Framing: meaning and application

The same price change can sound reasonable or unacceptable, depending on how you present it. Readers also bring their own frame, such as "I can't take this on as well". This article explains what framing is, how strong the evidence is and how to apply and test framing responsibly.

Framing is the way in which the presentation, wording or selection of information shapes how people interpret and evaluate a topic or choice. Cognitive framing is the frame in which a person places a choice for themselves.

Equivalence framing changes only the wording of logically equivalent information; emphasis framing changes which considerations come to the fore. The framing effect in risky choice is well replicated, but its size depends strongly on task and context.

Chong and Druckman describe framing as the process by which people develop a particular conceptualisation of an issue or reorient their thinking about it. A framing effect occurs when a change in the presentation of an issue produces a change of opinion (Chong & Druckman, 2007).

Two levels help. A frame in communication is the way a message highlights certain features, such as consequences or values. A frame in thought is the set of dimensions on which a person evaluates a topic. A framing effect occurs when the first changes the second.

Sociologist Erving Goffman used frames in Frame analysis (1974) for schemata of interpretation that allow people to locate, perceive, identify and label events (Goffman, 1974). This later led to research on frames in news and politics.

In decision research, Tversky and Kahneman showed that preferences shift predictably when the same decision problem is worded differently (Tversky & Kahneman, 1981). In their best-known problem, participants chose between two programmes against a disease expected to kill 600 people. When the outcomes were described as lives saved, 72 per cent of 152 participants chose the certain option. Described as deaths, 78 per cent of 155 participants chose the risky option. Expected outcomes were identical.

Levin, Schneider and Gaeth attributed seemingly contradictory findings to different operational definitions of framing. They distinguished three types of valence framing, in which information is presented in positive or negative terms (Levin et al., 1998):

• Risky choice framing: outcomes of options with different risk are described as gains or losses, as in the disease problem.
• Attribute framing: one characteristic of a product, service or event is worded positively or negatively, which affects its evaluation.
• Goal framing: a message stresses what you gain by acting or lose by not acting, which affects its persuasiveness.

These types work through different mechanisms, so when comparing studies, first check which type was examined.

Communication science developed a broader meaning. Entman described framing as selecting some aspects of a perceived reality and making them more salient in a text, so as to promote a particular problem definition, causal interpretation, moral evaluation or treatment recommendation (Entman, 1993).

Chong and Druckman set the two meanings side by side:

• Equivalence framing: different but logically equivalent phrases change preferences. The information stays the same; only the packaging changes.
• Emphasis framing: a message stresses qualitatively different but potentially relevant considerations, such as free speech versus public safety. Here, the content changes too.

With equivalence framing you test only the wording; with emphasis framing you choose which considerations to mention, and thus which to leave out.

Tversky and Kahneman defined a decision frame as a person's conception of the acts, outcomes and contingencies associated with a choice, controlled partly by the wording of the problem and partly by the norms, habits and personal characteristics of the decision-maker. We refer to that second part as cognitive framing: the frame in which a person places a choice for themselves.

Someone who thinks "I can't take this on as well" places a choice in the category "an extra burden on top of everything else". Within that frame, effort and risk weigh heavily and benefits fade. That is an interpretation, not a measured law.

Research on choice bracketing shows that people can evaluate choices one at a time (narrow bracketing) or together (broad bracketing), and that this grouping affects their behaviour (Read et al., 1999). Linking a choice to other ongoing demands is one possible form of such grouping.

Kühberger's meta-analysis covered 136 empirical papers with nearly 30,000 participants and 230 effect sizes on framing in risky choice. The average effect was small to moderate, varying widely between designs (Kühberger, 1998). A reanalysis using p-curve, which accounts for publication bias, found a corrected d = 0.52, compared with d = 0.31 originally. A new meta-analysis of 2016 experiments found d = 0.56 (Steiger & Kühberger, 2018). The measure d expresses a group difference in standard deviations.

In the large-scale Many Labs replication, with 36 samples and 6,344 participants, the gain-loss framing effect replicated consistently, with an effect of around d = 0.6, smaller than in the original study (Klein et al., 2014). That makes this effect reliable, but not every form of framing in every context.

According to Kühberger, the effect depended partly on whether a frame changed the reference point or only made outcomes more salient, and on the response mode: choosing or rating. Steiger and Kühberger also found that it varies with the task manipulation.

For framing in public communication, Chong and Druckman mention:

• Strong predispositions: firmly held views make people less susceptible to a frame.
• Source credibility: frames from credible sources are more likely to shift opinions.
• Competition: the effect depends on the relative strength of competing frames.

Laboratory effects therefore do not predict what will happen with your target group.

Tone of voice is how a sender sounds; framing is the frame a message offers. On tone, see What is tone of voice in a target group profile?.

You infer cognitive frames from how people talk about a choice and how they choose. A workable sequence (editorial advice):

1. Ask openly about the choice: let people describe in their own words what it means to them, before introducing your wording.
2. Code the frames: group statements by frame, such as "extra burden", "opportunity" or "obligation", using a codebook and a second coder.
3. Record what the frame does: which considerations come up, and which are missing?
4. Test the association: measure frame and choice separately in a larger sample.
5. Test experimentally: show comparable groups a message that does or does not address the frame.

Only step 5 says something about cause and effect. For capturing associations and people's own wording, see How do you measure associations?.

1. Start with the decision: which choice does the reader make, and what information do they need?
2. Keep the content complete: every version contains all relevant facts, even when you shift the emphasis.
3. Formulate a hypothesis for each frame: what do you expect the frame to change, and why?
4. Choose outcome measures in advance: comprehension and perceived fairness as well as choice.
5. Test with comparable groups: randomly assigned and with a pre-registered analysis plan.
6. Interpret with care: a result applies to that group, channel and moment.

For turning insights into testable messages, see From target group insights to testable messages.

Every message has a form; the question is whether it helps people decide well or misleads them. A frame becomes misleading when it omits relevant facts, gives a false impression of size or risk, or applies pressure that undermines freedom of choice. Equivalence framing can mislead too, for example if you deliberately choose wording you expect will make readers underestimate a size.

A practical test (editorial advice): could you explain your choice of frame to the reader without discomfort? For the broader considerations, see Neuromarketing ethics: key questions and responsibilities.

This example is fictional. No research results are known.

Situation and choice question. A fictional energy supplier is raising its fixed tariff. How should it present the change so that customers understand it and choose deliberately: stay, adjust or switch?

Available information. In exploratory interviews, customers describe such a letter as "yet another thing": "I can't take this on as well". Within that frame, they barely read the letter.

Suitable approach. The team creates two informationally equivalent versions. Both contain the new tariff, the monthly difference for an average household, the start date, the right to switch, the reason for the increase and one step to check one's own usage. Version A opens with the reason, version B with the monthly amount and that step. Pre-set outcome measures: correctly estimated monthly amount, perceived fairness and usage checks.

Possible interpretation. If version B readers estimate their monthly amount correctly more often, that opening may make the information more accessible to those who see the letter as a burden. It says nothing about later satisfaction.

Next step. The team repeats the test with another customer group first. A version that hides the increase or makes it look smaller is ruled out in advance.

• Treating framing as a trick: a frame changes interpretation but does not replace a good offer or honest information.
• Mixing up types: a result on gain-loss framing says little about emphasis framing.
• Adopting laboratory effects directly: effect sizes from standardised tasks do not automatically apply to your channel.
• Applying group results to individuals: an average effect does not predict how one person will respond.
• Reading association as cause: a frame that accompanies a choice does not necessarily cause it.

Framing describes how presentation and framing steer the interpretation of a choice. Equivalence framing changes only the wording, emphasis framing also the considerations, and cognitive framing describes the frame people bring themselves. The effect in risky choice is well replicated, but its size depends on task and context. So first study how your target group frames the choice, give complete information and test frames before wider use.