Psych Term Paper

Pacific Sociological Association

Measuring Social Norms Author(s): Sanford Labovitz and Robert Hagedorn Reviewed work(s): Source: The Pacific Sociological Review, Vol. 16, No. 3 (Jul., 1973), pp. 283-303 Published by: University of California Press Stable URL: http://www.jstor.org/stable/1388488 . Accessed: 22/11/2012 21:20

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

University of California Press and Pacific Sociological Association are collaborating with JSTOR to digitize, preserve and extend access to The Pacific Sociological Review.

http://www.jstor.org

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

MEASURING SOCIAL NORMS

SANFORD LABOVITZ University of Calgary

ROBERT HAGEDORN University of Victoria

Although social norms have been considered to be an important sociological concept (Gibbs, 1965; Morris, 1956; Jackson, 1966; Sumner, 1906; Homans, 1950; Blake and Davis, 1964), they are seldom measured in a systematic way. In the relatively few inquiries in which they are operationalized, there is little concensus on the most reliable and valid techniques. The purposes of this article are to explore the nature of social norm measurement and to demonstrate the utility of using more than one measurement technique. The basic idea is that different techniques are likely to be biased in different ways.' These biases may "cancel each other out" so that if two or more techniques produce similar conclusions, we can have more confidence in the techniques (for each acts as a validity check on the others) and more confidence in the conclusions. Further, given the theoretical vagueness of social norms, a variety of approaches designed to tap their various meanings seems to be a sound research strategy.

SOCIAL NORMS DEFINED

Most definitions of social norms are based on social evalu- ations or social sanctions or both (see the summary by Gibbs, Pacific Sociological Rev., Vol. 16 No. 3, July 1973 01973 Pacific Sociological Assn.

[2831

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[284] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

1965). Evaluations refer to statements of should or ought, or should not or ought not. If evaluations are stressed, norms are defined as standards of conduct that should or should not be followed. Sanctions refer to any kind of reactions to behavior, positive or negative, that attempt to alter the behavior, or increase or decrease its frequency. If sanctions are stressed, norms are defined as standards of conduct that are rewarded or punished. Stated otherwise, people receive physical or psycho- logical rewards and punishments for their behavior, which encourages or discourages them to conduct themselves in a similar way in the future.

THE IMPUTED IMPORTANCE OF SOCIAL NORMS

If most people behave in accordance with at least some norms-that is, if they behave as they should according to certain standards of society-then those norms may account, in part, for the behavioral patterns in a society; and different norms among societies may account, in part, for behavioral variation. Social norms, consequently, may be important links to order and predictability, which would make them critical factors in social control.

Furthermore, if social norms are learned, they may be modified over time; and if they are truly a causal link to behavior and interaction patterns, then normative modifications may account for certain social changes. Along this line, normative conflict and normative ambiguity also may lead to social changes such as a breakdown in the social order.

Besides their imputed link to social change and their unifying and controlling nature, conformity and deviance are usually defined in terms of following or not following social norms. To establish whether conformity or deviance has occurred, con- sequently, it is first necessary to establish the norms in question.

These conceptual treatments of the importance of social norms are largely speculative. The degree of predictive and explanatory power must be tested before their importance can

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [285]

be established, and any test depends on reliable and valid measurement.

MEASUREMENT OF SOCIAL NORMS

Techniques for measurement may be subsumed under four types: (1) questioning people, (2) inferring norms from be- havior, (3) postulated effects-that is, measuring the implica- tions of hypothetical norms, and (4) written documents. Each technique appears to have particular kinds of biases that may not characterize the others.

NORMS MEASURED BY DIRECT QUESTIONS

Questioning respondents may involve the direct approach ("Should people go to church?") or the use of a hypothetical or actual situation ("What should the teacher do to the student who was caught cheating on the final examination?"). Ques- tions, furthermore, may stress the evaluative aspects of social norms-that is, should or should not statements-or they may stress sanctions-that is, rewards and punishments. Studies may use one or a combination of such questions. The important point is, however, that responses to the questions determine the content and dimensions of social norms.

Questionnaire items are always subject to potential biases. For example, some respondents may lie or have faulty memories; some simply are not qualified to provide certain kinds of data; and some responses are invalid because respond- ents may give answers they feel the researcher wants or that represent their community rather than themselves.

An illustration of measuring social norms by questioning is a study by Turk (1965) concerning student participation in the classroom situation. In response to an article by Wrong (1961) stressing the "over-socialized conception of man," Turk feels that the author has overstated his case. In contrast to Wrong's thesis that norms conflict with individual motives, he hypothe-

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[286] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

sizes that agreement on social norms applied to others will be greater than agreement when self-applied.

Turk tested his agreement thesis on two introductory sociology classes at Duke University in the early 1960s. His problem was to specify the social norm on student participation in the classroom; specifically, on the amount of classroom time a student should use to express a lengthy opinion. To test the thesis, he obtained from each student the amount of time they felt they should use and the amount of time they felt others in the class should use. Each class was divided into two groups and each group answered only one of the following questions:

If you were to express a lengthy opinion in a regular one-period class at Duke, what is the longest you should take so that class time will not be monopolized? minutes and seconds. If a classmate were to express a lengthy opinion in a regular one-period class at Duke, what is the longest he should take so that class time will not be monopolized? minutes and seconds.

The results of the study support the thesis. Agreement, as measured by the variance, was greater with regard to others than with regard to oneself. Students, on the average, allotted themselves more time to express an opinion than they allotted to others.

Another study measuring social norms by direct questioning was carried out by the authors. The study measures the range of acceptable behavior of student involvement in determining the content of courses and is based on a conceptualization by Jackson (1960, 1964, 1966; Glick and Jackson, 1970). Jackson maintains that social norms are measured on two dimensions. First, there is a behavioral dimension specifying the amount or extent that is acceptable-for example, the degree to which a foreman supervised his workers, the number of times a person speaks in group therapy, or the degree to which citizens are involved in community affairs. The second dimension is an evaluation of the degree of approval or disapproval of the extent or range of behavior-for example, most people may approve of a person speaking from two to four times in group

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [287]

therapy, but disapprove of those who remain silent or repeated- ly speak out and, thereby, dominate the discussion. Note that Jackson uses approval rather than should in his specification of social norms.

In two classes-a senior sociology class at the University of Southern California and an introductory sociology class at the University of Victoria-students were asked the following: "According to an evaluation scale ranging from +3 (highly approve) to -3 (highly disapprove) and containing the values, +3, +2, 0, -1, -2, -3, what percentage of student suggestions on course content should be accepted by a professor: 0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100?" The mean evaluational response (approval to disapproval) was then calculated for each percentage and plotted as shown in Figure 1.

Not only does the figure show the acceptable and unaccept- able ranges of behavior for both classes, but it indicates the intensity of feeling associated with each behavioral distinction by the heights and depths of the curves. Perhaps the most striking difference between the two classes is the substantially greater intensity of feeling by the USC students. Their curve reaches a greater height (over +2 average), as well as a lower depth (almost the maximum negative evaluation of a -3 at both extremes). The second major distinction is the higher acceptable range of professional behavior (in accepting student suggestions) for the class at USC. Although their acceptable ranges overlap to some extent, the class at Victoria positively evaluates an acceptance of only about 15% of the student suggestions; for the USC class, the lowest positive evaluation is just above 30%. At the other extreme of acceptable behavior, UV positively evaluates up to about 55%, while USC approves, on the average, about 65%. It seems clear that not only do students in the USC class want a larger influence over what is taught than those in the class at UV, but they feel more strongly about it.

Besides measuring social norms by direct questions, a researcher may ask respondents to select the acceptable or unacceptable behavior in a hypothetical story. Rather than asking directly, "What is the norm governing a specific

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[288] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

Figure 1. MEAN EVALUATION SCORES FOR TWO CLASSESa ON THE PER- CENTAGE OF STUDENT SUGGESTIONS ON COURSE CONTENT THAT ARE ACCEPTED BY THE PROFESSOR

(highly approve) +3

+2

Evaluation 4-USC Class Scale +1

--UV Class

(indifferent) 0 10 20 30 40 50 60 70 80d 90 100

-1

-2 Percentage of Student Suggestions -2

(highly disapprove) -3

a. Senior sociology class at the University of Southern California in 1970, and an introductory sociology class at the University of Victoria in 1970.

behavior?" we ask, "What would the norm be if the following situation occurred?"

A good example of the use of a hypothetical story is a study carried out by Angell (1962). Based on a sample of 800 (out of approximately 3,000,000) in Detroit and contiguous areas in 1955-1956, he studied the "proper" standard of conduct in areas of religion, civil liberties, family privacy, and race relations. In the attempt to determine the social norms governing a given problem situation, respondents were asked to choose one of three alternatives after hearing a brief hypothet-

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [2891

ical story. The hypothetical story in the area of civil liberties is given below.

The story concerns the problem of freedom of speech- for atheists as opposed to the religious orientation of the society:

At a college, a student group plans to ask someone who does not believe in God (an atheist) to give a speech on his views toward religion. Some people want the president of the college to do something about this speech. What should the president do? (a) He should allow the speech to be made. (b) He should ask students to call off the speech, but not forbid it. (c) He should not allow the speech to be made.

Fifty percent of the respondents selected alternative a, 25% alternative b, and 21% alternative c. The results, consequently, support the freedom of speech for atheists, but they are not conclusive. Combining the responses to alternatives b and c shows that 46% of the sample are opposed to this type of freedom of speech for atheists-that is, they do not want the speech to be made, and they are willing to do something about it. It may be concluded tentatively that, although there is a tendency to support it, substantial dissensus characterizes the sample on the following social norm: "Freedom of speech should be afforded to atheists to the extent of allowing them to speak on campus."

Because norms seldom if ever apply universally, Samuel A. Stouffer (1949) tried to specify the situations in which some are applicable and also specify the situations characterized by normative conflict (for a more recent application of Stouffer's design, see Stoodley, 1959). The nature of the conflicting norms and the specific situation, according to the author, are important factors because they at least partially determine a person's choices among a set of behavioral alternatives.

The study is based on 196 students at Harvard and Radcliffe who were mostly undergraduates. The major concerns of the study are over the issues of cheating on examinations and whether the social norms about cheating change with the

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[2901 PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

situation. One of his major propositions is that persons strain to conform to social norms, even in conflicting situations.

Based on anonymous questionnaires, each student was told to imagine that he was proctoring an examination. During the examination, the proctor catches a student cheating (by copying answers from written notes). The students were asked what they would do in this situation under a given set of different circumstances. They were asked what they would do, for example, if they did not know the student, and then they were asked what they would do if the student was a roommate or a close friend. Under such situations, the alternative actions listed for the proctor were (1) "Take away his notes and exam book, dismiss him, and report him for cheating"; (2) "Take away his notes, let him finish the exam, but report him for cheating"; (3) "If he can be led to withdraw from the exam on some excuse, do not report him for cheating; otherwise report him"; (4) "Take away his notes, but let him finish the exam, and not report him for cheating."

Stouffer found that those actions approved by the students were not entirely consistent with what the students thought would be approved by university authorities. Students ap- proved, to illustrate, alternatives 3, 4, and 5 much more than their imputed approval to the university authorities. He found, further, that being a roommate or friend dramatically altered the approved actions. These results indicate that student norms and norms imputed to university authorities are not in complete agreement on cheating on examinations. The student, when acting as proctor, therefore, may be under conflicting norms- that is, the norms of authorities and students do not overlap to some extent. The extent to which they do not overlap results in conflict for the student in making a decision. In such conflict situations, many students selected a compromise alternative, and, if the student perceived an overlap between his preference and the preference of the authorities, he was quite likely to select the alternative in which both approved.

Despite these studies, the question remains, do norms predict behavior? Stated otherwise, to what extent do people conform

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [291]

to norms? Partially taking their cue from Stouffer's study, the authors have carried out a small inquiry on norms and behavior in cheating on examinations in college. Questionnaires were handed out in several classrooms, from 1968-1970, at the University of Southern California and California State College at Fullerton (now California State University, Fullerton). The number of students completing the anonymous questionnaires was 296, which represents almost a 100% return. Social norms were measured by two questions: "How often should college students cheat on exams?" (always, usually, sometimes, seldom, never); and "In your opinion, how often do other students think college students should cheat on exams?" (always, usually, sometimes, seldom, never). The first question is designed to tap the cheating norm of each student as he feels it applies; the second is designed to tap the imputed norm of other students. To ascertain the predictive nature of the cheating norm, two perceived behavioral questions were asked: "How often have you cheated on exams in college?" (very frequently, frequently, seldom, never); and "How often have you seen other college students cheating on exams?" (very frequently, frequently, seldom, never). About 76% of the respondents say that college students should never cheat on exams. This measure suggests that the social norm governing the examination situation is that college students should not cheat. About 50% of the students, however, report that they have cheated in their college career; and 82% report that they have seen other students cheating. These results strongly suggest that students are violating the norm to a large extent. Given, however, that besides the social norm there may be many factors that may influence a student to cheat (importance of the course to the student, ability of the student, fairness of the instructor, and the like), is the cheating norm at least partially constraining on the individual? That is, can behavior be predicted by the norm to some extent? Examination of the data indicates that the self-designated norm on cheating is related to a student's reported frequency of cheating (,y = .60). Fifty-seven percent of those students who feel you should never cheat

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[292] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

report that they, in fact, never do cheat; furthermore, 75% of the students who say you should cheat to some extent, report that they have, in fact, cheated to some extent.

NORMS INFERRED FROM BEHAVIOR

Perhaps the most prevalent way of designating social norms is by inferring from behavior. Social scientists observe some behavior pattern (a frequently repeated behavior) and impute a norm from it.2 In England, to illustrate, almost every marriage is monogamous, so a norm is imputed that marriages should be monogamous. In the United States, most marriages occur in June, so perhaps people should marry in June. Since small girls do play house, small girls should play house. Finally, since people do cheat on their income tax returns, they should cheat on them.

The social scientist inferring norms from behavior must be quite cautious of three factors. First, the "facts" may be incorrect. Do most people in fact cheat on their income tax returns? What percentage of the small girls in a society actually play house? And is this percentage substantially larger than that for small boys? Inferring norms from behavior is a difficult enough process without having misperceived the behavior in question. Second, even if the facts are correct, they may not be reflecting a social norm; other factors beside social norms may account for the behavior. Most people may be monogamous, for example, because this marital relationship is the most economi- cally feasible in certain types of societies; marriage rates may go up in June because this is the first month college students are out from school; and many people may cheat on their income tax returns because they are very likely to get away with it. We are not implying that these are the actual or only reasons for the behavior, but only that they are reasonable alternatives to the imputation of social norms. A third factor becomes extremely important in inferring norms from behavior when the researcher is trying to explain the behavior in question. If the explanation is in terms of the norm (norms constrain people to

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [293]

behave in certain ways), then the reasoning is tautological or circular in nature.

A survey by Converse and Campbell (1968) illustrates a rather sophisticated attempt to infer voting norms from the voting behavior of Jews, Catholics, labor union members, and blacks. Their study of the 1956 presidential election is based on interviews of a random sample of 1,772 adults in the United States living in private households. The authors concentrated on establishing the existence of group voting norms such as persons should vote as their group specifies or as the group leaders specify. In this case, the four groups of Jews, Catholics, labor union members, and blacks dominantly push a Democratic voting norm-that is, members should vote for the Democratic presidential nominee and for the slate of Democratic candidates in general.

If members of a group vote alike or are similar on any type of behavior, argue Converse and Campbell, they may either be influenced by a group norm which constrains them to behave that way, or they may be responding to the same life experiences. Responding to the same life experiences by behaving alike may merely reflect the notion that similar experiences lead to similar responses. If so, similar voting behavior is not due to group voting norms. Their task, consequently, is to demonstrate that members' voting behavior is not a response to similar life experiences.

To demonstrate that similar life experiences are not the sole cause of a group's homogeneous voting behavior, Converse and Campbell compared the "distinctiveness" of voting by group members when paired with nonmembers of similar character- istics. Catholics, to illustrate, are more likely to vote Demo- cratic than non-Catholics of equivalent ethnic background, sex, age, income, education, occupation, residence location, and region. Since these background characteristics are indicative of similar life experiences and are equated for Catholics and non-Catholics, the only major difference between the two groups is religious membership. It should be noted that the comparison did not use all possible characteristics to equate the

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[294] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

groups, and that religious membership also is a life experience. Although these two factors do not negate the interpretation that group norms are one of the causes of the distinctive voting behavior of Catholics, it should caution us in accepting the conclusion too readily. All four groups proved to be distinctive in voting to some extent-that is, when compared with nonmembers of similar characteristics, members of each group voted more homogeneously. These results are consistent with the group influence or group norm thesis.

Although the inferred norm is a possible interpretation for the relatively homogeneous voting behavior in each group, another possible interpretation is similar life experiences, because they could not control on all background character- istics. They may, consequently, have inferred norms where none actually exist, they have inferred the wrong norm, or, perhaps, they have inferred the correct norm, but it is too weak or unknown to have much influence on behavior. Either way, their conclusion that group voting norms influence members' voting behavior may be in error.

Another illustration of inferring norms from behavior is taken from several "quasi-experiments" reported by Harold Garfinkel (1964). Based on the disruption of everyday occur- rences, Garfinkel inferred the existence of what may be called common understandings or social expectancies between per- sons. These understandings or expectancies can be viewed as social norms, because they regulate behavior by designating what one should or should not do in a given situation. The task for sociology, according to Gardinkel, is to establish these understandings and expectancies-that is, social norms-that underlie everyday life. He attempts to do this by disrupting everyday activities and observing the consequences. The unmis- takable disruptive consequences of his quasi-experiments is consistent with the notion that interaction is sustained by the unnoticed rules of proper behavior (norms).

To illustrate Garfinkel's techniques consider the game of tick-tack-toe. A quasi-experiment may be set up by letting the naive subject put his X down first; the experimenter, instead of

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [2951

placing his O in an appropriate place, erases the subject's X and places it elsewhere. Now the task of the experimenter is merely to observe the behavior or reaction of the subject. A common reaction is to state that the experimenter is unfair or perhaps even to get angry; either way, the negative reaction leads to the assumption that a rule of the game has been violated-that is, a social norm has not been followed.

Other examples of his quasi-experiments are bargaining for standard priced merchandise; in your own household, assume that you are no longer a son or daughter, but a boarder, and address your parents formally; and, in ordinnary conversation, bring your face close to the other person's face. The inferred norm in the last example is that there is a proper distance that should be maintained when conversing.

MEASURING THE IMPLICATIONS OF POSTULATED NORMS

If social norms are treated hypothetically, their implications may be measured or tested. Only the imputed behavioral consequences are measured; if the consequences of the unmeas- ured standard are predicted, then they support the idea that there is, in fact, a social norm as specified.

Compared to the other three measuring techniques of questioning, inferring from behavior and written documents the use of postulated effects is a more indirect method for determining social norms. The measured factors are the possible consequences of the postulated norm. Because these conse- quences could be caused by conditions other than the norm in question,this technique seems more problematic than the three more direct measures. Even if the imputed consequences are observed, rather than validating the postulated norms, one can only say that the data are consistent with it.

Suppose social norms make human behavior predictable and establish order in society. What would be some of the behavioral consequences of a normless situation? The ability to anticipate or predict the behavior of others seems crucial to individual stability. If human behavior is haphazard or "ran-

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[296] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

dom," individuals are likely to experience extreme stress. Normlessness, furthermore, means that the established order is gone-society and its subgroups lack cohesiveness. Two likely results of normlessness, consequently, are low cohesiveness and personal stress. Without actually measuring the degree to which either a situation is normless or there is a normative breakdown, group cohesiveness and individual stress can be determined. Observations of low cohesiveness and high personal stress in areas ravished by flood or hurricane support the assumed normlessness of the situations.

A theory linking status integration to suicide by Gibbs and Martin (1964) provides a test of the implication of hypothetical norms. Their theory is based on five interrelated propositions from which one testable hypothesis is deduced: suicide rates vary inversely with status integration. Status integration refers to the occupancy of different positions in society in which the social norms and roles do not conflict. Being a male and an airline pilot, to illustrate, does not lead to much conflict, but being a mother and a businesswoman might. The first two statuses are highly integrated (male pilots), while the last two are not (businesswomen who are mothers). A high suicide rate would be predicted for this type of woman, because the expectations or social norms for businesswomen (devote a large amount of time to business) often conflict with those of being a mother (devote a large amount of time to the children).

The theory's five propositions and derived testable hypoth- esis are:

Proposition 1: The suicide rate of a population varies inversely with the stability and durability of social relationships within that population. Proposition 2: The stability and durability of social relationships within a population vary directly with the extent to which individuals in that population conform to the patterned and socially sanctioned demands and expectations-that is, social norms-placed upon them by others. Proposition 3: The extent to which individuals in a population conform to patterned and socially sanctioned demands and expecta-

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [297]

tions placed upon them by others varies inversely with the extent to which individuals in the population are confronted with role conflicts. Proposition 4: The extent to which individuals in a population are confronted with role conflicts varies directly with the extent to which individuals occupy incompatible statuses in that population. Proposition 5: The extent to which individuals occupy incompatible statuses in a population varies inversely with the degree of status integration in that population. Hypothesis: The suicide rate of a population varies inversely with the degree of status integration in that population.

Observations consistent with the deduced hypothesis lend support to the untestable propositions; support, however, does not mean the propositions are proved, but only that they are consistent with the results and could be correct. Gibbs and Martin's theory has proved to be predictive to some extent. The implied norms in the propositions, consequently, are partially supported by tests of the derived hypothesis.

NORMS FROM WRITTEN DOCUMENTS

The official documents of formal organizations, incorporated cities, counties, states, provinces, communes, countries, and international treaties and associations (like the United Nations) specify certain laws and rules that govern the people in question. These formally stated laws and rules usually have negative sanctions attached to them which indicate the nature of punishment for violators. Fines or imprisonment may be imposed for violating national laws; firing or laying off may accompany disregard for organizational rules; ostracism may be the fate of violators in certain small, relatively isolated societies; and breaking off diplomatic relations or even war may be the result of treaty violation. No matter what the sanctions or evaluations, sociologists may use statements in official docu- ments to indicate the existence of social norms. Societal laws against killing and stealing, organizational rules on authority

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[298] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

and communication, and international treaties on trade rela- tions may all be treated as social norms.

Actually, any written material can be used to ascertain the existence of social norms. Books, newspapers, magazines, and even the notes for speeches may be relevant for the study of norms. Such material can be scrutinized (e.g., by content analysis) for evaluational statements of should and should not and for statements on positive and negative sanctions.

A limitation in using written documents is that they may be outmoded or may not apply to the group in question. There is always a time lapse between any written document and the present, which may result in laws that are no longer enforced. These laws may still be "on the books," but they are not enforced and may not be known (by police or citizen), and therefore they do not govern the behavior of individuals.

A small inquiry using written documents on societal laws was carried out by Thorsten Sellin (1967) on the relation between homicide and capital punishment. Sellin claims that the effects of the death penalty (and especially its possible deterrence effect) cannot be ascertained by comparing states that have abolished it with states that have retained it. Those that have abolished the death peanalty do, in fact, have lower homicide rates, but the comparison ignores the vast economic, social, and political differences between the retentionist and abolitionist states. These factors, argues Sellin, may account for differences in the homicide rates rather than the abolition or retention of the death penalty.

To partially control (hold constant) economic, social, and political differences, Sellin compared homicide rates for con- tiguous retentionist and abolitionist states. Being contiguous, the states are more likely to have similar characteristics than are noncontiguous states. The underlying rationale is that, if the states that can be compared are different only on whether or not they have abolished the death penalty, then the effects of this factor can be directly determined without being con- founded by the presence of other factors. The use of contiguous states, of course, only partially controls for other factors, because even these states are at least somewhat distinctive.

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [299]

Based on several contiguous state comparisons, homicide death rates (per 100,000 population) were determined for the period, 1920-1963. The comparisons indicate that the death penalty does not affect homicide death rates. Abolitionist and retentionist states cannot be distinguished on the basis of comparing their respective rates of homicide.

DEGREE OF NORMNESS

Is a social norm a quality or a quantity? Is it best to orient social norm measures to statements that are differentiated in kind ("one should be married by the age of 30" or "one should believe in God"), or in degree ("75% of the population feels that one should be married sometime between the ages of 19 to 38" of "63% of the population feels that one should have a moderate to extensive belief in God")?

In general, sociological statements on social norms have stressed their qualitative nature. Either a norm exists for a specified group or it does not. This orientation seems unrealistic in characterizing social groups. Most groups are heterogeneous to some extent, and large groups tend to be characterized by subgroups, which may develop their own social norms. Further- more, as the Jackson model indicates, norms may refer differentially to a range of behavior. People may fluctuate in their intensity of feeling depending on the degree or severity of the behavior in question. These factors suggest that it may be valuable to measure social norms quantitatively to tap the narrow-to-wide variations in their dimensions.

Viewed quantitatively, in terms of the degree of "normness," the following dimensions may be considered: (1) the range of behavior specified, (2) percentage of the group supporting or not supporting a "should" statement, (3) the intensity of feeling associated with a "should" statement, (4) the intensity of feeling associated with different degrees of behavior, (5) the severity of the sanctions attached to the conformity or violation of a behavioral pattern, (6) the degree of conformity to a "should" statement, (7) the extent to which a "should"

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[300] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

statement is known among members of a social group, (8) the degree to which sanctions are uniformly applied, and (9) the degree to which a "should" statement is internalized. Although measuring the degree of "normness" may prove to be a sound strategy, other than the model proposed by Jackson, very little in the way of theorizing or inquiry has been done in this regard.

SUMMARY AND IMPLICATIONS

Each of the four techniques for measuring social norms (questions, inferring from behavior, postulation, and written documents) has obvious biases. Questions may produce invalid answers, because respondents may not know what should or should not be or they may not give truthful answers. Questions, furthermore, may not have the same meaning to each respond- ent. Measurement by inferring from behavior may be biased, because any specific behavior simply may not indicate a social norm. Marriage rates, for example, may increase in June in certain areas, not because one should get married, but because in this month college students begin their vacations. Besides the possibility that behavior may not reflect a social norm, the attempt to explain a behavior by a norm leads to circular reasoning if the norm is first measured by inferring from the behavior in question. A major bias from measurement by postulation is that the implications of hypothetical norms may be in error. Finally, the use of written documents may not adequately specify social norms, because they may be outdated or they may not apply to the group in question.

If two or more of these techniques are used to establish the nature or extent of social norms, then the effects of these biases tend to be cancelled. For example, suppose we want to know whether people in a given community should be church members. We could ask respondents directly ("Should people in this community be a member of a church?"), but a positive answer could be based on what the respondent thinks others in the community would say rather than what he believes. So we

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [301]

may attempt to measure the norm by noting whether, in fact, most people in the community do belong to a church (infer from behavior). If almost all do belong, this could indicate that one should belong. It could also indicate, however, a prevalent personal feeling that each wants to belong without indicating that others should belong. Some may be church members, furthermore, to establish business contacts or to take part in social activities rather than belonging because it is right and proper.

Suppose now we attempt to measure the nature of this potential norm by postulation. If people should be church members, then they are likely to send their children to Sunday school, and they are likely to have a comparatively low divorce rate. We could then observe the implications of such a norm by finding out the frequency of Sunday school attendance and comparing the divorce rate of the community to other community divorce rates where the dominant feeling is not to be a church member. Of course, even if supported, these implications may reflect something other than the social norm in question. Finally, we may check the community newspaper (a written document) and try to discern its sentiment regarding church membership. The newspaper, however, may express the sentiments of only a few in the community. If all four measurements, or any two or three, consistently show the same results (for example, that people should be church members), then confidence can be placed in the conclusion and the validity of the different measures is supported. At this modest stage in the development of sociological knowledge and given the vague conceptualization and treatment of social norms, the use of several different measurement techniques may prove to be a sound research strategy.

NOTES

1. For explications of this view, see Webb et al. (1966), Blalock (1971), Denzin (1970: chs. 3, 5), and Denzin (1971). The essential notion, if supported by sound

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

[302] PACIFIC SOCIOLOGICAL REVIEW / JULY 1973

rationales, is that a theoretical concept requires multiple operations or measurements. A single measure reflects several dimensions or parameters; seldom, if ever, will it reflect just one. Therefore, any particular measure is a function of more factors than just the concept it represents. With regard to the desired dimension, these other factors are biases. Because different measures tend to have different biases, the use of several measures may cancel out their effects. This process of "triangulation," consequently, is designed to reduce or eliminate rival interpretations to the empirical findings.

2. A classic sociological study in which informal work norms were inferred from behavior by the method of participant observation was carried out by Roethlisberger and Dickson (1939).

REFERENCES

Angell, Robert C. 1962 "Preferences for moral norms in three problem areas." Amer. J. of

Sociology 67 (May): 650-660. Blake, Judith and Kingsley Davis

1964 "Norms, values, and sanctions." Pp. 456-484 in R.E.L. Faris (ed.) Handbook of Modern Sociology. Chicago: Rand McNally.

Blalock, Hubert M. 1971 "Aggregation and measurement error." Social Forces 50 (December):

151-165. Converse, Philip and Angus Campbell

1968 "Political standards in secondary groups." Pp. 199-211 in D. Cartwright and A. Zander (eds.) Group Dynamics, New York: Harper & Row.

Denzin, Norman K. 1970 The Research Act. Chicago: Aldine-Atherton. 1971 "The logic of naturalistic inquiry." Social Forces 50 (December): 166-182.

Durkheim, Emile 1968 The Rules of Sociological Method. New York: Free Press.

Garfinkel, Harold 1964 "Studies of the routine grounds of everyday activities." Social Problems 11

(Winter): 225-250. Gibbs, Jack P.

1965 "Norms: the problem of definition and classification." Amer. J. of Sociology 70 (March): 586-594.

--- and Walter T. Martin 1964 Status Integration and Suicide. Eugene: Univ. of Oregon Press.

Glick, Oren W. and Jay Jackson 1970 "Effects of normative similarity on group formation among college

freshmen." Pacific Soc. Rev. 13 (Fall): 236-269. Homans, George C.

1950 The Human Group. New York: Harcourt, Brace.

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions

Labovitz, Hagedorn / MEASURING SOCIAL NORMS [3031

Jackson, Jay 1960 "Structural characteristics of norms." Pp. 131-163 in N. B. Henry (ed.)

Dynamics of Instructional Groups. Chicago: Univ. of Chicago Press. 1964 "The normative regulation of authoritative bahavior." Pp. 213-241 in W. J.

Gore and J. W. Dyson (eds.) The Making of Decisions. New York: Free Press.

1966 "A conceptual and measurement model for norms and roles." Pacific Soc. Rev. 9 (Spring): 35-47.

Morris, Richart T. 1956 "A typology of norms." Amer. Soc. Rev. 21 (October): 610-613.

Roethlisberger, F. J. and W. J. Dickson 1939 Management and the Worker. Cambridge, Mass.: Harvard Univ. Press.

Sellin, Thorsten 1967 "Homicides in retentionist and abolitionist states." Pp. 135-138 in T.

Sellin (ed.) Capital Punishment. New York: Harper & Row. Stoodley, Bartlett H.

1959 "A cross-cultural study of structure and conflict in social norms." Amer. J. of Sociology 65 (July): 39-48.

Stouffer, Samuel A. 1949 "An analysis of conflicting social norms." Amer. Soc. Rev. 14 (December):

707-717. Sumner, William Graham

1906 Folkways. Boston: Ginn. Turk, Herman

1965 "An inquiry into the undersocialized conception of man." Social Forces 43 (May): 518-521.

Webb, Eugene J., Donald T. Campbell, Richard D. Schwartz, and Lee Sechrest 1966 Unobtrusive Measures: Nonreactive Research in the Social Sciences.

Chicago: Rand McNally. Wrong, Dennis

1961 "The oversocialized conceptions of man in modern sociology." Amer. Soc. Rev. 26 (April): 183-193.

This content downloaded by the authorized user from 192.168.72.232 on Thu, 22 Nov 2012 21:20:52 PM All use subject to JSTOR Terms and Conditions