INT'L. J. AGING AND HUMAN DEVELOPMENT, Vol. 35(2) -123,1992
THE IMPACT OF STRESSFUL LIFE EVENTS AND SOCIAL SUPPORT ON DRINKING AMONG OLDER ADULTS: A GENERAL POPULATIONSURVEY*+
KAREN M. JENNISON, PH.D. University of Northern Colorado, Greeley
ABSTRACT
This article is an analysis of stressful life events, the buffering hypothesis, and alcohol use in a national sample of 1,418 respondents 60 years of age and over. The results indicate that older adults who experience stressful losses are significantly more likely to drink excessively than those who have not experienced such losses or who have experienced them to a lesser extent. Increased drinking among older adults may therefore be. a reaction to life circumstances in which alcohol represents an attempt to cope with traumatic loss, personal as well as within the kinship network. Supportive resources of spouse, family, friends, and church appear to have a stress-buffering effects that reduces the excessive-drinking response to life crisis. Data suggest, however, that older persons are vulnerable to the magnitude of losses experienced as they grow older and lose more of their family, friends, and peers. These stressors appear to seriously impact their drinking behavior and are not effectively buffered. Respondents report that drinking may increase during periods of prolonged exposure to emotionally depleting life change and loss, when supportive needs may exceed the capacities of personal and social support resources.
*Data used in the study were made available from the National Opinion Research Center (NORC) General Social Surveys. The analyses and interpretation of data presented herein are solely the author's. 'Revision of a paper presented at the National Conference on Aging in the 1990s.Alcohol and Other Drug Abuse, Novi, Michigan, April 1990. 99
0 1992, Baywood Publishing Co.,Inc.
doi: 10.2190/F6G4-XLV3-5KW6-VMBA http://baywood.com
100 / JENNISON
INTRODUCTI0 N A review of the literature on drinking behavior among elderly individuals, particularly within the last decade, suggests that many explanations of excessive drinking within this social group remain controversial. A major perspective that has been the subject of considerable debate, is the assumption that older people who experience stressful life event losses, including loss of loved ones, health, occupation, and status, are likely to drink excessively. This research has focused on the aging-drinking relationship and the impact of losses on drinking as people grow older [l-111. A growing number of researchers have come to believe that research emphasizing increased drinking as a response to stresses associated with aging is deficient; it is long on theory and short on empirical documentation [12]. The stressful loss research approach to deviant drinking among elderly individuals may, in fact, exaggerate the association of stress-linked alcohol abuse with chronological age, when other factors may provide more plausible explanations of drinking behavior in older groups [12-151. Yet, there have been few studies with national samples investigating the stressdrinking relationship. The purpose of this study is to present general information on national drinking patterns among noninstitutionalized elderly individuals and to test three major drinking-related perspectives: 1) older persons increase their alcohol consumption in response to major life transitions involving loss; 2) the effect of stressful life events may be buffered, or modified, for older persons who have sufficient support from various social groups such as intermediate or extended family, friends, or other groups [16]; and 3) the effects of other factors such as group affiliated drinking, alienation, and health status may have greater impact on elderly drinking than explanations based on the traumatic losses they experience over time [17-191. Stressful Life Events as Role Loss
The prevailing stress-loss hypothesis, widely tested in clinical studies and tested here in the general population, posits that individuals who experience stressful life losses will be more likely to drink excessively than those not experiencing losses, and that excessive drinking will be closely associated with a specific crisis situation [20]. The idea that most life transitions are stressful and that they lead to various coping strategies has frequently been researched [21]. Alcohol abuse among elderly individuals has been identified as a major response to the stresses associated with growing older (see Elwell and Matlbie-Crannell[4] for a discussion of stress and the elderly, and fennison and Johnson [20] for a discussion of the concept of stress in relation to alcohol use). There has been little empirical
STRESSAND DRINKINQ / 101
evidence, however, to support the theory that stressful losses among aging individuals are associated with increased, or problematic drinking [9,12,13,22]. Here, we are dealing only with loss type events; thus, the traditional concept of role loss or exit seems to be an appropriate way of linking social structure and the life course to drinking behavior [23]. Role loss occurs whenever any stable pattern of interaction and shared activity between two or more persons ceases [2]. Whether the event is viewed as loss, separation, departure or ending, and whether these losses are temporary or permanent, the process can be traumatic for an individual. Within the last few decades, stress and loss research has been developmentally integrated into the concept of “psychosocial transitions” [24-271. The critical factor for these researchers is not only the fact of loss, or crisis, but recognition of the possible duration of an individual’s progression through phases of the transition. Longer transitions are characterizedby chronically excessive demands which may seriously tax and undermine the resources available to the individual. Psychosocial transitions involve re-definitions of the individual’s assumptive world as well as life reorganization with its potentially devastating emotional impact. Because role loss in many instances is unexpected, “role discontinuity” occurs.With little preparation, the new role requirements and status arrangements may be sufficiently stressful to be classified as a crisis experience [B-311. A stable pattern of interaction with other people may be interrupted not only by one’s own stressful event, such as unemployment or illness, but also by such stressful events in the lives of other persons who are role partners. Extending the effects of the crisis experience of the individual to encompass the role losses of relatives is termed “network crisis events” [32], or “multiple crises” [33].
Affiliation Drinking behavior among the elderly, as among other groups, may be most completely explained when interpreted within a social context [34, 351. Group memberships and activities of various types are related to excessive drinking because drinking is often normative in these groups [36]. Drinking is known to increase in drinking-centered group activities where heavy social drinking takes place. Akers [37] defines drinking-centered groups as including tavern groups, formal and informal parties, neighbors, clubs, ceremonial and celebration gatherings, conventions, and similar group drinking events; these drinking groups are also attractive and rewarding for the drinker, and even give status to the skilled drinker (e.g., who can “drink like a man,” or “hold his liquor”). These groups legitimate drinking as a way of developing congenial social interaction. The more socially active, therefore, are also more likely to be heavy drinkers [38].
102 I JENNISON
Other examples of group-affiliational drinking, in which culturally conventionalized inebriety [39] is acceptable as long as it remains within the limits of the group norms, are found among: college students [40-441; lower-status occupational groups [45], white collar executives [46], college professors [47], salespeople and professional-technical employees [48], college fraternities and sororities [49], blue collar workers [36, 50, 511, and professionals in medicine, law, dentistry, nursing, and social work [52]. Even though most members of these groups do not develop alcoholism, focusing on affiliational group drinking tests the assumption that it is one factor related to problematic drinking [45]. In testing the affiliation hypothesis using national data, therefore, we expect that affiliation as well as social activity will be related to excessive drinking among the elderly, particularly among men and among higher socioeconomic status groups [38], since these are both categories with relatively few elderly abstainers.
Alienation At the opposite end of the continuum, many older drinkers and older problem drinkers have been characterized as individuals who experience alienation and isolation from the mainstream of society [9, 111, either by living arrangement (living alone), loss of family members (e.g., widowed, separated, divorced) or by reduced social participationbecause of retirement, lowered income, or diminished social support resources. The concept of alienation originated in 19th-century social thought where the entire democratic industrial order with its urbanism, secularization, complex technology, and inaccessible bureaucracy was viewed as threatening human dignity and social cohesiveness [53]. It was the antithesis of what enlightenment thinkers had imagined as the result of social progress. Thus alienation, as defined here, refers to powerlessness, estrangement, and social isolation, the perceived lack of control, or mastery over sociopoliticalevents that affect one’s life [19,54, 551. The alienation drinking connection arises out of the same social conditions which promote a generalized alienation from self, work, and society [a]. Alienation has been negatively correlated with social activity in previous studies. That is, the alienated individual who drinks heavily shows minimal levels of affiliation, social group memberships or social participation. This is especially the case among “early on-set,” or under 30-year-old heavy drinkers [49,56], elderly male problem drinkers [16], and escape drinkers, those who drink for relief from anxiety, depression, and personal problems [38]. Heavy drinking itself would be likely to increase a sense of powerlessness, especially if it merges into alcoholism with its attendant life problems.
STRESS AND DRINKING I 103
Empirical studies linking alienation with excessive drinking tend to reinforce the “powerlessness” dimension; less importance is attached to the estrangement or social isolation components of the alienation concept. Calicchia and Barresi [17], an exception to this trend, found that alcoholics as a group were significantly more alienated in the social isolation sense, as measured on Dean’s Social Alienation Scale, than nonalcoholics. But male alcoholics experienced, alienation more in the powerlessness form. Previous studies have shown that alcohol could be used as a coping mechanism to relieve anxiety brought on by stress in people with a high sense of powerlessness [57]. In leading studies of drinking and alienation, Seeman and colleagues [18, 19,551 tested three measures of alienation: work experience (low job complexity), powerlessness (generalized expectations for control, or mastery, over outcomes), and social isolation (degree of individual attachment to significant others: friends, relatives, neighbors) in a sample of 450 employed males in a metropolitan Los Angeles community and found in both the first and follow-up studies that powerlessness, as a form of alienation, is directly related to drinking and drinking problems; increased powerlessness is associated with increased drinking problems. Therefore, measures of alienation used in this study primarily assess the powerlessness dimension (sense of personal control or mastery over life). Older people who are alienated in the sense of subjective control, should be more likely to drink excessively for power considerations. Further, older people with high alienation, or low mastery [58], would be expected to exacerbate the distressing impact of major role loss events by drinking even more.
SOCIAL SUPPORT Social support may be defined as “support accessible to an individual through social ties to other individuals, groups and the larger community” [59, p. 1091. Social ties may be seen as having stress or anxiety reducing features which make coping with major role changes less difficult. Various formal and informal groups, of course, fit into this category. Older problem drinkers, for example, have been found to have few friends and supportive family ties [60], and diminished social, economic, and health resources in general, particularly older men alcoholics [61]. Older problem drinkers are also likely to be dissatisfied with their relationships with family members, spouses, and close friends [14]. Marital discord, or dissatisfaction, has been described as an “enduring strain” or chronic stressor [62], and is highly predictive of increased alcoholic use. Social resources play an important role in buffering the impact of life stress on health [63]; some studies indicate that fewer support resources negatively impact alcohol treatment outcomes as well, resulting in less successful treatment [61]. Therefore, the older
104 / JENNISON
person’s social connectedness may influence drinking behavior following traumatic losses. METHODOLOGY Sample
Data utilized in the study consist of a general population sample of 1,418 respondents sixty years old and over in four pooled surveys drawn from the National Opinion Research Center (NORC) General Social Surveys [MI: years 1978 (N= 339; 23.9%),1980 (N= 361; 25.5%), 1983 (N= 373; 26.3%)and 1984 (N = 345; 24.3%). GSS survey questions on drinking behavior are “rotating items”; they only appear periodically, and not always in the same year. The sampling years selected include matching questions on both alcohol and independent variables, and represent maximum item equivalence of cross measurement indicators in survey replication. The use of comparable pooled surveys [65] thus minimizes the undenesponse problems in data analysis due to the absence of a question in any one particular year. Basic data are cross-sectional baseline prevalence drinking patterns derived from GSS survey questions on self-reported use or experience with alcoholic beverages (liquor, wine, or beer) and the respondent’s subjective evaluation of these experiences (for example, perceived intoxication effects). The conceptual measures of drinking behavior are continuous rather than dichotomousor discrete. Additional characteristics of the research design and methodology used in the study have been described elsewhere [20]. Measuringthe Variables
For purposes of this study, following previous research, stressful role loss is defined as independent disaggregated events (divorce, unemployment-job loss, family deaths, hospitalization, or disability) happening to the respondent and to relatives last year and within the last five years prior to the research. In addition, independent aggregated role losses are operationalized as the total number of cumulative events (divorce, unemployment or job loss, family deaths, hospitalization, or disability) happening to the respondent and relatives combined, last year and within the last five years. The two measures of respondent-relative role loss have been combined in order to estimate the relationship of role related or “overlapping” strain and drinking behavior. Affdiation Measures are defined as the number of voluntary group memberships in specialized formally organized social groups or organizations in which membership is based on deliberate choice and from which members may
STRESS AND DRINKING I 105
resign (16-item dummy variable, reference or excluded group coded 0 = no membership). Social support measures, using Herzog’s [66]rationale, were categorized into two separate groups: 1)social indicators of objective estimates of role resources, and 2) life satisfaction measures, or social group indicators of subjective assessments of major role resources. Objective support resources, or familial variables, are 1)marital status (0 = married; 1 = otherwise); 2) family size (1; 8 or more), 3) number of siblings (0;8 or more), and church attendance (0 = never; 7 = several times a week). These are commonly documented measures of structural social support (the perception that support is available from the network of family and friends if needed) which have been used extensively in previous research on buffering and social support [67-741. Subjective assessments of the quality of life, or life satisfaction and happiness, are defined as the degree to which respondents are satisfied with 1) friendships, 2) family life (coded 0 = no satisfaction; 7 = very great deal of satisfaction), 3) personal finances (0 = low; 3 = high satisfaction), and 4) job situation (1 = very dissatisfied; 4 = very satisfied). Therefore high scores on life satisfaction measures are equated with high satisfaction. The quality of health, or personal estimate (0 = none; 4 = fair amount of satisfaction) of health status, is used as a separate independent variable with drinking behavior. Alienation, defined as powerlessness, or lack of control or mastery over sociopolitical events that affect one’s life, is operationalized as how much confidence the respondent has in people running the following American institutions: 1) banks and financial institutions, 2) major companies, 3) education, 4) the executive branch of the federal government, 5 ) Congress, and 6) the military (coded 1= a very great deal of confidence; 3 = hardly any confidence). Thus, low confidence is equated with high alienation. An additive index of alienation was developed (coefficient of reliability = .97) based on the model of Herzog, Rodgers, and Woodworth [66]. Employment status is operationalized as working full- or part-time, versus being unemployed, retired, in school, or a home maker. The U.S.Bureau of the Census 3-digit occupational classifications for 1970 occupations are used as indicators of socioeconomic status in the study. Prestige scores were assigned for occupations based on the National Opinion Research Center (NORC)rating system [a].
Statistical Measures We test the proposition that social support relationships will intervene (buffer) between role losses and drinking measures using analysis of variance and multiple regression techniques. Groups with statistically significant f-values on drinking behavior using analysis of variance and multiple comparison procedures
106 / JENNISON
(Alpha = < .05)were selected to be included in multiple regression models. These statistical methods treat the predictor variables of role loss, health, alienation, affiliation, employment status, and social support as well as the criterion variable alcohol as continuous. In multiple regression models, indicators of each type of independent variable are first tested separately with drinking to determine any independentmain effects between independent variables and drinking behavior. In the second step, in which the multiple interaction term (independent variable x social support) is forced into the equation (forced entry), alcohol drinking patterns are regressed on independent variables separately for each social support group to assess any degree of reduction in the regression coefficient between independent variables and excessive drinking behavior when the intervening support variable is included. Last, to assess the total effect of all hypothesized social resource buffers, social support indicators, role losses, and other independent variables are tested simultaneously in the regression model, along with socioeconomic variables, for every support group which produced any reduction in the role loss-excessive drinking relationship [32,66,69,74]. Previous research suggests that those who have supportive social networks are protected, or buffered, in times of crisis. If role loss buffering is demonstrated for older persons in the study, then the effects of role loss on drinking behavior should decrease when social support is controlled. Differentially, high role loss in the absence of social support should be associated with increases in vulnerability to alcohol. According to the buffering model and patterns of results consistent with the buffering postulation, social supports or resources should either partially reduce or totally (statistically) ameliorate the effects of role losses on drinking [69]. The impact of support on the independent variables-drinking relationship should also be maintained to a significant degree after controls for the effects of marital status, education, and total family income are established. The assumption is that those respondents with adequate social resources (actual and perceived) temper or moderate their drinking while those respondents with weak resources do not.
RESULTS
Sample Characteristics Sociodemographiccharacteristicsof the general elderly population and national prevalence drinking patterns are shown in Table 1. These patterns indicate that one-third of men respondents in the study abstain and nearly one-fifth report excessive drinking. Among older women, on the other hand, fifty percent report total abstention from drinking alcohol and only five percent say that they
STRESS AND DRINKING / 107
sometimes overindulge. Excessive drinking among elderly individuals, as with drinking in general, is disproportionately a male activity. In every age group, more men than women drink, and- among those who drink, there are more heavy drinkers among men than among women [75]. Drinking practices among the elderly, as in younger age groups, vary accordingly to ethnicity, marital status, educational achievement, socioeconomic status, occupation, and religious affiliation. In virtually all of these categories, however, excessive drinking appears to be primarily a male phenomenon. Although the percentage of drinkers among married men and women are roughly equal, married men are more likely to drink excessively than married women, at a nearly 3-to-1 ratio. Widowed elderly females, as found in previous research, were most likely to abstain [14]. In the “other” category (single, divorced, never married), men were 5 times as likely to report excessive drinking as were women (17% versus 3%). These findings are in line with previous research that found heavier drinking prevalent among single and divorced elderly individuals [3]. There are relatively higher proportions of abstainers among lower educated older persons (8th grade and below) in the general population and higher proportions of drinkers and heavier drinkers among groups with higher levels of education. Highly educated older persons, both men and women, are more likely to be among the “heavier” drinkers [76]. Professional and white-collar workers among both sexes (men = 17%; women = 9%)and managerial-administrators-salespeople among men (22%) were more likely to report overindulgent drinking. These groups, as noted in past research [48], are also at “high risk” for problematic drinking. People in the lower income brackets (below $5,000annual income) were more likely to be abstainers than those at higher income levels. Similarly, among religious groups, Catholics and respondents with no religious affiliation or “other” were more likely than Protestants and Jews to report excessive drinking. Independent Variables and Drinking Behavior Role loss, affiliation (or social integration), alienation, health status, and socioeconomic characteristics such as employment status have been hypothesized to impact drinking behavior among the elderly. One-way analysis of variance (F = p c .05) and multiple comparison (Alpha = p c .05) statistical techniques were used in the preliminary stage of data analysis to test for main effects of all hypothesized variables on drinking. The findings, shown in Table 2,indicate that older adults who experience stressful life event losses are significantly more likely to drink excessively than those who have not experienced such losses or who have experienced them to a lesser extent. These findings suggest, as one possible interpretation of the data, that increased drinking may be a reaction to life circumstances in which alcohol represents an attempt to cope with traumatic loss newly experienced in old age [76].
Race: White Black Marital: Married Widowed Other Education: Grammar Grades 7-8 High school Some college College grad Grad school
60t
Age:
(N)
54
41 28 32 13 17
71 63 44 38 30
40
(54) (25)
(90) (198) (403) (103)
4 12 23 11 25 16
55 56
53
23 35 51
6 2 5 9 15 4
54 43
42 47 49 57 63 67
7 4 3
45 42 48
(345) (410) (122)
18 14 17
49 52 52
33
54
48 68
34 31
5 5 4
45
Percentage
Drink Drink Moderately Excessively
Women (N= 877)
47 28
50
(798) (72)
(877)
17 14
17
(N)
51 38
50
(%I
Abstain
32 48
33
Percentage
Drink Drink Abstain Moderately Excessively
Men (N= 537)
Table 1. Drinking Behavior and Sociodemographic Characteristics of Men and Women Ages 60 and Over
(%)
(0
0
A
Other
$5-9,999 $10-14,999 $15-19,999 $20-24,999 $25,000> Religion: Protestant Catholic Jewish None
c$5,000
Occupation: Prof-tech Man-adm-sales Clerical Craftsmen Operatives Farmers-lab. Service wk. Income:
47 56 86 44 62
38 23 14 31 13
58
53 54
41 47 47
55 68 52 45 30 57
53
56 38 36 19 26 22
30 23 16 30 39 60 30
25 25
-
15 21
3 15 17 28 16 24
17 22 16 18 16 10 13
58 28 37 23 38
71 50 42 37 31 23
38 39 39 48 56 71 55
71 62
63
37 64
57 59 66
53
26 45
57 57 39 36 29 41
53
6
-
-
5 8
3 5 5 6 10 11
4
-
9 4 4 13 8
0
A
Respondent Divorce last 5 years Unemployed last 5 years Family deaths (mother, child, father, sibling, spouse)
ROLE LOSS
Respondent: Affiliation Alienation Physical health Employed
DISAGGREGATE MEASURES
Role of Loss
NS
,2465 ,4054
105.3852 4.1596 84.5203 3.7361
Between Group Sum of Squares
2 2
DF
.1232 ,2027
52.6926 2.0798 42.2602 1.8681
Mean Squares
4.4615 2.7113
13.3268 13.3268 15.7146 10.8830
F Ratio
Table 2 Analysis of Variance: Main Effects of 26 Independent Variables on Drinking Behavioa (60 Years Old and Over) (N = 1418)
NS
.01 .06
.01 .001 .001 .001
F P
THE IMPACT OF STRESSFUL LIFE EVENTS AND SOCIAL SUPPORT ON DRINKING AMONG OLDER ADULTS: A GENERAL POPULATIONSURVEY*+
KAREN M. JENNISON, PH.D. University of Northern Colorado, Greeley
ABSTRACT
This article is an analysis of stressful life events, the buffering hypothesis, and alcohol use in a national sample of 1,418 respondents 60 years of age and over. The results indicate that older adults who experience stressful losses are significantly more likely to drink excessively than those who have not experienced such losses or who have experienced them to a lesser extent. Increased drinking among older adults may therefore be. a reaction to life circumstances in which alcohol represents an attempt to cope with traumatic loss, personal as well as within the kinship network. Supportive resources of spouse, family, friends, and church appear to have a stress-buffering effects that reduces the excessive-drinking response to life crisis. Data suggest, however, that older persons are vulnerable to the magnitude of losses experienced as they grow older and lose more of their family, friends, and peers. These stressors appear to seriously impact their drinking behavior and are not effectively buffered. Respondents report that drinking may increase during periods of prolonged exposure to emotionally depleting life change and loss, when supportive needs may exceed the capacities of personal and social support resources.
*Data used in the study were made available from the National Opinion Research Center (NORC) General Social Surveys. The analyses and interpretation of data presented herein are solely the author's. 'Revision of a paper presented at the National Conference on Aging in the 1990s.Alcohol and Other Drug Abuse, Novi, Michigan, April 1990. 99
0 1992, Baywood Publishing Co.,Inc.
doi: 10.2190/F6G4-XLV3-5KW6-VMBA http://baywood.com
100 / JENNISON
INTRODUCTI0 N A review of the literature on drinking behavior among elderly individuals, particularly within the last decade, suggests that many explanations of excessive drinking within this social group remain controversial. A major perspective that has been the subject of considerable debate, is the assumption that older people who experience stressful life event losses, including loss of loved ones, health, occupation, and status, are likely to drink excessively. This research has focused on the aging-drinking relationship and the impact of losses on drinking as people grow older [l-111. A growing number of researchers have come to believe that research emphasizing increased drinking as a response to stresses associated with aging is deficient; it is long on theory and short on empirical documentation [12]. The stressful loss research approach to deviant drinking among elderly individuals may, in fact, exaggerate the association of stress-linked alcohol abuse with chronological age, when other factors may provide more plausible explanations of drinking behavior in older groups [12-151. Yet, there have been few studies with national samples investigating the stressdrinking relationship. The purpose of this study is to present general information on national drinking patterns among noninstitutionalized elderly individuals and to test three major drinking-related perspectives: 1) older persons increase their alcohol consumption in response to major life transitions involving loss; 2) the effect of stressful life events may be buffered, or modified, for older persons who have sufficient support from various social groups such as intermediate or extended family, friends, or other groups [16]; and 3) the effects of other factors such as group affiliated drinking, alienation, and health status may have greater impact on elderly drinking than explanations based on the traumatic losses they experience over time [17-191. Stressful Life Events as Role Loss
The prevailing stress-loss hypothesis, widely tested in clinical studies and tested here in the general population, posits that individuals who experience stressful life losses will be more likely to drink excessively than those not experiencing losses, and that excessive drinking will be closely associated with a specific crisis situation [20]. The idea that most life transitions are stressful and that they lead to various coping strategies has frequently been researched [21]. Alcohol abuse among elderly individuals has been identified as a major response to the stresses associated with growing older (see Elwell and Matlbie-Crannell[4] for a discussion of stress and the elderly, and fennison and Johnson [20] for a discussion of the concept of stress in relation to alcohol use). There has been little empirical
STRESSAND DRINKINQ / 101
evidence, however, to support the theory that stressful losses among aging individuals are associated with increased, or problematic drinking [9,12,13,22]. Here, we are dealing only with loss type events; thus, the traditional concept of role loss or exit seems to be an appropriate way of linking social structure and the life course to drinking behavior [23]. Role loss occurs whenever any stable pattern of interaction and shared activity between two or more persons ceases [2]. Whether the event is viewed as loss, separation, departure or ending, and whether these losses are temporary or permanent, the process can be traumatic for an individual. Within the last few decades, stress and loss research has been developmentally integrated into the concept of “psychosocial transitions” [24-271. The critical factor for these researchers is not only the fact of loss, or crisis, but recognition of the possible duration of an individual’s progression through phases of the transition. Longer transitions are characterizedby chronically excessive demands which may seriously tax and undermine the resources available to the individual. Psychosocial transitions involve re-definitions of the individual’s assumptive world as well as life reorganization with its potentially devastating emotional impact. Because role loss in many instances is unexpected, “role discontinuity” occurs.With little preparation, the new role requirements and status arrangements may be sufficiently stressful to be classified as a crisis experience [B-311. A stable pattern of interaction with other people may be interrupted not only by one’s own stressful event, such as unemployment or illness, but also by such stressful events in the lives of other persons who are role partners. Extending the effects of the crisis experience of the individual to encompass the role losses of relatives is termed “network crisis events” [32], or “multiple crises” [33].
Affiliation Drinking behavior among the elderly, as among other groups, may be most completely explained when interpreted within a social context [34, 351. Group memberships and activities of various types are related to excessive drinking because drinking is often normative in these groups [36]. Drinking is known to increase in drinking-centered group activities where heavy social drinking takes place. Akers [37] defines drinking-centered groups as including tavern groups, formal and informal parties, neighbors, clubs, ceremonial and celebration gatherings, conventions, and similar group drinking events; these drinking groups are also attractive and rewarding for the drinker, and even give status to the skilled drinker (e.g., who can “drink like a man,” or “hold his liquor”). These groups legitimate drinking as a way of developing congenial social interaction. The more socially active, therefore, are also more likely to be heavy drinkers [38].
102 I JENNISON
Other examples of group-affiliational drinking, in which culturally conventionalized inebriety [39] is acceptable as long as it remains within the limits of the group norms, are found among: college students [40-441; lower-status occupational groups [45], white collar executives [46], college professors [47], salespeople and professional-technical employees [48], college fraternities and sororities [49], blue collar workers [36, 50, 511, and professionals in medicine, law, dentistry, nursing, and social work [52]. Even though most members of these groups do not develop alcoholism, focusing on affiliational group drinking tests the assumption that it is one factor related to problematic drinking [45]. In testing the affiliation hypothesis using national data, therefore, we expect that affiliation as well as social activity will be related to excessive drinking among the elderly, particularly among men and among higher socioeconomic status groups [38], since these are both categories with relatively few elderly abstainers.
Alienation At the opposite end of the continuum, many older drinkers and older problem drinkers have been characterized as individuals who experience alienation and isolation from the mainstream of society [9, 111, either by living arrangement (living alone), loss of family members (e.g., widowed, separated, divorced) or by reduced social participationbecause of retirement, lowered income, or diminished social support resources. The concept of alienation originated in 19th-century social thought where the entire democratic industrial order with its urbanism, secularization, complex technology, and inaccessible bureaucracy was viewed as threatening human dignity and social cohesiveness [53]. It was the antithesis of what enlightenment thinkers had imagined as the result of social progress. Thus alienation, as defined here, refers to powerlessness, estrangement, and social isolation, the perceived lack of control, or mastery over sociopoliticalevents that affect one’s life [19,54, 551. The alienation drinking connection arises out of the same social conditions which promote a generalized alienation from self, work, and society [a]. Alienation has been negatively correlated with social activity in previous studies. That is, the alienated individual who drinks heavily shows minimal levels of affiliation, social group memberships or social participation. This is especially the case among “early on-set,” or under 30-year-old heavy drinkers [49,56], elderly male problem drinkers [16], and escape drinkers, those who drink for relief from anxiety, depression, and personal problems [38]. Heavy drinking itself would be likely to increase a sense of powerlessness, especially if it merges into alcoholism with its attendant life problems.
STRESS AND DRINKING I 103
Empirical studies linking alienation with excessive drinking tend to reinforce the “powerlessness” dimension; less importance is attached to the estrangement or social isolation components of the alienation concept. Calicchia and Barresi [17], an exception to this trend, found that alcoholics as a group were significantly more alienated in the social isolation sense, as measured on Dean’s Social Alienation Scale, than nonalcoholics. But male alcoholics experienced, alienation more in the powerlessness form. Previous studies have shown that alcohol could be used as a coping mechanism to relieve anxiety brought on by stress in people with a high sense of powerlessness [57]. In leading studies of drinking and alienation, Seeman and colleagues [18, 19,551 tested three measures of alienation: work experience (low job complexity), powerlessness (generalized expectations for control, or mastery, over outcomes), and social isolation (degree of individual attachment to significant others: friends, relatives, neighbors) in a sample of 450 employed males in a metropolitan Los Angeles community and found in both the first and follow-up studies that powerlessness, as a form of alienation, is directly related to drinking and drinking problems; increased powerlessness is associated with increased drinking problems. Therefore, measures of alienation used in this study primarily assess the powerlessness dimension (sense of personal control or mastery over life). Older people who are alienated in the sense of subjective control, should be more likely to drink excessively for power considerations. Further, older people with high alienation, or low mastery [58], would be expected to exacerbate the distressing impact of major role loss events by drinking even more.
SOCIAL SUPPORT Social support may be defined as “support accessible to an individual through social ties to other individuals, groups and the larger community” [59, p. 1091. Social ties may be seen as having stress or anxiety reducing features which make coping with major role changes less difficult. Various formal and informal groups, of course, fit into this category. Older problem drinkers, for example, have been found to have few friends and supportive family ties [60], and diminished social, economic, and health resources in general, particularly older men alcoholics [61]. Older problem drinkers are also likely to be dissatisfied with their relationships with family members, spouses, and close friends [14]. Marital discord, or dissatisfaction, has been described as an “enduring strain” or chronic stressor [62], and is highly predictive of increased alcoholic use. Social resources play an important role in buffering the impact of life stress on health [63]; some studies indicate that fewer support resources negatively impact alcohol treatment outcomes as well, resulting in less successful treatment [61]. Therefore, the older
104 / JENNISON
person’s social connectedness may influence drinking behavior following traumatic losses. METHODOLOGY Sample
Data utilized in the study consist of a general population sample of 1,418 respondents sixty years old and over in four pooled surveys drawn from the National Opinion Research Center (NORC) General Social Surveys [MI: years 1978 (N= 339; 23.9%),1980 (N= 361; 25.5%), 1983 (N= 373; 26.3%)and 1984 (N = 345; 24.3%). GSS survey questions on drinking behavior are “rotating items”; they only appear periodically, and not always in the same year. The sampling years selected include matching questions on both alcohol and independent variables, and represent maximum item equivalence of cross measurement indicators in survey replication. The use of comparable pooled surveys [65] thus minimizes the undenesponse problems in data analysis due to the absence of a question in any one particular year. Basic data are cross-sectional baseline prevalence drinking patterns derived from GSS survey questions on self-reported use or experience with alcoholic beverages (liquor, wine, or beer) and the respondent’s subjective evaluation of these experiences (for example, perceived intoxication effects). The conceptual measures of drinking behavior are continuous rather than dichotomousor discrete. Additional characteristics of the research design and methodology used in the study have been described elsewhere [20]. Measuringthe Variables
For purposes of this study, following previous research, stressful role loss is defined as independent disaggregated events (divorce, unemployment-job loss, family deaths, hospitalization, or disability) happening to the respondent and to relatives last year and within the last five years prior to the research. In addition, independent aggregated role losses are operationalized as the total number of cumulative events (divorce, unemployment or job loss, family deaths, hospitalization, or disability) happening to the respondent and relatives combined, last year and within the last five years. The two measures of respondent-relative role loss have been combined in order to estimate the relationship of role related or “overlapping” strain and drinking behavior. Affdiation Measures are defined as the number of voluntary group memberships in specialized formally organized social groups or organizations in which membership is based on deliberate choice and from which members may
STRESS AND DRINKING I 105
resign (16-item dummy variable, reference or excluded group coded 0 = no membership). Social support measures, using Herzog’s [66]rationale, were categorized into two separate groups: 1)social indicators of objective estimates of role resources, and 2) life satisfaction measures, or social group indicators of subjective assessments of major role resources. Objective support resources, or familial variables, are 1)marital status (0 = married; 1 = otherwise); 2) family size (1; 8 or more), 3) number of siblings (0;8 or more), and church attendance (0 = never; 7 = several times a week). These are commonly documented measures of structural social support (the perception that support is available from the network of family and friends if needed) which have been used extensively in previous research on buffering and social support [67-741. Subjective assessments of the quality of life, or life satisfaction and happiness, are defined as the degree to which respondents are satisfied with 1) friendships, 2) family life (coded 0 = no satisfaction; 7 = very great deal of satisfaction), 3) personal finances (0 = low; 3 = high satisfaction), and 4) job situation (1 = very dissatisfied; 4 = very satisfied). Therefore high scores on life satisfaction measures are equated with high satisfaction. The quality of health, or personal estimate (0 = none; 4 = fair amount of satisfaction) of health status, is used as a separate independent variable with drinking behavior. Alienation, defined as powerlessness, or lack of control or mastery over sociopolitical events that affect one’s life, is operationalized as how much confidence the respondent has in people running the following American institutions: 1) banks and financial institutions, 2) major companies, 3) education, 4) the executive branch of the federal government, 5 ) Congress, and 6) the military (coded 1= a very great deal of confidence; 3 = hardly any confidence). Thus, low confidence is equated with high alienation. An additive index of alienation was developed (coefficient of reliability = .97) based on the model of Herzog, Rodgers, and Woodworth [66]. Employment status is operationalized as working full- or part-time, versus being unemployed, retired, in school, or a home maker. The U.S.Bureau of the Census 3-digit occupational classifications for 1970 occupations are used as indicators of socioeconomic status in the study. Prestige scores were assigned for occupations based on the National Opinion Research Center (NORC)rating system [a].
Statistical Measures We test the proposition that social support relationships will intervene (buffer) between role losses and drinking measures using analysis of variance and multiple regression techniques. Groups with statistically significant f-values on drinking behavior using analysis of variance and multiple comparison procedures
106 / JENNISON
(Alpha = < .05)were selected to be included in multiple regression models. These statistical methods treat the predictor variables of role loss, health, alienation, affiliation, employment status, and social support as well as the criterion variable alcohol as continuous. In multiple regression models, indicators of each type of independent variable are first tested separately with drinking to determine any independentmain effects between independent variables and drinking behavior. In the second step, in which the multiple interaction term (independent variable x social support) is forced into the equation (forced entry), alcohol drinking patterns are regressed on independent variables separately for each social support group to assess any degree of reduction in the regression coefficient between independent variables and excessive drinking behavior when the intervening support variable is included. Last, to assess the total effect of all hypothesized social resource buffers, social support indicators, role losses, and other independent variables are tested simultaneously in the regression model, along with socioeconomic variables, for every support group which produced any reduction in the role loss-excessive drinking relationship [32,66,69,74]. Previous research suggests that those who have supportive social networks are protected, or buffered, in times of crisis. If role loss buffering is demonstrated for older persons in the study, then the effects of role loss on drinking behavior should decrease when social support is controlled. Differentially, high role loss in the absence of social support should be associated with increases in vulnerability to alcohol. According to the buffering model and patterns of results consistent with the buffering postulation, social supports or resources should either partially reduce or totally (statistically) ameliorate the effects of role losses on drinking [69]. The impact of support on the independent variables-drinking relationship should also be maintained to a significant degree after controls for the effects of marital status, education, and total family income are established. The assumption is that those respondents with adequate social resources (actual and perceived) temper or moderate their drinking while those respondents with weak resources do not.
RESULTS
Sample Characteristics Sociodemographiccharacteristicsof the general elderly population and national prevalence drinking patterns are shown in Table 1. These patterns indicate that one-third of men respondents in the study abstain and nearly one-fifth report excessive drinking. Among older women, on the other hand, fifty percent report total abstention from drinking alcohol and only five percent say that they
STRESS AND DRINKING / 107
sometimes overindulge. Excessive drinking among elderly individuals, as with drinking in general, is disproportionately a male activity. In every age group, more men than women drink, and- among those who drink, there are more heavy drinkers among men than among women [75]. Drinking practices among the elderly, as in younger age groups, vary accordingly to ethnicity, marital status, educational achievement, socioeconomic status, occupation, and religious affiliation. In virtually all of these categories, however, excessive drinking appears to be primarily a male phenomenon. Although the percentage of drinkers among married men and women are roughly equal, married men are more likely to drink excessively than married women, at a nearly 3-to-1 ratio. Widowed elderly females, as found in previous research, were most likely to abstain [14]. In the “other” category (single, divorced, never married), men were 5 times as likely to report excessive drinking as were women (17% versus 3%). These findings are in line with previous research that found heavier drinking prevalent among single and divorced elderly individuals [3]. There are relatively higher proportions of abstainers among lower educated older persons (8th grade and below) in the general population and higher proportions of drinkers and heavier drinkers among groups with higher levels of education. Highly educated older persons, both men and women, are more likely to be among the “heavier” drinkers [76]. Professional and white-collar workers among both sexes (men = 17%; women = 9%)and managerial-administrators-salespeople among men (22%) were more likely to report overindulgent drinking. These groups, as noted in past research [48], are also at “high risk” for problematic drinking. People in the lower income brackets (below $5,000annual income) were more likely to be abstainers than those at higher income levels. Similarly, among religious groups, Catholics and respondents with no religious affiliation or “other” were more likely than Protestants and Jews to report excessive drinking. Independent Variables and Drinking Behavior Role loss, affiliation (or social integration), alienation, health status, and socioeconomic characteristics such as employment status have been hypothesized to impact drinking behavior among the elderly. One-way analysis of variance (F = p c .05) and multiple comparison (Alpha = p c .05) statistical techniques were used in the preliminary stage of data analysis to test for main effects of all hypothesized variables on drinking. The findings, shown in Table 2,indicate that older adults who experience stressful life event losses are significantly more likely to drink excessively than those who have not experienced such losses or who have experienced them to a lesser extent. These findings suggest, as one possible interpretation of the data, that increased drinking may be a reaction to life circumstances in which alcohol represents an attempt to cope with traumatic loss newly experienced in old age [76].
Race: White Black Marital: Married Widowed Other Education: Grammar Grades 7-8 High school Some college College grad Grad school
60t
Age:
(N)
54
41 28 32 13 17
71 63 44 38 30
40
(54) (25)
(90) (198) (403) (103)
4 12 23 11 25 16
55 56
53
23 35 51
6 2 5 9 15 4
54 43
42 47 49 57 63 67
7 4 3
45 42 48
(345) (410) (122)
18 14 17
49 52 52
33
54
48 68
34 31
5 5 4
45
Percentage
Drink Drink Moderately Excessively
Women (N= 877)
47 28
50
(798) (72)
(877)
17 14
17
(N)
51 38
50
(%I
Abstain
32 48
33
Percentage
Drink Drink Abstain Moderately Excessively
Men (N= 537)
Table 1. Drinking Behavior and Sociodemographic Characteristics of Men and Women Ages 60 and Over
(%)
(0
0
A
Other
$5-9,999 $10-14,999 $15-19,999 $20-24,999 $25,000> Religion: Protestant Catholic Jewish None
c$5,000
Occupation: Prof-tech Man-adm-sales Clerical Craftsmen Operatives Farmers-lab. Service wk. Income:
47 56 86 44 62
38 23 14 31 13
58
53 54
41 47 47
55 68 52 45 30 57
53
56 38 36 19 26 22
30 23 16 30 39 60 30
25 25
-
15 21
3 15 17 28 16 24
17 22 16 18 16 10 13
58 28 37 23 38
71 50 42 37 31 23
38 39 39 48 56 71 55
71 62
63
37 64
57 59 66
53
26 45
57 57 39 36 29 41
53
6
-
-
5 8
3 5 5 6 10 11
4
-
9 4 4 13 8
0
A
Respondent Divorce last 5 years Unemployed last 5 years Family deaths (mother, child, father, sibling, spouse)
ROLE LOSS
Respondent: Affiliation Alienation Physical health Employed
DISAGGREGATE MEASURES
Role of Loss
NS
,2465 ,4054
105.3852 4.1596 84.5203 3.7361
Between Group Sum of Squares
2 2
DF
.1232 ,2027
52.6926 2.0798 42.2602 1.8681
Mean Squares
4.4615 2.7113
13.3268 13.3268 15.7146 10.8830
F Ratio
Table 2 Analysis of Variance: Main Effects of 26 Independent Variables on Drinking Behavioa (60 Years Old and Over) (N = 1418)
NS
.01 .06
.01 .001 .001 .001
F P
