Framing Effects Experiment
- Introduced by
- Iyengar & Kinder (effects tradition); Chong & Druckman (synthesis)
- Year
- 1987
- Type
- Randomized experiment isolating the causal effect of message frames on attitudes
- Family
- Media-effects experiments
What does it do?
A framing effects experiment is a randomized design that isolates the causal impact of how a message is framed — which considerations it emphasizes — on people's attitudes, judgments, or behavior.
By randomly assigning participants to read otherwise comparable messages that differ only in their frame, it provides the causal counterpart to the descriptive framing analysis of media content.
The intuition behind the method
Content analysis can show that news frames immigration as an economic issue in one outlet and a security issue in another, but it cannot prove the frame changed anyone's mind. The experiment can. Hand one random group the economic frame and another the security frame, hold everything else constant, and any difference in their resulting opinions must be caused by the frame — that is the logic of randomization. The design distinguishes equivalence frames (logically identical information presented differently, like '90% survive' vs '10% die') from emphasis frames (stressing different real considerations), and it routinely asks who is most susceptible by testing moderators.
History
Experimental study of framing in communication grew from Iyengar and Kinder's 1987 News That Matters, which used controlled experiments to demonstrate agenda-setting and priming effects of television news, and from Kahneman and Tversky's work on equivalence framing in decision making. Communication and political scientists, synthesized in Chong and Druckman's 2007 framing-theory review, developed competitive-framing designs showing that effects depend on frame strength, repetition, source, and audience predispositions, moving the field from 'do frames matter?' to 'when and for whom?'
When to use it
Use a framing effects experiment when you need causal evidence that a particular way of presenting an issue changes attitudes or behavior, not merely that frames exist in media. It is the design of choice for testing message strategies in political communication, health and risk communication, and persuasion, and for identifying which audiences are most susceptible. It assumes successful random assignment, that the manipulated frame is the only systematic difference between conditions, and that outcomes are measured validly. It is less appropriate when you cannot ethically or practically manipulate the frame, when you care about real-world prevalence of frames (use content analysis), or when external validity is paramount and the artificial single-exposure setting poorly mirrors a competitive media environment — concerns addressed by competitive-framing and over-time designs.
Limitations and considerations
- Single-exposure laboratory or survey-experiment settings may overstate effects relative to a noisy, competitive media environment.
- Framing effects are fragile: they often shrink when frames compete, when audiences have strong priors, or after delay.
- External validity is limited by artificial stimuli, convenience samples, and forced exposure unlike real media choice.
- Effects can be sensitive to stimulus wording, so a single operationalization may not generalize to the broader frame.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Randomization licenses causal claims about framing that observational content analysis cannot support.
- Cleanly separates equivalence from emphasis framing and isolates the frame from confounding message features.
- Built-in moderation testing reveals which audiences are susceptible and under what conditions.
- Flexible across domains — political, health, risk, and marketing communication — and scalable through online panels.
Applications
- Testing whether economic versus security frames shift public opinion on immigration or policy.
- Evaluating gain- versus loss-framed health messages for behaviors like screening or vaccination.
- Assessing risk-communication frames for climate change, disasters, or emerging technologies.
- Studying how source cues and competing frames moderate persuasion in political campaigns.
- Pretesting message strategies for advocacy, public-health, and marketing campaigns.
How it works
How it works
The researcher constructs two or more message versions that differ in frame while holding factual content, length, and style as constant as possible. Equivalence framing varies only the valence of logically identical information; emphasis framing stresses different substantive considerations relevant to the issue.
Participants are randomly assigned to a frame condition (and usually a control), so that on average the groups are equivalent on all confounders and any post-treatment difference can be attributed to the frame.
After exposure, outcomes are measured — issue attitudes, policy support, perceived importance, behavioral intentions — using validated scales, ideally with attention and manipulation checks to confirm participants processed the intended frame.
The average framing effect is the difference in mean outcomes between frame conditions; with a control group, each frame's effect is estimated relative to no message, and confidence intervals or tests quantify uncertainty.
Formula
\hat{\tau} = \bar{Y}_{\text{frame A}} - \bar{Y}_{\text{frame B}}Moderators — political knowledge, prior attitudes, source credibility, competing frames — are tested via interaction terms, since framing effects are notoriously conditional and often shrink or vanish when audiences encounter competing frames or strong priors.
Formula
Y = \beta_0 + \beta_1 F + \beta_2 M + \beta_3 (F\times M) + \varepsilon
Common pitfalls
- Confounding the frame with other message differences (length, evidence, tone), so the manipulation is impure.
- Omitting a control or manipulation check, leaving it unclear whether participants registered the intended frame.
- Testing frames in isolation and generalizing to competitive real-world settings where effects often attenuate.
- Mining many moderators without correction and reporting only the significant interactions.
Frequently asked questions
What is the difference between equivalence and emphasis framing?
Equivalence framing presents logically identical information in different terms — '90% survival' versus '10% mortality' — so any attitude difference reflects pure presentation, echoing Kahneman and Tversky's framing of risky choice. Emphasis framing stresses different substantive considerations that are all true — describing a rally in terms of free speech versus public safety. Communication framing research mostly studies emphasis frames, which is why effects depend so heavily on which considerations audiences find applicable and credible.
Why do framing effects often disappear in competitive settings?
Most framing experiments expose participants to a single frame, but real media environments present competing frames simultaneously. When people encounter both sides, frames can cancel out, and strong frames dominate weak ones regardless of repetition. Audiences with firm prior attitudes also resist reframing. This is why modern designs use competitive-framing conditions and test frame strength and source credibility, finding that one-sided effects overstate real-world influence.
How does a framing experiment relate to framing analysis of content?
They are complementary halves of framing research. Framing analysis describes which frames appear in media and how often — the supply side. The framing experiment tests whether exposure to a frame causally changes audiences — the effects side. A strong research program often uses content analysis to identify the frames actually circulating, then builds an experiment around those real frames to estimate their causal impact, linking prevalence to effect.
Related methods
Sources
Iyengar, S., & Kinder, D. R. (1987). News That Matters: Television and American Opinion. Chicago: University of Chicago Press.
ISBN: 9780226388571Copied.Chong, D., & Druckman, J. N. (2007). Framing theory. Annual Review of Political Science, 10, 103–126.
Open DOI referenceCopied.Entman, R. M. (1993). Framing: Toward clarification of a fractured paradigm. Journal of Communication, 43(4), 51–58.
Open DOI referenceCopied.