The SAGE Handbook of Persuasion

by James Price Dillard

CHAPTER 11

  Reactance Theory and Persuasion

  Brian L. Quick, Lijiang Shen, and James Price Dillard

  Sir Isaac Newton famously claimed that for every action there is an equal and opposite reaction. It is doubtful that he had persuasion in mind when formulating his third law of motion, but, the principle alerts us to the parallel possibility that suasory efforts may too instigate reactions that run counter to the intended action of the message. Quite apart from the laws of motion, the literature suggests that opposing reactions may vary greatly in degree. Some appeals are so effective that the counter-reaction is scarcely measurable. Others may produce no discernable persuasive change, perhaps because the action and reaction are closely balanced. In still other cases, the predominant response is counterpersuasion, a condition that has come to be known as a boomerang effect.

  All of these outcomes are of interest to persuasion researchers, but boomerangs and failure to persuade are both common and pragmatically frustrating. The perspective that is most frequently called on to give account of them is psychological reactance theory (PRT; Brehm, 1966). Consequently, this chapter reviews and synthesizes work on this landmark theory. It proceeds by first considering the theory’s original formulation alongside the empirical research done to test it. The second section focuses on refinement of the theory, before turning, in the third movement, to contemporary research. Following from that, directions are offered for future inquiry.

  Overview of Reactance Theory and Classic Research

  * * *

  There are four components to reactance theory: Freedom, threat to freedom, reactance, and restoration of freedom. Freedoms are beliefs about the ways in which one can behave. Brehm and Brehm (1981) argue that freedoms are “not ‘abstract considerations,’ but concrete behavioral realities” (p. 12). Nonetheless, freedom is defined broadly to include actions as well as emotions and attitudes (Brehm, 1966). In other words, freedom to do, freedom to feel, or freedom to hold a particular evaluation, or not to. Individuals possess freedoms only to the extent that they have knowledge of them and perceive that they are capable of enacting the behavior.

  Given that an individual perceives a specific freedom, anything that makes it more difficult for him or her to exercise that freedom constitutes a threat (Brehm, 1966; Brehm & Brehm, 1981). Even an impersonal event, such as the weather can be viewed as a threat, if it renders the exercise of a freedom more difficult. However, social influence as a threat is most pertinent to questions of persuasive communication. Important to understanding the theory is recognition that the term threat is used narrowly to mean threat-to-freedom, and not threats as used in everyday language, nor threats to well-being used in research on fear/threat appeals, or threats as a forewarning of social influence in research on inoculation.

  Psychological reactance is “the motivational state that is hypothesized to occur when a freedom is eliminated or threatened with elimination” (Brehm & Brehm, 1981, p. 37). This is the key mediator and central explanatory mechanism of the theory. Its proposed existence follows from the assumption that humans place a high value on choice, autonomy, and control. As reflected in many worn phrases, people seek to be the captains of their own destinies, the masters of their own fates.

  The fourth component of the theory is restoration. PRT contends that when a perceived freedom is eliminated or threatened with elimination, the individual will be motivated to reestablish that freedom. Direct restoration of the freedom involves doing the forbidden act. In addition, freedoms may be restored indirectly by increasing liking for the threatened choice (Brehm, Stires, Sensenig, & Shaban, 1966; Hammock & Brehm, 1966), derogating the source of threat (Kohn & Barnes, 1977; Schwarz, Frey, & Kumpf, 1980; Smith, 1977; Worchel, 1974), denying the existence of the threat (Worchel & Andreoli, 1974; Worchel, Andreoli, & Archer, 1976), or by exercising a different freedom to gain feeling of control and choice (Wicklund, 1974). (See Quick & Stephenson, 2007b for a measure of various means for restoring reactance.)

  Corollaries of Reactance Theory

  PRT is built around four principles (Brehm, 1966). The first specifies that reactance can only be aroused if individuals believe they have freedom over a particular outcome. An early study illustrating PRT’s first principle was conducted by Wicklund and Brehm (1968), in which participants were asked to evaluate a job applicant. Their results revealed differences among high competent individuals following exposure to the freedom-threatening condition. That is, competent individuals exposed to the freedom-threatening communication resulted in greater reactance than exposure to the non-freedom-threatening condition. No significant difference emerged for non-competent individuals (Wicklund & Brehm, 1968). In other words, for individuals aware of their abilities, exposure to freedom-threatening messages resulted in heightened reactance.

  Principle two pertains to the relationship between the importance of a freedom and the magnitude of reactance. Specifically, PRT asserts that as the threatened behavior increases in attractiveness, so does the amount of reactance aroused. Brehm and Weinraub (1977) recruited toddlers to participate in an experiment in which 2-year-olds were presented with two toys. However, one of these toys was placed behind a transparent barrier, whereas the other was next to the barrier. As expected, when the toys were different in shape, as opposed to similar in shape, consistent with the second principle, preference for the obstructed toy was heightened among boys. However, this hypothesized finding was not true for the girls. As will be shown in the subsequent pages, more recent work has shown a relationship between the importance of the threatened freedom and reactance arousal.

  The third principle states that reactance increases as the number of freedom threats increases. Although conceptually cogent, extant PRT research is inconsistent in supporting this principle (for a review, see Brehm & Brehm, 1981). One such study finding support for threat intensity on reactance was conducted by Heller, Pallak, and Picek (1973). In their project, in line with PRT, as persuasive intent and freedom threats increased, participants’ attitude change in the opposite direction spiked. These researchers reasoned that perceiving a message as a freedom threat results in maladaptive responses such as unfavorable attitudes toward the advocacy.

  The fourth principle maintains that as implied threats are present, the magnitude of reactance will increase (Brehm, 1966; Brehm & Brehm, 1981). Implied threats are often communicated in the form of a persuasive message. An earlier study conducted by Reich and Robertson (1979) examined littering behaviors following exposure to variations in freedom-threatening language. Specifically, in their three-study experiment, they found individuals presented with messages featuring explicit commands such as “Don’t you dare litter” resulted in greater littering than exposure to messages featuring appeals to social norms. As shown earlier, their study provides support for the fourth principle of PRT in that as implied threats increase, reactance arousal increases. However, it should be noted that the previously mentioned research in support of these four principles was conducted without measuring reactance. That is, these researchers merely hypothesized the presence of reactance given the outcomes believed to be linked to this aversive state. As will be shown later in this chapter, following a validated operationalization of psychological reactance (Dillard & Shen, 2005), more recent work has begun and continues to test the four theoretical principles advanced by Brehm (1966).

  Progress in Refining Reactance Theory

  * * *

  The Nature of Reactance

  Conceptually speaking, reactance was defined primarily in terms of its antecedents and outcomes. For example, reactance is the result of the collision between a freedom and a threat to freedom. Nearer the end of the process, reactance is the state that is responsible for efforts to restore freedom. In either case, it has been defined in terms of either what causes or what effect it has, but not what it is. Apart from a brief mention of the possibility that individuals “may be aware of hostile and aggressive feelings” (Brehm, 1966, p. 9), if the level of
reactance arousal is high, the nature of reactance itself is rarely addressed. Brehm’s apparent reluctance to explicate the principle mechanism of the theory may have followed his estimate of its potential for measurement. According to Brehm and Brehm (1981), “reactance has the status of an intervening, hypothetical variable. … We cannot [emphasis added] measure reactance directly, but hypothesizing its existence allows us to predict a variety of behavioral effects” (p. 37).

  The position that reactance cannot be measured is logically consequential. In effect, it assigns reactance to a black box. Message features go in the box and outcomes emerge, but the operations within the box itself remain mysterious and unknowable. When results match with theoretical predictions, then it is assumed that the process-in-a-box must have occurred. But, when results do not work out as expected, then the failure must be with the messages, the circumstances, the sample, or something other than the box. Indeed, it has been common to invoke reactance as the explanation for a failed persuasion attempt. In this regard, it became the default assumption for any persuasive effort that produced a boomerang or a no-effect finding. It is better to have a black-box theory than no theory at all. But, it is vastly preferable to understand and measure the workings inside the box. Only then can the entire theoretical process be open to empirical test. Hence, it was important to understand the nature of reactance in order that it might be assessed directly.

  Since the time of Brehm’s writings, scholars have applied ideas drawn from reactance theory to a number of different domains. In the course of analyzing and extending the theory, reactance has, implicitly and explicitly, been defined in several different ways. Dillard and Shen (2005) describe four views. In the first, reactance is purely cognitive. Research in the cognitive response tradition adopted this perspective (e.g., Petty & Cacioppo, 1986), as did work on the clinical manifestations of reactance (Kelly & Nauta, 1997). One advantage to conceiving of reactance in this way is that it immediately becomes measurable through a variety of self-report techniques. The means that is most obviously relevant to questions of persuasion is the widely used thought-listing technique (Petty & Cacioppo, 1986). This purely cognitive view suggests that reactance can be conceived of and operationalized as counter-arguing.

  Second, citing similarities between antecedents of reactance and cognitive appraisals that lead to anger, some writers suggest that reactance might be considered, in whole or in part, as an emotion (Dillard & Meijnders, 2002; Nabi, 2002). This claim aligned well with Brehm’s description of reactance as the experience of hostile and aggressive feelings (Seltzer, 1983; White & Zimbardo, 1980; Wicklund, 1974). In this view then, reactance is more or less synonymous with the family of concepts that index varying degrees of anger (e.g., irritation, annoyance, and rage). From this perspective, reactance might be operationalized in various ways, including asking individuals to make a judgment on a close-ended scale regarding the degree to which they are experiencing anger.

  A third logical option holds that reactance might be considered as both affect and cognition. Though unrelated to reactance per se, one example of this type of thinking can be seen in Leventhal’s (1970) parallel processing model. He posits that individuals have both cognitive and emotional reactions to persuasive health messages and that those reactions have unique effects on message acceptance. Evidence consistent with both points can be found in studies of cognitive and emotional responses to public service announcements (Dillard & Peck, 2000, 2001; Dillard, Plotnick, Godbold, Freimuth, & Edgar, 1996; Stephenson, 2003; Witte, 1994).

  A final possibility also suggests that reactance has both cognitive and affective components. However, unlike the previous position, which specified distinct effects, in this fourth perspective, cognition and affect are intertwined. In fact, they are intertwined to such a degree that their effects on persuasion cannot be disentangled. This view conceives of reactance as an alloy of its components, rather than a simple sum of distinct elements (as is implied by the previous position).

  Dillard and Shen (2005) report on a research project designed to test each of the four alternatives. The study itself utilized two messages: One that promoted the use of dental floss and the other that encouraged university students to curb their alcohol intake. For both messages, the data clearly favored the intertwined model over the alternatives. Hence, the authors concluded that reactance was in fact measureable and that it could be modeled as a combination of anger and negative cognitions.

  Several studies provided additional support for this conclusion. For example, Rains and Turner (2007) conducted an investigation that pitted the intertwined model against the dual-process model, as well as a fifth alternative in which anger preceded negative cognition. Across their two health-related studies, the intertwined model showed the best fit to the data. In a research project that examined exercise behaviors, Quick and Considine (2008) tested and found support for the intertwined model in a sample comprised of members of a health and fitness center. Additional, complementary findings can be seen in Quick and Stephenson (2007a) and Shen (2010a). Given the many variations across studies in terms of messages and samples, and the apparent consistency of findings, the intertwined model seems to provide a workable representation of reactance itself and, thus, an operational means of moving away from black-box theorizing.

  Measuring Reactance

  The method (discussed earlier) for measuring the components of reactance involves two procedures (Dillard & Shen, 2005). One uses Likert scales to assess anger. For example, respondents were asked: To what extent did the message that you just read make you feel … angry/irritated/annoyed/aggravated: 0 = none of this feeling, 4 = a great deal of this feeling. The complementary procedure asks respondents to list any thoughts that they had while reading the message. The results were coded by research assistants in a four-step process that unitized the data, then screened out self-reports of emotion as well as cognitions that were unrelated to the message or topic.

  Although not the primary focus of her study, Lindsey (2005) developed a measure of reactance to bone marrow donation messages using a four-item scale informed by Hong’s (Hong & Faedda, 1996) Psychological Reactance Scale. The items were: I am uncomfortable that I am being told how to feel about bone marrow donation; I do not like that I am being told how to feel about bone marrow donation; It irritates me that the message told me how to feel about bone marrow donation; I dislike that I am being told how to feel about bone marrow donation. In terms of the economics of research, the potential advantages of the four-item measure are apparent: It provides a brief index that could be easily incorporated into surveys without the expensive and laborious tasks of gathering and coding free-response data.

  Quick (2012) undertook a comparative analysis of the two methods in terms of reliability and validity (by evaluating each in a nomothetic network with a freedom threat, attitude, motivation, and source appraisal). Both approaches exhibited satisfactory reliability. Both were also sensitive to variations in forceful language and both predicted attitude, motivation, and source appraisal in theoretically meaningful ways. Quick recommends continued use of the Dillard and Shen (2005) measure on the basis of its relatively greater reliability and validity as well as the fact that the association between the two approaches is low: The four-item measures correlate only .23 and .22 with anger and negative cognitions respectively. Most importantly, he recommends the Dillard and Shen (2005) measure as it distinguishes reactance and freedom threat as distinct entities, whereas Lindsay’s (2005) measure combined these concepts. In order to be consistent with PRT (Brehm, 1966), a freedom threat must precede reactance. For this reason, combining a freedom threat with reactance does not permit researchers the opportunity to examine the relationship between them, which is discussed in greater detail later in this chapter. His recommendation, while reasoned, leaves open the question of how to more efficiently assess reactance.

  One option explored by Rains and Turner (2007) and by Quick and Stephenson (2008), involves participant cod
ing of their own cognitive responses. This generally precludes fine-grained analyses that unitize, then screen out emotions and irrelevant cognitions, because of the training time required to carry out the layers of classification. But, due to the fact that thought listing often contains ambiguous language, research participants may be in a uniquely valid position to judge the valence of their own reports. To wit, when confronted with a cognitive response such as “Is that claim supported by science?” researchers are forced to guess as to whether the query is critical or supportive. In principle, research participants have more direct access to their own evaluations, and can, conceivably, provide more valid judgments. Of course, focused research is needed to evaluate the veracity of this speculation.

  Reactance as an Individual Difference

  Psychological reactance was first conceived as situation specific (Brehm, 1966; Wicklund, 1974). Accordingly, most of the classic reactance research was done on situational reactance, such as alternative restriction and forced choice and in social influence settings (see Burgoon, Alvaro, Grandpre, & Voulodakis, 2002, for a review). However, Brehm and Brehm (1981) recognized that reactance could be conceptualized as a trait too, a position consistent with the theory’s assumption that people vary in the strength of their needs for autonomy and self-determination (Wicklund, 1974).

  Several attempts have been made to develop scales to measure reactance proneness (Dowd, Milne, & Wise, 1991; Hong, 1992; Hong & Faedda, 1996; Hong & Page, 1989; Merz, 1983). Merz (1983) developed the first self-report measure, which showed promising results, but later work yielded inconsistent factor structures (see Donnell, Thomas, & Buboltz, 2001; Hong & Ostini, 1989; Tucker & Byers, 1987). Subsequently, Dowd et al. (1991) developed the Therapeutic Reactance Scale for clinical purposes. However, the two-factor solution lacked a simple structure, and several of the items exhibited poor correspondence with the theoretical construct of reactance (cf. Buboltz, Thomas, & Donnell, 2002).