Personality and Individual Differences 32 (2002) 567–573
www.elsevier.com/locate/paid
Pessimistic explanatory style moderates the
effect of stress on physical illness
Benita Jackson a,*, Robert M. Sellers b, Christopher Peterson b
a
University of Missouri, Department of Psychology, 200 South Seventh Street, Room 124, Columbia, MO 65211, USA
b
University of Michigan, Department of Psychology, 525 East University Avenue, Ann Arbor, MI 48109-1109, USA
Received 13 November 2000; received in revised form 20 January 2001; accepted 1 March 2001
Abstract
Explanatory style is a cognitive personality variable that reflects the tendency to explain bad events
involving the self with causes that are internal to the self, stable across time, and global in effect. The
attribution reformulation of helplessness theory predicts that stress coupled with a pessimistic explanatory
style leads to negative outcomes, including physical illness, among at-risk individuals. This longitudinal
study of 198 college students examined whether pessimistic explanatory style interacts with perceived stress
to predict subsequent illness, even when controlling for baseline illness. Results confirmed this hypothesis.
# 2002 Published by Elsevier Science Ltd. All rights reserved.
Keywords: Explanatory style; Stress; Physical illness; Pessimism; Personality; Health; College students
What predicts the onset of illness? Current thinking points to an interaction among biological,
psychological, and situational risk factors (Justice, 1988; Ornstein & Sobel, 1989; Scheier & Carver,
1992). A representative model of these interactive influences is the attribution reformulation of
helplessness theory (Peterson & Bossio, 1991, 1993, 2001). According to this diathesis-stress
model, stress coupled with a pessimistic explanatory style leads to negative outcomes, including
illness, among individuals who are biologically or otherwise at risk (Abramson, Seligman, &
Teasdale, 1978).
Explanatory style is a cognitive personality variable that reflects the tendency to explain bad
events involving the self with causes that are internal to the self (‘‘it’s me’’), stable across time
(‘‘it’s going to last forever’’), and global in effect (‘‘it’s going to undercut everything’’) (Peterson &
Seligman, 1984). Explanatory style has been examined extensively as a risk factor for depression
(Sweeney, Anderson, & Bailey, 1986), but an emerging literature shows too that it foreshadows
* Corresponding author.
E-mail address: jacksonbe@missouri.edu (B. Jackson).
0191-8869/02/$ - see front matter # 2002 Published by Elsevier Science Ltd. All rights reserved.
PII: S0191-8869(01)00061-7
568
B. Jackson et al. / Personality and Individual Differences 32 (2002) 567–573
poor health measured in a variety of ways: symptom reports, doctor visits, physician exam,
immunosuppression, survival time with cancer, recurrence of heart disease, and untimely death
(e.g. Buchanan, 1995; Kamen-Siegel, Rodin, Seligman, & Dwyer, 1991; Levy, Lee, Bagley, &
Lippman, 1988; Lin & Peterson, 1990; Peterson, 1988; Peterson, Seligman, & Vaillant, 1988;
Peterson, Seligman, Yurko, Martin, & Friedman, 1998).
What research has not shown to date is that explanatory style interacts with stress to make
illness more likely. By and large, previous studies have not examined the diathesis-stress hypothesis. Instead, most studies have focused on testing the ‘‘main effect’’ of explanatory style on poor
health. In one study that did examine the role of stress in the form of major life events, no evidence was found for an interaction with pessimistic explanatory style (Dykema, Bergbower, &
Peterson, 1995). Although the helplessness reformulation may well be wrong with respect to the
diathesis-stress prediction, the design of this nonconfirming study makes it inconclusive. College
students were the research subjects, and they were followed for one month. The occurrence of
stressful life events for the entire time period was assessed at the end of the month, and when
during the past month these events took place was not measured.
One problem with this design is that very few of the students experienced the sorts of life events
during the study period that epidemiological research has linked to the onset of illness (Holmes &
Rahe, 1967). Stress as indexed by experiencing life events might have yielded insufficient variance
to do justice to the diathesis-stress hypothesis. As well, the life events scale used might have missed some of the events that were stressful for the participants. Finally, even if stress-inducing,
health-damaging life events did occur for some research participants, and even if they were correctly identified by the measure employed, the time period might have been too short to detect
such an effect, especially if the events occurred shortly before the end of the month, as some of
them surely would have.
Here, we report a more appropriate study. College students were again the research subjects,
but they were followed over a 9-week period. Perceived stress was measured at the beginning of
the study, along with explanatory style and a baseline measure of health status (to index generic
vulnerability to illness). Rather than asking about specific life events, we requested that research
participants each make a summary judgment of the stress they were experiencing at that time.
Subsequent health status was determined by weekly symptom reports, yielding — we hoped — a
stable estimate of physical well-being. The hypothesis of interest remained the diathesis-stress
prediction of the attribution reformulation of helplessness theory. Specifically, we predicted that
the relationship between stress and illness should be positive and stronger for individuals with
more pessimistic explanatory styles than for individuals with more optimistic explanatory styles.
1. Method
1.1. Participants and procedure
One hundred ninety-eight students from a public university in the mid-Atlantic region of the
United States participated in a 9-week study investigating the relationship between stress, coping,
and personality with physical illness and psychological distress. Participants were members of an
Introduction to Health Psychology course (n=205) in which students were given a choice
B. Jackson et al. / Personality and Individual Differences 32 (2002) 567–573
569
between participating in the study or writing three short papers. Approximately 97% of the class
chose to participate in the study. Of those who participated, 70% were female, 27% were male,
and 3% did not report their gender. Self-identified Whites comprised 69% of the sample, along
with 21% self-identified African Americans and 10% self-identified Asian Americans, Native
Americans, or Latina/os. Five percent of the sample were first-year students, 18% second-year
students, 42% third-year students, 29% fourth-year students, 3% fifth- or sixth-year students,
and 3% did not report their year in college.
During the third week of class, participants were given a health psychology journal that consisted of a series of psychological and health-related questionnaires. These measures were completed on a weekly basis over a 9-week period, with many measures repeated across weeks.
Participants were instructed to complete the journal at the end of each week of the study and to
turn in the completed portion of the journal to class on the following Tuesday.
1.2. Measures
Although there were numerous measures included in the health psychology journal, we
describe only those relevant to the present study.
1.2.1. Perceived Stress Scale (PSS)
During week one of the study, participants completed the PSS, which includes 14 items that
assess the degree to which individuals appraise their lives as stressful (Cohen, Kamarck, & Mermelstein, 1983). Participants responded to a 5-point scale (from 0, ‘‘never’’ to 4, ‘‘very often’’)
about how often they have had specific feelings or thoughts over the past month. Sample items
include: ‘‘In the last month, how often have you felt nervous and ‘stressed’?’’ and ‘‘In the last
month, how often have you felt on top of things?’’ A summary score was created by reversescoring the appropriate items and averaging across the 14 items. A higher score indicates greater
stress. In the present sample, the internal consistency of the PSS, estimated by Cronbach’s (1951)
coefficient alpha, was 0.85. Possible scores on this scale ranged from 0 to 4; actual reported scores
ranged from 0.50 to 3.36.
1.2.2. Expanded Attributional Styles Questionnaire (EASQ)
During week two of the study, participants completed the EASQ (Peterson & Villanova, 1988).
This measure presents respondents with 24 hypothetical bad events. They are asked to imagine
each event happening to them and to provide in writing the one major cause of this event. They
then rate the cause of the given event on a seven-point scale according to its internality, stability,
and globality. As in previous studies, these ratings were averaged across events and across the
three attributional dimensions to yield a single score (Peterson, Maier, & Seligman, 1993). Higher
scores on the scale represent a more internal, stable, and global — more pessimistic — explanatory
style. In the present sample, the composite explanatory style measure had a coefficient alpha of 0.88.
Possible scores on this scale ranged from 1 to 7; actual reported scores ranged from 2.61 to 5.65.
1.2.3. Physical illness
During each week of the study, participants completed an 11-item measure of physical illness
devised for the present study. The first three items asked respondents to rate the number of days,
570
B. Jackson et al. / Personality and Individual Differences 32 (2002) 567–573
from 0 to 7, in the past week in which they: (a) felt ill; (b) missed a class because of illness; and (c)
went to a doctor or health professional for a diagnosis and/or treatment of an illness. For the
final eight items, participants used a four-point scale (1= have not been bothered; 4=greatly
bothered) to rate the extent to which they were bothered during the past week with each of the
following health problems: colds, headaches, body aches, over-eating, under-eating, extreme
tiredness, insomnia, and dental problems.
Two composite scales of illness reports were generated. The first composite, which we called the
Previous Illness Index (PII), was created by standardizing all of the items in week one to place
them on the same metric and then averaging the standardized scores into a single composite
(alpha=0.69). Before standardizing, possible scores on the mean of the first three items on this
scale ranged from 0 to 7; actual reported scores on the mean of these items ranged from 0 to 5.
Possible scores on the mean of the remaining eight items ranged from 1 to 4; actual reported
scores on the mean of these items ranged from 1 to 3.38.
We created a second composite score, the Subsequent Illness Index (SII), by standardizing and
then averaging the items for weeks three through nine (alpha=0.81). Before standardizing, possible scores on the mean of the first three items on this scale ranged from 0 to 7; actual reported
scores on the mean of these items across weeks three through nine ranged from 0 to 2.72. Possible
scores on the mean of the remaining eight items ranged from 1 to 4; actual reported scores on the
mean of these items across weeks three through nine ranged from 1.08 to 2.27. For both the PII
and the SII, higher scores represent greater self-reported illness.
2. Results
The descriptive data presented in Table 1 show that this was a fairly healthy population. As
mentioned, there were relatively low levels of perceived stress and self-reported physical illness
(previous and subsequent). However, there was adequate variance in explanatory style, perceived
stress, previous illness, and subsequent illness to test our hypothesis. Gender and ethnicity were
Table 1
Means, standard deviations, and zero-order correlations
1
2
3
4
1. Previous illness
2. Stress
3. Explanatory style
4. Subsequent illness
–
0.42***
–
0.21***
0.17*
–
0.27***
0.15*
0.07
–
Mean
S.D.
0.00a
0.49
1.84
0.54
4.22
0.53
0.22
0.28
a
The mean of 0.00 for previous illness reflects that to create the Previous Illness Index, items of differing metrics
were standardized (resulting in a mean of zero), not that there was no previous illness in this sample.
*P< 0.05.
***P< 0.001.
B. Jackson et al. / Personality and Individual Differences 32 (2002) 567–573
571
not related to self-reported physical illness (previous or subsequent), and were not analyzed in
subsequent analyses.
Our hypothesis was that the interaction of pessimistic explanatory style and perceived stress
would predict subsequent physical illness, even when baseline illness was controlled. We used
hierarchical multiple regression to test both the main effects of explanatory style and previous
perceived stress, and their interaction, on subsequent illness, controlling for previous illness
(Table 2). The hierarchical regression model was constructed as follows: participants’ previous
illness was entered in Step 1; explanatory style and perceived stress were entered in Step 2; and
finally, the explanatory style by perceived stress interaction was entered at Step 3.
The explanatory style by perceived stress interaction accounted for a small but statistically
significant portion of the variance in predicting subsequent illness. The interaction of explanatory
style (higher scores indicating a more pessimistic style) and perceived stress predicted subsequent
illness, even when controlling for previous illness. Further, when regressions were calculated
separately for those with a more pessimistic explanatory style and for those with a more optimistic style (as determined by a median split on the EASQ), perceived stress predicted subsequent
illness for pessimistic individuals (=0.28, t=2.8, P< 0.005) but not for optimistic individuals
(= 0.07, t=-0.64, ns; Fig. 1). In sum, even when controlling for initial stress, pessimistic
explanatory style appears to exacerbate the impact of stress on illness.
3. Discussion and conclusion
These results support the diathesis-stress prediction of the attribution reformulation of helplessness theory: pessimistic explanatory style interacts with perceived stress to predict subsequent
physical illness, even when baseline illness is controlled. For those with an optimistic explanatory
style in this study, there was no relationship between stress and illness. These data also suggest
that researchers do not capture the whole story when examining only the main effects of either the
diathesis or the stressor on subsequent physical illness. Neither perceived stress nor pessimistic
attributional style alone predicted subsequent physical illness; only the interaction of perceived
stress and pessimistic attributional style predicted subsequent physical illness. While the interaction accounted for a small portion of the variance, it still accounted for more than the given
diathesis and stressor each did separately. Further, these data show that personality
Table 2
Hierarchical multiple regression analyses predicting subsequent illness
Step and predictor variable
1. Previous illness
2. Explanatory style
stress
3. Explanatory style stress
*P=0.05.
**P< 0.01.
***P< 0.001.
Subsequent illness
0.27***
0.04
0.02
R2
0.07***
0.00
0.02*
Final
0.23**
0.03
0.02
0.14*
572
B. Jackson et al. / Personality and Individual Differences 32 (2002) 567–573
Fig. 1. The relationship between perceived stress and subsequent illness by explanatory style.
characteristics — such as pessimistic explanatory style — can serve as a diathesis for subsequent
physical health outcomes. This suggests that researchers and clinicians could meaningfully incorporate personality measures to help understand the onset and course of physical health and illness.
As well, these data suggest that a more optimistic explanatory style may be a protective factor
against the deleterious influence that stress may have on physical health. However, it is still unclear
as to the processes by which explanatory style may be interacting with stress to effect health. One
possible process is through the appraisal and coping process. Sellers and Peterson (1993) found
evidence in a sample of college athletes that explanatory styles were related to coping behavior via
individuals’ appraisals of controllable events. Specifically, pessimistic individuals were more likely
to appraise these controllable events as being ones that they could handle, and in turn, reported
greater coping effort. It is possible that when individuals with more pessimistic explanatory styles
confront stress, their propensity to appraise stressful situations as manageable makes them particularly vulnerable to physical illness when they are unsuccessful in their attempts to reduce the stress.
In sum, this research provides a useful first step toward establishing that explanatory style
interacts with stress to influence physical health. Importantly, there are limitations to this study
that need to be remedied in future investigations. To see if this interaction of pessimistic explanatory style and stress may be generalized beyond a college sample, future research needs to use
samples more diverse across age span, perceived stress, and physical health. Using different
methods to assess explanatory style, stress, and physical health would be important to establish
that the effects found here are due to more than shared method variance. For example, future
studies could use peer, partner, and self-ratings of explanatory style, stress, and physical illness.
Physical illness could also be assessed from medical records. Also, it is important to see if this
interaction may be generalized to predicting other types and degrees of illness outcomes: accidents, more serious physical illnesses, and emotional distress. Finally, it remains to be explained
whether explanatory style is most linked to the onset of illness, its course, or its recurrence.
B. Jackson et al. / Personality and Individual Differences 32 (2002) 567–573
573
Acknowledgements
This work was supported in part by a University of Michigan Substance Abuse Research
Center Pre-Doctoral Research Fellowship and National Institute of Health Grant P50HL061202-01.
References
Abramson, L. Y., Seligman, M. E. P., & Teasdale, J. D. (1978). Learned helplessness in humans: critique and reformulation. Journal of Abnormal Psychology, 87, 49–74.
Buchanan, G. M. (1995). Explanatory style and coronary heart disease. In G. M. Buchanan, & M. E. P. Seligman,
Explanatory style (pp. 209–232). Hillsdale, NJ: Erlbaum.
Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social
Behavior, 24, 385–396.
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297–334.
Dykema, J., Bergbower, K., & Peterson, C. (1995). Pessimistic explanatory style, stress, and illness. Journal of Social
and Clinical Psychology, 14, 357–371.
Holmes, T. H., & Rahe, R. H. (1967). The social readjustment rating scale. Journal of Psychosomatic Research, 11,
213–218.
Justice, B. (1988). Who gets sick? How beliefs, moods, and thoughts affect your health. Los Angeles: Tarcher.
Kamen-Siegel, L., Rodin, J., Seligman, M. E. P., & Dwyer, J. (1991). Explanatory style and cell-mediated immunity.
Health Psychology, 10, 229–235.
Levy, S., Lee, J., Bagley, C., & Lippman, M. (1988). Survival hazards analysis in first recurrent breast cancer patients: 7
year follow-up. Psychosomatic Medicine, 50, 520–528.
Lin, E. H., & Peterson, C. (1990). Pessimistic explanatory style and response to illness. Behaviour Research and Therapy, 28, 243–248.
Ornstein, R., & Sobel, D. (1989). Healthy pleasures. Reading, MA: Addison-Wesley.
Peterson, C. (1988). Explanatory style as a risk factor for illness. Cognitive Therapy and Research, 12, 117–130.
Peterson, C., & Bossio, L. M. (1991). Health and optimism. New York: Free Press.
Peterson, C., & Bossio, L. M. (1993). Healthy attitudes: optimism, hope, and control. In D. Goleman, & J. Gurin (Eds.),
Mind/body medicine: how to use your mind for better health (pp. 351–366). Yonkers, NY: Consumer Reports Books.
Peterson, C., & Bossio, L. M. (2001). Optimism and physical well-being. In E. C. Chang (Ed.), Optimism and pessimism: Implications for theory, research, and Practice (pp. 127–145). Washington, DC: American Psychological
Association.
Peterson, C., Maier, S. F., & Seligman, M. E. P. (1993). Learned helplessness: a theory for the age of personal control.
New York: Oxford University Press.
Peterson, C., & Seligman, M. E. P. (1984). Causal explanations as a risk factor for depression: theory and evidence.
Psychological Review, 91, 347–374.
Peterson, C., Seligman, M. E. P., & Vaillant, G. E. (1988). Pessimistic explanatory style is a risk factor for physical
illness: a thirty-five year longitudinal study. Journal of Personality and Social Psychology, 55, 23–27.
Peterson, C., Seligman, M. E. P., Yurko, K. H., Martin, L. R., & Friedman, H. S. (1998). Catastrophizing and untimely death. Psychological Science, 9, 127–130.
Peterson, C., & Villanova, P. (1988). An expanded attributional style questionnaire. Journal of Abnormal Psychology,
97, 87–89.
Scheier, M. F., & Carver, C. S. (1992). Effects of optimism on psychological and physical well-being: theoretical
overview and empirical update. Cognitive Therapy and Research, 16, 201–228.
Sellers, R. M., & Peterson, C. (1993). Attributional style and coping among student-athletes during controllable
stressful events. Cognition and Emotion, 7, 431–441.
Sweeney, P. D., Anderson, K., & Bailey, S. (1986). Attributional style in depression: a meta-analytic review. Journal of
Personality and Social Psychology, 50, 974–991.