Following What People Think We Should Do Versus What People Actually Do

Distinguishing Between Descriptive and Injunctive Norms

When using normative explanations for behavior, theorists have often distinguished between at least two general categories of norms that serve as useful guidelines. Injunctive norms reflect behavior that most people perceive to be correct, whereas descriptive norms reflect behavior in which most people engage in a situation, regardless of the injunctive norms. Essentially, injunctive norms are what one “ought” to do, while descriptive norms are based on what others are actually doing at the time (Cialdini & Trost, 1998). An example of these different norms in action would be the issue of littering; most people believe that we should not litter, that it is bad for the environment, and that it is socially reprehensible. This is the injunctive norm at work. However, a look at any park or public space in a large city would tell us that littering is often a problem. Because we see others littering, and litter is already present in public spaces, in many contexts littering behavior seems normative at the descriptive level.

A substantial amount of empirical evidence has accumulated, indicating that both of these types of norms can have a substantial impact on behavior, sometimes in divergent and surprising ways (Cialdini, 2003; Cialdini et al., 1990, 2003; Goldstein, Cialdini, & Griskevicius, 2008; Jacobson, Mortensen, & Cialdini, 2011). Thus, there is clear evidence that is meaningful to distinguish between descriptive and injunctive norms and that both can influence behaviors and intentions. However, one issue that has received comparatively little attention in this literature is the conditions under which one of these types of norms might be expected to exert more influence than the other. Such a question is obviously of theoretical interest, in that understanding the moderators of the impact of different types of norms can provide greater insight into the underlying mechanisms responsible for their effects. Moreover, such a question also has applied relevance, given that in some behavioral contexts these norms might be in conflict and thus it could be difficult to construct messages that effectively use both norms.

Elaboration as a Moderator of the Impact of Norms

When considering potential moderators of the influence of descriptive and injunctive norms in compliance to persuasive messages, one candidate suggested by past theorizing is the extent to which people have to carefully think about (i.e., cognitively elaborate) the message conveying the norms. Theories such as the Elaboration Likelihood Model ([ELM] Petty & Cacioppo, 1986a, 1986b; Petty & Wegener, 1999) and the Heuristic-Systematic Model ([HSM] see Chaiken, 1987; Chaiken, Liberman, & Eagly, 1989; Chen & Chaiken, 1999) have demonstrated that variables affecting persuasion and social judgments can have different effects when information is extensively elaborated versus processed only superficially.

Interestingly, Cialdini (2003) has speculated that descriptive norms may require less effortful cognitive processing than injunctive norms. In the case of descriptive norms, one simply has to look to those around one’s self to judge appropriate behavior, whereas injunctive norms might require some understanding of the morals of a group in order to act upon injunctive norms, and this would require more cognitive elaboration. If these speculations are correct, one would predict that in nonthoughtful conditions descriptive messages should be more effective than injunctive messages. Conversely, when people engage in extensive consideration of a message, the advantage of descriptive norms over injunctive norms might be expected to decrease and perhaps the relative efficacy of these messages might reverse.

To date, these predictions have never been directly tested. Moreover, such a set of predictions might be too simple in that it is possible that the effects of these two types of messages, as well as their efficacy under different levels of elaboration, may be conditional on characteristics of the message recipient. That is, perhaps the content of descriptive messages could coordinate well with certain personal tendencies, whereas injunctive messages could coordinate well with others. For example, the personality literature has identified several traits that reflect in various ways the extent to which people look to their social environment for guides to their behaviors or rely on internal cues such as values and attitudes.

Two of the most widely studied of these types of traits are extraversion and self-monitoring. Individuals who score high on the trait of extraversion tend to be characterized by concern with, and attention to, things outside the self; with good social skills and numerous friendships (McCrae & Costa, 1999). Conversely, those scoring low on the extraversion scale tend to direct their attention inward and are more focused on their own thoughts and feelings rather than outward experiences. Likewise, self-monitoring is a personality trait that relates to self-presentation (Snyder & Gangestad, 1986). Individuals who are high self-monitors tend to adapt their responses to social cues readily and are highly keen to cues of others and the situation. In contrast, low self-monitors are more likely to express themselves based on their inner attitudes, thoughts, and feelings and are less influenced by external cues and the desire to be viewed in a certain way.

It is plausible that the effects of descriptive norm and injunctive norm messages might vary across these personality traits. For instance, one might expect descriptive messages to be especially effective for people who are dispositionally inclined to look to other people as a guide to their actions (e.g., extraverts and high self-monitors), whereas people who do not naturally look to others as a guide to their actions (e.g., introverts and low self-monitors) might not be particularly influenced by such messages. Likewise, injunctive messages, with their focus on values, might be especially compelling for people who rely on internal cues such as values, but less effective for people who are externally focused.

At a more complex level, these traits might also interact with level of elaboration to moderate the impact of descriptive and injunctive messages. For example, it could be that descriptive norms are, as speculated, especially effective under low elaboration because they provide an easy cue for how to act. However, the heuristic that others’ behaviors are a valid guide to action might be much more strongly held by extraverts and/or high self-monitors than introverts and/or low self-monitors. Thus, the advantage of descriptive messages over injunctive messages under low elaboration might be expected to be particularly pronounced for extraverts and/or high self-monitors. Conversely, perhaps injunctive norms are more successful than descriptive norms under high elaboration, but it may be that this advantage is most evident for people who are likely to see value-based appeals as central and compelling argument for action. Thus, the greater efficacy of injunctive messages over descriptive messages in high-elaboration conditions might emerge most strongly for introverts and/or low self-monitors. The objective of the present research was to explore the impact of descriptive and injunctive messages on behavioral intentions under high and low elaboration as well as to examine whether these effects differ as a function of the personality traits of extraversion and self-monitoring.

Method

Results

Behavioral Intentions

To test our key hypotheses regarding the impact of type of message, elaboration, extraversion, and self-monitoring, a 2 (Elaboration: High vs. Low) × 2 (Message Type: Descriptive vs. Injunctive) General Linear Model with mean-centered extraversion and self-monitoring included as continuous predictor variables and intentions as the dependent variable was specified. The model also included all possible interactions among the two manipulated variables and the two personality trait variables as predictors in the model.

None of our speculations explicitly suggested the emergence of any main effects. This expectation was largely supported, as three of the possible main effects failed to reach significance. However, there was a significant main effect of extraversion on intent to join the program, F(1, 230) = 7.81, p = .01. An examination of the correlation between extraversion and intentions revealed that increased extraversion was associated with higher intentions to participate in the program (r = .11).

As noted earlier, Cialdini’s (2003) speculations imply that a 2-way interaction between Message Type and Elaboration should occur, such that descriptive messages should be more persuasive than injunctive messages under conditions of low elaboration because they require less cognitive processing. What is less clear from this view is how descriptive and injunctive messages should perform relative to one another under high elaboration. There are two plausible patterns either of which would be consistent with this perspective. First, it is possible that descriptive and injunctive messages could be equally persuasive under conditions of high elaboration. Alternatively, a reversal could occur such that in conditions of high elaboration, the injunctive message is actually more persuasive than the descriptive message. Interestingly, consistent with the Cialdini (2003) perspective, a significant 2-way interaction did emerge, F(1, 230) = 6.53, p = .01. The means associated with this interaction are presented in Figure 1 . As can be seen on the left of Figure 1, as predicted, the descriptive message led to greater intentions to join (M = 3.86, CI_.95 = [3.61, 4.12]) than the injunctive message (M =3.48, CI_.95 = [3.20, 3.76]) when elaboration was low. Interestingly, under high elaboration, the injunctive message (M = 3.67, CI_.95 = [3.41, 3.94]) was significantly more effective than the descriptive message (M = 3.36, CI_.95 = [3.09, 3.63]).

OPEN IN VIEWER

Figure 1.

Two-way interaction between message type and elaboration on participant intent to join program.

Also of interest was a significant 2-way interaction between message type and self-monitoring F(1, 230) = 5.04, p = .03. To further clarify this interaction, we computed the simple slopes reflecting the association between self-monitoring and intentions under the two message conditions. These analyses indicated that for injunctive messages, increases in self-monitoring were associated with lower intentions to participate in the program (b = −.06, p = .03). Conversely, for the descriptive message, there was a nonsignificant positive association between self-monitoring and intentions (b = .03, p = .29).

No other effects in the model reached statistical significance. Of particular interest was the nonsignificant interaction between extraversion, message type, and elaboration, F(1, 230) = .26, p = .61. Equally important, the interaction between Self-Monitoring, Message Type, and Elaboration was also not significant, F(1, 230) = 2.60, p = .11. Thus, the moderating effects of elaboration on the relative efficacy of the types of messages were not qualified by either extraversion or self-monitoring.

General Discussion

The purpose of the present research was to examine whether the impact of descriptive versus injunctive normative messages differed depending on the level of cognitive elaboration, extraversion, and self-monitoring. The emergence of a significant Message Type × Elaboration Interaction and the fact that this 2-way interaction was not qualified by any higher-order interactions fits well with Cialdini’s (2003) speculations regarding the role of elaboration as a moderator of the relative efficacy of normative messages. Thus, the data in the present research are probably the most direct evidence to date for Cialdini’s (2003) suggestion that descriptive norms require less processing than injunctive norms in order to exert their impact. Extending his speculations, the present experiment suggested that, under conditions of high elaboration, intent to join the program was higher for those who read the injunctive rather than the descriptive message. As was noted earlier, Cialdini’s (2003) prior theorizing never explicitly addressed whether injunctive messages should be superior to descriptive messages under high elaboration or that the advantage of descriptive messages over injunctive should simply decrease or disappear.

Of course, one issue that the present study cannot definitively address is why descriptive norms work better than injunctive norms under low elaboration. It is unlikely that the present results can be attributed to the complexity of the two messages, because these were carefully constructed and edited to be of near identical length and layout, with similar language and phrasing used throughout. Another possible explanation could be that something idiosyncratic to our specific manipulations of elaboration produced the outcome. We used a combination of two common methods of manipulating elaboration: personal involvement and a cognitive load task. Our manipulation checks suggested that these manipulations created differences in elaboration. However, it is possible they also influenced other processes. Most notably, one might argue that our manipulation of personal involvement (i.e., whether the program was planned for immediate implementation at the participant’s university or another university) also altered the social rewards/punishments of adhering to the norm. Injunctive norms in particular have been postulated to play an important role because of their implications for social rewards and punishments. Perhaps the high involvement condition produced greater sensitivity to social rewards/punishments because the peer group was more immediate to the participant.

Although plausible, there are at least two reasons to doubt this explanation. First, normative impact as a result of social rewards/punishments tends to be most pronounced in public settings where adherence to the norm can be directly assessed by others (e.g., Deutsch & Gerard, 1955). The present study involved behavioral intention measures that were completed in an anonymous setting and for a program where adherence would be difficult for peers to assess. Moreover, failure to participate did not involve a violation of the injunctive norm per se (a healthy life style) and indeed adherence to norm in the present context could be achieved by many behaviors in addition to the advocated program. Thus, concrete social rewards/ punishments in either condition do not seem to be especially likely. A second objection to this explanation for the present results is that it cannot really account for the effect of elaboration on the descriptive message, which was, if anything, more pronounced than that on the injunctive message.

Thus, we think the most likely reason for the present effects is that descriptive norms require less scrutiny and so when cognitive resources are limited, looking to others’ behavior as a guide is a quick heuristic to make a decision. In contrast, Kallgren, Reno, and Cialdini (2000) have found that personal norms (injunctive norms related to individual values) are weak predictors of behavior unless they are at the forefront of one’s attention when the norm-relevant behavior arises. Therefore, conditions of high elaboration would be the ideal for injunctive norms to be effective. Further, because injunctive norms deal with values, it is probable that the individual must actively cognitively process these messages to assess whether they fit with their own values.

Assuming this interpretation is correct, the present study in many respects complements recent findings on other moderators of normative messages. Jacobson, Mortensen, and Cialdini (2011) found in two studies that when people engaged in an ego-depletion task prior to exposure to very brief normative messages, descriptive messages were more effective than injunctive messages. In contrast, people who were not ego-depleted prior to exposure showed a reverse pattern of effects. Interestingly, past research has suggested that ego-depletion can lead to decreased cognitive elaboration of messages (Wheeler, Brinol, & Hermann, 2007). This previous finding along with those of the present study and Cialdini’s (2003) earlier speculations raise the possibility that elaboration might have been the proximal causal mechanism underlying the effects of ego-depletion on normative messages in the Jacobson et al. (2011) studies. However, no measures of elaboration were included in these ego-depletion studies so it is impossible to know whether elaboration was altered, and of course, it is possible that ego-depletion could have altered other mechanisms responsible for the effects.

Conversely, one might argue the effects of our elaboration manipulation were caused by ego-depletion. However, given that ego-depletion is known to reduce cognitive elaboration and cognitive elaboration was clearly altered in our study, such an alternative interpretation would only be different from our present interpretation if the effects occurred as a result of some unmeasured mechanism that was independent of the observed differences in elaboration. Additionally, given that the elaboration manipulation occurred during exposure to the message rather than prior to message exposure (as is typically done in ego-depletion research), one must assume ego-depletion occurred sufficiently rapidly during exposure to have its effects.

A final point of interest worth noting were the results involving extraversion and self-monitoring. Neither of these traits moderated the interaction between message type and elaboration, suggesting that the relative advantages of the messages across elaboration conditions were robust across variations in these traits. That being said, these traits did produce some effects of interest. A main effect of extraversion suggested that increases in this trait were associated with greater intentions to participate. Such an effect suggests that extraverts simply found the general program more appealing or that extraverts are more responsive to normative messages of both types, which makes sense in that both types of normative messages are a form of social validation. Interestingly, self-monitoring produced no main effect but interacted with message type. This interaction suggested that low self-monitors were more influenced than high self-monitors by the injunctive message and that there was a nonsignificant reversal of this effect for the descriptive message. Such an effect demonstrates yet another moderator of the efficacy of different types of normative messages and is sensible in light of past research suggesting that low self-monitors are more likely to respond to internalized guides to behavior such as values that are stressed in injunctive messages.

Taken together, at the most basic level, these findings further highlight both the theoretical and applied value of distinguishing between descriptive and injunctive normative messages. By demonstrating that the impact of elaboration on the efficacy of these two types of messages is quite different, the present study serves to confirm past theorizing that these messages exert their effects via different psychological processes. At the applied level, this key finding suggests that when attempting to shape behavior, practitioners should carefully consider the likely level of elaboration that will occur in the targets of their messages when deciding whether to stress injunctive or descriptive norms in their messages. Thus, it is hoped that this study can serve as a foundation for further theoretical and applied investigations of the effects of descriptive and injunctive norms.