SM. SC;. Med. Vol. 22. No. 5. pp. 509-516. Printed in Great Britain
1986
0?77-9536/86
S3.00 + 0.00
Pergamon Press Ltd
UNEMPLOYMENT STRESS: LOSS OF CONTROL, REACTANCE AND LEARNED HELPLESSNESS AXDREW
BAUW’,RAYMONDFLEMIEG’and DIANE M.
FREDDY?
‘Uniformed Services University of the Health Sciences, F. Edward Hebert School of Medicine. 4301 Jones Bridge Road, Bethesda, MD 20814 and *University of Wisconsin at Milwaukee, Milwaukee, WI 53201, U.S.A.
Abstract-The present study was concerned with stress-related consequences of unemployment and the behavioral changes related to this experience of control loss. Subjects were sampled along a continuum of time since unemployment, including a control group of employed subjects. Results indicated evidence of stress responding among the unemployed subjects, measured as increased levels of urinary catecholamines and behavioral performance deficits. Further evidence is presented which analyzes subjects’ attributions and behaviors in terms of the theories of reactance and learned helplessness. Results support a biohasic response to loss of control with reactance manifested at early stages of control loss and learned
helplessness at later stages.
As researchers have come to more fully understand the role of occupational characteristics and job stress in health and well-being, recognition of the consequences of job loss and unemployment has been more pronounced. Where research was initially limited to studies of self-esteem and morale in the face of the Great Depression, it is now focused on the complex psychophysiological effects of stress associated with involuntary unemployment. Anticipation of job loss has been considered, as have the immediate and longer-term implications of being out of work [l-4]. The present literature discusses a number of indices of strain and pathology as they relate to unemployment and suggests that health may be impaired during periods of joblessness [5]. However, the processes by which unemployment is translated into psychological, behavioral and physiological changes have not been clearly specified. This paper is concerned with the ways in which expectations, attributions and other psychological events affect the stress associated with unemployment and how they may ultimately influence behavior and health. PREVIOUSRESEARCHON UNEXIPLOYMENT Early research on unemployment was conducted between 1930 and 1940, studying the plight of victims of the Great Depression. Most of these efforts were concerned with self-esteem and morale. Not unexpectedly, they found that involuntary joblessness was associated with lower self-esteem and morale [6-S]. Clearly, the events of the 1930s were unlike those of most other periods in history, and victims of that upheaval were affected in different ways than are today’s unemployed. However, the negative This research was facilitated by research grants from the Uniformed Services University of the Health Sciences (CO7205) and the National Science Foundation (BNS8317997). The opinions or assertions contained herein are the private ones of the authors and are not to be construed as official or reflecting the views of the Department of Defense or the Uniformed Services University of the Health Sciences.
impact of unemployment observed in those studies is consistent with more recent findings. Modern research on job loss and unemployment has considered a variety of effects, ranging from changes in attitudes and emotional response to hormonal changes and death. For example, Little [9] found positive attitude change among white-collar unemployed, and Goodchilds and Smith [lo] found decreased self-esteem among people who had lost their jobs, while Cobb [l l] reported elevated urinary norepinephrine, creatinine and uric acid among unemployed subjects relative to levels recorded for employed individuals. The point is that a number of measures of stress or other negative effects of unemployment have been used, often with little consideration of how they may be related. We will briefly review the literature on unemployment, grouping studies around these different measurement approaches. We will then present findings from a study that we have conducted and will attempt to integrate the studies based on different measures. Job loss and self-worth
A number of studies have examined attitudes and feelings about self-worth following job loss. Consistent with research during the 1930s recent studies have found that people who have been fired or laid off from work express more feelings of wonhlessness [ 10, 12). This finding is typically explained in terms of loss of one’s ability to provide and care for oneself and one’s family, as well as the general effects of loss of control. However, at least one study has not found negative self-evaluations following job loss, finding instead that these feelings were nrore positice 191. Surprisingly, these positive attitudes persisted and did not show the expected decrease with increases in the duration of_ unemployment. After about 8 months, unemployed subjects actually expressed more positive feelings than they had shortly after losing their jobs. This discrepant finding points out some problems with an approach to studying unemployment that relies only on self-report measures. The fact that subjects in this study were white-collar engineers may have led to this finding since they may hav-e had
reasonable financial reserves to draw upon. ..Uternatively, these responses may have been affected by dissonance reduction or the desire to maintain consistency of self-image. These individuals may have viewed themselves as competent and valuable to their field and, to avoid a discrepancy with the fact that they were out of work. may have emphasized the positive aspects of unemployment in an attempt to convince themselves that their situation was not inconsistent with what they really wanted. Regardless, it is very difficult to interpret findings such as these when they are not discussed in the context of other changes accompanying unemployment. Emotional response to unemployment Some studies have examined emotional changes following job loss and continuing through unemployment. Generally, this research suggests that being out of work is a negative experience. Warr [l3. 141 has reported evidence of declines in psychological health following job loss, as well as behavior changes during unemployment. O’Brien [ 151noted higher frequencies of reporting of troublesome somatic symptoms among unemployed subjects, and research on the general relationship between economic conditions and emotion suggests that emotional disturbances are more likely during periods of economic change [l6]. One study specifically examined the relationships between unemployment and depression by comparing scores on the Beck Depression Inventory among automobile workers who had or had not been laid off [l7]. The two groups did not differ on overall levels of depression, but multiple regression analysis indicated that depression was related to different predictor variables in the two groups. Among those still working, traditional predictors such as race. marital status and education were associated with depressed mood. For subjects who had been laid off, however, the best predictors of depression were economic variables such as perceived future of the economy, layoff status and percent of salary recovered from union benefits. Thus, depression appeared to be primarily related to economic repercussions of unemployment among those who had been laid off. Emotional response to unemployment probably varies from individual to individual. Kasl [18]. for example, reports brief elevations in depressed affect and anxiety after job loss for some workers. Kasl and Cobb [I91 also report, however, that for some workers emotional strain continues even after reemployment. Those variables responsible for individual variation in response to stress have been a focus for research in several areas, and likely include such factors as social support and coping style [20,21]. Research has also suggested that changes in behavior following job loss are influenced by age and length of unemployment [l3]. Physiological changes Research has also examined bodily changes preceding and following job loss. A comprehensive study of workers has considered a number of these changes [3.4, 1I]. Blue-collar. male workers were followed over a 2-year period beginning before plant closings and were compared with a control sample. Results suggested, among other things, that urinary norepi-
nephrine levels among the unemployed workers were higher than among the control group workers. Further, these differences were apparent before job loss (while it was anticipated) and up to 12 months following job loss. By 24 months after job loss. differences were no longer significant. Cobb [I I] also reported that consumption of coffee resulted in increases in norepinephrine among workers anticipating job loss and continued to increase levels shortly after job loss. This is consistent with typical effects of coffee or other caffeine-laden food or drink: the consumption of caffeine usually increases catecholamine levels [Xl. However. this effect of coffee on norepinephrine was gradually reduced over time. As unemployment became more chronic or as re-employment occurred, consumption of coffee resulted in smaller and smaller increases in norepinephrine. Two years after job loss, coffee did not result in any elevation of norepinephrine. The observed interaction between environmental stress and coffee consumption is consistent with research showing that caffeine administration leads to increases in blood pressure among crowded rats but not among control animals [23]. Thus. the effects of stressors such as unemployment or crowding may potentiate the effects of pharmacological agents or other stimuli. Reports of this research have also considered levels of serum uric acid and serum creatinine as they were related to job loss and unemployment [3, I I]. Serum uric acid levels may reflect arousal or reactivity to stimuli and were found to increase during the anticipatory phase for those facing job loss. When workers found re-employment. serum uric acid levels returned to levels comparable to those of control subjects. Cholesterol did not shovv reliable differences between the two groups [3]. Kasl and Cobb [A] reported that unemployed workers had higher systolic and diastolic blood pressure than did the employed controls. This increase was most marked during the anticipatory period, and pressures peaked at the point of actual job loss. However, the differences persisted beyond job loss, showing a gradual decline during the I2 months following job loss. Unemploynen t and patholog) An alternative approach to studying the effects of unemployment is to examine correlations between global economic indicators and indices of pathology. Several studies have considered the relationships between indicators of economic performance and indices of strain and ill health [5]. Brenner [24. 251 has conducted such studies, examining (among other things) relationships between economic decline and rate of unemployment and various measures of mental health. The findines of this research suggest that, generally, economyc decline is related to social strains and mental health problems. More specifically, unemployment rate was related to number of first admissions to mental hospitals, incidence of infant rate of deaths due to cardiovascular mortality, disease, and incidence of suicide and alcohol-related illness [24, 251. These findings are subject to traditional limitations on correlational studies, and are relatively insensitive to situational and individual
511
Unemployment and loss of control But they suggest that unemployment may have serious health-related outcomes. Brenner’s conclusions have not gone unchallenged. Subsequent studies [26,27] found that brief upturns in economic conditions also predicted rates of mortality and psychological distress, suggesting that change rather than unemployment rate per se is responsible for these associations. However, individual level studies have found that being unemployed was associated with symptom distress in interviews up to four months after job loss [28] and that evidence of physiological changes is present through a year of unemployment [4, I I]. This persistence suggests that change is not the sole contributor to distress following job loss. More importantly, Gravelle ef al. [29] have suggested that Brenner’s analysis was inadequate and “does not support the hypothesis that aggregate unemployment rates have a serious adverse effect on population mortality rates” (p. 678). They also noted that positive associations between unemployment and mortality may be related to any one or combination of variables that tend to covary with unemployment. Some of these variables are likely to be heavily influenced by unemployment, but others may contribute to job loss. Research on the elIects of unemployment has considered a range of variables, and the evidence at least suggests that job loss and subsequent unemployment has negative consequences for health and well-being. Studies have not considered behavioral consequences of unemployment but suggest that anticipation of job loss and the period immediately following are the most aversive times. However, there is also evidence indicating that problems continue as unemployment persists. This prolonged component has not received as much attention, but there is reason to believe that unemployment is a source of chronic stress. variation.
UNEMPLOYMENT
AND CHRONIC
STRESS
The effects of unemployment, for the most part, resemble those of other stressors, and, for that reason, it may be useful to consider research on other sources of stress when discussing unemployment. Physiological responses, such as increases in urinary norepinephrine or serum uric acid, are similar to those reported in studies of occupational stress or chronic uncertainty (e.g. [29,30]). Blood pressure changes are also possible consequences of chronic stress and the general pattern of physiological change associated with unemployment resembles response to stress [3 I, 321. Chronic stress appears to be mediated by several psychosocial factors. including social support and coping style [I 1, 19,33, 341. Appraisal or interpretation of stressors is also an important determinant of stress [35]. Therefore, appraisal of job loss and associated cognitive changes involving coping or interpretation should be important determinants of unemployment stress. Perceived loss of control
Among the more important dynamics associated with unemployment is the attendant loss of control and inability to attain significant outcomes. It is reasonable to assume that several aspects of job-
lessness center around a loss of control, ranging from symbolic reflections of an uncontrollable world to specific mundane problems such as paying the rent, obtaining health insurance and so on. Research on other stressors suggests that having control (or believing that one has it) reduces stress, while low perceptions of control are associated with more serious stress consequences (e.g. [36.37]). To the extent that losing one’s job reduces one’s perceptions of control over the environment, it should facilitate stress-related effects such as emotional distress and sympathetic arousal. Of perhaps greater interest are the possible links between unemployment and learned helplessness-a state of low motivation and depressed affect conditioned by prolonged exposure to uncontrollable surroundings [2]. When people are exposed initially to uncontrollable events, they may become aroused and angry and direct their behavior toward regaining control. As time passes and exposure to uncontrollable events continues, however, expectations of being able to regain control may wane. When expectations have dimunshed to the point where successful control seems improbable. attempts to regain it may cease and negative emotions increase. This sequential pattern of reactant and helpless behavior was proposed by Wortman and Brehm (381and suggests that expectations mediate response to chronic stress. METHOD
this pattern of expectation-based response also characterizes unemployment, one would expect to find that recently unemployed individuals are more concerned with regaining or asserting control and are more confident of doing so than are people who have been unemployed for a longer period. The longer one remains unemployed, however, the lower these expectations should be and behavior should be more helplessness-like. In other words, helplessness should intensify over time following job loss. In order to consider this possibility, a study was conducted to measure physiological and behavioral changes associated with job loss and unemployment. In addition, changes in cognitive structure, including attributions of responsibility for outcomes and expectations for control or success, were measured. The time period studied ranged from immediately after job loss through 5 months of prolonged unemployment. Using a cross-sectional design, unemployed subjects were considered < 3 weeks after job loss, 3-8 weeks after job loss. and, in the chronic condition, > 8 weeks after job loss. Half of the subjects in each group were exposed to an unsolvable (and, hence, uncontrollable) task during the session, while the others found the same task to be solvable. This manipulation of controllability of outcomes allowed observations of reactant and helpless responses to noncontingency. Ali unemployed groups were compared with a group of employed control subjects. It’
Subjects
Subjects were recruited to till four groups and were employed, unemployed for <3 weeks, unemployed for 3-8 weeks, or unemployed for >8 weeks. All subjects were volunteers. Unemployed subjects were recruited from a local unemployment otTice (the
ANDREW BALM et al.
512
Employment Securities Administration in Wheaton, Md) and employed subjects were recruited through an announcement in a local newspaper. Forty subjects, ten in each condition, participated in this study. .Cfeasures and procedures
Selection of measures was based on the various objectives of study, including demonstration of stress-related effects of unemployment, assessment of helplessness-like effects of unemployment, and examination of the relationships between appraisal of control and behavioral and physiological response during unemployment. Urine samples were collected in order to measure urinary epinephrine and norepinephrine and thereby index hormonal aspects of stress responding. An embedded figures task (EFT), requiring both concentration and persistence, was included to measure behavioral aspects of stress and helpless responding. Subjects were examined individually in a laboratory setting. Upon the subjects’ arrival, the experimenter explained the nature and requirements of the study. All subjects were provided with written descriptions of the study and informed consent was obtained. Then the subject provided a urine sample, which was immediately preserved and frozen. A questionnaire containing items concerning employment and demographic information was completed by the subject before tasks began. The first task was a version of the Levine task [39]. The subject was shown a series of slides with a figure on the right and a figure on the left of the slide. The object was for the subject to determine which of the parameters was the correct one. For example, in Fig. 1, if the letter on the right was chosen by the subject, it could be the correct choice because the correct parameter was either the smaller letter, the white letter, the ‘X’, the letter within the circle, or the letter on the right side of the screen. Each slide in the series changed with respect to these five parameters, but the -correct’ parameter remained the same for each series. In other words, for a given series, the correct parameter might be ‘the white letter’. The subject chooses either letter for each slide and can determine which parameter yields the correct response by a process of elimination (e.g. if S chooses the ‘X’ in the above, this choice may be correct because it is the ‘X’, the letter on the right, the smaller letter and so on). Feedback on choices on subsequent slides provided information allowing elimination of alternatives. This task was not included as a dependent measure, but rather as a means of manipulating subjects’ experiences in the session. Half of the subjects in each
I Fig.
T I. Sample
condition were provided with random feedback during the task, while the other half were given veridical feedback. Subjects given random feedback were unable to solve the task, while the others were able to solve it. During this task, subjects responded to questions designed to assess subjects’ concerns. After each series of slides, subjects ranked five types of information about the task in the order of their importance to the subject during the preceding series. These statements referred to possible causes of their performance-whether the task was too d@cttlr to solve, how much effort was required, whether subjects had the skills required to solve the task, whether the task could be solved by anyone (whether it was solvable) and what the experimenter learned about subjects based on their performance. Thus, the statements reflected concern with personal or internal factors (effort, skill) or external factors (task difficulty, solvability) and provided information about subjects’ expectations as the task proceeded. Following completion of seven series of Levine tasks, subjects were asked to work on the EFT. The task consisted of 16 complex geometric figures in which simpler target figures were hidden. Subjects were instructed to locate and trace the target figure hidden within each complex figure. The same measure of cognitive concerns used during the Levine task was administered following the EFT. RESULTS
One-way analyses of variance were performed on the data for catecholamine levels and task performance. Ratings of expectations were also subjected to analyses of variance and a repeated measures design was used when multiple measures were considered. Persistence on the embedded figures task was assessed by measuring the amount of time spent on each puzzle, and these data were analyzed using a repeated measures analysis of variance treating puzzle number as trials. All group comparisons were made using procedures suggested by Tukey’s tests [40]. In order to demonstrate that the samples of unemployed and employed subjects were comparable to one another and to rule out the possibility that background variables were responsible for differences among these groups, demographic data were collected and analyzed. These data suggested that, on variables related to issues other than employment, the groups were comparable. Most subjects were under 40 and 35% of all subjects were married. No significant differences among groups were found for length of time subjects had been married, family size, family size when growing up, education. type of residence or whether the residence was owned or rented. Further, income levels prior to becoming unemployed were comparable across unemployed groups and similar to reported income of the employed subjects. Levels of urinary epinephrine and norepinephrine revealed an effect of length of unemployment. Subjects who were unemployed tended to have higher levels of both hormones than did those who were not unemployed, and levels increased as the length of unemployment increased, F(3,36) = 3.72, P < 0.05 and F(3,36) = 3.56, P < 0.05.
x (=I
slide from the Levine task.
Unemployment and loss of control
0
NOREPI
q
EPI
NC ”
NC
I
I
I Employed
Time
I
I
3 weeks
course
3 weeks TO Zmonths
c Zmontnr
of unemploymenl
Fig. 2. Changes in urinary catecholamine levels for unemployed subjects and controls receiving contingent (C) and noncontingent (NC) feedback during the experiment.
subjects showed different More importantly, patterns of change in hormone levels during the experimental session. Specifically, those subjects who received contingent feedback during the Levine task showed net decreases in norepinephrine and epinephrine during the session, regardless of employment status. Subjects who received noncontingent feedback, on the other hand, showed different tendencies depending on employment. Those who had jobs or who had recently been unemployed (C 3 weeks) exhibited sharp increases in both epinephrine and norepinephrine, while those who had been unemployed at least three weeks showed decreases in levels of these hormones over the course 120
-
100
-
513
of the session, F(3,32)=2.091, P ~0.12 and F(3,32) = 4.283, P < 0.05, respectively (see Fig. 2). While contingent feedback did not result in increases in sympathetic arousal, noncontingent feedback did result in higher arousal levels when expectations for control were likely to be greater. Findings for persistence on the EFT, indexed as the amount of time spent working on it, revealed that the longer subjects had been unemployed, the less persistent they were, F(3,36) = 3.21, P < 0.05. Employed subjects (X = 64 seconds/puzzle) and recently unemployed subjects (X = 60 seconds/puzzle) spent more time working on the task than did subjects in either of the other groups (X = 50 seconds/puzzle; X = 44 seconds/puzzle). Employed subjects were also more accurate than any of the unemployed subjects, solving more of the puzzles than the others, F(3,36) = 3.39, P < 0.05. Examination of persistence on this task by subjects in short- and long-term unemployment conditions provides additional information about response to random, noncontingent feedback. As can be seen in Fig. 3, when contingent feedback was provided during the Levine task, subjects who had been unemployed ‘for < 3 weeks and those out of work for >2 months responded comparably across the 16 EFT items. However, when noncontingent feedback was provided, responses by subjects in these unemployment conditions diverged. Subjects who had recently lost work became more persistent over time and those who had been unemployed longer than two months became less persistent over time. This pattern is similar to that characterizing the catecholamine findings. Subjects were asked to rank order a set of five statements that reflected attribution for possible causes of performance on the Levine task and the EFT. Six sets of rankings were obtained during the Levine task so that changes in concerns could be detected. The data suggest that subjects who had been unemployed for >2 months were concerned with different issues than were control subjects or subjects who had been unemployed for ~3 weeks. As can be seen in Table 1, ratings of the importance of ability suggest that ability or skill concerns
__,O
-=3 weeks NC
P
Ti -
.’
ao-
E
5 2 60e $
x
40-
20 -
i-
.-.
I
I
-.-.
z.2 months NC
I l-2
I 3-4
I 5-6
r-a Problem
9-10
I 11-12
I 13-14
I 15-16
number
Fig. 3. Persistence on the EFT as a function of length of unemployment contingent, C = contingent).
and feedback (NC = non-
514 Table
ASDREW
I. Mean
ranks
assigned
the Levine
Employed
EffOH Task
cilwes
of performance
Un-
Un-
Un-
employed
employed
employed
3-8
wks
wks
>8
uks
2.2
2.5
2.8
2.6
2.1
2.6
3.2
3.1
3.4
3.2
2.4
4.1
3.6
3.4
2.9
become less important as length of unemployment increases, F(3,32) = 2.94, P < 0.05. Ratings of the importance of effort did not show this effect, though subjects who had been unemployed longer tended to be less concerned about this factor. Of perhaps greater interest was a pattern suggesting that the importance of these two internal attributions was not affected by the nature of feedback on the task for most groups, but that subjects who had been unemployed the longest reported very different concerns under contingent conditions (see Table 2). For those out of work the longest, exposure to contingent feedback (C) was associated with a higher ranking of ability than was exposure to noncontingent feedback (NC), F(3.32) = 10.60, P < 0.001. For effort, a three-way interaction suggested that the longest unemployed subjects were again influenced by feedback, becomong more concerned with effort over time when feedback was contingent and less concerned over time when feedback was noncontingent, F(l5.160) = 2.93, P < 0.01. Findings for external factors, task difficulty and solvability, were consistent but opposite of the data reflecting concern for internal causes of performance. Task difficulty was ranked more highly by subjects out of work the longest, F(3.32) = 3.21, P < 0.05 (see Table 1). Similarly, rankings of the statement reflecting concern with whether the task could be solved were higher as length of unemployment increased, F(3,32) = 5.29, P < 0.01. While concern for internal causes of performance decreased with duration of unemployment, concern for external causes increased with length of unemployment. Ranking of task difficulty and solvability were generally comparable across feedback conditions for subjects in the control group and the first two unemployed groups. Again, however, subjects unemployed for more than two months reported different rankings in contingent and noncontingent conditions. When provided with contingent feedback, those unemployed the longest ranked task difficulty comparably to the other groups, but when exposed to noncontingent feedback, they ranked task difficulty more highly than any other subjects, F(3,32) = 2.98, P < 0.05 (see Table 2). The same pattern held for
Table
2.
Mean
ranks
assigned
to
possible
causes
nonconringenl Employed control Feedback
NC
2.33
1.80
EiTO‘orl
2.90
2.13
Task
3.30
3.43
4.10
4.03
difficultv
Solvability
’
DISCUSSIOS
This study has presented evidence of stress eiTects of unemployment. Subjects who had been unemployed for longer periods of time showed up at our laboratory with higher levels of catecholamines in their urine. This ‘arousal’ of the sympathetic adrenal medullary system is a major indicator of stress responding [22]. Also. behavioral performance on the EFT was impaired for these subjects. Subjects who had been unemployed for longer periods of time persisted less on the task and also solved fewer puzzles than did employed. control subjects. Studies of stress have shown behavioral performance deficits (e.g. concentration problems, reduced motivation) for stressed individuals [36.41], and subjects who were unemployed in the present study exhibited this pattern of reduced behavioral performance. The tindings present indications of stress responding for the unemployed subjects. Unemployment
stress ad
helplessness
One possible source of this stress is the noncontingency experienced between efforts made toward finding a job and the outcome of such attempts. Assuming that subjects were looking for work and were not successful, job loss and unemployment can be viewed as a situation involving loss of control. Once a person loses his/her job, control over many aspects of life is lost (e.g. income, self-esteem). Brehm [l] posited that one’s response to loss of control is likely to be one of reactance. Reactant individuals increase their efforts at re-establishing control over that aspect of their life which has become noncontingent. People who have recently lost control over an aspect of their life are likely to make vigorous and repeated attempts at regaining that control. Seligman [2] has presented a theory of learned helplessness which also predicts responding in noncontingent situations. His findings argue that exposure to noncontingency results in the cessation of response in the noncontingent situation. Typically, dogs were trained in active avoidance chambers where jumping a barrier would result in cessation of
performance
conditions
(I = high,
on
the
Levine
task
under
contingenr
5 = low) IJnemplo>sd
Unemployed 3-8 wk
Unemployed 13 wk
C
Ability
of
er al.
rankings of solvability, though the interaction term for this analysis only approached significance. F(3.32) = 2.67, P < 0.065 (see Table 2). Rankings of these causes of performance after completing the EFT also revealed effects of feedback and length of unemployment. Ability was again ranked lower by those unemployed the longest, F(3,32) = 4.63, P < 0.01, and solvability was ranked more highly by these subjects than by any others, f(3.32) = 3.66, P < 0.01.
on
5 = low)
2.0 dificulty
Solvability
(I = high.
<3
conlrol ;\bility
to possible
task
Baw
>8
wk
C
NC
C
NC
C
SC
2.33
2.17
2.87
2.30
1.73
4.03
2.43
1.80
2.50
2.80
2.50
3.60
3.23
3.50
3.30
2.97
3 23
I.57
3.43
3.83
2.70
3.13
3.53
’-.- ‘0
and
Unemployment
electric shock. Dogs in this paradigm learn rather easily how to terminate shock (and also learn to prevent shock through the use of signaled avoidance). Seligman included a yoked control group of dogs, however. That is. these voked controls received the same intensity and duration of shocks as the experimental dogs, but their receipt of shock was contingent upon the behaviors of the experimental dogs and not their own. When these dogs were placed m the avoidance chambers, they did not learn to jump the barrier in order to avoid (or terminate) the shocks. Instead. the dogs remained still for the duration of the shock. Seligman explained that the dogs that had been yoked to actively responding dogs had lenrned that response to shock was ineffective. Therefore, when placed in a situation that actually had a contingent response (jumping a barrier) to shock avoidance. these dogs failed to learn the avoidance behavior. They had presumably already learned (from their experiences with being harnessed as yoked controls) that shock was not controlled by their behavior. Thus. they made no attempts at escape behaviors. This leaves two opposing theories concerning response to noncontingency. Brehm [I] predicts increased responding (or reactance) and Seligman [2] predicts reduction in response to noncontingent situations (learned helplessness). Wortman and Brehm [38] presented a formulation which weds these two notions. such that reactance occurs early-after exposure to noncontingency-and learned helplessness follows as length of exposure to noncontingent settings increases. The present study provides evidence for this notion. First, looking at the changes in catecholamine levels from before to after the study, contingent and noncontingent feedback produced different patterns of results. Under contingent conditions, all groups showed decreases in catecholamine levels. Noncontingent conditions showed a varied response, however, depending on length of unemployment. Employed controls and recently unemployed subjects responded to the noncontingent feedback with reactance-like increases in catecholamines, while those subjects who had been unemployed for >3 weeks exhibited decreases in catecholamine levels when exposed to the unsolvable task. This same pattern of reactance for the recently unemployed subjects was also revealed on the EFT. Subjects who were recently unemployed persisted as long as employed subjects, both when given contingent feedback on the Levine task and also when given noncontingent feedback on this task. The subjects who were unemployed the longest, however, showed a decrease in persistence on the EFT when they had been given noncontingent feedback. This is typical of the phenomenon of learned helplessness. Their responding dropped off to a point where, by the ninth and tenth puzzles, they were neither solving the puzzles nor spending any serious effort on the task (persistence times were 20 seconds or less per puzzle). Another interesting finding in this study involved differences in subjects’ perceptions of what was important for their performance on the Levine task. All groups. when given veridical feedback on the Levine, reported thinkmg that the infernal statements on the questionnaires filled out between trials on the Levine
and loss of control
515
task were most important. That is, when presented with a contingent task, subjects focused on their ability and effort as being important to their performance on the Levine task. This can be taken to reflect a focus on internal determinants of cause. When given noncontingent feedback, however, subjects unemployed for >2 months focused more on the solvability and difficulty of the task. This reflects a focus on e.rfernal determinants of cause. In this instance, it was the subjects who were unemployed the longest who were most accurate in their assessment of the Levine task. They listed solvability of the task as most important to them. The important feature of this finding lies in the difference in the unemployed groups’ rankings of internal and external statements. Under contingent feedback, the longer-term unemployed subjects rated internal statements concerning their ability and effort as most important to doing well. This is consistent with the internal focus of the rest of the groups, including the employed control group. But, under conditions of noncontingent feedback, it was only the long-term unemployed group that recognized task solvability as being important. The employed control group and the shorter-term unemployed subjects who were exposed to the same noncontingent feedback failed to recognize its importance to their performance on the task. Instead, these subjects remained more concerned with internal statements concerning the importance of their effort and ability in helping them to solve the task. It is possible that the longer-term unemployed subjects have been exposed to noncontingency in their efforts at finding work and coping with related problems, and this exposure may have led to their heightened sensitivity to the noncontingency in the Levine task. This possibility, that exposure to noncontingency might lead to more ready recognition of noncontingency in the environment, could be viewed as adaptive. However, Seligman’s [2] work has shown that learning noncontingency can have maladaptive consequences as well. In his experiments, dogs were exposed to noncontingent conditions and were then put into situations where shocks were contingent on their actions. The learning of noncontingency had generalized for the dogs such that they became helpless in situations where they need not have been. In the present study, the subjects who had been unemployed the longest and who received noncontingent feedback were more accurate in recognizing noncontingency in the Levine task. This behavtor can hardly be looked upon as maladaptive. However, these same subjects persisted the least on the EFT. Since the EFT was solvable, such behavior may be considered maladaptive. It is possible that exposure to noncontingency heightens one’s awareness of cues in the environment which are likely to covary with noncontingency [42,43]. However, inasmuch as these cues may not vary directly with noncontingency, some environments may be diagnosed as noncontingent when they are actually controllable. This may have happened to the subjects who were unemployed for a longer period of time and who also received noncontingent feedback. These subjects may have approached the EFT and decided that it, too, was uncontrollable; when such an
516
ASDREW
appraisal was made, they stopped trying to find solutions to the puzzles. It is possible that some underlying variable which has not been specified has affected both the performance of the unemployed subjects on our tests and the fact that they are unemployed. However, we view this as unlikely since all groups were similar on a variety of demographic variables. The data suggest, rather, that the experience of unemployment is stressful and presents the unemployed person with loss of control over several aspects of his/her life (e.g. income, self-esteem, family interactions). The data further suggest that the lack of contingency between responses and outcomes is instrumental in the development of the deficits associated with unemployment. The extent to which these patterns may persist or change with even longer-term unemployment is not known and awaits further study.
BAC~I et al.
REFERENCES 1.
‘1.
22.
23.
21. 25.
26.
27.
Brehm J. W. A Theory of Psychological Reacrance. Academic Press, New York, 1966.
2. Seligman M. E. P. Learned Helplessness: On Depression, Decelopmenl and Death. Freeman, San Francisco, 1975. 3. Kasl S., Cobb S. et al. Changes in serum uric acid and cholesterol levels in men undergoing job loss. J. Am. med. Ass. 206, 1500, 1968. 4. Kasl S. and Cobb S. Blood pressure changes in men undergoing job loss: preliminary report. Psychosom. Med. 32, 19-38, 1970. 5. Liem R. and Rayman P. Health and social costs of unemployment. Am. Psychol. 37, I 1161123, 1982. 6. Hall 0. M. Attitudes of unemployed and employed engineers. J. Personality 12, 222-228, 1933. 7. Kardiner A. The role of economic security in the adaptation of the individual. The Family 17, 187-197, 1936. 8. Eisenberg P. and Lazarsfeld P. The psychological effects of unemployment. Psychol. Bull. 35, 358-390, 1938. 9. Little C. B. Technical-professional unemployment: middle-class adautabilitv to personal crisis. Social. 0. 17, 262-274, 1976. 10. Goodchilds J. D. and Smith E. E. The effects of unemployment as mediated by social status. Sociometry 26, 287-295, 1977. II. Cobb S. Physiological changes in men whose jobs were abolished. J. Psychosom. Res. 18, 245-258, 1974. 12. Harris R. D. Unemployment and its effects on the teenager. Aust. Family Physn 9, 546-553, 1980. 13. Warr P. Reported behaviour changes after job loss. Br. J. sot. Psychol. 23, 271-275. 1984. 14. Warr P. Job loss, unemployment, and psychological well-being. In Role Transirions (Edited by Allen V. and Ean der Vliert E.). Plenum Press, New York, 1984. 15. O’Brien G. E. Adjustment of the unemployed. Working Paper Series, No. 29, 1978. 16. Ddoley D. and Catalan0 R. Economic change as a cause of behavioral disorder. Psvchol. Bull. 87,45M68. 1980). 17. Oliver J. M. and Pomicter C. Depression of automobile assembly-line workers as a function of unemployment variables. Am. J. Community Psychol. 9, 507-5 12, 198 I. 18. Kasl S. Changes in mental status associated with job loss and retirement. In Stress and Mental Disorder (Edited by Barrett J. E.). Raven Press, New York, 1979. 19. Kasl S. and Cobb S. Some mental health consequences of plant closing and job loss. In Mental Health and (he Economy (Edited by Freeman L. and Gordus J.). Upjohn, Kalamazoo, Mich., 1979. 20. Baum A. and Singer J. E. Psychosocial aspects of
28.
29.
30.
31. 32.
33. 34. 35. 36.
37.
38.
39. 40. 41.
42.
43.
health. stress, and illness. In Cognirrre Social Ps.wholog? (Edited by Hastorf .\. H. and Isen A. M.). Elsevier, North-Holland, New York, 1982. Gore S. The effect oi social support in moderating the health consequences of unemployment. J. Hlrh sot. Behac. 19, 157-164. 1978. Frankenhaeuser M. Experimental approaches to the study of catecholamines and emotions. In Emotions: Their Parameters and .Veasuremem (Edited by Levi L.). Raven Press, New York. 1975. Henry J. P. and Cassel J. C. Psychosocial factors in essential hypertension: recent epidemiologic and animal experimental et-idcnce. Am. J. Epid. 90, 171. 1969. Brenner M. H. .\f