Advances in Consumer Research Volume 6, 1979 Pages 139-145
SOCIAL CLASS DETERMINANTS OF LEISURE ACTIVITY
Robert B. Settle, San Diego State University
Pamela L. Alreck, San Diego State University
Michael A. Belch, San Diego State University
A survey of 975 metropolitan West Coast adults from the general public obtained participation rates for 100 leisure activities grouped as entertainment, competitive sports, non-competitive sports, and hobbies. Participation was cross-tabulated with five socioeconomic variables: self-rated social class, education, income, occupational class and occupational group, and with five demographic variables: age, sex, marital status, employment status, and family life cycle. The results, reported as the trend in participation rate across categories, revealed that (a) the demographics were better predictors than the socioeconomic factors, (b) both groups are systematically related to leisure behavior, and (c) education was by far the best single socioeconomic determinant of leisure choices.
Social class status has long been thought to influence the members' behavior. Casual observation of different social classes leads to the intuitive belief that behavior patterns differ from one class to the next, and several studies have actually detected such differences. Yet there are many confusing issues surrounding both the theoretical construct of social class and the behavior patterns that are supposed to be affected by social status. There is general agreement that social class membership is determined by a person's status on several socioeconomic variables but no clear consensus about what variables must be included or what weight should be placed on each. There are also problems in identifying the behavioral effects of class membership. Even though preferences may differ by social class, the individual's actions may be constrained by situational factors that are not related to social status. The objective of this project was to measure the effect of four variables related to social class on people's leisure behavior.
Social Class and Consumer Behavior
Attempts to relate social class differences to specific consumer behaviors have resulted in less than conclusive results. Martineau (1958) presented information supporting his contention that social class affects perceived risk, choice making, and store selection, among other factors. Levy (1966) has demonstrated differences between classes in respect to values, interpersonal attitudes, shopping behaviors, and media selection, among others. Mathews and Slocum (1969) established social class differences in respect to credit card usage. Rich and Jain (1968), on the other hand, noted little, if any differences between classes in respect to fashion interests, sources of shopper information, interpersonal influences, or other factors influencing shopping behavior.
A number of hypotheses can be formulated to account for the lack of consistent findings of these studies, as well as many others attempting to differentiate behavior patterns on the basis of social class. Inconsistent definitions of social class, confounding influences such as status inconsistency, and the changing structure of American society are but a few factors that may intervene in these studies. This study incorporates a measure of self-rated social class membership with educational status, family income, and two methods of categorizing occupational status, together with the demographic variables of age, sex, marital status, and family life cycle stage. This permits a comparison of the relative potency of the different variables in shaping behavior. The recency of the study may also account for the changing social class structure.
Social Class and Leisure Activities
Most definitions of leisure establish it as time free from work or work related demands. Leisure time is, then, discretionary time, and leisure activities are a matter of individual judgment and undirected choices (Voss, 1967, p. 101). As such, leisure should provide a much better indication of differences in preferences that may exist among social strata. Substantiation for this position can be found in the sociological literature where a number of studies have examined the effects of social class and occupational status on the selection of leisure activities. Without attempting to review all of the literature in this area, a sampling of the studies will provide a perspective.
Reisman (1954), White (1955), Clark (1956), Gerstyl (1961), and Burdge (1969) report similar findings, essentially demonstrating that differences in social class and/or occupational milieu are responsible for differences in preferred leisure activities. The conclusions of these studies suggest that occupational positions lead to differences in leisure activities within strata, that as prestige increases, involvement in the variety of social activities also increases in a linear relationship, and that individuals are more likely to engage in activities perceived as consistent with their social standing. MacDonald et. al. (1949) provided supporting evidence contending that children use leisure to prepare for further upward mobility.
Studies by Kaplan (1960), Cunningham et. al (1970), Murphy (1974), and Kelly (1975), using the same basic paradigm, provide contrary information to the studies just reported. These studies found that neither social class nor occupational prestige was an effective indicator of leisure preference. In summary, the conflicting results of these studies prohibits the establishing of firm conclusions. One possible explanation is evident, and that is the dynamic nature of contemporary American social structure. The majority of early studies report differences among social classes, while more recent studies provide findings of homogeneity among strata. This study may provide additional insight into the existence or absence of class differences in behavior today, based on a very recent measure of current behavior patterns.
Classification of an individual into a given social strata is relatively easy and unambiguous when several of the determining variables are all consistent with one another. For example, a person whose income, wealth, education, and prestige of occupation were all at about the fifth percentile for that society could be classified as lower class without much equivocation. It also seems likely that the individual's self-rating of social class would be to that category. When, on the other hand, the variables regarded as determining social class membership differ widely from one another, this status inconsistency may make classification difficult, if not impossible. How should a person whose income is in the sixtieth percentile and whose occupational prestige is in the tenth percentile be classified? Even if weights are established and a weighted average used for classification, the variance in status from one variable to the next may tend to blur the person's own perceptions of class membership. Such an individual might not be expected to behave in the same manner as another member of the same strata whose strata on the determining variables are identical.
Such factors as the increase in American affluence, the progress of labor organization resulting in higher wages for the working class, and other similar changes in the socioeconomic system may have increased status inconsistency to the point where the conglomerate concept of "social class" has less meaning than more unitary social status variables. Examination of the various forces leading to status inconsistency is beyond the scope of this project, however, the relative influence of some variables commonly regarded as determining social class can be measured empirically.
I. Patterns of leisure behavior will be systematically related to self-rated social class, income, education, and occupational status.
II. Education will exert the strongest influence on leisure activity and income will exert the weakest influence, of the individual variables.
To test the hypotheses, a survey questionnaire was individually self-administered to a convenience sample of about one thousand metropolitan West Coast adults from the general public. Field workers were assigned data collection quotas based on age, sex, and occupation to insure that the demographic distributions of the sample would correspond to census data for the same geographic region.
Each data collection instrument consisted of a cover letter explaining the nature of the project, a sixty-four item, trait-specific personality test, a series of one hundred items about leisure activities, ten common pastimes, twenty-nine questions about habits of time use, and eleven demographic items. The personality test and items relating to time patterns are the object of another study and are not reported here.
The one hundred leisure activity items were divided into four groups of twenty-five each, and classified as Entertainment, Competitive Sports, Non-Competitive Sports, and Hobbies, respectively. The activities that were the content of the items are shown in Tables 1 through 4, in the following section. Respondents were asked how many times per week, month, or year they participated in each of the activities in the first three categories. For the hobbies, they indicated the number of hours per day, week, or month they engaged in each, and this same measure was obtained for the ten pastimes listed. The demographic items were listed by category, except for education, occupation and income. Respondents reported the last year of school they completed and these data were later classified into categories. They filled in their occupational title and a phrase describing what they did on the Job, and these data were later classified into occupational class and group categories. Approximate annual family income was recorded on an optional basis in thousands of dollars, and these too, were converted to categories for analysis. Lastly, each respondent was classified into a stage of the family life cycle, based on marital status, age of youngest child, age of respondent, and employment status.
Nine hundred ninety-six questionnaires were equally divided among eighty-three student field workers who collected the data as part of a Consumer Behavior class project. Each received about two hours of instruction on the data collection and instrument prior to going into the field, and each had completed the brief psychological test included in the questionnaire and received their own individual, six-page, computer-generated interpretations. The field workers only tasks were to select respondents, enlist their cooperation, deliver and retrieve the questionnaire, and transfer the data to code sheets for optical scanning. Instructions within the questionnaire were addressed to the respondent and the instrument was individually self-administered. As an inducement to participate, the field workers explained the personality test to the potential respondent and each was provided with his or her own interpretations within a few weeks after completing the questionnaire. These interpretations, mailed directly to the respondents' homes, provided one hundred percent validation of the data collection.
Of the 996 questionnaires sent into the field, 984 were recovered. Seven questionnaires were rejected because they were not substantially complete, and two others were lost due to errors in data transfer, leaving a total sample size of 975 respondents.
Statistical analysis for this report focused only on the 100 leisure activities and the social class and demographic variables. The data description for the independent variables is provided in the form of frequency and percentage distributions of response, to reveal the nature of the sample obtained. Leisure activity items were converted to dichotomous variables according to whether or not the respondent had participated in the activity. These were then cross-tabulated with the discrete social class and demographic items and the Chi square statistic used to measure statistical significance at the .05 alpha level.
While sacrificing much of the detail contained in the data, the procedure for analysis serves to condense and abbreviate the results for clarity of presentation. Even so, there remained exactly 1,000 relationships to portray. Time and space limitations demanded a novel yet easily understood format. Each of the contingency tables was checked for statistical significance. For those tables which were significant, the percentage of each category that reported participating in the activity was noted.
In this report, the tables depicting these relationships are in matrix form with the rows defined by the various leisure activities within one category and the columns defined by either a social class or a demographic variable. In the body of the table, each cell represents the relationship between one dependent and one independent variable. The content of the cell indicates (a) whether or not the relationship was statistically significant at the .05 alpha level, and if so, (b) the direction of the relationship. If the relationship between the social class variable and the participation in the activity is not significant, the cell is blank. If the relationship is significant, the cell contains a single line.
The direction of the relationship between the dependent and the independent variable for a given cell is shown by the slope of the line. If the rate of participation in the activity tended to increase for higher levels of the social class or demographic variables, the line has a positive slope. If the rate of participation tended to decrease as the independent variable ascended, the line has a negative slope. For those contingency tables where the rate of participation fluctuated significantly across the categories of the independent variable but no clear and consistent trend was discernible, the cell contains a horizontal line. The six categories of employment status defy arrangement into a hierarchy, and consequently, a horizontal line was used to indicate a significant relationship without regard for direction of trend. The other independent variables ascend in the order in which the categories are presented for data description, from lowest to highest income, education, age, or family life cycle, etc. Sex ascends from male to female and marital status from single to married (though the arrangement is admittedly an arbitrary choice). The use of this unique report format permits portrayal of both the number of significant relation- ships and the direction of the relationships. Comparison of columns reveals the relative potency of each independent variable, and inspection of rows shows the nature of activities that are influenced.
The direction of the significant relationships between social strata and demographic variables, on the one hand, and the dependent variables of participation in the various leisure activities, on the other, are shown in Tables 1 through 4. The activities classified as entertainment, competitive sports, non-competitive sports, and hobbies will be discussed in turn.
Inspection of Table 1 indicates that the respondents' level of education was significantly related to participation in 20 of the 25 entertainments listed. The rate of participation increased with education for 18 of the 20 significant items, while visits to a nightclub, lounge or bar appeared to increase to middle levels of education, then decrease for higher levels. The relationship between education and visiting a swap meet was inverse; the higher the level of education, the less likely the respondent would participate.
Eleven of the 25 entertainment items provided a significant relationship to self-rated social class and all but one of the 11 were also systematically related to education. The direction of the relationships were identical for all but one; going out to dance was more often reported for lower and higher class respondents and less often reported for middle class people. The two occupational categories were identical in their patterns of systematic relationships with leisure entertainment activities, except that visits to swap meets were related to occupational groups but not to occupational classes. This seems to indicate that, for entertainment activities, the two-category classification of white collar/ blue collar proved to be as indicative as the finer increments: semi- and unskilled, skilled labor, technical and clerical, or professional. A total of 10 of the activities were significantly related to occupational groups and 9 to occupational class.
Income was systematically related to participation in only 8 of the entertainment activities listed. The only leisure activity relationship uniquely related to income was visits to a card room, casino or bingo game. There were 57 significant relationships between social strata variables and participation in leisure activities classified as entertainment (while only about 6 or 7 would be expected by chance at the .05 alpha level). These re-suits furnish support for the first hypotheses, that social stratification would be systematically related to participation in leisure activities. The second hypothesis, that education will provide the best indication of leisure activity and income the least, was also supported by the results.
RELATIONSHIPS BETWEEN SOCIAL STATUS AND ENTERTAINMENT
Of the demographic items, age was significantly related to 21 of the entertainment activities, as was family life cycle. The sex of the respondent was systematically related to only seven activities. In total, there were 72 significant relationships between demographic items and entertainment activity participation.
Competitive Sports Participation
The significant relationships between social strata and demographic variables and competitive sports participation are shown in Table 2. Only 21 of the 25 sports activities are shown because the last four items were for "other" activities to be written in by respondents. None provided a sufficient frequency for contingency table analysis.
RELATIONSHIPS BETWEEN SOCIAL STATUS AND COMPETITIVE SPORTS
The pattern of relationships revealed by Table 2 indicates that the demographic items were much more strongly related to participation in these sports activities than were the social strata items, by a ratio of 87 to 37. Nearly half of the significant associations were with education, while both income and occupational group each accounted for only 5, and occupational class was the least effective predictor of activity with 4 systematic relationships to the 21 dependent variables. As with the previous table, these results provide support for both of the hypotheses.
Non-Competitive Sports Participation
Table 3 contains the report of the association between participation in non-competitive sports and respondent status on the demographic variables and those regarded as determinants of social status. As with competitive sports, age, family life cycle, and employment status were strongly related to participation. There were 37 significant relationships between social status variables and the activities. Of those, 18 were with education, 8 with income, 5 with occupational group and 3 each with self-rated social class and occupational class.
RELATIONSHIPS BETWEEN SOCIAL STATUS AND NON-COMPETITIVE SPORTS
For both competitive and non-competitive sports, grouping occupations into four categories rather than just two provided only very slightly more prediction of participation in the activities. The number of significant relationships shown on Table 3 provides support for Hypothesis I. The second hypothesis received partial support in that education was the best predictor of behavior, however income was not the least potent, as hypothesized. Examination of non-competitive sports items reveals that many require expensive equipment or fees, which may account for the relationships to income. This was not the case for competitive sports included in the previous table.
Participation in Hobbies
The systematic relationships between participation in 22 hobbies and the independent variables are shown in Table 4. As with the others, the demographic variables provided more influence on participation than did the items associated with social status. In all, there were 17 significant relationships between participation and the social status measures (while only 5 or 6 would be expected purely by chance at the .05 level of significance). Only one hobby, clothing design and sewing, was systematically related to self-rated social class. Collecting things was related to income, and adult education participation was significantly related to occupational class, while four hobbies were related to occupational group.
RELATIONSHIPS BETWEEN SOCIAL STATUS AND HOBBIES
As with the other categories of leisure activity, education was by far the best single predictor, significantly associated with 10 of the hobbies. Income shared the weakest position with self-rated social class and occupational class. These results provide support for both of the hypotheses.
Summary of Results
Considering only those independent variables associated with social status, the educational level of the respondent was by far the most effective determinant of leisure activity, with significant relationships with 65 of the 91 activities listed. In descending order of effect, the other variables were occupational group with 24 relationships, income with 22, self-rated social class with 21, and occupational class with the least number, 17. The total number of significant associations was 149 of a possible 455. At the .05 alpha level, one would expect only about 23 to be significant purely by chance. These results provide support for the first hypothesis, that the patterns of leisure activity will be significantly influenced by variables associated with social status. It was also hypothesized that education would provide the best single predictor of participation in the leisure activities, while income would be the least effective predictor. This hypothesis received only partial support. While education was the best predictor, both occupational class membership and self-rated social class proved to have fewer relationships to the dependent variables than did income.
Examination of the direction of the relationships indicates that for the vast majority of activities, participation tends to be less for lower levels of social status and greater for higher levels. There were a minority of activities which showed a curvilinear relationship, where participation increased from lower to middle levels of social status, then decreased for the highest levels. There were a few activities, such as pool or attending a swap meet, that were more typical of lower levels than of higher levels of social status, but these were intuitively understandable.
The patterns of relationships between participation in leisure activities and the independent variables are very consistent with prior expectations in many respects. For example, demographic factors of age, sex, and family life cycle provide significant relationships with those leisure pursuits that demand strenuous physical activity as one would expect. These kinds of observations about the results are, of course, only "face" validity, but they do encourage the belief that the measurements are reliable and valid. Face validity is unimportant when present but very important when it is absent.
Education versus Income and Occupation
The major objective of this study was to measure the effect of four variables indicative of social status on consumers' selections of leisure activity. The study also sought to compare the relative potency of each in shaping behavior. Education proved to be the most influential variable, by far. Why might this be so?
In pursuit of the answer to this question, one might logically ask how education differs from the other variables. One very fundamental difference lies in the distinction between an indicator of social status and a determinant of social status. If a variable is an indicator, this implies only a correlation, but if it is a determinant, that implies causality. For example, in many communities the size of a person's front lawn is an excellent indicator of social status, but this is only an indicator and not a determinant. If a person from a high social station were to acquire a home with a very small front lawn, he would not sacrifice social status, he would only decrease the predictability of the indicator.
Self-rated social class status can be declared an indicator and not a determinant of social class without much danger of conflict. Few would argue that a self-perception of a particular social class membership would, in fact, propel the person to that particular station. This is not the case, however, for income and occupation or occupational prestige. These are often regarded as true determinants of social status. The contributions upon which this society is supposed to accord social status often result from one's occupation, and the relative value of the contribution is thought to be denominated in terms of monetary rewards, or income.
But what happens when income deviates from an accurate reflection of social contribution? Both income and the occupation providing it might remain as indicators of social class, out of inertia, however they would no longer be determinants of social status.
Several reasons can be cited for deviation of income from an accurate reflection of social contribution in our contemporary society. First, the socioeconomic system often uses income to compensate for occupations that are socially unattractive, such as when trash collectors are paid more than school teachers. Second, occupational organization and political power may force income disproportionate to the long-term social contributions of the occupation. Third, some vocations may provide very valuable social contributions but they may be so attractive in terms of intrinsic rewards that many will pursue them without demanding monetary rewards commensurate with contributions. To the degree that these conditions describe society today, income and occupation can be expected to diminish in importance as determinants of social status.
Education as a Direct Influence
To this point, it has been assumed that (a) social status influences consumer choices and (b) education is an accurate indicator or determinant of social class. Thus, the paradigm is one of a mediationist and the social class construct is an intervening variable between the educational process and overt behavior. An alternate perspective is that education directly affects choice behavior and at the same time, determines social status. (An even more radical view would be that education affects choices which, in turn, determine social status; an idea that Thorstein Veblen would have savored.) Regardless of one's choice of models, there is ample evidence that education is more of a developmental process than the adult pursuit of an occupation or the acquisition of income. We might expect the educational process to shape the tastes and values of the individual, directly changing preferences for the various forms of leisure activity. This proposition is not intended to deny the existence or effects of social stratification, but it does serve as a reminder that all influences on choice behavior are not mediated by social class membership.
Reconciliation and a Projection
The results of the earlier studies of social class and choice behavior indicated a significant relationship. Later studies were less successful in detecting a strong relationship. The results of this study indicate that there are probably still many differences among social strata; however, the measurement of the strata themselves should probably be based on variables other than income and occupational class or group, as in the past. It may prove fruitful in future research to first explore the relative effects of several candidate variables in determining social status, selecting those that perform the best. Having done so, the second step in the process would be to define strata in terms of these factors. Lastly, the investigation of the effects class membership might be studied in terms of choice behavior. Research experience in the area of social stratification over the past several years, including this study, seem to offer this lesson: in a dynamic setting with an ever-increasing velocity of change, the clarity of yesterday's identities and relationships may fade very quickly into ambiguity and obsolescence.
INDEPENDENT VARIABLE DATA DESCRIPTIONS
R. J. Burdge, "Levels of Occupational prestige and Leisure Activity," Journal of Leisure Research, 1 (Summer 1969), 262-274.
A. C. Clarke, "The Use of Leisure and its Relation to Levels of Occupational Prestige," American Sociological Review, 21 (June 1956), 301-307.
D. A. Cunningham, H. J. Montoye, H. L. Metzner, and J. B. Keller, "Active Leisure Activities as Related to Occupation,'' Journal of Leisure Research, 4 (Spring 1970), 104-111.
J. E. Gerstyle, "Leisure, Taste, and Occupational Milieu,'' in Erwin O. Smigel, ed., Work and Leisure. New Haven: College and University Press, 1963.
M. Kaplan, "The Use of Leisure," in C. Tibbits, ed., Handbook of Social Gerontology. Chicago: University of Chicago Press, 1960.
J. R. Kelly, "Life Style and Leisure Choices," The Family Coordinator, (April 1975), 185-190.
S. J. Levy, "Social Class and Consumer Behavior," in Joseph W. Newman, ed., On Knowing the Consumer. New York: John Wiley and Sons, 1966.
M. MacDonald, C. McGuire and R. J. Havighurst, "Leisure Activities and the Socioeconomic Status of Children," American Journal of Sociology, 59 (May 1949), 505-519.
P. Martineao, "Social Classes and Spending Behavior," Journal of Marketing, 23 (October 1958), 121-130.
H. L. Mathews and J. W. Slocum, Jr., "Social Class and Commercial Bank Credit Card Usage," Journal of Marketing, 33 (January 1969), 71-78.
J. Murphy, "Leisure Determinants of Life Style," Leisure Today, (November-December 1974), 3-5.
L. Reismann, "Class, Leisure and Social Participation," American Sociological Review, 19 (February 1954), 76-84.
S. V. Rich and S. C. Jain, "Social Class and Life Cycle as Predictors of Shopping Behavior," Journal of Marketing Research, 5 (February 1968), 41-49.
J. Voss, "The Definition of Leisure," Journal of Economic Issues, 1 (June 1967), 91-106.
R. C. White, "Social Class Differences in the Use of Leisure," The American Journal of Sociology, 61 (1955), 145-150.