Journalof YouthandAdolescence, Vol. 12, No. 4, 1983
Social Support Factors and Drinking Among College Student Males M a r k R . F o n d a e a r o ~.3 and Kenneth H e l l e r 2
Received January 25, 1983
The present investigation focused on social support and social competence among male college freshmen and the relation o f these variables to alcohol use and psychological adjustment. Recent critical analyses o f the social support literature suggest that studies in this area have generally failed to distinguish between different modes o f support. Therefore, measures pertaining to possible dimensions o f the social support construct (i.e., social network characteristics and perceived social supporO were administered to 137 male college freshmen, along with a measure o f social competence, and these data were factor analyzed. As a result, three interpretable factors were identified: Network Functions, Perceived Intimacy~Support, and Social Competence. Measures representing social network characteristics (e.g., network size, density, amount o f social contact), perceived support, and social competence were used to predict alcohol use and psychological symptomatology. Results indicated that alcohol use was positively related to social network characteristics that reflect high levels o f social interaction (e.g., network density, amount o f social contact) and measures o f social competence. Drinking was not significantly related to measures o f perceived social support. Psychological symptomatology was negatively related to measures o f perceived support, social competence, and network density. Thus, this study concludes that different modes o f support and different measures o f psychological adjustment should not be treated as if they are IDepartment of Psychology, Indiana University. Doctoral candidate in clinical psychology. Special interests are family and peer influences on juveniledelinquency and alcohol abuse. 2Professor of Psychology, Indiana University. Received Ph.D. in psychology from Pennsylvania State University. Current interests include the role of social ties and social support in coping with stress. 3Correspondence should be sent to Mark R. Fondacaro, Department of Psychology, Indiana University, Bloomington, Indiana 47405. 285 0047-2891/83/08 00-0285503.00/0 © 1983Plenum Publishing Corporation
286
Fondacaro and Heller
equivalent. A n d this study reaffirms a growing concern that the social context provides frequent opportunities f o r alcohol use and abuse in a college community.
INTRODUCTION High rates of alcohol use and abuse have become prevalent on college campuses throughout the United States. In a nationwide study of college student drinking, Hanson (1974) found that 80% of the males and 73% of the females regarded themselves as drinkers. Weschler (1979) reported that 66% of male college students randomly sampled from 34 New England colleges reported drinking at least weekly, and more than 12% of seniors drank daily. The frequency of drinking among Wechsler's female subjects was lower. Still, 50% of females drank weekly and 2% drank daily. College students drink for a variety of reasons, ranging from the desire to enhance and facilitate their social interactions, to using alcohol to help cope with academic and peer pressures (Jessor et al., 1968; Fondacaro, 1980). Apparently, drinking has become intimately woven into the social fabric of college life. Moos et al. (1976) found that heavydrinking college students engaged in more frequent social interactions than moderate drinkers or abstainers. In a later study they found that heavy drinkers who reduced the frequency and quantity of their alcohol use tended to disengage from the college community (Moos et al., 1977). The results reported by Moos and associates suggest that college students who have extensive social networks may be at increased "risk" for alcohol abuse. This finding is particularly striking, given that literature in the area of social support has emphasized the benefits of social support networks in buffering individuals from stress and reducing their risk for psychological and physiological disorders (Caplan, 1974; Cassel, 1975; Cobb, 1976). Numerous empirical studies have attempted to demonstrate the stressbuffering effect of "social support" for both clinical and nonclinical populations (e.g., Nuckolls et al., 1972; Tolsdorf, 1976; Gore, 1978; Hirsch, 1979; Wilcox, 1981). Although initial reviews of the social-support literature were quite optimistic (Cobb, 1976; Dean and Lin, 1977; Gottlieb, 1979), recent critical analyses suggest that many of the studies in this area are marred by serious methodological and conceptual flaws (Hirsch, 1979; Cohen and McKay, in press; Heller and Swindle, 1983; Thoits, 1982). Chief among the conceptual problems is the lack of a specific definition of the social support construct. Often "social support" is
Social $up0ort Factors
287
operationally defined quite differently across various studies, making it very difficult to interpret and compare findings. Terms such as "social support," "social network," and "social support system" are used as if they were equivalent, when in fact they may be tapping different facets of the support construct. In an attempt to more clearly define the social support construct, Heller and Swindle (1983) have proposed a model of social support and coping that emphasizes conceptual distinctions among several facets of the support construct. Their model is based largely on an integration of findings and a clarification of concepts in the areas of stress, support, and coping. Conceptual distinctions are made between social networks, perceived social support, and social competence. A social network is defined as the set of all significant others with whom one has social interactions (Hirsch, 1979). According to Marsella and Synder (1981), social networks can be characterized by four dimensions, each of which may be independently related to support, coping and psychological adaptation. The first dimension, structure, includes morphological variables such as the individual's position within a network, amount of social contact with network members, network size and network density (i.e., the number of dyadic relationships in a network in proportion to the number of possible relationships, given the network size). The dimension of interaction refers to variables that describe the relationship between various network members, including reciprocity and directionality. A third dimension, quality, describes affective characteristics of relationships such as intimacy and level of friendship. Finally, the function dimension describes specific functions served by network members such as companionship, problem-solving assistance, material aid, information, and emotional support. Heller and Swindle (1983) make a major distinction in their model between social networks and perceived social support; the latter is intended to describe the subjective impact of helping behavior on the individual. Perceived social support is based in part on the individual's perception of his/her social network and is defined as the cognitive appraisal that one's needs for support, information, and feedback are provided for by network members. Defined in this way, an individual may experience high levels of perceived social support without being a member of a large or active social network. The person's needs for support, information, and feedback may be met by a few close or intimate relationships. On the other hand, individuals who have large social networks and engage in frequent social interactions may' not necessarily feel supported. In support of this hypothesis, Corty and Young (1980) found that perceived support was not
288
Fondacaro and Heller
related to the amount of social contact with network members, but was inversely related to feelings of loneliness. Social networks and social support have usually been regarded as environmental variables. Heller and Swindle (1983) argue that an individual's level of perceived support also is dependent on personal characteristics such as social competence. Socially competent individuals are regarded as possessing the necessary skills for initiating and maintaining supportive relationships. In their model then, perceived social support (the appraisal that one is supported) is influenced both by social connections provided by the environment (social networks) and person variables that aid in the development of supportive relationships. The present study was guided by a belief that social network characteristics and perceived support are independent facets of the support construct. To test this assumption, a questionnaire that included measures of social network characteristics, perceived social support, and social competence was administered to a sample of male college freshmen. Information obtained from these measures was factor analyzed. It was hypothesized that (1) separate factors would emerge for each of the support dimensions (networks and perceived support) and also for social competence. Following the assumption that drinking occurs in college social settings primarily to facilitate social interaction, it was hypothesized that (2) alcohol use among freshmen males would be positively associated with network variables that reflect high levels of social interaction (e.g., network size, density, and amount of social contact). Subjects completed a measure of psychological symptomatology (The Brief Symptom Inventory, or BSI) in addition to measures of alcohol use. Traditionally, studies of the effects of social support on adjustment have used such measures, and generally found that those with high support report good adjustment. However, Heller and Swindle (1983) postulate that the relationship between reported symptomatology and perceived social support should be higher than the relationship between symptomatology and social network variables, since both symptomatology and perceived support can be considered part of the appraisal process. Thus, it was hypothesized that (3) psychological symptomatology would be negatively related to perceived support, but would not be significantly correlated with network measures reflecting high levels of social interaction (network size, amount of social contact). Finally, it was expected that socially incompetent individuals would lack the necessary skills to initiate and maintain supportive relationships, and would subsequently be at greater risk for psychological problems. Therefore, it was predicted that (4) social competence would be negatively associated with psychological symptomatology.
Social Support Factors
289
METHOD Subjects The subjects were 137 freshman male undergraduates at Indiana University. Subjects were recruited from an introductory psychology subject pool and received required experimental credits for their participation. Ten subjects failed to complete the questionnaire according to instructions and were, therefore, excluded from subsequent data analyses. Freshmen were chosen for study because it was expected that the transition from high school to college would be a stressful life event for many students, taxing their already developed coping patterns. The study focused on males because of their general neglect in the social support literature.
Materials Subjects were administered a questionnaire battery that included the following materials: Personal Information Sheet. This portion of the battery contained questions concerning demographic background (e.g., school year, employment status, present living situations), parental drinking practices, heterosexual intimacy, self-report of current drinking practices, and quality of relationships with father and mother. PerceivedSocial Support. This scale contains two subscales, each with 20 items, which measure the extent to which persons perceive that they are receiving adequate levels of support from friends (PSS-Fr) and family members (PSS-Fa). Reliability and validity data are reported in Procidano and Heller (1983). Social Competence Questionnaire. This situation-specific measure of social competence contains two nine-item scales that evaluate an individual's self-reported competence in handling dating and assertion situations. Reliability and validity data are reported in Levenson and Gottman (1978). Michigan Alcohol Screening Test (MAST). The MAST is a measure of problem drinking that has been used extensively in research on alcohol abuse among adults (Seizer, et ak, 1975). According to Seizer eta/., a total score of 0-4 points indicates nonproblem drinking, 5-6 suggests problem drinking, and 7 or higher indicates problem drinking. Social Relationship Questionnaire (SRQ). The SRQ includes measures of social network variables reported in Hirsch (1979) and in Mitchell and Trickett (1980) and Tolsdorf (1976). Fourteen network variables, representing three of the four network dimensions outlined by Marsella and
290
Fondacaro and Heller
Snyder (1981) (Structure, Quality, and Function), were included in this study. They are as follows.
Structure (1) Size. Subjects were asked to list the first names or initials of up to 15 individuals wi[h whom they interacted at least once during a two threeweek period. Size was calculated by summing the number of people listed. (2) Density. Subjects were asked to list the individuals in their social network in matrix form, placing an "X" to indicate where individuals were likely to interact with each other during a,two-three-week period. Density X was computed in the following way: Density = N ( N - 1)/2, where X = the number of linkages in the network and N = the network size. (3) Proportion of males listed in the network. (4) Proportion of network members who werefemales. (5) Proportion of family members listed in the network. (6) Amount of Social Contact (ASC). For each network member, subjects estimated the number of hours, rounded to the nearest half-hour, that they spent with that person over the last seven days. ASC was computed as the sum of these ratings across all network members.
Quality (7) Proportion of Intimate Relationships (INT). Subjects rated the intimacy or closeness of their relationship with each network member on a 4-point scale. INT = sum of intimacy ratings across all network members/N. (8) Proportion of Very Intimate Relationships (V-INT). V-INT represents the proportion of relationships that were given the highest intimacy rating by the subject.
Function (9) Companionship Support (CS). Subjects rated each member of their network on a 4-point scale (ranging from 1 = very unlikely to 4 = very likely) in terms of whether they would seek the individual's companionship if they wanted to enjoy themselves and have fun. (This same 4-point rating scale was used to compute the emotional support, informational support, material support, and problem-solving support variables.) Companionship support was calculated in the following manner: CS = sum of companionship ratings across all network members/N. (10) Emotional Support (ES). Subject rated whether each member of his network would listen to him when he was feeling down, encourage him, be understanding, just be with him. ES = sum of emotional support ratings across all network members/N. (11) Informational Support (IS). Subject rated whether each member of his social network was someone who would be likely to tell him about
Social SupportFactors
291
things that are good to know, like new places to go or places to get help. IS = sum of informational support ratings across all network members/N. (12) Material Support (MS). Subjects rated network members in terms of their likelihood of providing material aid (e.g., money, help in moving) if a subject needed it. MS = sum of material aid ratings across all network members/N. (13) Problem-Solving Support (PS). Subjects rated whether each network member would help them solve personal problems, give them advice, help them explore alternatives. PS = sum of problem-solving ratings/N. (14)Multiplexity (MP). MP was computed as the proportion of multiplex relationships in an individual's social network. Relationships that served only one function for the individual (e.g., companionship) were regarded as unifunctional. Relationships that were rated as likely or very likely to serve more than one function O. e., companionship, emotional support) were scored as multiplex. Thus, multiplexity was calculated in the following manner: MP = number of multiplex relationships/N. The following two measures were also included in the SRQ to assess the drinking status of network members.
Proportion of Heavy Drinkers (HD). HD = the proportion of network members who were rated by the subject as either heavy or problem drinkers. Proportion of Drinkers (Drink). Drink = the proportion of network members who were rated by the subject as drinkers.
Dependent Variables
Quantity-Frequency Index (Q-F). The Q-F index is a measure of daily alcohol consumption. It has been used extensively in research concerning alcohol use and abuse and is reported in Cahalan and Cisin (1968). Brief Symptom Inventory (BSI). The BSI is a 53-item measure of reported psychological symptoms. Respondents were asked to rate on a 5point scale how much discomfort each symptom has caused them during the previous week. Information on the psychometric properties of this inventory are reported in Derogatis et al. (1973). Self-Reported Drinking Status (SRDS). Subjects were asked to rate their own drinking status on a 5-point scale ranging from I = abstainer to 5 = problem drinker.
292
Fondacaro and Heller Procedure
Subjects completed the battery of questionnaires during a two-hour testing session. Two testing sessions were held; the first session was attended by 90 subjects, and the second session was attended by 47 additional subjects. All questionnaires were completed anonymously to ensure the confidentiality of the subjects' answers.
RESULTS Twenty-two variables (all pertaining to either social networks, perceived support, or social competence) were subjected to a common factor analysis (SPSS version PAZ; Nie et al., 1975). Seven factors, each with an eigenvalue greater than 1.00, accounted for approximately 69070 of the total variance. Six of these factors were extracted and subjected to an orthogonal varimax rotation. As a result, four factors with factor contributions greater than 1.00 emerged. Factor structures were based on variables loading 0.35 or greater on a rotated factor. The four factors, the variable loadings, and the percentage of the reduced variance accounted for are presented in Table I. Factor 1 accounts for 37.1 070 of the reduced variance and consists of variables representing "network functions" (i.e., companionship, emotional support, material support, problem-solving support, and multiplexity). Factor 2 accounts for 20.6°70 of the variance and consists of variables associated with perceived social support and intimacy (e.g., perceived peer support, closeness of relationships with parents, and proportion of intimate relationships with network members). Factor 3 accounts for 16.1 070of the variance and consists of two redundant variables: the proportion of males in one's network (positive loading) and the proportion of females in one's network (negative loading). This was labeled the Residual factor. Finally, variables associated with social competence (e.g., dating competence, assertiveness) loaded significantly on Factor 4, accounting for 11.4070 of the variance. Thus, Network Functions and Perceived Intimacy/Support emerged as separate factors or dimensions of the social support construct, and both were differentiated from social competence. Although network size, density and amount of social contact did not load significantly on any of the interpretable factors, they did correlate significantly with the dependent variables and were therefore included in subsequent correlational and regression analyses. The range of scores on the MAST revealed the pervasiveness of alcohol abuse in college. Over 35070 of the subjects received scores which suggested problem drinking (i.e., 5 or greater). Since the MAST was not
Social Support Factors
293
Table I. Social Network, Perceived Support, and Social Competence Variables and Factor Loadings Factor and variance accounted for
Variable
Loading
1. Network functions (37.1%)
Multiplexity "Emotional support Informational support Problem support Material support Companionship support
0.75 0.74 0.72 0.72 0.65 0.46
2. Perceived intimacy/support (20.6%)
Network intimacy Very intimates Intimacy with father Intimacy with mother Perceived support - friends Perceived support - family
0.70 0.64 0.61 0.53 0.49 0.45
3. Residual (16.1)
Proportion of males Proportion of females
0.96 -0.96
4. Social competence (11.4%)
Dating competence Assertiveness Heterosexual intimacy
0.86 0.58 0.39
validated on a college population, this measure may not accurately reflect the extent of alcohol abuse in a college sample. However, roughly the same proportion of subjects (35~/0) reported drinking the equivalent of 1.5 ounces of absolute alcohol per day on the Q-F. Approximately 10~o of subjects reported drinking an average of over 4 ounces of absolute alcohol per day. Pearson product-moment correlations between the network, support, and competence variables and the alcohol use and psychological symptoms variables are presented in Table II. The variables most highly correlated with self-reported drinking status were HD (proportion of network members who are heavy or problem drinkers), r = 0.44, p < 0.001, and Drink (the proportion of network members who are drinkers or "nonabstainers") r = 0.40, p < 0.001. As predicted by our second hypothesis, self-reported drinking status (SRDS) was positively associated with Density, r = 0.24, p < 0.005, and Amount of Social Contact (ASC), r = 0.19, p < 0.05. However, SRDS was not significantly correlated with network size, r = 0.09, p > 0.05. This correlation probably does not accurately reflect the degree of association between these variables, since over 57 % of the subjects listed the maximum number of network members requested, 15. Thus, the range of the network size variable was apparently truncated. There was a marginally significant negative association between self-
Fondacaro and Heller
294
Table ii. Correlations Between Network, Support, and Competence Variables, and Alcohol Use and Psychological Symptoms Self-reported drinking
Quantity/frequency of drinking
Brief symptom inventory
0.09 0.24b 0.06 -0.14 c 0.19 c 0.07 -0.09 -0.17c -0.13 -0.14 c 0.40 a 0.44 a 0.10 0.16 c
0.13 0.14 c 0.00 -0.16 c 0.26b 0.15 -0.06 -0.07 -0.07 -0.11 0.26b 0.36a 0.16 c 0.11
0.04 - 0.17 c -0.13 0.10 -0.08 0.04 -0.01 -0.09 -0.04 -0.03 -0.15c 0.04 0.03 -0.06
Perceived support Perceived Support - Family Perceived Support - Friends
0.00 0.12
0.02 0.08
-0.20 c -0.24 b
Social competence Dating Competence Assertion Competence
0.10 0.04
0.17c 0.10
-0.18 c -0.30 b
Network variables Size Density Males in Network Family in Network Amount of Social Contact Companion Support Information Support Problem support Emotional Support Multiplexity Drinkers Heavy Drinkers Intimates in Network Very-Intimates in Network
ap < bp < Cp
0.05, or problem-solving support, r = - . 0 7 , p > 0.10. Neither the Q-F index nor the drinking status variables were significantly correlated with either perceived family support (PSS-Fa) or perceived peer support (PSS-Fr). In summarizing these results, it appears that the likelihood of drinking increases with the number of drinking and heavy-drinking network members, the frequency of social contacts, network density, the proportion of close or intimate relationships, the dating competence of the individual, and, to a lesser extent, the number of network members. There is an inverse relation between drinking and the proportion of network members who are also family members. Moreover, individuals who characterize themselves as frequent or heavy drinkers are less likely to seek out problem-solving assistance or participate in multiplex relationships than their peers, even though they characterize their relationships as more intimate. Inspection of the correlation coefficients in Table II indicates that various dimensions of the support construct are differentially related to alcohol use and psychological symptomatology. Among the more striking relationships is the finding that network density, proportion of drinkers, and dating competence are positively related to the drinking measures and negatively related to the Brief Symptom Inventory. Moreover, the BSI was not related to social network variables (ASC, INT, FAM) that were significantly correlated with the drinking measures. As predicted in our third hypothesis, the BSI was negatively related to the perceived support scales (PSS-Fa, PSS-Fr), both of which were not significantly related to either of the drinking variables. As predicted in our fourth hypothesis, the PSI was negatively associated with both measures of social competence. Stepwise multiple regression analyses were employed to examine the combined effects of social network, perceived support, and social competence variables on drinking and psychological adjustment. The results of the multiple regression analyses are summarized in Table III. The predictor variables entered into the regression equations included the variables with the highest loadings on Factors 1, 2, and 4 (MF, INT, and Date, respectively). In addition, Density, ASC, network size, and proportion of family members were included in the drinking regressions because these variables were significantly related to drinking, even though they did not load significantly on any of the factors. Likewise, the Assert and Males variables were included in the regression predicting the BSI for the same reason. Only predictors that account for at least 1°70 of the variance in the dependent variable are reported in Table III. The results of the regression analyses indicate that the ability to predict either self-reported drinking status, Q-F, or BSI is only moderately increased when network, support, and competence variables are combined
296
Fondacaro and Heller
Table IIL Stepwise Multiple Regression: Predicting Drinking Measures and Psychological Symptoms from Network, Perceived Support, and Competence Variables Variable entered
Multiple R
R2
R 2 change
Overall Fa
Predicting self-reported drinking status 1. Density 2. Date 3. MP
0.24 0.28 0.32
0.06 0.08 0.10
0.06 0.02 0.02
7.35 5.31 4.79
Predicting the Q-F index 1. ASC 2. FAM 3. Date
0.26 0.29 0.31
0.07 0.08 0.10
0.07 0.01 0.02
7.09 5.74 4.45
Predicting the BSI 1. Assert 2. Density 3. Males
0.30 0.35 0.36
0.09 0.12 0.13
0.09 0.03 0.01
12.50 8.46 6.17
aAll Fs are significant a t p < 0.01.
as predictors. The major share of the SRDS variance is accounted for by the Density variable. Adding the Date and MF variables changes the multiple correlation from 0.24 to 0.32 and accounts for an additional 4e70 of the variance. Likewise, the majority of the Q-F variance is accounted for by only one variable, ASC. Adding the FAM and Date variables only accounts for an additional 2°7o of the variance. The majority of the variance in the drinking variables was accounted for by network variables (Density and ASC), but the best predictor of BSI was Assert, a social competence variable. Again, the majority of variance in this dependent variable is accounted for by only one predictor variable.
DISCUSSION
Recent critical analyses of the social support literature suggest that researchers generally have failed to make distinctions between different modes of support. In the present study, a factor analysis revealed that two separate statistically significant social support factors could be distinguished (Network Function and Perceived Intimacy/Support), and each of these could be further differentiated from measures of social competence. Factor 1, which accounted for the largest share of the variance (37.1°/0), was composed of several variables representing network functions. Factor 2 included variables that reflect intimacy and perceived social
Social SupportFactors
297
support. This finding suggests that an individual's perception of being supported is strongly associated with the quality or closeness of relationships with network members. Interestingly, an individual's cognitive appraisal that his support needs are being met by network members (represented by Factor 2) is apparently separate from whether he reports that he turns to network members for emotional support, problem-solving assistance, information, material aid or companionship (Factor 1). The final factor that emerged included variables associated with social competence. Heterosexual intimacy also loaded significantly on this factor. This finding is consistent with Heller and Swindle's argument (1983) that social competence is important for the initiation and maintenance of supportive relationships, and that network functions and perceived intimacy/support are distinct dimensions of the social support construct which can be differentiated from social competence. The results of this study support the work of Jessor et al. (1968) and Moos et al. (1976) in finding that college student drinking is less related to personal psychopathology and is more a function of the role of alcohol in college student life. While high levels of drinking and being in networks with others who drink might indicate poor adjustment in some contexts, among college students drinking status seems to reflect an active social life. With the easy availability of alcohol, drinking often is used as a prerequisite for social activity. The relation between "social support" and psychological adjustment also varied as a function of how adjustment was measured. Reported psychological symptomatology was related to an individual's cognitive appraisal of the quality of his interpersonal relationships (perceived support); drinking was more strongly related to the frequency or amount of social contact (network structure and interaction variables). Thus, our impression is that student drinkers generally are socially competent individuals. When faced with social and academic pressures, they may involve themselves in ongoing social life and cope with life adjustment problems within a social context. In the college community, the social context provides frequent opportunities for alcohol use. On the other hand, students who report high levels of psychological distress are generally less socially competent. When faced with stress, they may turn inward, and, therefore, be more likely to report psychological symptoms. The results indicate that individuals who admit experiencing psychological problems also report low levels of perceived support, perhaps suggesting that both of these variables are part of the cognitive appraisal process. The study was largely exploratory and, therefore, is subject to several limitations. Obviously, caution must be exercised in interpreting the results, given the correlational nature of the data and the generally low magnitude
298
Fondaearo and Heller
of the correlation coefficients obtained. None of the predictor variables, either alone or combined, accounted for the majority of the total variance in the dependent variables. It should be mentioned, however, that correlation coefficients above 0.30 are rare in this area of research. In any event, future research might investigate the relation between support, drinking, and psychological adjustment longitudinally in order to clarify the direction and causal nature of the relationships. A second limitation is the reliance solely on self-report measures to assess the independent and dependent variables. Response biases and shared method variance may be influencing the findings. Future studies should employ multiple methods (e.g., peer report, behavioral observation) to assess network variables, social competence, and psychological adjustment in order to increase the validity of the data and to ensure that these variables are measured independently of the subjects's own perception. This is particularly crucial for clarifying the relation between perceived social support and independent measures of network variables, competence, and adjustment. Finally, the generalizability of the present findings should be evaluated through cross-validation studies with new subject samples. This study focused on male college freshmen. It is encouraging to note, however, that our findings are consistent with those of Moos et al. (1976, 1977), who found that social interaction variables also were positively related to alcohol use among females and nonfreshmen.
REFERENCES Cahalan, D., and Cisin, I. H. (1968). American drinking practices: Summary of findings from a national probability sample: II. Measurement of massed versus spaced drinking. Quart. J. Stud. Alcohol29: 642-656. Caplan, G. (1974). Support Systems and Community Mental Health, Behavioral Publications, New York. Cassel, J. 0975). Social science in epidemiology: Psychosocial processes and "stress" theoretical formulation. In Struening, E. L., and Guttentag, M. (eds.), Handbook of Evaluation Research (Vol. 2), Sage, Beverly Hills, Calif. Cobb, S. (1976). Social support as a moderator of life stress. Psychosomat. Med. 38:300-314. Cohen, S., and McKay, G. (in press). Social support, stress and the buffering hypothesis: A theoretical analysis. In Baum, A., Singer, J. E., and Taylor, S. E. (eds.), Handbook of Psychology and Health (Vol. 4), Erlbaum, Hillsdale, N.J. Corty, E., and Young, R. D. (1980). Social contact and loneliness in a university population. Paper presented at the meetings of the Midwestern Psychological Association, May. Dean, A., and Lin, N. (1977). The stress buffering role of social support: Problems and prospects for systematic investigation. J. Nerv. Ment. Dis. 165: 403-417. Derogatis, L. R., Lipman, R. S., and Covi, L. (1973). A brief symptom inventory. Psychopharmacol, Bull. 9: 13-27.
Social Supporl Factors
299
Fondacaro, M. R. (1980). Social competence and situations in college that induce drinking. Unpublished manuscript, Indiana University. Gore, S. (1978). The effect of social support in moderating the health consequences of unemployment. J. HIth. Soc. Behav. 19: 157-165. Gottlieb, B. H. (1979). The primary group as supportive milieu: Applications to community psychology. Am. J. Commun. Psychol. 7: 469-480. Hanson, D. J. (1974). Drinking attitudes and behaviors among college students. J. Alcohol Drug Educ. 19:6-14. Heller, K., and Swindle, R. (1983). Social networks, perceived social support and coping with stress. In Felner, R. D., Jason, L. A., Moritsugu, J., and Farber, S. S. (eds.), Preventive Psychology: Theory, Research and Practice in Community Intervention, Pergamon Press, New York. Hirsch, B. J. (1979). Psychological dimensions of social networks: A multimethod analysis. Am. J. Commun Psychol. 7: 263-267. Jessor, R., Carman, R. S., and Grossman, P. H. (1968). Expectations of need satisfaction and drinking patterns of college students. Quart. J. Stud. Alcohol 29: 101-116. Levenson, R. W., and Gottman, J. M. (1978). Toward the assessment of social competence. J. Consult. Clin. Psychol. 46: 453-462. Marsella, A. J., and Snyder, K. K. (1981). Stress, social supports and schizophrenic disorders: Toward an interactional model. Schizophren. Bull. 7:152-163. Mitchell, R. E., and Trickett, E. J. (1980). Task force report: Social networks as mediators of social support. An analysis of the effects and determinants of social networks. Commun. Merit. Hlth. J. 16: 27-44. Mops, R. H., Mops, B. S., and Kulik, J. A. (1976). College-student abstainers, moderate drinkers, and heavy drinkers: A comparative analysis. J. Youth Adoles. 5: 349-360. Mops, R. H., Mops, B. S. and Kulik, J. A. (1977). Behavioral and self-concept antecedents and correlates of college-student drinking patterns. Int. J. Addict. 12: 603-615. Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, K., and Bent, D. H. (1975). SPSS: Statistical Package for the Social Science~ (2nd ed.). McGraw-Hill, New York. Nuckolls, K. B., Cassel, J., and Kaplan, B. H. (1972). Psychosocial assets, life crisis and the prognosis of pregnancy. Am. J. Epidemiol. 95:431-441. Procidano, M. E., and Heller, K. (1983). Measures of perceived social support from friends and from family: Three validation studies. Am. J. Commun. Psychol. 11: 1-24. Seizer, M. L., Vinokur, A., and Rooijen, L. (1975). A self-administered Short Michigan Alcoholism Screening Test (SMAST). 3'. Stud. Alcohol 36:117-126. Thoits, P. A. (1982). Conceptual, methodological and theoretical problems in studying social support as a buffer against life stress. J. HIth. Soc. Behav. 23: 145-159. Tolsdorf, C. C. (1976). Social networks, support, and and coping: An exploratory study. Faro. Process 15: 407-417. Weschler, H. (1979). Patterns of alcohol consumption among the young: High school, college, and general population studies. In Blanc, H. T., and Chafetz, M. E. (eds.), Youth, Alcohol, and Social Policy, Plenum, New York. Wilcox, B. L. (1981). Social support, life stress, and psychological adjustment: A test of the buffering hypothesis. Am, J. Commun. Psychol. 9: 371-386.
Social Support Factors and Drinking Among College Student Males M a r k R . F o n d a e a r o ~.3 and Kenneth H e l l e r 2
Received January 25, 1983
The present investigation focused on social support and social competence among male college freshmen and the relation o f these variables to alcohol use and psychological adjustment. Recent critical analyses o f the social support literature suggest that studies in this area have generally failed to distinguish between different modes o f support. Therefore, measures pertaining to possible dimensions o f the social support construct (i.e., social network characteristics and perceived social supporO were administered to 137 male college freshmen, along with a measure o f social competence, and these data were factor analyzed. As a result, three interpretable factors were identified: Network Functions, Perceived Intimacy~Support, and Social Competence. Measures representing social network characteristics (e.g., network size, density, amount o f social contact), perceived support, and social competence were used to predict alcohol use and psychological symptomatology. Results indicated that alcohol use was positively related to social network characteristics that reflect high levels o f social interaction (e.g., network density, amount o f social contact) and measures o f social competence. Drinking was not significantly related to measures o f perceived social support. Psychological symptomatology was negatively related to measures o f perceived support, social competence, and network density. Thus, this study concludes that different modes o f support and different measures o f psychological adjustment should not be treated as if they are IDepartment of Psychology, Indiana University. Doctoral candidate in clinical psychology. Special interests are family and peer influences on juveniledelinquency and alcohol abuse. 2Professor of Psychology, Indiana University. Received Ph.D. in psychology from Pennsylvania State University. Current interests include the role of social ties and social support in coping with stress. 3Correspondence should be sent to Mark R. Fondacaro, Department of Psychology, Indiana University, Bloomington, Indiana 47405. 285 0047-2891/83/08 00-0285503.00/0 © 1983Plenum Publishing Corporation
286
Fondacaro and Heller
equivalent. A n d this study reaffirms a growing concern that the social context provides frequent opportunities f o r alcohol use and abuse in a college community.
INTRODUCTION High rates of alcohol use and abuse have become prevalent on college campuses throughout the United States. In a nationwide study of college student drinking, Hanson (1974) found that 80% of the males and 73% of the females regarded themselves as drinkers. Weschler (1979) reported that 66% of male college students randomly sampled from 34 New England colleges reported drinking at least weekly, and more than 12% of seniors drank daily. The frequency of drinking among Wechsler's female subjects was lower. Still, 50% of females drank weekly and 2% drank daily. College students drink for a variety of reasons, ranging from the desire to enhance and facilitate their social interactions, to using alcohol to help cope with academic and peer pressures (Jessor et al., 1968; Fondacaro, 1980). Apparently, drinking has become intimately woven into the social fabric of college life. Moos et al. (1976) found that heavydrinking college students engaged in more frequent social interactions than moderate drinkers or abstainers. In a later study they found that heavy drinkers who reduced the frequency and quantity of their alcohol use tended to disengage from the college community (Moos et al., 1977). The results reported by Moos and associates suggest that college students who have extensive social networks may be at increased "risk" for alcohol abuse. This finding is particularly striking, given that literature in the area of social support has emphasized the benefits of social support networks in buffering individuals from stress and reducing their risk for psychological and physiological disorders (Caplan, 1974; Cassel, 1975; Cobb, 1976). Numerous empirical studies have attempted to demonstrate the stressbuffering effect of "social support" for both clinical and nonclinical populations (e.g., Nuckolls et al., 1972; Tolsdorf, 1976; Gore, 1978; Hirsch, 1979; Wilcox, 1981). Although initial reviews of the social-support literature were quite optimistic (Cobb, 1976; Dean and Lin, 1977; Gottlieb, 1979), recent critical analyses suggest that many of the studies in this area are marred by serious methodological and conceptual flaws (Hirsch, 1979; Cohen and McKay, in press; Heller and Swindle, 1983; Thoits, 1982). Chief among the conceptual problems is the lack of a specific definition of the social support construct. Often "social support" is
Social $up0ort Factors
287
operationally defined quite differently across various studies, making it very difficult to interpret and compare findings. Terms such as "social support," "social network," and "social support system" are used as if they were equivalent, when in fact they may be tapping different facets of the support construct. In an attempt to more clearly define the social support construct, Heller and Swindle (1983) have proposed a model of social support and coping that emphasizes conceptual distinctions among several facets of the support construct. Their model is based largely on an integration of findings and a clarification of concepts in the areas of stress, support, and coping. Conceptual distinctions are made between social networks, perceived social support, and social competence. A social network is defined as the set of all significant others with whom one has social interactions (Hirsch, 1979). According to Marsella and Synder (1981), social networks can be characterized by four dimensions, each of which may be independently related to support, coping and psychological adaptation. The first dimension, structure, includes morphological variables such as the individual's position within a network, amount of social contact with network members, network size and network density (i.e., the number of dyadic relationships in a network in proportion to the number of possible relationships, given the network size). The dimension of interaction refers to variables that describe the relationship between various network members, including reciprocity and directionality. A third dimension, quality, describes affective characteristics of relationships such as intimacy and level of friendship. Finally, the function dimension describes specific functions served by network members such as companionship, problem-solving assistance, material aid, information, and emotional support. Heller and Swindle (1983) make a major distinction in their model between social networks and perceived social support; the latter is intended to describe the subjective impact of helping behavior on the individual. Perceived social support is based in part on the individual's perception of his/her social network and is defined as the cognitive appraisal that one's needs for support, information, and feedback are provided for by network members. Defined in this way, an individual may experience high levels of perceived social support without being a member of a large or active social network. The person's needs for support, information, and feedback may be met by a few close or intimate relationships. On the other hand, individuals who have large social networks and engage in frequent social interactions may' not necessarily feel supported. In support of this hypothesis, Corty and Young (1980) found that perceived support was not
288
Fondacaro and Heller
related to the amount of social contact with network members, but was inversely related to feelings of loneliness. Social networks and social support have usually been regarded as environmental variables. Heller and Swindle (1983) argue that an individual's level of perceived support also is dependent on personal characteristics such as social competence. Socially competent individuals are regarded as possessing the necessary skills for initiating and maintaining supportive relationships. In their model then, perceived social support (the appraisal that one is supported) is influenced both by social connections provided by the environment (social networks) and person variables that aid in the development of supportive relationships. The present study was guided by a belief that social network characteristics and perceived support are independent facets of the support construct. To test this assumption, a questionnaire that included measures of social network characteristics, perceived social support, and social competence was administered to a sample of male college freshmen. Information obtained from these measures was factor analyzed. It was hypothesized that (1) separate factors would emerge for each of the support dimensions (networks and perceived support) and also for social competence. Following the assumption that drinking occurs in college social settings primarily to facilitate social interaction, it was hypothesized that (2) alcohol use among freshmen males would be positively associated with network variables that reflect high levels of social interaction (e.g., network size, density, and amount of social contact). Subjects completed a measure of psychological symptomatology (The Brief Symptom Inventory, or BSI) in addition to measures of alcohol use. Traditionally, studies of the effects of social support on adjustment have used such measures, and generally found that those with high support report good adjustment. However, Heller and Swindle (1983) postulate that the relationship between reported symptomatology and perceived social support should be higher than the relationship between symptomatology and social network variables, since both symptomatology and perceived support can be considered part of the appraisal process. Thus, it was hypothesized that (3) psychological symptomatology would be negatively related to perceived support, but would not be significantly correlated with network measures reflecting high levels of social interaction (network size, amount of social contact). Finally, it was expected that socially incompetent individuals would lack the necessary skills to initiate and maintain supportive relationships, and would subsequently be at greater risk for psychological problems. Therefore, it was predicted that (4) social competence would be negatively associated with psychological symptomatology.
Social Support Factors
289
METHOD Subjects The subjects were 137 freshman male undergraduates at Indiana University. Subjects were recruited from an introductory psychology subject pool and received required experimental credits for their participation. Ten subjects failed to complete the questionnaire according to instructions and were, therefore, excluded from subsequent data analyses. Freshmen were chosen for study because it was expected that the transition from high school to college would be a stressful life event for many students, taxing their already developed coping patterns. The study focused on males because of their general neglect in the social support literature.
Materials Subjects were administered a questionnaire battery that included the following materials: Personal Information Sheet. This portion of the battery contained questions concerning demographic background (e.g., school year, employment status, present living situations), parental drinking practices, heterosexual intimacy, self-report of current drinking practices, and quality of relationships with father and mother. PerceivedSocial Support. This scale contains two subscales, each with 20 items, which measure the extent to which persons perceive that they are receiving adequate levels of support from friends (PSS-Fr) and family members (PSS-Fa). Reliability and validity data are reported in Procidano and Heller (1983). Social Competence Questionnaire. This situation-specific measure of social competence contains two nine-item scales that evaluate an individual's self-reported competence in handling dating and assertion situations. Reliability and validity data are reported in Levenson and Gottman (1978). Michigan Alcohol Screening Test (MAST). The MAST is a measure of problem drinking that has been used extensively in research on alcohol abuse among adults (Seizer, et ak, 1975). According to Seizer eta/., a total score of 0-4 points indicates nonproblem drinking, 5-6 suggests problem drinking, and 7 or higher indicates problem drinking. Social Relationship Questionnaire (SRQ). The SRQ includes measures of social network variables reported in Hirsch (1979) and in Mitchell and Trickett (1980) and Tolsdorf (1976). Fourteen network variables, representing three of the four network dimensions outlined by Marsella and
290
Fondacaro and Heller
Snyder (1981) (Structure, Quality, and Function), were included in this study. They are as follows.
Structure (1) Size. Subjects were asked to list the first names or initials of up to 15 individuals wi[h whom they interacted at least once during a two threeweek period. Size was calculated by summing the number of people listed. (2) Density. Subjects were asked to list the individuals in their social network in matrix form, placing an "X" to indicate where individuals were likely to interact with each other during a,two-three-week period. Density X was computed in the following way: Density = N ( N - 1)/2, where X = the number of linkages in the network and N = the network size. (3) Proportion of males listed in the network. (4) Proportion of network members who werefemales. (5) Proportion of family members listed in the network. (6) Amount of Social Contact (ASC). For each network member, subjects estimated the number of hours, rounded to the nearest half-hour, that they spent with that person over the last seven days. ASC was computed as the sum of these ratings across all network members.
Quality (7) Proportion of Intimate Relationships (INT). Subjects rated the intimacy or closeness of their relationship with each network member on a 4-point scale. INT = sum of intimacy ratings across all network members/N. (8) Proportion of Very Intimate Relationships (V-INT). V-INT represents the proportion of relationships that were given the highest intimacy rating by the subject.
Function (9) Companionship Support (CS). Subjects rated each member of their network on a 4-point scale (ranging from 1 = very unlikely to 4 = very likely) in terms of whether they would seek the individual's companionship if they wanted to enjoy themselves and have fun. (This same 4-point rating scale was used to compute the emotional support, informational support, material support, and problem-solving support variables.) Companionship support was calculated in the following manner: CS = sum of companionship ratings across all network members/N. (10) Emotional Support (ES). Subject rated whether each member of his network would listen to him when he was feeling down, encourage him, be understanding, just be with him. ES = sum of emotional support ratings across all network members/N. (11) Informational Support (IS). Subject rated whether each member of his social network was someone who would be likely to tell him about
Social SupportFactors
291
things that are good to know, like new places to go or places to get help. IS = sum of informational support ratings across all network members/N. (12) Material Support (MS). Subjects rated network members in terms of their likelihood of providing material aid (e.g., money, help in moving) if a subject needed it. MS = sum of material aid ratings across all network members/N. (13) Problem-Solving Support (PS). Subjects rated whether each network member would help them solve personal problems, give them advice, help them explore alternatives. PS = sum of problem-solving ratings/N. (14)Multiplexity (MP). MP was computed as the proportion of multiplex relationships in an individual's social network. Relationships that served only one function for the individual (e.g., companionship) were regarded as unifunctional. Relationships that were rated as likely or very likely to serve more than one function O. e., companionship, emotional support) were scored as multiplex. Thus, multiplexity was calculated in the following manner: MP = number of multiplex relationships/N. The following two measures were also included in the SRQ to assess the drinking status of network members.
Proportion of Heavy Drinkers (HD). HD = the proportion of network members who were rated by the subject as either heavy or problem drinkers. Proportion of Drinkers (Drink). Drink = the proportion of network members who were rated by the subject as drinkers.
Dependent Variables
Quantity-Frequency Index (Q-F). The Q-F index is a measure of daily alcohol consumption. It has been used extensively in research concerning alcohol use and abuse and is reported in Cahalan and Cisin (1968). Brief Symptom Inventory (BSI). The BSI is a 53-item measure of reported psychological symptoms. Respondents were asked to rate on a 5point scale how much discomfort each symptom has caused them during the previous week. Information on the psychometric properties of this inventory are reported in Derogatis et al. (1973). Self-Reported Drinking Status (SRDS). Subjects were asked to rate their own drinking status on a 5-point scale ranging from I = abstainer to 5 = problem drinker.
292
Fondacaro and Heller Procedure
Subjects completed the battery of questionnaires during a two-hour testing session. Two testing sessions were held; the first session was attended by 90 subjects, and the second session was attended by 47 additional subjects. All questionnaires were completed anonymously to ensure the confidentiality of the subjects' answers.
RESULTS Twenty-two variables (all pertaining to either social networks, perceived support, or social competence) were subjected to a common factor analysis (SPSS version PAZ; Nie et al., 1975). Seven factors, each with an eigenvalue greater than 1.00, accounted for approximately 69070 of the total variance. Six of these factors were extracted and subjected to an orthogonal varimax rotation. As a result, four factors with factor contributions greater than 1.00 emerged. Factor structures were based on variables loading 0.35 or greater on a rotated factor. The four factors, the variable loadings, and the percentage of the reduced variance accounted for are presented in Table I. Factor 1 accounts for 37.1 070 of the reduced variance and consists of variables representing "network functions" (i.e., companionship, emotional support, material support, problem-solving support, and multiplexity). Factor 2 accounts for 20.6°70 of the variance and consists of variables associated with perceived social support and intimacy (e.g., perceived peer support, closeness of relationships with parents, and proportion of intimate relationships with network members). Factor 3 accounts for 16.1 070of the variance and consists of two redundant variables: the proportion of males in one's network (positive loading) and the proportion of females in one's network (negative loading). This was labeled the Residual factor. Finally, variables associated with social competence (e.g., dating competence, assertiveness) loaded significantly on Factor 4, accounting for 11.4070 of the variance. Thus, Network Functions and Perceived Intimacy/Support emerged as separate factors or dimensions of the social support construct, and both were differentiated from social competence. Although network size, density and amount of social contact did not load significantly on any of the interpretable factors, they did correlate significantly with the dependent variables and were therefore included in subsequent correlational and regression analyses. The range of scores on the MAST revealed the pervasiveness of alcohol abuse in college. Over 35070 of the subjects received scores which suggested problem drinking (i.e., 5 or greater). Since the MAST was not
Social Support Factors
293
Table I. Social Network, Perceived Support, and Social Competence Variables and Factor Loadings Factor and variance accounted for
Variable
Loading
1. Network functions (37.1%)
Multiplexity "Emotional support Informational support Problem support Material support Companionship support
0.75 0.74 0.72 0.72 0.65 0.46
2. Perceived intimacy/support (20.6%)
Network intimacy Very intimates Intimacy with father Intimacy with mother Perceived support - friends Perceived support - family
0.70 0.64 0.61 0.53 0.49 0.45
3. Residual (16.1)
Proportion of males Proportion of females
0.96 -0.96
4. Social competence (11.4%)
Dating competence Assertiveness Heterosexual intimacy
0.86 0.58 0.39
validated on a college population, this measure may not accurately reflect the extent of alcohol abuse in a college sample. However, roughly the same proportion of subjects (35~/0) reported drinking the equivalent of 1.5 ounces of absolute alcohol per day on the Q-F. Approximately 10~o of subjects reported drinking an average of over 4 ounces of absolute alcohol per day. Pearson product-moment correlations between the network, support, and competence variables and the alcohol use and psychological symptoms variables are presented in Table II. The variables most highly correlated with self-reported drinking status were HD (proportion of network members who are heavy or problem drinkers), r = 0.44, p < 0.001, and Drink (the proportion of network members who are drinkers or "nonabstainers") r = 0.40, p < 0.001. As predicted by our second hypothesis, self-reported drinking status (SRDS) was positively associated with Density, r = 0.24, p < 0.005, and Amount of Social Contact (ASC), r = 0.19, p < 0.05. However, SRDS was not significantly correlated with network size, r = 0.09, p > 0.05. This correlation probably does not accurately reflect the degree of association between these variables, since over 57 % of the subjects listed the maximum number of network members requested, 15. Thus, the range of the network size variable was apparently truncated. There was a marginally significant negative association between self-
Fondacaro and Heller
294
Table ii. Correlations Between Network, Support, and Competence Variables, and Alcohol Use and Psychological Symptoms Self-reported drinking
Quantity/frequency of drinking
Brief symptom inventory
0.09 0.24b 0.06 -0.14 c 0.19 c 0.07 -0.09 -0.17c -0.13 -0.14 c 0.40 a 0.44 a 0.10 0.16 c
0.13 0.14 c 0.00 -0.16 c 0.26b 0.15 -0.06 -0.07 -0.07 -0.11 0.26b 0.36a 0.16 c 0.11
0.04 - 0.17 c -0.13 0.10 -0.08 0.04 -0.01 -0.09 -0.04 -0.03 -0.15c 0.04 0.03 -0.06
Perceived support Perceived Support - Family Perceived Support - Friends
0.00 0.12
0.02 0.08
-0.20 c -0.24 b
Social competence Dating Competence Assertion Competence
0.10 0.04
0.17c 0.10
-0.18 c -0.30 b
Network variables Size Density Males in Network Family in Network Amount of Social Contact Companion Support Information Support Problem support Emotional Support Multiplexity Drinkers Heavy Drinkers Intimates in Network Very-Intimates in Network
ap < bp < Cp
0.05, or problem-solving support, r = - . 0 7 , p > 0.10. Neither the Q-F index nor the drinking status variables were significantly correlated with either perceived family support (PSS-Fa) or perceived peer support (PSS-Fr). In summarizing these results, it appears that the likelihood of drinking increases with the number of drinking and heavy-drinking network members, the frequency of social contacts, network density, the proportion of close or intimate relationships, the dating competence of the individual, and, to a lesser extent, the number of network members. There is an inverse relation between drinking and the proportion of network members who are also family members. Moreover, individuals who characterize themselves as frequent or heavy drinkers are less likely to seek out problem-solving assistance or participate in multiplex relationships than their peers, even though they characterize their relationships as more intimate. Inspection of the correlation coefficients in Table II indicates that various dimensions of the support construct are differentially related to alcohol use and psychological symptomatology. Among the more striking relationships is the finding that network density, proportion of drinkers, and dating competence are positively related to the drinking measures and negatively related to the Brief Symptom Inventory. Moreover, the BSI was not related to social network variables (ASC, INT, FAM) that were significantly correlated with the drinking measures. As predicted in our third hypothesis, the BSI was negatively related to the perceived support scales (PSS-Fa, PSS-Fr), both of which were not significantly related to either of the drinking variables. As predicted in our fourth hypothesis, the PSI was negatively associated with both measures of social competence. Stepwise multiple regression analyses were employed to examine the combined effects of social network, perceived support, and social competence variables on drinking and psychological adjustment. The results of the multiple regression analyses are summarized in Table III. The predictor variables entered into the regression equations included the variables with the highest loadings on Factors 1, 2, and 4 (MF, INT, and Date, respectively). In addition, Density, ASC, network size, and proportion of family members were included in the drinking regressions because these variables were significantly related to drinking, even though they did not load significantly on any of the factors. Likewise, the Assert and Males variables were included in the regression predicting the BSI for the same reason. Only predictors that account for at least 1°70 of the variance in the dependent variable are reported in Table III. The results of the regression analyses indicate that the ability to predict either self-reported drinking status, Q-F, or BSI is only moderately increased when network, support, and competence variables are combined
296
Fondacaro and Heller
Table IIL Stepwise Multiple Regression: Predicting Drinking Measures and Psychological Symptoms from Network, Perceived Support, and Competence Variables Variable entered
Multiple R
R2
R 2 change
Overall Fa
Predicting self-reported drinking status 1. Density 2. Date 3. MP
0.24 0.28 0.32
0.06 0.08 0.10
0.06 0.02 0.02
7.35 5.31 4.79
Predicting the Q-F index 1. ASC 2. FAM 3. Date
0.26 0.29 0.31
0.07 0.08 0.10
0.07 0.01 0.02
7.09 5.74 4.45
Predicting the BSI 1. Assert 2. Density 3. Males
0.30 0.35 0.36
0.09 0.12 0.13
0.09 0.03 0.01
12.50 8.46 6.17
aAll Fs are significant a t p < 0.01.
as predictors. The major share of the SRDS variance is accounted for by the Density variable. Adding the Date and MF variables changes the multiple correlation from 0.24 to 0.32 and accounts for an additional 4e70 of the variance. Likewise, the majority of the Q-F variance is accounted for by only one variable, ASC. Adding the FAM and Date variables only accounts for an additional 2°7o of the variance. The majority of the variance in the drinking variables was accounted for by network variables (Density and ASC), but the best predictor of BSI was Assert, a social competence variable. Again, the majority of variance in this dependent variable is accounted for by only one predictor variable.
DISCUSSION
Recent critical analyses of the social support literature suggest that researchers generally have failed to make distinctions between different modes of support. In the present study, a factor analysis revealed that two separate statistically significant social support factors could be distinguished (Network Function and Perceived Intimacy/Support), and each of these could be further differentiated from measures of social competence. Factor 1, which accounted for the largest share of the variance (37.1°/0), was composed of several variables representing network functions. Factor 2 included variables that reflect intimacy and perceived social
Social SupportFactors
297
support. This finding suggests that an individual's perception of being supported is strongly associated with the quality or closeness of relationships with network members. Interestingly, an individual's cognitive appraisal that his support needs are being met by network members (represented by Factor 2) is apparently separate from whether he reports that he turns to network members for emotional support, problem-solving assistance, information, material aid or companionship (Factor 1). The final factor that emerged included variables associated with social competence. Heterosexual intimacy also loaded significantly on this factor. This finding is consistent with Heller and Swindle's argument (1983) that social competence is important for the initiation and maintenance of supportive relationships, and that network functions and perceived intimacy/support are distinct dimensions of the social support construct which can be differentiated from social competence. The results of this study support the work of Jessor et al. (1968) and Moos et al. (1976) in finding that college student drinking is less related to personal psychopathology and is more a function of the role of alcohol in college student life. While high levels of drinking and being in networks with others who drink might indicate poor adjustment in some contexts, among college students drinking status seems to reflect an active social life. With the easy availability of alcohol, drinking often is used as a prerequisite for social activity. The relation between "social support" and psychological adjustment also varied as a function of how adjustment was measured. Reported psychological symptomatology was related to an individual's cognitive appraisal of the quality of his interpersonal relationships (perceived support); drinking was more strongly related to the frequency or amount of social contact (network structure and interaction variables). Thus, our impression is that student drinkers generally are socially competent individuals. When faced with social and academic pressures, they may involve themselves in ongoing social life and cope with life adjustment problems within a social context. In the college community, the social context provides frequent opportunities for alcohol use. On the other hand, students who report high levels of psychological distress are generally less socially competent. When faced with stress, they may turn inward, and, therefore, be more likely to report psychological symptoms. The results indicate that individuals who admit experiencing psychological problems also report low levels of perceived support, perhaps suggesting that both of these variables are part of the cognitive appraisal process. The study was largely exploratory and, therefore, is subject to several limitations. Obviously, caution must be exercised in interpreting the results, given the correlational nature of the data and the generally low magnitude
298
Fondaearo and Heller
of the correlation coefficients obtained. None of the predictor variables, either alone or combined, accounted for the majority of the total variance in the dependent variables. It should be mentioned, however, that correlation coefficients above 0.30 are rare in this area of research. In any event, future research might investigate the relation between support, drinking, and psychological adjustment longitudinally in order to clarify the direction and causal nature of the relationships. A second limitation is the reliance solely on self-report measures to assess the independent and dependent variables. Response biases and shared method variance may be influencing the findings. Future studies should employ multiple methods (e.g., peer report, behavioral observation) to assess network variables, social competence, and psychological adjustment in order to increase the validity of the data and to ensure that these variables are measured independently of the subjects's own perception. This is particularly crucial for clarifying the relation between perceived social support and independent measures of network variables, competence, and adjustment. Finally, the generalizability of the present findings should be evaluated through cross-validation studies with new subject samples. This study focused on male college freshmen. It is encouraging to note, however, that our findings are consistent with those of Moos et al. (1976, 1977), who found that social interaction variables also were positively related to alcohol use among females and nonfreshmen.
REFERENCES Cahalan, D., and Cisin, I. H. (1968). American drinking practices: Summary of findings from a national probability sample: II. Measurement of massed versus spaced drinking. Quart. J. Stud. Alcohol29: 642-656. Caplan, G. (1974). Support Systems and Community Mental Health, Behavioral Publications, New York. Cassel, J. 0975). Social science in epidemiology: Psychosocial processes and "stress" theoretical formulation. In Struening, E. L., and Guttentag, M. (eds.), Handbook of Evaluation Research (Vol. 2), Sage, Beverly Hills, Calif. Cobb, S. (1976). Social support as a moderator of life stress. Psychosomat. Med. 38:300-314. Cohen, S., and McKay, G. (in press). Social support, stress and the buffering hypothesis: A theoretical analysis. In Baum, A., Singer, J. E., and Taylor, S. E. (eds.), Handbook of Psychology and Health (Vol. 4), Erlbaum, Hillsdale, N.J. Corty, E., and Young, R. D. (1980). Social contact and loneliness in a university population. Paper presented at the meetings of the Midwestern Psychological Association, May. Dean, A., and Lin, N. (1977). The stress buffering role of social support: Problems and prospects for systematic investigation. J. Nerv. Ment. Dis. 165: 403-417. Derogatis, L. R., Lipman, R. S., and Covi, L. (1973). A brief symptom inventory. Psychopharmacol, Bull. 9: 13-27.
Social Supporl Factors
299
Fondacaro, M. R. (1980). Social competence and situations in college that induce drinking. Unpublished manuscript, Indiana University. Gore, S. (1978). The effect of social support in moderating the health consequences of unemployment. J. HIth. Soc. Behav. 19: 157-165. Gottlieb, B. H. (1979). The primary group as supportive milieu: Applications to community psychology. Am. J. Commun. Psychol. 7: 469-480. Hanson, D. J. (1974). Drinking attitudes and behaviors among college students. J. Alcohol Drug Educ. 19:6-14. Heller, K., and Swindle, R. (1983). Social networks, perceived social support and coping with stress. In Felner, R. D., Jason, L. A., Moritsugu, J., and Farber, S. S. (eds.), Preventive Psychology: Theory, Research and Practice in Community Intervention, Pergamon Press, New York. Hirsch, B. J. (1979). Psychological dimensions of social networks: A multimethod analysis. Am. J. Commun Psychol. 7: 263-267. Jessor, R., Carman, R. S., and Grossman, P. H. (1968). Expectations of need satisfaction and drinking patterns of college students. Quart. J. Stud. Alcohol 29: 101-116. Levenson, R. W., and Gottman, J. M. (1978). Toward the assessment of social competence. J. Consult. Clin. Psychol. 46: 453-462. Marsella, A. J., and Snyder, K. K. (1981). Stress, social supports and schizophrenic disorders: Toward an interactional model. Schizophren. Bull. 7:152-163. Mitchell, R. E., and Trickett, E. J. (1980). Task force report: Social networks as mediators of social support. An analysis of the effects and determinants of social networks. Commun. Merit. Hlth. J. 16: 27-44. Mops, R. H., Mops, B. S., and Kulik, J. A. (1976). College-student abstainers, moderate drinkers, and heavy drinkers: A comparative analysis. J. Youth Adoles. 5: 349-360. Mops, R. H., Mops, B. S. and Kulik, J. A. (1977). Behavioral and self-concept antecedents and correlates of college-student drinking patterns. Int. J. Addict. 12: 603-615. Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, K., and Bent, D. H. (1975). SPSS: Statistical Package for the Social Science~ (2nd ed.). McGraw-Hill, New York. Nuckolls, K. B., Cassel, J., and Kaplan, B. H. (1972). Psychosocial assets, life crisis and the prognosis of pregnancy. Am. J. Epidemiol. 95:431-441. Procidano, M. E., and Heller, K. (1983). Measures of perceived social support from friends and from family: Three validation studies. Am. J. Commun. Psychol. 11: 1-24. Seizer, M. L., Vinokur, A., and Rooijen, L. (1975). A self-administered Short Michigan Alcoholism Screening Test (SMAST). 3'. Stud. Alcohol 36:117-126. Thoits, P. A. (1982). Conceptual, methodological and theoretical problems in studying social support as a buffer against life stress. J. HIth. Soc. Behav. 23: 145-159. Tolsdorf, C. C. (1976). Social networks, support, and and coping: An exploratory study. Faro. Process 15: 407-417. Weschler, H. (1979). Patterns of alcohol consumption among the young: High school, college, and general population studies. In Blanc, H. T., and Chafetz, M. E. (eds.), Youth, Alcohol, and Social Policy, Plenum, New York. Wilcox, B. L. (1981). Social support, life stress, and psychological adjustment: A test of the buffering hypothesis. Am, J. Commun. Psychol. 9: 371-386.
