Who Talks?
The Social Psychology of Illness Support Groups
Kathryn P. Davison,
James W. Pennebaker
Sally S. Dickerson
More Americans try to change their health behaviors
through self-help than through all other forms of profes-
sionally designed programs. Mutual support groups, in-
volving little or no cost to participants, have a powerful
effect on mental and physical health, yet little is known
about patterns of support group participation in health
care. What kinds of illness experiences prompt patients to
seek each other's company ? In an effort to observe social
comparison processes with real-world relevance, support
group participation was measured for 20 disease catego-
ries in 4 metropolitan areas (New York, Chicago, Los
Angeles, and Dallas) and on 2 on-line Jbrums. Support
seeking was highest for diseases viewed as stigmatizing
(e.g., AIDS, alcoholism, breast and prostate cancer) and
was lowest for less embarrassing but equally devastating
disorders, such as heart disease. The authors discuss im-
plications for social comparison theory and its applications
in health care.
T he experience of illness is a profoundly social one.
Suffering elicits intense emotions and hence the
desire to talk to others. Through interpersonal ex-
changes, patients develop an understanding of their illness:
They may talk to friends, relatives, and professionals about
what their diagnosis and treatment may entail. Over the
course of their particular illness, relationships are strained
or broken, and new ones become valuable, such as those
with doctors, nurses, or physical therapists. For some, the
condition itself constitutes a dangerous secret that erects a
barrier between themselves and their support network.
Thus, patients' experiences of illness both influence, and
are influenced by, the social fabric that surrounds them.
A recent development in the social milieu of the
patient is the proliferation of mutual support groups for
people coping with illness. Across the country, in hospitals,
churches, empty offices, and even shopping malls, small
groups of individuals assemble to cope collectively with
their unique challenges. Such populist approaches often go
unnoticed by psychologists because of a bias toward pro-
fessionalism on the one hand and a lack of awareness on
the part of group participants that they are part of a larger
movement on the other (Jakobs & Goodman, 1989). The
psychological and physical health importance of this dif-
fuse community is striking. Through such groups, millions
of Americans attempt to overcome addiction, discuss inno-
February 2000 • American Psychologist
Copyright2000by the AmericanPsychologicalAssociation.lnc, 0003-066X/00/$5.00
Vol. 55, No. 2. 205-217 DOt: 10.1037//0003-066X.55.2.205
The Human Asset
University of Texas at Austin
University of California, Los Angeles
vations in insulin treatments, grieve for the loss of a breast,
or share fears about the possible progression of HIV to
AIDS. Some conditions are associated with more support
groups than others. What kinds of illness experiences
prompt the formation and maintenance of support groups?
What makes the support group appealing for cancer pa-
tients but not for cardiac patients? Why is mutual support
an intrinsic part of AIDS case management but utterly
lacking from hypertension treatment'?
Venues for Social Comparison
Festinger's (1954) social comparison theory postulated that
social behaviors could be predicted largely on the basis of
the assumption that individuals seek to have and maintain
a sense of normalcy and accuracy about their world. In
times of uncertainty, Festinger predicted that affiliative
behaviors would increase as people sought others' opinions
about how they should be thinking or feeling.
Empirical tests of social comparison theory have
found that affiliative behaviors increase under conditions of
high anxiety. Participants awaiting a fake medical proce-
dure presumed to include an electric shock expressed a
strong preference for the presence of another, especially
someone facing the same procedure (Schachter, 1959).
However, later studies about affiliative motivation raised
further doubts about social comparison motives (e.g.,
Teichman, 1973). Sarnoff and Zimbardo (1961) examined
preferences of college students who thought they were
awaiting an electric shock but also were told that they
would be required to suck bottle nipples as part of the
experimental procedure. In this more embarrassing context
of anxiety, participants indicated a marked decrease in the
desire to have another present. Sarnoff and Zimbardo con-
cluded that the kind of anxiety associated with humiliation
E d i t o r ' s note. Melissa G. Warren served as action editor for this article.
Kathryn P. Davison, The Human Asset, Dallas, Texas;
A u t h o r ' s note.
James W. Pennebaker, Department of Psychology, University of Texas at
Austin: Sally S. Dickerson, Department of Psychology, University of
California. Los Angeles.
Preparation of this article was aided by Grant MH52391 from the
National Institutes of Health. We are indebted to Laura King, David Buss,
and Josh Holahan tbr their comments on an earlier version of this article.
Correspondence concerning this article should be addressed to James
W. Pennebaker, Department of Psychology, University of Texas, Austin,
TX 78712. Electronic mail may be sent to pennebaker@psy.utexas.edu.
205
or embarrassment actually would decrease affiliative mo-
tivation. Anxiety apparently motivated socialization but
could not serve as a singular explanatory variable in the
social comparison process, because the emotion resulted
from different concerns.
Social comparison is intrinsic to the health care set-
ting, where anxiety levels are often high, and information,
when available, may not come in a form that patients
understand. Diagnosis, treatment, side effects of medica-
tion, or other kinds of life disruptions prompt patients to
talk with others undergoing a similar challenge.
Characterizing the Self-Help Category
The self-help concept stems from the assertion that people
facing a similar challenge can help each other simply by
coming together. The power of this approach lies in the
belief that a collective wisdom is born through the shared
experience of participants rather than through the profes-
sional training or style of the leader. The kinds of help
requested and offered in this context are largely free of
professional structures or assumptions (Riessman, 1985).
Self-help group participation costs its members little or
nothing.
Overall epidemiological figures on self-help in the
United States indicate a prevalence rate of approximately
3% to 4% of the population over a 1-year period, and
lifetime participation rates are estimated at around 25 mil-
lion (Kessler, Mickelson, & Zhao, 1997; Lieberman &
Snowden, 1993). Lifetime use of self-help is higher for
men than for women (3.6% vs. 2.4%). Caucasians are three
times as likely to participate as African Americans, with
Hispanic participation levels falling about midway between
the two. The mean age of participants is 43.1 years, and
mean education level is 12 years (slightly higher than the
national average of 11 years).
Due to the grass-roots nature of the self-help phenom-
enon, support groups exist in a wide range of styles and
structures. Precise boundaries around self-help categoriza-
tion are not only difficult, but impractical. Within the
self-help arena, considerable innovation and diversity of
style occur as participants tailor group processes over time
to reflect group needs and goals. Lieberman and Snowden
(1993) have reported that over 60% of observed groups
were professionally facilitated while simultaneously being
characterized as self-help. Such findings point to the diffi-
culty of categorical boundaries and the problematic nature
of research in real-world social patterns. Nevertheless,
groups that define themselves as self-help, whether they are
professionally led or not, are viewed by members as sell'-
help. For example, many hospital-based support programs
described as professionally facilitated are in fact hybrid
programs developed with guidelines from a national foun-
dation and idiosyncratically shaped by the time commit-
ments of the patient volunteers and the coordination skills
of a nurse or social worker employed by the hospital.
Group counts say little about health outcomes or other
quality-of-treatment measures, just as other reports of so-
cial participation, such as church membership or marital
status, say little about the quality of those experiences.
206
Although distinctions between professionally facilitated
and peer-mediated groups may be meaningful in outcome
studies aimed at discerning the active agents in curative
processes, it is important to bear in mind that as a measure
of basic value to participants, participation is its own index
of success: Groups without value cease to be groups. Mem-
bers vote with their feet.
The most comprehensive analysis of self-help groups
in general found that the leading reason for participation in
groups of any kind was the experience of physical illness
(Lieberman & Snowden, 1993). Even after excluding cop-
ing with substance abuse, groups for coping with physical
illness composed approximately 42% of the self-help-
group population, dwarfing statistics for bereavement,
crime victimization, parenting issues, and personal growth,
yet no studies have addressed the epidemiology of mutual
support by diagnostic category.
Effectiveness of Mutual Support in
Improving Physical a n d
Psychological Health
Support groups are an important means by which Ameri-
cans change their health behaviors. In some cases, mutual
support constitutes the exclusive "treatment" for a health
problem, as in the case of Alcoholics Anonymous (AA).
Self-help has also been found to be as effective as profes-
sional treatment for some forms of mental (e.g., depres-
sion) or physical (e.g., headache) illnesses that have tradi-
tionally been viewed as the domain of psychotherapy or
medicine (Gould & Clum, 1993). The stories told and heard
in this context carry the weight of shared experience, the
emotional potency of common suffering, and an avenue for
social learning. Rappaport (1993) has suggested that these
shared stories form a kind of group narrative that consti-
tutes a social identity, distinguishing the self-help group
from any kind of formal psychotherapy (see also Gergen &
Gergen, 1997). In a similar vein, Yalom (1995) has as-
serted that self-help groups offer a unique venue for
growth, social experimentation, and change.
Measurement of the efficacy of self-help is problem-
atic from the outset because of its inherently self-selected
nature. However, investigations of the effects of support-
group participation, even under random assignment, have
yielded positive results on the whole. Rheumatoid arthritis,
cancer, heart attack, and epilepsy patients in support
groups, for example, have exhibited more health benefits
than nonparticipating controls or controls on a waiting list
(Bradley et al., 1987; Dracup, 1985; Droge, Arntson, &
Norton, 1986; Telch & Telch, 1986). Mutual support in-
terventions can be highly cost effective: One study of
rheumatoid arthritis patients participating in peer-facili-
tated groups reported mean pain reductions of 20%, in-
creases in self-efficacy, and an average four-year cost sav-
ings of $648 per person (Lorig, Mazonson, & Holman,
1993). Pain reductions were strongly related to increases in
self-efficacy, an identified factor in positive health behavior
change (DiClemente, 1995). If typical self-help participa-
tion rates (3%) occurred among the 32 million arthritis
sufferers, the four-year savings in arthritis care alone could
February 2000 • American Psychologist
total $650 million. The pressures of managed care and
incentive structures such as capitation may increase atten-
tion to lower cost approaches to health management.
Support group participation has also been associated
with superior prognosis in breast cancer and heart disease.
In one highly publicized study, a sample of 88 metastatic
breast cancer patients received either group psychotherapy
or informational and nutritional support. Those randomly
assigned to the therapy condition had survival times ap-
proximately twice those of controls (1.50 vs. 0.75 years,
respectively; Spiegel, Bloom, Kraemer, & Gottheil, 1989).
Similar findings have been reported for a sample of patients
with malignant melanoma (Fawzy et al., 1993). Finally,
various programs have been adopted for heart patients that
have relied heavily on regular group support for patients
and, on occasion, their spouses. Across multiple studies,
markers of heart disease progression, including measures
of atherosclerosis and resting blood pressure, have demon-
strated clinically significant improvement (Ornish et al.,
1990).
The interventions cited above represent highly refined,
professionally developed, programmatic approaches to
coping with serious health threats. Although they bear
marked differences from patient-initiated groups, they may
also share some fundamental therapeutic qualities worth
noting, such as opportunities for disclosure, empathic con-
nection, shared goals, and psychological adjustments to life
challenges. They underscore the medical value of mean-
ingful, group-based programs whose psychological focus
elicits psychosocial and physical health benefits. Addi-
tional research can help clinicians identify the optimal
blend of patient and professional agendas to maximize
treatment success.
On the basis of these very limited reports, it is appar-
ent that group approaches to health care can play an instru-
mental role in health behavior change, treatment adherence,
cost control, and disease reversal. Such groups have sig-
nificant therapeutic potential, yet it is apparent that the
primary feature of mutual-support participants is their self-
selected nature. These considerations prompted a series of
investigations about the demand side of the support group
community--who participates and why.
Examining_Supportive Behaviors as a
Function of Illness Type
Illnesses vary along a number of psychologically meaning-
ful dimensions, such as life threat, embarrassment, and
behavioral impact on disease course. Although health con-
cerns are the most frequently cited reason for joining a
support group, no studies have examined how support-
group participation varies as a function of illness features
or type. We designed a series of three studies to assess (a)
the prevalence of support groups in 20 different patient
populations, (b) the activity rates of virtual support forums
for those same patient groups, and (c) the identifiable
factors associated with support seeking across diagnostic
category.
Our choice of methodology merits some discussion.
The selection of diseases, rather than individuals, as sub-
February 2000 • American Psychologist
jects and the use of groups as our units of analysis, rather
than counts of individuals, is somewhat unusual. However,
the experience of illness does engender, either permanently
or temporarily, an illness identity (for an excellent treat-
ment of these themes, refer to Kleinman, 1995). We wanted
to examine the social context of illness at the collective
level. To do so, we adapted a design that would allow
researchers to observe something of how groups coalesce.
In our first study, the goal was to identify patterns, if
any, of support group participation by illness category. To
address this question, four major metropolitan areas--New
York, Los Angeles, Dallas, and Chicago--were canvassed
in an eftort to identify every support group that existed for
each of the 20 studied conditions. Counts were subse-
quently adjusted for population and prevalence and tested
for consistency across cities.
An extension of the first question centered on the
newest avenue of social interface, the Internet. In our
second study, the primary goal was to assess identifiable
patterns of mutual support activity among patient popula-
tions in this virtual community. A related goal was to
understand in what ways virtual and actual support patterns
would resemble one another. For a period of two weeks,
every post made to support forums for the 20 conditions in
two on-line domains (America Online and the larger Inter-
net) was collected and counted as a measure of mutual
support.
in our third study, we sought to identify factors asso-
ciated with the search for mutual support, both face-to-face
and on-line. Judges' ratings were obtained on 16 dimen-
sions: suspected of playing a role in support behaviors.
Ratings were then correlated with support measures to
generate a sense of the issues that drive the search for
mutual support.
Study 1: AssessingSupport Patterns
in Metropolitan Areas
Four major metropolitan areas--Chicago, Dallas, Los An-
geles, and New York--were the focus of a semistructured
series of surveys to identify the number of support groups
in that area for patients suffering from 20 different health
problems. We contacted the mental health services agency
serving each of the surveyed cities and requested any
handbook or comprehensive compilation of support groups
for that area. After receiving the compilations, each agency
or contact listed under every diagnostic category studied
was called to confirm that the group existed and was not a
duplicate or expired listing. Counts resulting from these
calls were assembled into a support profile for each city.
Disease Selection Criteria
The conditions selected for study are listed in Table 1. The
selection process was driven by practical rather than theo-
retical reasons, with the goal of capturing a wide range of
problems that afflict people potentially capable of attending
a support group. Inclusion efforts focused on the most
prewtlent and deadly conditions, as well as those with
significant psychological and behavioral components, pre-
207
 bias a n d reflected the a p p r o a c h o f p o t e n t i a l users, F o r this
Table i study, a g r o u p w a s c o u n t e d as s e l f - h e l p if the cost o f
An Overview of the Prevalence and Death Toll p a r t i c i p a t i o n did not e x c e e d $8 p e r session, if the r e s p o n -
Associated With the Studied Diseases d e n t d e p i c t e d the g r o u p as self-help, a n d if the m e e t i n g s
w e r e held on a r e g u l a r basis, w h e t h e r the g r o u p was pro-
Prevalence Deaths
f e s s i o n a l l y f a c i l i t a t e d or not.
Rate Rate W e identified a n d c o n t a c t e d g r o u p s for c o n f i r m a t i o n
Disease type (in thousands) Rank (in thousands) Rank t h r o u g h the listings in local registries o f s e l f - h e l p a n d social
s e r v i c e f a c i l i t i e s . " T h e s e registries feature listing c a t e g o r i e s
AIDS 785 16 9.8 11
Alcoholism 13,760 5 8.0 12 a c c o r d i n g to d i a g n o s i s or need, s u c h as " b e r e a v e m e n t " or
Anorexia nervosa 530 18 0.5 15 " d i a b e t e s . " O n c e the g r o u p s w e r e identified, w e c o n t a c t e d
Arthritis 32,642 1 0.0 18 e a c h o n e and, in a s t a n d a r d i z e d protocol, d e t e r m i n e d the
Asthma 13,074 6 1.4 14 n u m b e r o f g r o u p s run t h r o u g h the p a r t i c u l a r a g e n c y , the
Breast cancer 1,769 13 12.0 9 f r e q u e n c y o f g r o u p m e e t i n g s , a n d the a v e r a g e g r o u p size. 2
Colon cancer 911 15 23.6 5
Lung cancer 394 19 40.8 3 Survey Results and Discussion
Prostate cancer 583 17 13.5 8 T h e s u r v e y p r o c e s s y i e l d e d a total o f 12,596 s u p p o r t g r o u p s
Chronic fatigue 1,000 14 0.0 18 in the four cities. T h e raw counts, the p o p u l a t i o n - a d j u s t e d
Chronic pain 30,000 2 0.0 18 c o u n t s , a n d the p r e v a l e n c e - a d j u s t e d i n d i c e s o f s u p p o r t o f f e r
Depression 9,400 8 11.0 10 a variety o f i n s i g h t s a b o u t the s c o p e a n d f r e q u e n c y o f
Diabetes 7,813 9 19.6 7
s u p p o r t s e e k i n g for h e a l t h c h a l l e n g e s (see T a b l e 2).
Emphysema 1,900 12 36.0 4
A s is e v i d e n t at the b o t t o m o f T a b l e 2, the four cities
Heart disease° 21,255 4 288.0 1
Hypertension 27,549 3 2.0 13 v a r i e d c o n s i d e r a b l y in the t e n d e n c y to p a r t i c i p a t e in sup-
Migraine 11,023 7 0.0 18 port groups. P o p u l a t i o n - a d j u s t e d g r o u p totals i n d i c a t e that
Multiple sclerosis 350 20 0.0 18 s o m e cities are c h a r a c t e r i z e d b y h i g h e r overall s u p p o r t
Stroke 3,000 11 56.0 2 levels than others. A c h i - s q u a r e a n a l y s i s o f the four city
Ulcer 4,569 10 22.0 6 g r o u p totals ( a d j u s t e d ) y i e l d e d a h i g h l y significant differ-
e n c e b e t w e e n e x p e c t e d a n d o b s e r v e d totals for the four
Note. All figures were derived from the 1992 and 1993 reports of the
National Centerfor Health Statistics and the 1996 report of the Centersfor cities, X2(3) = 491.31, p < .001. D a l l a s h a d the l o w e s t
Disease Control and Prevention(CDC), exceptfor the following: alcoholism level o f support, w i t h 139.5 g r o u p s p e r m i l l i o n , w h e r e a s
(Taylor, 1995); anorexianervosa(NationalAssociationoFAnorexiaNervosa C h i c a g o h a d the h i g h e s t m e a n o f 754.9. A l t h o u g h it is
and AssociatedDisorders, http://www, anad.org.Factsht97,htmI, retrievedJune
1996); chronic fatigue (CDC, 1994; Kroenke et al., 1988); chronic pain b e y o n d the s c o p e o f this study to identify the r e a s o n s for
(Taylor, 1995); depression(NationalMentalHealthAssociation,http://v,,",~ t h e s e v a r i a t i o n s in s u p p o r t levels across cities, it w o u l d b e
mhsourcecom/resource/mh/html, retrievedJune 1996); diabetes (National
Institutes of Health, 1995}; emphysema(AmericanLungAssociation,http:// helpful to u n d e r s t a n d the factors at w o r k - - l o c a l m e n t a l
www.lung.usa.org.noframes/learn/lung/lungemphysema.html, retrievedJune h e a l t h a p p r o p r i a t i o n s levels, f o u n d a t i o n support, p o p u l a -
1996); migraine (National HeadacheFoundation,h~p://www.headaches tion density, or historical a n d c u l t u r a l o r i e n t a t i o n s - - t h a t
org, retrievedJune 1996); multiplesclerosis (Multiple SclerosisFoundation,
http://www msfacts.org/html, retrievedJune 1996); and ulcer(U.S. Bureauof
the Census, 1996).
° The prevalenceratefor heartdiseasesis the sum of 13,490,000 myocardial F~r the four surveyed areas, the handbooks used were as follows:
infarctionand coronaryheartdiseasediagnosesand almost8,000,000 angina Los Angeles--Los Angeles Count' Social Service Resource Directory;
cases with or without identifiedheartproblems. New York--Soeial Service Directo O' of Greater New York Cio' and
Directoly oj"Se!/-Help Groups in New York City," Chicago---Directot-y of
Sel]:Hell) and Mutual Aid Groups; Dallas--untitled private compilation
provided by the Dallas County Mental Health Association. All listings
under the relevant categories were called unless there were more than 20
s u m a b l y r e s p o n s i v e to p s y c h o s o c i a l support. C o n d i t i o n s separate listings, in which case only the first 20 were contacted, and the
a s s o c i a t e d w i t h e x t r e m e old age or c o n t a g i o n w e r e ex- mean results of those first 20 were extrapolated to the rest of the listings.
cluded from consideration. For example, more than 20 agencies and individuals facilitate groups for
AIDS patients in the New York area. so the first 20 listed were contacted.
T h e p r e v a l e n c e a n d m o r t a l i t y rates for e a c h o f the The number of groups in existence was obtained for each listing, and then
studied c o n d i t i o n s w e r e c o l l e c t e d to a d j u s t s u p p o r t - g r o u p these numbers were averaged: the resulting average was multiplied by the
c o u n t s for the p o p u l a t i o n f r o m w h i c h they w e r e d r a w n . total number of listings.
U n l e s s o t h e r w i s e noted, the s o u r c e for all was the N a t i o n a l = When a group was called but there was no answer, there were three
possible outcomes: (a) If a message machine played a tape of scheduled
C e n t e r for H e a l t h Statistics ( N C H S , 1996). Statistics un- meetin,2-s, that intk)rmation was regarded as current confirmation and was
available from NCHS were obtained from foundations recorded as data. (b) If a message machine identified that listing (tele-
s e r v i n g the p a t i e n t g r o u p s (e.g., A m e r i c a n C a n c e r S o c i e t y ) phone number) as the contact for a support group, the researcher would
or f r o m e p i d e m i o l o g i c a l r e s e a r c h reports (refer to Notes record the groups as "1" and make an effort to call back and confirm. (c)
If a message machine answered and no mention was made of a group, or
section o f T a b l e 1).
in the case of no answer at all, the researcher made three further attempts
Survey Procedure to confirm the listing. No groups were counted unless confirmed. Finally,
if in the course of the survey the contact person volunteered the name of
T h e goal o f the r e s e a r c h was to identify as m a n y g r o u p s as a group or agency that was not featured in the researcher's registry, the
p o s s i b l e p e r c o n d i t i o n in a w a y that m i n i m i z e d s a m p l i n g researcher also contacted that group.

208 February 2000 • American Psychologist

Table 2
Support Group Totals by. Condition and City., Mean Groups Across Cities, and Prevalence-Adjusted Support
Indices Across Cities, Along With Their Ranks
Chicago Dallas LosAngeles New YorkCity All cities
No. of No. of No. of No. of
Condition groups Rank groups Rank groups Rank groups Rank Mean° Rank Indexb Rank

AIDS 21 9 35 2 155 2 257 2 18.96 2 241.52 2

Alcoholism 4,003 1 128 1 2,835 1 4,000 1 402.78 1 292.72 1
Anorexia 32 8 2 13 13 7 7 7 2.19 12 41.30 4
Arthritis 19 11 5 7 18 3 5 10 2.12 13 0.65 16
Asthma 19 11 22 3 6 10 6 9 4.06 9 3.10 14
Cancer, breast 57 4 12 4 18 3 97 3 10.48 3 44.49 3
Cancer, colon 37 5 2 13 4 11 2 12 5.79 5 21.55 6
Cancer, lung 1 16 2 13 1 15 1 15 4.93 6 9.33 9
Cancer, prostate 13 13 6 5 16 6 7 7 6.04 4 34.66 5
Chronic tatigue syndrome 19 11 1 15 1 15 0 18 0.94 14 9.38 8
Chronic pain 2 15 0 18 0 19 1 15 0.12 17 0.04 18
Depression 57 3 6 5 17 5 31 4 4.71 7 5.01 12
Diabetes 80 2 1 15 10 8 23 5 4.51 8 5.77 11
Emphysema 0 18 0 18 1 15 2 12 0.63 15 3.29 13
Heart disease 36 6 4 8 4 11 5 10 2.29 11 1.08 15
Hypertension 0 18 0 18 0 19 0 18 0.00 19 0.00 19
Migraine 0 18 0 18 0 19 0 18 0.00 19 0.00 19
Multiple sclerosis 3 14 3 10 1 15 1 15 0.58 16 16.61 7
Stroke 34 7 3 10 10 8 10 6 2.42 10 8.08 10
Ulcer 0 18 0 18 1 15 0 18 0.03 18 0.06 17
Group total 4,605 265 3,201 4,535
Population (in millions) 6.10 1.90 8.90 7.00
Mean groups per million 754.92 139.47 359.66 647.86
Note. Populationfiguresfor citiesare fromthe U.S.CensusBureauStatistics(ht~p://venus.census.gov/cdrom/Iookup,retrievedSeptember20, 1996).
° Meanswerecomputedby averaginggroupcountsper millionacrosscity. b Indexrepresentsprevalence-adjustedsupportmeasurederivedfromcomputingmean
groups per million(acrossthe four citiessampled)dividedby prevalence,multipliedby 1,3,000.

make mutual support more the norm in Chicago than in
Dallas.
To assess the relative consistency of supportiveness as
a construct by diagnostic category, we computed Spearman
rank correlations (r~) for the tour city counts. The resulting
correlations were high: A reliability analysis indicated an
alpha for the four cities of .89. Given this high degree of
intercorrelation, we collapsed across cities to generate a
population-adjusted mean and rank for each disease, in-
cluded in Table 2. Not surprisingly, alcoholism ranked first,
with AIDS second. The cancers followed, with breast can-
cer showing the highest levels of support and lung cancer
the lowest. Finally, collapsed population-controlled counts
were adjusted for prevalence (figures represented in Table
1) so that the resulting support indices also reflected the
population sizes from which the participants were drawn.
The preponderance of support groups in the alcohol-
ism category is at once predictable and noteworthy and so
deserves specific mention. Of the 12,596 total groups iden-
tified for all of the conditions studied, across all cities
sampled, AA groups constituted 10,966, or 87%, of the
group counts. These numbers, and hence the AA story, are
a testament to the potential strength and efficacy of mutual
support: The largest number and degree of health behavior
changes in the country are the product of a network of
largely anonymous, expert-free, cost-free groups whose
sole purpose is mutual support. It is also useful to note that
the AA philosophy asserts that group participation is a
lifetime commitment: Addiction is an incurable condition
managed only by enduring vigilance. Although a detailed
account of the AA literature is outside the focus of this
article, multiple factors have been identified that describe
self-help and treatment success in coping with addiction
(e.g., Tucker, 1995).
What is striking is the degree, or range, in support
seeking in general among illness categories after adjusting
for prevalence. AIDS patients, for instance, are 250 times
more likely to participate in a support group than hyper-
tension patients. Breast cancer patients have formed over
40 times as many support groups as heart disease patients,
whose conditions undeniably benefit from psychosocial
and behavioral changes. Such contrasts raise questions
about the motivations of participants.
Our methodology suffers from a few problems intrin-
sic to the self-help culture that make clear quantification
inherently difficult. Comorbidity in both disease and sup-

February 2000 ° American Psychologist 209

port makes clean counts problematic. For instance, stroke is
highly related to hypertension. It is evident, though, that
those with high blood pressure will seek support after a
stroke much more readily than before it. Individuals suf-
fering from hypertension, moreover, have a higher proba-
bility of also suffering from diabetes or heart disease. It is
unclear whether patients carrying more than one diagnosis
tend to be involved in more than one support group. In
some cases, support groups were not exclusively composed
of individuals in the diagnostic category of interest. For
example, in the case of support groups for anorexia, some
(e.g., Overeaters Anonymous) included participants with
bulimia and obesity problems as well.
A second unavoidable problem concerns the process
of adjustment for the prevalence rates of the diseases.
Given the relatively low participation rates of all patient
types in mutual support, the observed ranks are heavily
driven by the prevalence rates themselves. Despite the
acknowledged problems, the methodology was the most
valid means of assessing support group patterns without
introducing significant bias. The observed participation
patterns in these face-to-face groups prompted questions
about how virtual avenues of support might elicit similar or
different patterns among the same patient groups.
Study 2: Frequency and
Characteristics of On-Line Support
As of 1996, approximately 15% of Americans (about 40
million) already had Internet access, with growth estimated
at 20% per month (Ferguson, 1996). Moreover, interest in
on-line participation is significantly driven by health issues.
A recent Louis Harris poll conducted to assess the motives
behind new subscriptions to on-line services found that
health concerns were ranked first (reported by Ferguson,
1996). The social connections enabled by the advent of the
Internet constitute a new forum of social support that has
unknown, and largely unstudied, potential. For the patient,
the availability of on-line support introduces a new dimen-
sion of social connection and access to information. For
researchers, the observation of on-line support groups of-
fers a unique window for understanding the kinds of expe-
riences that patients wrestle with, from reluctant insurers to
patronizing doctors, as well as histories of suffering, con-
fusion, and misdiagnosis. A few examples from our Inter-
net posts are featured below.
Dear Carol, Hi! I understand your concerns. I am 47 and I just
survived my first and hopefully last heart attack. I too continue to
think of one day or the next will be my last day. I weep for this
because I have not danced at my daughter's wedding and I
planned to sit on a porch with my husband and I cling to this life
with all of the intensity that I can muster. But I know your fear
and I know your depression so very well. Whether or not I can
make it past another twelve months without another episode is
now the issue since I have survived the first month. Carol they just
don't know very much about women and heart disease--this
much I know for sure. (AOL subscriber, heart disease bulletin
board)
210
Was wondering if any women who have had breast cancer have
noticed a correlation between stress in their life and the onset of
the bc? Two years ago my husband and I separated and in the last
year--the year from hell--I experienced a car accident, the real-
ization that someone I loved was on the verge of schizophrenia, a
peeping Tom invading my privacy, my father's death, the loss of
my job and financial security, and then breast cancer. . . . After
awhile [sic] I wondered if it was worth living. I lost my zest and
couldn't find much joy in life . . . . Psychologically and emotion-
ally, as most of you know who've been through similar things, I
feel wiped out. I pray a lot and have good friends who help me as
they can and also pray for me. (contributor to an Internet support
forum for breast cancer)
Let me give you my idea on how to be a good CFS [chronic
fatigue syndrome] wife. Tell him to order pizza or something.
Have him clean the exotic molds in the refrigerator. Prepare
yourself. Stay in bed all day and make sure you are more inter-
esting than the vixens he works with all day . . . . Give him a
grocery list and tell him to buy whatever he wants to drink. Feel
free to cry about your day. Men love to comfort and protect. Go
to sleep the moment he arrives. If he wants a clean house, he can
visit a neighbor. (contributor to the chronic fatigue discussion
list--Intemet)
The reason they are calling your mother a Type II is because the
first hour was 212 and the second 183. Two hours after eating is
when your sugar levels should peak out. Because her levels were
above 120 two hours after eating, she is probably a Type II. Since
her bg [blood glucose] levels were down to 11 three hours later,
it probably means that your morn has lazy organs like I do. She
may be put on reeds, then again she may be able to control it with
diet and exercise. Maybe you should go with your mona to the
doctor and try to get all the information that you both need. The
doctor probably won't mind, and will probably be glad that her
family has taken such an interest in her well being. (response to
the daughter of a recently diagnosed diabetic on the Internet
diabetes forum)
Each individual account contributes to a larger col-
lective narrative that paints a portrait of identity by
diagnosis. With good reason, researchers in the fields of
psychology, sociology, and anthropology are focusing
on the narrative approach to understanding identity and
culture in various groups. A comparison of virtual and
actual support enriches a basic understanding of support-
seeking patterns on the part of individuals coping with
health problems.
The aim in this study was to identify those kinds of
health problems that prompt higher levels of on-line par-
ticipation and to determine what similarities would exist
between virtual and actual avenues of support. Our strat-
egy, as in Study l, was not to assess the availability of all
support resources for patients (e.g., friends, family, health
care professionals, and coworkers) but to gain a broader
perspective on those types or experiences of illness that
cause patients to seek out others facing the same challenge.
Using a method similar to that used for the city surveys, we
measured support levels for the 20 studied conditions by
counting all contributions to on-line forums by participants
February 2000 • American Psychologist
o v e r a t w o - w e e k period. 3 T w o on-line d o m a i n s w e r e cho-
sen for monitoring: Internet n e w s g r o u p s and their A m e r i c a
Online ( A O L ) counterparts, bulletin boards. F r o m those
two venues, 37 virtual support groups w e r e identified and
m o n i t o r e d for a period o f two c o n s e c u t i v e weeks. 4
The Internet, at the time o f the study, featured o v e r
40,000 newsgroups, with dozens o f new ones being gen-
erated daily. T o d e t e r m i n e which groups to monitor, we
c o m p i l e d a master list o f all existing English language
n e w s g r o u p s (nationwide and/or international only, not re-
gional) and then used k e y w o r d searches for each disease
category. F r o m the results of those searches, we could
identify all groups that were d e e m e d to be mutual support
groups as o p p o s e d to groups that w e r e forums for lobbyists,
researchers, or other expert exchange. If m o r e than one
Support group existed, we selected the group exhibiting the
highest v o l u m e of e x c h a n g e o v e r a two-day pilot period.
The c o m m e r c i a l nature o f A O L ' s organization distin-
guishes it to a large degree f r o m the groups found on the
Internet. Groups present on A O L ' s " H e a l t h C h a n n e l " re-
flect the c o m m e r c i a l or public service goals of large groups
that w o u l d like to have a presence on that server (e.g.,
Arthritis Foundation, A m e r i c a n L u n g Association, various
p h a r m a c e u t i c a l companies). The pattern of e m e r g e n c e o f
n e w s g r o u p s (or bulletin boards as they are k n o w n on A O L )
is just the opposite of that of the larger Internet. Internet
groups are o r g a n i z e d by interested individuals, so no sys-
tematic c o m m e r c i a l m o t i v e s are part o f that agenda. This
m a k e s for a source of unique patterning in the groups on
A O L and the Internet, which is discussed more fully later.
The raw totals and adjusted indices of support, along
with their ranks, are featured in Table 3. In those cases in
which a condition lacked a forum, the n u m b e r of posts was
r e c o r d e d as zero. Posts on A O L o v e r the t w o - w e e k period,
for all groups studied, totaled 2,043, whereas posts on the
larger Internet totaled 5,440. The highest rates o f activity
o b s e r v e d on A O L w e r e for multiple sclerosis, f o l l o w e d by
diabetes, and then by depression. The highest rates of
activity on the Internet were for chronic fatigue syndrome,
f o l l o w e d by diabetes, and then by breast cancer. The lowest
rates o f activity on both domains were o b s e r v e d for chronic
pain, e m p h y s e m a , and migraine. Chronic fatigue s y n d r o m e
sufferers exhibited e x t r e m e l y high activity levels on the
Internet, whereas no list was available for those sufferers
on A O L . This contrast m a y reflect the i m p e r f e c t meeting
ground b e t w e e n c o n s u m e r - d r i v e n organizational patterns
on the Internet and the p r o v i d e r - d r i v e n e v o l u t i o n o f A O L .
R a n k correlations b e t w e e n the two forums approached but
did not reach significance, r~(18) = .35, p = .13. This
finding m a k e s sense in light o f the fact that Internet and
A O L support groups reflect different user bases. U s i n g a
m e t h o d similar to that used for the city groups, we col-
lapsed the two on-line measures within each diagnostic
category and adjusted for p r e v a l e n c e in order to yield a
m o r e accurate reflection o f the relative p r e v a l e n c e of on-
line support behaviors. T h e s e figures are outlined in Table
3. R a n k correlations of prevalence-adjusted on-line and
face-to-face city indices were quite high, r~(18) = .70,
p < .01.
February 2000 • A m e r i c a n Psychologist
Several patterns e m e r g e d in the on-line forums that
distinguished them f r o m the metropolitan groups. First,
alcoholism, which d o m i n a t e d the face-to-face samples, ex-
hibited a m u c h l o w e r activity level on-line, ranking ninth
and seventh on A O L and the Internet, respectively. This is
understandable g i v e n the long tradition of actual support
e n j o y e d by A A m e m b e r s , w h o m a y find virtual support a
poor' substitute for the group experience. Second, chronic
fatigue syndrome, which did not e m e r g e as a diagnostic
category until 1988, had the highest activity level o f all of
the lnternet groups. In v i e w of the high rate of use by
multiple sclerosis sufferers as well, the on-line d o m a i n m a y
be particularly useful in bringing together those w h o suffer
from rare and debilitating conditions, in w h i c h getting
together physically w o u l d present a n u m b e r o f practical
barriers. Virtual support can be very attractive to those
w h o s e disability impairs mobility, and, m o r e striking, the
on-line c o m m u n i t y allows for anonymity. Potent social
factors like physical attractiveness, vocal characteristics,
ethnicity, and social skills are neutralized.
Despite these differences, raw counts for the two
on-line measures were positively correlated with the col-
lapsed measure of the city g r o u p s - - r ~ ( 1 8 ) = .45, p < .05,
for A O L ; r~(l 8) = .37, p = . 11, for the Internet (two-tailed
in alll c a s e s ) - - s u g g e s t i n g that an underlying construct of
support seeking does exist by diagnosis, with the different
:~From a methodological standpoint, it was unclear whether the
appropriate comparison in this domain would be the number of unique
contributors or the number of contributions in total. The resulting decision
to count contributions was a judgment call based on the goal of having an
index of active participation rather than participants. The virtual arena is
also populated with a large number of "lurkers," or individuals who read
others' posts but do not contribute. It was felt that the overall amount of
conversation was a more accurate reflection of social exchange than the
number of members. In any case, post hoc analyses of the number of
unique users per forum revealed that ranks of users per category and ranks
of contribution totals per category exhibited a correlation of .95, suggest-
ing that the different measures were comparable.
To determine which groups to monitor, we compiled a master list
of all existing newsgroups (nationwide and/or international only, not
regional) and performed keyword searches for each disease category.
From the results of those searches, all groups that were deemed to be
mutual support groups, as opposed to forums for lobbyists, doctors,
researchers, or other expert exchange, were identified. In cases in which
more than one support group was identified, the group that exhibited the
highest w~lume of exchange over a two-day pilot period was selected.
Most support groups for both physical and mental illness begin with the
prefix al:.support. For example, the arthritis support group is
alt.support.arthritis. The most notable exception to this trend was the
most active support group for breast cancer, which is breast-
cancer@morgan.usc.mun.ca. The only difficult aspect in identifying the
appropriate support forum emerged with respect to AIDS. In some senses,
although new cases reflect a change in the demographics of AIDS, it has
traditionally been strongly tied to the homosexual community. The news-
group alt.~upport.homosexual was monitored and found to have a sub-
stantial focus on the AIDS epidemic; however, to select the list as the
fbrum for support would have misrepresented a significant portion of the
discussion. The point to bear in mind is that AIDS, unlike any of the other
diseases, encompasses lifestyle and sexual issues that make its discussion
hard to contain, at least by diagnostic boundaries. Many groups were
forums in which AIDS was discussed. Sci.med.aids was chosen because of
its focus on A1DS and its lack of homosexual bias (which cannot really be
considered any advantage, merely noteworthy). A complete list of the
selected newsgroups is available from the authors.
211
Table 3
Totals of On-Line Support Contributions for AOL and the Internet by Disease, Along With Their Ranks, as Well
as Prevalence-Adjusted Indices of Overall On-Line Support Activity and Their Ranks
AOL Internet AOL and Internet

No. of No. of
Disease contributions Rank contributions Rank Adjusted ° Rank

AIDS 127 8 138 9 168.79 7

Alcoholism 71 9 270 7 12.39 11
Anorexia 61 10 251 8 294.34 4
Arthritis 156 6 77 12 3.57 15
Asthma 39 12 103 10 5.43 12
Cancer, breast 221 4 946 3 329.85 3
Cancer, colon 9 14 337 5 189.90 6
Cancer, lung 173 5 no list 18 219.54 5
Cancer, prostate 40 11 14 15 46.31 10
Chronic tatigue syndrome no list 18 1,035 1 517.50 2
Chronic pain no list 18 56 13 0.93 17
Depression 248 3 779 4 54.63 9
Diabetes 291 2 989 2 81.91 8
Emphysema no list 18 no list 18 00.0 20
Heart disease 144 7 7 16 3.55 16
Hypertension 10 13 no list 18 .18 19
Migraine no list 18 90 11 4.08 14
Multiple sclerosis 451 1 317 6 1,097.14 1
Stroke 0 18 31 14 5.17 13
Ulcer 2 15 no list 18 0.22 18
Total posts 2,043 5,440
Note. Ranks assigned indicate a ranking of 1 for highest observed values. AOL - America Online.
° On-line adjusted index was derived by computing mean total posts per condition (AOL + Internet/2), divided by prevalence, multiplied by 1,000.

avenues of support constituting a kind of method variance.
Both types of support are means by which social compa>
ison is enabled: Similar sufferers can tell their stories to
sympathetic audiences and can exchange tips about the
management of their conditions. Means of the two on-line
measures computed for each condition exhibited a strong
relation to the city patterns, r~(18) = .47, p < .05. In a
similar vein to the patterns observed in the metropolitan
groups, breast cancer, depression, and diabetes were con-
ditions in which talk (in the form of posts) occurred at
relatively high rates, whereas talk by migraine, ulcer, and
chronic pain sufferers was conspicuously absent from the
bulletin boards.
Because city group participation and on-line partici-
pation rates were significantly con'elated, we felt it would
be useful to generate one comprehensive support index,
across both on-line and actual groups, leaving out the
prevalence adjustment that so strongly shaped rankings
observed in Tables 2 and 3. Because of its extreme value,
alcoholism was excluded. To convey a simpler picture,
figures for all cancers were combined, and an overall
cardiovascular disease category was generated by combin-
ing hypertension, coronary heart disease, and stroke. Both
the population-adjusted city index (refer to Table 2) and the
mean on-line indices (Table 3) were converted into z
scores, Means of those z scores across the two support
types were computed and are depicted in Figure 1.
As is evident in Figure 1, other than alcoholics, cancer
patients exhibit the highest overall tendency to seek and
offer support. Also notable were support levels for AIDS,
diabetes, depression, and chronic fatigue. By contrast, ex-
tremely low levels of support were observed in the cases of
ulcer, emphysema, chronic pain, and migraine. Support
levels in the cardiovascular disorders, even as combined,
are only slightly higher than those for anorexia, a condition
whose prevalence is almost 1,000 times less prevalent.
Some conditions, then, are noteworthy for the amount of
talk they generate among sufferers, whereas others are no
less noteworthy in their relative silence.
If migraines are as responsive to self-help as to med-
ication, why don't migraine sufferers get together? What
kinds of suffering do not generate social comparison be-
haviors? Bearing in mind the shared and distinctive fea-
tures of the two support venues and the markedly high
consistency in tendencies to participate in support, our goal
was to uncover the motives behind such patterns.

212 February 2000 • American Psychologist


Figure 1
Summary Indices of Supportiveness Combining City
and On-LineMeasures
Note. Indices represent means of z scores associated with population-
adjusted city groups and mean (America Online and Internet) on-line totals. The
alI-CVD category was derived by combining coranary heart disease, hyperten-
sion, and stroke conditions. Alcoholism, although unquestionably the condition
characterized by the highest support levels, was excluded because its extreme
value would have suppressed all standard scores. CFS = chronic fatigue
syndrome; MS = multiple sclerosis; CVD = cardiovascular disease.
Study 3: Identifying Correlates of
Support Seekmg
Social comparison theory asserts that affiliative behaviors
increase in times of uncertainty or anxiety. The difficulty in
predicting behaviors on the basis of such considerations is that
the experience of anxiety issues from myriad causes--embar-
rassment, threat to life, pain, anticipated pain, financial bur-
den, and so on. Our goal in this study was to identify sources
of anxiety relevant to patients' experiences of illness in order
to understand which, if any, of these sources would be related
to support seeking. Three broad areas of investigation served
as the conceptual foundation of our inquiries: patient charac-
teristics, (both psychological and demographic), health care
burden (aspects of illness that increase medical care usage),
and social burden (the interpersonal dimensions of the illness
experience). The goal was to clarify whether support seeking
was primarily motivated by personal, social, or health-care-
induced anxiety,
A number of personal factors are involved in care-
seeking behaviors. For instance, some segments of the
population exhibit higher levels of concern about their
health status. The elderly and women, for example, use
health care services more extensively than do young adults
and men (Aday & Andersen, 1974). AA participants with
unsupportive social networks exhibit higher adherence lev-
els to the self-help program (Tucker, 1995). Individual
February 2000 • American Psychologist
differences, such as sociability or commitment to work, are
also related to support seeking. Prior studies of self-help
participants found that they have higher levels of social
skills than do their nonparticipating counterparts (Taylor,
Falke, Shoptaw, & Lichtman, 1986). One question, then,
was the degree to which some patient groups would be
perceived as generally friendlier than others.
A second consideration centered on illness factors that
determine patients' perceived self-care needs and treat-
ment-adherence patterns. The most widely cited model of
patients' schemas about illness suggests that such con-
structs fall into five dimensions: illness identity, time
course, consequences, causal models, and likelihood of
cure (Leventhal, Meyer, & Nerenz, 1980). Some items
characterized under headings of health care salience and
patient burden were generated to capture the anxiety-
relevant aspects of illness and treatment consistent with
Leventhal et al.'s model of illness schemata. Examples
include degree of life threat, ideas about cause, burden on
patient to manage, and social stigma.
Finally, some items were included to assess the inter-
personal impact of a given illness category. Examples
include embarrassment, stigmatization, and disfigurement.
We hypothesized that social anxiety might be assuaged
through support group participation.
Judges' ratings were obtained from four health care
experts indicating, on a scale of 1 to 7, to what degree each
description accurately depicted each of the 20 diseases or
the patients that suffered from them. All judges had at least
a master s degree level of training in nursing, public health,
or medicine. One rater was an MD; the three other raters
(one Phi) in nursing, one MS in nursing, and one MS in
public health) were nurses on faculty at a large teaching
hospital. Overall, agreement among the judges was ade-
quately high, yielding a mean alpha of .72 across the 16
items. We averaged each rating across judges and com-
puted subsequent correlations with support levels to iden-
tify which factors, if any, were associated with support
seeking.
As shown in Table 4, having an illness that is embar-
rassing, socially stigmatizing, or disfiguring leads people to
seek the: support of others with similar conditions. The
most striking associations, then, between support seeking
and illness features centered around the interpersonal con-
sequences of illness: Conditions described as embarrassing,
socially stigmatizing, and disfiguring were associated with
higher levels of support of both kinds. In fact, embarrass-
ment around discussion of the illness emerged as the stron-
gest association with support of both kinds. This pattern of
results, on the surface, conflicts markedly with prior work
in social comparison theory. Embarrassment has been
widely presumed to weigh against affiliative behavior
(Sarnoff & Zimbardo, 1961; Teichman, 1973), yet alien-
ation from one's usual support network may be precisely
the kind of social anxiety that in turn increases the value of
the mutual support context.
Aspects of illness associated with cost of treatment
and loss of life were positively associated with support
seeking in the city groups but not in the on-line forums.
2l3
 support groups between cities and (b) the tendencies ot
Table 4 patients with certain conditions to seek or not to seek
Correlates of Mutual Support: Patient, Illness, and support, particularly after adjusting for prevalence.
Social Aspects Comparable patterns emerged in the on-line domain.
Amount of on-line Amount of mutual In that case, a modest relationship was found between the
Rating item support activity (ranked) support (ranked) two venues of virtual social exchange. The highest levels of
support activity were exhibited by sufferers of multiple
Health care salience sclerosis, chronic fatigue syndrome, breast cancer, and
Cause known -.37 -.14 anorexia. The lowest activity levels were observed for
Patient must manage - . 13 -.01 sufferers of chronic pain, ulcer, hypertension, and emphy-
How disabling .04 .01
Costliness of sema. On-line and face-to-face support patterns were sig-
treatment .22 .52* nificantly correlated, suggesting that broad tendencies to
Attitude importance .40* .34 seek support do vary by diagnostic category. Standardized
A terminal illness .18 .38** measures of face-to-face and on-line support, when col-
Patient characteristic lapsed, revealed a broad picture of variations in support
Related to aging -.32 .10 seeking by diagnosis. Alcoholism, cancer, AIDS, depres-
Woman's illness .29 -.04 sion, and diabetes are conditions that have given rise to the
Friendlypatients -.31 -.13 formation and maintenance of mutual support forums.
Social burden These tendencies, according to correlations with
Disfiguring .34 .51 * judges' ratings, may be spurred by the interpersonal con-
Embarrassing .43" .67*
sequences of illness, specifically, embarrassment, stigma,
Symptoms
noticeable - . 17 -.14 and disfigurement. Actual groups were unique in their
Constitutes a stigma .30 .43** association with conditions rated as terminal and costly to
treat. This kind of seriousness index was not evident in
Note. Ranks of support behaviors and publications reflect that a rank of 1 is motivating on-line support. On-line forums appear to be
assignect to the lowest value, and higher rank values are assigned to higher
variable values. For all figures, n = 20. slightly more oriented around conditions poorly understood
* p < .05. ** p < .10 (two-tailed, all correlations are based on the Spear- and somewhat overlooked by the medical community.
man rank correlation coefficient). Those confronted with grave concerns may feel more
acutely the need to experience the physical presence, the
validation, and the sense of belonging that come with
actual encounters. Virtual support may have its limits.
According to social comparison theory, in conditions
The threat of death and the experience of costly medical of uncertainty or anxiety, affiliative motivation should be
treatment are the two factors most highly correlated with increased, except under conditions of embarrassment. In
participation in face-to-face support groups. The personal particular, individuals would prefer the company of others
features grouped as patient characteristics showed no sig- who are in a similar situation when experts are absent.
nificant relations to support behaviors of any kind. None of These postulations are fitting in the case of the on-line
the personal characteristics we considered--friendliness of forums, in which a relative amount of anonymity is present,
patients, illnesses directly related to aging, or illnesses that and thus confiding can occur without immediate social
strike women--were associated with support seeking. This repercussions. In the case of the city support groups, how-
result stands in contrast to earlier findings by Taylor et al. ever, hypotheses about lowered support seeking in condi-
(1986) reporting higher participation rates by women. On- tions associated with embarrassment were pointedly dis-
line support was unique in its correlation with importance confirmed. These groups are populated by individuals
of patient's attitude to outcome. This pattern of findings whose illnesses, either by their very nature or as a result of
suggests that virtual support occurs at higher rates among treatment (e.g., mastectomy), have forced them to experi-
patients whose conditions, although not necessarily life ence embarrassment and social stigmatization. The serious-
threatening, are debilitating in ways less responsive to ness of their conditions, the weight of their illness impact,
purely medical care. and the degree of readjustment required under the circum-
stances suggest motives only partially captured by terms
Discussion like ambiguity and anxiety. In these cases, the patients'
The aim of these studies was to identify patterns of patient experiences set them apart from their immediate social
support, both in face-to-face encounters and through the setting and propel them toward others who have been
use of on-line options. The tendency to participate in sup- similarly marked.
port groups was highly consistent by category across cities. The findings surrounding support-group participation
The highest levels of support were found in the cases of suggest that laboratory investigations of social motives lack
alcoholism, AIDS, breast cancer, and anorexia, and the some ecological validity: It is not a great challenge for a
lowest levels of support were found in the cases of hyper- student to anticipate the experience of emban'assment and
tension, migraine, ulcer, and chronic pain. Wide disparities decline the opportunity to share this with another in a
were observed in (a) the overall tendency to participate in highly controlled lab study. The experience of embarrass-

214 February 2000 • American Psychologist

ment in everyday life, however, carries with it the ongoing
burden of a barrier that results in chronic emotional arousal
accompanied by the suppressed desire to talk with others.
Such social barriers represent a significant health risk
(Cole, Kemeny, Taylor, & Visscher, 1996; Pennebaker,
1997).
On the basis of the observed patterns, anxiety is a
necessary but not sufficient condition for support seeking.
Heart attack survivors, for example, report high levels of
anxiety in the aftermath of their attacks, feeling as though
their lives could have suddenly ended and may end in the
near future (Cay, Vetter, Phillip, & Dugard, 1972), yet
cardiac patients are not seeking to share feelings with
others (ranking 15th in adjusted support). The marked lack
of support in heart disease patients is consistent with prior
observations of personality profiles associated with that
condition: The Type A construct portrays a hostile, inter-
personally suspicious, hurried, or depressed temperament
ill-suited to the kinds of disclosure that are standard in
support groups (for a review, see Smith, 1992).
Conversely, the profile of the cancer-prone personality
is one ideally suited to a support forum because of its
characterization as friendly, cooperative, and tending to
suppress expression of negative emotion (Greer & Watson,
1985; Temoshok, 1987). It is noteworthy that, although
more women will die from heart disease every year, and
although more women are currently living with a history of
heart disease, our findings indicate that breast cancer pa-
tients participated in support groups at rates 40 times higher
than all heart disease patients, male or female.
Directions for Future Research
It is interesting that the striking differences in support
seeking between heart disease and cancer patients harken to
some of the traditional personality-situation debate. As the
Type A and cancer-prone constructs suggest, certain fea-
tures of a "disease personality" may dictate how individu-
als avoid or seek out others. Conversely, the very nature of
the disease may differentially threaten friends of the person
with the disease. For example, one's friendship network
may be more frightened or helpless in hearing about cancer
or AIDS than in learning that one of their own has high
blood pressure or has suffered a heart attack (cf. Bolger,
Foster, Vinokur, & Ng, 1996; Petrie, Weinman, Sharpe, &
Buckley, 1996).
One way to gain insights about the validity of contro-
versial profiles such as Type A and the cancer-prone per-
sonality would be through closer analysis of linguistic
styles revealed in on-line forums. Disclosure styles have
been linked to markers of immune function and cardiovas-
cular reactivity (Christensen & Smith, 1990; Kamen-
Siegel, Rodin, Seligman, & Dwyer 1991). For example, a
recent study of on-line disclosure styles found that consis-
tent with prior observations of cancer patients (e.g.,
Watson, Pettingale, & Greer, 1984), virtually no negative
emotion was expressed in such forums (Davison & Penne-
baker, 1997). Studies of disclosure styles associated with
different patient populations will make a significant con-
February 2000 • American Psychologist
tribution toward the psychosomatic literature focused on
the relationship between personality and illness.
Another obvious extension of the research would fo-
cus on elucidating distinctions in disclosure styles between
those on-line groups that include an authority figure and
those that do not. Our results suggest that conditions lack-
ing some medical legitimacy are associated with higher
participation levels. As the virtual community evolves, it
may soon be possible to distinguish between the kinds of
discourse that occur among patients that also have a doctor
present and the kinds of exchanges occurring when such an
authority figure is lacking. Social comparison would indi-
cate that group cohesion is higher when the authority is
absent. Changes in disclosure styles when authority figures
are present may shed valuable light on what patients feel
able to discuss with doctors and what they can't.
Another direction for future research concerns the
relationship between Internet support participation and
other indices of health and health-related behaviors. Recent
studies have reported that higher rates of Internet use were
associated with poorer measures of social adjustment, in-
cluding loneliness and smaller social circles (Kraut et al.,
1998). A constant concern in social-support research has
been the distinction between support network size and
support network quality (Stroebe & Stroebe, 1996). Inter-
net forums represent a new hybrid for participants in which
size is large but quality is largely unknown. People suffer-
ing from illness benefit from social support. If Internet use
is a substitute for actual emotional support, then patients
may be engaging in shallow forms of exchange when more
substantial ties could be built face-to-face, to the peril of
both psychosocial and physical well-being. Comparisons
between virtual and actual social behaviors will provide
insights about how the Internet contributes to the configu-
ration of individuals' social worlds.
Implications for Policy and Practice
As with other correlational designs, the results obtained in
this broadly based analysis of patient support groups are
suggestive, not conclusive. Although strong correlations
were found between indices of social marginalization (em-
barrassment, disfigurement, stigma, and life threat) and
support-group participation levels, the reality of participa-
tion motives is most likely a more complex interaction of
illness, individual differences, and cultural norms. Founda-
tions, hospitals, patients, and doctors can significantly alter
the viability of support groups. The investigations convey
only a part of the picture. They do, however, indicate the
value of attempts to observe, in an ecologically valid way,
the social patterns of individuals who may not be aware
that they or their activities are part of a larger movement.
The value of the approach lies in observing social be-
haviors, even collective ones, from an epidemiological
perspective.
An aspect of the support group picture that needs
additional attention is analysis of the character and purpose
of the groups. Prior reports on other kinds of support
groups indicate that group purposes frequently evolve over
time into watchdog organizations, lobbying arms, or fund-
215
raising outlets (Chesler & Chesney, 1995). When a support
group converts from one of emotional and informational
support into one of social activism, what does it do for
members? The lifeline of member participation and group
life in general is highly variable. As noted earlier, A A
membership is a lifelong one. Other groups are more tem-
porary in orientation. As with illnesses, people have
schema about the duration of support needs. For psychol-
ogists, the dynamics of these groups over time offer in-
sights about social theory and about patient needs.
The self-help movement, both in face-to-face and vir-
tual arenas, has tremendous therapeutic potential. Appar-
ently, relative increases in social marginalization result in a
desire to compare notes with similar others, despite the
experience of embarrassment. As such, it may be possible
to draw on the cohesive power that stigma apparently
generates. Recruitment efforts for rehabilitation programs
or health behavior change may be more successful if ref-
erence to one or more of the stigmatizing aspects of a
condition is made. Adherence over time may be increased
if medical programs integrate more opportunity for ex-
change around the human side of health and illness. Out-
side health care, other marginalized groups, such as immi-
grants, may be more effectively integrated with a modest
push toward group involvement.
Although self-help and professional help are often
perceived as mutually exclusive, the data indicate that such
perceptions are misleading: Over 60% of groups describing
themselves as self-help are professionally facilitated. This
apparent contradiction should be more fully explored.
Group participants may not be resistant to professional
input; rather, they may need to speak and be heard about
issues not addressed within the health care setting. Several
important questions emerge from the lay and professional
perspectives on mutual support. What factors are important
to patients helping patients, as opposed to professionals
helping patients? What features of professional input con-
tribute to optimal outcomes? What combinations are opti-
mal for various patient groups? What are the cost benefits
associated with group approaches to health care? Physician
enlistment could double or triple support group participa-
tion rates, and psychological expertise could heighten ef-
fectiveness in emotional support, health behavior change,
and patient prognosis. The clinical potential of group sup-
port in combination with professional guidance is largely
unknown.
Within contemporary health care, support group par-
ticipation patterns are particularly relevant at this juncture
in history given the disparity between the costs of health
care and the unmet needs of many patients. The culture of
institutional health care is still far from embracing psycho-
logical support into health care delivery, despite over-
whelming empirical justification (for an excellent commen-
tary on these issues, see Morgeson, Seligman, Sternberg,
Taylor, & Manning, 1999). On the other hand, the culture
of psychology has been focused around treatment of men-
tal, not physical, ailments. Support groups represent one of
the most adaptive permutations of the health care dilemma,
allowing sufferers to gather the information most valuable
216
to t h e m - - s t o r i e s of similar experience and endurance.
These investigations indicate that support groups are par-
ticularly valued by individuals whose lives and social iden-
tities have been put at risk. A fuller appreciation of the
social context of illness enriches our theoretical under-
standings of social support and social comparison, while
offering practical insights about a more appropriate match
between health care delivery and the health care sought by
patients and their families.
Both the methods and results of these studies have
relevance for a number of concerns common to all psy-
chologists: social comparison, group formation, health
care, effects of stigma, and disclosure processes. The most
interesting aspect of the study is also the most problematic:
A constant concern for researchers is that balancing act of
conducting research that draws on prior theoretical and
empirical foundations while asking questions interesting
enough to have real-world relevance. Support groups con-
stitute a category with fuzzy boundaries, and as such they
make scientists uneasy. In the interest of elegance and
experinaental control, we often prefer mutually exclusive
categories and singular causal models. The goal is usually
to isolate the hypothesized active agent in a particular
phenomenon. At broader levels of analysis, though, such
approaches lack validity, ignoring the realities of a society
in which richly overlapping categories are everywhere. In
this case, the most rigorous methodology was also a less
controlled one. Support groups cannot be replicated in the
lab, but the tendency of some types of patients to seek each
other's company more than other types of patients, despite
the noise in the imperfect categorizations, emerges statis-
tically as a clear pattern replicated across cities. We believe
that psychologists should be encouraged to err more often
on the side of real-world complexity in order to serve more
effectively the society they attempt to observe.
REFERENCES
Aday, L A., & Andersen, R. (1974). A framework ~br the study of access
to medical care. Health Services Research, 9, 208-220.
Bolger, N., Foster, M., Vinokur, A. D., & Ng, R. (1996). Close relation-
ships and adjustments to a life crisis: The case of breast cancer. Journal
()f"Personality and Social Psychology, 70. 283-294.
Bradley, L. A.. Young, L. D., Anderson, K. O., Turner, R. A., Agudelo,
C. A.. McDaniel, L. K., Pisko, E. J., Semble, E. L., & Morgan, T. M.
(1987). Effects of psychological therapy on pain behavior of rheuma-
toid arthritis patients: Treatment outcome and six-month follow-up.
Arthritis and Rheumatism, 30, 1105-1114.
Cay, E. L., Vetter, N. J., Phillip, A. E., & Dugard, P. (1972). Psycholog-
ical status during recovery from an acute heart attack. Journal oJ
Psychosomatic Research, 16, 425-435.
Centers for Disease Control and Prevention. (1994). The facts about
chronic.[?ttigue syndrome. Retrieved June 1996 from the World Wide
Web: http://www.cdc.gov/ncidod/diseases/cfs/cfs_infoz.htm
Centers for Disease Control and Prevention. (1996). Chronic fatigue
syndrome demographics. Retreived June 1996 from the World Wide
Web: http://www.cdc.gov/ncidod/diseases/cfs/cfs_info2.htm
Chesler. M. A., & Chesney, B. K. (1995). Cancer and self-help: Bridging
the troubled waters of childhood illness. Madison: University of Wis-
consin Press.
Christensen, A. J., & Smith, T. W. (1990). Cynical hostility and cardio-
vascular reactivity during self-disclosure. Psychosomatie Medicine, 55,
193-21/2.
Cole, S. W., Kemeny, M. E., Taylor, S. E., & Visscher, B. R. (1996).
February 2000 • American Psychologist
 Elevated health risk among gay men who conceal their homosexual
identity. Health Psychology, 15, 243-251.
Davison, K. P., & Pennebaker, J. W. (1997). Virtual narratives: Illness
representations in online support groups. In K. J. Petrie & J. Weinman
(Eds.), Perceptions of health and illness: Current research and appli-
cations (pp. 463-486). London: Harwood Academic Press.
DiClemente, C. C. (1995). Self-efficacy and addictive behaviors. In J. E.
Maddux (Ed.), Self-efficacy, adaptation and adjustment (pp. 109-141 ).
New York: Plenum Press.
Dracup, K. (1985). A controlled trial of couples' group counseling in
cardiac rehabilitation. Journal ~( Cardiopulmona~, Rehabilitation, 5,
436-442.
Droge, D., Arntson, P., & Norton, R. (1986). The social support function
in epilepsy self-help groups. Small Group Behavior, 17, 139-163.
Fawzy, F. I., Fawzy, N. W., Hyun, C. S., Elashoff, R., Guthrie, D., Fahey,
J. L., & Morton, D. L. (1993). Malignant melanoma: Effects of an early
structured psychiatric intervention, coping, and affective state on re-
currence and survival six years later. Archives qf General Psychiatry,
50, 681-689.
Ferguson, T. (1996). Health online. New York: Addison-Wesley.
Festinger, L. A. (1954). A theory of social comparison processes. Human
Relations, 7, 117-140.
Gergen, K. J., & Gergen, M. M. (1997). Narratives of the self. In L. P.
Hinchman & S. K. Hinchman (Eds.), SUNY series in the philosophy of
the social sciences: Vol. 33. Memo~, identity, communi~: The idea of
narrative in the human sciences (pp. 16l 184). Albany: State Univer-
sity of New York Press.
Gould, R. A., & Clum, G. A. (1993). A meta-analysis of self-help
treatment approaches. Clinical Psychology Review, 13, 169-186.
Greer, S., & Watson, M. (1985). Toward a psychological model of cancer:
Psychological considerations. Social Science and Medicine, 20, 773-
777.
Jakobs, M. K., & Goodman, G. (1989). Psychology and self-help groups:
Predictions on a partnership, American Psychologist, 44, 536-545.
Kamen-Siegel, L., Rodin, J., Seligman, M. E., & Dwyer, J. (1991).
Explanatory style and cell-mediated immunity in elderly men and
women. Health Psychology, 10, 229-235.
Kessler, R. C., Mickelson, K. D., & Zhao, S. (1997). Patterns and
correlates of self-help group membership in the United States. Social
Policy, 27, 27-46.
Kleinman, A. (1995). Writing at the margin: Discourse between anthro-
pology and medicine. Berkeley: University of California Press.
Kraut, R., Patterson, M., Lundmark, V., Kiesler, S., Mukopadhyay, T., &
Scherlis, W. (1998). Internet paradox: A social technology that reduces
social involvement and psychological well-being? American Psychol-
ogist, 53, 1017-1031.
Kroenke, S. R., Wood, D. R., Mangelsdorff, A. D., Meier, N. J., & Powell,
J. B. (1988). Chronic fatigue in primary care: Prevalence, patient
characteristics, and outcome. Journal of the American Medical Asso-
ciation, 20, 929-934.
Leventhal, H., Meyer, D., & Nerenz, D. (1980). The common sense
representation of illness danger. In S. Rachman (Ed.), Contributions to
medical psychology (pp. 7-30). New York: Pergamon Press.
Lieberman, M. A., & Snowden, L. R. (1993), Problems in assessing
prevalence and membership characteristics of self-help group partici-
pants. Journal of Applied Behavioral Science, 29, 166-180.
Lorig, K., Mazonson, P. D., & Holman, H. R. (1993). Evidence suggesting
that health education for self-management in patients with chronic
arthritis has sustained health benefits while reducing health care costs.
Arthritis and Rheumatism, 36, 439-446.
National Institutes of Health. (1995). Diabetes overview [on-line]. Re-
February 2000 ° American Psychologist
trieved June 1996 from the World Wide Web: http://www.niddk.nih.
gov/health/diabetes/diabetes.htm
Morgeson, E. P., Seligman, M. E. P., Sternberg, R. J., Taylor, S. E., &
Manning, C. M. (1999). Lessons learned from a life in psychological
science: hnplications for young scientists. American Psychologist, 54,
106-116.
National Center for Health Statistics. (1992). Vital statistics of the United
States, 1992. Washington, DC: U.S. Government Printing Office.
National Center for Health Statistics. (1993). Health, U.S., 1992. Hyatts-
ville, MD: Public Health Service.
National Center for Health Statistics. (1996). U.S. Bureau of Census. July
population.figures. Retrieved September 20, 1996 from the World Wide
Web: http://venus.census.gov/cdrom/lookup
Ornish, D., Brown, S. E., Scherwitz, L. W., Billings, J. H., Armstrong,
W. T., Ports, T. A., McClanahan, S. M., Kirkeide, R. L., Brand, R. J.,
& Gould, K. (1990). Can lifestyle changes reverse coronary artery
disease? The lifestyle heart trial. Lancet, 336, 129-133.
Pennebaker, J. W. (1997). Opening up: The healing power of expressing
emotion New York: Guilford Press.
Petrie, K. J., Weinman, J., Sharpe, N., & Buckley, J. (1996). Role of
patients" view of their illness in predicting return to work and func-
tioning after myocardial infarction: Longitudinal study. British Medical
Journal, 312, 1191-1194.
Rappaport, J. (1993). Narrative studies, personal stories, and identity
transformation in the mutual help context. Journal of Applied Behav-
iorcd Science, 29, 239-256.
Riessman, F. (1985). New dimensions in self-help. Social Policy, 15, 2-4.
Sarnoff, I., & Zimbardo, P. G. (1961). Anxiety, fear, and social facilita-
tion. Jom'nal of Abnormal and Social Psychology, 62, 597-605.
Schachter, S. (1959). The psychology of affiliation. Stanford, CA: Stanford
University Press.
Smith, T. W: (1992). Hostility and health: Current status of a psychoso-
matic hypothesis. Health Psychology, I1, 139-150.
Spiegel, D., Bloom. H. C., Kraemer, J. R., & Gottheil, E. (1989). Effect
of psychosocial treatment on survival of patients with metastatic breast
cancer. Lancet, 8668, 888-891.
Stroebe, W., & Stroebe, M. (1996). The social psychology of social
support In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychol-
ogy: Handbook of basic principles (pp. 597-621). New York: Guilford
Press.
Taylor, S. E. (1995). Health psychology. New York: McGraw-Hill.
Taylor, S. E., Falke, R. L., Shoptaw, S. J., & Lichtman, R. R. (1986).
Social support, support groups, and the cancer patient. Journal oJ
Consulting and Clinical Psychology, 54, 608-615.
Teichman, Y. (1973). Emotional comparison and affiliation. Journal oJ
Experimental Social Psychology, 9, 591-605.
Telch, C. F., & Telch, M. J. (1986). Group coping skills instruction and
supportive group therapy for cancer patients: A comparison of strate-
gies, ,l~,,urnal of Consulting and Clinical Psychology, 54, 802-808.
Temoshok, L. (1987). Personality, coping style, emotion, and cancer:
Towards an integrative model. Cancer Surveys, 6, 545-567.
Tucker, J A. (1995). Predictors of help-seeking and the temporal rela-
tionship of help to recovery among treated and untreated recovered
problem drinkers. Addiction, 90, 805-809.
U.S. Bureau of the Census. (1996). Statistical abstract of the United
States: 1995. Washington, DC: U.S. Government Printing Office.
Watson, M., Pettingale, K. W., & Greer, S. (1984). Emotional control and
autonomic arousal in cancer patients. Journal of Psychosomatic Re-
search, 28, 467-474.
Yalom, I. D. (1995). The theory and practice of group psychotherapy.
New York: Basic Books.
217