Journal of Personality and Social Psychology1973, Vol. 27, No. J, 100-108

"FROM JERUSALEM TO JERICHO":A STUDY OF SITUATIONAL AND DISPOSITIONAL VARIABLES IN

HELPING BEHAVIOR1

JOHN M. DARLEY 2 AND C. DANIEL BATSON

Princeton University

The influence of several situational and personality variables on helping behav-ior was examined in an emergency situation suggested by the parable of theGood Samaritan. People going between two buildings encountered a shabbilydressed person slumped by the side of the road. Subjects in a hurry to reachtheir destination were more likely to pass by without stopping. Some subjectswere going to give a short talk on the parable of the Good Samaritan, otherson a nonhelping relevant topic; this made no significant difference in thelikelihood of their giving the victim help. Religious personality variables didnot predict whether an individual would help the victim or not. However,if a subject did stop to offer help, the character of the helping response wasrelated to his type of religiosity.

Helping other people in distress is, amongother things, an ethical act. That is, it is anact governed by ethical norms and preceptstaught to children at home, in school, and inchurch. From Freudian and other personalitytheories, one would expect individual differ-ences in internalization of these standardsthat would lead to differences between indi-viduals in the likelihood with which theywould help others. But recent research onbystander intervention in emergency situa-tions (Bickman, 1969; Darley & Latane,1968; Korte, 1969; but see also Schwartz &Clausen, 1970) has had bad luck in findingpersonality determinants of helping behavior.Although personality variables that one mightexpect to correlate with helping behavior havebeen measured (Machiavellianism, authoritar-ianism, social desirability, alienation, and so-cial responsibility), these were not predictiveof helping. Nor was this due to a generalizedlack of predictability in the helping situa-tion examined, since variations in the experi-mental situation, such as the availability ofother people who might also help, producedmarked changes in rates of helping behavior.

1 For assistance in conducting this research thanksare due Robert Wells, Beverly Fisher, MikeShafto, Peter Sheras, Richard Detweiler, and KarenGlasser. The research was funded by National ScienceFoundation Grant GS-2293.

2 Requests for reprints should be sent to JohnDarley, Department of Psychology, Princeton Uni-versity, Princeton, New Jersey 08540.

These findings are reminiscent of Hartshorneand May's (1928) discovery that resistanceto temptation, another ethically relevant act,did not seem to be a fixed characteristic ofan individual. That is, a person who waslikely to be honest in one situation was notparticularly likely to be honest in the next(but see also Burton, 1963).

The rather disappointing correlation be-tween the social psychologist's traditional setof personality variables and helping behaviorin emergency situations suggests the need fora fresh perspective on possible predictors ofhelping and possible situations in which totest them. Therefore, for inspiration weturned to the Bible, to what is perhaps theclassical helping story in the Judeo-Christiantradition, the parable of the Good Samaritan.The parable proved of value in suggestingboth personality and situational variablesrelevant to helping.

"And who is my neighbor?" Jesus replied, "Aman was going down from Jerusalem to Jericho,and he fell among robbers, who stripped him andbeat him, and departed, leaving him half dead.Now by chance a priest was going down the road;and when he saw him he passed by on the otherside. So likewise a Levite, when he came to theplace and saw him, passed by on the other side.But a Samaritan, as he journeyed, came to wherehe was; and when he saw him, he had compassion,and went to him and bound his wounds, pouringon oil and wine; then he set him on his own beastand brought him to an inn, and took care of him.And the next day he took out two dennarii and

100

SlTUATIONAL AND DlSPOSITIONAL VARIABLES IN HELPING BEHAVIOR 101

gave them to the innkeeper, saying, "Take careof him; and whatever more you spend, I willrepay you when I come back." Which of thesethree, do you think, proved neighbor to him whofell among the robbers? He said, "The one whoshowed mercy on him." And Jesus said to him,"Go and do likewise." [Luke 10: 29-37 RSV]

To psychologists who reflect on the para-ble, it seems to suggest situational and per-sonality differences between the nonhelpfulpriest and Levite and the helpful Samaritan.What might each have been thinking anddoing when he came upon the robbery victimon that desolate road? What sort of personswere they?

One can speculate on differences in thought.Both the priest and the Levite were religiousfunctionaries who could be expected to havetheir minds occupied with religious matters.The priest's role in religious activities is obvi-ous. The Levite's role, although less obvious,is equally important: The Levites were neces-sary participants in temple ceremonies. Muchless can be said with any confidence aboutwhat the Samaritan might have been think-ing, but, in contrast to the others, it wasmost likely not of a religious nature, forSamaritans were religious outcasts.

Not only was the Samaritan most likelythinking about more mundane matters thanthe priest and Levite, but, because he wassocially less important, it seems likely that hewas operating on a quite different time sched-ule. One can imagine the priest and Levite,prominent public figures, hurrying along withlittle black books full of meetings and ap-pointments, glancing furtively at their sun-dials. In contrast, the Samaritan would likelyhave far fewer and less important peoplecounting on him to be at a particular placeat a particular time, and therefore might beexpected to be in less of a hurry than theprominent priest or Levite.

In addition to these situational variables,one finds personality factors suggested aswell. Central among these, and apparentlybasic to the point that Jesus was trying tomake, is a distinction between types of reli-giosity. Both the priest and Levite are ex-tremely "religious." But it seems to be pre-cisely their type of religiosity that the parablechallenges. At issue is the motivation for one's

religion and ethical behavior. Jesus seems tofeel that the religious leaders of his time,though certainly respected and upstandingcitizens, may be "virtuous" for what it willget them, both in terms of the admiration oftheir fellowmen and in the eyes of God. NewTestament scholar R. W. Funk (1966) notedthat the Samaritan is at the other end ofthe spectrum:The Samaritan does not love with side glances atGod. The need of neighbor alone is made self-evident, and the Samaritan responds without othermotivation [pp. 218-219].

That is, the Samaritan is interpreted as resspending spontaneously to the situation, notas being preoccupied with the abstract ethicalor organizational do's and don'ts of religionas the priest and Levite would seem to be.This is not to say that the Samaritan is por-trayed as irreligious. A major intent of theparable would seem to be to present theSamaritan as a religious and ethical example,but at the same time to contrast his type ofreligiosity with the more common conceptionof religiosity that the priest and Leviterepresent.

To summarize the variables suggested asaffecting helping behavior by the parable, thesituational variables include the content ofone's thinking and the amount of hurry inone's journey. The major dispositional varia-ble seems to be differing types of religiosity.Certainly these variables do not exhaust thelist that could be elicited from the parable,,but they do suggest several research hypothe-ses.

Hypothesis 1. The parable implies thatpeople who encounter a situation possiblycalling for a helping response while thinkingreligious and ethical thoughts will be no morelikely to offer aid than persons thinking aboutsomething else. Such a hypothesis seems torun counter to a theory that focuses on normsas determining helping behavior because anormative account would predict that the in-creased salience of helping norms producedby thinking about religious and ethical ex-amples would increase helping behavior.

Hypothesis 2. Persons encountering a pos-sible helping situation when they are in a

102 JOHN M. BARLEY AND C. DANIEL BATSON

hurry will be less likely to offer aid thanpersons not in a hurry.

Hypothesis 3. Concerning types of religios-ity, persons who are religious in a Samaritan-like fashion will help more frequently thanthose religious in a priest or Levite fashion.

Obviously, this last hypothesis is hardlyoperationalized as stated. Prior research byone of the investigators on types of religios-ity (Batson, 1971), however, led us to dif-ferentiate three distinct ways of being reli-gious: (a) for what it will gain one (cf.Freud, 1927, and perhaps the priest andLevite), (b) for its own intrinsic value (cf.Allport & Ross, 1967), and (c) as a responseto and quest for meaning in one's everydaylife (cf. Batson, 1971). Both of the latterconceptions would be proposed by their ex-ponents as related to the more Samaritanlike"true" religiosity. Therefore, depending onthe theorist one follows, the third hypothesismay be stated like this: People (a) who arereligious for intrinsic reasons (Allport & Ross,1967) or (b) whose religion emerges out ofquestioning the meaning of their everydaylives (Batson, 1971) will be more likely tostop to offer help to the victim.

The parable of the Good Samaritan alsosuggested how we would measure people'shelping behavior—their response to a strangerslumped by the side of one's path. The victimshould appear somewhat ambiguous—ill-dressed, possibly in need of help, but also pos-sibly drunk or even potentially dangerous. '

Further, the parable suggests a means bywhich the incident could be perceived as areal one rather than part of a psychologicalexperiment in which one's behavior was undersurveillance and might be shaped by demandcharacteristics (Orne, 1962), evaluation ap-prehension (Rosenberg, 1965), or other po-tentially artifactual determinants of helpingbehavior. The victim should be encounterednot in the experimental context but on theroad between various tasks.

METHODIn order to examine the influence of these variables

on helping behavior, seminary students were askedto participate in a study on religious education andvocations. In the first testing session, personalityquestionnaires concerning types of religiosity were

administered. In a second individual session, thesubject began experimental procedures in one build-ing and was asked to report to another building forlater procedures. While in transit, the subject passeda slumped "victim" planted in an alleyway. Thedependent variable was whether and how the sub-ject helped the victim. The independent variableswere the degree to which the subject was told tohurry in reaching the other building and the talkhe was to give when he arrived there. Some sub-jects were to give a talk on the jobs in whichseminary students would be most effective, others,on the parable of the Good Samaritan.

SubjectsThe subjects for the questionnaire administration

were 67 students at Princeton Theological Seminary.Forty-seven of them, those who could be reached bytelephone, were scheduled for the experiment. Ofthe 47, 7 subjects' data were not included in theanalyses—3 because of contamination of the experi-mental procedures during their testing and 4 due tosuspicion of the experimental situation. Each subjectwas paid $1 for the questionnaire session and $1.50for the experimental session.

Personality MeasuresDetailed discussion of the personality scales used

may be found elsewhere (Batson, 1971), so thepresent discussion will be brief. The generalpersonality construct under examination was re-ligiosity. Various conceptions of religiosity havebeen offered in recent years based on differentpsychometric scales. The conception seeming to gen-erate the most interest is the Allport and Ross (1967)distinction between "intrinsic" versus "extrinsic"religiosity (cf. also Allen & Spilka, 1967, on "com-mitted" versus "consensual" religion). This bipolarconception of religiosity has been questioned byBrown (1964) and Batson (1971), who suggestedthree-dimensional analyses instead. Therefore, in thepresent research, types of religiosity were measuredwith three instruments which together provided sixseparate scales: (a) a doctrinal orthodoxy (D-0)scale patterned after that used by Clock and Stark(1966), scaling agreement with classic doctrines ofProtestant theology; (b) the Allport-Ross extrinsic(AR-E) scale, measuring the use of religion as ameans to an end rather than as an end in itself;(c) the Allport-Ross intrinsic (AR-I) scale, mea-suring the use of religion as an end in itself; (d)the extrinsic external scale of Batson's Religious LifeInventory (RELI-EE), designed to measure the in-fluence of significant others and situations in gen-erating one's religiosity; (e) the extrinsic internalscale of the Religious Life Inventory (RELI-EI),designed to measure the degree of "driveness" inone's religiosity; and (/) the intrinsic scale of theReligious Life Inventory (RELI-I), designed tomeasure the degree to which one's religiosity in-volves a questioning of the meaning of life arisingout of one's interactions with his social environment.

SlTUATIONAL AND DlSPOSITIONAL VARIABLES IN HELPING BEHAVIOR 103

The order of presentation of the scales in the ques-tionnaire was RELI, AR, D-O.

Consistent with prior research (Batson, 1971), aprincipal-component analysis of the total scale scoresand individual items for the 67 seminarians pro-duced a theoretically meaningful, orthogonally ro-tated three-component structure with the followingloadings:

Religion as means received a single very highloading from AR-E (.903) and therefore was definedby Allport and Ross's (1967) conception of thisscale as measuring religiosity as a means to otherends. This component also received moderate nega-tive loadings from D-O (-.400) and AR-I (-.372)and a moderate positive loading from RELI-EE(.301).

Religion as end received high loadings fromRELI-EI (.874), RELI-EE (.725), AR-I (.768),and D-O (.704). Given this configuration, and againfollowing Allport and Ross's conceptualization, thiscomponent seemed to involve religiosity as an end initself with some intrinsic value.

Religion as quest received a single very highloading from RELI-I (.945) and a moderate load-ing from RELI-EE (.75). Following Batson, thiscomponent was conceived to involve religiosityemerging out of an individual's search for meaningin his personal and social world.

The three religious personality scales examined inthe experimental research were constructed throughthe use of complete-estimation factor score coeffi-cients from these three components.

Scheduling of Experimental StudySince the incident requiring a helping response

was staged outdoors, the entire experimental studywas run in 3 days, December 14-16, 1970, between10 A.M. and 4 P.M. A tight schedule was used in anattempt to maintain reasonably consistent weatherand light conditions. Temperature fluctuation ac-cording to the New York Times for the 3 days dur-ing these hours was not more than 5 degrees Fahren-heit. No rain or snow fell, although the third daywas cloudy, whereas the first two were sunny. With-in days the subjects were randomly assigned to ex-perimental conditions.3

ProcedureWhen a subject appeared for the experiment, an

assistant (who was blind with respect to the person-ality scores) asked him to read a brief statementwhich explained that he was participating in a studyof the vocational careers of seminary students. Afterdeveloping the rationale for the study, the state-ment read:

3 An error was made in randomizing that increasedthe number of subjects in the intermediate-hurryconditions. This worked against the prediction thatwas most highly confirmed (the hurry prediction)and made no difference to the message variabletests.

What we have called you in for today is to pro-vide us with some additional material which willgive us a clearer picture of how you think thandoes the questionnaire material we have gatheredthus far. Questionnaires are helpful, but tend tobe somewhat oversimplified. Therefore, we wouldlike to record a 3-5-minute talk you give based onthe following passage. . . .

Variable 1: Message. In the task-relevant conditionthe passage read,

With increasing frequency the question is beingasked: What jobs or professions do seminarystudents subsequently enjoy most, and in whatjobs are they most effective? The answer to thisquestion used to be so obvious that the questionwas not even asked. Seminary students were beingtrained for the ministry, and since both society atlarge and the seminary student himself had a rela-tively clear understanding of what made a "good"minister, there was no need even to raise thequestion of for what other jobs seminary experi-ence seems to be an asset. Today, however, neithersociety nor many seminaries have a very clearlydefined conception of what a "good" minister isor of what sorts of jobs and professions are thebest context in which to minister. Many seminarystudents, apparently genuinely concerned with"ministering," seem to feel that it is impossible tominister in the professional clergy. Other students,no less concerned, find the clergy the most viableprofession for ministry. But are there other jobsand/or professions for which seminary experienceis an asset? And, indeed, how much of an assetis it for the professional ministry? Or, even morebroadly, can one minister through an "establish-ment" job at all?

In the helping-relevant condition, the subject wasgiven the parable of the Good Samaritan exactly asprinted earlier in this article. Next, regardless ofcondition, all subjects were told,

You can say whatever you wish based on thepassage. Because we are interested in how youthink on your feet, you will not be allowed to usenotes in giving the talk. Do you understand whatyou are to do? If not, the assistant will be gladto answer questions.After a few minutes the assistant returned, asked

if there were any questions, and then said:Since they're rather tight on space in this building,we're using a free office in the building next doorfor recording the talks. Let me show you how toget there [draws and explains map on 3 X S card].This is where Professor Steiner's laboratory is. Ifyou go in this door [points at map], there's asecretary right here, and she'll direct you to theoffice we're using for recording. Another of Pro-fessor Steiner's assistants will set you up for re-cording your talk. Is the map clear?

Variable 2: Hurry. In the high-hurry conditionthe assistant then looked at his watch and said, "Oh,

104 JOHN M. DARLEY AND C. DANIEL BATSON

you're late. They were expecting you a few minutesago. We'd better get moving. The assistant shouldbe waiting for you so you'd better hurry. It shouldn'ttake but just a minute." In the intermediate-hurrycondition he said, "The assistant is ready for you,so please go right over." In the low-hurry conditionhe said, "It'll be a few minutes before they're readyfor you, but you might as well head on over. If youhave to wait over there, it shouldn't be long."

The incident. When the subject passed through thealley, the victim was sitting slumped in a door-way, head down, eyes closed, not moving. As thesubject went by, the victim coughed twice andgroaned, keeping his head down. If the subjectstopped and asked if something was wrong or offeredto help, the victim, startled and somewhat groggy,said, "Oh, thank you [cough]. . . . No, it's all right.[Pause] I've got this respiratory condition [cough].. . . The doctor's given me these pills to take, andI just took one. . . . If I just sit and rest for afew minutes I'll be O.K. . . . Thanks very much forstopping though [smiles weakly]." If the subjectpersisted, insisting on taking the victim inside thebuilding, the victim allowed him to do so andthanked him.

Helping ratings. The victim rated each subject ona scale of helping behavior as follows:

0 = failed to notice the victim as possibly in needat all; 1 = perceived the victim as possibly in needbut did not offer aid; 2 = did not stop but helpedindirectly (e.g., by telling Steiner's assistant about thevictim); 3 — stopped and asked if victim neededhelp; 4 = after stopping, insisted on taking the vic-tim inside and then left him.

The victim was blind to the personality scalescores and experimental conditions of all subjects.At the suggestion of the victim, another categorywas added to the rating scales, based on his observa-tions of pilot subjects' behavior:

5 — after stopping, refused to leave the victim (after3-5 minutes) and/or insisted on taking him some-where outside experimental context (e.g., for coffeeor to the infirmary).

(In some cases it was necessary to distinguishCategory 0 from Category 1 by the postexperimentalquestionnaire and Category 2 from Category 1 onthe report of the experimental assistant.)

This 6-point scale of helping behavior and adescription of the victim were given to a panel of10 judges (unacquainted with the research) whowere asked to rank order the (unnumbered) cate-gories in terms of "the amount of helping behaviordisplayed toward the person in the doorway." Of the10, 1 judge reversed the order of Categories 0 and1. Otherwise there was complete agreement with theranking implied in the presentation of the scaleabove.

The speech. After passing through the alley andentering the door marked on the map, the subjectentered a secretary's office. She introduced him to

the assistant who gave the subject time to prepareand privately record his talk.

Helping behavior questionnaire. After recordingthe talk, the subject was sent to another experi-menter, who administered "an exploratory question-naire on personal and social ethics." The question-naire contained several initial questions about theinterrelationship between social and personal ethics,and then asked three key questions: (a) "When wasthe last time you saw a person who seemed to be inneed of help?" (b) "When was the last time youstopped to help someone in need?" (c) "Have youhad experience helping persons in need? If so, out-line briefly." These data were collected as a checkon the victim's ratings of whether subjects who didnot stop perceived the situation in the alley as onepossibly involving need or not.

When he returned, the experimenter reviewed thesubject's questionnaire, and, if no mention was madeof the situation in the alley, probed for reactions toit and then phased into an elaborate debriefing anddiscussion session.

DebriefingIn the debriefing, the subject was told the exact

nature of the study, including the deception in-volved, and the reasons for the deception were ex-plained. The subject's reactions to the victim and tothe study in general were discussed. The role ofsituational determinants of helping behavior wasexplained in relation to this particular incident andto other experiences of the subject. All subjectsseemed readily to understand the necessity for thedeception, and none indicated any resentment of it.After debriefing, the subject was thanked for histime and paid, then he left.

RESULTS AND DISCUSSIONOverall Helping Behavior

The average amount of help that a subjectoffered the victim, by condition, is shown inTable 1. The unequal-A7 analysis of varianceindicates that while the hurry variable wassignificantly (F = 3.56, dj — 2/34, p < .05)related to helping behavior, the messagevariable was not. Subjects in a hurry werelikely to offer less help than were subjectsnot in a hurry. Whether the subject was go-ing to give a speech on the parable of theGood Samaritan or not did not significantlyaffect his helping behavior on this analysis.

Other studies have focused on the questionof whether a person initiates helping actionor not, rather than on scaled kinds of helping.The data from the present study can also beanalyzed on the following terms: Of the 40subjects, 16 (40%) offered some form of

SlTUATIONAL AND DlSPOSITIONAL VARIABLES IN HELPING BEHAVIOR 105

direct or indirect aid to the victim (CodingCategories 2-5), 24 (60%) did not (CodingCategories 0 and 1). The percentages of sub-jects who offered aid by situational variablewere, for low hurry, 63% offered help, inter-mediate hurry 45%, and high hurry 10%;for helping-relevant message 53%, task-rele-vant message 29%. With regard to this moregeneral question of whether help was offeredor not, an unequal-./V analysis of variance(arc sine transformation of percentages ofhelpers, with low- and intermediate-hurryconditions pooled) indicated that again onlythe hurry main effect was significantly (F =5.22, p < .05) related to helping behavior;the subjects in a hurry were more likely topass by the victim than were those in less ofa hurry.

Reviewing the predictions in the light ofthese results, the second hypothesis, that thedegree of hurry a person is in determines hishelping behavior, was supported. The predic-tion involved in the first hypothesis concern-ing the message content was based on theparable. The parable itself seemed to suggestthat thinking pious thoughts would not in-crease helping. Another and conflicting pre-diction might be produced by a norm saliencetheory. Thinking about the parable shouldmake norms for helping salient and thereforeproduce more helping. The data, as hypothe-sized, are more congruent with the predictiondrawn from the parable. A person going tospeak on the parable of the Good Samaritan

TABLE 1MEANS AND ANALYSIS OF VARIANCE OF

GRADED HELPING RESPONSES

M

Message

Helping relevantTask relevantSummary

Hurry

Low

3.8001.6673.000

Medium

2.0001.6671.818

High

1.000.500.700

Sum-mary

2.2631.333

Analysis of variance

Source

Message (A)Hurry (B)A X B

Error

55

7.76620.8845.237

99.633

Af

122

34

MS

7.76610.4422.6192.930

P2.653.56*.89

Note. N = 40.* p < .05.

is not significantly more likely to stop tohelp a person by the side of the road than isa person going to talk about possible occupa-tions for seminary graduates.

Since both situational hypotheses are con-firmed, it is tempting to stop the analysis ofthese variables at this point. However, multi-ple regression analysis procedures were alsoused to analyze the relationship of all of theindependent variables of the study and thehelping behavior. In addition to often beingmore statistically powerful due to the use of

TABLE 2STEPWISE MULTIPLE REGRESSION ANALYSIS

Help vs. no help Graded helping

Step

1. Hurryb

2. Message0

3. Religion as quest4. Religion as means5. Religion as end

Individualvariable

r a

-.37.25

-.03-.03

.06

F

4.537*1.495.081.003.000

Overallequation

R

.37

.41

.42

.42

.42

F

5.884*3.834*2.5211.838*1.430

1. Hurry2. Message3. Religion as quest4. Religion as means5. Religion as end

Individualvariable

Y

-.42.25

-.16-.08-.07

F6.665*1.7191.297.018.001

Variableequation

R

.42

.46

.50

.50

.50

F8.196**5.083*3.897*2.848*2.213

Note. N = 40. Helping is the dependent variable, d/ = 1/34.« Individual variable correlation coefficient is a point biserial where appropriate.b Variables are listed in order of entry into stepwise regression equations.0 Helping-relevant message is positive.* p < .05.

**£ < .01.

106 JOHN M. DARLEY AND C. DANIEL BATSON

more data information, multiple regressionanalysis has an advantage over analysis ofvariance in that it allows for a comparison ofthe relative effect of the various independentvariables in accounting for variance in thedependent variable. Also, multiple regressionanalysis can compare the effects of continuousas well as nominal independent variables onboth continuous and nominal dependent varia-bles (through the use of point biserial corre-lations, rpi,) and shows considerable robust-ness to violation of normality assumptions(Cohen, 1965, 1968). Table 2 reports the re-sults of the multiple regression analysis usingboth help versus no help and the graded help-ing scale as dependent measures. In this tablethe overall equation Fs show the F value ofthe entire regression equation as a particularrow variable enters the equation. Individualvariable Fs were computed with all five inde-pendent variables in the equation. Althoughthe two situational variables, hurry and mes-sage condition, correlated more highly withthe dependent measure than any of the re-ligious dispositional variables, only hurry wasa significant predictor of whether one will helpor not (column 1) or of the overall amountof help given (column 2). These resultscorroborate the findings of the analysis ofvariance.4

Notice also that neither form of the thirdhypothesis, that types of religiosity will pre-dict helping, received support from these data.No correlation between the various measuresof religiosity and any form of the dependentmeasure ever came near statistical signifi-cance, even though the multiple regressionanalysis procedure is a powerful and not par-ticularly conservative statistical test.

Personality Difference among Subjects WhoHelped

To further investigate the possible influ-ence of personality variables, analyses were

4 To check the legimacy of the use of both analy-sis of variance and multiple regression analysis,parametric analyses, on this ordinal data, Kendallrank correlation coefficients were calculated betweenthe helping scale and the five independent variables.As expected T approximated the correlation quiteclosely in each case and was significant for hurryonly (hurry, T = — .38, p < .001).

carried out using only the data from subjectswho offered some kind of help to the victim.Surprisingly (since the number of these sub-jects was small, only 16) when this was done,one religiosity variable seemed to be signifi-cantly related to the kind of helping behavioroffered. (The situational variables had nosignificant effect.) Subjects high on the re-ligion as quest dimension appear likely, whenthey stop for the victim, to offer help of amore tentative or incomplete nature than aresubjects scoring low on this dimension (r =-.53, p < .05).

This result seemed unsettling for the think-ing behind either form of Hypothesis 3. Notonly do the data suggest that the Allport-Ross-based conception of religion as end doesnot predict the degree of helping, but thereligion as quest component is a significantpredictor of offering less help. This latterresult seems counterintuitive and out of keep-ing with previous research (Batson, 1971),which found that this type of religiosity cor-related positively with other socially valuedcharacteristics. Further data analysis, how-ever, seemed to suggest a different interpre-tation of this result.

It will be remembered that one helping cod-ing category was added at the suggestion ofthe victim after his observation of pilot sub-jects. The correlation of religious personalityvariables with helping behavior dichotomizedbetween the added category (1) and all ofthe others (0) was examined. The correlationbetween religion as quest and this dichoto-mous helping scale was essentially unchanged(>> = —.54, p < .05). Thus, the previouslyfound correlation between the helping scaleand religion as quest seems to reflect thetendency of those who score low on the questdimension to offer help in the added helpingcategory.

What does help in this added category rep-resent? Within the context of the experi-ment, it represented an embarrassment. Thevictim's response to persistent offers of helpwas to assure the helper he was all right, hadtaken his medicine, just needed to rest for aminute or so, and, if ultimately necessary, torequest the helper to leave. But the superhelpers in this added category often would

SlTUATIONAL AND DlSPOSITIONAL VARIABLES IN HELPING BEHAVIOR 107

not leave until the final appeal was repeatedseveral times by the victim (who was grow-ing increasingly panicky at the possibility ofthe arrival of the next subject). Since itusually involved the subject's attempting tocarry through a preset plan (e.g., taking thesubject for a cup of coffee or revealing to himthe strength to be found in Christ), and didnot allow information from the victim tochange that plan, we originally labeled thiskind of helping as rigid—an interpretationsupported by its increased likelihood amonghighly doctrinal orthodox subjects (r =.63, p < .01). It also seemed to have an in-appropriate character. If this more extremeform of helping behavior is indeed effectivelyless helpful, then the second form of Hy-pothesis 3 does seem to gain support.

But perhaps it is the experimenters ratherthan the super helpers who are doing the in-appropriate thing; perhaps the best character-ization of this kind of helping is as differentrather than as inappropriate. This kind ofhelper seems quickly to place a particularinterpretation on the situation, and the help-ing response seems to follow naturally fromthis interpretation. All that can safely besaid is that one style of helping that emergedin this experiment was directed toward thepresumed underlying needs of the victim andwas little modified by the victim's commentsabout his own needs. In contrast, anotherstyle was more tentative and seemed moreresponsive to the victim's statements of hisneed.

The former kind of helping was likely tobe displayed by subjects who expressedstrong doctrinal orthodoxy. Conversely, thisfixed kind of helping was unlikely among sub-jects high on the religion as quest dimen-sion. These latter subjects, who conceivedtheir religion as involving an ongoing searchfor meaning in their personal and social world,seemed more responsive to the victim's im-mediate needs and more open to the victim'sdefinitions of his own needs.

CONCLUSION AND IMPLICATIONSA person not in a hurry may stop and offer

help to a person in distress. A person in ahurry is likely to keep going. Ironically, he

is likely to keep going even if he is hurryingto speak on the parable of the Good Samari-tan, thus inadvertently confirming the pointof the parable. (Indeed, on several occasions,a seminary student going to give his talk onthe parable of the Good Samaritan literallystepped over the victim as he hurried on hisway!)

Although the degree to which a person wasin a hurry had a clearly significant effect onhis likelihood of offering the victim help,whether he was going to give a sermon on theparable or on possible vocational roles ofministers did not. This lack of effect of ser-mon topic raises certain difficulties for an ex-planation of helping behavior involving help-ing norms and their salience. It is hard tothink of a context in which norms concerninghelping those in distress are more salient thanfor a person thinking about the Good Sa-maritan, and yet it did not significantly in-crease helping behavior. The results were inthe direction suggested by the norm saliencehypothesis, but they were not significant. Themost accurate conclusion seems to be thatsalience of helping norms is a less strongdeterminant of helping behavior in the pres-ent situation than many, including the presentauthors, would expect.

Thinking about the Good Samaritan didnot increase helping behavior, but being in ahurry decreased it. It is difficult not to con-clude from this that the frequently citedexplanation that ethics becomes a luxury asthe speed of our daily lives increases is atleast an accurate description. The picture thatthis explanation conveys is of a person seeinganother, consciously noting his distress, andconsciously choosing to leave him in distress.But perhaps this is not entirely accurate, for,when a person is in a hurry, something seemsto happen that is akin to Tolman's (1948)concept of the "narrowing of the cognitivemap." Our seminarians in a hurry noticed thevictim in that in the postexperiment interviewalmost all mentioned him as, on reflection,possibly in need of help. But it seems thatthey often had not worked this out when theywere near the victim. Either the interpreta-tion of their visual picture as a person indistress or the empathic reactions usually

108 JOHN M. BARLEY AND C. DANIEL BATSON

associated with that interpretation had beendeferred because they were hurrying. Ac-cording to the reflections of some of the sub-jects, it would be inaccurate to say that theyrealized the victim's possible distress, thenchose to ignore it; instead, because of thetime pressures, they did not perceive thescene in the alley as an occasion for an ethicaldecision.

For other subjects it seems more accurateto conclude that they decided not to stop.They appeared aroused and anxious after theencounter in the alley. For these subjects,what were the elements of the choice thatthey were making? Why were the seminarianshurrying? Because the experimenter, whomthe subject was helping, was depending onhim to get to a particular place quickly. Inother words, he was in conflict between stop-ping to help the victim and continuing on hisway to help the experimenter. And this isoften true of people in a hurry; they hurrybecause somebody depends on their beingsomewhere. Conflict, rather than callousness,can explain their failure to stop.

Finally, as in other studies, personalityvariables were not useful in predicting whethera person helped or not. But in this study,unlike many previous ones, considerable vari-ations were possible in the kinds of helpgiven, and these variations did relate to per-sonality measures—specifically to religiosityof the quest sort. The clear light of hindsightsuggests that the dimension of kinds of help-ing would have been the appropriate place tolook for personality differences all along;whether a person helps or not is an instantdecision likely to be situationally controlled.How a person helps involves a more complexand considered number of decisions, includ-ing the time and scope to permit personalitycharacteristics to shape them.

REFERENCESALLEN, R. O., & SPILKA, B. Committed and consen-

sual religion. A specification of religion-prejudicerelationships. Journal for the Scientific Study ofReligion, 1967, 6, 191-206.

ALLPORT, G. W., & Ross, J. M. Personal religiousorientation and prejudice. Journal of Personalityand Social Psychology, 1967, 5, 432-443.

BATSON, C. D. Creativity and religious develop-ment: Toward a structural-functional psychologyof religion. Unpublished doctoral dissertation,Princeton Theological Seminary, 1971.

BICKMAN, L. B. The effect of the presence of otherson bystander intervention in an emergency. Un-published doctoral dissertation, City College ofthe City University of New York, 1969.

BROWN, L. B. Classifications of religious orientation.Journal for the Scientific Study of Religion, 1964,4, 91-99.

BURTON, R. V. The generality of honesty reconsid-ered. Psychological Review, 1963, 70, 481-499.

COHEN, J. Multiple regression as a general data-analytic system. Psychological Bulletin, 1968, 70,426-443.

COHEN, J. Some statistical issues in psychologicalresearch. In B. B. Wolman (Ed.), Handbook ofclinical psychology. New York: McGraw-Hill,1965.

DARLEY, J. M., & LATANE, B. Bystander interventionin emergencies: Diffusion of responsibility. Jour-nal of Personality and Social Psychology, 1968, 8,377-383.

FREUD, S. The future of an illusion. New York:Liveright, 1953.

FUNK, R. W. Language, hermeneutic, and word ofGod. New York: Harper & Row, 1966.

CLOCK, C. Y., & STARK, R. Christian beliefs andanti-Semitism. New York: Harper & Row, 1966.

HARTSHORNE, H., & MAY, M. A. Studies in the natureof character. Vol. 1. Studies in deceit. New York:Macmillan, 1928.

KORTE, C. Group effects on help-giving in an emer-gency. Proceedings of the 77th Annual Conventionof the American Psychological Association, 1969,4, 383-384. (Summary)

ORNE, M. T. On the social psychology of the psy-chological experiment: With particular reference todemand characteristics and their implications.American Psychologist, 1962, 17, 776-783.

ROSENBERG, M. J. When dissonance fails: On elimi-nating evaluation apprehension from attitude mea-surement. Journal of Personality and Social Psy-chology, 1965, 1, 28-42.

SCHWARTZ, S. H., & CLAUSEN, G. T. Responsibility,norms, and helping in an emergency. Journal ofPersonality and Social Psychology, 1970, 16, 299-310.

TOLMAN, E. C. Cognitive maps in rats and men.Psychological Review, 1948, 55, 189-208.

(Received January 6, 1972)