You are on page 1of 11

J O U R N A L OF

Personality
Social Psychology
Volume 37 January 1979 Number 1

Stressful Life Events, Personality, and Health:


An Inquiry Into Hardiness
Suzanne C. Kobasa
University of Chicago

Personality was studied as a conditioner of the effects of stressful life events on


illness onset. Two groups of middle and upper level executives had comparably
high degrees of stressful life events in the previous 3 years, as measured by the
Holmes and Rahe Schedule of Recent Life Events. One group (n = 86) suffered
high stress without falling ill, whereas the other (n = 75) reported becoming
sick after their encounter with stressful life events. Illness was measured by the
Wyler, Masuda, and Holmes Seriousness of Illness Survey. Discriminant function
analysis, run on half of the subjects in each group and cross-validated on the
remaining cases, supported the prediction that high stress/low illness executives
show, by comparison with high stress/high illness executives, more hardiness,
that is, have a stronger commitment to self, an attitude of vigorousness toward
the environment, a sense of meaningfulness, and an internal locus of control.

An exceptional number of studies in the last gested that stressful life events precipitate
20 years (cf. Dohrenwend & Dohrenwend, somatic and psychological disease. This article
1974; Gunderson & Rahe, 1974) have sug- considers the importance of personality as a
conditioner of the illness-provoking effects of
stress.
This article is based on the author's doctoral dis- During the last decade, investigators have
sertation (Kobasa, 1977), submitted to the Depart-
ment of Behavioral Sciences at the University of shown that the recent life histories of hos-
Chicago. The preparation of this manuscript was sup- pitalized persons contain significantly more
ported in part by Public Health Service Grant MH- frequent and serious stressful events than do
28839-01 from the National Institute of Mental histories of matched controls from the general
Health. The author wishes to thank Robert R. J.
Hilker, James Kennedy, and all of the executives population (e.g., Paykel, 1974) and that
who participated in the study. Special appreciation Navy personnel who begin a cruise with high
is extended to Salvatore R. Maddi who supervised stress scores suffer more illness episodes dur-
the project. Chase P. Kimball, David E. Orlinsky, ing the months at sea than do sailors who
and Marvin Zonis contributed many useful sugges-
tions as dissertation committee members. start out with low stress scores (Rahe, 1974).
Requests for reprints should be sent to Suzanne But the possibility of a causal connection be-
C. Kobasa, Department of Behavioral Sciences, Uni-
versity of Chicago, 5848 S. University Avenue, Chi- tween stress and illness is hardly a new idea.
cago, Illinois 60637. Physicians, philosophers, and persons simply

Copyright 1979 by the American Psychological Association, Inc. 0022-3514/79/J701-0001S00.75

1
SUZANNE C. KOBASA

concerned about staying healthy have long This warning is mitigated, however, by a sense
wondered about the etiological significance of of optimism that there is something to this
life events. The distinctiveness of recent re- stress and illness connection and a wish for
search lies in its attempt to define and mea- studies that are more sophisticated in design
sure stress. and methods of analysis. But in the popular
In the current studies, a life event is de- literature (e.g., Wolfe, 1972), final conclusions
nned as stressful if it causes changes in, and have already been drawn. Readers are pro-
demands readjustment of, an average person's vided with a self-administered stress test and
normal routine. This definition of stress has told the likelihood of illness associated with
relied upon the empirical demonstration each total stress score (e.g., a score above
(Holmes & Masuda, 1974; Holmes & Rahe, 300 means an 80% chance of getting sick).
1967) that there is a general consensus about Readers whose stress scores are high are
the degree to which specific life occurrences warned that if they wish to remain healthy
involve change and require readjustment. This they should avoid additional stressful encoun-
consensus emerged from the ratings of stress- ters, even to the point of shunning such every-
fulness of a long list of events made by thou- day necessities as driving on the Los Angeles
sands of subjects varying in age, sex, socio- Freeway.
economic status, race, cultural background, This advice seems inadequate, insofar as
education, and religion. From this consensus, modern life can be characterized as inherently
stressfulness weights for the life events were stressful (Brodsky, 1977; Toffler, 1970). A
obtained. person might indeed be able to keep from get-
Deriving from this research is the Schedule ting married or taking on a new mortgage,
of Recent Life Events (Holmes & Rahe, but how can one prevent the effects of pollu-
1967), which contains positive (e.g., mar- tion and overcrowding or of being assigned
riage), negative (e.g., illness of a family mem- to a different job when his or her company
ber), frequent (e.g., minor traffic violation), signs a new contract? Further, increasing
and rare (e.g., death of a child) occurrences. levels of stress tend to coincide with increas-
The associated Social Readjustment Rating ing opportunities and potential resources
Scale (Rahe, Lundberg, Theorell, & Bennett, (Kobasa, Hilker, & Maddi, Note 1). By
1971) gives the consensual weights for these avoiding stress, modern persons may be turn-
events (e.g., on a scale from 1 to 100, divorce ing away from a chance to better their lives.
gets a mean score of 73 and vacation, a score Before deciding to avoid stressful life at
of 13). The stressfulness of a particular per- any cost, one should consider the actual mag-
son's life is measured in this procedure by nitude of the relationship between stress and
finding out what events he or she has en- illness. Although correlations range from .20
countered and scoring the occurrences accord- to .78, the majority fall below .30, and in
ing to their consensually quantified weights. Rahe's naval data, the correlations are con-
This consensual approach to stress as mea- sistently around .12 (Rabkin & Struening,
surable environmental input has allowed easy 1976). Many stress and illness researchers, in
collection of large amounts of data and un- their preoccupation with group means, have
complicated analysis of the statistical rela- failed to question the distribution of their
tionship between stressful life events and data. Variability of both stress and illness
illness onset. The findings of stress research scores within groups has been observed to be
have had dramatic impact in both professional extreme in several studies (Wershow & Rein-
journals and popular magazines. In the jour- hart, 1974). One likely explanation for these
nals (e.g., Rabkin & Struening, 1976), one data is the presence of subjects with high
usually finds some caution urged about too stress scores who are not getting sick. Oddly,
quickly concluding a causal relationship be- such subjects have been overlooked in the
tween stress and illness on the basis of cor- popular and professional literature on stress
relational and methodologically weak studies. and illness.
STRESSFUL LIFE EVENTS

Studying Stressed but Healthy Persons ness of authentic living, White (1959) on
competence, Allport (1955) on propriate
In contrast, the present study considers striving, and Fromm (1947) on the produc-
how highly stressed subjects who remain tive orientation. Hardy persons are considered
healthy differ from those who show illness to possess three general characteristics: (a)
along with high stress. Studying the individual the belief that they can control or influence
who undergoes high degrees of stress without the events of their experience, (b) an ability
falling ill amounts to inquiring about the to feel deeply involved in or committed to the
mediating factors that affect the way one re- activities of their lives, and (c) the anticipa-
acts to stress. Holmes and Masuda (1974) tion of change as an exciting challenge to
and the majority of other stress investigators further development. Much research has al-
attempt to draw a direct causal link between ready shown the advantages in behavior of
the occurrence of stressful life events and the control (e.g., Lefcourt, 1973; Rodin & Langer,
onset of illness by reference to the physio- 1977; Rotter, Seeman, & Liverant, 1962;
logical model of a stress reaction formulated Seligman, 1975), commitment (e.g., Antonov-
by Hans Selye (1956). Stressful life events sky, 1974; Kobasa & Maddi, 1977; Lazarus,
are said to evoke "adaptive efforts by the hu- 1966; Lazarus, Averill, & Opton, 1974; Moss,
man organism that are faulty in kind or dura- 1973), and challenge (Fiske & Maddi, 1961;
tion, lower 'bodily resistance' and enhance the Maddi, 1967). In discussing the hypotheses
probability of disease occurrence" (Holmes presented below, the implications of theory
& Masuda, 1974, p. 68). Holmes and others, and research concerning these three general
however, fail to take into account what Selye characteristics are extended to considerations
goes on to say about individual differences of health and illness.
and the stress reaction. In the study described Hypothesis 1. Among persons under stress,
in this article, the more subtle points in those who have a greater sense of control
Selye's work are referred to, in the attempt over what occurs in their lives will remain
to consider factors in the stress reaction that healthier than those who jeel powerless in the
serve to deflect the negative impact of stress- face of external forces. Following the model
ful events. Mediators of the stress and illness proposed by Averill (1973) to explain his
connection, which probably include physio- laboratory observation that some organisms
logical predisposition, early childhood experi- are not debilitated by stressful stimuli, the
ences, and social resources, as well as the highly stressed but healthy person is hypothe-
mediator emphasized here, personality, are to- sized to have (a) decisional control, or the
gether responsible for what Selye calls the capability of autonomously choosing among
distinctive way in which each individual various courses of action to handle the stress;
"takes to" stressful life occurrences. (b) cognitive control, or the ability to inter-
The proposition of this study is that per- pret, appraise, and incorporate various sorts
sons who experience high degrees of stress of stressful events into an ongoing life plan
without falling ill have a personality structure and, thereby, deactivate their jarring effects;
differentiating them from persons who become and (c) coping skill, or a greater repertory
sick under stress. This personality difference of suitable responses to stress developed
is best characterized by the term hardiness. through a characteristic motivation to achieve
The conceptual source of the supposition, in across all situations. In contrast, the highly
contrast to the passive and reactive view of stressed persons who become ill are powerless,
humankind found in most stress and illness nihilistic, and low in motivation for achieve-
work, is a set of approaches to human be-
ment. When stress occurs, they are without
havior that Maddi (1976), in his categoriza-
tion of the major personality theories, calls recourse for its resolution, give up what little
fulfillment theories. The hardy personality control they do possess, and succumb to the
type formulated here builds upon the theoriz- incapacity of illness.
ing of existential psychologists (Kobasa & Hypothesis 2. Among persons under stress,
Maddi, 1977; Maddi, 197S) on the strenuous- those who feel committed to the various areas
SUZANNE C. KOBASA

of their lives will remain healthier than those be clear that the highly stressed but healthy
who are alienated. Committed persons have a individual is not engaging in irresponsible
belief system that minimizes the perceived adventurousness. At the core of the search
threat of any given stressful life event. The for novelty and challenge are fundamental
encounter with a stressful environment is miti- life goals that have become, in adulthood,
gated by a sense of purpose that prevents increasingly integrated in a widening diver-
giving up on one's social context and oneself sity of situations (Henry, 1968; Neugarten,
in times of great pressure. Committed persons 1974).
feel an involvement with others that serves Although personality is the primary con-
as a generalized resistance resource against cern of the study reported here, other kinds
the impact of stress (Antonovsky, 1974). of variables may well differentiate the highly
Committed persons have both a reason to and stressed and healthy from those who have
an ability to turn to others for assistance in fallen ill under stress. These would include a
times demanding readjustment. variety of psychological, social, physiological,
Although commitment to all areas of life— and environmental mediators. In order to in-
work, social institutions, interpersonal rela- crease understanding of those who fall ill and
tionships, family, and self—should be charac- stay healthy under high stress, the present
teristic of highly stressed persons who do not study included information concerning various
fall ill, one area is singled out as particularly demographic characteristics and perceptions
important for health. Staying healthy under of stressfulness of events.
stress is critically dependent upon a strong
sense of commitment to self. An ability to Method
recognize one's distinctive values, goals, and
priorities and an appreciation of one's capac- Overview
ity to have purpose and to make decisions The first task was the identification of high stress/
support the internal balance and structure low illness and high stress/high illness groups. For
that White and other theorists (cf. Coelho, this purpose, all members (# = 837) of a large sub-
Hamburg, & Adams, 1974) deem essential ject pool were mailed a stress and illness question-
naire. They were asked to report, by month and
for the accurate assessment of the threat year, which of a list of stressful life events and ill-
posed by a particular life situation and for ness episodes they had experienced in the previous
the competent handling of it. 3 years. A total stress score and a total illness score
Hypothesis 3. Among persons under stress, were obtained for each subject by the standard pro-
cedure of noting the items checked and multiplying
those who view change as a challenge will these by their frequencies and consensually denned
remain healthier than those who view it as a weights. Subjects were then assigned to one of the
threat. Persons who feel positively about two groups or put aside. Assigned to the high stress/
change are catalysts in their environment and low illness and high stress/high illness groups were
126 subjects above the median for total stress and
are well practiced at responding to the unex- below the median for total illness and 200 subjects
pected. Because they value a life filled with above the median for total stress and above the
interesting experiences, change seekers have median for total illness, respectively. Discarded sub-
well explored their environment and know jects included 322 low-stress cases and 22, or all,
of the female subjects. In further group refinement,
where to turn for resources to aid them in 40 high stress/high illness cases whose peak illness
coping with stress. They have a predisposi- score preceded rather than followed their peak stress
tion to be cognitively flexible, which allows score and 10 borderline cases from each of the two
them to integrate and effectively appraise the groups were put aside. Finally, because of a request
from the company for which all of the subjects
threat of new situations. Their basic motiva- worked that the sample to be administered personal-
tion for endurance allows them to persist even ity questionnaires be kept at a workable minimum,
when the new information is exceedingly in- a set of 100 subjects was randomly selected from the
116 remaining high stress/low illness cases and 100
congruous and, thereby, maximally provoking randomly selected from the 150 remaining high
of strain and illness (Moss, 1973). stress/high illness subjects.
Within 3 months of the stress and illness testing,
Given the characterization of hardiness the two groups were mailed another questionnaire.
contained in the first two hypotheses, it should This questionnaire contained several personality tests,
STRESSFUL LIFE EVENTS

along with questions about demographics and per- Each of the most ambiguous events was replaced
ception of life stressfulness. by two events, one presenting the positive form of
The data from approximately half of each group the Holmes and Rahe item and the other, the nega-
were used to test hypotheses concerning group dif- tive version. "Change in financial state," for example,
ferences in hardiness. Data from the remaining cases was translated into "improvements in financial state"
in each group were used to cross-validate the results. and "worsening of financial state." These specifica-
The division of the groups into test and cross- tions were given the seriousness weights of the items
validation cases was necessitated by the statistical from which they were derived. Other additions to the
technique relied upon in the study, discriminant Holmes and Rahe list were based on a pilot use of
function analysis, which has been characterized in the test with SO randomly selected executives. In
previous research by problems of generalizability response to the question "What other events have
due to instability of results (cf. Huberty, 1975). Al- you experienced during the past 3 years?", these
though powerful as a tool for looking at group dif- subjects reported 15 events not found on the original
ferences, discriminant function analysis, like regres list. Most of these referred to occurrences at work.
sion analysis, has not always provided results that Seriousness weights were assigned to these additions
hold up in the making of inferences from sample re- by the investigator and 20 other judges using the
sults to some population, and over repeated sam- ratio scale judgment procedure of Holmes and Rahe
plings. Until a replication of the study is possible, (1967).
the use of a "holdout sample" must be relied upon The illness items in the stress and illness question-
for an accurate test of the adequacy of the derived naire were taken from the Wyler, Masuda, and
discriminant function that defines group differences. Holmes (1968) Seriousness of Illness Survey. After
consultation with the medical director of the execu-
tives' company, 118 of the diseases listed in this
Subjects survey were chosen as applicable to the pool being
tested. Each illness item is characterized by a seri-
This study required a subject pool both large and ousness weight based on a consensual agreement of
stressed enough to obtain sufficiently large groups numerous and diverse judges (both medical and
for study. All of the middle and upper level execu- lay) and is obtained in a manner similar to the
tives of a large public utility served as the pool derivation of the Social Readjustment Rating Scale.
from which groups were selected. This utility, in the The reliability and validity of this scale as a com-
year prior to the onset of the study, had entertained plete listing of disease syndromes in a form accessible
serious discussion (in executive seminars, company to both laypersons and physicians, and as an accu-
publications, and consultations with the medical de- rate set of evaluations of the general seriousness of
partment) of the increasing numbers of stressful various distinct illnesses, has been established (cf.
events faced by executives. These events consisted Wyler, Masuda, & Holmes, 1970).
of changes instituted by the utility itself, like a pro-
gram of job evaluations that led to some promotions
and many more demotions, as well as requirements Measurement of Personality, Demographic,
for readjustment from external sources, such as the
federal government's affirmative action demands. and Perception Variables
These changes, coupled with the expected usual range
of personal and family stresses, suggested that stress A composite questionnaire, made up of all or parts
scores for the subject pool would be generally high. of four standardized and two newly constructed
Demographically, the pool was quite homogeneous. instruments, was designed to test the three personal-
The modal characteristics of the subjects were (a) ity hypotheses. The standardized tests were chosen
male gender; (b) 40 to 49 years of age; (c) married, for their theoretical relevance and empirical relia-
with two children; (d) on the third or middle man- bility and validity. All of the instruments are ap-
agement level, and having been there for 6 years propriate for use with a sample of executives (i.e.,
or more; (e) possessing at least a college degree; a group of well-educated adult professionals who
(f) wife not working outside the home; (g) usually are relatively free of gross psychopathology).
Protestant, and attending religious services very or The control dimension was measured through four
fairly often. different instruments. What has been called deci-
sional control or autonomy was measured through
the Internal-External Locus of Control Scale (Lef-
Measurement oj Stress and Illness court, 1973; Rotter, Seeman, & Liverant, 1962), and
the Powerlessness versus Personal Control scale of
The most frequently used scales in stress and ill- the Alienation Test (Maddi, Kobasa, & Hoover,
ness research, the Schedule of Recent Life Events Note 2). The latter instrument also provided a
and the Social Readjustment Rating Scale (Holmes way of measuring cognitive control (i.e., the abil-
& Rahe, 1967), were employed in this study. Addi- ity to find meaning in stressful life events) in its
tions to these scales were made on the basis of Nihilism versus Meaningfulness scale. Coping skill
pilot testing. The majority of additions were more (i.e., the availability of responses with which to deal
detailed specifications of the original items, modeled with stressful life events) was measured through the
after suggestions from other adapters of the test Achievement scale of the Personality Research Form
(Hough, Fairbank, & Garcia, 1976; Paykel, 1974). (Jackson, 1974; Wiggins, 1973).
6 SUZANNE C. KOBASA

The question of whether these forms of control ble in the literature (Holmes & Masuda,
are accompanied by a need to dominate the behavior 1974; Wyler et al., 1968, 1970) indicates that
of others was addressed through administration of
the Leadership Orientation scale of the California on the average, they had experienced a mod-
Life Goals Evaluation Schedules (Hahn, 1966) and erate amount of illness (comparable to the
the Dominance scale of the Personality Research threat of life and discomfort associated with
Form (Jackson, 1974).
The commitment dimension was measured by the
having a peptic ulcer or high blood pressure)
Alienation Test (Maddi et al., Note 2). Of relevance in the previous 3 years. During this period,
here are the alienation versus commitment scores they had, on the average, also encountered
this test yields in five areas of functioning (work, sufficient change and demands for readjust-
social institutions, interpersonal relationships, family, ment in their lives to constitute major life
and self). The degree of consistency across a sub-
ject's different forms of involvement was also tested. crisis (what Holmes has associated with an
This was done through the Role Consistency Test, 80% likelihood of getting seriously ill in the
adapted from the Gergen and Morse (1967) Self- near future).
Consistency Test to measure the compatibility among
the subject's reported five most important life roles. A Pearson product-moment correlation of
Measurement of the orientation to challenge re- .24 (p < .025) was obtained between the total
quired the administration of several tests. The de- stress and total illness scores. Canonical cor-
gree to which the executive sought out rather than relations were also run to determine whether
avoided stimulation from his environment was mea-
sured through the Preference for Interesting Experi- a stronger relationship might be obtained.
ences scale of the Hahn (1966) test and the Vegeta- Stress scores—both the standard weighted
tiveness versus Vigorousness scale of the Alienation scores and simple frequencies—for various
Test. The executive's ability to seek challenge even areas of life (work, home, community, etc.)
in the face of potential psychological, social, and
biological threat was indexed by Hahn's scale of were used as the predictor, and illness scores
Security Orientation. Cognitive flexibility versus cog- —both weighted and frequencies—for vari-
nitive rigidity was measured through the Need for ous body systems (cardiovascular, respira-
Cognitive Structure scale of the Personality Re- tory, etc.) were used as the criterion set.
search Form and the ability to persist even in the
most challenging environment, through the Need for Neither analysis of unweighted stress events
Endurance scale of the same instrument. Finally, and illnesses nor correlations of selected types
the degree to which an orientation toward challenge of stress with selected types of illness, how-
was characterized by a general sense of responsibil- ever, significantly strengthened the statistical
ity toward life's demands was measured through the
Adventurousness versus Responsibility scale of the relationship between stress and illness. The
Alienation Test. weak, but significant, correlation of .24 is
Subjects were asked about three demographic char- consistent with most of the available research
acteristics: age, job level (third, fourth, fifth, and reports (Rabkin & Struening, 1976).
officer), and number of years spent at current level.
Executives were also asked to rate on a scale of 1
(not at all stressful) to 7 (extremely stressful) how
stressful they usually think each of the following Testing oj Group Differences
areas of life is: work, financial concerns, social/com-
munity involvements, interpersonal relationships, Completed personality questionnaires were
family, and personal or inner-life concerns. returned by 86% of trie high stress/low ill-
ness and 75% of the high stress/high illness
Results subjects. Forty "test" subjects were randomly
selected from each group for analysis of dif-
Stress and Illness Scores ferences across personality, demographic, and
Of the 837 executives who were contacted perception variables, and 81 subjects were
by mail, 670 (80%) returned completed stress put aside for cross-validation purposes.
and illness questionnaires. The mean stress Mean differences between the test subjects
score was 399, with a standard deviation of of the two groups were evaluated by t test.
162, a range from 0 to 2,239, and a median None of the demographic variables and only
of 306. The mean illness score was 913, with one of the perception variables yielded sig-
a standard deviation of 1,115, a range from nificant differences, whereas several of the
0 to 6,900, and a median of 550. Comparison personality variables did so. Further evalua-
of these executives' scores with norms availa- tion of group differences was achieved by a
STRESSFUL LIFE EVENTS

Table 1
Differences Between High Stress/Low Illness* and High Stress/High Illness* Executives

High stress/ High stress/ Standardized


low illness high illness discriminant
f
* function
Variable M SD M SD value coefficient

Control
Nihilism 196.05 133.61 281.02 169.86 2.49** .73
External locus of control 5.92 4.10 7.90 4.61 2.03* .22
Powerlessness 301.15 188.93 388.47 188.44 2.11* —
Achievement 16.50 2.10 15.12 3.20 -1.20 —
Dominance 14.60 3.26 13.85 4.46 .86 —
Leadership 33.47 7.34 34.63 6.80 .73 .43
Commitment
Alienation from self 102.35 117.24 219.15 185.77 3.36** 1.04
Alienation from work 181.67 122.04 223.73 175.09 1.22 .43
Alienation from interpersonal 256.02 162.76 316.10 165.24 1.64
Alienation from family 158.47 139.02 198.72 144.33 1.27 —

Alienation from social 202.15 100.21 226.95 133.93 .94 —
Role consistency 29.22 6.42 29.50 6.44 .19 .30
Challenge
Vegetativeness 155.50 140.24 216.27 160.94 1.98* .99
Security 21.11 6.33 22.19 8.60 .34 .35
Cognitive structure 13.35 2.81 14.10 2.85 1.10 .21
Adventurousness 269.00 164.58 337.54 174.95 1.78* —
Endurance 15.97 2.35 14.37 3.19 -.96 —.
Interesting experiences 34.97 6.83 32.52 7.02 -.92
Perception of personal stress 3.00 1.21 3.83 1.73 2.46** .43

Note. For all variables, the higher the number, the greater the degree of the variable observed. Superior
hardiness is indicated by higher scores on achievement, role consistency, endurance, and interesting experi-
ences, and lower scores on nihilism, external locus, powerlessness, dominance, leadership, alienation (from
self, work, social institutions, interpersonal relationships, and family), vegetativeness, security, cognitive
structure, and adventurousness. A subject's scores on all areas of alienation, measured by the Alienation
Test, have a possible range of 0 to 1,200. Vegetativeness, nihilism, powerlessness, and adventurousness
scores, also from the Alienation Test, may range from 0 to 1,500. External locus has a low of 0 and an upper
limit of 23. The scales taken from the Jackson test—achievement, dominance, cognitive structure, and
endurance—have a minimum value of 0 and a maximum of 20. The California Life Goals scale—leadership,
security, and interesting experiences—may range from 0 to 60. Role consistency has a low of 0 and a high
of 40; perception of personal stress can range from 0 to 7.
a
« = 40.
* p < .05.
** p < .01.
discriminant function analysis done on all the discriminant function analysis and their sum-
personality variables plus the one perception mary statistics, including significance of dif-
variable that yielded a significant t score. ference between groups established by t test.
After all data are transformed into standard The results demonstrate that high stress/low
scores, discriminant function analysis com- illness executives can be distinguished from
putes a discriminant equation, or a linear high stress/high illness subjects. The 11 vari-
combination of weighted variables that pro- ables for which a standardized discriminant
duces the greatest statistically derivable dis- function coefficient is provided in the table
tance between the two groups. The larger the combine to form a significant function, with
weighting or discriminant coefficient of a a Wilks's Lambda of .64, significant at the
variable, the more powerful it is as a group .001 level, and a canonical correlation of .60.
discriminator. The Wilks's Lambda (see Huberty, 1975) is
Table 1 presents all variables submitted to a measure of the original variable's discrimi-
SUZANNE C. KOBASA

nant power, before it is removed by the dis- meaningfulness (as opposed to nihilism), and
criminant function. The canonical correlation their internal (as opposed to external) locus
is a measure of the association between the of control. One perception-of-stress variable
single discriminant function and the set of joins the important personality discrimina-
dummy variables that defines the two group tors—high stress/low illness executives find
memberships. Given the composition of this the personal sphere of their lives significantly
function, the general position concerning less stressful then do the high stress/high
hardiness is confirmed. High stress/low ill- illness subjects.
ness executives are, at least in some ways, Several personality variables appear to con-
more in control, more committed, and more tribute to the discriminant equation even
oriented to challenge than are high stress/ though they do not yield significant mean dif-
high illness executives. Looking at the varia- ferences by t test. Although it is risky to
bles that make the most significant contri- interpret the discriminant coefficients of these
bution to the discriminant equation and that variables, it is worth noting that the direc-
are also responsible for significant mean dif- tion of mean difference is in general consistent
ferences, the high stress/low illness executives with the hypotheses concrening hardiness as
are distinguished by their sense of commit- an insulation against illness.
ment to (or lack of alienation from) self, Along with their analytic function, the dis-
their sense of vigorousness (as opposed to criminant coefficients also serve a classifica-
vegetativeness) about life, their sense of tory and validation purpose. Multiplying a

Table 2
Differences Between All High Stress/Low Illness11 and All High Stress/High Illnessb
Executives

High stress/ High stress/


low illness high illness

Variable M SD M SD value

Control
Nihilism 217.01 141.03 262.23 161.77 1.82*
External locus of control 5.38 3.90 7.51 4.60 2.05*
Powerlessness 314.95 190.65 364.20 189.50 1.82*
Achievement 16.40 2.35 15.95 3.03 -1.02
Dominance 14.35 3.38 14.27 4.19 -.13
Leadership 33.89 6.56 34.53 6.29 .54
Commitment
Alienation from self 113.87 122.66 180.56 150.65 2.97**
Alienation from work 180.61 129.22 201.68 159.69 .89
Alienation from interpersonal 276.53 159.95 309.64 151.54 1.30
Alienation from family 168.93 145.83 200.99 144.62 1.35
Alienation from social 205.39 98.37 219.28 123.45 .76
Role consistency 28.80 6.78 29.59 6.84 .71
Challenge
Vegetativeness 160.04 133.34 215.26 153.54 1.82*
Security 21.52 6.83 22.63 7.43 .95
Cognitive structure 13.16 2.77 13.75 2.94 1.26
Adventurousness 276.88 165.58 330.84 158.91 1.78*
Endurance 15.01 2.95 14.53 3.15 -.96
Interesting experiences 34.29 6.31 32.80 7.00 -1.37
Perception of personal stress 2.72 1.31 3.61 1.79 3.48**

• n = 86.
*> n = 75.
* p < .05.
** p < .01.
STRESSFUL LIFE EVENTS

subject's scores on the discriminating varia- experience the changes that the transfer will
bles by the associated coefficients allows one bring about—learning to cope with new sub-
to predict the likelihood of the subject's mem- ordinates and supervisors, finding a new home,
bership in each of the groups. As a test of helping children and wife with a new school
the classificatory power of the derived dis- and neighborhood, learning new job skills, and
criminant equation, its coefficients are (a) so on. The hardy executive will approach the
reapplied to the test subjects used to derive necessary readjustments in his life with (a)
the equation and (b) applied to the scores a clear sense of his values, goals, and capa-
of the "holdout" cases. The discriminant bilities, and a belief in their importance (com-
function presented in Table 1 shows predic- mitment to rather than alienation from self)
tive capability both internally and externally. and (b) a strong tendency toward active in-
Applying the unstandardized versions of the volvement with his environment (vigorousness
discriminant function coefficients to the raw rather than vegetativeness). Hence, the hardy
scores used to derive the function, 78% of executive does more than passively acquiesce
the "test" cases are correctly classified (80% to the job transfer. Rather, he throws himself
of the high stress/low illness executives and actively into the new situation, utilizing his
75% of the high stress/high illness subjects). inner resources to make it his own. Another
This significantly correct (p < .025) classi- important characteristic of the hardy execu-
fication is matched in the external cross-vali- tive is an unshakable sense of meaningfulness
dation. Using the unstandardized coefficients and ability to evaluate the impact of a trans-
on the raw data from the "holdout" subjects, fer in terms of a general life plan with its
35 hits (77%) and 21 hits (60%) in the high established priorities (meaningfulness rather
stress/low illness and high stress/high illness than nihilism). For him, the job transfer
groups, respectively, are realized (p < .05). means a change that can be transformed into
These cross-validation results offer support a potential step in the right direction in his
for the stability and generalizibility of the overarching career plan and also provide his
results obtained through discriminant func- family with a developmentally stimulating
tion analysis. An examination of the statis- change. An internal (rather than external)
tics for the entire sample (test cases plus locus of control allows the hardy executive
holdouts), which are notably similar to those to greet the transfer with the recognition that
of the test cases in Table 1, illustrates the although it may have been initiated in an
strength of the discriminant analysis. Table office above him, the actual course it takes is
2 presents the mean values, standard devia- dependent upon how he handles it. For all
tions, and t values for the full high stress/low these reasons, he is not just a victim of a
illness and high stress/high illness groups. threatening change but an active determinant
of the consequences it brings about. In con-
Discussion trast, the executive low in hardiness will react
to the transfer with less sense of personal re-
This study of persons who do not fall ill source, more acquiescence, more encroach-
despite considerable stress suggests that per- ments of meaninglessness, and a conviction
sonality may have something to do with stay- that the change has been externally deter-
ing healthy. Using the five most significant mined with no possibility of control on his
discriminators of the high stress/low illness part. In this context, it is understandable that
executives (i.e., the variables that contribute the hardy executive will also tend to perceive
to the discriminant equation and produce sig-
the transfer as less personally stressful than
nificant ts) one can speculate on what hap-
pens when the hardy individuals meet a stress- his less hardy counterpart.
ful life event—how they evaluate the threat The mechanism whereby stressful life
posed by the event and cope with it. events produce illness is presumably physio-
A male executive having to deal with a job logical. Whatever this physiological response
transfer will serve as an example. Whether is, the personality characteristics of hardiness
hardy or not, the executive will anticipate and may cut into it, decreasing the likelihood of
10 SUZANNE C. KOBASA

breakdown into illness. Needless to say, de- questionnaire may reflect the psychologically
scription of the actual nature of physiological debilitating effects of illness. A prospective,
mechanisms and their links to personality and longitudinal study in which stress and per-
stressful occurrences will have to wait for sonality scores at Time 1 are used to predict
more sophisticated stress research. Until then, illness scores at Time 2 is obviously necessary.
however, two alternative explanations of the But the present study provides some basis for
present results should be considered. regarding the alternative explanation as less
It could be argued that there is a spurious likely than that offered here. Although not dis-
factor at work in the subjects' completion of cussed in this report, subjects found to be
questionnaires, one that determines what they low in stress and high in illness showed per-
say about personality, stress levels, and ill- sonality scores midway between the high
ness experience. Mechanic (1976) has dis- stress/low illness and high stress/high illness
cussed the possible distortion of stress data groups (Kobasa, 1977). It is of specific rele-
by a variable that he calls illness behavior. vance to the alternative explanation that the
From Mechanic's perspective, one could sug- low stress/high illness subjects were lower on
gest that the high stress/high illness execu- nihilism, alienation from self, vegetativeness,
tives have not really undergone any physio- and external locus of control than were high
logical breakdown but that they simply want stress/high illness subjects. This finding indi-
to act and be treated as if they are sick and cates that personality questionnaire responses
thereby withdraw from a life situation experi- indicative of lack of hardiness are not merely
enced as too stressful. The best evaluation of a response to illness.
this position would come from a study that Continued work in the area of stress is es-
employs physiological rather than self-report sential. There is still much to be learned
measures of illness and monitors physiological about the role of personality and other medi-
states before and after the occurrence of a ators in the connection between stressful life
stressful life event. Until such studies are events and illness or health. Should studies in-
available, reliance must be placed on data al- corporating physiological measures and em-
ready collected. Mechanic's explanation seems ploying prospective designs confirm the pres-
much more likely to apply to the self-report ent findings, then instead of merely warning
of mild, vague symptoms than to report of persons to avoid stressful lives, social scien-
more serious illness. For example, headaches tists will be able to illuminate ways of de-
and indigestion might well be reported by sub- veloping the personality characteristics that
jects engaging in illness behavior. But it seems can aid in a productive and healthy life led
unlikely that definite illnesses requiring medi- in the full complexity of modern, urbanized,
cal diagnosis, such as heart attack, cancer, industrialized societies.
detached retina, and even hypertension would
be erroneously reported. Illnesses of the latter Reference Notes
sort are responsible for high illness scores on 1. Kobasa, S. C., Hilker, R. R. J., & Maddi, S. R.
the self-report instrument used in the present Remaining healthy in the encounter with stress.
study. It is therefore not likely that the re- Paper presented at the meeting of the American
sults obtained can be explained completely Medical Association Congress on Occupational
Health, St. Louis, Missouri, September 1977.
by the concept of illness behavior.
2. Maddi, S. R., Kobasa, S. C., & Hoover, M. The
Another alternative explanation would sug- Alienation Test: A structured measure of a multi-
gest that the observed personality differences dimensional subjective state. Manuscript submitted
between the high stress/low illness and the for publication, 1978.
high stress/high illness executives result from
the latter group having experienced illness. References
Presumably, the high stress/high illness execu- Allport, G. W. Becoming: Basic considerations for a
tives were not more alienated, external in psychology of personality. New Haven, Conn.:
beliefs concerning locus of control, and unin- Yale University Press, 19SS.
Antonovsky, A. Conceptual and methodological prob-
terested in change before the illnesses oc- lems in the study of resistance resources and stress-
curred. Their responses to the personality ful life events. In B. S. Dohrenwend & B. P. Doh-
STRESSFUL LIFE EVENTS 11

renwend (Eds.), Stressful life events: Their nature Maddi, S. R. The existential neurosis. Journal of Ab-
and effects. New York: Wiley, 1974. normal Psychology, 1967, 72, 311-325.
Averill, J. R. Personal control over aversive stimuli Maddi, S. R. The strenuousness of the creative life.
and its relationship to stress. Psychological Bulletin, In I. A. Taylor & J. W. Getzels (Eds.), Perspec-
1973, SO, 286-303. tives in creativity. Chicago: Aldine, 197S.
Brodsky, C. M. Long-term work stress in teachers Maddi, S. R. Personality theories: A comparative
and prison guards. Journal of Occupational Medi- analysis (3rd ed.). Homewood, 111.: Dorsey Press,
cine, 1977,19, 133-138. 1976.
Coelho, G., Hamburg, D. A., & Adams, J. E. (Eds.). Mechanic, D. Stress illness and illness behavior. Jour-
Coping and adaptation. New York: Basic Books, nal of Human Stress, 1976, 2, 2-6.
1974. Moss, G. E. Illness, immunity, and social interaction.
Dohrenwend, B. S., & Dohrenwend, B. P. (Eds.). New York: Wiley, 1973.
Stressful life events: Their nature and effects. New Neugarten, B. L. The middle years. In S. Arieti
York: Wiley, 1974. (Ed.), American handbook of psychiatry. New
Fiske, D. W., & Maddi, S. R. (Eds.). Functions of York: Basic Books, 1974.
varied experience. Homewood, 111.: Dorsey Press, Paykel, E. S. Recent life events and clinical depres-
1961. sion. In E. K. Gunderson & R. H. Rahe (Eds.),
Fromm, E. Man for himself. New York: Holt, Rine- Life stress and illness. Springfield, 111.: Charles C
hart & Winston, 1947. Thomas, 1974.
Gergen, K. J., & Morse, S. J. Self-consistency: Mea- Rabkin, J. G., & Struening, E. L. Life events, stress,
surement & validation. Proceedings of the 75th and illness. Science, 1976,194, 1013-1020.
Annual Convention of the American Psychological Rahe, R. H. The pathway between subjects' recent
Association, 1967, 2, 207-208. (Summary) life change and their near-future illness reports:
Gunderson, E., & Rahe, R. (Eds.). Life stress and Representative results and methodological issues.
illness. Springfield, 111.: Charles C Thomas, 1974. In B. S. Dohrenwend & B. P. Dohrenwend (Eds.),
Hahn, M. E. California Life Goals Evaluation Sched- Stressful life events: Their nature and effects. New
ules. Palo Alto, Calif.: Western Psychological Ser- York: Wiley, 1974.
vices, 1966. Rahe, R. H., Lundberg, U., Theorell, T., & Bennett,
Henry, W. E. Personality change in middle and old L. K. The Social Readjustment Rating Scale: A
age. In E. D. Norbeck, D. Price-Williams, & W. comparative study of Swedes and Americans. Jour-
M. McCord (Eds.), The study of personality: An nal of Psychosomatic Research, 1971, SI, 241-249.
interdisciplinary appraisal. Chicago: Holt, Rine- Rodin, J., & Langer, E. J. Long-term effects of a
hart & Winston, 1968. control-relevant intervention with the institutional-
ized aged. Journal of Personality and Social Psy-
Holmes, T. H., & Masuda, M. Life change and illness chology, 1977, 35, 897-902.
susceptibility. In B. S. Dohrenwend & B. P. Doh-
Rotter, J. B., Seeman, M., & Liverant, S. Internal
renwend (Eds.), Stressful life events: Their nature vs. external locus of control of reinforcement: A
and effects. New York: Wiley, 1974.
major variable in behavior theory. In N. F. Wash-
Holmes, T. H., & Rahe, R. H. The Social Readjust- burne (Ed.), Decisions, values, and groups. Lon-
ment Rating Scale. Journal of Psychosomatic Re- don: Pergamon, 1962.
search, 1967,11, 213-218. Seligman, M. E. P. Helplessness. San Francisco: Free-
Hough, R. L., Fairbank, D. T., & Garcia, A. M. man, 1975.
Problems in the ratio measurement of life stress. Selye, H. The stress of life. New York: McGraw-Hill,
Journal of Health and Social Behavior, 1976, 17, 1956.
70-82. Toffler, A. Future shock. New York: Random House,
Huberty, C. J. Discriminant analysis. Review of Edu- 1970.
cational Research, 1975, 45, S43-S98. Wershow, H., & Reinhart, G. Life change and hos-
Jackson, D. N. Personality research form manual. pitalization—A heretical view. Journal of Psycho-
Goshen, N.Y.: Research Psychologists Press, 1974. somatic Research, 1974,18, 393-401.
Kobasa, S. C. Stress personality and health: A study White, R. W. Motivation reconsidered: The concept
of an overlooked possibility. Unpublished doctoral of competence. Psychological Review, 1959, 66,
dissertation, University of Chicago, 1977. 297-333.
Kobasa, S. C., & Maddi, S. R. Existential personality Wiggins, J. S. Personality and prediction: Principles
theory. In R. Corsini (Ed.), Current personality of personality assessment. Reading, Mass.: Addi-
theories. Itasca, 111.: Peacock, 1977. son-Wesley, 1973.
Lazarus, R. S. Psychological stress and the coping Wolfe, S. W. Avoid sickness—How life changes affect
process. New York: McGraw-Hill, 1966. your health. Family Circle, May 1972, pp. 30; 166-
Lazarus, R. S., Averill, J. R., & Opton, E. M., Jr. 170.
The psychology of coping: Issues of research and Wyler, A. R., Masuda, M., & Holmes, T. H. Serious-
assessment. In G. V. Coelho, D. A. Hamburg, & ness of Illness Rating Scale. Journal of Psycho-
J. E. Adams (Eds.), Coping and adaptation. New somatic Research, 1968, 11, 363-375.
York: Basic Books, 1974. Wyler, A. R., Masuda, M., & Holmes, T. H. Serious-
ness of Illness Rating Scale: Reproducibility. Jour-
Lefcourt, H. M. The function of the illusions of nal of Psychosomatic Research, 1970, 14, 59-64.
control and freedom. American Psychologist, 1973,
28, 417^25. Received January 9, 1978 •

You might also like