Copyright © 2003 American Psychological Society

105

CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE

It is rare when a public serviceannouncement (PSA) is believedto have the sort of effectivenessachieved by the most successfulmass media commercial messages,which typically benefit from muchlarger production budgets andbroadcast frequencies. Yet there isone PSA that is regularly creditedas having such status. Called the“Iron Eyes Cody spot” (after theNative American actor who starredin it), it begins with a shot of astately, buckskin-clad American In-dian paddling his canoe up a riverthat carries various forms of indus-trial and individual pollution. Aftercoming ashore near the littered sideof a highway, the Indian watches as abag of garbage is thrown, splatteringand spreading along the road, fromthe window of a passing car. Fromthe refuse at his feet, the camerapans up slowly to the Indian’s face,where a tear is shown trackingdown his cheek, and the slogan ap-pears: “People Start Pollution, Peo-ple Can Stop It.”

Broadcast for many years in the1970s and 1980s, the spot won nu-merous awards and millions uponmillions of dollars of donated air-time. Indeed, it has even beennamed the 16th best televisioncommercial of all time by

TV Guide

magazine (“The Fifty Greatest,”1999). However, despite the fame ofthis touching piece of public ser-vice advertising, research suggeststhat it contains features that maybe less than optimal, and perhapseven negative, in their impact onthe littering actions of those whosee it. In addition to the laudablemessage in the ad urging viewersto stop littering, there is the under-

lying message, as well, that a lot ofpeople

do

litter: Debris floats on theriver, litter lies at the roadside,trash is tossed from an automobile.

DESCRIPTIVE VERSUS INJUNCTIVE NORMS

Thus, the creators of the Iron EyesCody spot may well have pitted twokinds of norms against one another,

injunctive norms

(involving percep-tions of which behaviors are typi-cally approved or disapproved) and

descriptive norms

(involving percep-tions of which behaviors are typi-cally performed). Much research in-dicates that both kinds of normsmotivate human action; people tendto do what is socially approved aswell as what is popular. The wisdomof setting these two kinds of motiva-tions in line with (rather than inopposition to) one another within acommunication has direct implica-tions for the development of pro-environmental messages. Experi-ences that focus individuals on theall-too-frequent occurrence of an of-fense against the environment havethe potential to increase the occur-rence of that offense.

An Initial Experiment

To explore this possibility as itapplies to individuals’ decisions todespoil the environment, my col-leagues and I have conducted a va-riety of studies over the past sev-eral years. In one investigation(Cialdini, Reno, & Kallgren, 1990,Experiment 1), participants weregiven the opportunity to litter (ahandbill they found on their carwindshields) into either a previouslyclean or a fully littered environmentafter first witnessing a confederatewho either dropped trash into theenvironment or simply walked

Crafting Normative Messages to Protect the Environment

Robert B. Cialdini

1

Department of Psychology, Arizona State University, Tempe, Arizona

Abstract

It is widely recognized thatcommunications that activate so-cial norms can be effective inproducing societally beneficialconduct. Not so well recognizedare the circumstances underwhich normative informationcan backfire to produce the op-posite of what a communicatorintends. There is an understand-able, but misguided, tendency totry to mobilize action against aproblem by depicting it as re-grettably frequent. Informationcampaigns emphasize that alco-hol and drug use is intolerablyhigh, that adolescent suiciderates are alarming, and—mostrelevant to this article—that ram-pant polluters are spoiling theenvironment. Although theseclaims may be both true and wellintentioned, the campaigns’ cre-ators have missed somethingcritically important: Within thestatement “Many people aredoing this

undesirable

thing”lurks the powerful and under-cutting normative message“Many people

are

doing this.”Only by aligning descriptivenorms (what people typically do)with injunctive norms (whatpeople typically approve ordisapprove) can one optimizethe power of normative ap-peals. Communicators who failto recognize the distinction be-tween these two types of normsimperil their persuasive efforts.

Keywords

norms; environment; public ser-vice announcements

106 VOLUME 12, NUMBER 4, AUGUST 2003

Published by Blackwell Publishing Inc.

through it. By varying the state ofthe environment (clean vs. littered),we sought to manipulate the per-ceived descriptive norm for litter-ing in the situation. By manipulat-ing whether the confederate drop-ped trash into the environment, wesought to differentially focus partici-pants’ attention on the state of theenvironment and, consequently, tomanipulate the salience of the per-ceived descriptive norm there (i.e.,what most people did).

We had three main predictions.First, we expected that participantswould be more likely to litter intoan already littered environmentthan into a clean one. Second, weexpected that participants who sawthe confederate drop trash into afully littered environment wouldbe most likely to litter there them-selves, because they would have hadtheir attention drawn to evidence ofa pro-littering descriptive norm—that is, to the fact that people typi-cally litter in that setting. Conversely,we anticipated that participants whosaw the confederate drop trash intoa clean environment would be leastlikely to litter there, because theywould have had their attentiondrawn to evidence of an anti-litter-ing descriptive norm—that is, to thefact that (except for the confederate)people typically do not litter in thatsetting. This last expectation distin-guished our normative accountfrom explanations based on simplemodeling processes in that we werepredicting decreased littering afterparticipants witnessed a model litter.

As can be seen in Figure 1, thedata supported our experimentalhypotheses. Overall, there was morelittering in the littered environmentthan in the clean environment. Inaddition, the most littering occurredwhen participants saw a model droptrash into a littered environment;and, most tellingly, the least litteringoccurred when participants saw amodel drop trash into a clean envi-ronment.

Rethinking the Iron EyesCody PSA

At this point, it is appropriate tolook back at the Iron Eyes CodyPSA, as the findings of our studypoint to reasons for concern aboutthe effectiveness of that ad. Recallthat it depicts a character whosheds a tear after encountering anarray of litter. No doubt the tear isa powerful reminder of the injunc-tive norm against littering in U.S.culture. But accompanying the ben-eficial reminder is the potentiallydamaging message that many peo-

ple

do

litter. Thus, the resultant im-pact of the injunctive norm againstlittering may be undermined by theunintended presentation of a de-scriptive norm for littering. More-over, that presentation occurs in away that, according to the results ofour study, may be especially dam-aging. That is, the creators of the ad

seem to have been correct in theirdecision to show a dismaying in-stance of someone (the passing mo-torist) actively littering the environ-ment; but they may have beenmistaken in their decision to use analready-littered environment, as theobservation of another person litter-ing into a littered environment pro-duced the greatest littering in ourstudy. In contrast, the combinationof a (single) litterer and an other-wise clean environment generatedthe least littering from our partici-pants.

Were we to suggest a revision ofthe Iron Eyes Cody PSA, then, itwould be to make the procedurallysmall but conceptually meaningfulmodification of changing the de-picted environment from trashed toclean—and thereby changing theperceived descriptive norm regard-ing littering. Then, when the disap-proving tear appeared, viewers

Fig. 1. Percentage of participants littering as a function of the salience of the descrip-tive norm and the state of the environment.

Copyright © 2003 American Psychological Society

CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 107

would be exposed to injunctiveand descriptive norms guiding be-havior in the same direction.

ENVIRONMENTAL THEFT

In situations already character-ized by high levels of socially cen-sured conduct, the distinction be-tween descriptive and injunctivenorms offers a clear implication: It isa serious error to focus an audienceon the descriptive norm (i.e., whatis done in those situations); instead,public service messages should focusthe audience on the injunctive norm(i.e., what is approved or disap-proved in those situations). Take,for instance, the case of Arizona’sPetrified Forest National Park,which suffers from the estimatedtheft of more than a ton of woodper month by visitors. New arriv-als quickly learn of the past thieveryfrom prominently placed signage:“Your heritage is being vandalizedevery day by theft losses of petrifiedwood of 14 tons a year, mostly asmall piece at a time.”

Although it is understandablethat park officials would want to in-stigate corrective action by describ-ing the dismaying size of the prob-lem, such a message ought to be farfrom optimal. According to an in-formed normative account, it wouldbe better to design park signage tofocus visitors on the social disap-proval (rather than the harmfulprevalence) of environmental theft.Recently, my colleagues and I soughtto examine this hypothesis—that ina situation characterized by unfortu-nate levels of socially disapprovedconduct, a message that focuses re-cipients on the injunctive norm willbe superior to messages that focus re-cipients on the descriptive norm(Cialdini et al., 2003).

To test our expectation, we gainedpermission from Petrified Forest Na-tional Park officials to place secretlymarked pieces of petrified wood

along visitor pathways. During fiveconsecutive weekends, at the en-trance to each path, we displayed sig-nage that emphasized either descrip-tive or injunctive norms regardingthe theft of petrified wood from thepark. The descriptive-norm signstated, “Many past visitors have re-moved petrified wood from thePark, changing the natural state ofthe Petrified Forest.” This wordingwas accompanied by pictures ofthree visitors taking wood. In con-trast, the injunctive-norm signstated, “Please don’t remove thepetrified wood from the Park, in or-der to preserve the natural state ofthe Petrified Forest.” This wordingwas accompanied by a picture of alone visitor stealing a piece ofwood, with a red circle-and-barsymbol superimposed over his hand.Our measure of message effective-ness was the percentage of marked

pieces of wood stolen over the5-week duration of the study. As pre-dicted, the descriptive-norm messageresulted in significantly more theftthan the injunctive-norm message(7.92% vs. 1.67%).

2

RECYCLING

Should one conclude from theseresults that highlighting descriptivenorms is always likely to be a coun-terproductive tactic in environmen-tal information campaigns? No. Al-though highlighting descriptivenorms is detrimental when environ-mentally harmful behavior is preva-lent, this approach should be effec-tive when the prevalent behavior isenvironmentally beneficial. For ex-ample, if the majority of citizens con-serve energy at home, campaign de-velopers would be well advised toinclude such descriptive normativeinformation in their presentations in-tended to increase residential energyconservation. Of course, if the major-ity of citizens also approve of such

efforts, the campaign developerswould be wise to incorporate thisinjunctive normative informationas well.

Thus, the most effective norm-based persuasive approach underthese circumstances would be onethat enlists the conjoint influence ofdescriptive and injunctive norms.To examine the impact of an infor-mation campaign that combined theinfluence of injunctive and descrip-tive norms, my colleagues and I cre-ated three PSAs designed to in-crease recycling, an activity that wasboth performed and approved bythe majority of local residents in ourstudy area. Each PSA portrayed ascene in which the majority of de-picted individuals engaged in recy-cling, spoke approvingly of it, andspoke disparagingly of a single in-dividual in the scene who failed torecycle. When, in a field test, thesePSAs were played on the local TVand radio stations of four Arizonacommunities, a 25.35% net advan-tage in recycling tonnage was re-corded over a pair of control com-munities not exposed to the PSAs.

Although a 25% recycling ad-vantage is impressive from a prac-tical standpoint, that study did notallow for confident theoretical con-clusions about the causes of the ad-vantage. For instance, it was notpossible to determine the extent towhich our PSAs may have been ef-fective because of their normativeelements. After all, it is conceivablethat the PSAs were successful be-cause they included humorous andinformational components unre-lated to norms. In order to assesswhether and to what degree de-scriptive and injunctive norms—separately and in combination—contributed to the messages’ effec-tiveness, additional evidence wasnecessary. To that end, we con-ducted a study in which collegestudents viewed our three recy-cling PSAs and rated their impactalong several relevant dimensions(Cialdini et al., 2003).

108 VOLUME 12, NUMBER 4, AUGUST 2003

Published by Blackwell Publishing Inc.

That study was designed to de-termine whether our PSAs had theintended effect of conveying toviewers that recycling was preva-lent (descriptive norm) and ap-proved (injunctive norm), whetherthese perceived norms influencedviewers’ intentions to recycle, andwhether the two types of normsoperated similarly or differently toaffect recycling intentions. A statis-tical analysis of the results indi-cated that both normative and non-normative factors influenced theintent to recycle (see Fig. 2). Ofcourse, the finding that nonnorma-tive factors (prior attitude, new in-formation, humor) had causal im-pact is not incompatible with ourtheoretical position, as we certainlywould not claim that normative

factors are the only motivators ofhuman responding.

At the same time, it is encourag-ing from our theoretical perspec-tive that both injunctive and de-scriptive normative informationsignificantly influenced recyclingintentions. That is, as a result ofviewing the ads, the more partici-pants came to believe that recy-cling was (a) approved and (b)prevalent, the more they plannedto recycle in the future. It is note-worthy that, despite a strong corre-

lation (

r

�

.79) between partici-pants’ perceptions of the existingprevalence and approval of recy-cling, these two sources of motiva-tion had independent effects on re-cycling intentions. Such resultsaffirm the theoretical distinction

between descriptive and injunc-tive norms.

CONCLUSION

Public service communicatorsshould avoid the tendency to sendthe normatively muddled messagethat a targeted activity is sociallydisapproved but widespread. Norm-based persuasive communicationsare likely to have their best effectswhen communicators align de-scriptive and injunctive normativemessages to work in tandem ratherthan in competition with one an-other. Such a line of attack unitesthe power of two independentsources of normative motivation

Fig. 2. Impact of public service announcements intended to promote recycling. The arrows in thediagram depict the pathways through which viewers’ attitudes and perceptions affected their in-tentions to recycle. Alongside each arrow is the corresponding path coefficient, a measure ofcausal impact; all the path coefficients shown are significant at p � .05.

Copyright © 2003 American Psychological Society

CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 109

and can provide a highly success-ful approach to social influence.

At the same time, certain issuesremain to be clarified if communi-cators are to optimize the impact ofnorm-based messages. The firstconcerns the nature of the psycho-logical mechanisms that underliedescriptive and injunctive norms.The results of our last study sug-gest an intriguing difference be-tween them. Information about so-cial approval or disapproval affectedrecycling intentions by influencingassessments of the ads’ persuasive-ness (see Fig. 2). Information aboutrelative prevalence, in contrast, in-fluenced intentions directly, with-out affecting the perceived persua-siveness of the ads. Why shouldthat be the case? One possibility isthat because descriptive norms arebased in the raw behavior of otherindividuals, it is relatively easy toaccommodate to such norms with-out much cognitive analysis. In-deed, organisms with little cogni-tive capacity do so: Birds flock, fishschool, and social insects swarm.Injunctive norms, however, arebased in an understanding of themoral rules of the society (i.e., whatother people are likely to approve),and should therefore require morecognitive analysis to operate suc-cessfully. Hence, one might expectthat the impact of injunctive (but

not descriptive) normative informa-tion would be mediated throughcognitive assessments of the qual-ity or persuasiveness of the norma-tive information. Additional workis necessary to test this possibility.

A second important research is-sue concerns the problem of dimin-ished salience of the normativemessage at the time when a tar-geted behavior is likely to be per-formed. Often, the message is nolonger present when the desiredbehavior must take place. For ex-ample, PSAs are typically radio,television, and print communica-tions that are encountered at timesfar removed from the opportuni-ties to perform the socially desir-able actions that the PSAs promote.A crucial question to be answeredby future investigation is howcommunicators can structure theirmessages to maximize the likeli-hood that the motivational compo-nents of those messages will be sa-l i en t a t the t ime for ac t ion .Research that identifies persuasiveor mnemonic devices for achiev-ing this goal will be of immensebenefit to public service communi-cation efforts.

Recommended Reading

Bator, R.J., & Cialdini, R.B. (2000).The application of persuasion the-

ory to the development of effec-tive pro-environmental publicservice announcements.

Journal ofSocial Issues

,

56

, 527–541.Kallgren, C.A., Reno, R.R., & Cial-

dini, R.B. (2000). A focus theory ofnormative conduct: When normsdo and do not affect behavior.

Per-sonality and Social Psychology Bulle-tin

,

26

, 1002–1012.Schultz, P.W. (1999). Changing behav-

ior with normative feedback inter-ventions: A field experiment oncurbside recycling.

Basic and Ap-plied Social Psychology

,

21

, 25–38.

Notes

1. Address correspondence to Rob-ert B. Cialdini, Department of Psychol-ogy, Arizona State University, Tempe,AZ 85287-1104; e-mail: robert.cialdini@asu.edu.

2. These data are best understoodin the context of previous research in-dicating that the ratio of thefts to parkvisitors falls just under 3%.

References

Cialdini, R.B., Barrett, D.W., Bator, R., Demaine,L.J., Sagarin, B.J., Rhoads, K.v.L., & Winter,P.L. (2003).

Activating and aligning social normsfor persuasive impact

. Manuscript submitted forpublication.

Cialdini, R.B, Reno, R.R., & Kallgren, C.A. (1990).A focus theory of normative conduct: Recy-cling the concept of norms to reduce litteringin public places.

Journal of Personality and SocialPsychology

,

58

, 1015–1026.The fifty greatest TV commercials of all time.

(1999, July 3–9).

TV Guide

, pp. 2–34.