Psychological reactance and trait reactance: meaning and application

If you announce a measure and the reactions are fiercer than its content justifies, the feeling that others decide for people may weigh more than the change itself. This article explains psychological reactance and trait reactance, and how researchers measure them. You also read why controlling language can backfire and what is known about reducing reactance.

Psychological reactance is a motivational state that arises when someone sees a freedom they consider important threatened or eliminated, and that is directed at restoring that freedom.

Trait reactance differs: it is a person's relatively stable tendency to experience such threats more readily or strongly. The state arises in a situation and fades again; the trait describes differences between people. Both help explain why pressure sometimes achieves the opposite, but neither alone predicts how one individual responds.

Jack Brehm introduced the theory in 1966 in A theory of psychological reactance (Brehm, 1966). With S. S. Brehm he developed it further in 1981 in Psychological reactance: A theory of freedom and control (Brehm & Brehm, 1981). In their review of fifty years of research, Rosenberg and Siegel summarise the theory in four components (Rosenberg & Siegel, 2018):

1. Freedom: people have behaviours they know they are allowed and able to perform.
2. Threat: anything that impedes such a behaviour is a threat; anything that makes it entirely impossible is an elimination.
3. Reactance: the motivational state that follows. Its strength depends on characteristics of the freedom, such as its importance, and of the threat.
4. Restoration: the state drives people to regain the freedom, for example by engaging in the very behaviour that was restricted or because the threatened option starts to look more attractive.

This explains the boomerang effect: a message pushing someone one way can move them the other way. It can happen, but it is not a law. According to Steindl and colleagues, research since Brehm has yielded many new insights into what happens when people see their freedom threatened (Steindl et al., 2015).

Reactance was long considered not directly measurable. In 2005, Dillard and Shen proposed measuring it as anger combined with negative thoughts. In two studies of health messages, on flossing (N = 196) and alcohol use (N = 200), a composite index of the two fully mediated the effect of threat to freedom and trait reactance on attitude and intention (Dillard & Shen, 2005). This is known as the intertwined model.

In practice, participants rate how angry, irritated and annoyed a message made them. They also list their thoughts, which are coded as supportive, neutral or negative (Reynolds-Tylus, 2019). Perceived threat is often measured separately, as the theory places it before reactance.

A meta-analysis by Rains of 20 studies with 4,942 participants in total found that the intertwined model described the data best (Rains, 2013). Reynolds-Tylus notes that anger and negative thoughts were only modestly related there (r = 0.31), which raises the question of whether both measures capture one concept (Reynolds-Tylus, 2019).

Clinical psychologists later also treated reactance as a personality characteristic: the consistent tendency to perceive situations as if one's freedom is threatened (Rosenberg & Siegel, 2018).

• Merz questionnaire: in 1983 Merz published a German-language Fragebogen zur Messung der psychologischen Reaktanz in the German journal Diagnostica (Merz, 1983), with 18 items according to Rosenberg and Siegel. We have not consulted the original text.
• Hong Psychological Reactance Scale: Hong and Page found a four-factor structure among 257 students (Hong & Page, 1989). Hong and Faedda refined the scale to 11 items using data from 3,085 participants in Sydney (Hong & Faedda, 1996).
• Testing: across three studies, Shen and Dillard confirmed four factors, but also that the scale can be treated as one dimension at a higher level (Shen & Dillard, 2005).

The trait concept is contested: Rosenberg and Siegel point to the psychometric instability of well-known scales. So keep two questions apart: is someone responding to this message with reactance now (state), and is someone generally inclined to respond this way (trait)? A high trait score does not predict a response to one specific message.

Experienced autonomy and resistance type have separate articles. An objection can be substantive without any reactance.

According to the theory, a message that leaves little room threatens the freedom to decide for oneself. Words such as 'must' and 'mandatory' make that threat explicit. In a meta-analysis of 33 studies, Li and Shi found that highly freedom-threatening language, compared with low-threat language, went with more anger (r = 0.21) and more negative thoughts (r = 0.17). Both went with lower persuasiveness; gain or loss framing made no difference (Li & Shi, 2026). These are small to moderate associations.

By the same logic, scarcity, time pressure or limiting options can also threaten freedom of choice. Whether it does depends on the situation and has not been studied for every technique. See also Cialdini's principles.

What does the literature say about reducing reactance?

Reviews list message features that evoked less reactance in experiments (Reynolds-Tylus, 2019; Rosenberg & Siegel, 2018):

• Offering choice: several recommended behaviours rather than one mandatory action reduced perceived threat and reactance in some studies.
• Autonomy-supportive language: less controlling wording, with words such as 'perhaps' or 'possibly', or a closing line stating that the choice lies with the recipient. Such a line reduced perceived threat in the study by Miller and colleagues (Miller et al., 2007), but a study with radio spots found no effect.
• Narratives: stories are less readily recognised as persuasion attempts and evoked less reactance than expository texts.
• Forewarning (inoculation): warning in advance that a message may threaten one's freedom reduced reactance in one study. A follow-up found this only with low-controlling language, and a warning can also increase reactance.

Studies per technique are still few and replication is needed. Treat these techniques as testable hypotheses, not guarantees.

1. Name the freedom. Which choice might the target group see as threatened, and how important is it?
2. Separate content and form. Examine separately whether the objection concerns the measure itself or the way it is presented.
3. Create variants. Compare a controlling and an autonomy-supportive version with the same content.
4. Measure the state directly. Include threat to freedom, anger and negative thoughts, alongside the intended outcome.
5. Measure the trait separately. Use trait reactance as an explanatory factor, not to classify or exclude people.
6. Decide in advance what counts. Set outcome measure and relevant difference beforehand and test in the target group concerned.

See also how to translate target group insights into testable messages.

This example is fictional and contains no research results.

Situation. A municipality introduces a mandatory residents' parking permit in a residential neighbourhood. The draft letter opens with: "You are required to apply for a permit before 1 March. Without a permit you risk a fine."

Decision question. How can the municipality make the measure clear without its tone triggering unnecessary resistance?

Available information. The measure is fixed. Residents do choose how to apply (online or at the desk), which licence plate to register and how to use the visitor scheme.

Suitable approach. The research team drafts two letters with the same facts. Version A keeps the controlling opening. Version B first explains the purpose and names residents' real choices. Residents read one version; the team then measures perceived threat, anger, a short thought listing, understanding of the rules and intention to apply on time. Trait reactance is measured separately.

Possible interpretation. If version B evokes less anger with equal understanding, this points to an effect of form. If resistance stays the same, it is probably a substantive objection that calls for a different conversation. Version B offers only real choices.

Next step. The municipality chooses based on measures set in advance and follows actual application behaviour.

• Confusing state and trait. High trait reactance does not mean someone responds to every message with reactance, and a low score does not rule out a strong response.
• Calling all resistance reactance. Resistance can also stem from objections, distrust or habit.
• Overestimating effect sizes. Associations are small to moderate and apply to groups, not individuals.
• Forgetting limited samples. Much research relies on convenience samples of adolescents and students, and on health messages (Reynolds-Tylus, 2019). Other target groups require their own testing.
• Ignoring culture. People from a collectivistic background were less sensitive to threats to their individual freedom but more sensitive to threats to a collective freedom; for people from an individualistic background it was the other way round (Jonas et al., 2009).
• Using techniques as tricks. Offering choice is honest only if the choice is real; a sham choice can damage trust.

Psychological reactance explains why people sometimes resist a message because it limits their freedom, whatever its content. Keep the state and the trait apart, including in your measurement design. Controlling language goes together with more anger and negative thoughts, and choice, autonomy-supportive language and narratives can reduce reactance. Effects are modest and context-dependent, so test in your own target group.