Presidents, Prosperity, and Public Opinion

Transcription

1 Presidents, Prosperity, and Public Opinion DONALD R. KINDER RESIDENTIAL power rests partly and often precariously on public approval. Widespread support in the public augments a president's ability to bargain and to persuade. Confronted with a popular president, Congress, the private sector, the bureaucracy, the executive branch itself, all become more accommodating to presidential initiative (Neustadt, 1960; Edwards, 1976). And just how does a president manage to gain or lose public support? Beginning with Mueller (1970, 1973), many researchers have sought answers to this question through time-series analysis of Gallup Poll presidential performance ratings. The final word belongs, momentarily at least, to Kernell (1978). Capitalizing on the methodological sins of his predecessors, Kernell finds that shifts in presidential popularity correspond closely to the flux of events. Dramatic international crises typically increase support for the president (as Abstract A president skillful enough, or fortunate enough, to preside over a healthy economy is rewarded with public support. This paper examines two conceptions of the individual citizen that might underlie this relationship. A president's popularity might decline when economic times are bad because citizens in effect blame him for their personal hardships the pocketbook citizen hypothesis or because they see the president as failing to cope adequately with national economic problems, quite apart from the economic dislocations of private life the sociotropic citizen hypothesis. Across a variety of tests, results from national surveys covering the Nixon, Ford, and Carter presidencies consistently supported the sociotropic hypothesis. The paper concludes by suggesting several promising explanations for the findings, and by exploring their normative implications. Donald R. Kinder is Associate Professor in the Departments of Political Science and Psychology, Yale University. Data for this paper were provided principally by the Inter-University Consortium for Political Research, University of Michigan. The consortium, of course, bears no responsibility for the interpretations or conclusions here. For their comments on an earlier version of the manuscript the author would like to thank Richard Brody, Morris Fiorina, W. Russell Neuman, Laurie Rhodebeck, Janet A. Weiss, and particularly D. Roderick Kiewiet and an anonymous referee. Public Opinion Quarterly Vol. 45: S1 by The Truttetj of CotanbU University Published by EUevier North-HoDand, Inc. 003J-362X/81AK)45-iyjl.75

2 1 DONALD R. KINDER Jimmy Carter's recent Iranian experience powerfully illustrates). War (at least in the protracted form taken by the Korean and Vietnam conflicts) erodes a president's popularity. And most important for my purpose here, when economic conditions sour, so too does the public's evaluation of the incumbent president. From Truman to Carter, when unemployment and (especially) prices have increased, support for the president has declined. 1 This paper explores two conceptions of the individual citizen that might underlie the relationship between economic conditions and presidential popularity. The traditional view maintains that political preferences reflect the circumstances and predicaments of private life financial circumstances and predicaments most of all. From this perspective, the pocketbook becomes a principal source of political support. In evaluating the president, citizens ask, in effect, "What has the president done forme lately?" and "What can I reasonably expect the president to do for me in the future?" So defined, the pocketbook voter represents a pervasive assumption regarding the motivational underpinnings of participation in political life. No one has articulated this assumption more forcefully than Neustadt (1960): "What a President should be is something most men see by light of what is happening to them. Their notions of the part a president should play, their satisfaction with the way he plays it, are affected by their private hopes and fears. Behind their judgments of performance lie the consequences in their lives" (p. 95). Thus, a president's popularity falls most precipitously when politics intrudes adversely upon matters close to home upon "paychecks, grocery bills, children's schooling, sons at war" (p. 97). 2 Although the assumption of pocketbook politics has become for many of us a comfortable habit, a plausible case can nevertheless be made for an alternative conception, one that emphasizes citizens moved by sociotropic concerns (Kinder and Kiewiet, 1979, 1981; Meehl, 1977). In arriving at political preferences, sociotropic citizens worry about the nation's predicament, not their own. Whether chronically unemployed or financially secure, sociotropic citizens 1 On this point, analysts other than Kernel], exploring different sets of economic indicators and making somewhat different technical assumptions, nevertheless reach the same conclusion. See Brody (1980); Brody and Page (1973); Frey and Schneider (1978); Haight and Brody (1977); Kenski (1977); Monroe (1978); Sigelman (1979) provides a useful annotated bibliography. 1 Neustadt is hardly a lonely advocate. The pocketbook voter, insinuated throughout much contemporary political analysis, bears more than a superficial resemblance to Downs's (1957) rational voter, and is a close cousin of the "investment voter" put forward by Popkin and his colleagues (1976). And in a largely tacit way, the pocketbook voter appears here and there in the rapidly expanding literature that attempts to estimate the political consequences of macroeconomic performance (for example, Bloom and Price, 1975; Kemell, 1978; Kramer, 1971; Monroe, 1979).

3 PBESJDENTS, PROSPERITY, AND PUBLIC OPINION 3 form political judgments by paying attention to national economic conditions. They vote the country's pocketbook, not their own. In evaluating the president, sociotropic citizens ask: "What has the president done for the nation lately?" and, "What can I reasonably expect the president to do for the nation in the future?" Notice that in principle, both conceptions of citizenship sketched here can account for the relationship between economic conditions and presidential popularity reported by Kernell and others, but they do so in radically different ways. According to pocketbook politics, support for the incumbent president diminishes during hard times because more citizens are encountering economic problems in their own lives: the president's eroding support directly reflects citizens' personal economic deprivations and grievances. By the same (pocketbook) argument, the president enjoys high ratings during prosperity because citizens translate their own financial security into greater political support. Not so, according to the sociotropic account. Given citizens moved by sociotropic motives, the incumbent president's popularity declines as the economy sours because.more people grow increasingly glum and pessimistic about national economic conditions quite apart from any personal economic frustrations. Assessments of the national economy, not the circumstances and predicaments of private life, determine presidential ratings provided by sociotropic citizens. Evidence on the importance citizens in fact ascribe to pocketbook experiences and to sociotropic judgments in evaluating the incumbent president is difficult to find, and because of the simplicity of the analysis that generates the evidence, more difficult to interpret (Fiorina, 1978; Tufte, 1978; Wides, 1976). Nowhere are the two models systematically compared. No analysis explores the possible interplay between the two between economic experiences of private life and economic judgments of national life. The state of the literature is summarized perfectly by the discussion of economic discontent presented in The American Voter (Campbell et al., 1960). In an otherwise insightful analysis, Campbell and company developed an index "economic outlook" that combined pocketbook and sociotropic concerns, thereby making comparisons between the two impossible. My purpose here is to offer a systematic assessment of the empirical respectability of the two conceptions of citizenship: pocketbook and sociotropic. Evidence for this assessment is derived from four national surveys covering the Nixon, Ford, and Carter presidencies. In each case, answers will be provided to three general questions. First, what role (if any) do citizens' assessments of their own economic predicaments play in their evaluations of the incumbent president to what extent does support for the president derive from

4 4 ' DONALD R. KINDEH the pocketbook? Second, what role (if any) is played by citizens' assessments of national economic conditions to what extent is presidential popularity determined by sociotropic concerns? And finally, what is the character of the interplay (if any) between citizens' appraisals of their own economic circumstances and their judgments about economic conditions in the society at large to what extent do sociotropic judgments represent generalizations from pocketbook experiences? Procedure This research draws principally upon national surveys undertaken by the Center for Political Studies (CPS) during the fall of 1972, 1974, 1976, and Embedded in the extensive interviews that composed each of these studies were questions that enable an analysis of the pocketbook and sociotropic perspectives on presidential popularity. These questions are collected and displayed in Table 1. The first entry in Table 1 is a summary of responses to a measure of presidential popularity the key dependent variable in most of what follows. Popularity was assessed by the standard CPS 100-point "thermometer scale," with 0 representing "very cold," 50 "no feeling at all" and 100 "very warm." (The Gallup approval question was asked in all four years as well, making parallel analyses possible. Unless otherwise noted, such analyses produced results that simply Table 1. Presidential Popularity, Pocketbook Economic Discontents, and Sodotropk Economic Judgments: Items and Marginals Presidential popularity Feeling thermometer rating Mean S.D. Pocketbook economic discontents Family Progress Would you say that you (and your family) are better off or worse off financially than you were a year ago? Better (+2) Same (0) Worse (-2) % % % % Family Expectations Now looking ahead do you think that a year from now you (and your family living here) will be better off financially, or worse off, or just about the same as now? Better ( + 2) 36.8% 19.9% 32.5% 24.4% Same (0) Worse (-2)

5 PRESIDENTS, PROSPERITY, AND PUBLIC OPINION Table 1. (Continued) Family Impact of Recession Looking at this list tell me if any of these things happened to you personally OR to anyone in your family living here during the last couple of years: Had work hours reduced Had to work a different shift Took a cut in pay Had to take a job below qualifications Was temporarily laid off Was unemployed or lost a job? None One More than one Family Unemployment Experiences Working at present, never out of work in the past year (+1) Working at present, but out of work sometime in the past year (+6) Temporarily laid off at present, but had not been out of work in the past year (+7) Temporarily laid off at present, and had been out of work in the past year (+8) Unemployed at present, but had worked sometime in the past year (+9) Unemployed at present, and had not worked anytime in the past year (+10) Sociotropic economic judgments Business Conditions Would you say at the present time business conditions are better or worse than they were a year ago? Better now (+2) Same (0) Worse now (-2) Government Performance* As to the economic policy of the government I mean steps taken to fight inflation or unemployment would you say that the government is doing a good job, a fair job, or a poor job? Good ( + 2) Fair (0) Poor (-2) Party Competence" Do you think that the problems of inflation and unemployment would be handled better by the Democrats, the Republicans, or about the same by both? President's party (+2) Same (0) Opposition party (-2) N.A. 81.4% % % N.A. N.A. 80.2% % N.A. N.A. 47.0% % % % % % N.A. 38.7% % % * In 1976 and 1978, respondents were asked separate questions about inflation and unemployment. Percentages shown here reflect an average response across questions.

6 6 DONALD R. KINDES corroborated the findings from the analysis of thermometer ratings reported here.) Table 1 also sets out four measures of pocketbook economic concerns. The first two elicit respondents' subjective appraisals of their economic predicaments, one oriented to the past, the other to the future. Family Progress measures respondents' reading of recent trends in their economic circumstances; Family Expectations asks respondents to forecast their own financial futures. Both questions were included in all four surveys. The second pair of pocketbook variables captures personal economic dislocation in more objective terms. Family Impact of Recession, carried only in the 1976 and 1978 surveys, is an index based on a battery of items that summarizes the recent unemployment (and underemployment) experiences of the respondent's family. In its place in the 1972 and 1974 surveys is Family Unemployment Experiences, which registers somewhat less comprehensively the current status and recent past employment experiences of the respondent's head of household. In the analysis that follows, these four measures (three, in any single survey) together constitute the pocketbook basis of support for the president. The sociotropic side of things is represented by three variables, again displayed in Table 1. Business Conditions measures the respondent's judgment of recent trends in business conditions in general. Government Performance taps the respondent 1 s evaluation of how the current administration is dealing with economic problems. Party Competence is a measure of the respondent's judgment of the relative ability of the two parties to manage economic problems. The reader may wonder whether sociotropic judgments, defined in this way, are anything more than merely generalizations from the economic experiences of private life. They are. The interplay between pocketbook concerns and sociotropic judgments is discussed in greater depth below. For now it is sufficient simply to make the general point that the connection between the two is in fact tenuous: the median Pearson correlation coefficient between variables across categories was.12 (range:.01 to.22). Consequently, it is neither hopeless as a practical matter nor trivial in theoretical terms to distinguish between pocketbook and sociotropic concerns as they play into support for the incumbent president. The analysis itself comes in five interlocking pieces. In the first, the problem is to estimate the additive effects on presidential popularity due to pocketbook and sociotropic economic concerns (with controls on party identification and social background characteristics). Predictions from the two models are straightforward. By the assumption of pocketbook politics, a president's popularity should depend on the economic circumstances of citizens' personal lives. All other things

7 PRESIDENTS, PROSPHUTY, AND PUBLIC OPINION 7 being equal, people whose economic conditions have deteriorated, or who express pessimistic predictions about their economic futures, or who have recently run into employment difficulties, should give the incumbent president poor marks, while their economically secure and personally optimistic counterparts evaluate the president more favorably. According to the sociotropic citizen model, in contrast, evaluations of the president should reflect citizens' assessments of national economic conditions, not the ups and downs of personal life. The second piece of the analysis estimates the interactive effects of pocketbook and sociotropic economic concerns. Do sociotropic judgments take on greater importance for evaluating the president among those citizens who are themselves experiencing economic difficulties? To this point, estimates will be provided by OLS multiple regression, with the appropriate equations estimated separately in each of the four studies. In the third section of the analysis, the task is to estimate the indirect and reciprocal links among pocketbook concerns, sociotropic judgments, and presidential popularity, again within each of the four cross-sectional surveys. Central to this analysis is the explicit provision that presidential popularity and sociotropic economic judgments may influence each other. The fourth section of analysis maintains this concentration on the causal interplay between economic concerns and presidential popularity, but from another angle. The analysis capitalizes at this point on the panel features of the CPS study. The problem here becomes one of ascertaining the impact of changes in economic discontents on changes in evaluations of President Ford. The fifth and final section then takes a brief look at the first two years of the Carter presidency, tracking fluctuations in Carter's aggregate public support against changes over time in pocketbook economic discontents and in sociotropic economic judgments. Results POCKETBOOK AND SOCIOTROPIC BASES OF PRESIDENTIAL POPULARITY: ADDITIVE EFFECTS According to the CPS surveys summarized in Table 1, all three presidents enjoyed considerable public support. Nixon in the fall of 1972 on the eve of his greatest political victory, Ford in 1974 just after assuming the presidency and again in 1976 as he was about to relinquish it, and Carter in the fall of 1978 all drew largely favorable evaluations. But each president had his share of critics and detractors as well. How much of the cross-sectional variation in presidential popularity can be traced to pocketbook economic discontents and

8 8 DONALD H. KINDER how much to sociotropic economic judgments? The machinery for this determination is provided by OLS multiple regression, with the independent or predictor variables coming in two sets, pocketbook and sociotropic, as set out in Table 1. For purposes of statistical control, this first regression equation also includes a measure of party identification, along with a standard set of demographic variables (age, sex, race, and socioeconomic status, the last a composite index composed of the respondent's education, self-identified social class, and family income). Table 2 summarizes the returns from this first set of analyses, separately for each of the four surveys. As indicated there, pocketbook and sociotropic economic concerns, together with party identification, predicted a respectable portion of the variance in presidential popularity: from 23 percent in the case of Ford in 1974 to 32 percent for Ford in And between the two classes of economic indicators, it is clear that the predictive burden was carried largely by sociotropic judgments. Table 2. Predicting Presidential Popularity from Pocketbook Economic Discontents and Sociotropic Economic Judgments Pocketbook economic discontents Family progress Family expectations Family impact of recession Family unemployment experiences Sociotropic economic judgments Business conditions Government performance Party competence Partisanship Party identification R/R 2 (N) Nixon * * 4.837*.J5/.30 (588) Ford * * * 2.231* (864) Ford * * -.925* * * 3.413*.57/. 32 (1,407) Carter * * * * 3.160*.5O/.25 (1,724) NOTE: Entry is b, the unstandardized regression coefficient. Here as elsewhere, the measure of party identification is derived from the standard CPS question series. It takes a value of +1 if the respondent identifies with the party of the president (strong identifier, weak identifier, independents "leaning" toward the president's party, all three), 0 if the respondent is a (pure) independent, and -1 if the respondent identifies with the opposition party. Estimates are based on a regression model that also included a set of demographic characteristics as additional control variables (age, race, sex, and a composite measure of socioeconomic status, comprised of the respondents education, subjective social class, and family income). Finally, the number of respondents is depressed in 1972 because in that year the interview schedule was split into two overlapping but nonidentical forms, with each administered to one-half of the full sample. The economic questions typically appeared in just one form, so the analysis is confined to one halfsample. So it goes. *p <.05.

9 PRESIDENTS, PROSPERITY, AND PUBLIC OPINION 9 This point can be made in several ways. Consider first the ragged pattern of findings on the pocketbook side of the ledger. Of the four measures of personal economic discontent included in the regression equations, just one bore a consistent relationship to popularity. Respondents who expected their family's economic fortunes to improve evaluated the president more favorably than did those whose expectations were pessimistic. The coefficient representing family expectations ran uniformly in the predicted direction, attaining statistical respectability in three of four cases. Family expectations were reliably tied to evaluations of the president, therefore, though the political ramifications of this statistically secure relationship should not be exaggerated. Averaging across the four surveys, and holding constant sociotropic economic judgments, partisanship, social background characteristics, and the other pocketbook economic discontents, respondents who expected their family finances to improve over the next year (coded +2) evaluated the president about 5 thermometer degrees more warmly than did those who expected their economic condition to worsen (coded -2). Elsewhere in Table 2 the evidence is weaker for pocketbook politics. Neither the retrospective assessments citizens offered for their own economic predicaments nor their objective experiences with unemployment and underemployment had much to do with their evaluation of the incumbent president. In each case, statistically significant relationships showed up on just one occasion, in ratings of Ford in The modesty of the effects registered by pocketbook economic discontents has nothing to do with the simultaneous presence of the sociotropic measures. Considered just by themselves, the set of personal economic discontents account for an average of less than 4 percent of the variance in presidential thermometer ratings (R 2 =.034 in 1972,.021 in 1974,.066 in 1976, and.022 in 1978). Or to make the same point in a slightly different way, the average bivariate Tau-Beta between presidential popularity and pocketbook economic discontents was.07. In contrast, citizens' assessments of national economic conditions were consistently and powerfully related to their evaluation of the incumbent president. As indicated in Table 2, a president's standing hinged substantially on citizens' judgments regarding recent trends in general business conditions, on their impressions of which party was more capable in dealing with economic problems, and especially on their assessments of how adequately the government was managing economic performance. By the estimates displayed in Table 2, with all other variables held constant, a president received roughly 12 to 20

10 10 DONALD R. KINDEH thermometer degrees more support from citizens who thought the government was doing a good job on economic problems than from those who believed the government was doing a poor job. Sociotropic and pocketbook economic concerns can be compared in another way, this time by exploiting the hierarchical possibilities of multiple regression analysis. Deleting the set of variables representing pocketbook economic discontents from the full regression model reduced the overall predictability of presidential popularity by an average of about 2 percent (2.0 percent in 1972, 1.5 percent in 1974, 2.6 percent in 1976, 2.5 percent in 1978: each figure is the difference between./? 2 produced by the full regression model and that produced by the model with the pocketbook economic discontents excluded; Cohen and Cohen, 1975). When sociotropic judgments are excluded from the regression model, the predictability of presidential popularity is diminished by about 8-9 percent or roughly four times the contribution made by pocketbook economic discontents (5.5 percent in 1972; 9.4 percent in 1974; 8.3 percent in percent in 1978). By several statistical criteria, then, support for the president is tied substantially to sociotropic economic judgments, and rather little to the circumstances of personal economic life. 3 SOCIOTROPIC-POCKETBOOK INTERACTIONS According to the preceding analysis, citizens evaluate incumbent presidents largely without regard to the economic strains and successes of their own lives. Personal frustrations and fears of an economic character do not figure much into a president's popularity at least not in any simple way. But pocketbook economic discontents might interact with sociotropic judgments: interpretations of national economic conditions might become more consequential in evaluating the president among those who are economically distressed than among those who are economically secure. In order to test this interaction hypothesis, multiplicative terms were created to represent all possible pairings of pocketbook with sociotropic variables (six multiplicative terms in 1972 and 1974, and nine in 1976 and 1978). These terms were then treated as a single set in a hierarchical regression analysis, with the statistical significance of the interaction given by the improvement in the predictability of presidential evaluation that is associated with the introduction of the multiplicative terms (Cohen and Cohen, 1975). 3 A more exotic rendition of the pocketbook hypothesis is that when things go badly in personal life, citizens polarize their evaluation of the president, either blaming him for their hardships or idealizing him in the hope that he might set things right. As it turned out, personal economic discontents had no effect at all on how extremely a president was rated (across the four surveys, average /?* =.003; p >.25 for each).

11 PRESIDENTS, PROSPERITY, AND PUBLIC OPINION II By this criterion, the interaction hypothesis finds little support. Adding the multiplicative terms to the regression equation augments the prediction of presidential evaluation almost imperceptibly: across the four studies, R 2 inches upward by an average of less than.5 percent (range =.2 percent to.6 percent, p >.25 in all four cases). People who themselves were economically burdened generally paid no greater attention to the nation's economic conditions in evaluating the president than did the economically secure. A closer inspection of the individuial multiplicative terms, however, reveals glimmers of the predicted relationship. The particular interactions vary from one study to the next, and they never reach conventional levels of statistical significance (as indexed by b, the unstandardized regression coefficient associated with each multiplicative term). Yet they all take the same (and expected) form: the impact of a particular sociotropic judgment on presidential evaluation is enhanced among those most distressed in terms of a particular pocketbook discontent. In 1972, for example, the relationship between ratings of Nixon and evaluations of government economic performance was stronger among those who expected their own financial condition to worsen than among those who expected it to improve. This interaction approached but did not achieve statistical significance (for the multiplicative term, b = -.872, F = 2.21 p >.20), and in this way, is prototypical of the findings from the other surveys. Consequently, the interaction hypothesis receives mixed support. On the one hand, personal economic discontents do not substantially augment the consequentiality of sociotropic judgments. On the other hand, a term-by-term examination of the interaction hypothesis does suggest that under some circumstances (but not others), certain pocketbook discontents (but not others) will amplify the effects of certain sociotropic judgments (but not others) on presidential evaluations. The effects are mysteriously context-dependent in their particularities, they never surpass conventional levels of statistical significance, and in practical political terms, they can be disregarded altogether. Nevertheless, the sporadic support provided the interaction hypothesis is of theoretical interest. Furthermore, the findings, checkered as they are, identify one way that pocketbook discontents might take on political significance, given a different and more dire set of economic circumstances than examined here. INDIRECT AND RECIPROCAL INFLUENCES If the economic experiences of private life were to determine how citizens interpret national economic conditions, then the pocketbook would necessarily assume a larger place in our understanding of the

12 L2 DONALD R. KINDER Famlly Progress Family Unemployment Family Expectations A n \ \._ Sociotropic Judgments Party ldentlfication-^pre8 dent a( I Popular)ty^Soclal T Policy Prefer forces that shape a president's prestige. The power of pocketbook politics would in that case be indirect from personal economic circumstances to sociotropic economic judgments to support for the president but could nevertheless be substantial. One purpose of this section of the analysis is to estimate a model of presidential popularity that takes into account the interplay between pocketbook discontents and sociotropic judgments. A second purpose is to estimate reciprocal influence in particular, the extent to which presidential evaluations and sociotropic economic judgments affect each other. Causal reciprocity is certainly plausible. Just as believing the national economy is falling apart should affect evaluation of the incumbent president, so, too, should evaluation of the president color judgments of national economic conditions. The essential question here for sociotropic politics is whether judgments of national economic conditions are only rationalizations for evaluations of the president, with no political force of their own. A model of presidential popularity that incorporates both indirect and reciprocal effects is portrayed in the top panel of Figure 1. Central to the model are two new variables. Sociotropic Judgments is an index based on averaged replies to the sociotropic questions available in any single year. Policy Preferences is an averaged index based on replies to four seven-point policy questions included in each of the four studies (concerning the role of government in providing assis- Family Family FamBy Expectations Unem- Presidential Sociotropic ployment Judgments Popularity Progress Effects' of:.16*.06.09*.17* Given by Beta, p< *.18* the standardized regression coefficient. ence Effects' on Presidential Popularity of: Sociotropic Judgments -2T -ar -.14* Race SES Sex Predicted Presidential Evaluations of Citizens Holding Sociotropic Judgments That Are. Unfavorable Favorable (-1) (+1) 63.9* 70 5" 44.5* 77.2* 47.8* 76.3* S2.8" 75.4" Figure 1. Presidential Popularity, Pocketbook Discontents, and Sociotropic Judgments: Indirect and Reciprocal Effects (TSLS)

13 PRESIDENTS, PROSPERITY, AND PUBUC OPINION U tance to blacks and other minorities; the proper balance between the rights of the accused and the demands of efficiency in the criminal justice system; whether women should be principally home makers and child raisers; and whether it is the federal government's responsibility to guarantee jobs and a decent standard of living for all citizens). The policy preference measure is included to render the system of two equations identifiable. As indicated in Figure 1, the equation for sociotropic judgments is exactly identified (Policy Preferences is assumed to affect Presidential Popularity directly but not Sociotropic Judgments). The equation for presidential popularity is overidentified (Family Progress and Family Unemployment are assumed to affect Sociotropic Judgments directly but not Presidential Popularity). Estimates of the various effects were provided by a two-stage least squares procedure (James and Singh, 1978; Johnston, 1972), summarized in the lefthand and middle panels of Figure 1. With respect to the indirect role pocketbook discontents might play in shaping presidential evaluations, the evidence in Figure 1 is mainly discouraging. To be sure, here and there are statistically significant links between personal economic discontents and sociotropic economic judgments. For example, people who reported that their own financial situation had deteriorated were also somewhat more negative about national economic conditions. Of the several possible links, this was the most consistent, appearing reliably in three of the four years. But the general pattern displayed in Figure 1 indicates a tenuous connection between personal discontents and sociotropic judgments. As a consequence, the impact on presidential popularity registered by pocketbook discontents working through sociotropic judgments necessarily approaches the vanishing point. Figure 1 also presents the coefficients relevant to the proposition that sociotropic judgments both cause and are affected by presidential popularity. Reciprocity does find some support. A president's enthusiasts tend to place a rosier construction on economic conditions than do presidential critics (Beta ranges from -.03 to -.39; none reaches conventional levels of statistical significance, but this may be traced in part to the imprecision associated with 2SLS). At the same time, citizens' assessments of the national economic scene affected their evaluation of the incumbent president, reliably so in three of four studies (Beta ranges from -.06 to -.31). Thus the sociotropic prediction fails only in the case of Nixon in 1972, a reflection, perhaps, of the apparent vitality of the economy in the fall of 1972, along with the public's preoccupation with Vietnam and crime (Kinder and Kiewiet, 1981). As the economy worsened, sociotropic judgments began to shape the president's standing in the polls power-

14 DONALD R. KINDER fully so according to the figures displayed in the righthand panel of Figure 1. As indicated there, with controls on party identification, pocketbook discontents, policy preferences, and social background, citizens who expressed pessimism about national economic conditions were substantially less pleased with the incumbent president on the order of thermometer degrees than were their counterparts whose national economic assessments were more upbeat. ACCOUNTING FOR INDIVIDUAL CHANGE IN PRESIDENTIAL EVALUATIONS The statistical problem of disentangling the reciprocal relationship between sociotropic judgments and evaluations of the president can be approached in another and complementary way, by capitalizing upon the panel feature of the CPS study. That some respondents were interviewed across this entire period permits an inquiry into how changes over time in individuals' national economic judgments influence and are influenced by their changing evaluations of the incumbent president. In practice, this investigation will be confined to the period, to the Ford presidency. CPS respondents interviewed both in 1974 and 1976 (N = 1,201) showed no change of heart in their average evaluation of Ford. In 1974, panel respondents gave the new president an average rating of 63.5 degrees on the thermometer scale; two years later, among the same respondents, Ford's evaluation registered 62.6 degrees. 4 But at the individual level, slightly less than one third of the CPS panel gave President Ford exactly the same rating in 1976 as in The remainder divided evenly between those who evaluated Ford more favorably (36 percent) and those who evaluated him less favorably (37 percent). The question here is to what extent such change can be ascribed to change in judgments about the national economic condition. Figure 2 summarizes the relevant evidence, as generated by a cross-lagged panel correlation analysis. Shown there are correlations between evaluations of Ford and respondents' assessments of government performance and party competence, all as measured in 1974 and (These are partial correlations, with party identification, age, race, sex, and socioeconomic status controlled in an effort to improve stationarity; Kenny, 1975). 4 The CPS figures correspond to other national survey data. For example, the Gallup Poll indicated that Ford's popularity declined precipitously in the very early stages of his presidency, in apparent response to the Nixon pardon. By the time CPS interviewers went into the field following the 1974 election, Ford's support had bottomed out, and then remained at roughly that level over the next two years.

15 PRESDENTS, PROSPERITY, AND PUBLIC OPINION IS 1976 Government Performance g ^ arty Competence^.38 >Party Competence NOTE: Entries are partial Pearson correlation coefficients, with party identification (three-way), age, race, sex, and socioeconomic status controlled. AU coefficients are statistically significant (p <.05). Figure 2. Ford's Popularity and Sodotropk Judgments: Cross-Lagged Panel Correlation Analysis The synchronous correlations displayed in Figure 2 indicate, not surprisingly, that evaluation of Ford was associated with judgments of how well the government was doing on economic matters and of which party was better equipped to deal with economic problems, in both 1974 and 1976, and with controls placed on party identification and on demographic characteristics (the synchronous partial correlations range from.22 to.31). Of greater interest are the cross-lagged correlations. They indicate that the magnitude of the over-time links between evaluation of Ford and sociotropic judgments on the one hand, and the complementary link between sociotropic judgments and Ford's evaluation, on the other, are both nontrivial and essentially equal. This pattern is consistent with the interpretation that judgments of national economic conditions and evaluations of the president are reciprocally related. The cross-lagged correlation results suggest, as do the cross-sectional results from 1974 and 1976, that each influences the other, and in roughly equal measure. It is important to point out that the findings from the two sections needn't have converged. That they did and so precisely adds to the sociotropic case. Interpretation of the nation's economic predicament influences and in turn is influenced by evaluation of the president. TRACKING SOCIOTROPIC JUDGMENTS AND POCKETBOOK DISCONTENTS THROUGH TIME This final section turns from the behavior of individual citizens to consider, if briefly, the behavior of the American public to presidential popularity in its aggregate form. Figure 3 shows Carter's Gallup approval ratings (percentage approving percentage disapproving) from his inauguration through the early months of As indicated there, Carter launched his presidency amidst substantial public sup-

16 DONALD R. KINDEB 7J 76/77 77/78 78/7 9 Figure 3. Carter Approval, Pocketbook Discontents, and Sodotropk Judgment: port. According to Gallup data collected during the first week of February 1977, two-thirds of the American public approved his performance, less than 10 percent disapproved, with about one-quarter withholding judgment. Carter's "honeymoon period" soon ended. By the fall of 1977 his plurality of approvers over disapprovers had shrunk to roughly 20 percent, and by the spring of 1978 had disappeared altogether. Carter's slide in the polls was interrupted decisively at only a single point, with the apparent success of the Camp David Mideast peace negotiations. Camp David aside, however, the news from the polls for Carter over this period grew steadily grimmer. Plotted alongside Carter's Gallup approval ratings in Figure 3 are quarterly data gathered in telephone interviews with national samples by the Economic Change Program of the University of Michigan's Survey Research Center. Across the Carter presidency, these surveys have fortuitously included two measures of pocketbook economic discontents and two measures of sociotropic judgments. All four are shown in Figure 3, each transformed into a difference score, reflecting the percentage of respondents who offered a favorable evaluation minus the percentage who responded unfavorably. Contrasts between the two sets are striking. In the aggregate across this period, pocketbook economic discontents hardly budged. Expectations regarding the family's economic future declined slightly between early 1977 and early 1979, while retrospective assessments of trends in family economic conditions changed not at all. In comparison, evaluations of government performance on economic matters as well as judgments of business conditions in general were substantially more volatile. Moreover, the sociotropic judgments moved in concert with changes in Carter's popularity. In February 1977, judgments of national economic conditions were on balance positive; by early 1979, -15

17 PRESIDENTS, PROSPERITY, AND PUBLIC OPINION 17 they had become predominantly negative. Such crude covariation, does not prove, of course, that aggregate shifts in sociotropic judgments are responsible for shifts in aggregate levels of presidential popularity. By itself, the evidence presented in Figure 3 is only suggestive. But it is of one piece with the individual-level findings presented throughout the paper. Together, the evidence suggests that aggregate shifts in judgments regarding national economic conditions may have aggregate political consequences for the American public's evaluations of its presidents in particular. Summary and Conclusions Support for the president fluctuates with changes in general economic conditions. A president skilled enough, or fortunate enough, to preside over a healthy economy is rewarded with public popularity and, as Neustadt reminds us, with real political power as well. The purpose of this paper was to peer underneath this finding, to ask how we might understand the relationship between changes in economic conditions and presidential popularity in terms of the political and economic decisions made by individual citizens. By the evidence presented here, the public's evaluation of recent presidents Nixon, Ford, and Carter depends substantially more on what I have called sociotropic assessments than on pocketbook experiences. When the economy falters, support for the president erodes not so much because citizens blame the president for their private hardships however vivid, immediate, and otherwise important they might be but because citizens hold the president accountable for the deterioration of national economic conditions. In evaluating the president, citizens seem to pay principal attention to the nation's economic predicament, and comparatively little to their own. 5 Why might this be so? Why in particular are personal economic 3 One possible explanation for this pattern is methodological. By this account, the ragged showing made by the pocketbook voter hypothesis reflects not the true relationship between pocketbook concerns and evaluation of the president, but rather shortcomings in the assessment of personal economic discontent. With better and more comprehensive measures, so goes the argument, pocketbook motives would be elevated in importance to their proper station. It would be foolish to argue against this claim as a matter of principle. But in practice, improvements in the measurement of personal economic discontent, while desirable, are unlikely to radically enhance the predictive power of the pocketbook. For there is substantial evidence to support the construct validity of the indicators of personal economic discontent used here (this evidence is summarized in Kinder and Kiewiet, 1981; and in Klorman, 1978). The measurement of pocketbook discontents meets customary methodological standards, therefore: it is just that the real discontent the indicators tap seldom gets expressed in evaluations of presidential performance.

18 It DONALD R. KINDER grievances so insulated from evaluations of the president? And why do they impinge so faintly on citizens' assessments of national economic conditions? 6 One profitable way to pursue these questions may be by inquiring into the ordinary citizen's understanding of economic phenomena. Naive economic theory may define one interpretative framework within which the incumbent president is understood and evaluated. Of special interest here is how citizens explain economic problems their own and the nation's. If, for example, personal economic outcomes are widely understood to have nonpolitical causes and nonpolitical solutions-if the unemployed blame an idiosyncratic showdown in the local economy; if the financially secure ascribe their own success to effort and ability; if the economically distressed look not to government for help but to their own resources-then the faint traces left by the pocketbook on presidential popularity become more understandable (Kinder and Kiewiet, 1981; Scholzman and Verba, 1979; Sniderman and Brody, 1977). By the same token, if citizens believe that the incumbent president's policies affect national economic conditions, or that the president could implement policies that would soften the pernicious consequences of national economic problems, then it is no wonder that the president's fate at the polls reflects citizens' assessments of the nation's predicament. Exploring the explanations and solutions people offer for economic problems however simple-minded or sensible or baroque they turn out to be should deepen our understanding of the interplay between economic conditions and presidential popularity. Another, and complementary way of pursuing these questions is by directly investigating citizens' understanding of the presidential role the assumptions people make regarding what an ideal president should and should not do. My assumption here is that the potency of economic conditions to shape the public's view of the incumbent president hinges in part on the public's view of an ideal president. This conjecture is supported in a preliminary way by a national survey conducted in the spring of According to that evidence, solving economic problems does indeed rank very high on the American public's roster of presidential priorities (Kinder et al., 1981). More important, the relative weight individual citizens gave to economic matters radically affected their evaluations of the president. 6 The greater importance of sociotropic judgments reported here is part of a broader pattern. For evaluations of the president, as for congressional voting, presidential voting, partisanship, and disaffection from government (Kiewiet, 1980; Kinder and Kiewiet, 1979, 1981), sociotropic judgments dominate pocketbook discontents. (Also see Scholzman and Verba, 1979).

19 PRESIDENTS, PROSPERITY, AND PUBLIC OPINION 1» Among those who believed it essential for an ideal president to solve economic problems, assessments of national economic conditions were much more consequential in evaluations of incumbent President Carter than were the assessments offered by people whose conception of an ideal president emphasized other priorities. Of course, merely the appearance of coping successfully with a problem as vast and bewildering as the economy may have enormous symbolic significance. Hiring official inflation fighters and slashing the budget may convey to some the reassuring impression of order and control at the top. In any case, the visions of an ideal president expressed by citizens are more than cultural artifacts; conceptions of the ideal may reveal the standards by which incumbent presidents are evaluated. 7 That citizens rely principally on their assessments of national economic conditions and rather little on their own circumstances in evaluating the incumbent president leads us to wonder, finally, about the quality of sociotropic judgment. One initial point should be emphasized: a sociotropic politics does not demand that citizens undertake a systematic analysis of macroeconomic theory, nor that they appreciate fully the president's economic policies (though of course the ambitious among us may try to do so). What is required is much more tractable: merely that citizens can form impressions of how the economy is performing, of what national problems seem the most pressing, and of how the incumbent president is managing. As Page (1978) has pointed out, a large share of presidential campaign rhetoric is devoted to precisely these topics. Consequently a sociotropic calculus does not demand of citizens an irrational hunger for political knowledge. But having said that, just how faithfully do assessments of national economic conditions reflect economic reality? How vulnerable are citizens to their own expectations and confusions, to distortion on the part of media, or to orchestration on the part of politicians? The questions are important: sociotropic politics contains both danger and promise for democratic control. An appraisal of the normative significance of a politics based on sociotropic calculations requires further investigation most pointedly, research on the quality of sociotropic decision making. 7 My argument regarding citizens' understanding of presidential accountability has been confined to economics. It needn't be. Whether the sociotropic hypothesis will be sustained in other domains energy, foreign policy, and so forth remains an open and empirical question, of course. But for these domains as for economics, an inquiry into citizens' theories, both problem-centered and presidency-centered, seems one propitious way to build a deeper understanding of presidents and their publics.