Reactance Versus Rationalization

Rationalization and Reactance

The term rationalization refers to a range of psychological processes, all designed to make a specific target appear more palatable to the rationalizer. For example, cognitive dissonance research suggests that people will go to great lengths to maintain a view of their behavior as consistent with their preferences, often by enhancing the perceived desirability of a decision that has already been made or an action that has already been taken (Cooper, 2007; Festinger, 1957). Likewise, system-justification theory proposes that people are motivated to enhance the legitimacy of the system within which they operate (Jost, Banaji, & Nosek, 2004), viewing their given status quo as the way things should be (Kay et al., 2009). This stream of research suggests that restrictions on freedoms—or any new circumstances, for that matter—will most likely be met with rationalization. Although NHL hockey players may greatly value fighting, this research suggests that they would actually rationalize a ban on fighting.

A separate body of research, however, indicates that people will have the complete opposite response to restricted freedom. Reactance theory suggests that people are motivated to restore restricted freedoms, and respond negatively to others’ attempts to constrain their freedoms (Brehm, 1966, 1989; Wicklund, 1974). In a seminal study, participants rated a record as much more desirable if they learned that they could not choose to receive it as a gift, and had to choose a different record instead (Brehm, Stires, Sensenig, & Shaban, 1966; see also Chartrand, Dalton, & Fitzsimons, 2007; Snibbe & Markus, 2005). This stream of research suggests that restricted freedoms are met with backlash; according to this research, hockey players would lash out against a ban on fighting, exaggerating the value they attach to this practice.

Absoluteness

These two contradictory sets of findings have coexisted in the social psychological literature for decades. We tested a possible resolution that revolves around the construct of absoluteness. An absolute restriction is complete, certain, and permanent; for example, a new law that definitively bans all senior citizens from driving would be an absolute restriction. A nonabsolute restriction is incomplete, uncertain, or temporally limited; for example, the ban on senior citizens driving would be nonabsolute if it allowed some to continue driving, required further approval before coming into effect, was difficult to enforce, or was in danger of being repealed by its opponents. We tested the hypothesis that people respond to absolute restrictions with rationalization and to nonabsolute restrictions with reactance.

The literatures on rationalization and reactance are consistent with this proposal. Most rationalization research has involved events that already happened, or that would definitely happen; that is, most rationalization research has focused on absolute restrictions. In cognitive dissonance research, participants have typically rationalized an action that they already committed and could not change. System-justification theory proposes that people rationalize persistent features of the social, economic, and political status quos (e.g., racial and gender inequalities), which most people likely perceive as relatively permanent and unchangeable (see also Laurin, Shepherd, & Kay, 2010).

In contrast, a survey of the reactance literature reveals that little of it, if any, concerns absolute restrictions. Participants who have their freedom restricted when choosing a record can go out and purchase the record elsewhere. People who are exposed to coercive health messages, which typically produce reactance (Silvia, 2005), can effectively resist these attempts to constrain their beliefs and actions. Thus, we hypothesized that participants would react against nonabsolute restrictions, but that absolute restrictions would lead to rationalization, the exactly opposite reaction.

Study 1

In Study 1, we measured attitudes toward a restriction on participants’ driving rights. We manipulated this restriction’s absoluteness and predicted that participants in the absolute condition would rationalize, reporting more positive attitudes toward restrictions on driving rights compared with participants in a control condition. In contrast, we predicted that participants in the nonabsolute condition would display reactance, reporting more negative attitudes toward restrictions on driving rights compared with participants in a control condition.

Our view of reactance and rationalization suggests that both are motivational processes (see Kay et al., 2009; Wortman & Brehm, 1975); as such, they should emerge particularly strongly in response to restrictions viewed as relevant. People should feel no motivation to protect rights that they do not exercise, nor to adapt to new restrictions that do not affect them. We therefore included a measure of driving frequency in Study 1, predicting that the effects of condition would emerge most strongly among the most frequent drivers.

Results

In the first step of a hierarchical linear regression, we found that more frequent drivers reported marginally more negative attitudes toward reduced speed limits, β = −0.21, t(74) = 1.89, p = .06. Adding two dummy codes representing the three conditions explained a significant additional portion of the variance, ΔF(2, 72) = 6.23, p = .003. Participants who read about an absolute restriction rationalized it, reporting more positive attitudes toward reduced speed limits compared with control participants, β = 0.26, t(72) = 2.09, p = .04. In contrast, participants who read about a nonabsolute restriction reacted against it, reporting more negative attitudes toward reduced speed limits compared with control participants, β = −0.18, t(72) = 1.95, p = .06.

Finally, adding the interaction between condition and driving frequency explained significantly more variance, ΔF(2, 70) = 4.57, p = .01 (see Fig. 1). The effects of the absoluteness manipulations were driven by participants who drove frequently (defined as 1 SD above the mean frequency score)—absolute versus control: β = 0.41, t(70) = 2.50, p = .02; nonabsolute versus control: β = −0.40, t(70) = 2.36, p = .02. Among participants who drove infrequently (defined as 1 SD below the mean frequency score), we observed neither reactance nor rationalization, both βs < 0.12, both ts < 1, n.s.

OPEN IN VIEWER

Fig. 1.

Results from Study 1: attitudes toward a reduction in municipal speed limits as a function of driving frequency and condition. In the absolute condition, participants expected that such a restriction would definitely occur in their locality; in the nonabsolute condition, they expected that such a restriction had been proposed and would take effect if passed by the local government; and in the control condition, they were not given any expectations regarding implementation of such a restriction. “Frequent” and “infrequent” drivers were defined as those 1 standard deviation above and 1 standard deviation below the mean frequency score. Higher numbers indicate a more positive attitude toward a reduction in speed limits.

As predicted, participants rationalized a restriction on their freedom when the restriction was absolute, but reacted against the very same restriction when it was nonabsolute—that is, when it was not yet set in stone. This effect was moderated by an indicator of motivational involvement.

Study 2

In Study 2, we attempted to replicate our findings from Study 1 using a different restriction and a conceptually different dependent measure. In addition, we included two nonabsolute conditions that varied in their distance from absoluteness (i.e., how likely the restriction was to take effect), predicting that both nonabsolute conditions would elicit reactance.

Finally, we sought to address an alternative interpretation of Study 1, according to which participants in different conditions inferred different normative information. Participants in the absolute condition could have assumed that the legislation was objectively desirable, simply because it seemed to have encountered no opposition; participants in the nonabsolute condition could have assumed that the legislation was objectively less desirable, simply because we raised the possibility that it could encounter opposition. Thus, the pattern of results could have been due to this purely inferential process. Even though this alternative explanation does not account for the observed interaction between condition and self-reported driving frequency, we sought to rule it out more convincingly. Thus, in Study 2, we manipulated whether the information our American participants read applied to America or to India (i.e., whether the information was self-relevant or not), which resulted in a 2 (self-relevant vs. non-self-relevant) × 4 (absolute vs. nonabsolute and very likely vs. nonabsolute and somewhat likely vs. control) between-subjects design. This design yielded four pairs of conditions in which the non-self-relevant condition acted as a control condition for its self-relevant counterpart, with normative information held constant within each pair. Comparing self-relevant and non-self-relevant conditions that did not differ in normative information provided a stronger test of our motivational hypothesis.

Results

A 2 (self-relevant vs. non-self-relevant) × 4 (absolute vs. nonabsolute and very likely vs. nonabsolute and somewhat likely vs. control) between-subjects analysis of variance yielded a significant interaction, F(3, 250) = 8.24, p < .001 (see Fig. 2). Participants who read about an absolute restriction rationalized it, reporting more negative attitudes toward cell-phone use while driving if they thought the restriction was self-relevant than if they thought it was non-self-relevant, F(3, 250) = 10.92, p < .001. In contrast, participants who read about a nonabsolute restriction reacted against it, reporting more positive attitudes toward cell-phone use while driving if they thought the restriction was self-relevant than if they thought it was non-self-relevant; this pattern was found both when this nonabsolute restriction was very likely, F(3, 250) = 4.45, p = .005, and when it was somewhat likely, F(3, 250) = 4.62, p = .004. Participants who read no information about a possible restriction reported similar attitudes whether they thought the material they had read was self-relevant or not, F(3, 250) < 1, n.s.

OPEN IN VIEWER

Fig. 2.

Results from Study 2: participants’ attitudes toward using cell phones while driving as a function of condition. Participants in the absolute condition expected an absolute restriction on cell-phone use while driving to take effect, those in the nonabsolute conditions expected that such a restriction was either very likely or somewhat likely, and those in the control condition were given no expectations regarding such a restriction. In each of these four conditions, the materials participants read were either self-relevant (about the United States) or non-self-relevant (about India). Higher numbers indicate more positive attitudes toward cell-phone use while driving.

The manipulation of absoluteness affected attitudes in the self-relevant conditions, F(3, 250) = 7.26, p < .001, but not in the non-self-relevant conditions, F(3, 250) = 1.70, p = .17. In the self-relevant conditions, participants who read about an absolute ban on cell-phone use while driving reported more negative attitudes toward cell-phone use while driving relative to control participants, F(3, 250) = 3.55, p = .02; participants who read about a nonabsolute ban reported more positive attitudes than did control participants, whether the ban they read about was very likely, F(3, 250) = 2.77, p = .04, or somewhat likely, F(3, 250) = 5.46, p = .001. When the possible ban was self-relevant, participants in the two nonabsolute conditions reported similar attitudes, F(3, 250) < 1, n.s.

Thus, Study 2 replicated and extended findings from Study 1, while ruling out an alternative explanation by keeping normative information constant across experimental and control conditions. Participants who read about a restriction that would absolutely apply to them rationalized it: They downplayed the importance of the restricted freedom. Participants who read about a restriction that would likely, but not absolutely, apply to them reacted against it: They attached increased importance to the restricted freedom. Neither of these effects occurred when the restriction was situated in a context (India) that was not relevant to the participants.

General Discussion

We have illustrated people’s divergent reactions to different types of restrictions. When a restriction is definitive, people respond positively to it and minimize the importance they attach to the restricted freedom. In contrast, when there is a chance—even a slim one—that the restriction will not come into effect, people respond negatively to it and exaggerate the importance they attach to the restricted freedom. Both effects occurred specifically for self-relevant restrictions, which suggests that both are driven by motivational, rather than inferential, processes. These studies, therefore, help reconcile two seemingly opposing literatures—on rationalization and reactance—by identifying one factor that determines which process is more likely to occur.

The complete story, however, is probably more nuanced than what these two studies demonstrate. Some restrictions, even when absolute, might be too sudden or abhorrent to elicit rationalization: Imagine if the American government announced today that all marriages would henceforth be arranged by the state. Also, perceptions of absoluteness may matter much more than objective features of restrictions. Some individuals might perceive the most objectively absolute restrictions as nonabsolute, or the most objectively nonabsolute restrictions as absolute. Future research might address these issues involving perceptions of absoluteness.

These findings also have diverse practical applications, potentially shedding some light on the uprisings currently spreading throughout the Middle East. To the extent that a political regime feels absolute and permanent to its citizens, they will rationalize its actions and decisions. But once they learn that similar regimes have been toppled, and are therefore not as absolute as they once seemed, citizens may become reactant and find themselves closer to experiencing the fury and dissatisfaction required to motivate a revolution. Consistent with this perspective, prominent theories of collective action have emphasized the role of perceived cognitive alternatives to current social arrangements in motivating social change (e.g., Tajfel & Turner, 1986).

The two studies presented here help clarify the seemingly opposing predictions derived from theories of reactance and rationalization. Rather than arguing for or against the primacy of one of these two processes, we have demonstrated conditions under which each process is most likely to emerge. Thus, this research advances understanding of the psychological processes involved, and should also prove useful in the wide range of domains in which it is important to accurately predict how people will respond to attempts to restrict their behavior.