE 1 OMITTED]
Of course, these bivariate correlations are hardly the end of the story. Over 30 years of study has told us that other things affect approval. To assess the relationship between perceptions of Clinton's character and his approval ratings, we need to control for the effects of the economy and major events. To move to a multivariate approach, I estimated a standard model of approval, including inflation and unemployment rates, along with a series of variables denoting different types of positive and negative events as outlined in Ostrom and Simon (1985) and Newman (2002; see these articles for details on which events were included, how they were chosen, and which category they are in). (7) Following "conventional practice" among approval studies, I modeled approval by "including the lagged dependent variable ... among the regressors to capture the inertia component of presidential approval" (Nadeau et al. 1999, 123), although other estimation techniques generated similar results.
Table 2 presents the results of several models. (8) The first is a basic model excluding the character variables to establish a baseline for comparison. As piles of previous studies find, approval closely tracks unemployment and events, as the model accounts for about 87 percent of the variance in approval. Although only positive foreign events are statistically significant at the .05 level, several other events variables are significant at the. 10 level (two-tailed test), and group significance tests find that the model is significantly improved by including even the nonsignificant events variables (block F-test for event categories that are not significant individually, p = .04).
Model 2, which includes the three character variables, shows the important role character perceptions play once other factors affecting approval are controlled for. Assessments of Clinton's ability to get things done and whether he shares one's values both significantly affect approval. A ten-point increase in the percentage of people who thought Clinton could get things done would immediately translate into about three additional points of approval, while a ten-point decline in the percentage of people who thought Clinton shared their values would immediately depress approval by about four points.
In contrast, perceptions of Clinton's honesty and trustworthiness had no discernable impact on approval. It could be that assessments of Clinton's honesty and trustworthiness had the wrong sign and a large standard error because they are so highly correlated with the "shares you values" series, so I estimated the models excluding one or the other trait. As model 3 shows, even when the "shares your values" series was excluded, these perceptions still failed to register any effect. Perceptions of Clinton's honesty and trustworthiness simply appear unrelated to his aggregate approval ratings. However, when the "honest and trustworthy" series was excluded (model 4), the "shares your values" series maintained statistical significance at the .10 level (two-tailed test), although the magnitude of the effect was somewhat smaller than in model 2.
In general, these results bolster previous findings that the president's personal character matters for job performance evaluations. In fact, just as Newman (2002, 2003) found, these results suggest that the Lewinsky scandal actually took a toll on Clinton's approval. The dwindling perception that Clinton shared respondents' values significantly depressed his approval. Had the scandal not occurred, Clinton's approval probably would have been higher than it already was. Before the scandal, 49 percent of the public thought Clinton shared their values. One year after the scandal broke, only 35 percent thought Clinton shared their values, leading to a loss of about five percentage points of approval (14 * .37). Once the booming economy and positive assessments of Clinton's competence are controlled for, we see that the Lewinsky scandal hurt Clinton's approval after all.
On the other hand, over the same period, the public's view of Clinton's ability to get things done improved dramatically, from 60 percent before the scandal to 82 percent a year after it broke, translating into a gain of about six and a half points (22 * .30). Similarly, unemployment dropped by 0.4 percent over the same period, boosting approval by almost another point (-.4 * -2.14). On balance, then, improving conditions and perceptions of Clinton's ability to get things done compensated for the loss of approval resulting from declining percentages of those feeling that Clinton shared their values.
These results also tell us something about the ways events affect approval. Major events may alter public views of the president's capacities and qualifications and thereby have an indirect impact on approval. In other words, character perceptions may mediate the impact of events. The models bear this out. When character perceptions are excluded (model 1), the events as a group significantly add to the model's explanatory power (block F-test p-value is .01). However, once the character variables are included, none of the events variables are statistically significant on their own, and as a group their significance is questionable (block F-test p-values are between. 11 and. 18). In fact, when events are excluded altogether, the model does not change much (compare models 5 and 6 to 1 through 4), except the "shares your values" item retains statistical significance whether or not the "honest and trustworthy" item is included. (9) Thus, while we are a long way from a comprehensive model of how events translate into job approval, including character perceptions provides a more substantive feel to the model than simple indicators that an event took place.
Although these results shed new light on the question at hand, recall that these analyses must be taken as a first attempt to model the effects of character perceptions on aggregate approval ratings. The measures of character cover only one presidency and are too infrequent and irregular to warrant definitive statements about their effects. Caveats notwithstanding, finding effects even with poor measures, especially during a period when character perceptions appeared irrelevant, suggests quite strongly that these perceptions really do affect job approval in important ways. This should increase attention to these perceptions in discussions of presidential approval and boost the frequency and regularity of their measurement by pollsters.
These results continue to press the point that public views of the president's character help define evaluations of his job performance. As previous work has shown, the economy and major events profoundly shape the ways Americans think about their president's performance, but they are not the only factors related to approval ratings. Job approval also reflects a sense of the president's character. Where previous work showed this at the individual level, these results demonstrate the aggregate importance of character perceptions. Extant studies tell us that individuals with positive views of the president's character are more likely to approve of his job performance, but these results tell us that as public views of the president's character improve, his overall approval ratings rise. The individual-level relationships uncovered previously have aggregate, and consequently political, importance. Furthermore, we see this in the midst of a presidency when it appeared that at least some aspects of personal character were irrelevant to job approval ratings. We are beginning to see from a variety of angles that character assessments are integral elements of public thinking about the president.
Finding that character mattered for Clinton's approval ratings tells us something important about the American public. Many observers were shocked and even dismayed at what appeared to be a lack of response to the Lewinsky scandal. They echoed the confusion and frustration of Clinton's 1996 presidential opponent, Robert Dole, who asked "where's the outrage" (Renshon 2002). When examined in a multivariate context, it becomes clear that the public did in fact hold Clinton at least partially accountable for the Lewinsky affair and all it entailed. These results are hardly evidence of the outrage some conservatives were calling for, but they do suggest that character mattered more than we have so far appreciated. Thus, despite claims that the public drew a distinction between public and private, it appears that this distinction may not have been complete. It seems the public was willing to call Clinton to account for his lapses in some limited measure.
But the story is not so simple. There are some differences between these results and previous studies. Despite cross-sectional evidence that views of Clinton's integrity were related to job evaluations, views of Clinton's honesty and trustworthiness were not related to aggregate approval. Only opinions that Clinton can get things done and shares your values affected aggregate approval. Thus, we find continued evidence that public views of the president's competence are important, but mixed evidence on views of integrity. (10) That the item most clearly connected to the concept of integrity, the honest and trustworthy item, is unrelated to approval provides some support to the common claim that the public considered Clinton's "private" life irrelevant to his "public" duties.
On the other hand, the more complex "shares your values" item did affect approval. Although it is difficult to know exactly which values the measure captures, it does seem to tap some sense of Clinton's personal morality because these perceptions were highly correlated with sentiments about Clinton's honesty and trustworthiness and became increasingly negative during the Lewinsky scandal and Clinton's handling of it. Yet, the measure probably taps more, perhaps elements more related to representation or the president's "public" activities. Assessments of whether Clinton shares one's values may reflect views about how he might act on one's behalf. These ambiguities highlight the need for further conceptual clarity and empirical analysis to illuminate the nature of complex and multifaceted views of the president.
These results also raise some concerns about presidential politics. Some worry that public emphasis on character encourages presidents to focus on cultivating positive character images even at the expense of pursuing good public policy, leaving us with an "image is everything presidency" (Waterman, Wright, and St. Clair 1999). Because presidents presumably have greater control over public views of their character than they have over economic conditions and events abroad, Waterman, Wright, and St. Clair (1999, xiv) argue that rather than dealing with the substantive problems the nation is facing, presidents will often "choose the issues upon which to govern in order to promote a desired personal image," putting "the public relations cart before the policy horse." It is important to note that despite character's importance, Table 2 also shows that real economic conditions strongly affect approval. The public apparently demands results independent of its views of the president as a person.
Still, the influence of character perceptions and the significant normative issues associated with it highlights the importance of understanding what drives perceptions that the president can get things done or shares one's values. Are they grounded in substantive political, social, and economic realities or are they the product of carefully orchestrated public relations efforts, or more likely, what combination of each? To assess the implications of character's importance, future research will have to face the challenge of untangling complex relationships among political events and conditions, the ways they are portrayed in mass media and discussed in elite discourse, perceived and interpreted by the public, and integrated into its views of various aspects of the president's character and performance.
Exploring the political impact of public perceptions of presidential traits that Cohen (2001) introduced expands what we know about the link between the public and the president and what we need to learn to understand presidential politics better. This analysis is only a first step in incorporating character assessments into time series models of presidential approval. This first attempt adds to existing evidence that these assessments are politically significant and will hopefully spark further study of the public's perceptions of presidential character and their relation to presidential politics.
TABLE 1 Descriptive Statistics for Character Traits, 1993-1999 Standard Mean Deviation Min Max Obs Can get things done 57.72 12.66 41 82 25 Honest and trustworthy 45.06 8.67 24 61 37 Shares your values 44.65 5.48 35 53 19 Source: Gallup Poll, Various Dates, 1993-1999. TABLE 2 Character's Impact on Presidential Approval (1) (2) (3) Lagged approval 0.77 * 0.35 * 0.48 * (0.07) (0.12) (0.11) Inflation -1.18 -0.12 1.52 (0.94) (1.46) (1.28) Unemployment -1.62 * -2.14 * -2.05 * (0.61) (0.74) (0.76) Positive personal 2.89 -0.47 0.14 (2.79) (2.75) (2.82) Positive domestic 1.82 1.41 1.34 (1.06) (0.98) (1.01) Positive foreign 2.34 * 0.82 1.58 (1.04) (1.10) (1.08) Positive diplomatic 3.49 5.17 4.60 (2.88) (2.78) (2.85) Negative personal -1.58 -0.90 -0.94 (0.96) (0.94) (0.97) Negative domestic -2.35 -1.00 -1.45 (1.30) (1.24) (1.25) Gets things done 0.30 * 0.25 * (0.08) (0.08) Honest and trustworthy -0.13 0.00 (0.10) (0.08) Shares your values 0.37 * (0.17) Constant 24.32 * 19.24 * 20.97 * (6.02) (5.73) (5.84) Block E-test for Events (p-value) .01 .18 .11 Adj. R-squared 0.87 0.89 0.88 Q statistic 3.234.44 26.35 26.26 Q test p-value 0.00 0.82 0.82 Observations 72 72 72 (4) (5) (6) Lagged approval 0.42 * 0.30 * 0.36 * (0.11) (0.11) (0.11) Inflation 0.17 -0.23 0.23 (1.46) (1.35) (1.33) Unemployment -2.59 * -2.09 * -2.59 * (0.67) (0.73) (0.67) Positive personal 0.35 (2.71) Positive domestic 1.52 (0.99) Positive foreign 1.53 (0.99) Positive diplomatic 4.09 (2.69) Negative personal -0.62 (0.92) Negative domestic -1.08 (1.24) Gets things done 0.25 * 0.33 * 0.30 * (0.08) (0.07) (0.07) Honest and trustworthy -0.13 (0.08) Shares your values 0.21 0.42 * 0.26 * (0.13) (0.16) (0.13) Constant 20.95 * 17.63 * 19.45 * (5.64) (5.47) (5.41) Block E-test for Events (p-value) .15 Adj. R-squared 0.89 0.88 0.88 Q statistic 27.18 23.97 28.99 Q test p-value 0.79 0.89 0.71 Observations 72 72 72 OLS coefficients with standard errors are in parentheses. * Significant at 5%. The Q test evaluates whether the residuals are white noise (p-value is the probability of being white noise).
(1.) Variable number 980481 in the 1998 National Election Study.
(2.) In his initial treatment of presidential character traits, Kinder (1986, 236) identified four traits, competence, leadership, integrity, and empathy, but collapsed competence and leadership into one category and empathy and integrity into another because of high correlations among these traits. For definition purposes, I follow the convention of referring to the combination of these and other traits as "character." That is, character is the umbrella covering specific traits such as integrity and competence.
(3.) See Renshon (2002) for five footnotes supporting and explaining his claims.
(4.) McAvoy (2003) has compiled a much longer time series of personal trait items. However, he trades gains in length of the series for breadth of data because his character measure is one-dimensional. This is a real trade-off because past research consistently finds character assessments to be multidimensional (e.g., Kinder 1986; Newman 2003). Still, McAvoy's data allow for a much more general treatment than is possible in this analysis of the Clinton presidency.
(5.) As Cohen (2001, 735) notes, the trait "cares about the needs of people like you" was also asked several times (16 in all). However, the item was asked fewer than other items, varied less over time, and was highly correlated with the "gets things done" series (r = .80). Further, this series, unlike the others, appears to be endogenous to approval. Lags of approval predict the "cares about the needs of people like you" series, but lags of this trait series do not predict approval independent of lagged approval. For these reasons, I focus on the other three traits. When this fourth trait is included, similar results are obtained, but the "cares about the needs of people like you" variable is not statistically significant.
(6.) See Green's Samplemiser program, available from his website, http://research.yale.edu/vote/.
(7.) Categories include domestic, international, diplomatic, and personal events, both positive and negative. There were no negative foreign or diplomatic events during the years of study.
(8.) The first month was excluded by including lagged approval as an independent variable. The character data are only available through early 1999.
(9.) Models 5 and 6 also show that the results of the other models do not depend on the specific events coding scheme employed.
(10.) This is consistent with previous research, which consistently finds that competence assessments influence job evaluations more than do integrity assessments (e.g., Greene 2001; Newman 2003).
Baum, Matthew, and Samuel Kernell. 2001. Economic class and popular support for Franklin Roosevelt in war and peace. Public Opinion Quarterly 65 (2): 198-229.
Brody, Richard A. 1991. Assessing the president: The media, elite opinion, and public support. Stanford, CA: Stanford University Press.
Cohen, Jeffrey E. 2001. Change and stability in public assessments of personal traits, Bill Clinton, 1993-99. Presidential Studies Quarterly 31 (4): 733-41.
Cronin, Thomas E., and Michael A. Genovese. 1998. President Clinton and character questions. Presidential Studies Quarterly 28 (4): 892-97.
Edwards, George C. III. 1983. The public presidency: The pursuit of popular support. New York: St. Martin's Press.
Edwards, George C. III, and Tami Swenson. 1997. Who rallies? The anatomy of a rally event. Journal of Politics 59 (1): 200-12.
Gerber, Alan S., and Donald E Green. 1998. Rational learning and partisan attitudes. American Journal of Political Science 42 (3): 794-818.
Gilens, Martin. 1988. Gender and support for Reagan: A comprehensive model of presidential approval. American Journal of Political Science 32 (1): 19-49.
Glad, Betty. 1998. Evaluating presidential character. Presidential Studies Quarterly 28 (4): 861-72.
Green, Donald P., Alan S. Gerber, and Suzanna L. De Boef. 1999. Tracking opinion over time: A method for reducing sampling error. Public Opinion Quarterly 63 (2): 178-92.
Greene, Steven. 2001. The role of character assessments in presidential approval. American Politics Research 29 (2): 196-210.
Gronke, Paul, and Brian Newman. 2003. FDR to Clinton, Mueller to ??: A field essay on presidential approval. Political Research Quarterly 56 (4): 501-12.
Hargrove, Erwin C. 1998. Public purposes and private pursuits: Questions about Bill Clinton. Presidential Studies Quarterly 28 (4): 851-55.
Jamieson, Kathleen Hall, and Sean Aday. 1998. When is presidential behavior public and when is it private? Presidential Studies Quarterly 28 (4): 856-60.
Kinder, Donald R. 1986. Presidential character revisited. In Political Cognition, edited by Richard R. Lau and Douglas O. Sears. Hillsdale, NJ: L. Erlbaum Associates.
McAvoy, Gregory E. 2003. Substance vs. style: Distinguishing presidential job performance from personal approval. Paper presented at the Annual Meeting of the American Political Science Association, Philadelphia.
Nadeau, Richard, Richard G. Niemi, David E Fan, and Timothy Amato. 1999. Elite economic forecasts, economic news, mass economic judgments, and presidential approval. Journal of Politics 61 (1): 109-35.
Newman, Brian. 2002. Bill Clinton's approval ratings: The more things change, the more they stay the same. Political Research Quarterly 55 (4): 781-804.
--. 2003. Integrity and presidential approval, 1980-2000. Public Opinion Quarterly 67 (3): 33567.
Ostrom, Charles W., Jr., and Dennis M. Simon. 1985. Promise and performance: A dynamic model of presidential popularity. American Political Science Review 79 (2): 334-58.
Pfiffner, James E 1998. Sexual probity and presidential character. Presidential Studies Quarterly 28 (4): 881-86.
Rahn, Wendy, John H. Aldrich, Eugene Borgida, and John Sullivan. 1990. A social-cognitive model of candidate appraisal. In Information in Democratic Processes, edited by John Ferejohn and M. Kuklinski. Chicago: University of Chicago Press, pp. 136-59.
Renshon, Stanley A. 2002. The public's response to the Clinton scandals, part I: Inconsistent theories, contradictory evidence. Presidential Studies Quarterly 32 (1): 169-84.
Sullivan, John, John H. Aldrich, Eugene Borgida, and Wendy Rahn. 1990. Candidate appraisal and human nature: Man and Superman in the 1984 election. Political Psychology 11 (3): 459-84.
Waterman, Richard W., Robert Wright, and Gilbert K. St. Clair. 1999. The image-is-everything-presidency: Dilemmas in American leadership. Boulder: Westview Press.
Brian Newman is Assistant Professor of Political Science at Pepperdine University. He studies presidential politics, public opinion, and representation. His work has appeared in such journals as Legislative Studies Quarterly, Public Opinion Quarterly, and Political Research Quarterly.…
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: The Polls: Presidential Traits and Job Approval: Some Aggregate-Level Evidence. Contributors: Newman, Brian - Author. Journal title: Presidential Studies Quarterly. Volume: 34. Issue: 2 Publication date: June 2004. Page number: 437+. © 1999 Center for the Study of the Presidency. COPYRIGHT 2004 Gale Group.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.