Citation
Assessing presidential behavior in economic downturns and election years from 1961 through 1988

Material Information

Title:
Assessing presidential behavior in economic downturns and election years from 1961 through 1988
Creator:
Smaha, Joseph K
Publication Date:
Language:
English
Physical Description:
vii, 95 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Economic inflation ( jstor )
Economic models ( jstor )
Foreign affairs ( jstor )
Gross national product ( jstor )
Presidential elections ( jstor )
Saliency ( jstor )
Speeches ( jstor )
Travel ( jstor )
Unemployment ( jstor )
Voting ( jstor )
Dissertations, Academic -- Political Science -- UF
Elections -- Effect of recessions on -- United States ( lcsh )
Political Science thesis Ph.D
Presidents -- Attitudes -- United States ( lcsh )
Recessions -- Political aspects -- United States ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1994.
Bibliography:
Includes bibliographical references (leaves 89-94).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Joseph K. Smaha.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
021022882 ( ALEPH )
33002994 ( OCLC )

Downloads

This item has the following downloads:


Full Text













ASSESSING PRESIDENTIAL BEHAVIOR IN ECONOMIC DOWNTURNS
AND ELECTION YEARS FROM 1961 THROUGH 1988

















By

JOSEPH K. SMAHA


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1994


































Copyright 1994

by

Joseph K. Smaha















ACKNOWLEDGEMENTS

Many individuals have made a significant contribution to the development and execution of this research project and to all of them I owe much gratitude. First, I wish to thank the following faculty, graduate students, and personnel at the University of Florida: Dr. M. Margaret Conway, Dr. Wayne Francis, Dr. Walter Rosenbaum, Dr. Leslie Thiele, Dr. David Colburn, Dr. Carol West, Dr. Robert Emerson, Dr. Mark Yang, Dr. Robert McMahon, Debra Moody, Jeffrey Katz, Elizabeth Williams, Ken Mease, the staff at Library West (especially Government Documents), Janet Galvez, and Patricia Turner.

My appreciation is also extended to the staffs of

Senator Bob Graham, Senator Connie Mack, Congresswoman Karen Thurman, the Congressional Research Service, the Bureau of Labor Statistics, the Ronald Reagan Library, and also to Virginia Shope (Army War College Library) and Carl Galbraith (Department of Commerce).

I also wish to thank Glynova Howe (New York Times), Rob Turson (Roper Center), Dr. Gary King (Harvard University), Dr. Lyn Ragsdale (University of Arizona), Cathy Akritas (Minitab, Inc.), and Donna Woodward (SAS Institute, Inc.).









A sincere appreciation is extended to my wife Katherine and to my children Sara and Christopher for their continuous support and for having the patience to ignore me when my research did not proceed smoothly.

Lastly, I wish to acknowledge my deceased parents,

Joseph and Teresa Smaha, whose love, support, and encouragement greatly contributed to making graduate school a reality for me.

My sincerest apologies to anyone I may have forgotten. Naturally, all errors are mine.
















TABLE OF CONTENTS


page


ACKNOWLEDGEMENTS .....................................

ABSTRACT .............................................

CHAPTERS


I.
II.


III.

IV.

V.

APPENDICES

A.

B.

C.

D.


INTRODUCTION ...............................

PREVIOUS RESEARCH ON SELECTED ASPECTS OF PRESIDENTIAL BEHAVIOR .....................

Presidential Job Approval ..................
Presidential Elections .....................
Congressional Elections ....................
Leading Congress ...........................
Foreign Affairs and Travel Abroad .......... Speeches and Press Conferences .............

METHODOLOGY ................................

RESULTS ....................................

CONCLUSIONS ................................



WAR ........................................

MAJOR FOREIGN POLICY INITIATIVES ...........

VARIABLES ..................................

CORRELATION MATRIX WITH NINE ECONOMIC VARIABLES .................................


REFERENCES ...........................................

BIOGRAPHICAL SKETCH ..................................


iii vi



1


6

6
11
16 19 22 24 27 50 69



80 82 86 88 89 95















Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

ASSESSING PRESIDENTIAL BEHAVIOR IN ECONOMIC DOWNTURNS
AND ELECTION YEARS FROM 1961 THROUGH 1988 by

Joseph K. Smaha

August 1994



Committee Chair: M. Margaret Conway, Ph.D. Major Department: Political Science



This is a quantitative study of the American presidency. The subjects are six presidents, John F. Kennedy through Ronald W. Reagan. The author examines whether there is evidence to support the political axiom that presidents attempt to divert the nation's attention from economic affairs to foreign affairs when the economy is performing poorly. Two competing hypotheses are also presented and tested.

Secondary research areas are whether presidents

increase the number of foreign affairs speeches given and the number of days spent traveling abroad in a presidential election year, and whether presidents increase their pursuit








of major foreign policy initiatives when the economy is performing poorly or during a presidential election year.

The results show that there was insufficient evidence to support the political axiom even though the competing hypotheses were not accepted. Neither did the evidence support the hypotheses in our secondary research areas. A new variable, economic saliency, which measures attitudes about the national economy, performed quite well. Economic saliency is highly influenced by changes in unemployment and inflation. In turn, presidential job approval is significantly influenced by economic saliency.

The evidence suggests that economic conditions (unemployment and inflation) are extremely important to the American public and that presidents would be wise to devote greater attention to economic affairs, rather than foreign affairs, when trying to enhance their popularity or electoral chances.

Note: contact author for revisions.


vii














CHAPTER I
INTRODUCTION
Studies of the American presidency have largely been

historical and anecdotal. Whether by political scientists or historians, most past studies of the presidency analyzed single administrations or, less common, compared performance in a particular area over several presidencies. These studies were rarely driven by theory, rarely invoked statistical analysis, and were often methodologically unsound. Many other works were simply biographical or autobiographical recollections of presidents and their administrations, hence subject to extreme bias.

In part, political scientists were hesitant to develop theories and apply quantitative research techniques to the study of presidential behavior because of the small number of cases available. Only forty-one men have held the office to date.'

Because of advances made in statistical techniques, the increasing amount of information available to scholars, the growth of the political behavior subfield, and efforts within the discipline to develop good theory, political scientists have more recently begun to advocate presidential


'Forty-two presidents when Grover Cleveland is counted twice (1885-1889, 1893-1897).









2
research which "may suggest statistical patterns across presidential terms" (King and Ragsdale 1988, 5). Primarily, this study is an attempt to discern whether patterns of behavior exist in some relationships between presidential action, economic performance, and presidential elections. Secondary issues will also be addressed.

It has often been suggested by scholars, politicians, and the media that when domestic economic conditions are poor, presidents will attempt to divert the nation's attention from economic matters to foreign affairs. This is an untested political axiom, but worth examining because of its broad acceptance within the population (especially among elites) and the insight it may provide regarding presidential behavior.

A poor economy causes social, economic, and political disruptions and spurs news stories which depict the plight of the middle and lower classes, the unemployed, and the homeless. These tragedies may affect the president's job approval rating, his re-election chances, and his party's success in congressional elections, because the electorate holds the chief executive responsible for poor economic conditions.

A decrease in presidential job approval may also weaken the president's ability to promote his agenda within Congress. The result may be a further erosion of the public's confidence in their chief executive and non-cooperation









3
between the executive and legislative branches. Presidents may, therefore, attempt to shift attention from the economy (a policy area difficult to control) to foreign affairs (where events may be more easily manipulated) in expectation that their job approval rating and their legislative effectiveness will be enhanced. This kind of behavior is analogous to impression management or image control discussed in social psychological literature (see Schlenker 1980).

A theory of presidential behavior can now be articulated. If the above generalizations are correct, then as the economy deteriorates it should become a more salient issue to the public. As a result, the president's job approval rating (or popularity)2 should decrease. We would then expect the president to participate less frequently in press conferences (where he may be questioned directly about the ailing economy) so that he may avoid additional negative publicity. Conversely, we would expect the president to devote more attention to foreign affairs (a policy area in which the executive dominates) and travel abroad more frequently in an attempt to generate positive publicity for himself and his administration. We now model this theory of presidential behavior.






2Throughout this paper presidential job approval and presidential popularity are synonymous terms.











Press
Conferences
Decrease
Presidential
When the Economic Job Foreign Affairs Economy -- Saliency -- Approval -> Speeches Declines Increases Decreases Increase Travel Abroad
Increases


Figure 1. Behavioral Model


Two competing hypotheses are suggested from the discussion of the theory which deserve attention and testing. First, some may argue that when economic conditions worsen presidents will address the economy forthrightly rather than avoid it. Therefore, a dependent variable which measures the number of economic speeches given by the president will be introduced into our model. Second, the number of foreign affairs speeches given and the number of days spent traveling abroad may increase when the United States is involved in war. To test this competing hypothesis a variable which measures the presence or absence of war will be introduced.

Two other areas of interest will also be addressed in this study. One, do presidents give more foreign affairs speeches and devote more time to foreign travel in a presidential election year regardless of economic conditions? It may be that presidents use the publicity which surrounds these types of events as a method for improving their standing (or that of their party's presidential nominee) with the American public in order to enhance their electoral








5
chances. Two, do presidents increase their pursuit of major foreign policy initiatives in either economic downturns or presidential election years in order to increase their level of support with the American public? When the economy is performing poorly, presidents may announce a major foreign policy initiative in order to regain lost points in the popularity polls or to increase their standing (or that of their party's presidential nominee) in the polls during a presidential election year.



The reader should consider that a primary goal of this study is to determine whether statistical patterns of behavior exist over several presidencies. That is, do presidents behave similarly in specific situations? It is not the intent of the author to investigate patterns of behavior within singular presidencies. Similarly, controlling for the party of the president would yield too few cases to generalize from.3 We now turn to a review of some major literature regarding the American presidency.












3We are studying the behavior of three Democratic presidents and three Republican presidents.















CHAPTER II
PREVIOUS RESEARCH ON SELECTED ASPECTS
OF PRESIDENTIAL BEHAVIOR

A large body of research has been published which suggests the existence of a relationship between the economy and (1) presidential job approval, (2) the electoral chances of the president, (3) the electoral chances of the president's party in congressional elections, and (4) the president's ability to promote policy within Congress. Each of these relationships will now be discussed.



Presidential Job Approval

It is commonly believed that an economy which is performing poorly will have a negative effect upon the job approval rating of the president. And while "political scientists have concentrated on inflation, unemployment, and income as the major factors affecting economic voting and presidential popularity" (Lanoue 1988, 42), disagreement exists as to which economic indicators influence the public's evaluation of presidential job performance. Disagreement also exists concerning the measurement of these variables and whether lagging should be incorporated. These latter issues will be addressed in the methodology chapter.









7

Economic variables are not the only ones that have an effect upon presidential popularity. Many researchers would agree that contextual effects such as personality, political and personal scandals, and military conflicts, among others, play an important role (see, for example, Kenski 1977, Norpoth 1984, and Kenney and Rice 1988). For our purposes we are only concerned with economic effects.

In an early study Frey and Schneider (1978) found that unemployment, inflation, and disposable income were important indicators or measures for the public when they evaluated presidential job performance:

A one percentage point increase in the rate of
inflation decreases presidential popularity by one half to one percentage point; an increase in unemployment of one percentage point decreases popularity by somewhat more than four percentage
points; and a one percentage point increase in the growth rate of nominal consumption increases presidential popularity by about seven-tenths of a
percentage point. (Frey and Schneider 1978, 177178)

Monroe (1978) confirmed the significance of the lagged effects of inflation on popularity and found a lagged effect for military expenditures as well. The significance of military expenditures is intuitively appealing because defense spending has often been a "hot" campaign issue in presidential elections (i.e., 1980, 1988) and because of the impact that the level of military expenditures has on domestic economic performance. Interestingly, Monroe found no effects for either unemployment or real personal income.









8

Unemployment also proved unfruitful in Kernell's (1978) study; however, consumer prices (inflation) did demonstrate levels of significance. Consumer prices were also significant in Fiorina's (1981) research and his results suggest that unemployment does in fact affect presidential popularity. The research conducted by Shapiro and Conforto (1980a) also confirmed the importance of consumer prices and unemployment:

We found that changes in consumer prices and the
percentage of the work force unemployed affect the
public's perception and evaluation of economic
conditions, and that this popular assessment in
turn affects disapproval of presidential performance. (Shapiro and Conforto 1980a, 65)

Shapiro and Conforto found no similar effect for real disposable personal income (also see Kenski 1980, and Shapiro and Conforto 1980b, for a commentary and reply).

Inflation (lagged one quarter) was found by Norpoth (1984) to significantly influence presidential approval ratings; however, no strong effect was found for unemployment. In his study of economic conditions, presidential popularity, and their effects on midterm congressional elections, Abramowitz reported that:

In 1982, an election year in which economic issues were very salient, both personal financial outlook
and national economic outlook had significant
effects on evaluations of President Reagan. However, of the four economic variables, expectations
regarding the future of the national economy had
by far the strongest impact on Reagan evaluations.
(Abramowitz 1985, 37)








9

This negative evaluation of President Reagan led to significant losses for Republican incumbents who were seeking re-election to the U.S. House of Representatives. Economic conditions and their effects on presidential and congressional elections will be addressed later.

By combining inflation and unemployment into a "misery index" Ostrom and Simon (1985) found a significant and consistent impact upon popular support for the president. Brace and Hinckley (1991) also used the misery index in their research and found that it worked "to reduce presidential approval across administrations with an effect statistically significant at the .01 level" (p.1003). It is not clear, however, whether voters actually perceive these two economic indicators in the combined manner that Ostrom and Simon and Brace and Hinckley have presented them.

Kenney and Rice (1988) note that the most common

economic indicators used by political scientists are unemployment, inflation, personal income, and gross national product. In their study of presidential rankings Kenney and Rice reported that "an improving economy can contribute to a high greatness rating" by scholars (p.165). Their measure of a healthy economy was the average annual change in per capita exports. This measure was chosen because most of the economic indicators used today were not in use during the terms of our earliest presidents.








10

Some scholars disagree with the assertion that the economy affects presidential popularity. Stimson (1976) argues that presidential popularity rides an up and down cycle and is independent from presidential behavior. His model predicts a high approval rating at the beginning of the president's term, followed by a long, significant decline, with a small rally at the close of the term--but never attaining the original level.

Brace and Hinckley (1991) also found that presidential popularity rides an up and down cycle and that it is artificially inflated at the beginning of each new term. However, as previously noted, their findings show that presidential popularity is affected by economic conditions (i.e., the misery index). Brace and Hinckley contradict Stimson by demonstrating that presidential evaluations are rooted in real events and conditions rather than just time.

The presidency of Ronald Reagan has also confounded

Stimson's theory. The mean approval rating for Reagan during the first quarter of 1981 was 55.3 percent. It dropped to a low of 38 percent during the first quarter of 1983 and rebounded to 60 percent in the fourth quarter of 1984 (the end of Reagan's first term). Similarly, during the first quarter of 1985, Reagan's mean approval rating was 60.5 percent. It dropped to a low of 46 percent in the first









11

quarter of 1987 and rebounded to 57 percent in the fourth quarter of 1988 (the end of Reagan's second term).'

The strong increases at the close of President Reagan's two terms may be atypical, but they should not be dismissed as aberrant since Stimson evaluated the terms of only five presidents and the total number of men who have occupied the office is relatively small.



Presidential Elections

That economic performance during a presidential election year is an important issue for candidates and voters should not be surprising. Research has shown that a variety of economic variables may be important to voters when they elect their chief executive in November.

Fair (1978) developed an equation for predicting the Democratic share of the two-party vote for president. The data he examined spanned presidential elections from 19161976. His tests of three economic variables--the unemployment rate, the growth rate of real GNP (gross national product) per capita, and the growth rate of the GNP deflator--demonstrated that, in an election year, the unemployment rate and the growth rate of real GNP per capita "have an important effect on votes for president," though the latter performed slightly better (p.171). In his update



'All figures were calculated by the author from data presented in Edwards with Gallup (1990).








12

which included results from 1980, Fair (1982) found that "the two main factors contributing to the low predicted vote for Carter were the low growth rate (-4.9%) and the high inflation rate (8.7%)" (p.322). And when the 1984 figures were incorporated his equation still performed quite well:

The growth rate coefficient is .0102, which means
that every one percentage point increase in the
growth rate leads to a 1.02 percentage point increase in the vote share for the incumbent
party. The inflation rate coefficient is -.0034,
which means that every one percentage point
increase in the inflation rate leads to a 0.34
percentage point fall in the vote share for the
incumbent party. (Fair 1988, 170)

Again, it was the growth rate in real GNP per capita that had the strongest effect upon the popular vote. The effects of inflation were negligible.

In his evaluation of ten presidential elections between 1948 and 1984 Erikson (1989) reported that the "cumulative weighted average of annual percentage change in per capita disposable income over the previous four years" was the best predictor of the outcome (p.568). Furthermore, "the vote is determined almost entirely by the amount of prosperity that the incumbent party delivers" (p.568). Erikson is cautious, however, and reminds his audience that both candidate quality and campaign effectiveness are just as important and should not be discounted. That is, economic factors are not the sole determinants of voter choice.

Markus (1988), too, found that personal income plays an important role in voter choice. In his study of National









13

Election Studies survey data (1956-1984) Markus discovered that:

Each 1 percent real increase in per capita disposable income increases the probability of a proincumbent vote by .023 (+- .004), net. Moreover, since objective national economic conditions are
identical for every voter in a given election,
that effect translates into a 2.3 percentage point
change in the incumbent's share of the two-party
vote for each 1 percent change in real PDI.
(Markus 1988, 151)

As previously discussed a poor economy may affect the

job approval rating of presidents. As a result, presidential job approval may be a good predictor of presidential elections. Sigelman (1979) found, for those elections he analyzed,2

a clearcut positive relationship between
presidential popularity and the percentage of the popular vote won by the incumbent president in the
seven elections since 1940. The correlation
between presidential popularity and election outcomes is +.737, which indicates a very substantial
relationship. (Sigelman 1979, 533)

Additionally, Sigelman found that "an incumbent whose

last preelection popularity figure is 45 percent or less can expect to lose or to eke out the narrowest of victories in the popular vote column" (p.534).

Lewis-Beck and Rice (1982) expanded upon the work of Sigelman and reported that the popularity figure in June preceding the November election is a better predictor of the


2Election years: 1940, 1944, 1948, 1956, 1964, 1972, and 1976.








14

popular vote. Their model reduced the average predictive error and was wrong only for Harry Truman (1948).

Presidents have long known that a poor economy can lead to defeat on election day. Are you better off now than you were four years ago? has become a common campaign slogan. Tufte (1978) has shown that presidents will attempt to stimulate the economy in election years by increasing real income through transfer payments such as old age, survivors, disability, and health insurance benefits, and veterans' benefits. Furthermore, Tufte believes that administrations will pursue policies which will decrease unemployment or inflation and postpone the closing of military bases and veterans' hospitals.

Oppenheimer (1980) disagrees that these short-term increases in transfer payments stimulate the economy. He does concede, however, that politicians believe that these transfer payments affect their electoral chances. This is a crucial factor when trying to understand the relationship between presidents, the economy, and elections. Presidents believe that if the economy is not strong or steadily improving before election day, their chances for victory will be diminished. The following will demonstrate the importance presidents and presidential candidates place on strong economic performance during an election year.









15

Vice President Richard Nixon, who sought the presidency in 1960, fully understood the role economics would play in the campaign:

And as the national economy began to turn upward
in the winter and spring of 1959, the Administration's standing rose accordingly--and along with
it, the personal stock of all of us associated
with it. (Nixon 1962, 303)

Nixon was warned by economist Arthur Burns that the

economy would dip shortly before the November 1960 election. However, the Eisenhower administration chose to stay the course. Nixon reflected:

Unfortunately, Arthur Burns turned out to be a
good prophet. The bottom of the 1960 dip did come
in October and the economy started to move up
again in November--after it was too late to affect
the election returns. In October, usually a month of rising employment, the jobless rolls increased
by 452,000. All the speeches, television broadcasts and precinct work in the world could not
counteract that one hard fact. (Nixon 1962, 310311)

When the economy remained sluggish during the 1976

presidential campaign Jimmy Carter pounded away at the Ford administration's failure to turn it around (see Witcover 1977). President Ford was warned that the economic recovery would stall shortly before the election and that something had to be done to stimulate the economy or there would be political repercussions. Even though the polls showed that uncertainty about the economy was hurting his campaign President Ford decided not to alter course. Ford later recalled, "I decided to gamble instead that the pause








16

wouldn't hit until after election day. . . . I lost" (Ford 1979, 429).

And again, during the presidential race of 1992, the economy would be the focus of the campaign. For James Carville, Bill Clinton's chief political strategist, one issue stood out:

THE ECONOMY, STUPID, read the sign he posted to
remind everyone that nothing else was more
important in their daily war games. (Newsweek
1992, 78)



Congressional Elections

Some research has demonstrated that economic conditions may directly affect the outcome of congressional elections-especially elections involving members of the president's party. Indirect affects have also been shown to exist through the public's evaluation of the president's job performance. It is commonly known that senators and congressmen often distance themselves from the president during presidential and congressional campaigns when the economy is performing poorly or when presidential job approval is low.

Kramer (1971), in his evaluation of votes for candidates for the U.S. House of Representatives between 1896 and 1964, found that:

Economic fluctuations, in particular, are important influences on congressional elections, with economic upturn helping the congressional candidates of the incumbent party, and economic decline benefitting the opposition. In quantitative terms, a 10% decrease in per capita real personal income









17

would cost the incumbent administration 4 or 5
percent of the congressional vote, other things
being equal. With a "swing ratio" of 2, this would
translate into a loss of around 40 House seats--a
respectable shift. (Kramer 1971, 140-141)

Kramer found no effects for either unemployment or inflation.

Tufte (1978) reported results similar to Kramer's.

Tufte also believes that midterm congressional elections are "a referendum on the performance of the president and his administration's management of the economy" (p.115). He found that:

An election-year change of 1.0 percent in real
disposable income has typically produced a change
of 1.1 percent in the national vote for congressional candidates of the in-party. . . . the difference between a mediocre and a buoyant electionyear economy has counted for shifts of three or four percentage points in the vote, which translates into an equivalent swing of 25 to 45 House
seats. (Tufte 1978, 119)

With changes in real disposable per capita income as their prime economic indicator, Hibbing and Alford (1981) discovered that the economy had a smaller impact on congressional elections than what Kramer and Tufte had discerned. Hibbing and Alford show that "changes in the economy have significant effects on the electoral fortunes of only a relatively small subset of congressmen--specifically, incumbents of the president's party" (p.434).

Not all research has been supportive of the hypothesis that economic conditions affect congressional elections. For example, Fiorina (1978), when using individual-level data









18

from responses to survey questions which asked if respondents' financial situation was getting better, worse, or stayed the same, found "little support for the traditional view that midterm elections constitute a referendum on the incumbent administration's handling of the economy" and that there was "no evidence that perceptions of the state of the country's economy and/or its personal impact on people had any relation to their congressional vote" (pp.435, 438). Fiorina suggests that voters are probably reacting to issues other than the economy when deciding whom to vote for in congressional elections and that voters probably blame the president, not members of Congress, for poor economic conditions.

Erikson (1989) confirmed Fiorina's findings. Using per capita income growth as his measure, Erikson found that "the same electorate that votes its collective pocketbook for president virtually ignores the economy when voting for Congress" (p.397).

Besides effects from the state of the economy, research demonstrates that congressional elections can be affected by the job approval ratings of the president. Kernell (1977) reports that for every nine-point change in presidential job approval the share of the vote for the president's party in congressional elections will change 1.3 or 1.4 percentage points (depending upon the party). In his study of midterm congressional elections Campbell (1985) discovered that seat









19

losses for the president's party are partially associated with the president's job approval rating:

� . . seat losses tend to be greater when the
President is unpopular at the midterm. For every
additional percentage point favorable to the President in the midterm Gallup poll, one can
expect the President's party to save about one
seat. (Campbell 1985, 1155)

In a study of the midterm congressional elections of 1974, 1978, and 1982, Abramowitz (1985) in part concluded that "economic conditions may have indirectly affected the vote through their effects on presidential evaluations" (p.37). The effects of economic conditions were strongest in 1982:

* . .President Reagan's relatively low public
evaluation appears to have reduced the Republican share of the popular vote by about 2.5 percentage points compared with what would be expected in a "normal" midterm election. (Abramowitz 1985, 41)



Leading Congress

Intuitively, it seems logical that if a president is not popular with the American public (as measured by national opinion polls) he will have difficulty getting his legislative agenda through Congress. What member of Congress wants to be associated with the policies of an unpopular chief executive? Surprisingly, the research yields mixed results.

Rivers and Rose (1985) focused their research on the size of the president's legislative agenda--the annual number of requests for action. Their findings show that








20

"a 1 percent increase in a president's public support level increases the president's legislative approval rate by approximately 1 percent (holding program size fixed)" (p.183).

Rohde and Simon (1985) studied presidential vetoes from 1945-1980. In part, they reviewed the public standing of the president (as measured by Gallup surveys) and congressional support on roll call votes. Regarding congressional attempts to override a presidential veto, Rohde and Simon report that "public support is a significant influence on both the probability of an override attempt and the probability that the attempt will be successful" (p.424).

As previously mentioned not all research supports the

hypothesis that decreases in presidential popularity lead to decreases in congressional support. Bond and Fleisher (1984) examined a sample of House and Senate roll call votes on which the president took a position from 1957-1980. Their dependent variable was presidential success in Congress3 and the independent variable was the percentage approving the president's job performance as measured by Gallup surveys. Their findings demonstrate that there are "very few instances when the President's popularity with the public is even moderately related to his influence in Congress"


3Bond and Fleisher tested two measures of presidential success in Congress: one, presidential wins and losses on roll call votes and two, the percentage of support the president received on each roll call vote.








21

(p.303). In their conclusion, Bond and Fleisher state that we should not be surprised by this finding. They suggest that Neustadt (1986) might have been right--that it is the president's ability to persuade, bargain, and compromise in his relationships with Congress which matters most.

In a similar study Zeidenstein (1985) reversed the variables to examine whether legislative success affects presidential popularity. The dependent variable was presidential popularity and the independent variable was key vote wins and losses for the president in Congress. Zeidenstein reported that:

. . . Presidential popularity on a given date has
no greater influence over whether a president wins or loses subsequent key votes than a key vote win
or loss has influence in changing a president's
popularity after the vote. In both cases, the
relationships range from weak to non-existent for most administrations in both the House and Senate.
It appears that whether a president is an able legislative leader does not directly affect his
popularity, and his popularity does not directly
affect his legislative leadership. (Zeidenstein
1985, 294)

Zeidenstein's findings may not be unexpected. Political scientists generally agree that the public is not wellinformed on political matters. If this is an accurate assessment, then we would not expect that congressional vote outcomes would influence the public's evaluations of the president's job performance.

If a poor economy causes presidential popularity to decline or leads to electoral defeats or an inability to lead Congress, do presidents then attempt to use foreign








22

affairs and travel abroad to improve their standing with the public and Congress? And, do the frequencies of presidential speechmaking and presidential press conferences change?



Foreign Affairs and Travel Abroad

By examining Gallup survey results before and after major international events Lee (1977) found that a relationship exists between these kinds of events and the president's popularity. Lee reports that a majority of the public will usually support presidential action in foreign affairs at its inception (rally 'round the flag), but that public support is often short-lived.

In consonance with our theory, Ostrom and Job (1986),

in their study of the political use of force, concluded that "the president is more prone to use force in times of economic stress" (p.557). Could this be an attempt by the president to increase his standing with the country? This may be, since they also reported that:

� . . the higher the level of presidential
approval, the more likely the president will be
concerned with the possibility that the use of
force, if unsuccessful, could reduce his personal
resources. (Ostrom and Job 1986, 557)

That war affects the president's popularity ratings and his reputation should not be surprising. This relationship is suggested in a study of presidential greatness rankings (Kenney and Rice 1988). Kenney and Rice reported that "involvement in war will move a president up 11 places in








23

the rankings on average" (p.165). War can also negatively affect presidential popularity--especially when casualties begin to mount (see Mueller 1973).

Foreign travel may also be a method for improving

presidential popularity. Darcy and Richman (1988) believe that "travel abroad is seen by many as a device Presidents can use to gain favorable attention, thereby influencing their popular standing" (p.85). They examined Gallup survey data before and after such events (1953-1980) and found that foreign travel (especially longer trips) had a significant influence on presidential popularity, but, generally, only for Republicans. Unfortunately, Darcy and Richman offered no explanation for this anomaly.

If foreign travel improves a president's standing then we would expect presidents to take more trips. Interestingly, there may be no increasing trend to take more of them. Simon and Ostrom (1989) examined presidential travel from 1953-1987 and reported that "on average, the president can be expected to take two or three foreign trips per year" (p.71). Furthermore, they found that foreign travel must be accompanied by a significant event before it impacts on presidential job approval.

Some international events and foreign trips may have

significant short-term effects upon presidential popularity. If an approval-enhancing event closely preceded an election it could alter the predicted outcome.








24

Speeches and Press Conferences

Our theory states, in part, that when the economy is performing poorly the president will devote more attention to foreign affairs in his speeches and decrease the frequency of press conferences. Research on presidential speechmaking and press conferences has produced mixed findings.

Ragsdale (1984) developed an equation for predicting when a president would give a major speech. Her definition of "major speech" is a presidential address given before a live television or radio audience. The data base included speeches from 1949-1980. Contrary to our theory Ragsdale found that "as the unemployment situation worsens, presidents are less likely to deliver a major address" and that "presidents appear to be more likely to address the nation when troops are being removed and the foreign conflict is presumably winding down" (pp.976-977).

The reader should note that the set of speeches used in the present study is significantly larger than the set used by Ragsdale because we include speeches which were not given live before a television or radio audience. Consequently, our research may yield different results (see methodology chapter).

Simon and Ostrom (1989) found no evidence that speechmaking is on the rise. In fact, they report that "presidential use of television has declined since the Johnson and








25

Nixon administrations" (p.71). Their findings also demonstrated that speeches by themselves did not enhance presidential support, but they did when "given in conjunction with a trip and an approval-enhancing event" (p.74). On average the increase in presidential approval was 4.17 rating points (Simon and Ostrom 1989, 74).

Contrary to Ragsdale (1984), Simon and Ostrom (1989)

believe that "it is the event rather than the speech that is responsible for triggering the rally effect" (p.79). Their findings give support to the theory that presidents may seek a foreign policy success or engage in foreign travel for the purpose of improving presidential popularity.

In a study of presidential press conferences Lammers (1981) found that foreign policy difficulties led to a decrease in the number of press conferences held. And, in contradiction to our theory, he reported that:

Despite their periodic denials, presidents do seem
to read the polls as they schedule [press] conferences--but more often to increase rather than
decrease [press] conference frequency as their
popularity falls. (Lammers 1981, 270)

The reader should note that research in this area is scant,4 although the findings by Lammers should not be dismissed and our research may yield different results.

Some research has demonstrated that poor economic conditions may negatively affect the president's popularity,


4For a short, nonquantitative history and assessment of presidential press conferences, see Smith (1990).








26

his electoral chances, his party's chances in congressional elections, and his ability to lead Congress. Major international events, speeches, and foreign travel may help the president improve his standing with the public. These aspects of presidential behavior have not been brought together under the umbrella of one study, nor have some of the aforementioned studies included the more recent presidents, nor has our theory been previously tested.

In the following chapter we will specifically develop the hypotheses, operationalize the variables, and discuss the method of analysis.














CHAPTER III
METHODOLOGY

This is a quasi-experimental design in which the subjects are six United States presidents, John F. Kennedy (1961) through Ronald W. Reagan (1988).' We begin with Kennedy because his administration was the first to televise press conferences on a regular basis and the number of households with televisions surpassed the 90 percent mark at roughly the same time that he took office (see chapters 3 and 4 in Kernell 1986). This is an important distinction because press conferences now became national media events and a method for presidents to communicate directly and immediately with a very large proportion of the public.

The data end with the Reagan administration because

complete information for several variables was not available for the term during which George Bush was president. Information for all variables in this study is complete for all quarters unless otherwise noted.

To reiterate, our theory states that as the economy

worsens it becomes a more salient issue to the public. As a result, the president's job approval will go down. In



'Though Reagan continued to serve as president through the first three weeks of January 1989, data for the first quarter of 1989 are omitted.








28

response, the president will give fewer press conferences and try to divert the nation's attention from domestic to foreign affairs by (1) giving more speeches which concern foreign affairs and (2) traveling abroad more frequently.

Nine independent variables which measure economic

performance are incorporated. Economic saliency and presidential job approval are intervening variables and both measure responses to survey questions. The number of press conferences held, the number of foreign affairs speeches given, and the number of days spent traveling abroad are all dependent variables.

The first step entails the regression of economic

saliency on our nine economic measures. Monetary variables are standardized in 1987 dollars.2 The following will describe each variable, its code, and source.

Economic saliency (ECSAL) is defined as the mean

quarterly percent of those who cited an economic problem in response to the survey question "What do you think is the most important problem facing the [this] country today?" An economic problem is defined as: cost of living, inflation, taxes, unemployment, balanced budget, depression, recession, interest rates, and the trade deficit, among others. These are broad, national economic problems which concern citizens as opposed to specific social problems such as welfare or


2Standardization adjusts for inflation.








29

poverty, even though the latter surely have economic repercussions.

Responses to the surveys were assigned to the quarters in which the interviews occurred. Results from interviews which spanned two quarters were discarded since we could not determine which quarter to assign them to. Responses to variations of the question were not included, nor were responses which were not from surveys of the general population.' This survey question has been asked irregularly by the major polling organizations. Of the 112 quarters in the data base, 54 show missing data for this variable (for a loss of 48 percent). The mean quarterly figure can exceed 100 percent due to multiple responses.

Source: most of the data are from responses to Gallup surveys (Gallup Organization, no date). Data for eight quarters4 are from responses to surveys conducted by CBS News/New York Times (Staff, New York Times, no date).

Nine independent variables which measure economic

performance have been chosen. The first two, unemployment and inflation, have often been used by scholars in similar studies (see review of literature) and certainly affect




3Some of the original data were from surveys of selected subgroups, rather than the general population, and were not included in our data set.

4Quarter/year: 2/1981, 3/1981, 2/1986, 4/1986, 1/1987, 4/1987, 2/1988, and 3/1988.









30

personal economic well-being. They also receive much media attention.

1. The unemployment rate (UNEMP) is a measure of those who have been unemployed fifteen weeks or longer as a percent of the total work force. This is a quarterly figure for all workers, 16 years of age or older, and it is seasonally adjusted.

Source: Citibase (1978).5

2. The inflation rate (INFL) is the mean quarterly

percent rate of change in the Consumer Price Index (CPI). The CPI is a monthly measure of the cost of a hypothetical basket of goods for all urban consumers, for all items, and is seasonally adjusted. The base period is 1982-84 = 100.6

Source: for the CPI, Citibase (1978); see footnote 5. The percent rates of change were calculated by the author using a statistical program.'

The third economic indicator chosen is the prime lending rate. Though not often used in similar studies, it is included here because of the impact the rate has on the


'The primary source for this and other economic variables is the U.S. Department of Commerce, Bureau of Economic Analysis, but it is cited in the text and bibliography as Citibase because this is the data file from which the figures were generated. Though the copyright date is 1978 the data available surpass that year.

6The CPI has not yet been calculated with a base year beyond 1982-84.
7The author used Minitab, Release 8, for the personal computer. Minitab is a registered trademark of Minitab, Inc., State College, PA.









31

purchase of large consumer goods such as automobiles, boats, housing, and real estate, and on credit card interest rates. It, too, receives a great amount of media attention.

3. The prime lending rate (PRIME) is the rate of interest charged by banks on short-term business loans to their best corporate customers, per annum. The data are recorded on a mean quarterly basis.

Source: Citibase (1978); see footnote 5.

Our next economic indicators are per capita real disposable income, per capita gross national product, and per capita national defense expenditures. Even if the average citizen is not familiar with the meaning or interpretation of these three indicators, most will perceive the repercussions when these indicators change significantly. Change in real disposable income affects the purchasing level of individuals and families. Significant decreases in the levels of GNP and national defense expenditures can lead to a sluggish economy--even recession. These economic declines can be devastating to industries, communities, and families. It is not clear from prior research (see review of literature), however, whether it is the levels or the percent rates of change in these three variables which is more significant. Therefore, both the levels and the percent rates of change will be tested (thereby creating three additional variables).









32

Data for real disposable income are available on a per capita basis. It was necessary to calculate the per capita figures for gross national product and national defense expenditures. This was done by dividing the quarterly figure for each variable by the current estimate of the population of the United States.8

4. Per capita real disposable income (INC) is the quarterly measure of the average worker's real disposable income. Real disposable income is defined as personal income less personal taxes. Personal taxes are comprised of income, property, and inheritance taxes (see McConnell 1975, 182). The quarterly figure in 1987 dollars, annual rate, is recorded.

Source: Datadisk (1984).9

5. Per capita gross national product (GNP) is the

quarterly measure of the productivity of the nation (gross national product) divided by the current population estimate. This figure is recorded in 1987 dollars.

Source: Citibase (1978); see footnote 5.

6. Per capita national defense expenditures (DEF) is the quarterly measure of national defense expenditures by




aPopulation figures for 1967 through 1988 are from the Bureau of the Census and reproduced from a data file (Data Resources, Inc.). Figures for 1961 through 1966 are from U.S. Department of Commerce, Bureau of the Census (1979).

9Even though the copyright date is 1984 the data available surpass that year.








33

the federal government in 1987 dollars divided by the current population estimate.

Source: the data for 1972 through 1988 are from U.S. Department of Commerce (1992, Table 3.11, 118-119). There are no data for defense expenditures in 1987 dollars prior to 1972, therefore, expenditures have been estimated. The method chosen" prescribes taking the actual dollar amount spent in each quarter (1961-1971) and dividing it by the implicit price deflator (1987 = 100) for the federal government for the corresponding quarter (1961-1971). Actual dollar figures were obtained from U.S. Department of Commerce (1992, Table 1.1, 1-2). The price deflator was obtained from U.S. Department of Commerce (1992, Table 7.14, line 12, 298299).

7. Rate of change in per capita real disposable income (CINC) is the quarterly percent rate of change in per capita real disposable income (see number 4 above).

Source: calculated by the author using a statistical program; see footnote 7.

8. Rate of change in per capita gross national product (CGNP) is the quarterly percent rate of change in per capita gross national product (see number 5 above).

Source: calculated by the author using a statistical program; see footnote 7.



'(As recommended by Mr. Carl Galbraith, an economist with the U.S. Department of Commerce.









34

9. Rate of change in per capita national defense

expenditures (CDEF) is the quarterly rate of change in per capita national defense expenditures (see number 6 above).

Source: calculated by the author using a statistical program; see footnote 7.

From the above variables we can now formulate our first equation:"



Economic Saliency (ECSAL) = a + bl(UNEMP) + b2(INFL) +

b3(PRIME) + b4(INC) + b5(GNP) + b6(DEF) + b7(CINC)

+ b8(CGNP) + b9(CDEF) + e



Nine hypotheses can now be tested:

H1: As the unemployment rate increases, economic saliency

increases (bl > 0).

H2: As the inflation rate increases, economic saliency

increases (b2 > 0).

H3: As the prime lending rate increases, economic saliency

increases (b3 > 0).

H4: As per capita real disposable income increases, economic saliency decreases (b4 < 0).

H5: As per capita gross national product increases, economic saliency decreases (b5 < 0).





"For all equations in this paper, a = Y-intercept, b = slope, and e = error term.









35

H6: As per capita national defense expenditures increase,

economic saliency decreases (b6 < 0).

H7: As the rate of change in per capita real disposable

income increases, economic saliency decreases (b7 < 0). H8: As the rate of change in per capita gross national

product increases, economic saliency decreases

(b8 < 0).

H9: As the rate of change in per capita national defense

expenditures increases, economic saliency decreases

(b9 < 0).

Our second intervening variable, presidential job

approval, will now be regressed on economic saliency. The president's job approval rating will be measured by tracking positive responses to a question frequently asked in national opinion surveys.

Presidential job approval (APPROV) is the mean quarterly percent of those who cited approval when asked "Do you approve or disapprove of the way is handling his job as president?" This question is asked very frequently; only three quarters2 have missing data (for a loss of sixtenths of 1 percent). Responses to variations of the question were not included nor were responses from polls which did not utilize a sample of the general population. As was the case for economic saliency, responses from interviews which spanned two quarters were discarded.


12Quarter/year: 3/1964, 3/1972, and 3/1976.









36

Source: all of the data are from Gallup surveys

(Edwards with Gallup 1990), except for the third quarter of 1972 which is from a poll conducted by the Opinion Research Corporation.13

From presidential job approval and economic saliency we formulate the equation:



Presidential Job Approval (APPROV) = a + bl(ECSAL) + e



We have one hypothesis to test:

H10: As economic saliency increases, presidential job

approval decreases (bl < 0).

To complete the test of our theory we will now regress our three dependent variables--the number of press conferences held, the number of foreign affairs speeches given (three categories), and the number of days spent traveling abroad, on presidential job approval. The following will describe each dependent variable.

1. Presidential press conferences (PRESS) is the number of press conferences held each quarter by the president.

Source: data for Kennedy through Carter are from the series The Cumulated Indexes to the Public Papers of the Presidents of the United States. Data for Ronald Reagan were provided by Staff, Reagan Library (1990).



13The figure was provided by Mr. Rob Turson, Roper Center, in a telephone conversation.








37
2. Foreign affairs speeches given by the president are recorded quarterly in three categories: (1) major, (2) secondary, and (3) total, or the sum of major and secondary. The following will describe the coding process.

As a starting point we begin with, and then expand

upon, the definitions developed by Lammers (1982), Ragsdale (1984, 1987), and Simon and Ostrom (1989). For our purposes a speech is any address given by the president which is one thousand words or more in length. This is roughly two pages of single-spaced text. However, to avoid eliminating an important speech which is less than one thousand words, flexibility and contextual analysis are required. For example, President Johnson's televised address to the nation on August 4, 1964 announcing that U.S. military forces would respond to attacks in the Gulf of Tonkin was much less than one thousand words, but was included in our data set. To have omitted this speech, or similar speeches of national importance, would have harmed our data base.

Speeches made by the president, whether required (i.e., the State of the Union) or discretionary, may affect the opinions people hold. Nationally televised or not, when a president speaks publicly some segment of the population will be affected and the message will be carried by some part of the news media. For these reasons all speeches made by the president, regardless of location or level of media attention, are considered.








38
A major foreign affairs speech (MFAS) is defined as any speech, or part thereof, which concerns diplomacy, summitry, treaties, executive agreements, national defense matters (including troop deployments), and crises, and is given either before a live television audience, a joint session of Congress, a press conference,4 a nominating convention, an international body (i.e., the Organization of American States), or a foreign government (i.e., the British Parliament). The speech, or part thereof, must address at least one foreign affairs issue or problem.

A secondary foreign affairs speech (SFAS) is defined as having the same content, but not given before a live television audience, the Congress, at the beginning of a press conference, at a nominating convention, or before an international body or foreign government.

Total foreign affairs speeches (TFAS) is defined as the sum of all major and secondary speeches.

Because our theory could not predict which of these three categories of speeches would demonstrate the best relationship, each will be tested. For now, total foreign affairs speeches will be used in our equations.

Source: for all presidents we will utilize the series Public Papers of the Presidents of the United States which


141f a prepared address given by the president at the beginning of a press conference meets the aforementioned requirements, then it is counted as a speech and included in our data set.









39
compiles all of the presidents' speeches, press conferences, and executive agreements, etc.

3. Presidential travel abroad (TRAV) is the number of days each quarter that the president or president-elect is traveling outside the United States.

Source: U.S. Department of State (1990).15

From our dependent and intervening variables we can now formulate three equations with corresponding hypotheses for testing. To reiterate, we theorize that when presidential job approval decreases, presidents will give fewer press conferences, give more foreign affairs speeches, and spend more time traveling abroad.



Presidential Press Conferences (PRESS) = a + bl(APPROV) + e



Hl1: As presidential job approval increases, presidential

press conferences increase (bl > 0).

And,



Total Foreign Affairs Speeches (TFAS) = a + bl(APPROV) + e



H12: As presidential job approval increases, total foreign

affairs speeches decrease (bl < 0).

And,

15Several scholars have used the indexes to the New York Times for this variable, but the indexes understate the true number of days presidents have spent traveling abroad.









40
Presidential Travel Abroad (TRAV) = a + bl(APPROV) + e



H13: As presidential job approval increases, presidential

travel abroad decreases (bl < 0).

We now address the two competing hypotheses presented in the introduction. First, rather than avoid economic matters as theorized presidents may forthrightly address a declining economy. Therefore, we will introduce a dependent variable which measures the number of economic affairs speeches given by the president (three categories) and regress it on presidential job approval.

Economic affairs speeches given by the president are recorded quarterly and in three categories: (1) major, (2) secondary, and (3) total, the sum of major and secondary. The following will describe the coding process.

The codes for these speeches are major (MEAS), secondary (SEAS), and total (TEAS). The reasons for three categories and their definitions are the same as those for foreign affairs speeches except, of course, where content is concerned. An economic speech concerns cost of living, inflation, unemployment, recession, and the trade deficit, among others. These are broad problems and are meant to coincide with the responses to the "most important problem" survey question previously discussed. The speech, or part thereof, must address at least one economic affairs issue or problem.








41

Source: we will utilize the series Public Papers of the Presidents of the United States.

For the present total economic affairs speeches will be used in our equation. We now have the following equation and hypothesis to test:



Total Economic Affairs Speeches (TEAS) = a + bl(APPROV) + e



H14: As presidential job approval increases, the number of

total economic affairs speeches decreases (bl < 0).

The second competing hypothesis states that when the

nation is involved in war presidents will give more speeches which concern foreign affairs and devote more time to foreign travel because of the national and international attention the military conflict will generate. Therefore, a variable which measures the presence or absence of war will be introduced and added to two of our previous equations.

War (WAR) measures whether the armed forces of the

United States are engaged in combat; 1 = yes, 0 = no. The following criteria are used to define combat: (1) the event lasts at least five days, (2) live fire occurs between U.S. armed forces and an enemy, and (3) the event is not solely an evacuation. The reader is referred to Appendix A for a list of events which meet these criteria.

Source: Congressional Research Service (1989, 1992) and Clodfelter (1992).









42

We now have two additional hypotheses to test:



Total Foreign Affairs Speeches (TFAS) = a + bl(APPROV) +

b2(WAR) + e



H15: When war is present, the number of total foreign

affairs speeches increases (b2 > 0).

And,



Presidential Travel Abroad (TRAV) = a + bl(APPROV) +

b2(WAR) + e



H16: When war is present, presidential travel abroad

increases (b2 > 0).

Two other areas of inquiry will now be considered. First, do foreign affairs speeches and foreign travel increase during presidential election years, regardless of economic conditions? We now introduce a new variable.

Presidential election year (ELEC) denotes whether a

presidential election year is present; 1 = yes, 0 = no. The presidential election years are 1964, 1968, 1972, 1976, 1980, 1984, and 1988. If we restricted this variable to incumbent presidents who had served a full term and who had sought re-election, then only three presidents would have qualified: Nixon (1972), Carter (1980), and Reagan (1984). Generalizations made from three cases would be untenable.









43

The following equations and hypotheses are now considered:



Total Foreign Affairs Speeches (TFAS) = a + bl(ELEC) + e



H17: In a presidential election year, the number of total

foreign affairs speeches increases (bl > 0).

And,



Presidential Travel Abroad (TRAV) = a + bl(ELEC) + e



HI8: In a presidential election year, presidential travel

abroad increases (bi > 0).

Our second area of interest is whether presidents

increase their pursuit of major foreign policy initiatives in either economic downturns or presidential election years. A new dependent variable, major foreign policy initiatives, will be introduced and tested. It is necessary to create this new variable because the variable major foreign policy speeches (MFAS) does not suffice. Many foreign policy speeches coded as "major" do not announce or propose actions taken or to be taken by the United States in the areas of foreign or military affairs. It is this characteristic which is of interest now.

Major foreign policy initiatives (MFPOL) are speeches made by the president which announce or propose a signifi-








44
cant initiative or action in either foreign or military affairs. These are recorded quarterly. The reader is referred to Appendix B for a list of initiatives.

Source: the list of initiatives was developed by using the rally events compiled by Brody (1984, Table 1, 43) and the speeches and documents recorded in the series Public Papers of the Presidents of the United States (United States, President 1961-1989).

The first test involves regressing major foreign policy initiatives (MFPOL) on our nine economic variables. The following equation and hypotheses are developed.



Major Foreign Policy Initiatives (MFPOL) = a + bl(UNEMP) +

b2(INFL) + b3(PRIME) + b4(INC) + b5(GNP) + b6(DEF) +

b7(CINC) + b8(CGNP) + b9(CDEF) + e


H19: When the unemployment rate increases, major foreign

policy initiatives increase (bl > 0).

H20: When the inflation rate increases, major foreign policy

initiatives increase (b2 > 0).

H21: When the prime lending rate increases, major foreign

policy initiatives increase (b3 > 0).

H22: When per capita real disposable income increases, major

foreign policy initiatives decrease (b4 < 0).

H23: When per capita gross national product increases, major

foreign policy initiatives decrease (b5 < 0).








45

H24: When per capita national defense expenditures increase,

major foreign policy initiatives decrease (b6 < 0).

H25: When the rate of change in per capita real disposable

income increases, major foreign policy initiatives

decrease (b7 < 0).

H26: When the rate of change in per capita gross national

product increases, major foreign policy initiatives

decrease (b8 < 0).

H27: When the rate of change in per capita national defense

expenditures increases, major foreign policy initiatives decrease (b9 < 0).

Our next test involves regressing major foreign policy initiatives on the variable presidential election year. The following is presented.



Major Foreign Policy Initiatives (MFPOL) = a + bl(ELEC) + e



H28: When a presidential election year is present, major

foreign policy initiatives increase (bl > 0).

Several additional variables were created which were

helpful during the recording of the data and they may prove useful for future research.

Year (YR) -- a four-digit number.

Quarter (QTR) -- 1 = first, 2 = second, 3 = third, 4 = fourth.









46

A dummy variable was created for each president: JFK

(John F. Kennedy), LBJ (Lyndon B. Johnson), RMN (Richard M.

Nixon), GRF (Gerald R. Ford), JEC (Jimmy E. Carter), and RWR

(Ronald W. Reagan). Quarters which were shared by two presidents are coded for missing data.16

The reader is referred to Appendix C for a complete

list of variables and their codes.

Whether some variables should be tested for lagged

effects is a methodological dilemma. With regards to one

area of presidency research--economic effects on presidential popularity--the following is considered.

In their study of economic performance and presidential

popularity, Norpoth and Yantek found that:

The evidence. . . .is too mixed to support a lag
structure for either inflation or unemployment.
(Norpoth and Yantek 1983, 796)

In a separate study, Norpoth found that a

; . . tentative specification of the influence of
economic performance on presidential popularity includes a contemporaneous effect for both unemployment and inflation, as well as a one-quarter lagged effect for inflation. (Norpoth 1984, 262)

16The fourth quarter of 1963 was shared by presidents Kennedy and Johnson because Kennedy was assassinated. The third quarter of 1974 was shared by presidents Nixon and Ford due to the resignation of Nixon. So that the actions of two presidents would not be combined in the same quarter, the following variables which concern presidential behavior are coded for missing data for the aforementioned quarters: presidential press conferences, presidential travel abroad, major foreign affairs speeches, secondary foreign affairs speeches, total foreign affairs speeches, major economic affairs speeches, secondary economic affairs speeches, total economic affairs speeches, major foreign policy initiatives, war, and presidential job approval.









47

Norpoth found no lagged effects for unemployment.

Based on the findings of Norpoth and Yantek (1983), Brace and Hinckley (1991) avoided lagged effects and, as noted in the review of literature, still found significant economic effects (i.e., the misery index) on presidential popularity. Ostrom and Simon (1989), in their evaluation of popular support for President Reagan, reported that unemployment had a significant impact on public evaluations and that it was "over eight times greater than that of inflation" (p.378).

Obviously, the findings are mixed. Yet, there were some significant results when a lagging function was utilized. Citizens may not necessarily react immediately to a declining economy, rather, they may reflect on their (or the country's) economic situation for a short while before their anger or disappointment is projected upon the president or other political institutions. Conversely, present opinions may be affected by past conditions and events. Lagging is an appropriate methodological tool for assessing or measuring these types of behaviors and this tool is utilized in the present study.

We have twelve equations with a total of 28 hypotheses. Several methodological procedures will now be restated for clarity. Both foreign affairs and economic affairs speeches have been divided into three categories--major, secondary, and total. Each will be tested because we could not predict









48
which category would demonstrate the best relationship with our other variables. Similarly, we will test both the levels and the percent rates of change in per capita real disposable income, per capita gross national product, and per capita national defense expenditures.

The data for this study constitute a time series and

they are recorded on a quarterly basis. When appropriate the mean for each quarter is used. Quarterly data are common in econometric models because they tend to smooth the data points. This should soften the effects of occasionally large variations within some variables. Since this study is concerned with patterns of behavior over a period of nearly three decades the significance of any one event is unimportant.

Monetary variables have been standardized in 1987 dollars and the consumer price index is standardized as 1982-84 = 100.

When a president is leaving office in January his speeches, press conferences, etc. for January are not included in the data for the first quarter; only data for the new president are recorded.

A SAS17 statistical program (release 6.08) will be

utilized for analyzing the data. The PROC REG routine calculates ordinary least squares (OLS) estimates. When a lag"SAS is a registered trademark of SAS Institute Inc., Cary, NC.








49

ging function is needed the PROC AUTOREG routine will be used. The latter uses OLS for the linear component and the Yule-Walker method (generalized least squares) for the autoregressive component of the models.

Our tables present data for several statistics: the

unstandardized parameter estimates (b), the standard errors (s.e.), the probabilities (p-values), which are one-tailed tests,18 the R-square statistic, the Durbin-Watson statistic for autocorrelation,"9 and sample size (N).

The following chapter will present the results from the various regressions. Chapter 5 will discuss the conclusions drawn from our analysis.



















"Our variables are significant when the p-values are <= .05.

19Ideally, the Durbin-Watson statistic should be 2; however, there is a range about this numeral which is acceptable. To determine whether the Durbin-Watson statistics from our regressions fell in the acceptable range, Table VIII, Critical Points of the Durbin-Watson Test for Autocorrelation, was used as a reference (see Wonnacott and Wonnacott 1979, 544).














CHAPTER IV
RESULTS

Before the results of the regressions are presented we report that there are significant levels of multicollinearity among our nine economic variables (see Appendix D for a correlation matrix). In a regression framework unemployment is the strongest explanatory variable for economic saliency (r= 0.84). The inflation rate also significantly affects economic saliency (r= 0.50) and it (inflation) does not correlate with unemployment (r= 0.07).

Our first model (see methodology chapter) regresses the variable economic saliency on nine economic indicators. Because economic saliency has a small sample size (N= 58), the number of regressors must be limited. Therefore, we will first regress economic saliency on six economic indicators while omitting the three variables which measure the percent rates of change in per capita real disposable income, per capita GNP, and per capita national defense expenditures (Model I, Table 1). Our second regression will include the three change variables, but omit per capita real disposable income, per capita GNP, and per capita national defense expenditures (Model II, Table 1).











Table 1. Regression of Economic Saliency
On Nine Economic Indicators


Variable

Intercept Unemployment Inflation Prime Rate Income GNP


Defense


MODEL I b, (s.e.

-62.232 (40.189)

13.179 *** (1.985)

7.447 ** (3.198)


0.339 (0.697)

-0.011 (0.011)

0.010 (0.009)

-0.019 * (0.011)


MODEL II b,(s.e.)

-55.586 (6.821)

14.049 *** (1.150)

11.972 *** (2.387)


MODEL III b. (s.e.)

-33.238 (22.106)

12.106 *** (1.374)

9.253 *** (2.189)


-0.201 (0.659)


0.0009 (0.001)


-0.017 * (0.010)


Change Inc Change GNP Change Def


R-square Durbin-Watson
N


P <= .05, P <= .01, P < .001,


one-tail one-tail one-tail


-0.209 (1.826)

-0.616
(1.616)

0.748 (0.642)


0.87 1.44 58


*
**


0.86 1.52 58


0.87 1.47 58









52

Table 2. Regression of Economic Saliency On Nine Economic Indicators, Continued


MODEL IV MODEL V MODEL VI Variable b,(s.e.) b,(s.e.) b,(s.e.)

Intercept -25.412 -58.341 157.468
(19.376) (6.067) (7.263)

Unemployment 12.591 ** 14.064 ** 14.034 **
(1.205) (0.897) (1.098)

Inflation 9.517 ** 12.165 ** 11.642 **
(2.152) (1.590) (1.930) Defense -0.018 *
(0.010)


R-square 0.87 0.86 0.87 Durbin-Watson 1.46 1.42 2.04 N 58 58 57

* P <= .05, one-tail
�* P < .001, one-tail



From Model I we see that unemployment, inflation, and

per capita national defense expenditures are significant and their signs are in the correct direction. The Durbin-Watson statistic (1.44) indicates the presence of autocorrelation. And in our second test (Model II), only unemployment and inflation are significant. Again, the Durbin-Watson statistic indicates autocorrelation (1.52).

Per capita real disposable income is highly correlated with per capita GNP (r= 0.99). To test whether per capita real disposable income would become significant if per capita GNP was omitted, per capita real disposable income was added to a model with unemployment, inflation, and per cap-








53
ita national defense expenditures (Model III). Per capita real disposable income did not reach significance, however (p-value= 0.22).

From Models I, II, and III we see that unemployment,

inflation, and per capita national defense expenditures significantly affect economic saliency. Per capita national defense expenditures are highly correlated with unemployment (r= -0.60) and inflation (r= -0.60). Model IV regresses economic saliency on unemployment, inflation, and per capita national defense expenditures. All three variables are significant and the R-square is 0.87 (Durbin-Watson= 1.46).

Model V omits per capita national defense expenditures. Both unemployment and inflation remain very significant, and the R-square (0.86) and Durbin-Watson statistic (1.42) are almost identical to those in Model IV. It is now evident that per capita national defense expenditures should be excluded from our final model because it does not add any significant explanatory power.

Now that we have specified the correct model we must

correct for autocorrelation. Model VI regresses the variable economic saliency on unemployment and inflation with a lag of one quarter. The Durbin-Watson statistic is now in the acceptable range (2.04). Our parameter estimates remain very significant and the R-square statistic changed very little.

We report that for each 1 percent increase in unemployment, economic saliency increases 14 percentage points and,








54

for each 1 percent increase in inflation, economic saliency increases by roughly 12 percentage points. The effects are delayed one quarter. Together, unemployment and inflation explain most of the variance of economic saliency (R-square=

0.87).

Two of our research hypotheses are accepted. First, as the unemployment rate increases, economic saliency increases

(H1) and second, as the inflation rate increases, economic saliency increases (H2). We do not reject the null hypotheses for the remainder (H3 through H9).

Our next model (Table 3) regresses presidential job approval on economic saliency. From Model I we see that economic saliency is significant and the sign is in the correct direction. The Durbin-Watson statistic (0.83) indicates significant autocorrelation, however. From Model II we see that lagging one quarter eliminated the autocorrelation problem. The Durbin-Watson statistic increased to 1.90, which is in the correct range. The variable conomic saliency remained significant and the sign is in the correct direction. Also note that the R-square statistic increased from 0.16 to 0.52.

Our final model (Model II) shows that, for each 1 percent increase in economic saliency, presidential job approval decreases by 0.11 percentage points. Or, for each ten point increase in economic saliency, presidential job approval will decrease roughly 1 percentage point. The









55

effect is delayed one quarter. We accept our hypothesis that when economic saliency increases, presidential job approval decreases (H10).



Table 3. Regression of Presidential Job Approval On Economic Saliency


MODEL I MODEL II Variable b, s.e.) b,(s.e.)

Intercept 59.069 57.420
(2.887) (3.469) Ec. Saliency -0.158 ** -0.117 *
(0.049) (0.061)


R-square 0.16 0.52 Durbin-Watson 0.83 1.90 N 55 54

* P <= .05, one-tail �* P <= .001, one-tail



Before we present the remaining regressions, we should first determine whether unemployment and/or inflation have direct affects on presidential job approval (see Table 4). Model I regresses presidential job approval on economic saliency, unemployment, and inflation (with one lag to correct for autocorrelation). We see that inflation is significant (p-value= 0.001), while economic saliency and unemployment are not. When presidential job approval was regressed on just economic saliency and inflation (one lag), only








56
inflation was significant again (p-value < .001). The p-value for economic saliency was 0.15.'

The reason economic saliency became insignificant in the two previous regressions is that when this variable is placed on the right side of an equation with unemployment and/or inflation, multicollinearity ensues. Economic saliency is highly correlated with both unemployment (r= 0.84) and inflation (r= 0.50), therefore, we will remove this variable and continue our tests for direct affects.

Model II regresses presidential job approval on unemployment and inflation (with two lags to correct for autocorrelation). Again, only inflation is significant (p-value= 0.003). From Model III, which regresses presidential job approval on just inflation (with two lags to correct for autocorrelation), we see that for each one point increase in inflation, presidential job approval decreases

3.2 percentage points. In Models II and III inflation accounts for a significant portion of the variance in presidential job approval (R-square= 0.73).

We have presented evidence which shows that unemployment and inflation indirectly affect presidential job approval through economic saliency, and that inflation directly affects presidential job approval. No direct effects from unemployment were found. Our final model (see Figure 2, chapter 5) will depict both paths.


'This regression is not shown.









57

Table 4. Regression of Presidential Job Approval
on Economic Saliency, Unemployment and
Inflation


Variable

Intercept Ec. Saliency Unemployment Inflation


MODEL I b. (s.e.

62.639 (8.787)

-0.062 (0.111)

0.133 (1.815)

-6.474 *** (2.015)


MODEL II b,(s.e.)

63.934 (6.170)




-1.100 (0.941)

-3.255 ** (1.172)


MODEL III b,(s.e.)

57.246 (2.151)







-3.274 ** (1.176)


R-square Durbin-Watson
N

* P <= .05,
** P < .01,
*** P <= .001,


0.51 1.78 54

one-tail one-tail one-tail


0.73 1.89 104


0.73 1.86 104


The final tests of our theory necessitates the regression of presidential press conferences, foreign affairs speeches (three categories), and presidential travel abroad on presidential job approval. We begin with presidential press conferences.

Table 5 presents results from three regressions of the variable presidential press conferences on presidential job approval. In Model I, presidential job approval is significant and the sign is in the correct direction, but autocorrelation is present (Durbin-Watson= 0.85). Lagging one









58

quarter (Model II) and two quarters (Model III) greatly improved the Durbin-Watson statistics, but presidential job approval lost significance in both models. The p-values increased to 0.08 (Model I) and to 0.12 (Model II). We therefore reject the hypothesis that when presidential job approval increases, presidential press conferences increase (H11).



Table 5. Regression of Presidential Press Conferences
On Presidential Job Approval


MODEL I MODEL II MODEL III Variable b,(s.e.) b,(s.e.) b_s.e.)

Intercept 0.595 1.124 1.512
(1.200) (1.696) (1.724) Job Approval 0.051 ** 0.042 0.035
(0.022) (0.030) (0.030) R-square 0.04 0.36 0.37 Durbin-Watson 0.85 2.12 2.00 N 107 106 105

* P <= .05, one-tail
** P <= .01, one-tail



Next we regressed our three categories of foreign

affairs speeches on presidential job approval. We begin with major speeches. Major foreign affairs speeches was not significant even when lagged one and two quarters. The lowest p-value reached for presidential job approval was 0.11 at two lags. Both secondary and total foreign affairs speeches









59

were significant and they yielded almost identical results. However, autocorrelation was present in both models.

Table 6 presents the results of secondary foreign affairs speeches (Model I) and total foreign affairs speeches (Model II) regressed on presidential job approval; both models are lagged once. Presidential job approval is significant in both models and the signs are in the correct direction. The Durbin-Watson statistics are in the proper range as well.

Because the results for the two models are nearly

identical, and because total foreign affairs speeches combines two categories, secondary foreign affairs speeches will be chosen for our final model.



Table 6. Regression of Foreign Affairs Speeches On Presidential Job Approval


MODEL I MODEL II Variable b.(s.e.) b,(s.e.)

Intercept 20.373 21.898
(4.655) (4.906) Job Approval -0.180 ** -0.163 *
(0.085) (0.089) R-square 0.13 0.13 Durbin-Watson 2.04 2.05 N 106 106

* P <= .05, one-tail �* P <= .01, one-tail








60
From Model I we see that for each 1 percent increase in presidential job approval, secondary foreign affairs speeches decrease by 0.18 per quarter (approximately onefifth of one speech). Or, for each 10 percent increase in presidential job approval, secondary foreign affairs speeches decrease roughly two per quarter (this finding will be addressed in the concluding chapter). We accept the hypothesis that when presidential job approval increases, (secondary) foreign affairs speeches decrease (H12).

Lastly, we regressed presidential travel abroad on

presidential job approval. From Table 7 we see that presidential job approval is significant and that the sign is in the correct direction. The Durbin-Watson statistic (2.20) is also in the acceptable range. For each 1 percent increase in presidential job approval, presidential travel abroad decreases by 0.06 days (less than two hours) per quarter. We accept the hypothesis that when presidential job approval increases, presidential travel abroad decreases (H13). We will address the significance of this finding later.

The next phase of our analysis concerns the two competing hypotheses. First, presidents may choose to address a poor economy in their public speeches rather than avoid it. We now regress our three categories of economic affairs speeches on presidential job approval.









61

Table 7. Regression of Presidential Travel
Abroad On Presidential Job Approval


MODEL I
Variable b,(s.e.)

Intercept 6.839 (2.095)

Job Approval -0.063 * (0.038)


R-square 0.02 Durbin-Watson 2.20 N 107

* P <= .05, one-tail





From Model I, Table 8 we see that major economic

affairs speeches is significant, but the sign is positive, which is in the wrong direction. Lagging one and two quarters did not reverse the sign.

When secondary economic affairs speeches was regressed on presidential job approval (Model II), the variable was significant and the sign was in the correct direction, but autocorrelation was present (Durbin-Watson= 0.99). Lagging one and two quarters improved the Durbin-Watson statistic, but our levels of significance increased to 0.07 in both regressions.

Model III regresses total economic affairs speeches on presidential job approval. The sign is in the correct direction, but presidential job approval did not reach our









62

significance level (p-value= 0.08) and lagging one and two quarters increased the p-values to 0.10 in both regressions. Our hypothesis (H14), therefore, is not accepted; economic affairs speeches do not increase when presidential job approval decreases.



Table 8. Regression of Economic Affairs Speeches
On Presidential Job Approval


MODEL I MODEL II MODEL III Variable b,.(s.e.) b,(s.eA ___ .) Intercept 0.130 21.461 21.591
(0.489) (5.738) (5.783) Job Approval 0.020 ** -0.169 * -0.149
(0.009) (0.105) (0.106) R-square 0.04 0.02 0.01 Durbin-Watson 1.96 0.99 0.98 N 107 107 107

* P <= .05, one-tail
�* P <= .01. one-tail



Our second competing hypothesis concerns whether either foreign affairs speeches (three categories) or presidential travel abroad, when regressed on presidential job approval, increase when war is present. We added the variable war to the right side of our previous equations.

We first address the regressions of three categories of foreign affairs speeches on presidential job approval and war. Rather than present the reader with multiple tables, suffice it to say that the variable war did not become sig-








63
nificant in any of our regressions. Even when lagged one and two quarters, the p-values for war did not go below 0.20. We therefore reject the hypothesis that when war is present, the number of foreign affairs speeches increases (H15).

We had similar results when we regressed presidential travel abroad on presidential job approval and war. The p-value for war was not significant (0.08) and lagging was unsuccessful (p-value= 0.07 at one lag and 0.08 at two lags). We reject the hypothesis that when war is present, presidential travel abroad increases (H16), and we conclude this section by reporting that none of our competing hypotheses were accepted.

We now address our two remaining areas of research

interest. First, do either foreign affairs speeches (three categories) or presidential travel abroad increase in a presidential election year?

We begin with foreign affairs speeches. When major

foreign affairs speeches was regressed on presidential election year, presidential election year was not significant (p-value= 0.23). Lagging increased the p-value to 0.27 at one lag and to 0.28 at two lags.

We then regressed secondary (Model I) and total (Model II) foreign affairs speeches on the variable presidential election year (Table 9). The parameter estimates and p-values for presidential election year are nearly identical in both models and the signs are in the correct direction.









64

Because the results are similar, and because total speeches combines two categories, we select secondary foreign affairs speeches for the final model.



Table 9. Regression of Foreign Affairs Speeches On Presidential Election Year


MODEL I MODEL II MODEL III Variable b.(s.e.) b.(s.e.) b,(s.e.)

Intercept 8.500 10.878 8.284
(0.874) (0.911) (1.144)

Election YR 8.714 ** 9.014 ** 9.542 **
(1.734) (1.806) (2.014) R-square 0.18 0.18 0.26 Durbin-Watson 1.38 1.33 2.05 N 110 110 109

* P <= .05, one-tail
** P < .001, one-tail




Secondary foreign affairs speeches (Model I) must be corrected for autocorrelation (Durbin-Watson= 1.38). Model III regresses secondary foreign affairs speeches on presidential election year with a lag of one quarter. The variable election year remained significant, the sign is in the correct direction, and the Durbin-Watson statistic (2.05) is in the correct range. From the data presented in Model III, we now construct the following equation:


Secondary Foreign Affairs Speeches = a + bl(Election YR) = 8.28 + 9.54(1)
= 17.82








65

From our equation we calculate that in a presidential election year there are approximately 18 secondary foreign affairs speeches given by the president per quarter; in nonpresidential election years the number is approximately 8 per quarter [8.28 + 9.54(0)]. The increase in secondary foreign affairs speeches as a result of a presidential election year being present is roughly ten per quarter. Because we have a lag of one quarter, the effect is demonstrable only in the second, third, and fourth quarters of the election year. We therefore accept the hypothesis that in a presidential election year, the number of (secondary) foreign affairs speeches increases (H17).

Second, we regressed the variable presidential travel abroad on presidential election year. The p-values for presidential election year did not go below 0.14 even when lagged one and two quarters and the sign in all three regressions was negative, which was in the wrong direction. We reject the hypothesis that in a presidential election year, presidential travel abroad increases (H18).

Our last area of inquiry concerns major foreign policy initiatives. We desire to know whether major foreign policy initiatives increase when the economy is performing poorly or when a presidential election year is present.

First, we regressed the variable major foreign policy initiatives on our nine economic variables. From Model I,








66
Table 10, we see that only the p-values for unemployment and the percent rate of change in per capita GNP are significant; however, using the entire model for explanatory purposes would be inelegant. Regardless, autocorrelation is present (Durbin-Watson= 1.71).

The p-value for per capita real disposable income is nearly significant (0.06) and this variable is highly correlated with per capita GNP (r= 0.99). Per capita real disposable income may become significant when regressed with unemployment and the percent rate of change in per capita GNP (see Model II). Model III regresses major foreign policy initiatives on just unemployment and the percent rate of change in per capita GNP; the latter are the two variables from Model I which are significant.

From Table 10 we see that none of the variables in

Models II and III reached our significance level. Lagging was not fruitful, though the Durbin-Watson statistics did improve. We conclude that the number of major foreign policy initiatives does not change as a result of declining economic conditions, and we do not reject our null hypotheses (H19-H27).

Lastly, we regressed major foreign policy initiatives on the variable presidential election year. Presidential election year was not significant (p-value= 0.20) and lagging caused our p-values to slightly increase. We reject the











Table 10. Regression of Major


Initiatives On Nine


Foreign Policy Economic Indicators


Variable

Intercept


Unemployment Inflation Prime Rate


Income


GNP


Defense Change Inc Change GNP


Change Def



R-square Durbin-Watson
N


MODEL I b, (s. e.)

-2.374 (2.081)

0.202 * (0.106)

-0.055 (0.178)

-0.031 (0.038)

-0.0009 (0.0005)

0.0007 (0.0004)

0.0001 (0.0005)

0.138 (0.094)

-0.214 ** (0.096)

-0.006 (0.030)


0.07 1.71 109


MODEL II b.(s.e.)

0.673 (0.444)

0.060 (0.051)


MODEL III bos.e.)


0.406 (0.286)

0.038 (0.044)


-0.00003 (0.00004)


-0.078 (0.072)


0.02 1.67 109


-0.073 (0.072)


0.01 1.66 109


* P <= .05,
** P <= .01,


one-tail one-tail


Initiatives On Nine








68
hypothesis that major foreign policy initiatives increase in a presidential election year (H28).

This concludes the regression of our models. The following chapter will present the conclusions drawn from the foregoing analysis, discuss their implications, and present suggestions for future research.














CHAPTER V
CONCLUSIONS

The results from our regressions lead to a modification of our theory of presidential behavior (see Figure 2). We found that unemployment and inflation are important economic indicators for the public. When either unemployment or inflation increases, economic saliency increases. This saliency is recorded through a modification of the responses to the "most important problem" question asked in national opinion surveys. This new variable measures the attitude quite well, though the delay (lag) is one quarter. A short delay in the processing of economic information by the public seems reasonable, however, and is supported by previous research on the presidency (see review of literature and below).





Travel
Abroad
Unemployment Increases Increases > Economic Presidential
Saliency ---> Job Approval
Inflation - Increases Decreases Secondary Increases Foreign Affairs
T Speeches
Increase


Figure 2. Modified Behavioral Model








70
When the public's level of economic saliency goes up, its collective attitude concerning the president goes down (with a delay of one quarter). This attitude is measured by responses to the "presidential job approval" question frequently asked in national opinion surveys. This finding offers evidence to support the belief held by many politicians and political scientists that the public holds the president accountable for poor economic conditions (see, for example, Nixon 1962, Tufte 1978, and Norpoth 1984).

In the present study economic saliency is an intervening variable between unemployment, inflation, and presidential job approval. Unemployment and inflation indirectly affect presidential job approval through economic saliency, and inflation also directly affects presidential job approval. These relationships are analogous to findings reported in previous research. With regards to the variable presidential job approval, both Monroe (1978) and Norpoth (1984) found lagged effects for inflation, and Fiorina (1981) confirmed the relationship (non-lagged) with unemployment. Both of these economic indicators (non-lagged) were reported by Frey and Schneider (1978) and Shapiro and Conforto (1980a) to significantly affect presidential job approval. And, when unemployment and inflation were combined into a "misery index" it, too, significantly affected presidential job approval (Ostrom and Simon 1985, Brace and Hinckley 1991).








71

We subsequently discovered that when presidential job approval decreases, presidents will increase their time spent traveling abroad and increase the number of secondary foreign affairs speeches given (the latter with a delay of one quarter), though the increases for both variables were quite small. The relationship between presidential job approval and presidential travel abroad supports the findings of Darcy and Richman (1988); however, our results are irrespective of political party.

A judgment must be made regarding the importance of the latter two findings. If presidents consciously increase their attention to foreign affairs in their speeches and increase the number of days spent traveling abroad as methods for improving their standing in public opinion polls and their electoral chances, then we would expect to see increases which reasonable people would agree were substantial. It is difficult to conclude that an increase of 0.06 days of foreign travel per quarter and 0.18 secondary foreign affairs speeches per quarter, for each one point decrease in presidential job approval, would produce the desired effect.' It seems unlikely that the public would notice such small changes in presidential behavior. This conclusion may support the research of Simon and Ostrom



'For the entire data set the mean number of days per quarter that the president spends traveling abroad is 3.37. Similarly, the mean number of secondary foreign affairs speeches is 10.71 per quarter.








72
(1989). They reported that it is approval-enhancing events, not speeches or foreign travel, which positively affects presidential popularity.

Contrary to our theory, and to the findings reported by Lammers (1981), we found no relationship between presidential press conferences and presidential job approval. Lammers had found that when presidential popularity decreased, presidential press conferences increased.

None of our competing hypotheses were accepted. Economic affairs speeches did not increase when presidential job approval decreased, and neither foreign affairs speeches nor presidential travel abroad were affected by the presence of war. We conclude this section by stating that insufficient evidence exists to support our theory that presidents attempt to divert the nation's attention to foreign affairs when the economy is performing poorly.

Regarding our remaining areas of research interest, we found that (secondary) foreign affairs speeches increased in presidential election years, though presidential travel abroad did not. Apparently, presidents are inclined to remain at home during a presidential election year to devote their energies to the campaign.

Surprisingly, major foreign policy initiatives did not increase as a result of poor economic conditions nor did they increase in presidential election years. It may be that presidents believe that economic conditions are foremost in








73
the minds of the American public and that attempts to increase their own popularity or electoral chances through foreign affairs speeches, foreign travel, or major foreign policy initiatives would be fruitless.

We will now address the significance of the finding

that secondary foreign affairs speeches increase by ten per quarter during a presidential election year. Some may argue that the increase is not due to greater attention to foreign affairs by presidents in presidential election years, but the result of a general increase in speechmaking while presidents are on the campaign trail. One method of testing this assertion is to regress our three categories of economic affairs speeches on the variable presidential election year and compare the results.

Presidential election year was not significant when regressed with major economic affairs speeches (p-value= 0.09). Table 11 presents results from the regressions of secondary and total economic affairs speeches on presidential election year.

From Model I (secondary economic affairs speeches) and Model II (total economic affairs speeches) we see that presidential election year is significant and the signs are in the correct direction. The results for both regressions are nearly identical and autocorrelation is present. In keeping with our previous decision-making processes, secondary economic affairs speeches is chosen for our final model.









74

Model III presents the regression of secondary economic affairs speeches on presidential election year, but with a lag of one quarter. The variable is significant, the sign is in the correct direction, and autocorrelation has been corrected (Durbin-Watson= 2.21).



Table 11. Regression of Economic Affairs Speeches On Presidential Election Year


MODEL I MODEL II MODEL III Variable b.bs.e.) b.s.e.)

Intercept 9.048 10.207 8.665
(1.277) (1.281) (1.746)

Election YR 13.415 ** 13.792 ** 14.960 **
(2.531) (2.539) (2.969) R-square 0.20 0.21 0.30 Durbin-Watson 1.31 1.28 2.21 N 110 110 109

* P <= .05, one-tail
** P < .001, one-tail




From the data presented in Model III, Table 11, we create the following equation:



Secondary Economic Affairs Speeches = a + bl(Election YR) = 8.66 + 14.96(1)
= 23.62



We calculate that, in a presidential election year, there are approximately 24 secondary economic affairs








75
speeches given by the president per quarter. In non-presidential election years the number of speeches is approximately nine per quarter [8.66 + 14.96(0)]. The increase in secondary economic affairs speeches as a result of a presidential election year being present is roughly 15 per quarter. Because the variable is lagged one quarter, the effect is demonstrable only in the second, third, and fourth quarters of the election year. It seems reasonable to infer that the increase in secondary foreign affairs speeches can be attributed to a general increase in speechmaking by presidents during the political campaign (presidential election year).

From Table 12 we see that presidents give greater

attention to economic affairs in their speeches during the election year. This may be due to the importance voters place on economic conditions (especially unemployment and inflation) when they evaluate the president's job performance and when deciding whom to vote for in the general election. This additional piece of evidence supports our previous findings and speculations.


Table 12. Comparison of the Presidents' SDeeches
In Non-presidential Election Years
and Presidential Election Years

Type of Non-presidential Presidential Speech Election Years Election Years Increase

For. Affairs 8 per qtr. 18 per qtr. 10 (125%) Econ. Affairs 9 per qtr. 24 per qtr. 15 (166%)








76
In the present study we have reported that unemployment and inflation matter greatly to the American public. These two economic variables affect economic saliency, which in turn affects presidential job approval. Previous research has shown that presidential job approval affects the president's electoral chances (Sigelman 1979, Lewis-Beck and Rice 1982) and his party's chances in congressional elections (Kernell 1977, Campbell 1985).

We have shown that increases in presidential job

approval are not associated with increases in foreign travel and foreign affairs speeches. Rather, presidential approval declined as the latter two variables increased. And, presidents do not increase the number of major foreign policy initiatives when the economy is performing poorly or during presidential election years.

Regarding economics and foreign affairs, it is evident that presidents would be wise to devote greater attention to economic affairs since these are obviously the issues foremost in the minds of the public. The problem for the chief executive, however, is that the number of tools at his disposal for correcting a declining economy is quite small and in most cases he must rely on the cooperation of Congress or the bureaucracy or the Federal Reserve Board. Additionally, we found that the number of economic speeches given by the president does not affect presidential job approval. It may








77
be that actual economic conditions matter greatly to the public, while rhetoric does not.

From our analyses two topics for future research are suggested. First, we found that unemployment and inflation account for most of the variance (R-square= 0.87) in econnomic saliency and that economic saliency accounts for a substantial portion of the variance (R-square= 0.52) in presidential job approval. Researchers should focus on an explanation for the remaining variance in presidential job approval.' Possible explanatory variables include personality or charisma, crises, scandals, and treatment by the media.

Second, what can presidents do to improve their job

approval ratings? Greater attention to foreign affairs and foreign travel is obviously not the answer because our results show that increases in foreign affairs speeches and foreign travel are associated with decreases in presidential job approval. We also reported that presidents do not attempt to initiate major foreign policy when the economy is slumping. Yet, even if presidents did increase the number of major foreign policy initiatives, the results from the regressions of foreign affairs speeches on presidential job approval suggest that an increase in the number of major





2When presidential job approval was regressed just on inflation, the R-square was 0.73.








78

foreign policy initiatives would also be associated with a decrease in presidential job approval.

The evidence from this study suggests that economic affairs are more important than foreign affairs to the American public. Specifically, presidents should attempt to keep the unemployment and inflation rates as low as possible. Although administrations may attempt to enlarge the incomes of individuals and families by increasing transfer payments (see Tufte 1978), success greatly depends upon cooperation from others within the political structure. More importantly, these transfer payments do not decrease unemployment or inflation. In fact, additional payments could fuel inflation. And, as previously mentioned, attempts to increase presidential job approval through policy successes in Congress may be unsuccessful (see Bond and Fleisher 1984, Zeidenstein 1985).

If unemployment and inflation cannot be controlled, a president may have to wait for an economic slump to work itself out or for fate to present him with a potential rally event close to the election and hope for a positive response from the public. In either case, the president is at the mercy of uncontrollable forces. Political scientists should conduct research for the purpose of discovering which types of behavior, situations, or events will increase presidential popularity.








79

This research project has presented the results of an investigation into nearly thirty years of presidential behavior; however, we have analyzed the actions of only six presidents--a rather small study group to generalize from. If institutions have memories and precedent is a guiding force, then we would expect that, at least for the near future, presidents will continue to behave like their predecessors in similar situations. Yet, we know that what appears to be politically conventional today is often tomorrow's discarded theory.















APPENDIX A
WAR'



EVENT START DATE END DATE
(qtr/yr) (qtr/yr)


Vietnam 4/1961 1/1973 Lebanon 3/1982 1/1984 Persian Gulf 3/1987 3/1988







Events whose dates overlapped those recorded were discarded. For example, the U.S. invasion of Grenada occurred in the fourth quarter of 1983, which overlaps the Lebanon conflict. Since the variable measures only the presence or absence of war, several military conflicts are not listed here.

There is disagreement as to when the Vietnam war began and ended, but an explanation is in order for the dates chosen. U.S. military advisors had been in Vietnam throughout the 1950s. In December of 1961 two U.S. Army helicopter companies were deployed to South Vietnam, hence the reason

'Data for this appendix are from Congressional Research Service (1989, 1992) and Clodfelter (1992).

80









81

why the fourth quarter of 1961 was chosen as the start date. Many scholars have elected August of 1964 as the start date because this was when the Gulf of Tonkin crisis occurred and when President Johnson responded with heavy bombing. The following data show why August 1964 may be inappropriate.



Table 13. Vietnam War Statistics


1961


1962


1963


1964


U.S. troop strength in Vietnam Killed Air sorties


3,164


11


11,326


31

2,334


16,263 23,310


78 6,929


147 5,362


(Source: Clodfelter 1992, 1229)




The first quarter of 1973 is chosen as the end date

because "by March 29 the last of the 24,200 U.S. servicemen remaining in South Vietnam had been withdrawn" (Clodfelter 1992, 1276).















APPENDIX B
MAJOR FOREIGN POLICY INITIATIVES'




Kennedy

1-30-61 Reappraise entire defense structure; fight cold

war.

4-20-61 Discusses Bay of Pigs events. 5-17-61 Strengthening of North Atlantic Treaty Organization (NATO).

5-25-61 New defense requests for military assistance; new

attempts at disarmament.

9-25-61 Plan for disarmament; end of nuclear testing;

situations in Laos and West Berlin. 3-2-62 Resumption of nuclear testing. 4-18-62 Plan for disarmament. 10-22-62 Quarantine of Cuba. 7-26-63 Nuclear test-ban treaty. [total: 9]

Johnson

8-4-64 Forces respond to attacks in Gulf of Tonkin. 4-7-65 Statement of Vietnam policy.



'Sources consulted: Brody (1984) and United States, President (1961-1989).









83

5-2-65 U.S. forces sent to the Dominican Republic. 5-4-65 Request for $700 million for the defense of Vietnam and the Dominican Republic.

1-31-66 Resumption of bombing in Vietnam. 3-31-68 De-escalating the Vietnam conflict. 6-12-68 U.S. will sign nuclear non-proliferation treaty. 6-13-68 Consular convention signed by the U.S. and the

U.S.S.R.

7-1-68 Signing of nuclear non-proliferation treaty. 10-31-68 Halts bombing of North Vietnam. [total: 10]

Nixon

5-14-69 Outline of peace proposal for Vietnam. 11-24-69 Signing of non-proliferation of nuclear weapons

treaty.

3-5-70 Ratification and entry into force of the nonproliferation of nuclear weapons treaty.

4-20-70 Withdrawal of 150,000 troops from Vietnam. 4-30-70 New military initiatives against the communists in

Cambodia and Vietnam.

10-7-70 New peace proposal for Vietnam. 7-15-71 Announcement of diplomatic invitation to China. 1-25-72 Peace plan for Vietnam. 2-10-72 Itinerary for China trip. 5-8-72 Mining of North Vietnam's harbors. 5-19-72 Announcement of trip to the U.S.S.R.









84

1-23-73 Peace agreement with North Vietnam. [total: 12]

Ford

3-6-75 Assistance to Cambodia and Vietnam. 5-15-75 Concerning SS Mayaguez and rescue of her crew. 5-28-76 Treaty with the U.S.S.R. regarding underground

nuclear explosions.

[total: 3]

Carter

8-12-77 Agreement in principle on Panama Canal treaties. 9-7-77 Signing of Panama Canal treaties. 9-8-77 United States-Canada agreement on the natural gas

pipeline.

10-4-77 Pledges to use nuclear weapons only defensively;

decrease the number of nuclear weapons.

3-10-78 Signing of nuclear non-proliferation agreement. 4-7-78 Deferment of the production of neutron weapons. 9-17-78 Camp David accords signed. 12-15-78 Diplomatic relations with China established. 3-26-79 Egyptian-Israeli peace treaty signed. 6-18-79 Signing of SALT II (Strategic Arms Limitation

Talks).

1-4-80 Sanctions against the U.S.S.R. for invading

Afghanistan; delay of ratification of SALT II. 4-7-80 Sanctions against Iran. 4-25-80 Failed rescue of American hostages in Iran.








85

9-17-80 New agreements with China. [total: 14]


Reagan

10-1-81 5-9-82 8-20-82 10-9-82 11-22-82 3-23-83 3-30-83 10-4-83



10-25-83 4-4-84 5-1-85 11-14-85 1-7-86 4-14-86 12-8-87



3-11-88 3-16-88 4-18-88 6-4-88


Sale of aircraft (AWACs and Fl5s) to Saudi Arabia. Proposal for arms reductions. Troops to deploy to Lebanon. Sanctions against Poland. Proceeding with the MX missile. Strategic Defense Initiative (SDI) proposed. Proposal to reduce nuclear forces. New START (Strategic Arms Reduction Talks) initiatives.

Deployment of forces to Grenada. Proposal to ban chemical weapons. Economic sanctions against Nicaragua. Concerning U.S.- U.S.S.R. summit in Geneva. Economic sanctions against Libya. Airstrikes against Libya. Signing of Intermediate-range Nuclear Forces (INF) treaty with the U.S.S.R. Sanctions against Panama. Troops to be deployed to Honduras. Military confrontations with Iran. Concerning the INF and START talks with the Soviet Union.


[total: 19] Grand total: 67















APPENDIX C
VARIABLES




YR Year

QTR Quarter ELEC Presidential Election year JFK John F. Kennedy LBJ Lyndon B. Johnson RMN Richard M. Nixon GRF Gerald R. Ford JEC Jimmy E. Carter RWR Ronald W. Reagan PRESS Presidential Press Conferences TRAV Presidential Travel Abroad MFAS Major Foreign Affairs Speeches SFAS Secondary Foreign Affairs Speeches TFAS Total Foreign Affairs Speeches MEAS Major Economic Affairs Speeches SEAS Secondary Economic Affairs Speeches TEAS Total Economic Affairs Speeches MFPOL Major Foreign Policy Initiatives WAR War

APPROV Presidential Job Approval










ECSAL Economic Saliency PRIME Prime Lending Rate UNEMP Unemployment Rate INFL Inflation Rate INC Per Capita Real Disposable Income GNP Per Capita Gross National Product DEF Per Capita National Defense Expenditures CINC Percent Rate of Change in Per Capita Real

Disposable Income

CGNP Percent Rate of Change in Per Capita Gross

National Product

CDEF Percent Rate of Change in Per Capita National

Defense Expenditures
















APPENDIX D
CORRELATION MATRIX WITH NINE ECONOMIC VARIABLES (Pearson's r)








PRIME UNEMP INFL GNP DEF

PRIME 1.00 0.49 0.62 0.63 -0.56 UNEMP 0.49 1.00 0.07 0.41 -0.60 INFL 0.62 0.07 1.00 0.38 -0.60 GNP 0.63 0.41 0.38 1.00 -0.59 DEF -0.56 -0.60 -0.60 -0.59 1.00 INC 0.64 0.50 0.34 0.99 -0.62 CINC -0.23 -0.10 -0.30 -0.13 0.12 CGNP -0.35 -0.07 -0.29 -0.09 0.12 CDEF 0.21 0.17 0.01 0.16 0.02






INC CINC CGNP CDEF

PRIME 0.64 -0.23 -0.35 0.21 UNEMP 0.50 -0.10 -0.07 0.17 INFL 0.34 -0.30 -0.29 0.01 GNP 0.99 -0.13 -0.09 0.16 DEF -0.62 0.12 0.12 0.02 INC 1.00 -0.12 -0.11 0.16 CINC -0.12 1.00 0.43 -0.08 CGNP -0.11 0.43 1.00 0.08 CDEF 0.16 -0.08 0.08 1.00














REFERENCES


Abramowitz, Alan. 1985. "Economic Conditions, Presidential
Popularity, and Voting Behavior in Midterm Congressional Elections." Journal of Politics 47:31-43.

Bond, Jon R., and Richard Fleisher. 1984. "Presidential Popularity and Congressional Voting: A Reexamination of Public Opinion As A Source of Influence In Congress."
Western Political Ouarterly 37:291-306.

Brace, Paul, and Barbara Hinckley. 1991. "The Structure of
Presidential Approval: Constraints within and across
Presidencies." Journal of Politics 53(4):993-1017.

Brody, Richard. 1984. "International Crises: A Rallying
Point for the President?" Public Opinion DecemberJanuary:41.

Campbell, James E. 1985. "Explaining Presidential Losses in
Midterm Congressional Elections." Journal of Politics
47:1140-1157.

Citibase (data file). 1978. New York, NY: Citibank, N.A.

Clodfelter, Micheal. 1992. Warfare and Armed Conflicts: A
Statistical Reference to Casualty and Other Figures, 1618-1991. Vol.II. Jefferson, NC: McFarland and Company,Inc.

Congressional Research Service. 1989. Instances of Use of
United States Armed Forces Abroad, 1798-1989. Washington, DC: Library of Congress.

. 1992. Use of Force by the United States: Case
Studies, 1950-1991. Washington, DC: Library of Congress.

The Cumulated Indexes to the Public Papers of the Presidents
of the United States, John F. Kennedy, 1961-1963. 1977.
Millwood, NY: KTO Press.

The Cumulated Indexes to the Public Papers of the Presidents
of the United States, Lyndon B. Johnson, 1963-1969.
1978. Millwood, NY: KTO Press.










The Cumulated Indexes to the Public Papers of the Presidents
of the United States, Richard M. Nixon, 1969-1974.
1978. Millwood, NY: KTO Press.

The Cumulated Indexes to the Public Papers of the Presidents
of the United States, Gerald R. Ford. 1974-1977. 1980.
Millwood, NY: Kraus International Publications.

The Cumulated Indexes to the Public Papers of the Presidents
of the United States, Jimmy Carter, 1977-1981. 1983.
Millwood, NY: Kraus International Publications.

Darcy, Robert E., and Alvin Richman. 1988. "Presidential
Travel and Public Opinion." Presidential Studies Quarterly 18(l):85-90.

Data Disk (data file). 1984. Cambridge, MA: Cambridge Planning and Analytics.

Data Resources, Inc. (data file). Lexington, MA: Data
Resources, Inc.

Edwards, George C. III, with Alec M. Gallup. 1990. Presidential Approval: A Sourcebook. Baltimore, MD: Johns
Hopkins University Press.

Erikson, Robert S. 1989. "Economic Conditions and the Presidential Vote." American Political Science Review 83
(2):567-573.

Fair, Ray C. 1978. "The Effect of Economic Events on Votes
for President." The Review of Economics and Statistics
60(2):159-173.

� 1982. "The Effect of Economic Events on Votes for
President: 1980 Results." The Review of Economics and
Statistics 64:322-325.

� 1988. "The Effect of Economic Events on Votes for
President: 1984 Update." Political Behavior 10(2):168179.

Fiorina, Morris P. 1978. "Economic Retrospective Voting in
American National Elections: A Micro-Analysis." American Journal of Political Science 22:426-443.

� 1981. Retrospective Voting in American National
Elections. New Haven, CT: Yale University Press.

Ford, Gerald R. 1979. A Time to Heal: The Autobiography of
Gerald R. Ford. New York, NY: Harper and Row.










Frey, Bruno S., and Friedrich Schneider. 1978. "An Empirical
Study of Politico-Economic Interaction in the United
States." The Review of Economics and Statistics 60:174183.

Gallup Organization (unpublished report). America's Opinion
On: Most Important Problem, 1935-1992. Princeton, NJ:
Gallup Organization.

Hibbing, John R., and John R. Alford. 1981. "The Electoral
Impact of Economic Conditions: Who Is Held Responsible?" American Journal of Political Science 25(3):423439.

Kenney, Patrick J., and Tom W. Rice. 1988. "The Contextual
Determinants of Presidential Greatness." Presidential
Studies Ouarterly 18(l):161-169.

Kenski, Henry C. 1977. "The Impact of Economic Conditions on
Presidential Popularity." Journal of Politics 39:764773.

. 1980. "Economic Perception and Presidential Popularity: A Comment." Journal of Politics 42:68-75.

Kernell, Samuel. 1977. "Presidential Popularity and Negative
Voting: An Alternative Explanation of the Midterm Congressional Decline of the President's Party." American
Political Science Review 71:44-66.

. 1978. "Explaining Presidential Popularity." American Political Science Review 72:506-522.

_ 1986. Going Public: New Strategies of Presidential
Leadership. Washington, DC: CQ Press.

King, Gary, and Lyn Ragsdale. 1988. The Elusive Executive:
Discovering Statistical Patterns in the Presidency.
Washington, DC: CQ Press.

Kramer, Gerald H. 1971. "Short-Term Fluctuations in U.S.
Voting Behavior, 1896-1964." American Political Science
Review 65:131-143.

Lammers, William W. 1981. "Presidential Press-Conference
Schedules: Who Hides and When?" Political Science Ouarterly 96(2):261-278.

te. 1982. "Presidential Attention-Focusing Activities." In The President and the Public, ed. Doris A.
Graber. Philadelphia, PA: Institute for the Study of
Human Issues.










Lanoue, David J. 1988. From Camelot to the Teflon President:
Economics and Presidential Popularity Since 1960. New
York, NY: Greenwood Press.

Lee, Jong R. 1977. "Rallying Around the Flag: Foreign Policy
Events and Presidential Popularity." Presidential Studies Quarterly 7(4):252-256.

Lewis-Beck, Michael S., and Tom W. Rice. 1982. "Presidential
Popularity and Presidential Vote." Public Opinion Quarterly 46:534-537.

Markus, Gregory. 1988. "The Impact of Personal and National
Economic Conditions on the Presidential Vote: A Pooled Cross-Sectional Analysis." American Journal of Political Science 32:137-154.

McConnell, Campbell R. 1975. Economics: Principles, Problems, and Policies. New York, NY: McGraw-Hill.

Monroe, Kristen R. 1978. "Economic Influences on Presidential Popularity." Public Opinion Quarterly 42:360-369.

Mueller, John E. 1973. War, Presidents, and Public Opinion.
New York, NY: Wiley.

Neustadt, Richard E. 1986. Presidential Power: The Politics
of Leadership From FDR to Carter. New York, NY:
Macmillan.

Newsweek. 1992. "The War Room Drill." November-December:78.

Nixon, Richard M. 1962. Six Crises. Garden City, NY:
Doubleday.

Norpoth, Helmut. 1984. "Economics, Politics, and the Cycle
of Presidential Popularity." Political Behavior 6(3):
253-273.

Norpoth, Helmut, and Thom Yantek. 1983. "Macroeconomic Conditions and Fluctuations of Presidential Popularity: The Question of Lagged Effects." American Journal of
Political Science 27:785-807.

Oppenheimer, Joe A. 1980. "Small Steps Forward for Political
Economy." World Politics 33:121-151.

Ostrom, Charles W. Jr., and Brian Job. 1986. "The President
and the Political Use of Force." American Political
Science Review 80(2):541-566.










Ostrom, Charles W. Jr., and Dennis M. Simon. 1985. "Promise
and Performance: A Dynamic Model of Presidential Popularity." American Political Science Review 79:334-358.

� 1989. "The Man in the Teflon Suit? The Environmental Connection, Political Drama, and Popular Support in the Reagan Presidency." Public Opinion Quarterly 53:
353-387.

Ragsdale, Lyn. 1984. "The Politics of Presidential Speechmaking, 1949-1980." American Political Science Review
78:971-984.

. 1987. "Presidential Speechmaking and the Public
Audience: Individual Presidents and Group Attitudes."
Journal of Politics 49:704-736.

Rivers, Douglas, and Nancy L. Rose. 1985. "Passing the President's Program: Public Opinion and Presidential Influence in Congress." American Journal of Political Science 29:183-196.

Rohde, David W., and Dennis M. Simon. 1985. "Presidential
Vetoes and Congressional Response: A Study of Institutional Response." American Journal of Political Science
29:397-427.

Schlenker, Barry R. 1980. Impression Management. Monterey,
CA: Brooks/Cole Publishing Co.

Shapiro, Robert Y., and Bruce M. Conforto. 1980a. "Presidential Performance, the Economy, and the Public's
Evaluation of Economic Conditions." Journal of Politics
42:49-67.

� 1980b. "Economic Perception and Political Behavior: Reply to Professor Kenski." Journal of Politics
42:76-81.

Sigelman, Lee. 1979. "Presidential Popularity and Presidential Elections." Public Opinion Quarterly 43:532534.

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.

Smith, Carolyn. 1990. Presidential Press Conferences: A
Critical Approach. New York, NY: Praeger Publishers.




Full Text

PAGE 1

ASSESSING PRESIDENTIAL BEHAVIOR IN ECONOMIC DOWNTURNS AND ELECTION YEARS FROM 1961 THROUGH 1988 By JOSEPH K. SMAHA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1994

PAGE 2

Copyright by Joseph K. 1994 Smaha

PAGE 3

ACKNOWLEDGEMENTS Many individuals have made a significant contribution to the development and execution of this research project and to all of them I owe much gratitude. First, I wish to thank the following faculty, graduate students, and personnel at the University of Florida: Dr. M. Margaret Conway, Dr. Wayne Francis, Dr. Walter Rosenbaum, Dr. Leslie Thiele, Dr. David Colburn, Dr. Carol West, Dr. Robert Emerson, Dr. Mark Yang, Dr. Robert McMahon, Debra Moody, Jeffrey Katz, Elizabeth Williams, Ken Mease, the staff at Library West (especially Government Documents), Janet Galvez, and Patricia Turner. My appreciation is also extended to the staffs of Senator Bob Graham, Senator Connie Mack, Congresswoman Karen Thurman, the Congressional Research Service, the Bureau of Labor Statistics, the Ronald Reagan Library, and also to Virginia Shope (Army War College Library) and Carl Galbraith ( Department of Commerce ) . I also wish to thank Glynova Howe (New York Times), Rob Turson (Roper Center), Dr. Gary King (Harvard University), Dr. Lyn Ragsdale (University of Arizona), Cathy Akritas (Minitab, Inc.), and Donna Woodward (SAS Institute, Inc.). iii

PAGE 4

A sincere appreciation is extended to my wife Katherine and to my children Sara and Christopher for their continuous support and for having the patience to ignore me when my research did not proceed smoothly. Lastly, I wish to acknowledge my deceased parents, Joseph and Teresa Smaha, whose love, support, and encouragement greatly contributed to making graduate school a reality for me. My sincerest apologies to anyone I may have forgotten. Naturally, all errors are mine. iv

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGEMENTS iii ABSTRACT vi CHAPTERS I . INTRODUCTION 1 II. PREVIOUS RESEARCH ON SELECTED ASPECTS OF PRESIDENTIAL BEHAVIOR 6 Presidential Job Approval 6 Presidential Elections 11 Congressional Elections 16 Leading Congress 19 Foreign Affairs and Travel Abroad 22 Speeches and Press Conferences 24 III. METHODOLOGY 27 IV. RESULTS 50 V. CONCLUSIONS 69 APPENDICES A . WAR 80 B. MAJOR FOREIGN POLICY INITIATIVES 82 C. VARIABLES 86 D. CORRELATION MATRIX WITH NINE ECONOMIC VARIABLES 88 REFERENCES 89 BIOGRAPHICAL SKETCH 95 V

PAGE 6

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ASSESSING PRESIDENTIAL BEHAVIOR IN ECONOMIC DOWNTURNS AND ELECTION YEARS FROM 1961 THROUGH 1988 by Joseph K. Smaha August 1994 Committee Chair: M. Margaret Conway, Ph.D. Major Department: Political Science This is a quantitative study of the American presidency. The subjects are six presidents, John F. Kennedy through Ronald W. Reagan. The author examines whether there is evidence to support the political axiom that presidents attempt to divert the nation's attention from economic affairs to foreign affairs when the economy is performing poorly. Two competing hypotheses are also presented and tested. Secondary research areas are whether presidents increase the number of foreign affairs speeches given and the number of days spent traveling abroad in a presidential election year, and whether presidents increase their pursuit vi

PAGE 7

of major foreign policy initiatives when the economy is performing poorly or during a presidential election year. The results show that there was insufficient evidence to support the political axiom even though the competing hypotheses were not accepted. Neither did the evidence support the hypotheses in our secondary research areas. A new variable, economic saliency, which measures attitudes about the national economy, performed quite well. Economic saliency is highly influenced by changes in unemployment and inflation. In turn, presidential job approval is significantly influenced by economic saliency. The evidence suggests that economic conditions (unemployment and inflation) are extremely important to the American public and that presidents would be wise to devote greater attention to economic affairs, rather than foreign affairs, when trying to enhance their popularity or electoral chances. Note: contact author for revisions. vii

PAGE 8

CHAPTER I INTRODUCTION Studies of the American presidency have largely been historical and anecdotal. Whether by political scientists or historians, most past studies of the presidency analyzed single administrations or, less common, compared performance in a particular area over several presidencies. These studies were rarely driven by theory, rarely invoked statistical analysis, and were often methodologically unsound. Many other works were simply biographical or autobiographical recollections of presidents and their administrations, hence subject to extreme bias. In part, political scientists were hesitant to develop theories and apply quantitative research techniques to the study of presidential behavior because of the small number of cases available. Only forty-one men have held the office to date.^ Because of advances made in statistical techniques, the increasing amount of information available to scholars, the growth of the political behavior subfield, and efforts within the discipline to develop good theory, political scientists have more recently begun to advocate presidential ^Forty-two presidents when Grover Cleveland is counted twice (1885-1889, 1893-1897). 1

PAGE 9

2 research which "may suggest statistical patterns across presidential terms" (King and Ragsdale 1988, 5). Primarily, this study is an attempt to discern whether patterns of behavior exist in some relationships between presidential action, economic performance, and presidential elections. Secondary issues will also be addressed. It has often been suggested by scholars, politicians, and the media that when domestic economic conditions are poor, presidents will attempt to divert the nation's attention from economic matters to foreign affairs. This is an untested political axiom, but worth examining because of its broad acceptance within the population (especially among elites) and the insight it may provide regarding presidential behavior. A poor economy causes social, economic, and political disruptions and spurs news stories which depict the plight of the middle and lower classes, the unemployed, and the homeless. These tragedies may affect the president's job approval rating, his re-election chances, and his party's success in congressional elections, because the electorate holds the chief executive responsible for poor economic conditions . A decrease in presidential job approval may also weaken the president's ability to promote his agenda within Congress. The result may be a further erosion of the public's confidence in their chief executive and non-cooperation

PAGE 10

3 between the executive and legislative branches. Presidents may, therefore, attempt to shift attention from the economy (a policy area difficult to control) to foreign affairs (where events may be more easily manipulated) in expectation that their job approval rating and their legislative effectiveness will be enhanced. This kind of behavior is analogous to impression management or image control discussed in social psychological literature (see Schlenker 1980). A theory of presidential behavior can now be articulated. If the above generalizations are correct, then as the economy deteriorates it should become a more salient issue to the public. As a result, the president's job approval rating (or popularity)^ should decrease. We would then expect the president to participate less frequently in press conferences (where he may be questioned directly about the ailing economy) so that he may avoid additional negative publicity. Conversely, we would expect the president to devote more attention to foreign affairs (a policy area in which the executive dominates) and travel abroad more frequently in an attempt to generate positive publicity for himself and his administration. We now model this theory of presidential behavior. ^Throughout this paper presidential job approval and presidential popularity are synonymous terms.

PAGE 11

4 Press Conferences Decrease Foreign Affairs Speeches Increase Travel Abroad Increases Figure 1. Behavioral Model Two competing hypotheses are suggested from the discussion of the theory which deserve attention and testing. First, some may argue that when economic conditions worsen presidents will address the economy forthrightly rather than avoid it. Therefore, a dependent variable which measures the number of economic speeches given by the president will be introduced into our model. Second, the number of foreign affairs speeches given and the number of days spent traveling abroad may increase when the United States is involved in war. To test this competing hypothesis a variable which measures the presence or absence of war will be introduced. Two other areas of interest will also be addressed in this study. One, do presidents give more foreign affairs speeches and devote more time to foreign travel in a presidential election year regardless of economic conditions? It may be that presidents use the publicity which surrounds these types of events as a method for improving their standing (or that of their party's presidential nominee) with the American public in order to enhance their electoral When the Economy — > Declines Economic Saliency — > Increases Presidential Job Approval — Decreases

PAGE 12

5 chances. Two, do presidents increase their pursuit of major foreign policy initiatives in either economic downturns or presidential election years in order to increase their level of support with the American public? When the economy is performing poorly, presidents may announce a major foreign policy initiative in order to regain lost points in the popularity polls or to increase their standing (or that of their party's presidential nominee) in the polls during a presidential election year. The reader should consider that a primary goal of this study is to determine whether statistical patterns of behavior exist over several presidencies. That is, do presidents behave similarly in specific situations? It is not the intent of the author to investigate patterns of behavior within singular presidencies. Similarly, controlling for the party of the president would yield too few cases to generalize from.^ We now turn to a review of some major literature regarding the American presidency. ^We are studying the behavior of three Democratic presidents and three Republican presidents.

PAGE 13

CHAPTER II PREVIOUS RESEARCH ON SELECTED ASPECTS OF PRESIDENTIAL BEHAVIOR A large body of research has been published which suggests the existence of a relationship between the economy and (1) presidential job approval, (2) the electoral chances of the president, (3) the electoral chances of the president's party in congressional elections, and (4) the president's ability to promote policy within Congress. Each of these relationships will now be discussed. Presidential Job Approval It is commonly believed that an economy which is performing poorly will have a negative effect upon the job approval rating of the president. And while "political scientists have concentrated on inflation, unemployment, and income as the major factors affecting economic voting and presidential popularity" (Lanoue 1988, 42), disagreement exists as to which economic indicators influence the public's evaluation of presidential job performance. Disagreement also exists concerning the measurement of these variables and whether lagging should be incorporated. These latter issues will be addressed in the methodology chapter. 6

PAGE 14

7 Economic variables are not the only ones that have an effect upon presidential popularity. Many researchers would agree that contextual effects such as personality, political and personal scandals, and military conflicts, among others, play an important role (see, for example, Kenski 1977, Norpoth 1984, and Kenney and Rice 1988). For our purposes we are only concerned with economic effects. In an early study Frey and Schneider (1978) found that unemployment, inflation, and disposable income were important indicators or measures for the public when they evaluated presidential job performance: A one percentage point increase in the rate of inflation decreases presidential popularity by one half to one percentage point; an increase in unemployment of one percentage point decreases popularity by somewhat more than four percentage points; and a one percentage point increase in the growth rate of nominal consumption increases presidential popularity by about seven-tenths of a percentage point. (Frey and Schneider 1978, 177178) Monroe (1978) confirmed the significance of the lagged effects of inflation on popularity and found a lagged effect for military expenditures as well. The significance of military expenditures is intuitively appealing because defense spending has often been a "hot" campaign issue in presidential elections (i.e., 1980, 1988) and because of the impact that the level of military expenditures has on domestic economic performance. Interestingly, Monroe found no effects for either unemployment or real personal income.

PAGE 15

8 Unemployment also proved unfruitful in Kernell's (1978) study; however, consumer prices (inflation) did demonstrate levels of significance. Consumer prices were also significant in Fiorina's (1981) research and his results suggest that unemployment does in fact affect presidential popularity. The research conducted by Shapiro and Conforto (1980a) also confirmed the importance of consumer prices and unemployment: We found that changes in consumer prices and the percentage of the work force unemployed affect the public's perception and evaluation of economic conditions, and that this popular assessment in turn affects disapproval of presidential performance. (Shapiro and Conforto 1980a, 65) Shapiro and Conforto found no similar effect for real disposable personal income (also see Kenski 1980, and Shapiro and Conforto 1980b, for a commentary and reply) . Inflation (lagged one guarter) was found by Norpoth (1984) to significantly influence presidential approval ratings; however, no strong effect was found for unemployment. In his study of economic conditions, presidential popularity, and their effects on midterm congressional elections, Abramowitz reported that: In 1982, an election year in which economic issues were very salient, both personal financial outlook and national economic outlook had significant effects on evaluations of President Reagan. However, of the four economic variables, expectations regarding the future of the national economy had by far the strongest impact on Reagan evaluations. (Abramowitz 1985, 37)

PAGE 16

9 This negative evaluation of President Reagan led to significant losses for Republican incumbents who were seeking re-election to the U.S. House of Representatives. Economic conditions and their effects on presidential and congressional elections will be addressed later. By combining inflation and unemployment into a "misery index" Ostrom and Simon (1985) found a significant and consistent impact upon popular support for the president. Brace and Hinckley (1991) also used the misery index in their research and found that it worked "to reduce presidential approval across administrations with an effect statistically significant at the .01 level" (p.l003). It is not clear, however, whether voters actually perceive these two economic indicators in the combined manner that Ostrom and Simon and Brace and Hinckley have presented them. Kenney and Rice (1988) note that the most common economic indicators used by political scientists are unemployment, inflation, personal income, and gross national product. In their study of presidential rankings Kenney and Rice reported that "an improving economy can contribute to a high greatness rating" by scholars (p.l65). Their measure of a healthy economy was the average annual change in per capita exports. This measure was chosen because most of the economic indicators used today were not in use during the terms of our earliest presidents.

PAGE 17

10 Some scholars disagree with the assertion that the economy affects presidential popularity. Stimson (1976) argues that presidential popularity rides an up and down cycle and is independent from presidential behavior. His model predicts a high approval rating at the beginning of the president's term, followed by a long, significant decline, with a small rally at the close of the term — but never attaining the original level. Brace and Hinckley (1991) also found that presidential popularity rides an up and down cycle and that it is artificially inflated at the beginning of each new term. However, as previously noted, their findings show that presidential popularity is affected by economic conditions (i.e., the misery index) . Brace and Hinckley contradict Stimson by demonstrating that presidential evaluations are rooted in real events and conditions rather than just time. The presidency of Ronald Reagan has also confounded Stimson 's theory. The mean approval rating for Reagan during the first quarter of 1981 was 55.3 percent. It dropped to a low of 38 percent during the first quarter of 1983 and rebounded to 60 percent in the fourth quarter of 1984 (the end of Reagan's first term). Similarly, during the first quarter of 1985, Reagan's mean approval rating was 60.5 percent. It dropped to a low of 46 percent in the first

PAGE 18

11 quarter of 1987 and rebounded to 57 percent in the fourth quarter of 1988 (the end of Reaqan's second term) The strong increases at the close of President Reagan's two terms may be atypical, but they should not be dismissed as aberrant since Stimson evaluated the terms of only five presidents and the total number of men who have occupied the office is relatively small. Presidential Elections That economic performance during a presidential election year is an important issue for candidates and voters should not be surprising. Research has shown that a variety of economic variables may be important to voters when they elect their chief executive in November. Fair (1978) developed an equation for predicting the Democratic share of the two-party vote for president. The data he examined spanned presidential elections from 19161976. His tests of three economic variables — the unemployment rate, the growth rate of real GNP (gross national product) per capita, and the growth rate of the GNP deflator — demonstrated that, in an election year, the unemployment rate and the growth rate of real GNP per capita "have an important effect on votes for president," though the latter performed slightly better (p.l71). In his update ^All figures were calculated by the author from data presented in Edwards with Gallup (1990).

PAGE 19

12 which included results from 1980, Fair (1982) found that "the two main factors contributing to the low predicted vote for Carter were the low growth rate (-4.9%) and the high inflation rate (8.7%)" (p.322). And when the 1984 figures were incorporated his equation still performed quite well: The growth rate coefficient is .0102, which means that every one percentage point increase in the growth rate leads to a 1.02 percentage point increase in the vote share for the incumbent party. The inflation rate coefficient is -.0034, which means that every one percentage point increase in the inflation rate leads to a 0.34 percentage point fall in the vote share for the incumbent party. (Fair 1988, 170) Again, it was the growth rate in real GNP per capita that had the strongest effect upon the popular vote. The effects of inflation were negligible. In his evaluation of ten presidential elections between 1948 and 1984 Erikson (1989) reported that the "cumulative weighted average of annual percentage change in per capita disposable income over the previous four years" was the best predictor of the outcome (p.568). Furthermore, "the vote is determined almost entirely by the amount of prosperity that the incumbent party delivers" (p.568). Erikson is cautious, however, and reminds his audience that both candidate quality and campaign effectiveness are just as important and should not be discounted. That is, economic factors are not the sole determinants of voter choice. Markus (1988), too, found that personal income plays an important role in voter choice. In his study of National

PAGE 20

13 Election Studies survey data (1956-1984) Markus discovered that: Each 1 percent real increase in per capita disposable income increases the probability of a proincumbent vote by .023 (+.004), net. Moreover, since objective national economic conditions are identical for every voter in a given election, that effect translates into a 2.3 percentage point change in the incumbent's share of the two-party vote for each 1 percent change in real PDI. (Markus 1988, 151) As previously discussed a poor economy may affect the job approval rating of presidents. As a result, presidential job approval may be a good predictor of presidential elections. Sigelman (1979) found, for those elections he analyzed,^ ... a clearcut positive relationship between presidential popularity and the percentage of the popular vote won by the incumbent president in the seven elections since 1940. The correlation between presidential popularity and election outcomes is +.737, which indicates a very substantial relationship. (Sigelman 1979, 533) Additionally, Sigelman found that "an incumbent whose last preelection popularity figure is 45 percent or less can expect to lose or to eke out the narrowest of victories in the popular vote column" (p.534). Lewis-Beck and Rice (1982) expanded upon the work of Sigelman and reported that the popularity figure in June preceding the November election is a better predictor of the ^Election years: 1940, 1944, 1948, 1956, 1964, 1972, and 1976.

PAGE 21

14 popular vote. Their model reduced the average predictive error and was wrong only for Harry Truman (1948). Presidents have long known that a poor economy can lead to defeat on election day. Are you better off now than you were four years ago? has become a common campaign slogan. Tufte (1978) has shown that presidents will attempt to stimulate the economy in election years by increasing real income through transfer payments such as old age, survivors, disability, and health insurance benefits, and veterans' benefits. Furthermore, Tufte believes that administrations will pursue policies which will decrease unemployment or inflation and postpone the closing of military bases and veterans' hospitals. Oppenheimer (1980) disagrees that these short-term increases in transfer payments stimulate the economy. He does concede, however, that politicians believe that these transfer payments affect their electoral chances. This is a crucial factor when trying to understand the relationship between presidents, the economy, and elections. Presidents believe that if the economy is not strong or steadily improving before election day, their chances for victory will be diminished. The following will demonstrate the importance presidents and presidential candidates place on strong economic performance during an election year.

PAGE 22

15 Vice President Richard Nixon, who sought the presidency in 1960, fully understood the role economics would play in the campaign: And as the national economy began to turn upward in the winter and spring of 1959, the Administration's standing rose accordingly — and along with it, the personal stock of all of us associated with it. (Nixon 1962, 303) Nixon was warned by economist Arthur Burns that the economy would dip shortly before the November 1960 election. However, the Eisenhower administration chose to stay the course. Nixon reflected: Unfortunately, Arthur Burns turned out to be a good prophet. The bottom of the 1960 dip did come in October and the economy started to move up again in November — after it was too late to affect the election returns. In October, usually a month of rising employment, the jobless rolls increased by 452,000. All the speeches, television broadcasts and precinct work in the world could not counteract that one hard fact. (Nixon 1962, 310311) When the economy remained sluggish during the 1976 presidential campaign Jimmy Carter pounded away at the Ford administration's failure to turn it around (see Witcover 1977). President Ford was warned that the economic recovery would stall shortly before the election and that something had to be done to stimulate the economy or there would be political repercussions. Even though the polls showed that uncertainty about the economy was hurting his campaign President Ford decided not to alter course. Ford later recalled, "I decided to gamble instead that the pause

PAGE 23

16 wouldn't hit until after election day. ... I lost" (Ford 1979, 429). And again, during the presidential race of 1992, the economy would be the focus of the campaign. For James Carville, Bill Clinton's chief political strategist, one issue stood out: THE ECONOMY, STUPID, read the sign he posted to remind everyone that nothing else was more important in their daily war games. (Newsweek 1992, 78) Congressional Elections Some research has demonstrated that economic conditions may directly affect the outcome of congressional elections — especially elections involving members of the president's party. Indirect affects have also been shown to exist through the public's evaluation of the president's job performance. It is commonly known that senators and congressmen often distance themselves from the president during presidential and congressional campaigns when the economy is performing poorly or when presidential job approval is low. Kramer (1971), in his evaluation of votes for candidates for the U.S. House of Representatives between 1896 and 1964, found that: Economic fluctuations, in particular, are important influences on congressional elections, with economic upturn helping the congressional candidates of the incumbent party, and economic decline benefitting the opposition. In quantitative terms, a 10% decrease in per capita real personal income

PAGE 24

17 would cost the incumbent administration 4 or 5 percent of the congressional vote, other things being equal. With a "swing ratio" of 2, this would translate into a loss of around 40 House seats — a respectable shift. (Kramer 1971, 140-141) Kramer found no effects for either unemployment or inflation. Tufte (1978) reported results similar to Kramer's. Tufte also believes that midterm congressional elections are "a referendum on the performance of the president and his administration's management of the economy" (p.ll5). He found that: An election-year change of 1.0 percent in real disposable income has typically produced a change of 1.1 percent in the national vote for congressional candidates of the in-party. . . . the difference between a mediocre and a buoyant electionyear economy has counted for shifts of three or four percentage points in the vote, which translates into an equivalent swing of 25 to 45 House seats. (Tufte 1978, 119) With changes in real disposable per capita income as their prime economic indicator, Hibbing and Alford (1981) discovered that the economy had a smaller impact on congressional elections than what Kramer and Tufte had discerned. Hibbing and Alford show that "changes in the economy have significant effects on the electoral fortunes of only a relatively small subset of congressmen — specifically, incumbents of the president's party" (p.434). Not all research has been supportive of the hypothesis that economic conditions affect congressional elections. For example, Fiorina (1978), when using individual-level data

PAGE 25

18 from responses to survey questions which asked if respondents' financial situation was getting better, worse, or stayed the same, found "little support for the traditional view that midterm elections constitute a referendum on the incumbent administration's handling of the economy" and that there was "no evidence that perceptions of the state of the country's economy and/or its personal impact on people had any relation to their congressional vote" (pp.435, 438). Fiorina suggests that voters are probably reacting to issues other than the economy when deciding whom to vote for in congressional elections and that voters probably blame the president, not members of Congress, for poor economic conditions . Erikson (1989) confirmed Fiorina's findings. Using per capita income growth as his measure, Erikson found that "the same electorate that votes its collective pocketbook for president virtually ignores the economy when voting for Congress" (p.397). Besides effects from the state of the economy, research demonstrates that congressional elections can be affected by the job approval ratings of the president. Kernell (1977) reports that for every nine-point change in presidential job approval the share of the vote for the president's party in congressional elections will change 1.3 or 1.4 percentage points (depending upon the party). In his study of midterm congressional elections Campbell (1985) discovered that seat

PAGE 26

19 losses for the president's party are partially associated with the president's job approval rating: . . . seat losses tend to be greater when the President is unpopular at the midterm. For every additional percentage point favorable to the President in the midterm Gallup poll, one can expect the President's party to save about one seat. (Campbell 1985, 1155) In a study of the midterm congressional elections of 1974, 1978, and 1982, Abramowitz (1985) in part concluded that "economic conditions may have indirectly affected the vote through their effects on presidential evaluations" (p.37). The effects of economic conditions were strongest in 1982; . . . President Reagan's relatively low public evaluation appears to have reduced the Republican share of the popular vote by about 2.5 percentage points compared with what would be expected in a "normal" midterm election. (Abramowitz 1985, 41) Leading Congress Intuitively, it seems logical that if a president is not popular with the American public (as measured by national opinion polls) he will have difficulty getting his legislative agenda through Congress. What member of Congress wants to be associated with the policies of an unpopular chief executive? Surprisingly, the research yields mixed results . Rivers and Rose (1985) focused their research on the size of the president's legislative agenda — the annual number of requests for action. Their findings show that

PAGE 27

20 "a 1 percent increase in a president's public support level increases the president's legislative approval rate by approximately 1 percent (holding program size fixed)" (p.l83) . Rohde and Simon (1985) studied presidential vetoes from 1945-1980. In part, they reviewed the public standing of the president (as measured by Gallup surveys) and congressional support on roll call votes. Regarding congressional attempts to override a presidential veto, Rohde and Simon report that "public support is a significant influence on both the probability of an override attempt and the probability that the attempt will be successful" (p.424). As previously mentioned not all research supports the hypothesis that decreases in presidential popularity lead to decreases in congressional support. Bond and Fleisher (1984) examined a sample of House and Senate roll call votes on which the president took a position from 1957-1980. Their dependent variable was presidential success in Congress^ and the independent variable was the percentage approving the president's job performance as measured by Gallup surveys. Their findings demonstrate that there are "very few instances when the President's popularity with the public is even moderately related to his influence in Congress" ^Bond and Fleisher tested two measures of presidential success in Congress: one, presidential wins and losses on roll call votes and two, the percentage of support the president received on each roll call vote.

PAGE 28

21 (p.303). In their conclusion. Bond and Fleisher state that we should not be surprised by this finding. They suggest that Neustadt (1986) might have been right — that it is the president's ability to persuade, bargain, and compromise in his relationships with Congress which matters most. In a similar study Zeidenstein (1985) reversed the variables to examine whether legislative success affects presidential popularity. The dependent variable was presidential popularity and the independent variable was key vote wins and losses for the president in Congress. Zeidenstein reported that: . . . Presidential popularity on a given date has no greater influence over whether a president wins or loses subsequent key votes than a key vote win or loss has influence in changing a president's popularity after the vote. In both cases, the relationships range from weak to non-existent for most administrations in both the House and Senate. It appears that whether a president is an able legislative leader does not directly affect his popularity, and his popularity does not directly affect his legislative leadership. (Zeidenstein 1985, 294) Zeidenstein's findings may not be unexpected. Political scientists generally agree that the public is not wellinformed on political matters. If this is an accurate assessment, then we would not expect that congressional vote outcomes would influence the public's evaluations of the president's job performance. If a poor economy causes presidential popularity to decline or leads to electoral defeats or an inability to lead Congress, do presidents then attempt to use foreign

PAGE 29

22 affairs and travel abroad to improve their standing with the public and Congress? And, do the frequencies of presidential speechmaking and presidential press conferences change? Foreign Affairs and Travel Abroad By examining Gallup survey results before and after major international events Lee (1977) found that a relationship exists between these kinds of events and the president's popularity. Lee reports that a majority of the public will usually support presidential action in foreign affairs at its inception (rally 'round the flag), but that public support is often short-lived. In consonance with our theory, Ostrom and Job (1986), in their study of the political use of force, concluded that "the president is more prone to use force in times of economic stress” (p.557). Could this be an attempt by the president to increase his standing with the country? This may be, since they also reported that: . . . the higher the level of presidential approval, the more likely the president will be concerned with the possibility that the use of force, if unsuccessful, could reduce his personal resources. (Ostrom and Job 1986, 557) That war affects the president's popularity ratings and his reputation should not be surprising. This relationship is suggested in a study of presidential greatness rankings (Kenney and Rice 1988). Kenney and Rice reported that "involvement in war will move a president up 11 places in

PAGE 30

23 the rankings on average" (p.l65). War can also negatively affect presidential popularity — especially when casualties begin to mount (see Mueller 1973). Foreign travel may also be a method for improving presidential popularity. Darcy and Richman (1988) believe that "travel abroad is seen by many as a device Presidents can use to gain favorable attention, thereby influencing their popular standing" (p.85). They examined Gallup survey data before and after such events (1953-1980) and found that foreign travel (especially longer trips) had a significant influence on presidential popularity, but, generally, only for Republicans. Unfortunately, Darcy and Richman offered no explanation for this anomaly. If foreign travel improves a president's standing then we would expect presidents to take more trips. Interestingly, there may be no increasing trend to take more of them. Simon and Ostrom (1989) examined presidential travel from 1953-1987 and reported that "on average, the president can be expected to take two or three foreign trips per year" (p.71). Furthermore, they found that foreign travel must be accompanied by a significant event before it impacts on presidential job approval. Some international events and foreign trips may have significant short-term effects upon presidential popularity. If an approval-enhancing event closely preceded an election it could alter the predicted outcome.

PAGE 31

24 Speeches and Press Conferences Our theory states, in part, that when the economy is performing poorly the president will devote more attention to foreign affairs in his speeches and decrease the frequency of press conferences. Research on presidential speechmaking and press conferences has produced mixed findings . Ragsdale (1984) developed an equation for predicting when a president would give a major speech. Her definition of "major speech" is a presidential address given before a live television or radio audience. The data base included speeches from 1949-1980. Contrary to our theory Ragsdale found that "as the unemployment situation worsens, presidents are less likely to deliver a major address" and that "presidents appear to be more likely to address the nation when troops are being removed and the foreign conflict is presumably winding down" (pp. 976-977 ) . The reader should note that the set of speeches used in the present study is significantly larger than the set used by Ragsdale because we include speeches which were not given live before a television or radio audience. Consequently, our research may yield different results (see methodology chapter) . Simon and Ostrom (1989) found no evidence that speechmaking is on the rise. In fact, they report that "presidential use of television has declined since the Johnson and

PAGE 32

25 Nixon administrations" (p.71). Their findings also demonstrated that speeches by themselves did not enhance presidential support, but they did when "given in conjunction with a trip and an approval-enhancing event" (p.74). On average the increase in presidential approval was 4.17 rating points (Simon and Ostrom 1989, 74). Contrary to Ragsdale (1984), Simon and Ostrom (1989) believe that "it is the event rather than the speech that is responsible for triggering the rally effect" (p.79). Their findings give support to the theory that presidents may seek a foreign policy success or engage in foreign travel for the purpose of improving presidential popularity. In a study of presidential press conferences hammers (1981) found that foreign policy difficulties led to a decrease in the number of press conferences held. And, in contradiction to our theory, he reported that: Despite their periodic denials, presidents do seem to read the polls as they schedule [press] conferences — but more often to increase rather than decrease [press] conference freguency as their popularity falls, (hammers 1981, 270) The reader should note that research in this area is scant,* although the findings by hammers should not be dismissed and our research may yield different results. Some research has demonstrated that poor economic conditions may negatively affect the president's popularity. ‘For a short, nonquantitative history and assessment of presidential press conferences, see Smith (1990).

PAGE 33

26 his electoral chances, his party's chances in congressional elections, and his ability to lead Congress. Major international events, speeches, and foreign travel may help the president improve his standing with the public. These aspects of presidential behavior have not been brought together under the umbrella of one study, nor have some of the aforementioned studies included the more recent presidents, nor has our theory been previously tested. In the following chapter we will specifically develop the hypotheses, operationalize the variables, and discuss the method of analysis.

PAGE 34

CHAPTER III METHODOLOGY This is a quasi-experiinental design in which the subjects are six United States presidents, John F. Kennedy (1961) through Ronald W. Reagan (1988).^ We begin with Kennedy because his administration was the first to televise press conferences on a regular basis and the number of households with televisions surpassed the 90 percent mark at roughly the same time that he took office (see chapters 3 and 4 in Kernell 1986). This is an important distinction because press conferences now became national media events and a method for presidents to communicate directly and immediately with a very large proportion of the public. The data end with the Reagan administration because complete information for several variables was not available for the term during which George Bush was president. Information for all variables in this study is complete for all quarters unless otherwise noted. To reiterate, our theory states that as the economy worsens it becomes a more salient issue to the public. As a result, the president's job approval will go down. In ^Though Reagan continued to serve as president through the first three weeks of January 1989, data for the first quarter of 1989 are omitted. 27

PAGE 35

28 response, the president will give fewer press conferences and try to divert the nation's attention from domestic to foreign affairs by (1) giving more speeches which concern foreign affairs and (2) traveling abroad more freguently. Nine independent variables which measure economic performance are incorporated. Economic saliency and presidential job approval are intervening variables and both measure responses to survey guestions . The number of press conferences held, the number of foreign affairs speeches given, and the number of days spent traveling abroad are all dependent variables. The first step entails the regression of economic saliency on our nine economic measures. Monetary variables are standardized in 1987 dollars.^ The following will describe each variable, its code, and source. Economic saliency (ECSAL) is defined as the mean quarterly percent of those who cited an economic problem in response to the survey question "What do you think is the most important problem facing the [this] country today?" An economic problem is defined as: cost of living, inflation, taxes, unemployment, balanced budget, depression, recession, interest rates, and the trade deficit, among others. These are broad, national economic problems which concern citizens as opposed to specific social problems such as welfare or ^Standardization adjusts for inflation.

PAGE 36

29 poverty, even though the latter surely have economic repercussions . Responses to the surveys were assigned to the quarters in which the interviews occurred. Results from interviews which spanned two quarters were discarded since we could not determine which quarter to assign them to. Responses to variations of the question were not included, nor were responses which were not from surveys of the general population.^ This survey question has been asked irregularly by the major polling organizations. Of the 112 quarters in the data base, 54 show missing data for this variable (for a loss of 48 percent) . The mean quarterly figure can exceed 100 percent due to multiple responses. Source: most of the data are from responses to Gallup surveys (Gallup Organization, no date). Data for eight quarters* are from responses to surveys conducted by CBS News/New York Times (Staff, New York Times, no date). Nine independent variables which measure economic performance have been chosen. The first two, unemployment and inflation, have often been used by scholars in similar studies (see review of literature) and certainly affect ^Some of the original data were from surveys of selected subgroups, rather than the general population, and were not included in our data set. “Quarter/year : 2/1981, 3/1981, 2/1986, 4/1986, 1/1987, 4/1987, 2/1988, and 3/1988.

PAGE 37

30 personal economic well-being. They also receive much media attention. 1. The unemployment rate (UNEMP) is a measure of those who have been unemployed fifteen weeks or longer as a percent of the total work force. This is a guarterly figure for all workers, 16 years of age or older, and it is seasonally adjusted. Source; Citibase (1978).® 2. The inflation rate (INFL) is the mean quarterly percent rate of change in the Consumer Price Index (CPI). The CPI is a monthly measure of the cost of a hypothetical basket of goods for all urban consumers, for all items, and is seasonally adjusted. The base period is 1982-84 = 100.® Source: for the CPI, Citibase (1978); see footnote 5. The percent rates of change were calculated by the author using a statistical program.’ The third economic indicator chosen is the prime lending rate. Though not often used in similar studies, it is included here because of the impact the rate has on the ®The primary source for this and other economic variables is the U.S. Department of Commerce, Bureau of Economic Analysis, but it is cited in the text and bibliography as Citibase because this is the data file from which the figures were generated. Though the copyright date is 1978 the data available surpass that year. ®The CPI has not yet been calculated with a base year beyond 1982-84. ’The author used Minitab, Release 8, for the personal computer. Minitab is a registered trademark of Minitab, Inc., State College, PA.

PAGE 38

31 purchase of large consumer goods such as automobiles, boats, housing, and real estate, and on credit card interest rates. It, too, receives a great amount of media attention. 3. The prime lending rate (PRIME) is the rate of interest charged by banks on short-term business loans to their best corporate customers, per annum. The data are recorded on a mean guarterly basis. Source: Citibase (1978); see footnote 5. Our next economic indicators are per capita real disposable income, per capita gross national product, and per capita national defense expenditures. Even if the average citizen is not familiar with the meaning or interpretation of these three indicators, most will perceive the repercussions when these indicators change significantly. Change in real disposable income affects the purchasing level of individuals and families. Significant decreases in the levels of GNP and national defense expenditures can lead to a sluggish economy — even recession. These economic declines can be devastating to industries, communities, and families. It is not clear from prior research (see review of literature), however, whether it is the levels or the percent rates of change in these three variables which is more significant. Therefore, both the levels and the percent rates of change will be tested (thereby creating three additional variables) .

PAGE 39

32 Data for real disposable income are available on a per capita basis. It was necessary to calculate the per capita figures for gross national product and national defense expenditures. This was done by dividing the quarterly figure for each variable by the current estimate of the population of the United States.® 4. Per capita real disposable income (INC) is the quarterly measure of the average worker's real disposable income. Real disposable income is defined as personal income less personal taxes. Personal taxes are comprised of income, property, and inheritance taxes (see McConnell 1975, 182). The quarterly figure in 1987 dollars, annual rate, is recorded. Source: Datadisk (1984).® 5. Per capita gross national product (GNP) is the quarterly measure of the productivity of the nation (gross national product) divided by the current population estimate. This figure is recorded in 1987 dollars. Source: Citibase (1978); see footnote 5. 6. Per capita national defense expenditures (DBF) is the quarterly measure of national defense expenditures by ®Population figures for 1967 through 1988 are from the Bureau of the Census and reproduced from a data file (Data Resources, Inc.). Figures for 1961 through 1966 are from U.S. Department of Commerce, Bureau of the Census (1979). ®Even though the copyright date is 1984 the data available surpass that year.

PAGE 40

33 the federal government in 1987 dollars divided by the current population estimate. Source: the data for 1972 through 1988 are from U.S. Department of Commerce (1992, Table 3.11, 118-119). There are no data for defense expenditures in 1987 dollars prior to 1972, therefore, expenditures have been estimated. The method chosen^° prescribes taking the actual dollar amount spent in each quarter (1961-1971) and dividing it by the implicit price deflator (1987 = 100) for the federal government for the corresponding quarter (1961-1971). Actual dollar figures were obtained from U.S. Department of Commerce (1992, Table 1.1, 1-2). The price deflator was obtained from U.S. Department of Commerce (1992, Table 7.14, line 12, 298299) . 7 . Rate of change in per capita real disposable income (CINC) is the quarterly percent rate of change in per capita real disposable income (see number 4 above). Source: calculated by the author using a statistical program; see footnote 7. 8 . Rate of change in per capita gross national product (CGNP) is the quarterly percent rate of change in per capita gross national product (see number 5 above). Source: calculated by the author using a statistical program; see footnote 7. ^°As recommended by Mr. Carl Galbraith, an economist with the U.S. Department of Commerce.

PAGE 41

34 9 . Rate of change in per capita national defense expenditures (CDEF) is the quarterly rate of change in per capita national defense expenditures (see number 6 above). Source: calculated by the author using a statistical program; see footnote 7. From the above variables we can now formulate our first equation Economic Saliency (ECSAL) = a + bl(UNEMP) + b2(INFL) + b3(PRIME) + b4(INC) + b5(GNP) + b6(DEF) + b7(CINC) + b8(CGNP) + b9(CDEF) + e Nine hypotheses can now be tested: HI: As the unemployment rate increases, economic saliency increases (bl > 0). H2: As the inflation rate increases, economic saliency increases (b2 > 0). H3: As the prime lending rate increases, economic saliency increases (b3 > 0). H4 : As per capita real disposable income increases, economic saliency decreases (b4 < 0). H5: As per capita gross national product increases, economic saliency decreases (b5 < 0). ^Â’^For all equations in this paper, a = Y-intercept, b = slope, and e = error term.

PAGE 42

35 H6: As per capita national defense expenditures increase, economic saliency decreases (b6 < 0). H7 : As the rate of change in per capita real disposable income increases, economic saliency decreases (b7 < 0). H8 : As the rate of change in per capita gross national product increases, economic saliency decreases (b8 < 0) . H9: As the rate of change in per capita national defense expenditures increases, economic saliency decreases (b9 < 0) . Our second intervening variable, presidential job approval, will now be regressed on economic saliency. The president's job approval rating will be measured by tracking positive responses to a guestion frequently asked in national opinion surveys. Presidential iob approval (APPROV) is the mean quarterly percent of those who cited approval when asked "Do you approve or disapprove of the way is handling his job as president?" This question is asked very frequently; only three quarters^^ have missing data (for a loss of sixtenths of 1 percent) . Responses to variations of the question were not included nor were responses from polls which did not utilize a sample of the general population. As was the case for economic saliency, responses from interviews which spanned two quarters were discarded. ^^Quarter/year : 3/1964, 3/1972, and 3/1976.

PAGE 43

36 Source: all of the data are from Gallup surveys (Edwards with Gallup 1990), except for the third quarter of 1972 which is from a poll conducted by the Opinion Research Corporation . From presidential job approval and economic saliency we formulate the equation: Presidential Job Approval (APPROV) = a + bl(ECSAL) + e We have one hypothesis to test: HIO: As economic saliency increases, presidential job approval decreases (bl < 0). To complete the test of our theory we will now regress our three dependent variables — the number of press conferences held, the number of foreign affairs speeches given (three categories), and the number of days spent traveling abroad, on presidential job approval. The following will describe each dependent variable. 1. Presidential press conferences (PRESS) is the number of press conferences held each quarter by the president. Source: data for Kennedy through Carter are from the series The Cumulated Indexes to the Public Papers of the Presidents of the United States . Data for Ronald Reagan were provided by Staff, Reagan Library (1990). ^^The figure was provided by Mr. Rob Turson, Roper Center, in a telephone conversation.

PAGE 44

37 2. Foreign affairs speeches given by the president are recorded quarterly in three categories: (1) major, (2) secondary, and (3) total, or the sum of major and secondary. The following will describe the coding process. As a starting point we begin with, and then expand upon, the definitions developed by hammers (1982), Ragsdale (1984, 1987), and Simon and Ostrom (1989). For our purposes a speech is any address given by the president which is one thousand words or more in length. This is roughly two pages of single-spaced text. However, to avoid eliminating an important speech which is less than one thousand words, flexibility and contextual analysis are required. For example, President Johnson's televised address to the nation on August 4, 1964 announcing that U.S. military forces would respond to attacks in the Gulf of Tonkin was much less than one thousand words, but was included in our data set. To have omitted this speech, or similar speeches of national importance, would have harmed our data base. Speeches made by the president, whether required (i.e., the State of the Union) or discretionary, may affect the opinions people hold. Nationally televised or not, when a president speaks publicly some segment of the population will be affected and the message will be carried by some part of the news media. For these reasons all speeches made by the president, regardless of location or level of media attention, are considered.

PAGE 45

38 A major foreign affairs speech (MFAS) is defined as any speech, or part thereof, which concerns diplomacy, summitry, treaties, executive agreements, national defense matters (including troop deployments), and crises, and is given either before a live television audience, a joint session of Congress, a press conference,^* a nominating convention, an international body (i.e., the Organization of American States), or a foreign government (i.e., the British Parliament). The speech, or part thereof, must address at least one foreign affairs issue or problem. A secondary foreign affairs speech (SFAS) is defined as having the same content, but not given before a live television audience, the Congress, at the beginning of a press conference, at a nominating convention, or before an international body or foreign government. Total foreign affairs speeches (TFAS) is defined as the sum of all major and secondary speeches. Because our theory could not predict which of these three categories of speeches would demonstrate the best relationship, each will be tested. For now, total foreign affairs speeches will be used in our eguations. Source: for all presidents we will utilize the series Public Papers of the Presidents of the United States which ^*If a prepared address given by the president at the beginning of a press conference meets the aforementioned reguirements , then it is counted as a speech and included in our data set.

PAGE 46

39 compiles all of the presidents' speeches, press conferences, and executive agreements, etc. 3. Presidential travel abroad (TRAV) is the number of days each quarter that the president or president-elect is traveling outside the United States. Source: U.S. Department of State (1990).^® From our dependent and intervening variables we can now formulate three equations with corresponding hypotheses for testing. To reiterate, we theorize that when presidential job approval decreases, presidents will give fewer press conferences, give more foreign affairs speeches, and spend more time traveling abroad. Presidential Press Conferences (PRESS) = a + bl(APPROV) + e Hll: As presidential job approval increases, presidential press conferences increase (bl > 0). And, Total Foreign Affairs Speeches (TFAS) = a + bl(APPROV) + e H12: As presidential job approval increases, total foreign affairs speeches decrease (bl < 0). And, ^^Several scholars have used the indexes to the New York Times for this variable, but the indexes understate the true number of days presidents have spent traveling abroad.

PAGE 47

40 Presidential Travel Abroad (TRAV) = a + bl(APPROV) + e H13: As presidential job approval increases, presidential travel abroad decreases (bl < 0). We now address the two competing hypotheses presented in the introduction. First, rather than avoid economic matters as theorized presidents may forthrightly address a declining economy. Therefore, we will introduce a dependent variable which measures the number of economic affairs speeches given by the president (three categories) and regress it on presidential job approval. Economic affairs speeches given by the president are recorded quarterly and in three categories: (1) major, (2) secondary, and (3) total, the sum of major and secondary. The following will describe the coding process. The codes for these speeches are major (MEAS) , secondary (SEAS), and total (TEAS). The reasons for three categories and their definitions are the same as those for foreign affairs speeches except, of course, where content is concerned. An economic speech concerns cost of living, inflation, unemployment, recession, and the trade deficit, among others. These are broad problems and are meant to coincide with the responses to the "most important problem" survey question previously discussed. The speech, or part thereof, must address at least one economic affairs issue or problem.

PAGE 48

41 Source: we will utilize the series Public Papers of the Presidents of the United States . For the present total economic affairs speeches will be used in our equation. We now have the following equation and hypothesis to test: Total Economic Affairs Speeches (TEAS) = a + bl(APPROV) + e H14: As presidential job approval increases, the number of total economic affairs speeches decreases (bl < 0). The second competing hypothesis states that when the nation is involved in war presidents will give more speeches which concern foreign affairs and devote more time to foreign travel because of the national and international attention the military conflict will generate. Therefore, a variable which measures the presence or absence of war will be introduced and added to two of our previous equations. War (WAR) measures whether the armed forces of the United States are engaged in combat; 1 = yes, 0 = no. The following criteria are used to define combat: (1) the event lasts at least five days, (2) live fire occurs between U.S. armed forces and an enemy, and (3) the event is not solely an evacuation. The reader is referred to Appendix A for a list of events which meet these criteria. Source: Congressional Research Service (1989, 1992) and Clodfelter (1992).

PAGE 49

42 We now have two additional hypotheses to test: Total Foreign Affairs Speeches (TFAS) = a + bl(APPROV) + b2(WAR) + e H15: When war is present, the number of total foreign affairs speeches increases (b2 > 0). And, Presidential Travel Abroad (TRAV) = a + bl(APPROV) + b2(WAR) + e H16: When war is present, presidential travel abroad increases (b2 > 0). Two other areas of inquiry will now be considered. First, do foreign affairs speeches and foreign travel increase during presidential election years, regardless of economic conditions? We now introduce a new variable. Presidential election year (ELEC) denotes whether a presidential election year is present; 1 = yes, 0 = no. The presidential election years are 1964, 1968, 1972, 1976, 1980, 1984, and 1988. If we restricted this variable to incumbent presidents who had served a full term and who had sought re-election, then only three presidents would have qualified: Nixon (1972), Carter (1980), and Reagan (1984). Generalizations made from three cases would be untenable.

PAGE 50

43 The following equations and hypotheses are now considered: Total Foreign Affairs Speeches (TFAS) = a + bl(ELEC) + e H17: In a presidential election year, the number of total foreign affairs speeches increases (bl > 0). And, Presidential Travel Abroad (TRAV) = a + bl(ELEC) + e H18: In a presidential election year, presidential travel abroad increases (bl > 0). Our second area of interest is whether presidents increase their pursuit of major foreign policy initiatives in either economic downturns or presidential election years. A new dependent variable, major foreign policy initiatives, will be introduced and tested. It is necessary to create this new variable because the variable major foreign policy speeches (MFAS) does not suffice. Many foreign policy speeches coded as "major” do not announce or propose actions taken or to be taken by the United States in the areas of foreign or military affairs. It is this characteristic which is of interest now. Major foreign policy initiatives (MFPOL) are speeches made by the president which announce or propose a signifi-

PAGE 51

44 cant initiative or action in either foreign or military affairs. These are recorded quarterly. The reader is referred to Appendix B for a list of initiatives. Source: the list of initiatives was developed by using the rally events compiled by Brody (1984, Table 1, 43) and the speeches and documents recorded in the series Public Papers of the Presidents of the United States (United States, President 1961-1989). The first test involves regressing major foreign policy initiatives (MFPOL) on our nine economic variables. The following equation and hypotheses are developed. Major Foreign Policy Initiatives (MFPOL) = a + bl(UNEMP) + b2(INFL) + b3(PRIME) + b4(INC) + b5(GNP) + b6(DEF) + b7(CINC) + b8(CGNP) + b9(CDEF) + e H19: When the unemployment rate increases, major foreign policy initiatives increase (bl > 0). H20: When the inflation rate increases, major foreign policy initiatives increase (b2 > 0). H21: When the prime lending rate increases, major foreign policy initiatives increase (b3 > 0). H22: When per capita real disposable income increases, major foreign policy initiatives decrease (b4 < 0). H23: When per capita gross national product increases, major foreign policy initiatives decrease (b5 < 0).

PAGE 52

45 H24: When per capita national defense expenditures increase, major foreign policy initiatives decrease (b6 < 0). H25: When the rate of change in per capita real disposable income increases, major foreign policy initiatives decrease (b7 < 0). H26: When the rate of change in per capita gross national product increases, major foreign policy initiatives decrease (b8 < 0). H27: When the rate of change in per capita national defense expenditures increases, major foreign policy initiatives decrease (b9 < 0). Our next test involves regressing major foreign policy initiatives on the variable presidential election year. The following is presented. Major Foreign Policy Initiatives (MFPOL) = a + bl(ELEC) + e H28: When a presidential election year is present, major foreign policy initiatives increase (bl > 0). Several additional variables were created which were helpful during the recording of the data and they may prove useful for future research. Year (YR) — a four-digit number. Quarter (QTR) — 1 = first, 2 = second, 3 = third, 4 = fourth.

PAGE 53

46 A dummy variable was created for each president: JFK (John F. Kennedy), LBJ (Lyndon B. Johnson), RMN (Richard M. Nixon), GRF (Gerald R. Ford), JEC (Jimmy E. Carter), and RWR (Ronald W. Reagan). Quarters which were shared by two presidents are coded for missing data.^® The reader is referred to Appendix C for a complete list of variables and their codes. Whether some variables should be tested for lagged effects is a methodological dilemma. With regards to one area of presidency research — economic effects on presidential popularity — the following is considered. In their study of economic performance and presidential popularity, Norpoth and Yantek found that: The evidence. . . .is too mixed to support a lag structure for either inflation or unemployment. (Norpoth and Yantek 1983, 796) In a separate study, Norpoth found that a . . . tentative specification of the influence of economic performance on presidential popularity includes a contemporaneous effect for both unemployment and inflation, as well as a one-quarter lagged effect for inflation. (Norpoth 1984, 262) ^®The fourth quarter of 1963 was shared by presidents Kennedy and Johnson because Kennedy was assassinated. The third quarter of 1974 was shared by presidents Nixon and Ford due to the resignation of Nixon. So that the actions of two presidents would not be combined in the same quarter, the following variables which concern presidential behavior are coded for missing data for the aforementioned quarters: presidential press conferences, presidential travel abroad, major foreign affairs speeches, secondary foreign affairs speeches, total foreign affairs speeches, major economic affairs speeches, secondary economic affairs speeches, total economic affairs speeches, major foreign policy initiatives, war, and presidential job approval.

PAGE 54

47 Norpoth found no lagged effects for unemployment. Based on the findings of Norpoth and Yantek (1983), Brace and Hinckley (1991) avoided lagged effects and, as noted in the review of literature, still found significant economic effects (i.e., the misery index) on presidential popularity. Ostrom and Simon (1989), in their evaluation of popular support for President Reagan, reported that unemployment had a significant impact on public evaluations and that it was "over eight times greater than that of inflation" (p.378). Obviously, the findings are mixed. Yet, there were some significant results when a lagging function was utilized. Citizens may not necessarily react immediately to a declining economy, rather, they may reflect on their (or the country's) economic situation for a short while before their anger or disappointment is projected upon the president or other political institutions. Conversely, present opinions may be affected by past conditions and events. Lagging is an appropriate methodological tool for assessing or measuring these types of behaviors and this tool is utilized in the present study. We have twelve eguations with a total of 28 hypotheses. Several methodological procedures will now be restated for clarity. Both foreign affairs and economic affairs speeches have been divided into three categories — major, secondary, and total. Each will be tested because we could not predict

PAGE 55

48 which category would demonstrate the best relationship with our other variables. Similarly, we will test both the levels and the percent rates of change in per capita real disposable income, per capita gross national product, and per capita national defense expenditures. The data for this study constitute a time series and they are recorded on a quarterly basis. When appropriate the mean for each quarter is used. Quarterly data are common in econometric models because they tend to smooth the data points. This should soften the effects of occasionally large variations within some variables. Since this study is concerned with patterns of behavior over a period of nearly three decades the significance of any one event is unimportant. Monetary variables have been standardized in 1987 dollars and the consumer price index is standardized as 1982-84 = 100. When a president is leaving office in January his speeches, press conferences, etc. for January are not included in the data for the first quarter; only data for the new president are recorded. A SAS^Â’ statistical program (release 6.08) will be utilized for analyzing the data. The PROG REG routine calculates ordinary least squares (OLS) estimates. When a lag^'SAS is a registered trademark of SAS Institute Inc., Cary, NC.

PAGE 56

49 ging function is needed the PROC AUTOREG routine will be used. The latter uses OLS for the linear component and the Yule-Walker method (generalized least squares) for the autoregressive component of the models. Our tables present data for several statistics: the unstandardized parameter estimates (b), the standard errors (s.e.), the probabilities (p-values), which are one-tailed tests, the R-square statistic, the Durbin-Watson statistic for autocorrelation,^® and sample size (N) . The following chapter will present the results from the various regressions. Chapter 5 will discuss the conclusions drawn from our analysis. ^®Our variables are significant when the p-values are <= . 05 . ^®Ideally, the Durbin-Watson statistic should be 2; however, there is a range about this numeral which is acceptable. To determine whether the Durbin-Watson statistics from our regressions fell in the acceptable range. Table VIII, Critical Points of the Durbin-Watson Test for Autocorrelation, was used as a reference (see Wonnacott and Wonnacott 1979, 544).

PAGE 57

CHAPTER IV RESULTS Before the results of the regressions are presented we report that there are significant levels of multicollinearity among our nine economic variables (see Appendix D for a correlation matrix) . In a regression framework unemployment is the strongest explanatory variable for economic saliency (r= 0.84). The inflation rate also significantly affects economic saliency (r= 0.50) and it (inflation) does not correlate with unemployment (r= 0.07). Our first model (see methodology chapter) regresses the variable economic saliency on nine economic indicators. Because economic saliency has a small sample size (N= 58), the number of regressors must be limited. Therefore, we will first regress economic saliency on six economic indicators while omitting the three variables which measure the percent rates of change in per capita real disposable income, per capita GNP, and per capita national defense expenditures (Model I, Table 1). Our second regression will include the three change variables, but omit per capita real disposable income, per capita GNP, and per capita national defense expenditures (Model II, Table 1). 50

PAGE 58

51 Table 1. Regression of Economic Saliency On Nine Economic Indicators Variable Intercept Unemployment Inflation Prime Rate Income GNP Defense Change Inc Change GNP Change Def MODEL I b. f s.e. 1 -62.232 (40.189) 13.179 *** (1.985) 7.447 ** (3.198) 0.339 (0.697) 0.011 ( 0 . 011 ) 0.010 (0.009) -0.019 * ( 0 . 011 ) MODEL II b. f s.e . ) -55.586 (6.821) 14.049 *** (1.150) 11.972 *** (2.387) 0.201 (0.659) -0.209 (1.826) -0.616 (1.616) 0.748 (0.642) MODEL III b. fs.e. 1 -33.238 (22.106) 12.106 *** (1.374) 9.253 *** (2.189) 0.0009 ( 0 . 001 ) -0.017 * ( 0 . 010 ) R-square 0.87 Durbin-Watson 1.44 N 58 0.86 1 . 52 58 0.87 1.47 58 * p <= .05, one-tail ** p <= .01, one-tail *** p < .001, one-tail

PAGE 59

52 Table 2. Regression of Economic Saliency On Nine Economic Indicators. Continued Variable MODEL IV b . f s . e . ) MODEL V b . f s . e . 1 MODEL VI b . C s . e . 1 Intercept -25.412 (19.376) -58 . 341 (6.067) 157.468 (7.263) Unemployment 12.591 ** (1.205) 14.064 ** (0.897) 14.034 ** (1.098) Inflation 9.517 ** (2.152) 12.165 ** (1.590) 11.642 ** (1.930) Defense -0.018 * (0.010) R-sguare Durbin-Watson N 0.87 1.46 58 0.86 1.42 58 0.87 2.04 57 * P <= .05, one-tail ** P < .001, one-tail From Model I we see that unemployment, inflation, and per capita national defense expenditures are significant and their signs are in the correct direction. The Durbin-Watson statistic (1.44) indicates the presence of autocorrelation. And in our second test (Model II), only unemployment and inflation are significant. Again, the Durbin-Watson statistic indicates autocorrelation (1.52). Per capita real disposable income is highly correlated with per capita GNP (r= 0.99). To test whether per capita real disposable income would become significant if per capita GNP was omitted, per capita real disposable income was added to a model with unemployment, inflation, and per cap-

PAGE 60

53 ita national defense expenditures (Model III). Per capita real disposable income did not reach significance, however (p-value= 0.22). From Models I, II, and III we see that unemployment, inflation, and per capita national defense expenditures significantly affect economic saliency. Per capita national defense expenditures are highly correlated with unemployment (r= -0.60) and inflation (r= -0.60). Model IV regresses economic saliency on unemployment, inflation, and per capita national defense expenditures. All three variables are significant and the R-square is 0.87 ( Durbin-Watson= 1.46). Model V omits per capita national defense expenditures. Both unemployment and inflation remain very significant, and the R-square (0.86) and Durbin-Watson statistic (1.42) are almost identical to those in Model IV. It is now evident that per capita national defense expenditures should be excluded from our final model because it does not add any significant explanatory power. Now that we have specified the correct model we must correct for autocorrelation. Model VI regresses the variable economic saliency on unemployment and inflation with a lag of one quarter. The Durbin-Watson statistic is now in the acceptable range (2.04). Our parameter estimates remain very significant and the R-square statistic changed very little. We report that for each 1 percent increase in unemployment, economic saliency increases 14 percentage points and.

PAGE 61

54 for each 1 percent increase in inflation, economic saliency increases by roughly 12 percentage points. The effects are delayed one quarter. Together, unemployment and inflation explain most of the variance of economic saliency (R-square= 0.87) . Two of our research hypotheses are accepted. First, as the unemployment rate increases, economic saliency increases (HI) and second, as the inflation rate increases, economic saliency increases (H2). We do not reject the null hypotheses for the remainder (H3 through H9). Our next model (Table 3) regresses presidential job approval on economic saliency. From Model I we see that economic saliency is significant and the sign is in the correct direction. The Durbin-Watson statistic (0.83) indicates significant autocorrelation, however. From Model II we see that lagging one quarter eliminated the autocorrelation problem. The Durbin-Watson statistic increased to 1.90, which is in the correct range. The variable conomic saliency remained significant and the sign is in the correct direction. Also note that the R-square statistic increased from 0.16 to 0.52. Our final model (Model II) shows that, for each 1 percent increase in economic saliency, presidential job approval decreases by 0.11 percentage points. Or, for each ten point increase in economic saliency, presidential job approval will decrease roughly 1 percentage point. The

PAGE 62

55 effect is delayed one quarter. We accept our hypothesis that when economic saliency increases, presidential job approval decreases (HIO). Table 3 . Regression of Presidential Job Approval On Economic Saliency MODEL I Variable b. f s.e. 1 MODEL II b, f s.e. 1 Intercept 59.069 (2.887) 57.420 (3.469) Ec. Saliency -0.158 ** (0.049) -0.117 * (0.061) R-square 0.16 Durbin-Watson 0.83 N 55 * P <= .05, one-tail ** P <= .001, one-tail 0.52 1.90 54 Before we present the remaining regressions, we should first determine whether unemployment and/or inflation have direct affects on presidential job approval (see Table 4). Model I regresses presidential job approval on economic saliency, unemployment, and inflation (with one lag to correct for autocorrelation). We see that inflation is significant (p-value= 0.001), while economic saliency and unemployment are not. When presidential job approval was regressed on just economic saliency and inflation (one lag), only

PAGE 63

56 inflation was significant again (p-value < .001). The p-value for economic saliency was 0.15.^ The reason economic saliency became insignificant in the two previous regressions is that when this variable is placed on the right side of an equation with unemployment and/or inflation, multicollinearity ensues. Economic saliency is highly correlated with both unemployment (r= 0.84) and inflation (r= 0.50), therefore, we will remove this variable and continue our tests for direct affects. Model II regresses presidential job approval on unemployment and inflation (with two lags to correct for autocorrelation). Again, only inflation is significant (p-value= 0.003). From Model III, which regresses presidential job approval on just inflation (with two lags to correct for autocorrelation), we see that for each one point increase in inflation, presidential job approval decreases 3.2 percentage points. In Models II and III inflation accounts for a significant portion of the variance in presidential job approval (R-square= 0.73). We have presented evidence which shows that unemployment and inflation indirectly affect presidential job approval through economic saliency, and that inflation directly affects presidential job approval. No direct effects from unemployment were found. Our final model (see Figure 2, chapter 5) will depict both paths. ^This regression is not shown.

PAGE 64

57 Table 4. Regression of Presidential Job Approval on Economic Saliency. Unemployment and Inflation Variable MODEL I b, f s.e. 1 MODEL II b. fs.e. > MODEL III b, f s.e. 1 Intercept 62.639 (8.787) 63.934 (6.170) 57.246 (2.151) Ec. Saliency -0.062 (0.111) Unemployment 0.133 (1.815) -1.100 (0.941) Inflation -6.474 *** (2.015) -3.255 ** (1.172) -3.274 ** (1.176) R-square Durbin-Watson 0.51 1.78 0.73 1.89 0.73 1.86 N 54 104 104 * P <= .05, one-tail ** P < .01, one-tail *** p <= .001, one-tail The final tests of our theory necessitates the regression of presidential press conferences, foreign affairs speeches (three categories), and presidential travel abroad on presidential job approval. We begin with presidential press conferences. Table 5 presents results from three regressions of the variable presidential press conferences on presidential job approval. In Model I, presidential job approval is significant and the sign is in the correct direction, but autocorrelation is present ( Durbin-Watson= 0.85). Lagging one

PAGE 65

58 quarter (Model II) and two quarters (Model III) qreatly improved the Durbin-Watson statistics, but presidential job approval lost significance in both models. The p-values increased to 0.08 (Model I) and to 0.12 (Model II). We therefore reject the hypothesis that when presidential job approval increases, presidential press conferences increase (Hll) . Table 5. Regression of Presidential Press Conferences On Presidential Job Approval Variable MODEL I b, f s.e. 1 MODEL II b . f s . e . I MODEL III b, f s.e. I Intercept 0.595 (1.200) 1.124 (1.696) 1.512 (1.724) Job Approval 0.051 ** (0.022) 0.042 (0.030) 0.035 (0.030) R-square Durbin-Watson N 0.04 0.85 107 0.36 2.12 106 0.37 2.00 105 * P <= .05, one-tail ** p <= .01, one-tail Next we regressed our three categories of foreign affairs speeches on presidential job approval. We begin with major speeches. Major foreign affairs speeches was not significant even when lagged one and two quarters. The lowest p-value reached for presidential job approval was 0.11 at two lags. Both secondary and total foreign affairs speeches

PAGE 66

59 were significant and they yielded almost identical results. However, autocorrelation was present in both models. Table 6 presents the results of secondary foreign affairs speeches (Model I) and total foreign affairs speeches (Model II) regressed on presidential job approval; both models are lagged once. Presidential job approval is significant in both models and the signs are in the correct direction. The Durbin-Watson statistics are in the proper range as well. Because the results for the two models are nearly identical, and because total foreign affairs speeches combines two categories, secondary foreign affairs speeches will be chosen for our final model. Table 6. Regression of Foreign Affairs Speeches On Presidential Job Approval MODEL I MODEL II Variable b. f s.e. 1 b. f s.e. 1 Intercept 20.373 (4.655) 21.898 (4.906) Job Approval -0.180 ** (0.085) -0.163 * (0.089) R-sguare Durbin-Watson N 106 0.13 2.04 106 0 .13 2.05 * P <= .05, one-tail ** P <= .01, one-tail

PAGE 67

60 From Model I we see that for each 1 percent increase in presidential job approval, secondary foreign affairs speeches decrease by 0.18 per quarter (approximately onefifth of one speech). Or, for each 10 percent increase in presidential job approval, secondary foreign affairs speeches decrease roughly two per quarter (this finding will be addressed in the concluding chapter). We accept the hypothesis that when presidential job approval increases, (secondary) foreign affairs speeches decrease (H12). Lastly, we regressed presidential travel abroad on presidential job approval. From Table 7 we see that presidential job approval is significant and that the sign is in the correct direction. The Durbin-Watson statistic (2.20) is also in the acceptable range. For each 1 percent increase in presidential job approval, presidential travel abroad decreases by 0.06 days (less than two hours) per quarter. We accept the hypothesis that when presidential job approval increases, presidential travel abroad decreases (H13). We will address the significance of this finding later. The next phase of our analysis concerns the two competing hypotheses. First, presidents may choose to address a poor economy in their public speeches rather than avoid it. We now regress our three categories of economic affairs speeches on presidential job approval.

PAGE 68

61 Table 7. Regression of Presidential Travel Abroad On Presidential Job Approval MODEL I Variable b. f s.e. 1 Intercept 6.839 (2.095) Job Approval -0.063 * (0.038) R-square Durbin-Watson N 107 0.02 2.20 * P <= .05, one-tail From Model I, Table 8 we see that major economic affairs speeches is significant, but the sign is positive, which is in the wrong direction. Lagging one and two quarters did not reverse the sign. When secondary economic affairs speeches was regressed on presidential job approval (Model II), the variable was significant and the sign was in the correct direction, but autocorrelation was present ( Durbin-Watson= 0.99). Lagging one and two quarters improved the Durbin-Watson statistic, but our levels of significance increased to 0.07 in both regressions . Model III regresses total economic affairs speeches on presidential job approval. The sign is in the correct direction, but presidential job approval did not reach our

PAGE 69

62 significance level (p-value= 0.08) and lagging one and two quarters increased the p-values to 0.10 in both regressions. Our hypothesis (H14), therefore, is not accepted; economic affairs speeches do not increase when presidential job approval decreases. Table 8. Regression of Economic Affairs Speeches On Presidential Job Approval Variable MODEL I b, f s.e. 1 MODEL II b . fs . e . I MODEL III b^Xs. e . ) Intercept 0.130 (0.489) 21.461 (5.738) 21.591 (5.783) Job Approval 0.020 ** (0.009) -0.169 * (0.105) -0.149 (0.106) R-square Durbin-Watson N 0.04 1.96 107 0.02 0.99 107 0.01 0.98 107 * P <= .05, one-tail ** p <= ,01. one-tail Our second competing hypothesis concerns whether either foreign affairs speeches (three categories) or presidential travel abroad, when regressed on presidential job approval, increase when war is present. We added the variable war to the right side of our previous equations. We first address the regressions of three categories of foreign affairs speeches on presidential job approval and war. Rather than present the reader with multiple tables, suffice it to say that the variable war did not become sig-

PAGE 70

63 nificant in any of our regressions. Even when lagged one and two quarters, the p-values for war did not go below 0.20. We therefore reject the hypothesis that when war is present, the number of foreign affairs speeches increases (H15). We had similar results when we regressed presidential travel abroad on presidential job approval and war. The p-value for war was not significant (0.08) and lagging was unsuccessful (p-value= 0.07 at one lag and 0.08 at two lags). We reject the hypothesis that when war is present, presidential travel abroad increases (H16), and we conclude this section by reporting that none of our competing hypotheses were accepted. We now address our two remaining areas of research interest. First, do either foreign affairs speeches (three categories) or presidential travel abroad increase in a presidential election year? We begin with foreign affairs speeches. When major foreign affairs speeches was regressed on presidential election year, presidential election year was not significant (p-value= 0.23). Lagging increased the p-value to 0.27 at one lag and to 0.28 at two lags. We then regressed secondary (Model I) and total (Model II) foreign affairs speeches on the variable presidential election year (Table 9). The parameter estimates and p-values for presidential election year are nearly identical in both models and the signs are in the correct direction.

PAGE 71

64 Because the results are similar, and because total speeches combines two categories, we select secondary foreign affairs speeches for the final model. Table 9. Regression of Foreign Affairs Speeches On Presidential Election Year Variable Intercept Election YR MODEL I b. (s.e. ') 8.500 (0.874) 8.714 ** (1.734) MODEL II b . f s . e . 1 10.878 (0.911) 9.014 ** (1.806) MODEL III b. fs.e. 1 8.284 (1.144) 9.542 ** (2.014) R-sguare 0.18 Durbin-Watson 1.38 N 110 0.18 0.26 1.33 2.05 110 109 * P <= .05, one-tail ** P < .001, one-tail Secondary foreign affairs speeches (Model I) must be corrected for autocorrelation (Durbin-Watson= 1.38). Model III regresses secondary foreign affairs speeches on presidential election year with a lag of one guarter. The variable election year remained significant, the sign is in the correct direction, and the Durbin-Watson statistic (2.05) is in the correct range. From the data presented in Model III, we now construct the following eguation: Secondary Foreign Affairs Speeches = a + bl( Election YR) = 8.28 + 9.54(1) = 17.82

PAGE 72

65 From our equation we calculate that in a presidential election year there are approximately 18 secondary foreign affairs speeches given by the president per quarter; in nonpresidential election years the number is approximately 8 per quarter [8.28 + 9.54(0)]. The increase in secondary foreign affairs speeches as a result of a presidential election year being present is roughly ten per quarter. Because we have a lag of one quarter, the effect is demonstrable only in the second, third, and fourth quarters of the election year. We therefore accept the hypothesis that in a presidential election year, the number of (secondary) foreign affairs speeches increases (H17). Second, we regressed the variable presidential travel abroad on presidential election year. The p-values for presidential election year did not go below 0.14 even when lagged one and two quarters and the sign in all three regressions was negative, which was in the wrong direction. We reject the hypothesis that in a presidential election year, presidential travel abroad increases (H18). Our last area of inquiry concerns major foreign policy initiatives. We desire to know whether major foreign policy initiatives increase when the economy is performing poorly or when a presidential election year is present. First, we regressed the variable major foreign policy initiatives on our nine economic variables. From Model I,

PAGE 73

66 Table 10, we see that only the p-values for unemployment and the percent rate of change in per capita GNP are significant; however, using the entire model for explanatory purposes would be inelegant. Regardless, autocorrelation is present (Durbin-Watson= 1.71). The p-value for per capita real disposable income is nearly significant (0.06) and this variable is highly correlated with per capita GNP ( r= 0.99). Per capita real disposable income may become significant when regressed with unemployment and the percent rate of change in per capita GNP (see Model II). Model III regresses major foreign policy initiatives on just unemployment and the percent rate of change in per capita GNP; the latter are the two variables from Model I which are significant. From Table 10 we see that none of the variables in Models II and III reached our significance level. Lagging was not fruitful, though the Durbin-Watson statistics did improve. We conclude that the number of major foreign policy initiatives does not change as a result of declining economic conditions, and we do not reject our null hypotheses (H19-H27) . Lastly, we regressed major foreign policy initiatives on the variable presidential election year. Presidential election year was not significant (p-value= 0.20) and lagging caused our p-values to slightly increase. We reject the

PAGE 74

67 Table 10. Regression of Manor Foreign Policy Initiatives On Nine Economic Indicators Variable MODEL I b . f s . e . MODEL II b. fs.e. I MODEL III b.s.e. 1 Intercept -2.374 (2.081) 0.673 (0.444) 0.406 (0.286) Unemployment 0.202 * (0.106) 0.060 (0.051) 0.038 (0.044) Inflation -0.055 (0.178) Prime Rate -0.031 (0.038) Income -0.0009 (0.0005) -0.00003 (0.00004) GNP 0.0007 (0.0004) Defense 0.0001 (0.0005) Change Inc 0.138 (0.094) Change GNP -0.214 ** (0.096) -0.078 (0.072) -0.073 (0.072) Change Def -0.006 (0.030) R-sguare Durbin-Watson N 0.07 1.71 109 0.02 1.67 109 0.01 1.66 109 * P <= .05, ** P <= .01, one-tail one-tail

PAGE 75

68 hypothesis that major foreign policy initiatives increase in a presidential election year (H28). This concludes the regression of our models. The following chapter will present the conclusions drawn from the foregoing analysis, discuss their implications, and present suggestions for future research.

PAGE 76

CHAPTER V CONCLUSIONS The results from our regressions lead to a modification of our theory of presidential behavior (see Figure 2). We found that unemployment and inflation are important economic indicators for the public. When either unemployment or inflation increases, economic saliency increases. This saliency is recorded through a modification of the responses to the "most important problem" question asked in national opinion surveys. This new variable measures the attitude quite well, though the delay (lag) is one quarter. A short delay in the processing of economic information by the public seems reasonable, however, and is supported by previous research on the presidency (see review of literature and below) . Unemployment Increases > Economic Presidential Saliency > Job Approval > Inflation > Increases Decreases Increases A 1 Travel Abroad Increases Secondary Foreign Affairs Speeches Increase Figure 2 . Modified Behavioral Model 69

PAGE 77

70 When the public's level of economic saliency goes up, its collective attitude concerning the president goes down (with a delay of one guarter) . This attitude is measured by responses to the "presidential job approval" question frequently asked in national opinion surveys. This finding offers evidence to support the belief held by many politicians and political scientists that the public holds the president accountable for poor economic conditions (see, for example, Nixon 1962, Tufte 1978, and Norpoth 1984). In the present study economic saliency is an intervening variable between unemployment, inflation, and presidential job approval. Unemployment and inflation indirectly affect presidential job approval through economic saliency, and inflation also directly affects presidential job approval. These relationships are analogous to findings reported in previous research. With regards to the variable presidential job approval, both Monroe (1978) and Norpoth (1984) found lagged effects for inflation, and Fiorina (1981) confirmed the relationship (non-lagged) with unemployment. Both of these economic indicators (non-lagged) were reported by Frey and Schneider (1978) and Shapiro and Conforto (1980a) to significantly affect presidential job approval. And, when unemployment and inflation were combined into a "misery index" it, too, significantly affected presidential job approval (Ostrom and Simon 1985, Brace and Hinckley 1991).

PAGE 78

71 We subsequently discovered that when presidential job approval decreases, presidents will increase their time spent traveling abroad and increase the number of secondary foreign affairs speeches given (the latter with a delay of one quarter) , though the increases for both variables were quite small. The relationship between presidential job approval and presidential travel abroad supports the findings of Darcy and Richman (1988); however, our results are irrespective of political party. A judgment must be made regarding the importance of the latter two findings. If presidents consciously increase their attention to foreign affairs in their speeches and increase the number of days spent traveling abroad as methods for improving their standing in public opinion polls and their electoral chances, then we would expect to see increases which reasonable people would agree were substantial. It is difficult to conclude that an increase of 0.06 days of foreign travel per quarter and 0.18 secondary foreign affairs speeches per quarter, for each one point decrease in presidential job approval, would produce the desired effect.^ It seems unlikely that the public would notice such small changes in presidential behavior. This conclusion may support the research of Simon and Ostrom ^For the entire data set the mean number of days per quarter that the president spends traveling abroad is 3.37. Similarly, the mean number of secondary foreign affairs speeches is 10.71 per quarter.

PAGE 79

72 (1989). They reported that it is approval-enhancing events, not speeches or foreign travel, which positively affects presidential popularity. Contrary to our theory, and to the findings reported by Lammers (1981), we found no relationship between presidential press conferences and presidential job approval. Lammers had found that when presidential popularity decreased, presidential press conferences increased. None of our competing hypotheses were accepted. Economic affairs speeches did not increase when presidential job approval decreased, and neither foreign affairs speeches nor presidential travel abroad were affected by the presence of war. We conclude this section by stating that insufficient evidence exists to support our theory that presidents attempt to divert the nation's attention to foreign affairs when the economy is performing poorly. Regarding our remaining areas of research interest, we found that (secondary) foreign affairs speeches increased in presidential election years, though presidential travel abroad did not. Apparently, presidents are inclined to remain at home during a presidential election year to devote their energies to the campaign. Surprisingly, major foreign policy initiatives did not increase as a result of poor economic conditions nor did they increase in presidential election years. It may be that presidents believe that economic conditions are foremost in

PAGE 80

73 the minds of the American public and that attempts to increase their own popularity or electoral chances through foreign affairs speeches, foreign travel, or major foreign policy initiatives would be fruitless. We will now address the significance of the finding that secondary foreign affairs speeches increase by ten per quarter during a presidential election year. Some may argue that the increase is not due to greater attention to foreign affairs by presidents in presidential election years, but the result of a general increase in speechmaking while presidents are on the campaign trail. One method of testing this assertion is to regress our three categories of economic affairs speeches on the variable presidential election year and compare the results. Presidential election year was not significant when regressed with major economic affairs speeches (p-value= 0.09). Table 11 presents results from the regressions of secondary and total economic affairs speeches on presidential election year. From Model I (secondary economic affairs speeches) and Model II (total economic affairs speeches) we see that presidential election year is significant and the signs are in the correct direction. The results for both regressions are nearly identical and autocorrelation is present. In keeping with our previous decision-making processes, secondary economic affairs speeches is chosen for our final model.

PAGE 81

74 Model III presents the regression of secondary economic affairs speeches on presidential election year, but with a lag of one quarter. The variable is significant, the sign is in the correct direction, and autocorrelation has been corrected (Durbin-Watson= 2.21). Table 11. Regression of Economic Affairs Speeches On Presidential Election Year Variable Intercept Election YR MODEL I b. f s.e. 1 9.048 (1.277) 13.415 ** (2.531) MODEL II b , (s.e. ) 10.207 (1.281) 13.792 ** (2.539) MODEL III b . ( s . e . 1 8.665 (1.746) 14.960 ** (2.969) R-square 0.20 Durbin-Watson 1.31 N 110 0.21 1.28 110 0.30 2.21 109 * P <= .05, one-tail ** P < .001, one-tail From the data presented in Model III, Table 11, we create the following equation: Secondary Economic Affairs Speeches = a + bl( Election YR) = 8.66 + 14.96(1) = 23.62 We calculate that, in a presidential election year, there are approximately 24 secondary economic affairs

PAGE 82

75 speeches given by the president per quarter. In non-presidential election years the number of speeches is approximately nine per quarter [8.66 + 14.96(0)]. The increase in secondary economic affairs speeches as a result of a presidential election year being present is roughly 15 per quarter. Because the variable is lagged one quarter, the effect is demonstrable only in the second, third, and fourth quarters of the election year. It seems reasonable to infer that the increase in secondary foreign affairs speeches can be attributed to a general increase in speechmaking by presidents during the political campaign (presidential election year) . From Table 12 we see that presidents give greater attention to economic affairs in their speeches during the election year. This may be due to the importance voters place on economic conditions (especially unemployment and inflation) when they evaluate the president's job performance and when deciding whom to vote for in the general election. This additional piece of evidence supports our previous findings and speculations. Table 12. Comparison of the Presidents' Speeches In Non-presidential Election Years and Presidential Election Years Type of Non-presidential Presidential Speech Election Years Election Years Increase For. Affairs Econ. Affairs 8 per qtr. 9 per qtr. 18 per qtr. 24 per qtr. 10 (125%) 15 (166%)

PAGE 83

76 In the present study we have reported that unemployment and inflation matter greatly to the American public. These two economic variables affect economic saliency, which in turn affects presidential job approval. Previous research has shown that presidential job approval affects the president's electoral chances (Sigelman 1979, Lewis-Beck and Rice 1982) and his party's chances in congressional elections (Kernell 1977, Campbell 1985). We have shown that increases in presidential job approval are not associated with increases in foreign travel and foreign affairs speeches. Rather, presidential approval declined as the latter two variables increased. And, presidents do not increase the number of major foreign policy initiatives when the economy is performing poorly or during presidential election years. Regarding economics and foreign affairs, it is evident that presidents would be wise to devote greater attention to economic affairs since these are obviously the issues foremost in the minds of the public. The problem for the chief executive, however, is that the number of tools at his disposal for correcting a declining economy is quite small and in most cases he must rely on the cooperation of Congress or the bureaucracy or the Federal Reserve Board. Additionally, we found that the number of economic speeches given by the president does not affect presidential job approval. It may

PAGE 84

77 be that actual economic conditions matter greatly to the public, while rhetoric does not. From our analyses two topics for future research are suggested. First, we found that unemployment and inflation account for most of the variance (R-square= 0.87) in econnomic saliency and that economic saliency accounts for a substantial portion of the variance (R-square= 0.52) in presidential job approval. Researchers should focus on an explanation for the remaining variance in presidential job approval.^ Possible explanatory variables include personality or charisma, crises, scandals, and treatment by the media . Second, what can presidents do to improve their job approval ratings? Greater attention to foreign affairs and foreign travel is obviously not the answer because our results show that increases in foreign affairs speeches and foreign travel are associated with decreases in presidential job approval. We also reported that presidents do not attempt to initiate major foreign policy when the economy is slumping. Yet, even if presidents did increase the number of major foreign policy initiatives, the results from the regressions of foreign affairs speeches on presidential job approval suggest that an increase in the number of major ^When presidential job approval was regressed just on inflation, the R-square was 0.73.

PAGE 85

78 foreign policy initiatives would also be associated with a decrease in presidential job approval. The evidence from this study suggests that economic affairs are more important than foreign affairs to the American public. Specifically, presidents should attempt to keep the unemployment and inflation rates as low as possible. Although administrations may attempt to enlarge the incomes of individuals and families by increasing transfer payments (see Tufte 1978), success greatly depends upon cooperation from others within the political structure. More importantly, these transfer payments do not decrease unemployment or inflation. In fact, additional payments could fuel inflation. And, as previously mentioned, attempts to increase presidential job approval through policy successes in Congress may be unsuccessful (see Bond and Fleisher 1984, Zeidenstein 1985). If unemployment and inflation cannot be controlled, a president may have to wait for an economic slump to work itself out or for fate to present him with a potential rally event close to the election and hope for a positive response from the public. In either case, the president is at the mercy of uncontrollable forces. Political scientists should conduct research for the purpose of discovering which types of behavior, situations, or events will increase presidential popularity.

PAGE 86

79 This research project has presented the results of an investigation into nearly thirty years of presidential behavior; however, we have analyzed the actions of only six presidents — a rather small study group to generalize from. If institutions have memories and precedent is a guiding force, then we would expect that, at least for the near future, presidents will continue to behave like their predecessors in similar situations. Yet, we know that what appears to be politically conventional today is often tomorrow's discarded theory.

PAGE 87

APPENDIX A WAR^ EVENT START DATE END DATE (qtr/yr) (qtr/yr) Vietnam 4/1961 1/1973 Lebanon 3/1982 1/1984 Persian Gulf 3/1987 3/1988 Events whose dates overlapped those recorded were discarded. For example, the U.S. invasion of Grenada occurred in the fourth quarter of 1983, which overlaps the Lebanon conflict. Since the variable measures only the presence or absence of war, several military conflicts are not listed here. There is disagreement as to when the Vietnam war began and ended, but an explanation is in order for the dates chosen. U.S. military advisors had been in Vietnam throughout the 1950s. In December of 1961 two U.S. Army helicopter companies were deployed to South Vietnam, hence the reason ^Data for this appendix are from Congressional Research Service (1989, 1992) and Clodfelter (1992). 80

PAGE 88

81 why the fourth quarter of 1961 was chosen as the start date. Many scholars have elected August of 1964 as the start date because this was when the Gulf of Tonkin crisis occurred and when President Johnson responded with heavy bombing. The following data show why August 1964 may be inappropriate. Table 13. Vietnam War Statistics 1961 1962 1963 1964 U.S. troop strength in Vietnam 3,164 11,326 16,263 23,310 Killed 11 31 78 147 Air sorties — 2,334 6,929 5,362 (Source: Clodfelter 1992 , 1229) The first quarter of 1973 is chosen as the end date because "by March 29 the last of the 24,200 U.S. servicemen remaining in South Vietnam had been withdrawn" (Clodfelter 1992, 1276).

PAGE 89

APPENDIX B MAJOR FOREIGN POLICY INITIATIVES^ Kennedy 1-30-61 Reappraise entire defense structure; fight cold war. 4-20-61 Discusses Bay of Pigs events. 5-17-61 Strengthening of North Atlantic Treaty Organization (NATO). 5-25-61 New defense reguests for military assistance; new attempts at disarmament. 9-25-61 Plan for disarmament; end of nuclear testing; situations in Laos and West Berlin. 3-2-62 Resumption of nuclear testing. 4-18-62 Plan for disarmament. 10-22-62 Quarantine of Cuba. 7-26-63 Nuclear test-ban treaty, [total: 9] Johnson 8-4-64 Forces respond to attacks in Gulf of Tonkin. 4-7-65 Statement of Vietnam policy. ^Sources consulted: Brody (1984) and United States, Pres ident (1961-1989). 82

PAGE 90

83 5-2-65 U.S. forces sent to the Dominican Republic. 5-4-65 Request for $700 million for the defense of Vietnam and the Dominican Republic. 1-31-66 Resumption of bombing in Vietnam. 3-31-68 De-escalating the Vietnam conflict. 6-12-68 U.S. will sign nuclear non-proliferation treaty. 6-13-68 Consular convention signed by the U.S. and the U.S.S.R. 7-1-68 Signing of nuclear non-proliferation treaty. 10-31-68 Halts bombing of North Vietnam. [total: 10] Nixon 5-14-69 Outline of peace proposal for Vietnam. 11-24-69 Signing of non-proliferation of nuclear weapons treaty. 3-5-70 Ratification and entry into force of the nonproliferation of nuclear weapons treaty. 4-20-70 Withdrawal of 150,000 troops from Vietnam. 4-30-70 New military initiatives against the communists in Cambodia and Vietnam. 10-7-70 New peace proposal for Vietnam. 7-15-71 Announcement of diplomatic invitation to China. 1-25-72 Peace plan for Vietnam. 2-10-72 Itinerary for China trip. 5-8-72 Mining of North Vietnam's harbors. 5-19-72 Announcement of trip to the U.S.S.R.

PAGE 91

84 1-23-73 Peace agreement with North Vietnam. [total: 12] Ford 3-6-75 Assistance to Cambodia and Vietnam. 5-15-75 Concerning SS Mayaguez and rescue of her crew. 5-28-76 Treaty with the U.S.S.R. regarding underground nuclear explosions. [total : Carter 3] 8-12-77 Agreement in principle on Panama Canal treaties. 9-7-77 Signing of Panama Canal treaties. 9-8-77 United States-Canada agreement on the natural gas pipeline . 10-4-77 Pledges to use nuclear weapons only defensively; decrease the number of nuclear weapons . 3-10-78 Signing of nuclear non-proliferation agreement. 4-7-78 Deferment of the production of neutron weapons. 9-17-78 Camp David accords signed. 12-15-78 Diplomatic relations with China established. 3-26-79 Egyptian-Israeli peace treaty signed. 6-18-79 Signing of SALT II (Strategic Arms Limitation Talks) . 1-4-80 Sanctions against the U.S.S.R. for invading Afghanistan; delay of ratification of SALT II. 4-7-80 Sanctions against Iran. 4-25-80 Failed rescue of American hostages in Iran. 4-25-80

PAGE 92

85 9-17-80 New agreements with China [total: 14] Reagan 10-1-81 Sale of aircraft (AWACs and F15s) to Saudi Arabia. 5-9-82 Proposal for arms reductions. 8-20-82 Troops to deploy to Lebanon. 10-9-82 Sanctions against Poland. 11-22-82 Proceeding with the MX missile. 3-23-83 Strategic Defense Initiative (SDI) proposed. 3-30-83 Proposal to reduce nuclear forces. 10-4-83 New START (Strategic Arms Reduction Talks) initiatives . 10-25-83 Deployment of forces to Grenada. 4-4-84 Proposal to ban chemical weapons . 5-1-85 Economic sanctions against Nicaragua. 11-14-85 Concerning U.S.U.S.S.R. summit in Geneva. 1-7-86 Economic sanctions against Libya. 4-14-86 Airstrikes against Libya. 12-8-87 Signing of Intermediate-range Nuclear Forces (INF) treaty with the U.S.S.R. 3-11-88 Sanctions against Panama. 3-16-88 Troops to be deployed to Honduras. 4-18-88 Military confrontations with Iran. 6-4-88 Concerning the INF and START talks with the Soviet Union . [total: 19] Grand total: 67

PAGE 93

APPENDIX C VARIABLES YR Year QTR Quarter ELEC Presidential Election year JFK John F. Kennedy LBJ Lyndon B. Johnson RMN Richard M. Nixon GRF Gerald R. Ford JEC Jimmy E. Carter RWR Ronald W. Reagan PRESS Presidential Press Conferences TRAV Presidential Travel Abroad MFAS Major Foreign Affairs Speeches SFAS Secondary Foreign Affairs Speeches TFAS Total Foreign Affairs Speeches MEAS Major Economic Affairs Speeches SEAS Secondary Economic Affairs Speeches TEAS Total Economic Affairs Speeches MFPOL Major Foreign Policy Initiatives WAR War APPROV Presidential Job Approval 86

PAGE 94

87 ECSAL PRIME UNEMP INFL INC GNP DEF CINC CGNP CDEF Economic Saliency Prime Lending Rate Unemployment Rate Inflation Rate Per Capita Real Disposable Income Per Capita Gross National Product Per Capita National Defense Expenditures Percent Rate of Change in Per Capita Real Disposable Income Percent Rate of Change in Per Capita Gross National Product Percent Rate of Change in Per Capita National Defense Expenditures

PAGE 95

APPENDIX D CORRELATION MATRIX WITH NINE ECONOMIC VARIABLES (Pearson's r) PRIME PRIME 1.00 UNEMP 0.49 INFL 0.62 GNP 0.63 DEF -0.56 INC 0.64 CINC -0.23 CGNP -0 . 35 CDEF 0.21 INC PRIME 0.64 UNEMP 0 . 50 INFL 0.34 GNP 0.99 DEF -0.62 INC 1.00 CINC -0.12 CGNP -0.11 CDEF 0.16 UNEMP INFL 0.49 0.62 1.00 0.07 0.07 1.00 0 . 41 0.38 -0.60 -0.60 0.50 0.34 -0.10 -0.30 -0.07 -0.29 0.17 0.01 CINC CGNP -0.23 -0.35 -0.10 -0.07 -0.30 -0.29 -0.13 -0.09 0.12 0.12 -0.12 -0.11 1.00 0.43 0.43 1.00 -0.08 0.08 GNP DEF 0.63 -0.56 0.41 -0.60 0.38 -0.60 1.00 -0.59 -0.59 1.00 0.99 -0.62 -0.13 0.12 -0.09 0.12 0.16 0.02 CDEF 0.21 0.17 0.01 0.16 0.02 0.16 -0.08 0.08 1.00 88

PAGE 96

REFERENCES Abraiaowitz, Alan. 1985. "Economic Conditions, Presidential Popularity, and Voting Behavior in Midterm Congressional Elections." Journal of Politics 47:31-43. Bond, Jon R. , and Richard Fleisher. 1984. "Presidential Popularity and Congressional Voting: A Reexamination of Public Opinion As A Source of Influence In Congress." Western Political Quarterly 37:291-306. Brace, Paul, and Barbara Hinckley. 1991. "The Structure of Presidential Approval: Constraints within and across Presidencies." Journal of Politics 53 ( 4 ): 993-1017 . Brody, Richard. 1984. "International Crises: A Rallying Point for the President?" Public Opinion DecemberJanuary : 41 . Campbell, James E. 1985. "Explaining Presidential Losses in Midterm Congressional Elections." Journal of Politics 47:1140-1157. Citibase (data file). 1978. New York, NY: Citibank, N.A. Clodfelter, Micheal. 1992. Warfare and Armed Conflicts: A Statistical Reference to Casualty and Other Figures. 1618-1991 . Vol.II. Jefferson, NC: McFarland and Company , Inc . Congressional Research Service. 1989. Instances of Use of United States Armed Forces Abroad. 1798-1989 . Washington, DC: Library of Congress. . 1992. Use of Force by the United States: Case Studies. 1950-1991 . Washington, DC: Library of Congress . The Cumulated Indexes to the Public Papers of the Presidents of the United States, John F. Kennedy. 1961-1963 . 1977. Millwood, NY: KTO Press. The Cumulated Indexes to the Public Papers of the Presidents of the United States. Lvndon B. Johnson. 1963-1969 . 1978. Millwood, NY: KTO Press. 89

PAGE 97

90 The Cumulated Indexes to the Public Papers of the Presidents of the United States. Richard M. Nixon. 1969-1974 . 1978. Millwood, NY: KTO Press. The Cumulated Indexes to the Public Papers of the Presidents of the United States. Gerald R. Ford. 1974-1977 . 1980. Millwood, NY: Kraus International Publications. The Cumulated Indexes to the Public Papers of the Presidents of the United States. Jimmy Carter. 1977-1981 . 1983. Millwood, NY: Kraus International Publications. Darcy, Robert E. , and Alvin Richman. 1988. "Presidential Travel and Public Opinion." Presidential Studies Quarterly 18(1) :85-90. Data Disk (data file). 1984. Cambridge, MA: Cambridge Planning and Analytics. Data Resources, Inc. (data file). Lexington, MA: Data Resources, Inc. Edwards, George C. Ill, with Alec M. Gallup. 1990. Presidential Approval: A Sourcebook . Baltimore, MD: Johns Hopkins University Press. Erikson, Robert S. 1989. "Economic Conditions and the Presidential Vote." American Political Science Review 83 (2) :567-573. Fair, Ray C. 1978. "The Effect of Economic Events on Votes for President." The Review of Economics and Statistics 60(2) :159-173. . 1982. "The Effect of Economic Events on Votes for President: 1980 Results." The Review of Economics and Statistics 64:322-325. . 1988. "The Effect of Economic Events on Votes for President: 1984 Update." Political Behavior 10(2): 168179. Fiorina, Morris P. 1978. "Economic Retrospective Voting in American National Elections: A Micro-Analysis." American Journal of Political Science 22:426-443. . 1981. Retrospective Voting in American National Elections. New Haven, CT: Yale University Press. Ford, Gerald R. 1979. A Time to Heal: The Autobiography of Gerald R. Ford . New York, NY: Harper and Row.

PAGE 98

91 Frey, Bruno S., and Friedrich Schneider. 1978. "An Empirical Study of Politico-Economic Interaction in the United States." The Review of Economics and Statistics 60:174183. Gallup Organization (unpublished report). Americans Opinion On: Most Important Problem. 1935-1992 . Princeton, NJ : Gallup Organization. nibbing, John R. , and John R. Alford. 1981. "The Electoral Impact of Economic Conditions: Who Is Held Responsible?" American Journal of Political Science 25(3):423439. Kenney, Patrick J. , and Tom W. Rice. 1988. "The Contextual Determinants of Presidential Greatness." Presidential Studies Quarterly 18 ( 1 ): 161-169 . Kenski , Henry C. 1977. "The Impact of Economic Conditions on Presidential Popularity." Journal of Politics 39:764773. . 1980. "Economic Perception and Presidential Popularity: A Comment." Journal of Politics 42:68-75. Kernell, Samuel. 1977. "Presidential Popularity and Negative Voting: An Alternative Explanation of the Midterm Congressional Decline of the President's Party." American Political Science Review 71:44-66. . 1978. "Explaining Presidential Popularity." American Political Science Review 72:506-522. . 1986. Going Public: New Strategies of Presidential Leadership . Washington, DC: CQ Press. King, Gary, and Lyn Ragsdale. 1988. The Elusive Executive: Discovering Statistical Patterns in the Presidency . Washington, DC: CQ Press. Kramer, Gerald H. 1971. "Short-Term Fluctuations in U.S. Voting Behavior, 1896-1964." American Political Science Review 65:131-143. Lammers, William W. 1981. "Presidential Press-Conference Schedules: Who Hides and When?" Political Science Quarterly 96(2) :261-278. . 1982. "Presidential Attention-Focusing Activities." In The President and the Public , ed. Doris A. Graber. Philadelphia, PA: Institute for the Study of Human Issues.

PAGE 99

92 Lanoue, David J. 1988. From Camelot to the Teflon President; Economics and Presidential Popularity Since 1960 . New York, NY: Greenwood Press. Lee, Jong R. 1977. "Rallying Around the Flag: Foreign Policy Events and Presidential Popularity." Presidential Studies Quarterly 7 ( 4 ) : 252-256 . Lewis-Beck, Michael S., and Tom W. Rice. 1982. "Presidential Popularity and Presidential Vote." Public Opinion Quarterly 46:534-537. Markus, Gregory. 1988. "The Impact of Personal and National Economic Conditions on the Presidential Vote: A Pooled Cross-Sectional Analysis." American Journal of Political Science 32:137-154. McConnell, Campbell R. 1975. Economics: Principles. Problems. and Policies . New York, NY: McGraw-Hill. Monroe, Kristen R. 1978. "Economic Influences on Presidential Popularity." Public Opinion Quarterly 42:360-369. Mueller, John E. 1973. War. Presidents, and Public Opinion . New York, NY: Wiley. Neustadt, Richard E. 1986. Presidential Power: The Politics of Leadership From FDR to Carter . New York, NY: Macmillan . Newsweek . 1992. "The War Room Drill." November-December : 78 . Nixon, Richard M. 1962. Six Crises . Garden City, NY: Doubleday. Norpoth, Helmut. 1984. "Economics, Politics, and the Cycle of Presidential Popularity." Political Behavior 6(3): 253-273. Norpoth, Helmut, and Thom Yantek. 1983. "Macroeconomic Conditions and Fluctuations of Presidential Popularity: The Question of Lagged Effects." American Journal of Political Science 27:785-807. Oppenheimer, Joe A. 1980. "Small Steps Forward for Political Economy." World Politics 33:121-151. Ostrom, Charles W. Jr., and Brian Job. 1986. "The President and the Political Use of Force." American Political Science Review 80(2) :541-566.

PAGE 100

93 Ostrom, Charles W. Jr., and Dennis M. Simon. 1985. "Promise and Performance: A Dynamic Model of Presidential Popularity." American Political Science Review 79:334-358. . 1989. "The Man in the Teflon Suit? The Environmental Connection, Political Drama, and Popular Support in the Reagan Presidency." Public Opinion Quarterly 53: 353-387. Ragsdale, Lyn. 1984. "The Politics of Presidential Speechmaking, 1949-1980." American Political Science Review 78:971-984. . 1987. "Presidential Speechmaking and the Public Audience: Individual Presidents and Group Attitudes." Journal of Politics 49:704-736. Rivers, Douglas, and Nancy L. Rose. 1985. "Passing the President's Program: Public Opinion and Presidential Influence in Congress." American Journal of Political Science 29:183-196. Rohde, David W. , and Dennis M. Simon. 1985. "Presidential Vetoes and Congressional Response: A Study of Institutional Response." American Journal of Political Science 29:397-427. Schlenker, Barry R. 1980. Impression Management . Monterey, CA: Brooks/Cole Publishing Co. Shapiro, Robert Y., and Bruce M. Conforto. 1980a. "Presidential Performance, the Economy, and the Public's Evaluation of Economic Conditions." Journal of Politics 42:49-67. . 1980b. "Economic Perception and Political Behavior: Reply to Professor Kenski . " Journal of Politics 42:76-81. Sigelman, Lee. 1979. "Presidential Popularity and Presidential Elections." Public Opinion Quarterly 43:532534. 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. Smith, Carolyn. 1990. Presidential Press Conferences: A Critical Approach . New York, NY: Praeger Publishers.

PAGE 101

94 Staff, New York Times (unpublished). Survey Data, 1980-88 . Staff, Reagan Library (unpublished). 1990. Press Conferences of Ronald Reaaan . Stimson, James A. 1976. "Public Support for American Presidents: A Cyclical Model." Public Opinion Quarterly 40: 1 21 . Tufte, Edward R. 1978. Political Control of the Economy . Princeton, NJ: Princeton University Press. United States, President. 1961-1989. Public Papers of the Presidents of the United States . Washington, DC: Office of the Federal Register. U.S. Department of Commerce, Bureau of the Census. 1979. Estimates of the Population of the U.S. and Components of Change: 1940 to 1978 . Washington, DC: U.S. Government Printing Office. U.S. Department of Commerce, Economics and Statistics Administration, Bureau of Economic Analysis. 1992. National Income and Product Accounts of the United States. Volume 2. 1959-88 . Washington, DC: U.S. Government Printing Office. U.S. Department of State, Office of the Historian, Bureau of Public Affairs. 1990. Visits Abroad of the Presidents of the United States. 1906-1989 . Washington, DC: Department of State. Witcover, Jules. 1977. Marathon: The Pursuit of the Presidency 1972-1976 . New York, NY: Viking Press. Wonnacott, Ronald J., and Thomas H. Wonnacott. 1979. Econometrics . Second ed. New York, NY: John Wiley and Sons. Zeidenstein, Harvey. 1985. "President's Popularity and Their Wins and Losses On Major Issues in Congress: Does One Have Greater Influence Over the Other?" Presidential Studies Quarterly 15:287-300.

PAGE 102

BIOGRAPHICAL SKETCH Joseph K. Smaha was born and raised in Portland, Maine. He earned a B.A. in political science from the University of Southern Maine in 1979, an M.A in political science from Boston College in 1983, and a Ph.D. in political science from the University of Florida in 1994. Mr. Smaha 's master's thesis is entitled Education and the New Federalism and is a study of the effects of the Reagan administration's education policies on the public school system of Portland, Maine. Mr. Smaha was a student senator at the University of Southern Maine from 1978-1979 and served as president of the Political Science Graduate Student Association at the University of Florida from 1992-1993. He is a member of the American Political Science Association, and the Center for the Study of the Presidency. Mr. Smaha is married to the former Katherine M. Bavelaar of Milo, Maine. They have two children, Sara Michelle and Christopher James. His parents are the late Joseph K. and Teresa A. Smaha and he has one sibling, Terri A. Manship. 95

PAGE 103

I certify that I have read this study and that in my opinion it conforms to acceptable standards of p?esin?atior and is fully adequate, in a dissertation for the degree of Doctor of Philosophy. M. Margaret Conway ^ M. Margaret Conway Professor of Political Science I certify that I have read this study and that in my opinion it conforms to acceptable standards as uresentation and is fully adequate, in scope and quality, TdiSe^tation for the degree of Doctor of Philosophy. I ir sL 4 ^ Wayne IJ. Francis Professor of Political Science I certify that I have read this study and that in my opinion it conforms to acceptable standards oresentation and is fully adequate, in scope and quality, rdIISe?ta?Ln for the degree of ^ctor of . Philosophy . Walter A. Rosenbaum Professor of Political Science I certify that I have read this study and that in my opinion it conforms to acceptable standards of J p?eLntation and is fully adequate, in a dissertation for the degree of Doctor of Philosophy. Leslie P. Thiele _ Assistant Professor of Political Science I certify that I have read this study and that in my opinion it conforms to acceptable standards David R. Colburn Professor of History

PAGE 104

This dissertation was submitted to the Graduate Faculty of the Department of Political Science in the College of Liberal Arts and Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August 1994 Dean, Graduate School