Wagging the Doves? Peace-Promoting Actions as a Source of Presidential Support David T Burbach MIT Security Studies Program * Prepared for the Annual Meeting of the American Political Science Association August 28, 2003 Philadelphia, PA ABSTRACT Dramatic uses of force are widely thought to increase public support for national leaders, but less attention has been given to the possibility of “uses of peace” as sources of rallies. Media and public opinion theories suggest that, as visible and popular events, diplomatic activities should also be status-enhancing. This study tests that proposition by measuring the impact of “peace- promoting actions” on the approval ratings of U.S. Presidents from 1953-2000. A dataset of 85 dramatic peace events is constructed, and statistical tests used to measure the magnitude and duration of approval changes following them. Peace events cause an average 2% increase, and just over 5% for the 25 largest events, decaying with a half life of 3 months. Media quantity and elite opinion strongly influence the response, with gains >10% or losses of nearly 5% being possible. Despite these political gains, there is no evidence of “diversionary peace”: peace events are not more frequent in response to low approval ratings, recessions, scandals, or impending elections. !"#" $%&$ ’()*++ ,, - .$ - +/(01 ’$23 $1 4$$ 565 #4# " 6# " 7’ . "5 . 5$# 1 56 5 $ !$ " "#" $1
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
������������ ������������ �����
Wagging the Doves?
Peace-Promoting Actions as a Source of Presidential Support
David T Burbach
MIT Security Studies Program*
Prepared for the Annual Meeting of the American Political Science Association
August 28, 2003
Philadelphia, PA
ABSTRACT Dramatic uses of force are widely thought to increase public support for national leaders, but less attention has been given to the possibility of “uses of peace” as sources of rallies. Media and public opinion theories suggest that, as visible and popular events, diplomatic activities should also be status-enhancing. This study tests that proposition by measuring the impact of “peace-promoting actions” on the approval ratings of U.S. Presidents from 1953-2000. A dataset of 85 dramatic peace events is constructed, and statistical tests used to measure the magnitude and duration of approval changes following them. Peace events cause an average 2% increase, and just over 5% for the 25 largest events, decaying with a half life of 3 months. Media quantity and elite opinion strongly influence the response, with gains >10% or losses of nearly 5% being possible. Despite these political gains, there is no evidence of “diversionary peace”: peace events are not more frequent in response to low approval ratings, recessions, scandals, or impending elections.
H1 “Peace-Promoting” activity by presidents leads to increases in presidential approval ratings (i.e., they create “rallies” as uses of force often do)
H2
Peace activities will cause greater increases in presidential approval when A) media coverage is more intense
B) the presidential action is successful
C) the president’s actions receive support from Congress and
opinion leaders in the media
H3
Peace activities should be more frequent during periods of presidential political need, such as:
a) low approval ratings,
b) recessions,
c) scandals
d) impending elections.
Hypothesis 1: “Peace-Promoting” activity by presidents leads to increases in presidential approval ratings (i.e., they create “rallies” as uses of force often do)
Hypothesis 1 simply summarizes the argument of the previous section: since they meet
the criteria for producing “rallies”, we should observe increases in presidential approval
following dramatic peace-promoting activities by presidents.
Hypothesis 2a: Peace activities will cause greater increases in presidential
approval when media coverage is more intense Hypothesis 2b: Peace activities will cause greater increases in presidential
approval when successful Hypothesis 2c: Peace activities will cause greater increases in presidential approval
when supported by opinion leaders Hypothesis 2 derives from the “media-priming” model, identifying conditions that will
produce larger or smaller approval increases for presidents. More media coverage should lead to
������������ ������������ �����/
higher salience for the peace activity or foreign policy generally, and so should lead to larger
increases in approval. On the other hand, drawing attention to a disaster will not be helpful:
successful initiatives should produce more favorable results for presidents. Citizens can also be
expected to follow the lead of experts, such as members of Congress or media commentators, in
evaluating presidential actions.
Hypothesis 3: Peace activities should be more frequent during periods of presidential
political need, such as low approval ratings, recessions, scandals, or impending elections
If peace activity can boost presidential ratings, then we would expect presidents to avail
themselves of the opportunity to improve their poll numbers during times of political misfortune.
This prediction does not come directly from the media priming model, but instead applies the
logic of the diversionary war theory to peacemaking: politically-motivated presidents will use
the tools available to them to improve their standing and reelection prospects, and thus we should
see an increase in the frequency of peace events when presidents face domestic difficulties.
Diversionary war studies have come to mixed conclusions about the existence of diversionary
motivations behind U.S. uses of force, so it would be worth extending the analysis to see if the
timing of other foreign policy actions by presidents appears to be politically inspired.
Methods and Sources of Data
The effect of peace events on presidential approval was measured using the same general
strategy that has been employed by studies of the “rally effect” that supposedly follows uses of
������������ ������������ �����(
force: a set of relevant events was identified, then statistical tests were used to measure the
changes in presidential approval ratings following those events.
Identifying Peace Events
Although there are a number of existing datasets of uses of force or conflicts more
generally, that is not the case for peace initiatives. Earlier presidential approval studies that
included cooperative actions did so on an ad hoc basis (Mueller 1973;Kernell 1978;MacKuen
1983), or did not provide a list of their events at all (Marra and others 1990;Brace and Hinckley
1993). More general datasets of conflict and cooperation such as COPDAB (Azar ) and WEIS
(McClelland 1999)are also inadequate, due to their limited years of coverage (neither extends
past 1978), their inclusion of many small, low-visibility events, and the fact that they focus on
actual international actions, rather than political initiatives (which would include proposals that
do not ultimately bear fruit). For similar reasons, using treaties and agreements would not be
satisfactory, since in may cases initiation or changes of position in talks are the more relevant
political event than the actual signing of an agreement.
Therefore, a new dataset was constructed of presidential events that were visible, dramatic,
and “peace promoting”. These criteria were operationalized as follows:
• Visibility: events had to receive five or more New York Times front-page stories in a 14 day period.
• Dramatic: events had to be:7
- A specific, significant accomplishment or action, such as completing a treaty, cessation of bombing, etc
Figure 1 -- Peace Events by Year To illustrate the sort of events that were chosen and the approval changes that followed
them, Table 2 lists representative events and their associated change in presidential approval
ratings (measured from the last poll prior to first poll after). Some significant gains have been
seen; Richard Nixon’s numbers went up 16 points when the Vietnam War finally ended. On the
other hand, presidents have not always benefited from diplomatic activity: George Bush’s
diplomatic forays in 1991 and 1992 did nothing to help the steady erosion of his poll numbers,
and Eisenhower actually lost more than 10% when the Korean War ended.
������������ ������������ �����*
TABLE 2: Approval Changes Following Selected Events
EVENT Date Approval Change
End of Vietnam War 1/73 +16%
Initiatives to improve relations with Soviet Union 5/89 +14%
“Atoms for Peace” proposal 12/53 +10
Johnson attends Manila Conference, offers Vietnam peace plan 10/66 +5
Reagan reopens arms talks with USSR 11/81 +5
Nixon’s visit to China 2/72 +4%
Ford attends Helsinki CSCE Summit 7/75 -8%
Bush attends Middle East peace conference in Madrid 11/91 -10%
To show the full spectrum of immediate changes, figure 2 presents a histogram showing
the distribution of approval changes over the whole set of peace events. These measures are
from the last poll prior to first poll after, and so are not strictly comparable – sometimes days
have elapsed since an event, sometimes weeks to months. The statistical analysis to follow will
take those timing issues and other factors into account, but figure 2 does suggest that peace
events have helped presidents more often than not, but on the average have been followed by
only minor immediate increases in approval.
������������ ������������ �����,
-1 0 -5 0 5 10 15 2 0 2 5
P o st-E ven t C h an g e in Ap p ro val (% )
0
2
4
6
8
1 0
Num
ber
of E
vent
s
Figure 2 – Distribution of Approval Changes Following Peace Events Variables and Coding
Each event was coded on the following variables:
PEACE_EVENT
1 for each event
DATE
The date that the principal U.S. activity was reported in the New York Times.
COVERAGE
An ordinal variable representing coverage in the New York Times:
1 5 to 15 front-page stories in 14 days
2 15 to 25 front-page stories in 14 days
3 More than 25 front-page stories in 14 days
������������ ������������ �����)
PEACE_MAJOR
1 for each event that met the following criteria: a medium or high level of media coverage (COVERAGE = 2 or 3), and either a major presidential statement (e.g., prime-time speech) or foreign trip associated with the event.
. SUCCESS
coded +1, 0, or –1 depending on whether the event (in its initial stages, at least), appeared to be successful for the U.S. The judgment was based on actual outcomes, not assessment of Presidential activity (i.e., if the dominant media message was that the President had done well but something bad still happened, that would be coded –1). For example, the 1960 Paris and the 1986 Reykjavik summits were coded as failures, while the 1972 Moscow and 1987 Washington summits were successes.
OPINION
Variable indicating the balance of reaction in editorials and columns to each event, as reported in the New York Times. If the majority of editorials and columns were favorable, +1. If the majority are negative, -1; if there is no clear majority or if no opinion is expressed, 0.
CONGRESS_OWN, CONGRESS_OPP (H2c) Congressional reaction by the president’s own party, and by the opposition. Coded +1 if the majority of reactions expressed by members of the party were supportive, -1 if the majority were negative, and 0 if mixed or none were recorded. The source for Congressional opinion is also the New York Times Index,10 but opinions were counted even if they were not reported on the front page.
CONGRESS_EXPECTED, CONGRESS_REVERSE (H2c) Opinions expressed in Congress should carry more weight when they are not in the expected direction. To capture this effect, these two variables were coded as follows CONGRESS_EXPECTED is the sum of: +1 if CONGRESS_OWN=+1 -1 if CONGRESS_OPP=-1 CONGRESS_REVERSE is the sum of: -1 if CONGRESS_OWN=-1 +1 if CONGRESS_OPP=+1
OPINION_INDEX Sum of OPINION, CONGRESS_EXPECTED, CONGRESS_REVERSE (i.e., a –3 to +3 scale).
Each event was assigned into one of these five categories. Events are coded 1 for their assigned category, 0 for the other two. WAR_TERMINATION events were the termination of major wars, which meant Korea, Vietnam, the Gulf War, and Kosovo. DE-ESCALATION included events that reduced the level of American involvement in an ongoing conflict, or constituted steps towards a negotiated end to the conflict, such as the initiation of peace talks. Most events in this category related to Vietnam, such as bombing pauses, troop withdrawal announcements, and progress in peace talks. ARMS_CONTROL events were summits where major arms control treaties or agreements took place, initiation of new talks, completion of treaties when a dramatic event in and of themselves (not taking place at a summit), or announcement of major proposals for arms control or other actions to make nuclear war less likely (de-targeting, “hotline”, etc). RELATIONS events signified improvements in relations with adversaries– meaning principally China and the USSR. Included summits with no specific arms control agreements, or policy changes by the U.S. that mark improved relations (e.g., recognition of China). MEDIATION events were cases of the U.S. engaging at the White House level to serve as intermediary in conflicts not involving the U.S. (principally, the Middle East)
������������ ������������ ����/+
Statistical Approach The effects of dramatic peace events on presidential approval were modeled as
exponentially decaying shocks superimposed on an underlying level of approval set by economic
conditions and other control variables. This approach was chosen for several reasons. First, the
duration of any gains from peace activities are important to a president’s political calculations: a
gain of 5% that lasts for months may be more significant than a gain of 10% that lasts for days.
Relatively few studies have estimated both the magnitude and duration of approval changes from
events,11 and so this is a useful methodological contribution. This approach also controls for
background trends in approval ratings, such as when peace event happens as the economy is
rapidly worsening or improving. Finally, the approach is less sensitive to sampling error or
errors in the dating of events than the commonly used method of using the change from the last
approval poll prior to first poll after an event (i.e., using only two polls per event).
The control model was similar to specifications used by previous studies of presidential
approval (Kernell 1978;Marra and others 1990;MacKuen; Erickson, and Stimson 1992;MacKuen
1983). The form chosen was a linear, autoregressive (i.e., AR1) specification (Erickson;
MacKuen, and Stimson 2002;Beck 1992). The principal control variables were economic
conditions, along with early term “honeymoons”, and the impact of war deaths and major
scandals. One slight difference is that individual polls, rather than monthly averages, were the
unit of analysis.12 This has little effect on the control model but is important to allow more
accurate modeling of rally dynamics. The specific variables and functional form were:
The share of the public indicating that they approve of the way the president is handling his job according to each Gallup survey. Expressed in percentage points – that is, 0 to 100, rather than fractions for ease of interpretation.13
UNEMPLOYMENT
The national unemployment rate for the month in which a given poll took place.
QUARTERLY_INFLATION
Three-month moving average of the annualized monthly change in the Consumer Price Index. A moving average is used because of the volatility of the series.14
CONSUMER_EXPECT
Index of the “Business Conditions Expected in the Next 12 Months” item from the University of Michigan Survey of Consumers. 15
INAUGURATION
Takes on 1 for the first poll after a new President takes office, zero otherwise
VIETNAM_SQRT
The square root of the cumulative casualties (in thousands) during the Vietnam War, from January 1965, to the end of Johnson’s term. From Mueller (1973).16
WATERGATE
Dummy variable for the Watergate scandal: 1 starting March of 1973, and 2 from the “Saturday Night Massacre” in October of 1973 until Nixon’s resignation in August 1974.
Dummy variable for the Iran-Contra scandal; 1 from November of 1986 until the end of August 1987 (when Congressional hearings concluded).
The model without any effect from events is specified in equation 1:
1-t1t
8
7654
210
Control*Approval:1
*
****
**:1
φ
θθθθθ
θθθ
+=
+++++
++=
t
t
tttt
ttt
Controlb
IranContra
WatergatetVietnamSqronInauguratipectConsumerEx
nflationQuarterlyIntUnemploymeControla
Where t represents each poll in the 1953-2000 sequence, and φ1 is the coefficient on the
autoregressive term. Also, the control coefficients are shown as � rather than � to distinguish
them from coefficients on event effects. For brevity, Controlt will be used in place of the
individual control variables and their coefficients in subsequent equations.
Event effects are added as exponentially decaying shocks, with an initial magnitude that
is a function of the independent variables. To allow for the possibility that “rally” effects decay
more rapidly than do approval changes from changes in economic conditions, a separate decay
constant φ2 is estimated. The model does not treat polling periods as equal for event effects.
Instead, event effects decay according to the actual time between polls – the model will predict
that virtually all of a shock should still be present if the first poll is just a day later, while much
of the effect will have diminished by the time of the first poll if weeks or months have elapsed.
The time between polls is termed Intervalt, measured in months of 30.5 days. The magnitude of
the disturbance at time t is thus equal to φ2Intervalt times the disturbance at time (t-1). Moreover,
the magnitude of the disturbance at the first poll after an event will not be equal to the
instantaneous effect, since polls sometimes come weeks after an event. The first observed
������������ ������������ ����/(
magnitude is thus a function of Delayt, where Delay is the time (in months) between an event
and the first poll following the event.
iDelayt 21-t
Interval2t *tEventEffec* tEventEffec:2 t φφ BX+=
Where Xt and B are vectors of independent variables and coefficients, and EventEffectt-1
is the residual effect of previous events. A final complication is the possibility that multiple
events will happen within one polling period – each with a different delay until the next poll. To
allow for this, the equation is re-specified, with the independent variables using events as the unit
of analysis and i representing a given event, and during a polling period t the new disturbance is
the sum of the disturbance from all events i that take place between poll (t-1) and poll t:
���
����
�+= �
=
it
t
DelayLastEvent
FirstEventi21-t
Interval2t *tEventEffec* tEventEffec:3 t φφ BX i
Finally, this disturbance is added into the approval model. Since it seems unlikely that
one president’ s “ rally” would carry over to the next, the term (1-Inaugurationt) is added, and
used to null out prior event effects at any polling period when a new president takes office.
εφβββ
φθθθθθ
θθθφ
++++
−++
+++++++
=
�=
it
t
DelayLastEvent
FirstEventiini
tt
tttt
tt
VariableNVariable1
onInauguratiIranContra
WatergatetVietnamSqronInauguratipectConsumerEx
nflationQuarterlyIntUnemployme
210
1-tInterval
28
7654
2101-t1
t
*)*...*(
)1(*tEventEffec* *
****
**Control*
Approval:4
t
������������ ������������ ����/8
Or, using Controlt to signify the control variables, and setting EventMagnitude equal to the sum of the independent variables and coefficients:
( )
VariableNVariable1tudeEventMagni
where
tudeEventMagni
onInauguratiControl
nii
DelayLastEvent
FirstEventii
tt
it
t
*...*
*
)1(*tEventEffec* Control*
Approval:5
10
2
1-tInterval
21-t1
t
t
βββ
εφ
φφ
++=
++
−++
=
�=
In the form of eq 5, the only terms that change with different independent variables are
the components of EventMagnitude. Therefore, for brevity when describing specific regressions
only the form of EventMagnitude will be given, but in each case that term is a component of
equation 5, which is the actual equation being regressed. This form can not be estimated using
the standard AR(1) extensions to OLS, but it behaves much like a standard linear auto-regressive
model and can be readily estimated with maximum likelihood methods .17
Analysis (I): Presidential Gains from Peace Events General Effects of Peace Events To test Hypothesis 1, that peace events are generally followed by increases in
presidential approval, three regressions were run. In each case, the full model was as specified in
equation 5; only the EventMagnitude term changed from one regression to another. For the
initial attempts to measure the average effect of peace events, event magnitudes were specified
95% confidence intervals for effect predictions. Variables in bold significant at a .05 level or above. Coefficients on the control variables show the expected signs, are mostly significant, and
are generally consistent with the findings of earlier presidential approval studies.
The results show that peace events are followed, on average, by increases in presidential
approval ratings. A positive and statistically significant effect is found in all three regressions.
Without controlling for uses of force, an average 4% increase is seen – but that effect decays
������������ ������������ ����/,
rapidly; only a 1% increase would remain after a month. With a control for uses of force, only a
2% increase is expected, though with a half life of three months (about the same half life as has
been found for uses of force alone). Though that 2% result is statistically significant, it is not of
much political significance.
More dramatic effects are seen when only the 25 largest events are included in the
regression. As results for equation 6b show, major peace events can be expected to provide
presidents a 5.4% boost (+/- 3.4%) – nearly as large as the 6.2 (+/- 3.2%) gain from major uses
of force. In either case, the “ rally” is expected to decay with a half life of three months. Figure 2
illustrates the expected decay patterns.
0 2 4 6������
-10
-5
0
5
10
����� ���� ���
����������
����������������������
Figure 2: Predicted Peace Event Effects
0 2 4 6������
-10
-5
0
5
10
����� ���� ���
�����������
�����������������
0 2 4 6������
-10
-5
0
5
10
����� ���� ���
� ������������
������������������
Figure 3 – Predicted Effects of Peace Events Dramatic “ uses of peace” apparently affect presidential ratings in a manner similar to that
of uses of force. The size of these rallies is modest: the 2% gain from an average peace event
would hardly be noticeable given the other influences on presidential ratings. A sudden 5% gain,
however, is politically meaningful, but even so the effect is small and short-lived compared to
major negative factors such as a recession or major scandal. From the coefficients in table 3, for
������������ ������������ ����/)
example, the Iran-Contra scandal had a long term impact of about –13% on Reagan’ s approval
ratings.
Media/Opinion Variables and Peace Events
As stated in Hypothesis 2, the media priming model predicts that the public reaction to
events will be conditioned by the coverage, success, and elite reaction of those events. To test
this, regressions were run with the media and opinion variables as shown in equations 7a and 7b:
ii
iiii
iii
iiii
ForceMajorexOpinionInd
SuccessCoveragePeaceEventtudeEventMagnib
ForceMajorposeCongressOpnCongressOwComment
SuccessCoveragePeaceEventtudeEventMagnia
**
***:7
****
***:7
54
321
7654
321
βββββ
βββββββ
++++=
++++++=
Unfortunately, the opinion variables showed classic signs of multicolinearity:
coefficients showing unexpected sign, high instability under specification changes, and
substantial correlation between each other. Therefore, regressions were also done with the
OpinionIndex variable, which is the sum of the editorial and Congressional variables, producing
an index that ranged from –3 to +3. This variable was much better behaved, as demonstrated in
the results in Table 4.
������������ ������������ ����/0
Table 4: Event Effects by Media/Opinion, Event Type
95% confidence intervals for effect predictions. Variables in bold significant at a .05 level or above. As predicted, the peace activity produces larger approval gains for presidents when
coverage is more intensive, and when supported by opinion leaders. Success, on the other hand,
did not matter much, though studies of the use of force have also found that.
������������ ������������ ����(+
The effect of coverage is particularly strong: if elite opinion is neutral, then an event just
meeting the 5-day threshold for New York Times coverage would have an expected effect of just
0.7% on the president’ s ratings, which a high coverage event (more than 25 stories) would have
an expected 6.0% immediate increase. Elite opinion can also make a substantial difference in the
public’ s reaction, adding or subtracting 5.1% with 100% support or 100% criticism. In practice,
unanimous support or opposition is rare; the average value for OpinionIndex is 0.5. There were
eight cases of full support (OpinionIndex = +3), and three of total criticism (OpinionIndex = -3).
The net result is that in the best possible case – high coverage, success, complete support
– a president’ s ratings would be expected to increase 11.5% initially, decaying by half every
seven weeks. In a worst-case scenario, unsuccessful, criticized peace activity could lower a
president’ s ratings by 5%. Figure 4 shows these predicted effects over time.
0 2 4 6������
-15
-10
-5
0
5
10
15
����� ���� ���
������ ��������� ��������� ��������� ���
Figure 2: Peace Effects by Media/Opinion Variables
0 2 4 6������
-15
-10
-5
0
5
10
15
����� ���� ���
������� ���������� ���������� ���������� ���
0 2 4 6������
-15
-10
-5
0
5
10
15
����� ���� ���
��� ���� ������ ���� ������ ���� ������ ���� ���
Figure 4 – Predicted Effects by Media/Opinion Variables
Effect of Different Types of Peace Activity
Further regressions were run with peace events disaggregated by type. The media
priming model predicts that presidential approval will show greater increases when presidential
������������ ������������ ����(�
actions are more in line with public preferences. Studies suggest this is true for uses of force:
intervention for internal change is less popular than other missions ((Jentleson and Britton
1998;Jentleson 1992)), and internal change events due lead to smaller increases (or even losses)
in approval ((Oneal and others 1996;Burbach 2003)).
Columns 2 and 3 of table 4 show results from regressions run by type of event, with all
peace events and major events only. The most striking result is the huge coefficient on War
Termination, with an instantaneous approval increase of 20 to 30% predicted. The end of wars
are important events, but such a large coefficient seems unreasonable, and there are reasons to
doubt the result. Most importantly, there are only four cases of war termination, and so the result
is based on very few data points. The four cases and their immediate approval changes were:
Korea –15, Vietnam +16, Gulf War +9, Kosovo 0%. Note that fairly rapid half-lives were found,
so that a 23% immediate prediction would typically mean a 15-20% by the time of the first
subsequent poll. The large WarTermination effect is probably driven by the extraordinary rally
of the Gulf War in 1991; Bush’s ratings went up more than 25%, far more than the 6% predicted
by the MajorForce variable. With the Gulf War making up 1/4th of the WarTermination variable,
and the end-of-war event happening so quickly after the start of the war, WarTermination picks
up much of the excess war-related rally. Looking at the individual data, with only four examples
about all we can say is that the ends of wars do not have a consistent effect, though they can
cause significant changes in presidential ratings for better or for worse.
Looking to other types of peace activity, there are some differences, though not as clearly
as with uses of force. Most activities provide modest gains on average. ARMS CONTROL
gives a consistent 2 or 3% immediate increase. DEESCALATION and IMPROVED
RELATIONS activities have little effect when events of all magnitudes are considered, but
������������ ������������ ����(/
highly prominent reductions in conflicts or improvements in relations are followed by gains of
5% or more.
MEDIATION actually hurts presidents; the coefficient was consistently negative in all
specifications that were tried. The negative effect of serving as a mediator for third parties is
surprising, given that presidents seem eager to play the role of peace maker. Looking at the data,
though, only in the case of Carter’s 1978 Camp David summit was mediation followed by
approval gains for a president. This would be consistent with a public focus on direct U.S.
interests in foreign policy – solving other nations’ conflicts may not be seen as relevant by many
Americans.
Summary: “ Rallies” From Peace Events The results generally confirm hypotheses 1 and 2: peace events do have positive effects
on presidential approval, and are affected as predicted by media coverage and elite opinion. The
gains from diplomacy are generally modest; even for events receiving heavy media coverage, the
expectation is for only a 5% initial gain lasting for a few months. Such approval gains would not
rescue a failing president, but on the other hand, they are not much less than the gains available
from the use of force. If the relatively small approval increases resulting from most uses of force
generate fears that presidents use military force to boost their poll numbers, then it does seem
plausible that the gains from diplomacy found here would also inspire manipulation. The next
section tests that proposition.
������������ ������������ ����((
Analysis (II): Political Timing of Peace Events
Hypothesis 3 predicted that presidents would engage in “ diversionary peace” Since
peace activity yields political benefits, there would appear to be opportunities for presidents to
exploit statesmanship for their own personal gain.
Methods and Variables
To test for political motivations in the timing of peace activities, the frequency of peace
activity was modeled as a function of politically relevant variables, and control variables, using a
Poisson event-count model for the number of peace events observed in any given polling
interval, an approach that has often been used in the diversionary war literature (Gowa
1998;Waterman and Meernik 1996;Fordham 1998a;Brace and Hinckley 1993).19 In addition to
the variables described earlier, these regressions required the following additional variables were
used:
USSR_TENSION
Coded 1 for periods of heightened tension, -1 for periods of unusual cooperation,
1/1/1992 1/29/2001 End of cold war (USSR disbanded) 0
WARTIME 1 during periods when the U.S. was at war. Specifically, 1953 up to the Korean
armistice, February 1965 to January 1973 (Vietnam), August 1990 to March 1991 (Gulf War), March to June 1999 (Kosovo).
ELECTION YEAR 1 for the 12 months prior to a Presidential election ELECTION PRESIDENT 1 for the final three months prior to a presidential election ELECTION MIDTERM 1 for the three months prior to a midterm election POST COLDWAR 1 for the period after December, 1991 (the breakup of the Soviet Union). EISENHOWER …. CLINTON Dummy variables for each president SCANDALS
������������ ������������ ����(9
1 during periods of major scandals, as identified from (Schultz 1999).20 3MONTH APPROVAL 3 month moving average of presidential approval, up to the poll prior to the polling
period in questions (approval immediately before each event is not used to avoid the possibility that approval ratings are affected by the run-up to the events).
The following equations were regressed, with and without individual presidential effects
included:
����
����
�
+++++++
+++
=
rPostColdWarmElectMidte
esidentElectarElectionYeScandals
WartimeTensionUSSRMoApproval
pectConsumerExntUnemployme
IntervalCountPeaceEvent tt
**
Pr***
*_*3*
**
exp*:8
109
876
543
210
βββββ
ββββββ
����
����
�
++++++++
+++
=
ClintonJFKrmElectMidte
esidentElectarElectionYeScandals
WartimeTensionUSSRMoApproval
pectConsumerExntUnemployme
IntervalCountPeaceEvent tt
*...**
Pr***
*_*3*
**
exp*:9
18109
876
543
210
ββββββ
ββββββ
Note that Intervalt is used as a weighting factor due to the unequal polling periods.
EISENHOWER is not included in the presidential effects since a constant term is included (i.e., each presidential dummy represents their difference from Eisenhower), and POST COLDWAR is dropped when presidential effects are included since it is almost identical to CLINTON (POST COLDWAR would effectively be a dummy variable for the 1992).
Analysis of Results Regression results from equations 8 and 9 are shown below in table 6.
ElectionMidterm -0.1128 0.3927 -0.0880 0.4011 N 704 704
Log-likelihood -139.6 -137.3 Likelihood ratio
test (Pr(Y|�=0))
0.0009 0.0076
Variables in bold significant at a .05 level or above; Variables underlined at a 0.30 level. Dummy variable for Eisenhower omitted since constant included; read as if Eisenhower effect is zero.
The control variables for the current U.S. international situation have the expected
effects: peace events are more common during wars (when peace proposals and de-escalation
are possible), less common when U.S./Soviet tensions are high, and become significantly more
������������ ������������ ����(,
common after the end of the Cold War (though this could be a Clinton effect as much as a post-
Cold War effect).
Political variables generally have insignificant effects, and often in the direction opposite
that would be predicted for “ diversionary peace” . This can be seen more clearly in Figure 4,
below, which shows peace event frequency predicted for the unit changes (unless otherwise
specified) in variables, with all other variables held constant. The reference line shows the
predicted rate for Eisenhower, during a non-wartime, non-election period, with USSR TENSION
at 0, and approval and economic variables at their means.
E ise nho w erE lectY e ar
M idterm 3 m oP resid ent 3m o
S can dalW artim e
U S S RC lin ton
C arterR eces sio n
Ap rv -1 5%0 .0
0 .1
0 .2
0 .3
0 .4
0 .5
Pea
ce E
vent
s / M
onth
Figure 5 – First Differences On Peace Event Rates Presidential elections campaigns appear to depress diplomatic activity; the frequency
drops dramatically in the last months before an election, as does the rate of uses of force.
Presidents may fear charges of “ October surprises” , the White House may be too focused on the
������������ ������������ ����()
campaign to conduct major policy initiatives, and potential diplomatic partners may prefer to
wait until after elections when it is clear who their future negotiating partner will be.
Scandals, recessions, and low approval ratings have little effect on the frequency of peace
activity. Presidents are no more likely to conduct dramatic diplomacy when their poll numbers
are down than they are at other times. Presidents are also no more likely to do so during
recessions, which is an interesting finding: most authors report that the use of force does become
more common during recessions. It is not clear why no evidence of political timing is seen.
Presidents might simply refuse to let domestic political calculations affect their foreign policy
decisions. The White House may have a wide range of other political tools at its disposal that are
more effective than overtly diversionary uses of peace or force. It also seems likely that there is
strategic interaction involved: Presidents are well aware that potentially self-serving actions
during times of political need are widely criticized by the press and opposition politicians.
A variety of alternative specifications were tried using higher order terms for approval,
and interactions between approval and economic conditions.21 These specifications did not
significantly outperform the simple model shown in table 6, but in no case did the “ diversionary
peace” pattern appear. Instead, a majority of the models showed peace activity being more
frequent when presidential approval was already high, and for peace activity to become
particularly uncommon during the combination of low presidential approval and poor economic
conditions. In no case was there anything like the strong relationship that has been observed
between economic slowdowns and an increased use of force.
If the greater propensity to use force during recessions is due to diversionary motivations,
it seems odd that peace events do not also become more common, since they too can enhance
10/14/94 Iraq: U.S. Forces Deployed to Reinforce Kuwait ("Vigilant Warrior") DEFENSE 1 3
11/21/94 Bosnia: NATO Air Strikes (Bihac) INTERNAL 1 1.5
08/31/95 Bosnia: NATO Air Campaign ("Deliberate Force") INTERNAL 2 3
12/15/95 Bosnia: Deployment of IFOR Peacekeepers INTERNAL 2 0.5
09/03/96 Iraq: Missile Strikes after Kurdistan Incursion INTERNAL 2 -0.5
08/20/98 Sudan/Afghanistan Missile Strikes after Embassy Bombings PROTECT 2 -0.5
12/16/98 Iraq: Air Strikes After Inspectors Expelled ("Desert Fox") DEFENSE 2 -1
03/24/99 Kosovo War INTERNAL 3 0
06/12/99 Kosovo Peacekeeping (KFOR) INTERNAL 2 1
������������ ������������ ����8,
Bibliography
Adams, Sherman, 1961. Firsthand Report: The Story of the Eisenhower Administration. New York: Greenwood Press.
Azar, Edward E. computer data file (ICPSR 7767). Conflict and Peace Data Bank (COPDAB) 1949-1978. 1993 (3rd release): University of Maryland, Center for International Development and Conflict Management.
Baker, William D. and John R. Oneal, 2001. "Patriotism or Opinion Leadership: The Nature and Origins of the Rally 'Round the Flag Effect." Journal of Conflict Resolution 45, no. 5 (2001): 661-87.
Baum, Matthew A, 2002. "The Constituent Foundations of the Rally-Round-The-Flag Phenomenon." International Studies Quarterly 46 (2002): 263-98.
Beck, Nathaniel, 1992. "Comparing Dynamic Specifications: The Case of Presidential Approval."Political Analysis, ed James A Stimson. Ann Arbor: University of Michigan Press.
Beschloss, Michael, 1997 . Taking Charge: The Johnson White House Tapes, 1964-1965. New York: Simon and Schuster.
Michael Beschloss and Strobe Talbott, 1993. At the Highest Levels: The Inside Story of the End of the Cold War. Boston: Little, Brown and Co.
Brace, Paul and Barbara Hinckley, 1993. "Presidential Activities From Truman to Reagan: Timing and Impact." Journal of Politics 55, no. 2 (1993): 382-98.
Brody, Richard A., 1984 . "International Crises: a Rallying Point for the President?" Public Opinion 6 (1984): 41-43. ________, 1991. Assessing the President . Stanford, CA: Stanford University Press. McGeorge Bundy, 1988. Danger and Survival: Choices About the Bomb in the First Fifty Years. New York:
Vintage. Bundy, William, 1998. A Tangled Web: The Making of Foreign Policy in the Nixon Presidency. New York: Hill
and Wang. Burbach, David T. "Diversionary Temptations: Presidential Incentives and the Political Use of Force." Ph.D.
dissertation. Massachusetts Institute of Technology, 2003. Cameron, A. Colin and Pravin K. Trivendi, 1998. Regression Analysis of Count Data. Cambridge, England:
Cambridge University Press. Cannon, Lou, 1991. President Reagan: Role of a Lifetime. New York: Simon and Shuster. Center for Media and Public Affairs, 1998. "Clinton's China Syndrome." Media Monitor 12, no. 4 (1998). Clifford, Clark, 1991. Counsel To the President. New York: Basic Books. Coser, Lewis A, 1956. The Functions of Social Conflict. Glencoe, Ill: Free Press. Dassel, Kurt and Eric Reinhardt, 1999. "Domestic Strife and the Initiation of Violence at Home and Abroad."
American Journal of Political Science 43, no. 1 (1999): 56-85. Davies, Graeme, 2002. "Domestic Strife and the Initiation of International Conflicts, 1950-1982." Journal of
Conflict Resolution 46, no. 5 (2002): 672-92. DeRouen, Karl Jr, 2000. "Presidents and the Diversionary Use of Force: A Research Note." International Studies
Quarterly 44 (2000): 317-28. DeRouen, Karl R., 1995. "The Indirect Link: Politics, The Economy, and the Use of Force." Journal of Conflict
Resolution 39 (1995): 671-95. Edwards, George C., 1990 . Presidential Approval. Baltimore: Johns Hopkins University Press. Edwards, George C., William Mitchell, and Reed Welch, 1995. "Explaining Presidential Approval: The
Significance of Issue Salience." American Journal of Political Science 39 (1995): 108-34. Edwards, George C. III and Tami Swenson, 1997. "Who Rallies: The Anatomy of a Rally Event." Journal of
Politics 59, no. 1 (1997): 200-212. Enterline, Andrew J and Kristian Gleditsch, 2000. "Threats, Opportunity, and Force: Repression and Diversion of
Domestic Pressure, 1948-1982." International Interactions 26 (2000): 21-53. Erickson, Robert S, Michael B MacKuen, and James A Stimson, 2002. The Macro Polity. New York: Cambridge
University Press. Fitzgerald, Frances, 2000. Way Out There In the Blue: Reagan, Star Wars, and the End of the Cold War. New York:
Simon & Schuster. Fordham, Benjamin, 1998 . "Partisanship, Macroeconomic Policy, and U.S. Uses of Force." Journal of Conflict
Resolution 42, no. 4 (1998): 418-39. ________, 1998. "The Politics of Threat Perception and the Use of Force: A Political Economy Model of U.S.
Uses of Force." International Studies Quarterly 41 (1998): 255-82. ________, 2001. "Another Look at 'Parties, Voters, and the Use of Force Abroad'." Journal of Conflict Resolution
������������ ������������ ����8)
46 (2001): 572-96. Fordham, Benjamin O. and Christopher C. Sarver, 2001. "The Militarized International Disputes Dataset and
United States Uses of Force." International Studies Quarterly 45 (2001): 455-66. Garthoff, Raymond, 1985. Detente and Confrontation: American-Soviet Relations From Nixon to Reagan.
Washington, D.C.: Brookings Institution. Garthoff, Raymond, 1994. The Great Transition: American-Soviet Relations and the End of the Cold War.
Washington, D.C.: Brookings. Gelpi, Christopher, 1997. "Democratic Diversions: Government Structure and the Externalization of Domestic
Conflict." Journal of Conflict Resolution 41, no. 2 (1997): 255-82. Gilliam, Franklin D. and Shanto Iyengar, 2000. "Prime Suspects: The Influence of Television News on the
Viewing Public." American Journal of Political Science 44 (2000): 560-573. Goemans, H. E., 2000. "Fighting For Survival: The Fate of Leaders and the Duration of War." Journal of Conflict
Resolution 44 , no. 5 (2000): 555-79. Gowa, Joanne, 1998. "Politics at the Water's Edge: Parties, Voters and the Use of Force Abroad." International
Organization 52, no. 2 (1998): 307-24. Graham, Thomas W., 1989. American Public Opinion on NATO, Extended Deterrence, and the Use of Nuclear
Weapons. Lanham, MD: University Press of America. David Halberstam, 1972. The Best and the Brightest. New York: Random House. David Halberstam, 2001. War In a Time of Peace. New York: Scribner. H. R. Haldeman, 1978. The Ends of Power. New York: Times Books. Hersh, Seymour M., 1983. The Price of Power: Kissinger in the Nixon White House. New York: Summit Books. Hinckley, Ronald H., 1992. People, Polls, and Policymakers: American Public Opinion and National Security.
New York: Lexington Books. Hughes, Emmet John, 1962. The Ordeal of Power: A Political Memoir of the Eisenhower Years. New York: H.
Wolff. Hugick, Larry and Alec Gallup, 1991. "Rally Events and Presidential Approval." Gallup Poll Monthly (1991): 15-23. Iyengar, Shanto, 1991. Is Anyone Responsible? How Television Frames Political Issues. Chicago: University of
Chicago. Iyengar, Shanto and Donald Kinder, 1987. News That Matters. Chicago: University of Chicago Press. James, Patrick and Anathasios Hristoulas, 1994. "Domestic Policy and Foreign Policy: Evaluating a Model of
Crisis Activity for the United States." Journal of Politics 56 (1994): 327-48. James, Patrick and Jean-Sebastian Rioux, 1998. "International Crises and Linkage Politics: The Experiences of the
United States, 1953-1994." Political Research Quarterly 51 (1998): 781-812. Jentleson, Bruce W., 1992. "The Pretty Prudent Public: Post Post-Vietnam American Opinion on the Use of
Military Force." International Studies Quarterly 36 (1992): 49-74. Jentleson, Bruce W and Rebeca L Britton, 1998. "Still Pretty Prudent." Journal of Conflict Resolution 42 (1998):
395-418. Kant, Immanuel, 1949. "Perpetual Peace."The Philosophy of Kant: Moral and Political Writings Carl Friedrich.
New York: Random House. Stanley Karnow, 1983. Vietnam: A History. New York: Viking. Kernell, Samuel, 1978. "Explaining Presidential Popularity." American Political Science Review 72 (1978): 506-22. King, Gary, 1988. "Statistical Models for Political Science Event Counts: Bias in Conventional Procedures and
Evidence for the Exponential Poisson Regression Model." American Journal of Political Science 32, no. 3 (1988): 838-63.
Kissinger, Henry, 1979. White House Years. New York: Little, Brown. Knopf, Jeffrey W., 1998 . "Domestic Sources of Preferences for Arms Cooperation: The Impact of Protest."
Journal of Peace Research 35 (1998): 677-95. Krosnick, John A. and Donald R Kinder, 1990. "Altering the Foundations of Support for the President Through
Priming." American Political Science Review 84 (1990): 497-512. Lee, J. R., 1977. "Rallying 'Round the Flag: Foreign Policy Events and Presidential Popularity." Presidential
Studies Quarterly 7 (1977): 252-56. Levy, Jack S., 1989. "The Diversionary Theory of War: a Critique."Handbook of War Studies, ed Manus I.
Midlarsky, 259-88. Ann Arbor: University of Michigan press. Lewy, Gunter, 1978. America in Vietnam. Oxford: Oxford University Press. Lian, Bradley and John R. Oneal, 1993. "Presidents, the Use of Force, and Public Opinion." Journal of Conflict
Resolution 37 (1993): 277-300. Lowi, Theodore, 1979. The End of Liberalism. New York: W.W. Norton.
������������ ������������ ����80
MacKuen, Michael, 1983. "Political Drama, Economic Conditions, and the Dynamics of Presidential Popularity." American Political Science Review 27 (1983): 165-92.
MacKuen, Michael, Robert Erickson, and James A Stimson, 1992. "Peasants or Bankers? The American Electorate and the U.S. Economy." American Political Science Review 86 (1992): 597-611.
Marra, Robin, Charles Ostrom, and Dennis Simon, 1990. "Foreign Policy and Presidential Popularity." Journal of Conflict Resolution 34 (1990): 588-623.
McClelland, Charles. data file, ICPSR 5211. World Event/Interaction Series (WEIS), 1966-1978. University of Southern California, 1999.
McCombs, Maxwell and D. Shaw, 1972. " The Agenda-Setting Function of the Mass Media 36 (1972): 176-85. Meernik, James, 2001. "Domestic Politics and the Political Use of Military Force by the United States." Political
Research Quarterly 54, no. 4 (2001): 889-904. Miller, Joanne and Jon A. Krosnick, 2000. "News Media Impact on the Ingredients of Presidential Evaluations:
Politically Knowledgable Citizens Are Guided By a Trusted Source." American Journal of Political Science 44, no. 2 (2000): 295-309.
Mueller, John, 1994. Policy and Opinion In The Gulf War. Chicago: University of Chicago Press. Mueller, John E., 1973. War, Presidents, and Public Opinion. New York: Wiley. Nixon, Richard, 1978. RN: The Memoirs of Richard Nixon. New York: Grosset and Dunlap. Don Oberdorfer, 1991. The Turn. New York: Touchstone. Oneal, J. R., B. Lian, and J. H. Joyner, 1996. "Are the American People 'Pretty Prudent'? Public Responses to U.S.
Uses of Force, 1950-1988." International Studies Quarterly 40 (1996): 261-80. Oneal, John R and Anna Lillian Bryan, 1995. "The Rally Round the Flag Effect in U.S. Foreign Policy Crises."
Political Behavior 17 (1995): 379-401. Ostrom, Charles W. and Dennis M. Simon, 1985. "Promise and Performance: A Dynamic Model of Presidential
Popularity." American Political Science Review 79 (1985): 334-58. Page, Benjamin and Robert Shapiro, 1992. The Rational Public: Fifty Years of Trends in American's Policy
Preferences. Chicago: University of Chicago. Pan, Zhongdang and Gerald Kosicki, 1997. "Priming and Media Impact on the Evaluations of the President's
Performance." Communications Research 24, no. 1 (1997): 3-30. Popkin, Samuel, 1994. The Reasoning Voter: Communication and Persuasion in Presidential Campaigns. Chicago:
University of Chicago. Prins, Brandon, 2001. "Domestic Politics and Interstate Disputes: Examining U.S. MID Involvement and
Reciprocation, 1870-1992." International Interactions 26, no. 4 (2001): 411-38. Rogers, Everett M., William B. Hart, and James W. Dearing, 1997. "A Paradigmatic History of Agenda-Setting
Research." Do the Media Govern?, eds Shanto Iyengar and Richard Reeves. Thousand Oaks, CA: Sage Publications.
Sagan, Scott D. and Jeremi Suri, 2003. "The Madman Nuclear Alert: Secrecy, Signaling, and Safety in October 1969." International Security 27, no. 4 (2003): 150-183.
Arthur Schlessinger, 1993. The Almanac of American History . Greenwich, CT: Brompton Books. Schultz, Jeffrey D., 1999. Presidential Scandals. Washington, DC: Congressional Quarterly. Seaborg, Glenn, 1987. Stemming the Tide: Arms Control in the Johnson Years. Lexington, MA: Lexington Books. Seaborg, Glenn T, 1981. Kennedy, Khruschev, and the Test Ban.Benjamin S Loeb. Berkeley, CA: University of
California. Simmel, Georg, 1955. Conflict. trans. Kurt Wolff. Glencoe, Ill: Free Press. Small, Melvin, 1988. Johnson, Nixon, and the Doves. New Brunswick, NJ: Rutgers University Press. Stein, Art, 1976. "Conflict and Cohesion: A Review of the Literature." Journal of Conflict Resolution 20, no. 1
(1976): 143-72. Talbott, Strobe, 1984. Deadly Gambits. New York: Random House. Talbott, Strobe, 2002. The Russia Hand. New York: Random House. Waterman, Peter and James Meernik, 1996. "The Myth of the Diversionary Use of Force by American Presidents."
Political Research Quarterly 49 (1996): 573-90. Wicker, Tom. "The Johnson Blitz." New York Times (New York), 2 November 1966, 36. Zaller, John R., 1992. The Nature and Origin of Mass Opinion. Cambridge: Cambridge University Press.