🎖️

Framing War

How Partisan Identity & Elite Cues Shape

Public Support for U.S. Military Interventions

Survey Experiment · N = 2,079 Political Science · IR Research

Roadmap

Research Overview

📌

Research Question

How do partisan identity & elite cues causally shape public support for U.S. military interventions?

🔬

Methodology

5-condition vignette survey experiment with ordinal & multinomial logistic regression analysis

⚖️

Core Tension

Security vs. humanitarian justifications — which frame wins public support?

📊

Key Finding

Humanitarian frames — especially from co-partisan elites — consistently outperform security-based rationales

Theory

Theoretical Foundations

Holsti (1992)

"Constraints Model"

Public opinion as a significant constraint on foreign policy — especially in prolonged engagements. Mass discontent forced strategy shifts in Vietnam, Iraq & Afghanistan.

Jentleson (1992)

"Pretty Prudent Public"

Support hinges on intervention objective: (1) Humanitarian, (2) Security-driven, (3) Regime-change. Security frames consistently elicit the highest approval (Jentleson & Britton, 1998).

Gelpi (2010)

Elite Cues & Surprises

Surprising events significantly impact opinion more than partisan cues alone. Top-down opinion formation where partisan narratives shape perceptions of military efficacy and moral justification.

Grieco et al. (2011): UN endorsement increases support among otherwise skeptical publics, highlighting legitimacy's role in shaping intervention attitudes.

Literature

Partisan Identity as Cognitive Filter

Republicans

Favor interventionist policies when framed as national security necessities. Higher casualty tolerance when governing party endorses the mission. Stronger response to co-partisan humanitarian cues (OR = 1.42).

Democrats

Show more hesitation unless interventions are framed around multilateralism or humanitarian imperatives. Lower opposition partisan casualty tolerance. More sensitive to civilian casualties.

Druckman (2001)

Unified partisan elite stance → stable public opinion. Conflicting elite narratives → highly polarized mass attitudes. Iraq War bipartisan erosion as proof.

Media Fragmentation Effect

Digital media intensifies polarization — ideologically congruent news ecosystems deepen entrenchment. Corrective information effectiveness depends on elite credibility (Silverman et al., 2020).

Berinsky (2009)

Security-based justifications leverage fear-based cognitive biases — individuals more likely to endorse force when perceiving direct threat. Post-9/11 effect pronounced.

Context

Historical Trends in Intervention Support

Cold War Era — Vietnam Syndrome

Anti-communist sentiment drove initial support, but eroded as costs escalated. Created lasting public skepticism toward engagements lacking clear strategic benefits (Holsti, 2004).

Late Cold War — Grenada (1983) & Panama (1989)

Brief duration + low casualties = sustained approval. Reagan's "Peace Through Strength" narrative effectively framed security justifications. 73% public approval for Grenada operation.

Post-Cold War — Somalia & Bosnia (1992–1999)

"Black Hawk Down" incident sharply eroded support for open-ended humanitarian operations. Bosnia (1995) & Kosovo (1999) sustained support only with clear objectives and minimal U.S. casualties.

Post-9/11 — Iraq & Afghanistan (2001–2021)

Fear amplification drove 80%+ initial approval. Support collapsed by 2006 as Iraq descended into sectarian violence. Afghanistan 2021 withdrawal enjoyed broad bipartisan support (~65%) despite humanitarian costs.

Methods

Experimental Research Design

Control

No partisan frame, no justification

DemHu

Democrat + Humanitarian frame

DemNatSec

Democrat + National Security frame

RepHu

Republican + Humanitarian frame

RepNatSec

Republican + National Security frame

📋 Dependent Variables

Outcome 1: "How likely are you to support U.S. military intervention in Country A?" — 5-point Likert scale

Outcome 2: Primary concern driving stance: Civilian Casualties | Long-term Entanglement | Regional Stability | International Norms

📐 Statistical Methods

Ordinal Logistic Regression — primary analysis for Outcome 1; interaction terms isolate partisan moderation effects

Multinomial Logistic Regression — models concern prioritization (Outcome 2) across treatment × PID conditions

Key Statistics

Study at a Glance

0

Respondents

0

Experimental Conditions

42%

Higher Odds (RepHu vs Control)

56%

Dem boost under RepHu frame

Likelihood Ratio Test

χ² = 28.64

df = 9, p < 0.001 — interaction model significantly better fit

RepHu Additive Model Coeff.

β = 0.350***

OR = 1.4185 — strongest & most significant single predictor

PID Effect

β = −0.112**

OR = 0.894 — Democrats/Independents marginally less supportive overall

Results — Figure 2 Analog

Odds Ratios by Treatment Group

Likelihood of selecting higher Likert response vs. Control (OR > 1.0 = more supportive)

RepHu ★

OR = 1.42 ***

DemHu

OR = 1.21

RepNatSec

OR = 1.04

DemNatSec

OR = 1.008

OR = 1.0 (Baseline)

Key insight: Humanitarian frames (DemHu & RepHu) consistently exceed OR 1.0 and achieve statistical significance. National security frames hover near baseline — contradicting the classic "Pretty Prudent Public" hypothesis.

Results — Figure 3 Analog

Partisan × Frame Interaction Effects

🔵 Democratic Respondents

DemHu frameOR > 1.0 ✓ (moderate)

RepHu frameOR > 1.0 (wide CI)

NatSec framesNear 1.0 (NS)

Dems respond more favorably to DemHu (+33% OR for independents baseline). RepHu still meaningful but weaker CI.

🔴 Republican Respondents

RepHu frameHighest OR ✓✓

DemHu frameBelow 1.0 ↓ attenuated

NatSec framesNear 1.0 (NS)

Republicans strongly respond to co-partisan humanitarian cues (OR up to 1.56 in interaction model). In-party source × moral appeal = peak persuasion.

Interaction Model (χ² = 28.64, p < .001): Partisan alignment magnifies humanitarian frames' persuasive power. "Both parties respond to humanitarian appeals, but most powerfully from their own side."

Results — Outcome 2

What Concerns Drive Opposition?

Multinomial Regression Findings

RepHu → Long-term Entanglement ↓↓

Coeff: −0.727***, OR = 0.484. Republican humanitarian framing reduces fears of prolonged engagement by ~52%. CI: [−1.064, −0.389]

DemNatSec → International Norms ↑

Coeff: 0.579**, OR = 1.784. Democrat security framing elevates concern for international agreements. CI: [0.061, 1.096]

DemHu → Regional Stability ↓

OR = 0.696. Democrats under humanitarian framing less concerned about regional consequences; focus shifts to civilian casualties instead.

Partisan Concern Patterns

Republican Respondents

Higher sensitivity to Long-term Entanglement in security conditions. Conservative skepticism toward protracted foreign interventions. But RepHu reduces this concern (OR = 0.484).

Democratic Respondents

More balanced concern distribution, but stronger reaction to Civilian Casualties in humanitarian conditions. Liberal constituencies prioritize humanitarian costs over strategic risks.

Independents

Intermediate pattern: align with Dems on Civilian Casualties, resemble Republicans on entanglement concerns. Most susceptible to framing shifts.

Preamble Questions

Pre-Existing Beliefs as Moderators

Four "forcegrid" items capturing baseline military attitudes — all significantly improve model fit when added as covariates.

Military strength ensures peace

OR = 1.33***

LR test: 66.67, p < .001. Strongest predictor. Military assertiveness correlates significantly with interventionist support. AIC = 6000.64

War is sometimes necessary

OR = 1.47***

LR test: 106.60, p < .001. Strongest overall model fit (AIC = 5960.70). "War necessity" belief is the single most potent attitude predictor.

Military skepticism

OR = 0.90**

LR test: 7.71, p = .006. Skepticism reduces support. Democrats with strong reservations: OR drops to 0.587. Republicans under NatSec: OR = 1.31 (polarizing).

Diplomacy over military

OR = 1.10*

LR test: 6.28, p = .012. Paradoxically small positive effect. Dems under NatSec framing: OR = 0.90 (diplomatic messaging moderates). Republican PID: OR = 1.16 (no mitigation).

Regression Output

Additive Model — Ordinal Logistic Regression

DV: Support for U.S. Military Intervention (5-pt Likert). N = 2,079

Predictor	Coefficient	Std. Error	Odds Ratio	Significance
DemHu	0.194	(0.125)	1.214	—
DemNatSec	0.008	(0.125)	1.008	—
RepHu	0.350	(0.126)	1.419	*** p < .01
RepNatSec	0.041	(0.124)	1.042	—
PID (partisan ID)	−0.112	(0.050)	0.894	** p < .05

Interpretation: Only the Republican/Humanitarian frame reaches conventional significance (t = 2.78). National security frames (both parties) show negligible effects near OR = 1.0. The negative PID coefficient (OR = 0.894) indicates Democrats/Independents are marginally less inclined toward higher Likert responses than Republicans overall.

Literature

Casualty Sensitivity & War Fatigue

The "Casualty Tolerance" Model

Gelpi, Feaver & Reifler (2010): Americans accept losses if they believe the mission will ultimately succeed. Explains why Iraq War support held 2003–2005 despite rising casualties, then collapsed in 2006 as progress narrative broke down.

War Fatigue Threshold

Kreps & Maxey (2017): War fatigue sets in without clear exit strategy. Afghanistan's 20-year mission "objective ambiguity" drove sustained public opposition — support declined from ~80% in 2001 to under 50% by 2021.

Civilian Casualty Effect

Sagan & Valentino (2017): Public support declines sharply when interventions cause significant foreign civilian casualties, even when national security justifications exist. Moral costs increasingly influence legitimacy calculations.

Contemporary Preference Shift

Growing preference for diplomacy, economic sanctions, and multilateral over unilateral action. More receptive to limited, precision-based operations (drone strikes, cybersecurity) than ground interventions. Rising China/Russia tensions may reverse this.

Theory

Cognitive Anchoring & Attitudinal Stability

Once formed, opinions on military force exhibit high stability — resisting shifts even against countervailing evidence (Berinsky, 2009). This rigidity intensifies in partisan environments.

🧠

Ideological Filtering

Strong militarists interpret interventions as security necessities; skeptics remain resistant regardless of evolving conditions (Zaller, 1992). Elite framing amplifies existing biases rather than reshaping core attitudes.

🎯

"War Optimism" Variable

Gelpi, Feaver & Reifler (2009): individuals who see military force as effective foreign policy tool consistently support intervention more — overrides short-term framing effects.

Implications for Framing

→

Framing effects are less effective when they contradict deeply held views — audiences selectively retain reinforcing information

→

Jentleson's "force grid" — baseline beliefs about force use fundamentally shape the interpretive lens before any elite cue is processed

→

Rathbun (2018): interventions aligned with pre-existing moral structures gain traction; attempts to shift fundamental attitudes meet substantial resistance

→

International institutional endorsement (UN) can modify preferences at the margins for otherwise skeptical audiences (Grieco et al., 2011)

Conclusions

What the Study Resolves

🕊️

Finding 1

Humanitarian > Security Frames

Contrary to Jentleson (1992), humanitarian justifications consistently outperform security frames. Effect is particularly pronounced for co-partisan Republican/Humanitarian cues (OR = 1.42, p < .01).

🔗

Finding 2

Partisan Alignment Amplifies

Elite cues and partisan identity work together — not independently. Interaction model fit (χ² = 28.64, p < .001) confirms partisanship moderates the persuasive power of each frame.

🔄

Finding 3

Framing Shifts Concern Focus

Beyond support levels, framing redirects the type of concern. RepHu reduces entanglement fears (OR = 0.484); DemNatSec elevates international norms attention (OR = 1.784).

Policy Implication: Policymakers seeking bipartisan support should frame interventions around moral humanitarian imperatives — especially through co-partisan elite messengers — rather than defaulting to security-based rhetoric. Pre-treatment military attitudes (war necessity, assertiveness) are the strongest predictors of ultimate support.

🎖️

Thank You

Framing War — Questions & Discussion

N = 2,079 Respondents 5 Experimental Conditions Ordinal + Multinomial Regression

Key takeaway: "Humanitarian frames — especially from co-partisan elites — consistently and significantly boost public support for military intervention, challenging the conventional wisdom that security rationales are most persuasive."