Guilt as a Mechanism of Persuasion

Daniel J O’keefe. The Persuasion Handbook: Developments in Theory and Practice. Editor: James Price Dillard & Michael Pfau. Sage Publications. 2002.

This chapter discusses the role of guilt as a mechanism of social influence. An introductory section offers a general sketch of the nature of guilt. The next two sections summarize extant research findings concerning guilt and anticipated guilt as mechanisms of Persuasion, drawn from a variety of research venues. A concluding section fits together several puzzling research findings and identifies some promising foci for future research attention.


Guilt can be understood broadly as a negative emotional state aroused when an actor’s conduct is at variance with an actor’s own standards (Baumeister, Stillwell, & Heatherton, 1994; Miceli, 1992). A paradigmatic guilt-arousing circumstance is one in which a Person has acted in some manner inconsistent with his or her own conception of proper conduct. For example, the sorts of situations that Persons recall as especially associated with guilt feelings are ones that involve conduct such as lying, stealing, failing to Perform duties, neglecting others, failing to maintain a diet or exercise plan, and cheating (Keltner & Buswell, 1996; Tangney, 1992). Thus, guilt involves some self-Perceived shortfall with respect to one’s own standards, where the focus of attention is some particular behavior.

The reactions characteristically associated with guilt make it especially well-suited to exploitation for purposes of social influence. Among the beliefs and feelings distinctively associated with guilt (as opposed to other emotions) are reactions such as “thinking that you were in the wrong,” “thinking that you shouldn’t have done what you did,” “feeling like undoing what you have done,” “wanting to make up for what you’ve done wrong,” and “wanting to be forgiven” (Roseman, Wiest, & Swartz, 1994, p. 215). When Persons recall guilt experiences, they commonly describe themselves as wanting to make amends, feeling responsible, feeling as though they had violated some moral standard, and wishing they had acted differently (Tangney, Miller, Flicker, & Barlow, 1996). Plainly, guilt (in contrast to, say, sadness) has a distinctive action-motivating character.

This action-motivating aspect of guilt presumably is connected to guilt’s being based in a transgression of the actor’s own standards. A Person whose behavior violates some standard (norm or value) that he or she does not accept might acknowledge responsibility for the conduct but presumably will not experience guilt (Miceli, 1992, p. 99) and so might not feel quite the same motivation to make amends or take corrective action. But when the action is inconsistent with the self’s own standards, then guilt (and its associated behavioral motivations) can be aroused.

Arousing Guilt

Guilt can straightforwardly be put to the service of social influence by the influencing agent’s arousing guilt in the target, which in turn motivates the target’s Performance of the desired action. There are two ways in which an influencing agent might attempt to arouse guilt in an influence target. One is for the influencer to draw the target’s attention to some existing inconsistency between the target’s standards and the target’s previous conduct, and the other is for the influencer to induce the target to act in a way that creates such an inconsistency.

Drawing Attention to an Existing Inconsistency

An influencing agent can exploit some existing inconsistency between the target’s previous behavior and the target’s own standards simply by drawing the target’s attention to the inconsistency; the target’s resulting guilt feelings then provide a basis for shaping the target’s future behavior. Three different areas of empirical research illuminate this sort of guiltbased influence mechanism: studies of guilt arousal in interpersonal relationships, research concerning guilt appeals in Persuasive messages, and studies of hypocrisy induction.

Guilt Arousal in Interpersonal Relationships. In everyday life, guilt arousal and attempted guilt arousal occur primarily in the context of close relationships (Baumeister, Reis, & Delespaul, 1995, Study 2; Baumeister, Stillwell, & Heatherton, 1995, Study 2; Jones, Kugler, & Adams, 1995; Vangelisti, Daly, & Rudnick, 1991). People seek to arouse guilt in others primarily for purposes of influence—as a means of inducing the target to undertake some action, refrain from some action, or stop some ongoing action (Vangelisti et al., 1991).

There are a variety of specific ways in which a Person may attempt to arouse guilt, such as indicating that the target is not meeting some obligation that is part of the target’s relationship with the influencer, pointing out that the target’s behavior does not reflect the target’s knowledge of appropriate conduct, and displaying some sacrifice being made on the influencer’s part on behalf of the target (for a discussion of different techniques of guilt induction, see especially Vangelisti et al., 1991; see also Miceli, 1992, and Sommer & Baumeister, 1997). However, each of the different techniques “may generally be considered a variation on saying ‘see how you are hurting me’” (Sommer & Baumeister, 1997, p. 43), and in the context of close relationships, hurting the other would plainly represent a transgression of the target’s own standards. Drawing the target’s attention to such conduct thus offers the prospect of guilt arousal and subsequent behavioral influence.

There has been little research directed at assessing the Persuasive effectiveness of guilt induction in interpersonal relationships. Some self-report evidence suggests that targets of such guilt induction do Perceive the aroused guilt to have had an impact on their subsequent behavior (Baumeister, Stillwell, & Heatherton, 1995, Study 1). However, targets of guilt induction also appear to resent this use of guilt (Baumeister, Stillwell, & Heatherton, 1995, Study 2; see also Rubin & Shaffer, 1987).

Guilt Appeals in Persuasive Messages. Guilt appeals in Persuasive messages, such as advertisements, commonly have two parts. One presents material designed to evoke guilt, characteristically through drawing attention to some existing inconsistency between the receiver’s standards and the receiver’s actions, and the other describes the message’s recommended viewpoint or action, which is meant to offer the prospect of guilt reduction. For example, a consumer advertisement might seek to make receivers feel guilt about the plight of the homeless and then ask for a charitable donation (which offers the prospect of reducing the guilt). Such guilt appeals can vary in any number of ways, but several studies of guilt appeal messages have examined the effects (on guilt arousal and Persuasive outcomes) of variations in the explicitness of the guilt appeal (e.g., Coulter & Pinto, 1995; Pinto & Priest, 1991; Yinon, Bizman, Cohen, & Segev, 1976). The fundamental experimental contrast in this research thus is between relatively direct explicit guilt appeals and relatively less explicit ones.

A meta-analytic review of this research has found that more explicit guilt appeals do arouse significantly greater guilt than do less explicit appeals (expressed as a correlation, the mean observed effect corresponds to r = .43). However, more explicit guilt appeals are significantly less Persuasive than their less explicit counterparts (mean r = Σ.26) (O’Keefe, 2000). This review also found no support for the supposition that moderately explicit appeals might yield both greater guilt and greater Persuasion than would either more or less explicit appeals. In general, when guiltbased appeals (of any level of explicitness) successfully arouse relatively greater guilt, those appeals are unlikely to be Persuasive.

The finding that appeals arousing greater guilt do not enjoy correspondingly greater Persuasive success is rather striking. After all, guilt is characteristically associated with feelings of wanting to change one’s actions, and hence appeals that arouse greater guilt might naturally be expected to effect greater change. A plausible explanation of the observed effect is that the more explicit guilt appeals might have evoked annoyance, resentment, anger, irritation, or similar reactions. As noted previously, studies of interpersonal guiltbased influence attempts have reported evidence of negative reactions such as anger and resentment (Baumeister, Stillwell, & Heatherton, 1995, Study 2; Rubin & Shaffer, 1987). And similar reactions have been reported in studies of guiltbased Persuasive appeals (Coulter, Cotte, & Moore, 1997; Coulter & Pinto, 1995; Englis, 1990; Pinto & Priest, 1991). Thus, although explicit guilt appeals may create greater guilt, they may also arouse other negative feelings that interfere with Persuasive success.

Hypocrisy Induction. The Persuasive effects of drawing attention to inconsistencies between the target’s conduct and the target’s standards is also illustrated by research on hypocrisy induction. In these studies, hypocrisy condition participants are led to advocate some position they support but are reminded of their failure to act consistently with that view. The expectation is that participants will alter their behavior so as to become more consistent with their beliefs. For instance, in Aronson, Fried, and Stone’s (1991) investigation, hypocrisy condition participants were asked to describe a situation in which they failed to use condoms; this served as a reminder of their past behavior. They then composed and delivered a short speech advocating condom use. Such participants subsequently avowed significantly greater intentions to increase their use of condoms than did participants in control conditions (for similar effects, see Dickerson, Thibodeau, Aronson, & Miller, 1992; Fried & Aronson, 1995; Stone, Aronson, Crain, Winslow, & Fried, 1994).

This research has usually been treated as exemplifying dissonance-based influence mechanisms. The hypocrisy induction is presumed to create dissonance through the “inconsistency between advocated Personal standards and past inconsistent behaviors” (Fried & Aronson, 1995, p. 926). Plainly, however, such inconsistency is a paradigmatic guilt-inducing circumstance, and hence the observed effects of hypocrisy induction might reflect guilt arousal effects rather than dissonance arousal effects. For present purposes, it is enough to notice that, viewed from a guiltbased Perspective, hypocrisy induction research illustrates the Persuasive power of drawing the attention of influence targets to existing inconsistencies between their conduct and their standards.

Summary. Arousing guilt by drawing the target’s attention to existing inconsistencies between the target’s conduct and the target’s own standards is potentially a successful mechanism of influence but can also be counterproductive. In particular, interpersonal guilt induction and guiltbased Persuasive appeals seem especially prone to evoke negative reactions that may undermine the success of influence attempts.

Creating an Inconsistency

Instead of drawing the target’s attention to previous conduct that is inconsistent with the target’s standards, an influencing agent might arouse guilt by inducing the target to act in a fashion inconsistent with the target’s standards, thus creating the inconsistency that arouses guilt in the target. Two areas of research may illustrate this sort of guiltbased influence mechanism: research on the transgression-compliance hypothesis and studies of the door-in-the-face (DITF) influence strategy.

Transgression-Compliance. The natural action-motivating quality of guilt, and particularly the guilt-related impulse to alter one’s actions and make up for what one has done, suggests that Persons who commit transgressions (and so presumably experience guilt) will be more likely than Persons who have not committed such transgressions to engage in helping behavior such as complying with a request. This transgression-compliance hypothesis has been studied very extensively (see, e.g., Boster et al., 1999; Carlsmith & Gross, 1969; Freedman, Wallington, & Bless, 1967; McMillen, Jackson, & Austin, 1974). A representative research design is that of Konoske, Staple, and Graf (1979), who had transgression condition participants apparently upset a graduate student’s carefully arranged computer cards. Subsequently, participants were asked by a confederate to make telephone calls to prospective participants. Transgression condition participants volunteered to make significantly more calls than did participants in a nontransgression control condition. Notice that in this research paradigm, guilt is aroused not by the influencer’s calling attention to some past transgression by the target but rather through the influencer’s actually creating the target’s transgression (or the appearance thereof).

A recent meta-analysis of this research concluded that transgressions have a powerful effect on subsequent guilt and compliance. Compared to Persons in no-transgression control conditions, Persons who have committed transgressions feel significantly greater guilt (the mean effect corresponds to r = .45) and are significantly more likely to comply (mean effect of r = .28) (O’Keefe, 2000). Moreover, the transgression-compliance effect is especially robust in the sense that it obtains under a variety of conditions. It occurs no matter whether the transgression is accidental (e.g., knocking over someone’s computer cards) or purposeful (e.g., telling a lie), it occurs no matter whether the subsequent compliance involves a direct request (e.g., being asked to volunteer for an experiment) or simply an opportunity to help (e.g., encountering someone who has just spilled a stack of papers), it occurs irrespective of whether the request (or helping opportunity) is presented by the victim of the transgression or by someone else, and it occurs irrespective of whether compliance benefits the victim (O’Keefe, 2000). That is, under all of these conditions, there is a dependably positive effect of transgression on compliance, with no dependable difference in the size of the effects under these varying circumstances. These findings confirm the powerfulness of transgression-induced guilt as a motivator of compliance and testify to its vigor.

However, the negative feelings engendered by transgression can be neutralized prior to the compliance opportunity and thereby lose their compliance-motivating power. A number of transgression-compliance studies contain an experimental variation in which some event intervenes between transgression and compliance that offers the prospect of neutralizing or alleviating negative feelings (e.g., Cialdini, Darby, & Vincent, 1973; Dietrich & Berkowitz, 1997; McMillen, 1971). The intervening events vary across these studies (including things such as bolstering the participant’s self-esteem, having the victim excuse the transgression, and having an opportunity to express one’s feelings about one’s conduct) but share the property of potentially reducing or nullifying any negative feelings aroused by the transgression. A meta-analytic review has discovered that the observed general effect of transgression on compliance evaporates when such an event intervenes between transgression and compliance (O’Keefe, 2000).

The Door-in-the-Face Influence Strategy. In the DITF influence strategy, a relatively large initial request is made of a Person, which the Person declines. Then a smaller request is made, with the hope that the Person’s having declined the initial request will make the Person more likely to comply with the second request. For example, in Cialdini et al.’s (1975, Experiment 1) classic investigation, unaccompanied individuals were approached on a campus sidewalk by a student who asked that each Person spend 2 hours a week, for a minimum of 2 years, as an unpaid volunteer counselor at the County Juvenile Detention Center. This request was always refused. The second request was that the receiver serve as an unpaid volunteer chaperone, spending 2 hours one afternoon or evening taking a group of juveniles from the detention center to the zoo. When only the smaller request was presented, 17% agreed; however, in the DITF condition, 50% agreed. An extensive body of subsequent research has confirmed that the DITF strategy can indeed dependably yield such enhanced compliance, with a mean effect (expressed as r) of about .10 to .15 (for reviews, see Dillard, Hunter, & Burgoon, 1984; Fern, Monroe, & Avila, 1986; O’Keefe & Hale, 1998, 2001).

It does not seem to have been widely appreciated that the DITF format can be seen to parallel the format of transgression-compliance designs. In each situation, there is initially a transgression (e.g., harm is inflicted on another, a lie is told, a prosocial request is refused), and subsequently a compliance opportunity is presented (commonly in the form of a request). And in each situation, the occurrence of the transgression enhances compliance. Hence, just as transgression-compliance results naturally suggest a guiltbased explanation, so might DITF results, as proposed by O’Keefe and Figgé (1997). In their analysis, DITF success occurs because refusal of the initial request arouses guilt and acceptance of the second request reduces such guilt. That is, the influencer invites the target’s transgression (through the refused initial request), thereby arousing guilt, and then proffers a guilt reduction mechanism (in the form of second-request compliance).

This guilt-based explanation appears consistent with the observed patterns of DITF effects. For example, DITF effects are stronger when the requests come from prosocial organizations (e.g., charities, environmental groups) as opposed to for-profit organizations (see Dillard et al., 1984; O’Keefe & Hale, 1998, 2001). From the Perspective of a guiltbased analysis, this effect can be seen to reflect the greater guilt likely engendered by declining prosocial (as opposed to for-profit) requests (for more extensive discussion of factors moderating DITF effects, see O’Keefe & Hale, 1998, 2001). It remains to be seen whether more direct investigations of guilt’s role in DITF processes will confirm the soundness of a guiltbased explanation (for some complexities, see O’Keefe & Figgé, 1999), but the parallels with transgression-compliance situations certainly are suggestive of similar underlying processes.

Summary. guiltbased social influence mechanisms involving the influencing agent’s creating an inconsistency between the target’s standards and the target’s conduct (i.e., creating a transgression) can be quite powerful means of influence. Moreover, these influence mechanisms do not appear to have the same potential for evoking negative reactions (resentment or anger) as do guiltbased mechanisms based on pointing out the inconsistency.

Anticipated Guilt

Plainly, aroused guilt can be an important mechanism of influence. But there is another (as yet little studied) way in which guilt can figure in social influence, namely, through the behavioral effects of anticipated guilt feelings.

The general idea that anticipated feelings can shape behavioral choices has received support in various behavioral domains. Anticipated emotions have been found to play a role in shaping intentions and actions in studies of weight regulation (Bagozzi, Baumgartner, & Pieters, 1998), salesPerson behavior (Brown, Cron, & Slocum, 1997), entertainment choices (Zillmann & Bryant, 1985, 1994), and drug and alcohol use (Richard, van der Pligt, & de Vries, 1996a). Intentions or behaviors concerning mammography participation (Lechner, de Vries, & Offermans, 1997), consumer purchases (Simonson, 1992), and safe sex practices (Bakker, Buunk, & Manstead, 1997; Richard, van der Pligt, & de Vries, 1996b) have been found to be related specifically to anticipated regret (for general discussions of research concerning anticipated regret, see van der Pligt & Richard, 1994; van der Pligt & de Vries, 1998). And although there is little direct evidence concerning guilt specifically, Birkimer, Johnston, and Berry (1993) did find that Persons’ estimates of how guilty they would feel if they were to engage in various health risk behaviors are related to avoidance of those behaviors; people avoid the actions that they expect would make them feel guilty.

Where anticipated guilt might influence conduct, a corresponding avenue for Persuasion is available. An influencing agent might encourage the target’s anticipated guilt feelings as a way of shaping the target’s conduct. There is some evidence that consumer advertisers are aware of this alternative guiltbased influence mechanism. In sampling 24 popular magazines, Huhmann and Brotherton (1997) found that guiltbased consumer advertising uses both advertisements that try to arouse guilt and advertisements that draw attention to anticipated guilt (using appeals that “offer consumers an opportunity to avoid a transgression.”)

Not much is yet known about exactly what alternative means (e.g., message variations) might successfully arouse anticipated guilt, or about the nature of other elements (e.g., a recommended course of action) that might be needed to connect such anticipated guilt to desired influence outcomes, or about when or whether efforts at creating anticipated guilt might evoke the negative reactions sometimes associated with guilt arousal mechanisms. All of these are plainly useful foci for future research attention. But as one illustration of the unappreciated potential of anticipated guilt’s role in social influence, consider its possible contribution to DITF effects. O’Keefe and Figgé’s (1997) analysis proposed that second-request compliance in DITF settings is motivated by a desire to reduce the guilt created by first-request refusal (and so might be said to be motivated by the anticipation of guilt reduction); however, second-request compliance might instead (or also) be motivated by a desire to avoid additional guilt anticipated to arise from refusing the second request (for some relevant evidence, see O’Keefe & Figgé, 1999). In any case, the general point to be noticed is that anticipated guilt feelings may prove to be as useful a basis of social influence strategies as are actual guilt feelings.

Puzzles and Conjectures

A number of the research findings concerning guilt as a mechanism of influence are rather puzzling. As just one example, consider that transgression’s compliance-enhancing effects are not influenced by whether the victim of the transgression benefits from the compliant act (O’Keefe, 2000). Given that one of the feelings characteristically associated with guilt is “wanting to make amends” (Tangney et al., 1996), one might expect that transgressions would be relatively more powerful motivators of compliance in circumstances where the victim benefits from compliance. This curious finding (and others) can perhaps be illuminated by reconsidering the nature of guilt.

Guilt Reconsidered

It is worth reanalyzing the close relationship between guilt and one’s sense of self. Put briefly, I want to suggest that guilt motivates self-affirmation because guilt-inducing actions represent threats to self-integrity. This idea can be seen as derived from (or as fitting within) self-affirmation theory (Steele, 1988; Steele & Liu, 1983). Self-affirmation theory proposes that Persons seek to maintain an image of the self as “adaptively and morally adequate”; hence, when faced with “information that threatens the Perceived adequacy or integrity of the self,” a Person’s self-affirmation processes are activated and continue until the self-image is restored (Steele, 1988, p. 262). This framework has been used to explain a number of diverse phenomena (e.g., how name-calling can enhance compliance [Steele, 1975]) and, in particular, has been offered as an alternative interpretation of much dissonance-based research (see, e.g., Steele & Spencer, 1992; Steele, Spencer, & Lynch, 1993).

Although self-affirmation theory appears to have given little explicit attention to guilt, it seems plain that the framework offers a promising way of understanding guilt. The general idea that guilt-inducing actions involve self-discrepancy is rather common (if sometimes implicit) and can be detected in a variety of theoretical discussions (e.g., Higgins, 1987; Kugler & Jones, 1992; Miceli, 1992). In fact, it is almost definitionally part of guilt that it involves conduct that is somehow discrepant from the actor’s image of self (specifically, from the standards to which the actor holds his or her own conduct).

This aspect of guilt is important for understanding the consequences of guilt arousal. Given that guilt-inducing actions challenge self-integrity, guilt should encourage a search for self-affirmation. Indeed, in the various research lines discussed previously, the observed sequelae to guilt induction all can be seen to represent means of self-affirmation. In transgression-compliance research formats and DITF situations, helping-related compliance plainly offers the prospect of reaffirming the self (“yes, I’m a good and helpful Person”); in guilt appeals, the recommended course of action attempts to supply a means of self-affirmation (“yes, I’m a good parent”); in hypocrisy induction research formats, behavioral change reaffirms the self’s values (“yes, I really do believe in safe sex practices”); and in interpersonal relationships, self-affirmation can be obtained through changing the behavior that is the basis of the influencer’s complaint (“yes, I’m a good partner”). In short, then, the conjecture here is that guilt works as a mechanism of Persuasion because it provokes self-affirmation processes.

Some Puzzles Illuminated

This way of approaching guilt promises to shed some light on various research findings concerning the role of guilt in social influence. These findings concern the moderators of transgression-compliance effects, negative reactions to guiltbased Persuasive efforts, and the magnitude of the observed effects.

Moderators of Transgression-Compliance Effects. Consider again the seemingly curious finding that transgression enhances compliance independent of whether the Person victimized by the transgression is helped by the act of compliance. This seems an unexpected result if one focuses on the guilt-related feelings of “wanting to make amends,” “wanting to make up for what you’ve done wrong,” and the like. But if, instead, people who are feeling guilty want first and foremost self-affirmation, then this finding is a little less puzzling. One way of achieving such self-affirmation might indeed be to “make amends”—to somehow try to “make it up to” the victimized party (and so do something that benefits the victim). But this is not the only way in which self-affirmation might be obtained. Any helping behavior (no matter whether the victim benefits from it) might provide such affirmation of the self’s worth.

The same reasoning explains why the transgression-compliance effect is not influenced by whether the victim of the transgression makes the compliance request (O’Keefe, 2000). Again, affirmation of the self can be accomplished in various ways. Responding affirmatively to any helping request (whether from the victim or from someone else) can provide the desired self-affirmation. Indeed, there need not even be an explicit request for help; transgression leads to enhanced helping even when, instead of an explicit request, there is simply a behavioral opportunity to help (as when encountering someone who has just dropped a stack of papers [O’Keefe, 2000]). When a Person who has committed a transgression is thereby more likely to spontaneously help in a circumstance where the victim neither presents the helping opportunity nor benefits from the helping, it is difficult to resist the conclusion that generalized self-affirmation processes are at work.

Finally, attention to self-affirmation processes also nicely accommodates the previously mentioned finding that the transgression-compliance effect can be eliminated by an intervening positive event (O’Keefe, 2000). This effect is another indication that guilt feelings, even if aroused by one’s actions toward another, can be neutralized in ways other than compensatory action toward that other. Although people may genuinely want to make amends, the underlying function (of reactions to guilt feelings) appears to be self-repair. Hence, when self-repair is obtained in a way other than compliance or compensatory action (as through those intervening positive events), the Person does not need to engage in compliance to obtain the self-affirming result.

In short, a number of otherwise curious findings concerning transgression’s effects on compliance appear to be illuminated by recognizing the self-affirmation processes set in motion by guilt-inducing events. Guilt engages fundamentally self-directed, and not other-directed, processes. Some of the feelings characteristically associated with guilt do concern other-directed actions (e.g., wanting to make amends, wanting to undo what one has done), but these feelings arise because such actions provide a means of self-affirmation.

Negative Reactions to Guiltbased Persuasive Efforts. A variety of puzzles are connected with the negative reactions (e.g., anger, annoyance, resentment) sometimes associated with the use of guilt as an influence mechanism. It should be noticed that, as yet, there is little detailed evidence concerning the nature of the mechanisms underlying such reactions, the particular situations that might encourage such reactions, and so forth. This plainly provides an important area for future research. But the evidence in hand does offer two findings that need explanation. One is that the negative reactions occur at all, and the other is the apparent difference among various guilt arousal mechanisms in their propensity to evoke such reactions.

It seems that guilt arousal attempts have a distinctive capacity to evoke reactions such as resentment, anger, and annoyance. One might initially hypothesize that any attempted emotional manipulation will evoke similar reactions. However, there is no evidence to suggest that, say, empathy-based or altruism-based appeals are as prone to generate such effects (in fact, for precisely the opposite indication, see Rubin & Shaffer, 1987). Another more specific version of this hypothesis might be that attempted arousal of a negative emotion can induce such reactions. But even this seems insufficiently specific. For example, there is no indication that annoyance is routinely evoked by fear appeals in the way it seems to be by guilt appeals. And although the apparent resentment aroused in the target may have some similarities to reactance (Brehm & Brehm, 1981), a reactance-based account would need to indicate why specifically guilt arousal attempts would arouse reactance more commonly or forcefully than would influence attempts invoking some other basis of influence (e.g., fear, pity, empathy, self-interest). An additional complexity is introduced by the apparent differences among guilt arousal mechanisms in their propensity to arouse negative reactions. For instance, resentment and anger appear more likely to be evoked by relatively explicit guilt appeals in Persuasive messages than by hypocrisy induction manipulations.

A focus on the connection between guilt and the self provides a useful basis for some conjectures concerning these findings. Because guilt involves self-discrepancies, an influencer who explicitly attempts to arouse guilt is, in a sense, questioning the target’s self by pointing out the inconsistency between the target’s conduct and the target’s principles. That is, attempted guilt arousal involves a challenge to the target’s self in ways that other influence mechanisms do not—and for precisely that reason has greater capacity to evoke resentment, anger, or annoyance. After all, I might (with some reason) think that others are not entitled to tell me what is discrepant from my self; only I get to decide that. It’s one thing if I notice weaknesses in my self-integrity; it’s something else if others deign to point them out to me.

Thus, one ought to expect a difference between self-generated guilt (i.e., guilt arising from the target’s having independently noticed the discrepancy between the target’s conduct and the target’s principles) and other-generated guilt (i.e., guilt arising from someone else’s having pointed out the discrepancy to the target) with respect to accompanying anger or annoyance. When the target recognizes the inconsistency himself or herself without having it pointed out explicitly by the influencing agent (as in transgression-compliance formats, the DITF strategy, or hypocrisy induction manipulations), resentment or anger seem less likely to be evoked. However, when the influencing agent explicitly draws attention to the inconsistency (as in common interpersonal guilt induction methods or relatively explicit guiltbased persuasive appeals), such negative reactions are more likely. That is, the more apparent the influencer’s intention to arouse guilt, the more likely it is that the target will experience negative reactions such as resentment and anger. From this vantage point, it is unsurprising that studies of guiltbased Persuasive appeals should have found that more explicit guilt appeals are less Persuasive than less explicit ones (O’Keefe, 2000) given that presumably the more explicit efforts at arousing guilt will be more likely to produce negative reactions.

However, relationally significant others may have a special status as guilt inducers. Such others are in an especially good position to know what is discrepant from one’s self and simultaneously may enjoy something of a privileged position with respect to pointing out discrepancies precisely because they are relationally significant. Hence, although influence targets report feelings of resentment from guiltbased interpersonal influence attempts, they also report changing their behavior in response to those attempts (Baumeister, Stillwell, & Heatherton, 1995). Indeed, one may sense something of a progression here: Consumer advertising that employs explicit guilt appeals arouses guilt but also resentment sufficient to undermine Persuasion, guilt arousal by relationally significant others also creates both guilt and resentment but can still enjoy Persuasive success, and nonobvious means of guilt arousal (as in the transgression-compliance or hypocrisy induction format) create guilt without much resentment and hence can have unalloyed Persuasive impact.

This analysis of the negative reactions attendant to guilt induction attempts suggests a possible difference between the use of aroused guilt and the use of anticipated guilt in the service of social influence. The use of anticipated guilt as an influence mechanism may be less susceptible to reactions such as resentment and anger precisely because the use of anticipated guilt does not require claiming that the influence target has already committed a transgression.

To bring out the difference here, consider the very different stances (identities or roles) available to the influencing agent when attempting to arouse guilt or when attempting to employ anticipated guilt as a mechanism of influence. In the case of anticipated guilt, the influencing agent can act almost as a coach or helper: “I want to help you think through what you want, help you consider how you will feel if you do or don’t do X, help you to see what actions will be consistent with yourself, or help you see what you really want to do here by encouraging you to contemplate how you will feel.” But in the case of guilt arousal, the influencer characteristically will have to adopt an oppositional accusatory stance: “You did this bad, self-inconsistent thing, and you should change or fix it.” The resentment that might naturally be generated by an accusatory stance is Perhaps less likely to be aroused when the influencer adopts a coaching role— and thus the use of anticipated guilt as an influence mechanism might not necessarily be subject to the same negative reactions sometimes aroused by the use of aroused guilt.

It plainly will be useful for future research to focus on the negative reactions to guilt induction attempts, especially as these appear to have the capacity to undermine the successful use of guilt as a mechanism of influence. Specification of exactly what negative reactions occur, identification of the specific features of guilt induction efforts that appear to evoke or minimize such reactions, information about the Persuasion-relevant consequences of the negative reactions, and clarification of the underlying mechanisms will be welcomed.

Magnitude of Effects. There is one other aspect of these research findings worth noticing, namely, the relative size of compliance effects observed in standard transgression-compliance formats and in DITF implementations. As discussed previously, these two lines of research are conceptually parallel; there is initially a transgression of some sort (e.g., a lie is told, a prosocial request is refused), and subsequently a helping opportunity arises (typically in the form of a request). But transgression-compliance studies have produced a significantly larger mean effect size (mean r = .28, k = 31, 95% CI bounds of .22 and .34 [O’Keefe, 2000]) than have DITF studies (mean r = .10, k = 88, 95% CI bounds of .05 and .14 [O’Keefe & Hale, 1998]). Given the apparent parallelism between the two circumstances, the difference in effect size wants explanation.

One straightforward possibility is that the kinds of “transgressions” common in DITF research formats (i.e., refusal of rather large prosocial requests) do not represent as great a threat to self-integrity, and hence do not generate as much guilt, as do the transgressions common in transgression-compliance research (e.g., lying or damaging someone’s equipment). Rationales may be ready at hand for why one might sensibly decline to volunteer to spend 2 hours a week for a minimum of 2 years working with disadvantaged youths, but not for why one broke someone’s camera. Thus, the need for self-affirmation may typically be greater in transgression-compliance studies than in DITF studies, resulting in larger compliance effects.

Research Priorities

A number of promising research questions have been mentioned in the preceding discussion. As a way of suggesting some priorities for research attention in this area—and a way of underscoring just how much remains to be learned about guilt as a social influence mechanism—these can usefully be collected under four broad research foci.

First, clarification is needed concerning how message variations produce differential guilt arousal. Although experimenters have proved capable of arousing different levels of guilt in message recipients, little attention has been given to careful description of the message features associated with such variation in effect. A complete understanding of the operation of guiltbased Persuasive appeals will require some account of how message variations are associated with guilt arousal variations. In the absence of such an account, we may find ourselves able to specify the Persuasive consequences of arousing guilt in receivers but unable to specify how to bring about guilt.

Second, clarification is needed of the negative reactions that appear capable of undermining guiltbased influence. Just what are the relevant negative reactions (e.g., anger, resentment, both, something else entirely)? What mechanisms (e.g., reactance, Perhaps arising from negative face threat) underlie these evoked reactions? Exactly how and why do these reactions undermine the success of guiltbased influence? Are there identifiable features of influence situations that affect the likelihood of these negative reactions? For example, is it in fact the case that the more apparent the influencer’s intention to arouse guilt, the more likely it is that the target will experience negative reactions?

Third, it will be useful to consider the possible roles of self-affirmation processes in guilt. As one example, there may be distinguishable varieties of guilt-evoking threats to self-integrity, with corresponding variation in compliance effects. As another example, the existence of alternative means of self-affirmation may create challenges to the successful use of guilt as an influence mechanism; a consumer advertisement that arouses guilt might encourage self-affirmation efforts of various sorts but not necessarily the specific course of action sought by the advertisement.

Finally, the role of anticipated guilt in social influence deserves greater research attention. Studies of the behavioral effects of other anticipated emotions plainly suggest the usefulness of such attention. But it is not known what message variations might dependably influence the anticipation of guilt, nor is it known whether anticipated guilt will provide the magnitude of influence potential that aroused guilt does, nor is it known whether (as speculated here) influence mechanisms based in anticipated guilt are less likely to evoke the negative reactions commonly attendant to guilt arousal mechanisms.


Guilt is plainly a potentially powerful mechanism of Persuasive influence, but many aspects of guiltbased social influence have received only slight research attention. In some ways, the inattention to guilt is unsurprising given that studies of communicative social influence have commonly focused on “rational” or “logical” aspects of influence (e.g., Fishbein & Ajzen, 1975). But continuing exploration of how aroused guilt and anticipated guilt operate as mechanisms of influence is surely justified. In particular, linking research on guiltbased influence to more general theoretical models of the self and its management appears to yield promising lines of inquiry. Placing guilt in the context of these broader processes helps to explain a number of findings concerning guiltbased influence (e.g., why transgression’s effects on compliance appear unaffected by what would seem to be plausible moderators, why guiltbased influence distinctively appears capable of evoking negative reactions) and offers the prospect of stitching together a number of previously unconnected lines of research.