paying the human costs of war
DESCRIPTION
Discusses how public support of the war in Iraq has varied more with perceptions of the likelihood of strategic success than with the number of casualties.TRANSCRIPT
Paying the HumanCosts of War
Christopher GelpiPeter D. Feaver
Jason ReiflerDuke University &
Triangle Institute for Security Studies
Casualty Aversion as aConstraint on American Power
Mueller (1973, 1994): support for war drops with log of casualties
Developed into conventional wisdom that public will not tolerate casualties Beirut 1983, Gulf War 1991, Somalia 1993
Widespread assumption of “casualty phobia” Kosovo 1999
Caused Saddam to doubt US resolve in March 2003
Casualty Aversion:A Contingent Constraint?
Casualty phobia myth persisted despite evidence to the contrary Academic consensus on public cost/benefit
approach to support for war Key Question: What causes variation in
sensitivity to casualties? Jentleson: Policy Objective (PPO) Larson: Domestic Elite Consensus Kull: International Consensus Feaver & Gelpi: Likelihood of Success
Approval of Korean War and Battle Deaths
Key appears to be perceived success
Big drop in support is Aug-Dec 1950 – losing to PRC
Slope is positive from Feb-Aug 1951 - US recovers
Slope is negative in 1952 - stalemate sets in
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
Aug-50 Jan-51 Jun-51 Nov-51 Apr-52 Sep-520
10
20
30
40
50
60
70
Cumulative Hostile Deaths Percent Supporting
Approval of Vietnam Warand Battle Deaths
Impact of casualties seems to depend on success
Pre-Tet slope only significant because of drop Mar-May 1966 Matches increase in
those saying war is a “mistake”
Post-Tet slope triples majority of population
now says war is a “mistake”
0
10
20
30
40
50
60
70
Aug-65
Dec-6
5
Apr-66
Aug-66
Dec-6
6
Apr-67
Aug-67
Dec-6
7
Apr-68
Aug-68
Dec-6
8
Apr-69
Aug-69
Dec-6
9
Apr-70
Aug-70
Dec-7
0
Apr-71
05000100001500020000250003000035000400004500050000
Percent Supporting Cumulative Hostile Deaths
Presidential Approval andCasualties in Iraq
40
45
50
55
60
65
70
75
Mar
-03
Apr-03
May
-03
Jun-0
3
Jul-0
3
Aug-03
Sep-
03
Oct-0
3
Nov-0
3
Dec-0
3
Jan-
04
Feb-
04
Mar
-04
Apr-04
May
-04
Jun-0
4
Jul-0
4
Aug-04
Sep-
04
Oct-0
4
Nov-0
4
Ap
pro
val
0
100
200
300
400
500
600
700
Casu
alt
ies
Bush Approval Major Combat DeathsInsurgency Deaths Post-Sovereignty Deaths
Measures for Analysis of Weekly Approval and Deaths in Iraq
Presidential Approval - Weekly data Data from The Polling Report
Casualties – Log of deaths in Iraq Data from the US Military
Minutes of Media coverage of Iraq Data from The Tyndall Report
Growth in Dow Jones Industrial Index Dummy variables:
Onset of Insurgency Capture of Hussein Release of Kay Report Renewed Iraqi Sovereignty
Predicting Presidential Approval With Casualties In Iraq
40
45
50
55
60
65
70
75
Mar
-03
Apr-0
3
May
-03
Jun-0
3
Jul-0
3
Aug-03
Sep-0
3
Oct-0
3
Nov-0
3
Dec-0
3
Jan-0
4
Feb-
04
Mar
-04
Apr-0
4
May
-04
Jun-0
4
Jul-0
4
Aug-04
Sep-0
4
Ap
pro
va
l
Bush Approval Predicted Approval
Summary: Impact of Iraq War on Presidential Approval Impact of US Casualties:
During battlefield combat, approval went up with casualties
During insurgency, approval went down with casualties
After sovereignty casualties have no effect on approval
Turning points: End of May 2003 when “insurgency” became dominant
media frame Transfer of sovereignty to new government – an
indicator of success? Impact of Media Coverage
Rally effect before war & during major combat Coverage of democratic presidential candidates Media frame: battlefield, insurgency, and sovereignty
Comparing Sources of Casualty Tolerance in the Real World
Aggregate Data show casualty tolerance varies What drives this variation?
Experimental research shows support for all four mechanisms
“Success” seems an anecdotal fit, but… Now we need to:
Compare the relative importance of these mechanisms
Show they influence attitudes in a real conflict
A Model of Public Attitudes Toward Casualties In Iraq
Tolerance ForCasualties
US Right toAttack
US WillSucceed
Education
Demographics(Age, Race, Gender)
PartyIdentification
Four Clusters of AttitudesToward the War in Iraq Vietnam Syndrome: Iraq war wrong, we
are likely to lose Bush Base: Iraq war right, we are likely to
win Noble Failure: Iraq war right, we are likely
to lose Pottery Barn: Iraq war wrong, we are likely
to win
Logit Analysis ofTolerable Casualties In Iraq
Variable Coefficient T - Statistic
Right to Attack 0.98 (0.75)
Will Succeed 0.642 (4.80)***
Right X Succeed 0.268 (4.09)***
Female -0.294 (2.48)**
Minority -0.615 (3.68)***
Education 0.293 (4.65)***
Age 0.182 (3.17)***
Party ID -0.083 (2.26)**
N 1006
The Sources of Tolerance for Casualties in Iraq: Interactive Effects
Not At AllLikely
Not VeryLikely Somewhat
Likely Very Likely
Strongly Disapprove
Somewhat Disapprove
Somewhat Approve
Strongly Approve
0
10
20
30
40
50
60
70
80
90
100
Probability of Tolerating 1500 US Deaths
Expected Success
Rightness or Wrongness of War
The Demographic Sources of Casualty Tolerance in Iraq
ExplanatoryVariable
Change in Explanatory Variable
Δ Probability of Tolerating 1500 US Deaths
Age 18-29 to 60+ Years Old
+12%
Gender Male to Female -6%
Race White to Non-White -15%
Level of Education No H.S. to College Degree
+20%
Party Identification Strong Rep. to Strong Dem.
-11%
Reality Check:Our Survey & Election Outcome 2004
Our Survey Actual Vote
Bush 51.84% 51%
Kerry 48.16% 48%
Logit Analysis of Presidential Vote Choice: November 2004
Variable Coefficient T - Statistic
Right to Attack 0.599 (3.03)***
Will Succeed 0.027 (0.14)
Right X Succeed 0.417 (3.80)***
Female 0.291 (1.66)*
Minority -0.216 (0.93)
Education -0.102 (1.13)
Age 0.089 (1.04)
Party ID -0.793 (12.59)**
N 1006
Iraq and Presidential Vote Choice 2004
Not At AllLikely Not Very
LikelySomewhat
Likely Very Likely
Strongly Disapprove
Somewhat Disapprove
Somewhat Approve
Strongly Approve
0
10
20
30
40
50
60
70
80
90
100
Probability of Voting For Bush
Expected Success
Rightness or Wrongness of War
The Demographic Sources of Presidential Vote Choice in 2004
ExplanatoryVariable
Change in Explanatory Variable
Δ Probability of Voting for Bush
Age 18-29 to 60+ Years Old
+6%
Gender Male to Female -7%
Race White to Non-White -5%
Level of Education No H.S. to College Degree
-6%
Party Identification Strong Rep. to Strong Dem.
-81%
What Does It Mean to beSuccessful in Iraq?
25%
4%
28%
22%
4%
9%
8%
Iraqis ProvideSecurity
Iraqi EconomyRebuilt
Iraq Democratic
Iraqis Normal Lives
Iraqi Govt. No WMD
Iraqi Govt. No Linksto Terrorism
Iraq Not Threat toNeighbors
How Will We KnowWe Are Succeeding In Iraq?
19%
6%
14%
40%
8%
4%4%
4% 1%
Services Iraqis HaveAvailableHow Well Iraq's EconomyIs DoingHow Soon Elections AreHeldWhether Iraqis AreCooperating with the USAttacks Against USSoldiersUS Casualties
How Many Insurgents AreKilled/ArrestedHow Much Money the USSpendsRefused
Likelihood of “Success” and Whether the War in Iraq was the “Right Thing”
Right | US Will Succeed Thing | 1 2 3 4 | Total-----------+--------------------------------------------+---------- 1 | 84 155 49 8 | 296 | 80.77 51.16 11.89 2.72 | 26.59 -----------+--------------------------------------------+---------- 2 | 12 68 91 15 | 186 | 11.54 22.44 22.09 5.10 | 16.71 -----------+--------------------------------------------+---------- 3 | 3 63 167 60 | 293 | 2.88 20.79 40.53 20.41 | 26.33 -----------+--------------------------------------------+---------- 4 | 5 17 105 211 | 338 | 4.81 5.61 25.49 71.77 | 30.37 -----------+--------------------------------------------+---------- Total | 104 303 412 294 | 1,113 | 100.00 100.00 100.00 100.00 | 100.00
Pearson chi2(9) = 631.8776 Pr = 0.000
A Model of Public Attitudes Toward “Success” and “The Right Thing”
US Right toAttack
US WillSucceed
SupportPreemptive
force
BelieveSaddam
linksto WMD,terrorism
Care aboutInt’l
Consensuson force
Confidentin US &
IraqiLeaders
US hasplan tosucceed
Party ID,Demographics
Education,
Domestic elite consensusto stay in Iraq
Two-Stage Least Squares Analysis Whether US Was Right to Attack Iraq
Variable Coefficient T - Statistic
US Will Succeed 0.743 (10.34)**
Iraq Had WMD 0.114 (3.34)**
Link to Al-Qaeda 0.108 (3.60)**
Link to War on Terror
0.158 (4.34)**
Multilateralism -0.111 (3.02)**
Bush Doctrine 0.127 (2.87)**
Party ID -0.071 (4.50)**
N 964
Two-Stage Least Squares Analysis Whether US Will Succeed Iraq
Variable Coefficient T - Statistic
US Right to Attack 0.451 (8.34)***
Confident in US Leaders 0.274 (5.12)***
US Plan to Win the War 0.046 (1.99)**
Importance of Force 0.164 (3.64)***
US Elite Consensus 0.102 (2.44)**
Party ID -0.001 (0.10)
N 964
Conclusions Aggregate data show public willingness to
bear costs of war has varied significantly over time
Experimental evidence supports all hypothesized influences on costs and benefits Jentleson, Larson, Kull, Feaver & Gelpi
Data on attitudes toward Iraq war suggest that weighting of factors depends on the decision being made
“Do the right thing?” or“Just Win Baby?” Expectations of success is trumps for casualty
tolerance and support for ongoing mission Prospective judgment about future outcome Pottery Barn twice as likely as Noble Failure to support
casualties Rightness/wrongness is trumps for
determining vote choice Retrospective judgment about wisdom of a
leader’s decision Majority of Noble Failure support Bush but
overwhelming majority of Pottery Barn support Kerry
The Structure of Public Attitudes Toward Iraq Attitudes toward “Success” and “Right
Thing” structured along reasonable dimensions Consistent with hypotheses in literature
Attitudes are well organized, but are they immune to new information?
Attitudes cause one another, but stronger flow is “success” as a cause of “right thing”
Supplemental Slides
Sources of Data Historical aggregate data on support for
Korea, Vietnam, Somalia Recent aggregate Presidential Approval
data during current in Iraq Individual level data on attitudes toward
current war in Iraq Series of surveys from October 2004 through
October 2004 Flaw in literature: mostly aggregate data
Experimental Support for the Impact of Primary Policy Objective
Do you support “the United States taking military action to replace the government of Yemen…” If it were threatening the shipping of oil
through the Persian Gulf? 47% approve (FPR Mission)
If it were engaging in ethnic cleansing and forced slavery?
61% approve (HI mission) If it were providing terrorist bases to Al-
Qaeda? 71% approve (WT Mission)
Framing Effects and Support for a Hypothetical Invasion of Yemen
Combat Global Terror 0.256 (2.40)*
Promote Human Rights 0.176 (1.85)
Prevent Hostile Superpower 0.154 (1.59)
Human Rights x Frame 0.437 (2.82)**
Terrorism x Frame 0.598 (3.47)**
US Power x Frame 0.504 (3.31)**
Human Rights Frame -0.944 (2.40)*
Terrorism Frame -1.403 (3.70)**
Expected Casualties -0.120 (1.92)
Expected Success 0.773 (8.64)**
Experimental Support for theImpact of Domestic and Int’l Consensus
Do you support “military action to defend the democratic government of East Timor against an insurrection …”? If Congress, UN, and NATO endorse?
74% approve
If Congress opposes? 48% approve
If UN and NATO oppose? 41% approve
If Congress, UN, and NATO all oppose? 24% approve
Elite Consensus and Support for Hypothetical Use of Force in E. Timor
Coefficient T-statistic
All ElitesSupport Force
3.165 (48.56)**
Domestic Elites Oppose Force
-0.414 (4.41)**
Int’l Elites Oppose Force
-0.551 (6.06)**
Both Sets of Elites Oppose
-1.105 (12.15)**
Expected Success, Casualties, and Support for Hypothetical Use of Force
Probability Respondent Strongly Approves of Using Force
Expect 50 Deaths
Expect 500 Deaths
Expect 5,000 Deaths
JCS Extremely Confident
.26 .20 .15
JCS SomewhatConfident
.19 .13 .09
JCS Not Very Confident
.12 .08 .05
Casualty Tolerance andConfidence in Success
1 vs. 2 1 vs. 3 1 vs. 4
JCS Confidence 0.184 (2.50)* 0.667 (8.28)** 0.719 (8.97)**
Expected Deaths 0.109 (1.51) 0.490 (6.19)** 0.654 (8.28)**
Casualty Tolerance -0.361 (7.41)** -0.619(10.77)** -0.933(15.19)**
Ready for Force 0.138 (1.01) 0.537 (3.78)** 0.977 (7.04)**
Party ID 0.120(2.75)** 0.281 (5.93)** 0.405 (8.53)**
Age 0.000 (0.10) -0.002 (0.50) 0.006 (1.59)
IR Knowledge 0.001 (0.02) 0.084 (1.03) -0.062 (0.76)
Sex -0.134 (1.09) -0.233 (1.73) -0.093 (0.69)
Constant 0.448 (0.94) -2.030 (3.83)** -3.182(5.97)**
Observations 2973 2973 2973
Approval of Vietnam Warand Battle Deaths
Impact of casualties seems to depend on success
Pre-Tet slope only significant because of drop Mar-May 1966 Matches increase in
those saying war is a “mistake”
Post-Tet slope triples majority of population
now says war is a “mistake”
0
10
20
30
40
50
60
70
Aug-65
Feb-6
6
Aug-66
Feb-6
7
Aug-67
Feb-6
8
Aug-68
Feb-6
9
Aug-69
Feb-7
0
Aug-70
Feb-7
1
05000100001500020000250003000035000400004500050000
Percent Supporting Cumulative Hostile Deaths
Prais-Winsten Analysis of Battle Deaths and Support for the Vietnam War
Variable Model 1 Model 2
Log of Battle Deaths
-3.667 (2.79)* -3.956 (5.74)**
Log of Battle Deaths X Tet
-8.270 (2.85)**
Tet Offensive 77.486 (2.59)*
Constant 76.541 (6.17)** 84.969 (14.62)**
R-squared / N 0.71 / 24 0.92 / 24
Approval of Korean War and Battle Deaths
Key appears to be perceived success
Big drop in support is Aug-Dec 1950 – losing to PRC
Slope is positive from Feb-Aug 1951 - US recovers
Slope is negative in 1952 - stalemate sets in
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
Aug-50 Jan-51 Jun-51 Nov-51 Apr-52 Sep-52
0
10
20
30
40
50
60
70
Cumulative Hostile Deaths Percent Supporting
Capturing Saddamand Perceptions of Success in Iraq
MSNBC/WSJ Poll fielded our “success” question
Got a “natural experiment” with the capture of Saddam
Bush received 8% boost in “very likely” to succeed
Casualty tolerance should rise if this perception persists 0%
20%
40%
60%
80%
100%
12/13 12/14
Very Likely
SomewhatLikely
Not Very Likely
Not at AllLikely
DK / Refused