The polls: can presidential rhetoric affect the public's economic perceptions? (Features).

Cohen, Jeffrey E. ; Hamman, John A.

A small but important literature has found that major presidential speeches may improve the level of public approval toward the president under certain conditions (Ragsdale 1984, 1987; Brace and Hinckley 1992). (1) Ragsdale (1984, 980) reports that each major speech lifts presidential approval by about 3 percentage points. In a refinement, Brace and Hinckley (1992) find that major speeches during a president's first term will boost his popularity by about 6 percentage points (p. 56) but will display no impact during the second term (p. 60). Ostrom and Simon (1989) and Simon and Ostrom (1989) argue that speeches will have an impact on approval under limited conditions and find that the public responds with greater levels of approval for the president only when an approval-enhancing event accompanies the speech. All agree, though, that when popularity boosts occur, they are short-lived, disappearing almost as fast as they arrived. (2)

This influential literature has laid an important foundation for understanding the impact of presidential speeches on public opinion but leaves important terrain on the topic unexplored. First, published studies may not have paid enough attention to how major speeches should be classified. We argue below that different types of speeches may have different impacts on public opinion. Thus, while Brace and Hinckley (1992, 95) distinguish between foreign policy and non-foreign-policy speeches, they bundle foreign policy speeches with other foreign policy activities; they do not compare the impact of foreign policy speeches on public opinion with other types of speeches, which we do below. Simon and Ostrom (1989, 76) offer a more refined categorization of speeches, five types based on their content, but they do not suggest why one would expect the different types of speeches to affect public opinion differently. In any event, they detect no differences in impact on public opinion across their speech types. Theoretically, more work needs to be devoted to conceptualizing the linkages between speech type and public opinion. We argue that foreign policy speeches will have greater impact on the public than other types, for instance, economic or domestic policy speeches, because foreign policy ones are better at portraying the president as a strong leader, a public image that is necessary for presidential leadership of public opinion.

Also, current studies focus almost exclusively on public support for the president as the dependent variable. Boosting public support is important to presidents, but it is not the only aspect of public opinion that presidents would like to influence. Speech effect studies may benefit by exploring the effects of speeches on other aspects of public opinion important to presidents. As Kernell (1993) argues, presidents also "go public" to alter public thinking about policies or to influence public impressions of the president. Hinckley (1990) stresses the symbolic aspects of presidential speech, and Cohen (1995, 1997) shows that presidential speech may be used to influence the public's agenda. Presidents may also speak to affect the public's mood, such as its orientation toward the future, its sense of optimism or pessimism, our topic in this article.

Finally, analyses may benefit from conceptualizing popularity other than as the dependent variable or ultimate end of speech making. Critics often complained that Ronald Reagan, for instance, tried to instill a "false" sense of well-being and future optimism in citizens. However, one may argue that Reagan might have calculated that an optimistic public would allow him greater latitude in policy choice and would be more likely to follow his lead. He might have further calculated that his ability to foster a sense of optimism would be greatest when he was popular with the public. Hence, instilling an atmosphere of optimism was part of his overall leadership strategy, and he used his popularity to instill such optimism.

In this article, we analyze the impact of presidential speeches on public expectations about the economy. We argue that economic expectations can be viewed as a general indicator of public optimism/pessimism about the future, a dimension of public opinion important to presidents. We classify speeches into three major types (economic, domestic, and foreign), hypothesizing that foreign policy speeches made when presidents are popular provide the best occasion for instilling a sense of public optimism about the future.

A Model of Presidential Speech Making Effects on Public Opinion

Our model is based on several widely held propositions in the literature that concern expectations for presidential leadership, the obstacles that must be overcome to provide that leadership, and the tactics that presidents may employ to surmount them. Presidents "go public" to affect public thinking about a host of political objects and subjects such as the president's own standing in the polls, policy initiatives, and evaluations of the state of the nation.

Strong forces impel such presidential leadership efforts, including public expectations for presidential leadership (Wayne 1982) and congressional expectations that the president help that branch set its agenda. Moreover, evaluations of presidents are rooted in their effectiveness as leaders and what they accomplish in office. Plus, the president's legacy and place in history depend on his leadership and ability to get his policies implemented.

The institutional context in which the president finds himself inhibits his ability to fulfill these expectations for leadership and in part underlie his efforts to mold public opinion. Jones (1994) offers a perspective for understanding the institutional context of the presidency and the implications of that context for presidential behavior. He argues that our system is a "separated" one. Separation of powers and checks and balances, which allow the legislature to block and frustrate presidential actions, are examples of that separation. So even though the public expects presidential leadership, separation between the branches makes leadership difficult and problematic for the president and limits the president's ability to direct public policy.

Thus, a gap exists between what is expected of the president and the power that he possesses (Waterman 1993). Consequently, presidents look for ways to supplement their political resources to enhance their political influence and policy leadership. One way is to mobilize public support or otherwise create a climate of public opinion favorable to their leadership efforts (Kernell 1993).

Public Regard toward the Chief Executive

The likely success of speech making as a presidential leadership strategy rests in large part upon how the public perceives the president. The visibility of the president in the public mind serves as the foundation on which presidential influence over public opinion is built, and research shows that there is a "positivity bias" in how the public views the president (Greenstein 1974; Edwards and Wayne 1997, 98-104). (3) This positivity bias derives in part from the public's desire for strong presidential leadership. Strong presidential leadership helps to dispel public fears and unease about politics, policy, and the state of the nation. Presidents develop an image of strong leadership by appearing effective, decisive, and in control (Cronin and Genovese 1998). In other words, strong leadership promotes a sense of security or well-being within the mass public (Edelman 1974; Greenstein 1974).

Relying on the president for such security also implies that the public will follow the president, everything else being equal, when the public views the president as a strong leader; when a president loses or fails to create an image of strong leadership, his ability to lead the public diminishes. As we argue below, foreign policy speeches provide better opportunities to present a strong presidential leadership image than other types of speeches.

Speech Type and the Presidential Leadership Image

In the ensuing analysis, we categorize major speeches that focus on one topic into one of three types: economic, domestic, and foreign. Each type of speech is made under different circumstances. These differing circumstances affect the president's leadership image. Foreign policy speeches provide the best opportunity to enhance a president's leadership image. Economic policy speeches tend to be offered during times of economic stress. Presidents rarely go before the nation in a major economic speech to talk about how well things are going. Although presidents are likely to emphasize what they are doing on the economic policy front, generally their motivation to speak stems from weakness in the economy. Such economic problems undermine the presidential leadership image, implying perhaps a failure of presidential policy leadership, a president not up to the task of steering the economy toward prosperity. Presidents speak about the economy to try to restore their image by showing themselves taking action and being on top of the situation. Still, economic policy speeches contain some bad news even if the president tries to underplay the bad news with pronouncements about what he will do or is doing. The mixed message of economic policy speeches limits their effectiveness as vehicles for presidential influence of the mass public.

Similarly, domestic policy speeches are often stimulated by internal crises, which again undermine the image of strong, effective presidential leadership. During the January 1978 through December 1994 period, Reagan spoke about drug problems (September 14, 1986) and his troubled Supreme Court nominee, Robert Bork (October 14, 1987). Bush spoke about drug abuse (September 5, 1989) and the Los Angeles riots (May 1, 1992). Clinton's sole major domestic policy speech during his first two years highlighted his health care reform proposal. Only one domestic policy speech, Reagan's Challenger disaster speech (January 28, 1986) does not fit the mold of relating to a major crisis, although that concerned a tragic incident.

Thus, presidents make major domestic policy speeches to demonstrate that they are aware of and on top of the issue and that they are doing something about it. But bad news stimulates them to go before the public. Similar to economic policy speeches, domestic policy ones are ambiguous in that they present a portrait of a president at once beleaguered but at the same time taking action. Consequently, presidents should not be able to move public opinion by speaking on such occasions. In effect, speaking about the economy and domestic policy usually is more about damage control than leadership of public opinion.

In contrast, foreign policy speeches unambiguously portray the president in a strong leadership role and, thus, afford a cleaner opportunity to influence public opinion. Foreign policy speeches tend to be made during international events and crises that one may consider to be rally events. Even when not directly associated with a rally event, these speeches present opportunities for the president to present himself as a strong leader taking decisive action (e.g., Reagan's September 26, 1986, speech on national security and his April 14, 1986, announcement of the air strike against Libya). (4)

Foreign policy speeches are also less likely to evoke political divisions among the nation's political leaders than domestic or economic speeches. This support behind the president (or lack of public opposition to him) reinforces the image of presidential leadership. When presidents take positions in domestic and economic speeches, they often become partisans for one point of view. Thus, we hypothesize that foreign policy speeches will influence public opinion, unlike economic and domestic policy speeches, which will not. Presidents give foreign policy speeches almost twice as often as they give major economic and domestic policy speeches. This comparative frequency suggests that presidents might be aware of the impact of such speeches on public opinion.

Popularity and the Ability to Influence Public Opinion

The president's standing with the public may affect his ability to influence public opinion. A naive model of presidential leadership might argue that the institutional position of the president alone secures his influence among the public. Our model, in contrast, argues that the institutional position of the presidency serves only as a foundation for potential presidential leadership. Context affects the president's ability to influence public opinion. Not allowing for such contextual effects is similar to saying that Jimmy Carter and Ronald Reagan were equally able to influence the public, an absurd claim.

One important context is the president's standing with the public. As one communication scholar posits, "A president will have more success in his dealings with the news media and the public when his approval rating with the public is high than he will when his approval rating is low" (Wanta 1991, 673). Presidential popularity may indicate the degree of credibility that the president has with the public. Research indicates that credible sources are more persuasive than less credible sources (Zaller 1992, 45-46). (5) Moreover, popularity may indicate that the public not only regards the president as a credible source but also respects him as a leader. Hence, we hypothesize that foreign policy speech making will be most effective when presidents are popular. Popularity is not hypothesized to affect presidential influence via economic and domestic policy speeches because those speeches do not convey a strong leadership image to begin with. In this sense, popularity is truly contextual, as we hypothesize no independent effects of popularity.

Data and Model

Our concern is with presidential ability to affect public optimism, but coming by such data is difficult. Only rarely are relevant questions directly posed to the public. In contrast, good data series exist on specific and substantively focused indicators of public optimism and future orientations. We use one such measure, public expectations of the economy, as our indicator of the public's mood with regard to the future.

While most of the literature on public attitudes toward the economy views this and other economic attitudes as being purely economic indicators, we suggest that one may also view economic expectations as an indicator of the public mood or optimism more generally. It is not likely that the public will be optimistic about the future if it feels pessimistic about the economy's future course, and a strong case can be made that economic orientation is one of the core foundations for the public's overall sense of optimism or pessimism. However, we recognize that other attributes may affect the public's overall sense of well-being and optimism. (6) Thus, we suggest that more than just economic factors might affect economic expectations because the economic expectations series is more than just a series on economic orientations. (7)

Unlike past research on economic perceptions (MacKuen, Erikson, and Stimson 1992, 1996; Norpoth 1996a, 1996b; Clarke and Stewart 1994), which uses quarterly data, we use monthly data, like past research on presidential speech effects (Ragsdale 1984, 1987; Brace and Hinckley 1992). We require monthly data because it is unlikely that presidential speech effects can be isolated in quarterly data--many other events/factors are likely to affect public opinion in the larger time aggregations. This decision restricts our data series to January 1978, when monthly readings on consumer confidence and expectations began, and ends with December 1994, a total of 204 monthly readings.

From January 1978 through December 1994, presidents issued fifty-seven major speeches. Of these, thirty-seven concentrated on foreign affairs, fifteen on economics, and five on domestic concerns. Multipurpose speeches that emphasize more than one policy are excluded. For a list of these speeches, see Ragsdale (1996, 161-63). We do not include major speeches that are forced on the president, like the state of the union address, because such speeches do not offer the president the same opportunity to use them strategically. Strategic use of speeches is a necessary component of understanding presidential attempts to influence public opinion. Forced speeches do not allow such strategic considerations to come into full consideration, and thus, presidential ability to influence public opinion is likely to be muted during such speech occasions.

Analysis

We specify a baseline equation with variables that previous studies indicate influence the public's economic expectations. Then we add the presidential speech and popularity variables. Our research strategy is conservative, looking for speech effects in addition to the impacts of other factors that affect economic expectations.

Our baseline model begins with the idea that aggregate opinion changes slowly and incrementally. Rarely will we see bursts of change. This is because of the decision process that individuals engage in when making such judgments as economic expectations. That is, individuals rely on past judgments, which are updated with new information as such information becomes available and is viewed as relevant. Through this process, the past is discounted (but not completely forgotten) and new information is added to the individual's decision foundation, which in turn affects expectations. Thus, current expectations and current economic confidence become the base upon which future orientations are built. We can model such a process as follows:

(1) [Expectations.sub.t] = f([Expectations.sub.t-1], Consumer [Sentiment.sub.t], New Information).

Two types of new information are used in updating expectations about the economy--real-world economic events and presidential speeches. Below, we model the real-world economic conditions of unemployment and inflation, which are two economic conditions that seem to most directly affect a large number of people and for which the public may be generally most aware. Research on the impact of economic conditions on public opinion and voting behavior stresses the importance of inflation and/or unemployment.

Current consumer sentiment is added to our model because those who speak from a sociotropic perspective suggest that economics may affect public judgements, not through their actual effects, but through public perceptions (Kiewiet 1983; MacKuen, Erikson, and Stimson 1992, 1996). In one conceptualization, economic perceptions (here consumer sentiment) may mediate the effects of the real economy. It is less important to us whether the real economy directly or indirectly affects economic expectations because our focus is on the impact of presidential speech. Thus, we model both types of economic factors into our estimations.

Our analysis also spans all or part of five presidential administrations and four different presidents-Carter, Reagan (1, 2), Bush, and Clinton. To account for effects unique to each administration and trend effects common to all, we add administration and year dummy variables as controls. Year 4 of each presidential term and the Carter administration are the criterion categories used in the estimation. (8)

With these considerations in mind, we specify a baseline model to account for monthly consumer economic expectations:

(2) [Expectations.sub.t] = Constant + [Expectations.sub.t-1] + Consumer Sentiment + [Unemployment.sub.t] + [Inflation.sub.t] + Reagan 1 + Reagan 2 + Bush + Clinton + Year 1 + Year 2 + Year 3 + error term.

The results of the estimation are presented in the left-hand column of Table 1. Ordinary least squares (OLS) is employed because we detected no residual error problem. Furthermore, we diagnose all variables in the equation for stationarity. (9) One variable, inflation, indicated a mild nonstationarity problem, which we corrected by first differencing (Hamilton 1994). The results show that three of the baseline variables have strong effects on economic expectations, all in the hypothesized direction: past economic expectations, current consumer sentiment, and current unemployment. (10) Also, Reagan's first term and year two of presidential administrations are statistically significant, indicating economic expectations were about 3 percent higher during Reagan's first term and about 2 percent lower during the second year in office. Overall, the baseline model estimation is strong, producing an [R.sup.2] of .90. It is against this well-specified equation that we test for any additional speech-making impacts on consumer expectations.

Major Speeches versus Types of Speeches

First, we add presidential speeches to the baseline equation without differentiating the type of speech. Presidential popularity, lagged one month, is also added into the equation, as it becomes an important variable later on. Results of this analysis are presented in the center column of Table 1. Again, appropriate estimation diagnostics are employed. Results indicate that major speeches, as an undifferentiated set, do not affect public economic expectations (b = 0.37, [s.sub.b] = 0.65).

A major part of our argument is that foreign policy speeches will have a stronger effect on expectations than either economic or domestic policy speeches. We estimated a third equation, this time inserting separate variables for each speech type (Table 2). The results are clear: neither economic nor domestic policy speeches affect public expectations, but foreign policy speeches do. Each foreign policy speech lifts public expectations by almost 2 percent (b = 1.83). We also estimated separate equations for each speech type in Table 3. Each pair of columns displays the impact of economic, foreign, or domestic policy speeches. The results are essentially the same-only foreign policy speeches (b = 1.86) matter. While not massive, the impact is signed consistent with our hypothesis and the effect is statistically significant.

The Conditioning Impact of Presidential Approval

To test the popularity context hypothesis, we multiply our speech variables by presidential popularity during the month in which the speech was made and include these multiplicative terms in our equations for major speech and the three different types of speeches. The interaction model equation takes the following general form:

(3) [Expectations.sub.t] = Constant + [Expectations.sub.t-1] + Consumer Sentiment + [Unemployment.sub.t] + [Inflation.sub.t] + Reagan 1 + Reagan 2 + Bush + Clinton + Year 1 + Year 2 + Year 3 + Trend + Presidential [Popularity.sub.t-1] + Presidential [Speech.sub.t-1] + (Type of Presidential [Speech.sub.t-1] x Popularity t-l) + error term.

The right-hand column in Table 1 reports results of the conditioning effect of popularity on major speech making. For estimation purposes, popularity, the major speech variable, plus the baseline variables are retained in the equation. The results show that the three variables-presidential popularity, presidential speech, and the interaction term-are statistically insignificant. The standard errors of major speech, popularity, and interaction term are large, indicating that none of the coefficients are statistically significant. The results from the interaction equation for the different speech types are given in the right-hand column of Table 2. Similarly, none of the coefficients have t-ratio values that exceed 1.00 except the foreign policy speech interaction term where t-ratio equals 1.80. Table 4 presents results when only one type of speech and its interaction with popularity is entered into the estimation. Results on Table 4 essentially reproduce those presented on Table 2.

What to make of these results? If we are not careful, we might, on the basis of the estimated coefficients and their standard errors, conclude popularity has no conditioning effect on the relationship between speeches and public expectations. Even for foreign policy speeches, the t-value for the interaction coefficient is a mere 1.80 and only significant at the .07 level. However, Friedrich (1982, 820) cautions that valid conclusions about the magnitude and statistical significance of interaction terms can only be obtained by examining conditional slopes and t-tests calculated within the observed ranges of variable values. "Statistically insignificant [b.sub.1]'s, [b.sub.2]'s, and [b.sub.3's] may nevertheless combine to produce statistically significant conditional effects" (p. 821). Adding a multiplicative term requires that we test the statistical significance of the conditional effect of one variable on another at a particular value of a third variable. Friedrich demonstrates this can be done with the following equation:

(4) t = [[b.sub.1] + [b.sub.3] ([X.sub.1])]/SQRT[var([b.sub.1]) + [([X.sub.2]).sup.2] x var([b.sub.3]) + 2([X.sub.2]) x cov([b.sub.1][b.sub.3])],

where [b.sub.1] is the slope of the presidential speech-type variable, [b.sub.3] is the slope of the interaction term, and [X.sub.2] is an observed value of presidential popularity. We retrace our steps following this procedure and find no statistically significant relationships between major speeches and expectations regardless of the value of presidential popularity. Popularity does not enhance the president ability to boost the public's expectations about the economy when a major speech is being delivered. Similar noneffects are found for domestic and economic policy speeches

However, popularity conditions foreign policy speech effects on public expectations about the economy. As a president's popularity approaches 50 percent, foreign policy speeches positively change expectations by 2 percent. At the zenith of popularity (i.e., popularity = 84 percent), foreign policy speeches result in more than a 4 percent shift in public expectations (b = 4.63, [s.sub.b] = 1.64). We repeat the analysis for the equations estimating each type of speech effect separately. The results are essentially the same. (11)

Conclusions

We have offered several critiques of the literature on the public opinion effects of presidential speech making. Extant literature does not distinguish among types of major presidential speeches, looks only at the impact of speech making on presidential approval to the neglect of other indicators of public thinking, and does not account for how the context of a speech may condition its impact on the public. Our article addresses each of these points.

First, we distinguish among foreign, economic, and domestic policy speeches, arguing that only foreign policy speeches will display strong impacts on public opinion because foreign policy speeches unambiguously portray the president as a strong leader. Second, we look at speech impacts on economic expectations, steering the literature away from its exclusive focus on presidential approval. Our position is that although presidents are interested in their approval levels, other aspects of public thinking, from policy positions to agenda priorities, are also of interest to the president. We argue that while economic expectations are in part an estimate of public attitudes toward the economy, they also indicate, in part, levels of public optimism about the future, a mood that may affect the president's ability to lead. Last, we argue that speech context can influence the persuasiveness of a presidential speech. In particular, we suggest that popular presidents are more persuasive than less popular presidents. In this sense, we can think of popularity not only as a dependent variable but also an intervening one. Our results support these contentions. Only foreign policy speeches affect public expectations about the economy. The popularity context conditioned the impact of foreign policy speeches. The impact of a foreign policy speech increases as presidential popularity rises. Only when presidents are popular do we observe a substantial impact of a foreign policy speech on economic expectations.

Finally, we need to place our findings into theoretical context. Although the speech effects were found to be statistically significant predictors of the public's economic expectations, what are we to make of the 4 percent boost that attends when presidents are very popular? Is this an increment worth the effort? Why do presidents persist in trying to reach the public when they obtain so little?

One reason may be that presidents misperceive the impact of speeches on public opinion, overstating the likely impact. Over time, one would expect that presidents would learn of the limited effect of going public with major speeches and thus alter that strategy. However, the presidency, due to the rotation in office of incumbents and aides, has a weak memory. Moreover, selective perception of cases when such going public paid off more handsomely, while forgetting or discounting those occasions where major speeches had little or no effect, may predispose presidents to the potential payoff of making a major speech.

Another possibility is that presidents are primarily concerned with elite perceptions of the impact of their speeches. It may be the case that Washington and other relevant elites magnify the effect of a speech on public opinion, taking a sign of small but measurable movements as an indicator of presidential momentum and ability to move the public even more if the president wants to. So how elites perceive even modest shifts in public opinion can be particularly significant. As Cohen (1997, 228) reports, similarly sized changes were recorded in 1993, when Vice President Gore faced off against Ross Perot in a televised debate about the North American Free Trade Agreement. The Washington political community read the modest shifts as indications of movement in the president's direction and also that the administration now possessed momentum on the issue.

Third, speech making is relatively easy for presidents. A large support apparatus now exists. It is geared toward churning out presidential statements in quantity. Thus, despite the opportunity costs of speaking versus doing something else, it may not cost presidents much to speak often. As Hart (1987) argues, presidents may actually find speaking easier than doing other things, such as governing. Moreover, presidents and others may think that speaking and governing are now one and the same. Thus, we may need to rethink the presidential calculus associated with speaking and other activities.

This leads to our last point. We need to develop a more theoretically rich account of why presidents speak to begin with. We need to specify better and more precisely the costs and benefits of speech making as an inducement for this type of presidential activity. By developing these points, we should arrive at a finer understanding of the connection between presidents and the public through public speech. TABLE 1 The Monthly Impact of Major Presidential Speeches on Consumer Expectations, 1978-94 Baseline Model Variable b [s.sub.b] Consumer expectations (t - 1) 0.61 ** 0.05 Consumer Sentiment Index 0.41 ** 0.06 Unemployment rate 1.16 ** 0.40 Consumer Price Index (first difference) -2.23 * 1.34 First -1.01 0.86 Second -3.01 ** 0.87 Third -1.60 0.92 Reagan 1 2.96 * 1.41 Reagan 2 0.54 1.19 Bush 1.42 0.88 Clinton 0.26 1.07 Major speech (t - 1) NA NA Popularity (t - 1) NA NA Major Speech x Popularity NA NA Constant -16.80 ** 5.53 [R.sup.2] .901 SEE 4.19 F 156 Sig. F .00 LaGrange multiplier 3.21 n 202 Major Speech Variable b [s.sub.b] Consumer expectations (t - 1) 0.59 ** 0.05 Consumer Sentiment Index 0.43 ** 0.06 Unemployment rate 0.97 ** 0.41 Consumer Price Index (first difference) -2.83 ** 1.39 First -0.79 0.96 Second -2.65 ** 0.94 Third -1.13 0.97 Reagan 1 3.92 ** 1.47 Reagan 2 1.18 1.26 Bush 1.60 1.06 Clinton 0.80 1.14 Major speech (t - 1) 0.37 0.65 Popularity (t - 1) 0.001 0.04 Major Speech x Popularity NA NA Constant -16.05 ** 5.81 [R.sup.2] .906 SEE 4.15 F 130 Sig. F .00 LaGrange multiplier 3.26 n 191 Major Speech Interaction Variable b [s.sub.b] Consumer expectations (t - 1) 0.58 ** 0.05 Consumer Sentiment Index 0.43 ** 0.06 Unemployment rate 0.94 ** 0.41 Consumer Price Index (first difference) -2.79 ** 1.40 First -0.71 0.96 Second -2.59 ** 0.94 Third -1.03 0.98 Reagan 1 4.14 ** 1.49 Reagan 2 1.23 1.27 Bush 1.64 1.06 Clinton 0.79 1.14 Major speech (t - 1) -2.50 2.98 Popularity (t - 1) -0.02 0.05 Major Speech x Popularity 0.05 0.05 Constant -15.15 ** 5.89 [R.sup.2] .906 SEE 4.15 F 121 Sig. F .00 LaGrange multiplier 2.74 n 191 Source: Speeches; Ragsdale (1996, 160-63). Ordinary least squares, * Significant at >. 10. ** Significant at > .05 (two-tailed). TABLE 2 The Monthly Impact on Consumer Expectations of Presidential Speeches Differentiating for Speech Type, 1978-94 All Speech Types Variable b [s.sub.b] Consumer expectations (t - 1) 0.58 ** 0.05 Consumer Sentiment Index 0.44 ** 0.06 Unemployment rate 1.03 ** 0.41 Consumer Price Index (first difference) -2.76 ** 1.39 First -0.56 0.96 Second -2.75 ** 0.93 Third -1.52 0.97 Reagan 1 3.89 ** 1.45 Reagan 2 1.06 1.25 Bush 1.83 1.05 Clinton 0.67 1.27 Popularity 0.00 0.04 Economic speech (t - 1) -0.65 1.36 Economic Speech x Popularity NA NA Foreign policy speech 1.83 ** 0.68 Foreign Policy Speech x Popularity NA NA Domestic policy speech 0.95 1.90 Domestic Policy Speech x Popularity NA NA Constant -16.94 ** 5.82 [R.sup.2] .910 SEE 4.08 F 117 Sig. F .00 LaGrange multiplier 1.72 n 191 All Speech Types plus Interactions Variable b [s.sub.b] Consumer expectations (t - 1) 0.57 ** 0.05 Consumer Sentiment Index 0.46 ** 0.06 Unemployment rate 0.99 ** 0.41 Consumer Price Index (first difference) -2.15 1.41 First -0.33 0.97 Second -2.67 * 0.95 Third -1.45 * 0.98 Reagan 1 4.32 ** 1.48 Reagan 2 1.24 1.38 Bush 2.06 1.34 Clinton 0.86 2.13 Popularity -0.04 0.05 Economic speech (t - 1) 11.11 10.87 Economic Speech x Popularity -0.23 0.21 Foreign policy speech -2.93 2.77 Foreign Policy Speech x Popularity 0.09 * 0.05 Domestic policy speech -13.09 10.61 Domestic Policy Speech x Popularity 0.25 0.18 Constant -15.94 5.93 [R.sup.2] .913 SEE 4.04 F 100 Sig. F .00 LaGrange multiplier 1.06 n 191 Source: Speeches; Ragsdale (1996, 160-63). Ordinary least squares, * Significant at > .10. ** Significant at > .05 (two-tailed). TABLE 3 The Monthly Impact on Consumer Expectations of Presidential Speeches Estimating Speech Types Separately, 1978-94 Economic Speeches (a) Variable b [s.sub.b] Consumer expectations (t - 1) 0.46 ** 0.05 Consumer Sentiment Index 0.52 ** 0.07 Unemployment rate 0.94 ** 0.48 Consumer Price Index (first difference) -2.69 * 1.46 First -0.43 1.18 Second -2.56 ** 1.17 Third -1.11 1.19 Reagan 1 5.97 ** 1.79 Reagan 2 2.43 1.57 Bush 2.28 1.33 Clinton 1.40 1.44 Popularity 0.02 0.05 Speech type (t - 1) -1.30 1.30 Constant -17.15 ** 6.50 [R.sup.2] .909 SEE 4.09 F NA Sig. F NA LaGrange multiplier NA n 191 Foreign Policy Speeches Variable b [s.sub.b] Consumer expectations (t - 1) 0.58 ** 0.05 Consumer Sentiment Index 0.45 ** 0.06 Unemployment rate 1.01 ** 0.40 Consumer Price Index (first difference) -2.65 ** 1.37 First -0.64 0.93 Second -2.82 ** 0.92 Third -1.54 0.96 Reagan 1 3.83 ** 1.44 Reagan 2 1.06 1.24 Bush 1.85 1.04 Clinton 0.70 1.12 Popularity -0.003 0.14 Speech type (t - 1) 1.86 ** 0.67 Constant -17.60 ** 5.72 [R.sup.2] .909 SEE 4.07 F 1.36 Sig. F 0.00 LaGrange multiplier 1.50 n 191 Domestic Policy Speeches (b) Variable b [s.sub.b] Consumer expectations (t - 1) 0.45 ** 0.05 Consumer Sentiment Index 0.52 ** 0.06 Unemployment rate 0.91 ** 0.48 Consumer Price Index (first difference) -2.47 * 1.44 First 0.63 1.18 Second 2.69 ** 1.18 Third -1.05 1.20 Reagan 1 5.99 ** 1.80 Reagan 2 2.43 1.59 Bush 2.27 1.34 Clinton 1.45 1.45 Popularity 0.10 0.15 Speech type (t - 1) 1.42 1.82 Constant -17.21 ** 6.54 [R.sup.2] .909 SEE 4.09 F NA Sig. F NA LaGrange multiplier NA n 191 Source: Speeches; Ragsdale (1996, 160-63). (a.) Under ordinary least squares (OLS) estimation, LaGrange multiplier equaled 3.74 and was significant at .0528, so equation was estimated under the assumptions of an AR1 auto-regressive process. (b.) Under OLS estimation, LaGrange multiplier equaled 3.71 and was significant at .0539, so equation was estimated under the assumptions of an AR1 auto-regressive process. OLS, * Significant at >. 10. ** Significant at > .05 (two-tailed). TABLE 4 by Speech Type with Presidential Popularity Interaction Term, 1978-94 Economic Speeches Variable b [s.sub.b] Consumer expectations (t - 1) 0.57 ** 0.05 Consumer Sentiment Index 0.43 ** 0.06 Unemployment rate 0.96 ** 0.41 Consumer Price Index (first difference) -2.92 ** 1.42 First -0.51 0.97 Second -2.58 ** 0.95 Third -1.08 0.96 Reagan 1 4.05 ** 1.47 Reagan 2 1.30 1.26 Bush 1.59 1.06 Clinton 0.79 1.14 Popularity 0.01 0.04 Speech type (t - 1) 11.13 11.04 Speech x Popularity -0.24 0.22 Constant -15.78 ** 5.85 [R.sup.2] .907 SEE 4.14 F 122 Sig. F .00 LaGrange multiplier 3.46 n 191 Foreign Policy Speeches Variable b [s.sub.b] Consumer expectations (t - 1) 0.59 ** 0.05 Consumer Sentiment Index 0.44 ** 0.06 Unemployment rate 0.96 ** 0.41 Consumer Price Index (first difference) -2.36 * 1.37 First -0.40 0.94 Second -2.51 ** 0.93 Third -1.40 0.96 Reagan 1 4.10 ** 1.44 Reagan 2 1.16 1.23 Bush 1.97 1.04 Clinton 0.65 1.11 Popularity -0.04 0.04 Speech type (t - 1) -2.87 2.73 Speech x Popularity 0.09 * 0.05 Constant -15.91 ** 5.76 [R.sup.2] .911 SEE 4.04 F 128 Sig. F .00 LaGrange multiplier 1.27 n 191 Domestic Policy Speeches Variable b [s.sub.b] Consumer expectations (t - 1) 0.58 ** 0.05 Consumer Sentiment Index 0.43 ** 0.06 Unemployment rate 1.00 ** 0.41 Consumer Price Index (first difference) -2.69 * 1.40 First -0.80 0.95 Second -2.67 ** 0.93 Third -1.07 0.96 Reagan 1 3.99 ** 1.47 Reagan 2 1.28 1.26 Bush 1.63 1.06 Clinton 0.98 1.16 Popularity 0.01 0.04 Speech type (t - 1) -11.29 10.85 Speech x Popularity 0.21 0.19 Constant -16.27 ** 5.86 [R.sup.2] .906 SEE 4.15 F 121 Sig. F .00 LaGrange multiplier 3.24 n 191 Source: Speeches; Ragsdale (1996, 160-63). Ordinary least squares, * Significant at >. 10. ** Significant at > .05 (two-tailed).

(1.) The literature on presidential rhetoric and speech making is actually quite massive. A thorough review can be found in Stuckey (1998). However, Edwards (1996, 208-9) points out in his critical review of seven major studies that there is a "lack of documentation of any kind on behalf of their [the authors] assertions regarding the effects of presidential rhetoric" on public opinion. As our concern is with such impacts, we do not deal directly with this literature.

(2.) There is also another subliterature that looks at the impact of specific presidential speeches on public opinion. That literature is more equivocal about the impact of presidential speeches, with some studies finding such effects, others finding no effect, and a few even finding negative effects. See Brody and Shapiro (1989), Edwards (1983), Kernell (1993, 168-82), Mondak (1993), Rosen (1973), Sigelman (1980), Sigelman and Sigelman (1981), and Thomas and Sigelman (1985).

(3.) Empirical support exists for the notion that the public is generally positively predisposed toward the presidency and its incumbent. Public support for the president surges about 10 points from his election margin to the first postinaugural popularity reading. As Edwards and Wayne (1997, 106) argue, "People want their new president to succeed and usually give him the benefit of the doubt." Similarly, the rally phenomenon may indicate underlying positive predispositions to the president, which become activated during times of stress and crisis (Callaghan and Virtanen 1993; Edwards and Swenson 1997). One study even found that 56 percent thought that the president should be supported even if he made the wrong decision (Kernell, Sperlich, and Wildavsky 1975,153).

(4.) In making our argument about the rallying effects of foreign policy speeches, we are well aware that rally events are rare, and rarely does public opinion surge dramatically even when so-called or conventionally defined rally events occur. This gives pause to the magnitude of expected presidential speech effects on the mass public. On rallies, see Edwards and Swenson (1997).

(5.) Popular approval of the president is admittedly not a direct measure of presidential credibility. Our argument is that popular presidents are considered by the public to be credible sources.

(6.) Crime, the state of international tensions, the rectitude of public figures, and their leadership qualities may also affect the public sense of future prospects. Collecting quality data on these matters to build an index of public orientations is beyond what exists in the available survey record. Thus, we rely on those economic expectations data arguing that they capture the major component of what we mean when we say that the public is optimistic or pessimistic.

(7.) For instance, compare the perspectives found in Haller and Norpoth (1994); MacKuen, Erikson, and Stimson (1992, 1996); Norpoth (1996a, 1996b); Clarke and Stewart (1994), and Katona (1975).

(8.) For more on this technique, see Brace and Hinckley (1991, 1993).

(9.) Dickey-Fuller unit root tests were run on each variable. Results can be obtained from the authors.

(10.) We also experimented with lagged terms for the inflation and unemployment variables, but results of those experiments indicated a weaker fit than when we use current measures of those variables.

(11.) We also utilized the F-test prescribed in Pindyck and Rubinfeld (1991, 110-12) to test for whether the interaction for foreign policy speeches and popularity contributed to the overall explanatory power of the equation. The equation takes the following form: [(ES[S.sub.R] - ES[S.sub.UR]/q]/[ES[S.sub.UR]/(N-k)], where ESS is the residual sum of squares for the restricted and unrestricted equations, q is the difference in independent variables between the two equations and the number of degrees of freedom in the numerator for the F-test, and N- k is the degrees of freedom in the denominator. For the foreign policy interaction effect, F = 5.68 with the critical value at .05 level of significance for 2 degrees of freedom in the numerator and 177 degrees of freedom in the denominator.

References

Brace, Paul, and Barbara Hinckley. 1991. The structure of presidential approval: Constraints within and across presidencies. Journal of Politics 53 (November): 993-1017.

--. 1992. Follow the leader: opinion polls and the modern presidency. New York: Basic Books.

--. 1993. Presidential activities from Truman through Reagan: Timing and impact. Journal of Politics 55 (May): 382-98.

Brody, Richard, and Catherine Shapiro. 1989. Policy failure and public support: The Iran Contra affair and public assessments of President Reagan. Political Behavior 11 (4): 353-69.

Callaghan, Karen J., and Simo Virtanen. 1993. Revised models of the "rally phenomenon": The case of the Carter presidency. Journal of Politics 55 (August): 756-64.

Clarke, Harold D., and Marianne C. Stewart. 1994. Prospections, retrospections, and rationality: The "bankers" model of presidential approval reconsidered. American Journal of Political Science 38 (November): 1104-23.

Cohen, Jeffrey E. 1995. Presidential rhetoric and the public agenda. American Journal of Political Science 39 (February): 87-107.

--. 1997. Presidential responsiveness and public policy-making: The public and the policies that presidents choose. Ann Arbor: University of Michigan Press.

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

Edelman, Murray. 1974. The politics of persuasion. In Choosing the president, edited by J. D. Barber, 149-73. Englewood Cliffs, NJ: Prentice Hall.

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

--. 1996. Presidential rhetoric: What difference does it make? In Beyond the rhetorical presidency, edited by M. Medhurst, 199-217. College Station: Texas A&M University Press.

Edwards, George C., III, and Tami Swenson. 1997. Who rallies? The anatomy of a rally event. Journal of Politics 59 (February): 200-12.

Edwards, George C., III, and Stephen J. Wayne. 1997. Presidential leadership: Politics and policy making. 4th ed. New York: St. Martin's.

Friedrich, Robert J. 1982. In defense of multiplicative terms in multiple regression equations. American Journal of Political Science, 26:797-833.

Greenstein, Fred I. 1974. What the president means to Americans: Presidential "choice" between elections. In Choosing the president, edited by J. D. Barber, 121-47. Englewood Cliffs, NJ: Prentice Hall.

Haller, Brandon H., and Helmut Norpoth. 1994. Let the good times roll: The economic expectations of American voters. American Journal of Political Science 38:625-50.

Hamilton, James D. 1994. Time series analysis. Princeton, NJ: Princeton University Press.

Hart, Roderick P. 1987. The sound of leadership: Presidential communication in the modern age. Chicago: University of Chicago Press.

Hinckley, Barbara 1990. The symbolic presidency: How presidents portray themselves. New York: Routledge.

Jones, Charles O. 1994. The presidency in a separated system. Washington, DC: Brookings Institution.

Katona, George. 1975. Psychological economics. New York: Elsevier.

Kernell, Samuel. 1993. Going public: New strategies of presidential leadership. 2d ed. Washington, DC: Congressional Quarterly.

Kernell, Samuel, Peter Sperlich, and Aaron Wildavsky. 1975. Public support for presidents. In Perspectives on the presidency, edited by A. Wildavsky, 148-81. Boston: Little, Brown.

Kiewiet, D. Roderick. 1983. Macroeconomics and micropolitics: The electoral effects of economic issues. Chicago: University of Chicago Press.

MacKuen, Michael B., Robert Erikson, and James A. Stimson. 1992. Peasants or bankers? The American electoral and the U.S. economy. American Political Science Review 86:597-611.

--. 1996. Comment. Journal of Politics 58:793-801.

Mondak, Jeffrey J. 1993. Source cues and policy approval: The cognitive dynamics of public support for the Reagan agenda. American Journal of Political Science 37:186-212.

Norpoth, Helmut. 1996a. Presidents and the prospective voter. Journal of Politics 58:776-92.

--. 1996b. Rejoinder. Journal of Politics 58:802-5.

Pindyck, Robert S., and Daniel L. Rubinfeld. 1981. Econometric models and economic forecasts. 2d ed. New York: McGraw-Hill.

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 53:353-87.

Ragsdale, Lyn. 1984. The politics of presidential speechmaking, 1949-1980. American Political Science Review 78:971-84.

--. 1987. Presidential speechmaking and the public audience: Individual presidents and group attitudes. Journal of Politics 49:704-36.

--. 1996. Vital statistics on the presidency. Washington, DC: Congressional Quarterly.

Rosen, Corey M. 1973. A test of presidential leadership of public opinion: The split-ballot technique. Polity 6:282-90.

Sigelman, Lee. 1980. Gauging the public response to presidential leadership. Presidential Studies Quarterly 10:427-33.

Sigelman, Lee, and Carol K. Sigelman. 1981. Presidential leadership of public opinion: From benevolent leader to kiss of death? Experimental Study of Politics 7 (3): 1-22.

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

Stuckey, Mary E. 1998. The rhetorical presidency: Deepening vision, widening exchange. In Communication yearbook 21, edited by M. Roloff and G. Paulson, 40541. Thousand Oaks, CA: Sage.

Thomas, Dan, and Lee Sigelman. 1985. Presidential identification and policy leadership: Experimental evidence on the Reagan case. In The presidency and public policy making, edited by G. C. Edwards III, S. A. Shull, and N. C. Thomas, 3749. Pittsburgh, PA: University of Pittsburgh Press.

Wanta, Wayne. 1991. Presidential approval ratings as a variable in the agenda-building process. Journalism Quarterly 68:672-79.

Waterman, Richard W., ed. 1993. The presidency reconsidered. Itasca, IL: F.E. Peacock.

Wayne, Stephen J. 1982. Expectations of the president. In The president and the public, edited by Doris Graber, 17-38. Philadelphia: Institute for the Study of Human Affairs.

Zaller, John. 1992. The nature and origins of mass opinion. New York: Cambridge University Press.

Jeffrey E. Cohen is professor of political science at Fordham University. His most recent book, Presidential Responsiveness and Public Policy-Making, received the 1998 Richard E. Neustadt Award of the Presidency Research Group of the American Political Science Association.

John A. Hamman is associate professor of political science at Southern Illinois University at Carbondale. He is the coauthor (with Jeffrey E. Cohen) of "Presidential Ideology and the Public Mood, 1956-1994" in The Presidency, edited by Diane Heath and Lori Cox Han (State University of New York Press forthcoming) and of "Reelection and Congressional Support: Presidential Motives in Distributive Politics (American Politics Quarterly 1997), coauthor (with Jon R. Bond, Jeffrey F. Cohen, and Richard Fleisher) of "State-Level Presidential Approval and Senatorial Supports' (Legislative Studies Quarterly 2000), and author of "Bureaucratic Accommodation of Congress and the President: Elections and the Distribution of Federal Assistance" (Political Research Quarterly 1993).