The Health Belief Model

and Preventive Health Behavior

Irwin M. Rosenstock, Ph.D.

School of Public Health, University of Michigan

The Health Belief Model was originally formulated to explain

(preventive) health behavior. As defined by Kasl and Cobb,’ health
behavior is “any activity undertaken by a person who believes himself
to be healthy for the purpose of preventing disease or detecting disease
in a n asymptomatic stage.” This is in contrast with illness behavior,
defined as “any activity undertaken by a person who feels ill, for the
purpose of defining the state of his health and of discovering suitable
remedy,” and sick-role behavior, “the activity undertaken by those
who consider themselves ill for the purpose of getting well.”
The present paper is confined to the first of these areas - health
behavior.
I t should first be noted that the three modes of behavior are not
discontinuous. Hardly anyone can be found who, upon intensive
questioning, would report himself free of all symptoms. Similarly, the
edges between illness behavior and sick-role behavior are blurred.
Nevertheless, the distinctions are valuable because they refer to modal
mental states which help to account for behavior.
STUDIES OF HOW PEOPLE
USE HEALTH SERVICES
Consideration may first be given to the relationship between studies
of how preventive health services are used and a n understanding of
why they are used. Do studies of how people use services explain why
people use health services? I n approaching a n answer to this question,
a careful distinction should be drawn between studies of utilization
whose findings are intended to have immediate application and
studies of utilization which are intended to serve as means to still
other research ends. In the first case, information is sought to serve as
a basis for formulating and implementing public policy in the health
area. Utilization data obtained for such purposes have proved
invaluable in the health
Utilization studies undertaken as means to achieve the broader aim
of increased understanding of why services are used, however, have
generally failed to accomplish their purpose. Little can be learned
from these studies about why people use or fail to use certain services.
Evidence in support of this conclusion has been drawn from studies of

354 Health Education Monographs VOL. 2, NO. 4

high a n d low users of free medical examinations,J detection tests for
cervical cancer,6 polio i m m ~ n i z a t i o n ,dental
~ services,*>.'Jphysicians'
services,J.Luhospital services,1i and from studies of the characteristics
of those who do and those who do not delay in seeking diagnosis a n d
treatment of cancer.1."

Dem ographic Varia tio m

Analyzing the major findings of studies on the patterns of use of
preventive a n d detection services permits c e r t a i n s u m m a r y
generalizations about the association of personal characteristics with
the use of services. I n general, such services a r e used most by younger
or middle-aged people, by females, by those who a r e relatively better
educated and have higher income (though perhaps not the very highest
levels of education a n d income). Striking differences may nearly
always be found in acceptance rates between whites a n d non-whites,
with whites generally showing higher acceptance rates, although
occasional exceptions occur.
A review of the previously cited d a t a on utilization of diagnostic a n d
treatment services suggests a pattern quite similar to t h a t obtained in
connection with preventive a n d detection services. I n general, more
females than males visit the physician a n d the dentist a n d incur
hospitalization, even when hospitalization for pregnancy is excluded.
Higher socioeconomic groupings (defined in terms of educational a n d
income level) a r e also more likely to obtain medical, dental, a n d
hospital services, although the associations between income a n d
utilization a r e becoming less marked.'..$
With reference to race, whites show much higher utilization rates
than non-whites in all three utilization categories (physician visits,
dental visits, a n d hospitalization).
T h e nature of the association between age a n d utilization of
treatment services is generally different from t h a t found between age
and seeking preventive a n d detection services, probably reflecting the
effect of objective medical a n d dental need.
With respect to characteristics of those who delay in seeking
diagnosis a n d treatment of cancer, similar patterns emerge. I n
general, persons who delay a r e older, of low educational status and, a t
least in some studies, males."
I t h a s recently been suggested t h a t there is a n increasing rate of
health supervision visits; t h a t is, visits to practitioners i n the absence
of symptoms. A question may be raised as to whether such increases, if
indeed they a r e occurring, show the typical social class gradient, with
those of higher income accounting for most of the increase. There a r e
no definitive d a t a on the subject but inferences can be drawn from a
combination of findings from several sources. Herman 1J notes, a s have
many others, t h a t while higher rates of disease a n d mortality still

Health Education Monographs Winter I974 355

persist among those with very low incomes, their frequency of visiting
the physician is considerably lower than for the more affluent,
healthier group. I t is commonly known that even when immunizations
are free, higher income families show a much better rate of protection
than do poorer families; moreover, while ambulatory services
generally show a lower utilization rate by lower income households,
poor people are overrepresented among hospital patients; their
hospitalization rates are as high as, or higher than, those of upper
income levels and their length of stay longer on the average. This
probably reflects the failure to receive treatment a t earlier stages of
disease and disability.

Attributing Cause to Demographic Correlates

While some of these data are not as recent as one would like, there
seems to be no reason for concluding that the poor, as yet, are showing
any marked increase in their likelihood of seeking early health care
compared to their prior behavior or to the behavior of the more
affluent. This is not to say that the removal of economic and social
barriers will not increase the use of health services; indeed, they may
well do so. In a n experiment reported by Alpert, et aI,l4 it was shown
that after exposure to comprehensive, personalized health care, low
income families (i.e., median income of $4,100) became more satisfied
with the services received, reported a n increased likelihood of using a
family doctor or pediatrician for selected medical problems of
children, and reported a greater likelihood of using the telephone as a
first contact. Nevertheless, it is still questionable whether such
attitudinal changes will result in patterns of use of health services that
are like those of the more affluent.
Several groups deny that there is any question; for them differences
in utilization patterns are entirely explainable by income differences.
A report by Colombo et al, is a good e ~ a m p l e . ~Ins 1967, a poverty
group, supported by the Office of Economic Opportunity, was admitted
into the Portland Region of the Kaiser Foundation. During a period of
nearly a year of observation, their use of medical services (3.9
encounters per person) was reportedly remarkably close to that of a
sample of the overall health plan membership (4.1 encounters per
person), and, with minor exceptions, the nature of the visits were quite
similar a s indicated by the most frequent diagnoses made by
physicians. The authors conclude that a strong emphasis on preventive
care exists both in the general health plan population and in the
O.E.O. population. The basis for this important conclusion is worthy of
detailed examination.
A close examination of methods of selection of poor families into the
plan and the mode of analysis casts some doubt on the conclusion that
the poor do in fact exhibit a strong emphasis on preventive care. The

356 Health Education Monographs VOL. 2, NO. 4

target area specified “was composed of neighborhoods that had a
majority of families who were not defined as poor. These areas
contained approximately 4,000 poor families.” In short, the poor
population selected represented a minority of the residents of the
neighborhoods from which they were drawn. The authors also indicate
that since many of the neighbors of the poor families selected were
already members of the Kaiser Health plan, the program was already
familiar to the indigent residents. We do not know whether the desire
for and utilization of health services by poor, urban people living in
predominantly non-poor neighborhoods are typical of the nation’s
poverty population. No information is available on the prior
utilization rates of the poor families.
Furthermore, since only 1,200 of the 4,000 identified poor families
could be served, a number of selection priorities were imposed of
which the two most important were (1) large families with small
children and ( 2 ) families with known acute health problems, but with
no existing medical care source. One wonders whether these selection
priorities did not almost guarantee greater utilization of services than
had all 4,000 poor families been included.
Selection priorities to include those in ereatest need are quite
sensible in terms of health need, but the procedure casts doubt on the
comparability of the O.E.O. group with a random sample of the total
health plan membership. If the random sample of health plan
membership included persons less in need, then their slight superiority
over the O.E.O. population in total patient encounters masks a
tendency for the general population to seek preventive services much
more frequently than the poverty population. In the Portland-Kaiser
study, children under six, from O.E.O. families, had 4.2 encounters
with physicians while in the general membership the number of such
encounters was 6.1. Colombo et a1 report a utilization rate of well-
baby and child care of 235.1 per thousand for the O.E.O. population
and only 159.3 per thousand for the general health plan membership,
but these figures are very misleading as there were nearly twice as
many children under six in the O.E.O. population a s in the general
health plan membership. Taking this discrepancy into account, one
might have anticipated a utilization rate in the O.E.O. population for
well child care of more than 300 per thousand.
The authors also present data on utilization of immunizations which
apparently shows substantially more utilization by the O.E.O.
population than by the health plan membership. Once again these
conclusions are derived from total patient encounters without allowing
for the overweighting of children in the O.E.O. group. Since the four
immunizations reported are primarily intended for those under six
years of age, recomputation of the data using as the denominator the
total number of children under age six in each of the two groups

Health Education Monographs Winter I974 35I

revealed that the immunization experience of the two groups was
virtually identical. However, since it is reasonable to believe that the
O.E.O. population had been under-doctored prior to their enrollment
in the Kaiser plan, if finances and organization were all important one
might have expected a much higher rate of immunization among the
poverty group than in the health plan membership whose children had
been seen more or less regularly during the period preceding the study.
In short, the O.E.O. group probably did not make preventive visits as
frequently as the more affluent, when differences in the numbers of
children are taken into account and when the apparently greater need
in the O.E.O. population is considered.
In summary, this report that purports to show that utilization rates
of health services by poor people increases to the levels of the non-poor
under proper organizational and prepayment patterns is unconvincing.
In fact, reexamination and reinterpretation of the data and the
sampling procedures lead to the conclusion that improved delivery
and payment mechanisms - however desirable and necessary they
may be for the poor - do not result in the extent of preventive health
behavior exhibited by the more affluent.
A recent article by Monteiro presents arguments similar to those
of Colombo et al.lS The author points out that, through 1968,
physician visits reported in national data no longer seemed directly
dependent on income. The lowest (under $3,000) and highest (above
$10,000) income groups made the most visits with intermediate
incomes exhibiting fewer visits. Monteiro reports a study in Rhode
Island, conducted in 1967, 1968, 1969, with a follow-up in 1971, which
confirms the national data. The author herself notes that there may be
wide differences as to what is defined as illness, a topic she did not
study. Thus, the poor may tolerate more severe morbidity before
restricting their activity and in fact go to the doctor with much more
serious illness than the non-poor. But, since it has been shown that the
poor have substantially higher levels of morbidity than the non-poor,
if income were unrelated to utilization one might expect that they
should visit the physician not only as often as the non-poor, but much
more often. Equally crucial criticisms can be leveled a t Monteiro’s
interpretations, for she merely used gross counts of doctor visits, a
measure that has always been regarded a s fairly crude. If she had
studied the purposes of the visits or the use of other professional
providers, for example dentists, she would have found what has so
often been reported in the national health survey and by other
investigators, namely that the poor show substantially lower
utilization than the non-poor of each of the following kinds of
preventive and curative services.
1. Percent under age 17 receiving routine physical examinations.
2. Percent under 17 visiting the pediatrician.

358 Health Education Monographs VOL, 2, NO. 4

 3. Percent of women visiting a n Ob/Gyn.
4. Percent visiting the dentist within the last year and, within that
category, percent obtaining dental restorations vs. extractions.
5. Percent of women voluntarily seeking Pap smears.
6. Percent of adults seeking regular chest X-rays.
7. Percent obtaining polio immunizations.
8. Percent brushing their teeth regularly.
What Monteiro's study seems to show is that publicly financed
care will increase utilization among poor people experiencing
substantial medical need but the level of utilization for less disabling
conditions, and the use of preventive services, remains highly
correlated with income; the poor receive far less of such services than
the non-poor. I t may also be emphasized that, within comparable
income categories, utilization of services tends to be substantially
higher for whites than non-whites.
Perception of Symptoms as an
Intermediate Variable in Utilization
Although most studies of utilization do not throw light on why
people use health services, one area of research can be identified in
which quite sophisticated efforts have been made to understand health
and illness behavior as a function of personal characteristics; a n area
described by Kasl and Cobb as "variables affecting the perception of
symptoms." I Several other workers attempt to link personal and
subcultural variables to the individual's likelihood of perceiving a n
event as a symptom or to his mode of responding to a symptom. For
instance, Koos found a social class gradient in terms of the likelihood
of interpreting a particular sign as a symptom.17 Stoeckle, Zola,and
Davidson and Zola studied the effects of ethnic values upon the
specific decision to seek medical attention and on the differential
interpretation of objectively similar symptoms.l"1t' Freidson
illustrated the different processes through which members of different
social groups move in obtaining diagnosis (lay and professional) and
in seeking care." Suchman attempted a n interesting and promising
approach which links demographic factors to social structure, both of
these to medical orientation and in turn to health and medical care.21
Studies of the kinds performed by Koos, Stoeckle, Zoia, Freidson,
and Suchman are far superior in their ability to explain than are the
more traditional analyses of relationships between demographic
factors and the utilization of services. This superiority lies in the
proposed linking mechanisms between personal characteristics and
behavior. These studies also demonstrate that health decision making
is a process in which the individual moves through a series of stages or
phases. Interactions with persons or events a t each of these stages
influence the individual's decisions and subsequent behavior.

Health Education Monographs Winter I974 359

 Yet, even these more sophisticated approaches limit their focus to
illness behavior; that is, to behavior undertaken in response to
symptoms. The findings are, thus, of unknown relevance to the
situation confronting the person who must decide whether to seek
preventive or detection services before the appearance of events that
he interprets as symptoms. Suchman explicitly notes the failure to his
concepts of social structure and health orientation to account for
preventive health actions." Stimulating the development of a
preventive orientation in the public is a t the heart of most educational
programs in public health.

EVIDENCE FOR AND AGAINST THE MODEL

Although many recent investigations have identified explanatory
variables which are similar to one or another variable contained in the
model, seven major projects were originally undertaken whose design
was largely or entirely determined by the behavioral model.2"z&"Of
these, four were retrospective while three were prospective
studies."&"' The retrospective research projects have in common the
crucial characteristic that data about respondents' beliefs and
behavior are gathered during the same interview, and the beliefs are
assumeci to have existed in a point in time prior to the behavior. That
assumption is a questionable one at best and will be considered after a
review of the retrospective research.
One other problem in the interpretation of the studies should be
noted. With the exception of the Hochbaum study 54 and the National
Study of Health Attitudes and the research has been
based on quite small samples. Sometimes sample size has been limited
by financial or other insuperable obstacles. However, in some cases
difficulties in categorizing responses or in obtaining responses to every
necessary item have reduced samples to dangerously low proportions.
Finally, the concept of motivation, now considered central to the
Health Belief Model, was not specifically studied in these seven early
investigations. The best documented of the early retrospective studies
were performed by Hochbaum and Kegeles,zs and these will be
reviewed in some detail.

The Retrospective Studies

Hochbaum clearly originated research on the Health Belief Model.
Beginning in 1952, he studied more than 1200 adults in three cities in
a n attempt to identify factors underlying the decision to obtain a chest
X-ray for the detection of tuberculosis. He tapped beliefs in sus-
ceptibility to tuberculosis and beliefs in the benefits of early detection.
Perceived susceptibility to tuberculosis contained two elements. It
included, first, the respondent's beliefs about whether tuberculosis was
a real possibility in his case, and second, the extent to which he

360 Health Education Monographs VOL. 2, NO. 4

accepted the fact that one may have tuberculosis in the absence of all
symptoms. Consider first the findings for the group of persons that
exhibited both beliefs, that is, belief in their own susceptibility to
tuberculosis and the belief that overall benefits would accrue from
early detection. In that group 82 percent had had a t least one
voluntary chest X-ray during a specified period preceding the
interview. On the other hand, of the group exhibiting neither of these
beliefs, only 21 percent had obtained a voluntary X-ray during the
criterion period. Thus, four out of five people who exhibited both
beliefs took the predicted action, while four of five people who
accepted neither of the beliefs had not taken the action.
Thus, Hochbaum appears to have demonstrated with considerable
precision that a particular action is a function of the two interacting
variables - perceived susceptibility and perceived benefits.
The belief in one’s susceptibility to tuberculosis appeared to be the
more powerful variable studied. For the individuals who exhibited this
belief without accepting the benefits of early detection, 64 percent had
obtained prior voluntary X-rays. Of the individuals accepting the
benefits of early detection without accepting their susceptibility to the
disease, only 29 percent had prior voluntary X-rays.
Hochbaum failed to show clearly that perceived severity plays a role
in the decision-making process. This may be due to the fact that his
study was not designed to identify perceived severity with any high
degree of accuracy and his measures of severity proved not to be
sensitive. However, he did identify 16 individuals who seemed
intensely afraid of TB. None of these had had a single voluntary X-ray
during the preceding eight-year period. In addition, those respondents
who appeared indifferent to the disease were among those who tended
not to feel susceptible, and consequently not to take X-rays. Finally,
those who exhibited some “mid-range’’ level of fear participated to a
slightly greater extent than those a t the very high or low end of the
scale.
Kegeles 85 dealt with the conditions under which members of a
prepaid dental care plan will come in for preventive dental check-ups
or for prophylaxis in the absence of symptoms. He attempted to
measure the respondent’s perceived susceptibility to a variety of dental
diseases, the perceived severity of these conditions, his beliefs about
the benefits of preventive action and his perceptions of barriers to
those actions.
While findings usually support the importance of the model
variables, their general applicability is greatly limited by a n unusually
large loss in the sample. The study was initiated with a sample of 430,
but those without teeth, those for whom information was not available
to determine whether past dental visits had been made for preventive
purposes or for treatment of symptoms and those whose positions

Health Education Monographs Winter I974 361

could not be coded on all three belief variables were excluded. The
crucial analysis could thus be made only on 77 individuals. Within the
major limitations implied by the small sample size and by their likely
nonrepresentativeness, Kegeles showed that with sucessive increases in
the number of beliefs exhibited by respondents from none to all three,
their frequency of making preventive dental visits also increased. The
actual findings show that: ( 1 ) of only three persons who were low on
all three variables none made such preventive visits; ( 2 ) of 18 who
were high on any one variable but low on the other two, 61 percent
made such visits; ( 3 ) of 38 persons high on two beliefs and low on one,
66 percent made preventive visits; and, finally, ( 4 ) of 18 persons who
were high on all three variables, 13 or 78 percent made preventive
dental visits. Similar patterns of findings based on much larger
samples were obtained in a n analysis of relationships between
behavior and each of a series of single variables, that is, susceptibility,
severity, benefits and barriers.
The findings ot tne two remaining retrospective studies will not be
reviewed in detail but are in most respects quite similar to the two
that have been r e v i e ~ e d . ~ eIn; ~each
~ case evidence that supports the
model has been obtained, although the sample sizes were not large.
In summary, while no one of the original studies provided convincing
confirmation of the model variables, each produced internally
consistent findings which were in the predicted direction. Taken
together they thus provided reasonably strong support for the model.
As indicated, any interpretations made of the findings of the
retrospective studies are based on a n assumption. The hypothesis that
behavior is determined by a particular constellation of beliefs can only
be tested adequately where the beliefs are known to have existed prior
to the behavior that they are supposed to determine. However, the
retrospective projects were undertaken in situations which
necessitated identifying the beliefs and prior behavior a t the same
point in time. This approach has always been known to be quite
dangerous. Work on cognitive dissonance:2 supported these suspicions
and suggested that the decision to accept or reject a health service may
in and of itself modify the individual’s perceptions in areas relevant to
that health action. Obviously, a two-phase study was needed in which
beliefs would be identified at one point in time, and behavior
measured later.

The Prospectioe Studies

Such a study was attempted in the fall of 1957, around the topic of
the impact of Asian influenza on American community As one of
a series of related studies, Leventhal, Hochbaum, and Rosenstock
investigated the impact of the threat of influenza on families through
the use of a design that was intended to permit a test of the model in a

362 Health Education Monographs VOL. 2, NO. 4

prospective manner. In this phase of the study, 200 randomly selected
respondents in each of two medium-size cities in the United States
were interviewed twice. The first interview was intended to be made
before most people had the opportunity to seek vaccination or to take
any other preventive action and before much influenza-like illness had
occurred in the communities. The second interview w a s to be made
after all available evidence indicated that the epidemic had subsided.
In fact, only partial success was achieved in satisfying these
conditions because community vaccination programs as well as the
spread of the epidemic moved much faster than had been anticipated.
For these reasons the sample on which the test could be made was
reduced to 86. This sample of 86 respondents had, at the time of initial
interview, neither taken preventive action relative to influenza nor had
they experienced influenza-like illness in themselves or in other
members of their families. Twelve of the 86 scored relatively high on a
combination of beliefs in their own susceptibility to influenza and the
severity of the disease.36 Five of these 12 subsequently made preventive
preparations relative to influenza. On the other hand, at the time of
the first interview, the remaining 74 persons rejected either their own
susceptibility to the disease or its severity or both. Of these, only eight,
or 11 percent, subsequently made preparations relative to
Although the samples on whom comparable data could be obtained
were very small and possibly not representative, the differences were
statistically significant beyond the one percent level of significance.
Analyses of the available data thus suggest that prior beliefs in
susceptibility and severity are instrumental in determining subsequent
action.
A second prospective study- was a follow-up by Kegeles 28 on the
study reported earlier.’6 Three years after the initial collection of data
on a sample of more than 400 in 1958, a mail questionnaire was sent
to each person in the sample as well as to a comparable control group
to obtain information about the three most recent dental visits. The
objective of the follow-up was to determine whether the beliefs
identified during the original study were associated with behavior
during the subsequent three-year period.
Kegeles found that perceptions of seriousness, whether considered
independently or together with other variables, were not at all
associated with subsequent behavior. Perceptions of benefits taken
alone were not related to subsequent behavior. However, the
perception of susceptibility did show a correlation with making
subsequent preventive dental visits. Of those who had earlier seen
themselves as susceptible, 58 percent made subsequent preventive
dental visits while 42 percent who had not accepted their susceptibility
made such visits. When beliefs about susceptibility and benefits were
combined, a more accurate prediction was possible of who would or

Health Education Monographs Winter I974 363

would not make preventive dental visits. Considering only those who
scored high on susceptibility, and cross-tabulating against beliefs in
benefits, 67 percent of those high on both beliefs made subsequent
preventive visits while only 38 percent low in benefits made such visits.
Thus, the combination of susceptibility and benefits was demonstrated
to be important in predicting behavior.
The results of the six studies cited above lend support to the
importance of several of the variables in the model as explanatory or
predictive variables. However, a seventh major investigation conflicted
in most respects with the findings of earlier studies.a",al The study
included analyses of beliefs and behavior of a probability sample of
nearly 1500 American adults studied in 1963, and the subsequent
behavior of a 50 percent subsample studied 15 months later. Perceived
susceptibility, severity, and benefits, whether taken singly or in
combination, did not account for a major portion of the variance in
subsequent preventive and diagnostic behavior, in regard to dental
visits, X-rays for TB, or check-ups for cancer, although predictions
based on the belief in benefits taken alone frequently approached
significance. The study findings did not disclose any explanation for
the failure to obtain findings similar to those of the earlier described
studies, but the more recent national study was conducted in a setting
which distinguished it from all the other reported studies in one
respect that may have been crucial.

Health Education as
an Independent Variable
I n the earlier described studies, the settings were such that the
population in each case had been offered the opportunity to take
action through directed messages and circumstances that could have
served as cues to stimulate action. In Hochbaum's study,%4mass
media had been used in three study cities to urge the population to
obtain chest X-rays. I n the Kegeles studies,"*.Joevery member of the
population was offered free or inexpensive dental treatment and was
urged to use it. I n the Leventhal, Hochbaum, and Rosenstock study,zm
the population had been alerted by newspapers and by public health
officials to the desirability of obtaining influenza immunizations. In
the Heinzelmann study,Ze the patients had been urged to use penicillin
prophylactically. The Flach study 27 offered the population a free test
for cervical cancer. I n short, in all the prior studies the population had
been exposed to information which both indicated the availability of a
health procedure and which, in most cases, urged them to avail
themselves of that procedure.
In contrast, such conditions did not occur for the national sample in
the most recent national study. With respect to the several health
problems covered in the study, neither the sample nor the United

364 Health Education Monographs VOL. 2, NO. 4

States adult population which it represents, had been uniformly
exposed to intensive campaigns to inform them about available
services and to persuade them to use such services. Nor can the
assumption reasonably be made that preventive and diagnostic
services were equally available to all. The absence of clear-cut cues to
stimulate action as well as unequal opportunity to act, may, in large
measure account for the failure to replicate the earlier results.
However, those possibilities must be treated as hypotheses which will
need to be tested in new research.
Two more recent studies lend further support to the general
explanatory model of health behavior although its specific variables
have been treated more innovatively than in earlier research. Haefner
and Kirscht attempted experimentally to increase people’s readiness
JJ

to follow preventive health practices by presenting them with messages

about selected health problems that were intended both to increase
their perceived susceptibility and/or severity regarding the health
problems and their beliefs in the efficacy of professionally
recommended behavior. Significantly more persons exposed to such
messages visited a physician for a check-up in the eight months
following the experimental manipulation than in a control group not
exposed to the messages. This significant difference held only for visits
made in the absence of symptoms, i.e., preventive health behavior. For
individuals reporting actual symptoms during the interval, the rate of
physician visits was the same in the experimental and control group.
While income as such was not treated, the sample represented a
universe of non-academic University employees, a group above the
poverty level but, in general, far from affluent. This study,
incidentally, provided evidence that it is possible to modify the
perceived threat of disease; that is, the combination of perceived
susceptibility to and severity of diseases as well as the perceived
efficacy of professional intervention, and that such modification leads
to predictable changes in health behavior.
Still more recently, Becker, Kaback, Rosenstock, and Ruth applied
the Health Belief Model to the area of genetic ~creening.3~ Beginning
in 1971, a n identified Jewish population in the Baltimore-Washington
area was invited to participate in screening for the Tay-Sachs trait
which has a frequency of about one in 30 Jews of Ashkenazi ancestry
(compared with one in 300 among non-Jews). About one in 900 Jewish
couples could thus be expected to be a t risk of having a child with
Tay-Sachs Disease, a n incurable condition which is invariably fatal in
early childhood. Of course the probability of having such a diseased
child for at-risk couples is one in four. The disease can, however, be
diagnosed in utero through amniocentesis at a stage when abortion is
feasible. Thus, the situation presents all the conditions for observing
the role of the components of the Health Belief Model in predicting

Health Education Monographs Winter I974 365

preventive health behavior. Furthermore, since the relatively rare
disease and the diagnostic test were largely unknown to the lay puolic,
it is a reasonable inference that the majority had had little contact
with the disease, with screening or with amniocentesis and that they
had few relevant beliefs about it in advance of the program.
The education of the target community began six to eight weeks
before initiation of mass screening. Multiple educational approaches
were used to saturate the communities with accurate and clear
information. These included the press, T V , radio, letters from Rabbis,
fliers from community organizations, medical presentations to the
community, telephone calls from trained volunteers, brochures from
physicians and other special mailings. Since lists of the target
population were available, it could be assured that all members of the
target group - couples of childbearing age - were exposed to a t least
some of these educational activities. As applied to the Tay-Sachs
situation, the explanatory variables were defined as follows: Health
motive was for the first time explicitly introduced into the model to
explain health behavior. In the present case, motive included two
components: (1) a positive response indicating a desire to have
(additional) children and ( 2 ) a set of generalized items about typical
health behavior, such as the frequency with which the person ’thinks
about his own health and whether he generally goes to the physician
right away if he feels sick. Perceived susceptibility included the
person’s belief that he could carry the Tay-Sachs gene and transmit it
to his progeny. Severity was interpreted as the individual’s views of the
potential impact of learning that he was a carrier, especially as
regards future family planning. The definition of perceived benefits
was in terms of a personal evaluation of how much good it would do
the potential carrier to be screened for the trait. Did he really need to
know or want to know his carrier status? Costs or barriers to action
were not directly measured in this study. They might include, however,
usual monetary or convenience factors as well as threats we currently
know very little about, for example, the impact on an individual of
learning that he is a carrier of some recessive trait. How does it affect
his self-image, his perception of his health and of his well being? Does
it affect his marriage? How does it influence future family planning?
I n all, nearly 7,000 adults were screened during the first year of the
study, all drawn from lists of synagogue membership, and names in
predominantly Jewish neighborhoods. All adults who appeared for
screening were asked to complete a brief questionnaire just before
going through the screening process; 500 of these were selected a t
random as the participant sample. In addition, 500 questionnaires
were mailed to a random sample of nonparticipants who had been
invited in for screening; here the response rate was 82 percent. I t
should be noted that non-respondents as well a s respondents had

366 Health Education Monographs VOL. 2, NO. 4

received intensive informational material on Tay-Sachs disease and
screening.
T h e analysis showed t h a t the participants, compared to
nonparticipants, were significantly younger, had had fewer children,
were less likely to have completed their families, and were slightly
better educated. Turning to the health belief variables, the
participants differed sharply in perceived susceptibility, the first
component of health motivation - 82 percent of those who expressed
the desire to have future children participated in the screening
program while less than 19 percent who did not desire future children
participated; there was no significant difference in participation
according to perceived severity, the second, less direct motivational
measure used. Mean score on perceived susceptibility to being a
carrier was highly correlated with participating in the screening
program while perceived severity was negatively associated with
participation.
When the three foregoing variables were combined it became
apparent that while each of the three is associated with participation,
perceived susceptibility interacted with the desire to have future
children while perceived severity played a n independent, additive role;
for people who desire additional children, moderate perceived
susceptibility and low perceived severity best explains participation in
the program. Among those who are not motivated to have additional
children, high susceptibility and low severity best explains
participation. Irrespective of motivation, the combination of high
perceived susceptibility and low perceived severity best accounts for
participation.
Benefits-to-barriers ratio. Among those individuals who indicated
that they planned to have more children, the non-participants more
than the participants indicated that the discovery that either or both
husband and wife were carriers would change their future child 5

planning behavior; frequently they reported they would have no

additional children. One possible interpretation of this finding related
to beliefs exhibited by participants and non-participants about the
transmission and detection of Tay-Sachs disease and about
reproductive alternatives. A question on the impact of learning that
one member of a married couple was a carrier obtained quite different
answers from participants and non-participants. Participants were
much less likely than non-participants to alter their plans. More of the
participants had apparently learned that carrier status in only one
member of the couple poses no dangers. However, in response to the
question on the impact if both parents were found to be carriers, while
participants were again less likely to change their reproductive plans
than nonparticipants, they did indicate they would reduce the number
of children they would have or that they would use “other”

Health Education Monographs Winter 19 74 367

approaches. In nearly every case where the “other” category was used,
participants went on to explain that they would elect to use
amniocentesis (fetal diagnostic test) in order to continue to have
children. Very few of the non-participants displayed knowledge of the
availability of amniocentesis; rather, they tended to indicate that, in
the event either member of both members of a couple were found to be
carriers, they would not have further children.
Since more participants than non-participants learned about the
“fetal diagnostic test,” it may be inferred that screening conferred
considerable benefits for participants: (1) they could rule out the
possibility that both parents carried the recessive gene or ( 2 ) if both
proved to be carriers, amniocentesis could rule out the possibility that
the fetus had the disease, or (3) if the child were diseased, they could
elect to abort it. On the other hand, while nearly all the study
respondents held attitudes favoring abortion in the event that a fetus
had TSD, the non-participants could not have seen as much benefit in
screening, since they did not give evidence of having learned about
amniocentesis.
Perceived barriers. Barriers to screening, though not studied
explicitly, were minimized in the present case by offering the test a t
low cost to a relatively affluent group and a t convenient times and
locations. Such financial and situational factors could, however, be
important for other target groups.
I n summary, more of those who agreed to participate, compared to
non-participants, (1) believed they could be Tay-Sachs carriers (high
perceived susceptibility), (2) stated that learning that they were
carriers would not affect their family planning (low perceived
severity), and (3) abortion was appropriate if the fetus should prove to
be diseased (high perceived benefits). Fewer of the non-participants,
on the other hand, believed that they were susceptible. Although they
favored abortion to the same extent as the participants, more of the
non-participants indicated that learning they were carriers would be
quite disruptive of their lives and that it would seriously affect their
future family planning. Limitations in the length of the questionnaire
precluded studying why the non-participants felt that learning of their
carrier status would be considerably more disruptive than was true for
participants. But, given t h a t orientation, their behavior is
understandable.
One final consideration should be emphasized. I t is believed that in
this case perceived severity associated with the Tay-Sachs trait
reached such high levels as to become dysfunctional. ( I t will be
recalled that a similar phenomenon had been observed in Hochbaum’s
original study.) Although the nature of the data in previous studies
have precluded the use of standard parametric analytic techniques, it
has always been believed that what is needed for behavior is “an

368 Health Education Monographs VOL. 2, NO. 4

optimal” balance of perception of health motive, vulnerability,
severity, and the psychological benefitlcost ratio. Where the balance
among these is either quite ‘‘low’’ or quite “high,” professionally
recommended behavior is not to be expected. The truth of this
assertion can only come out of the studies which use measures
sensitive to variations in the degree to which each variable is present.

Additional Evidence
As indicated earlier, various studies of preventive health behavior
have dealt with one or another component of the Health Belief Model
as a more or less incidental part of their design.
In a 1959 analysis of public response to polio vaccinations, drawing
on a number of independent investigations, Rosenstock, Derryberry,
and Carriger 7 concluded that the variables: perceived susceptibility,
perceived severity, perceived benefits (safety and effectiveness of the
vaccine) accounted for major portions of the variation in seeking polio
vaccination, though the authors of the several studies had not been
guided in their designs by the Health Belief Model.
The same review confirmed the importance of social pressures in the
decision to seek vaccination a s Hochbaum had earlier shown was the
case for the decision to seek chest X-rays.
Heinzelmann and Bagley 35 reported on reasons for participation in
physical activity programs and perceived outcomes. The two most
important reasons given for participation were the desire to feel better
and to lessen the chance of a heart attack. And one of the major
perceptual outcomes of participation w a s a feeling of decreased
vulnerability to specific health threats, including heart attacks.
Fink et a1 36 provide data that suggested that the perception of
personal vulnerability to cancer and a concern with its severity
distinguish participants from non-participants in a breast cancer
screening program.
Battistella 37 purports to show that a measure of readiness to
initiating physician visits and to obtaining checkups, presumably
adapted from Hochbaum’s concepts a s well as others, yielded only
slight relationships. Since Battistella’s work is criticized in some detail
in Kirscht’s chapter, only a few brief summary statements are needed
here. Despite the fact that an inappropriate statistical test was chosen,
the relationships he showed in the Hochbaum data were substantial;
variables preferred by the author turned out to be less predictive. In a
second paper 38 Battistella ruled out a whole host of typical
sociological variables (age, economic status, etc.) a s explaining delay
in seeking care. Only two variables were promising - both health
beliefs. The first was the perceived efficacy of care and the second was
worry about health.

Health Education Monographs Winter 1974 369

 Suchman agrees that the “motivation to change one’s health
practices depends, to a large extent, upon the individual’s feelings of
personal vulnerability and the seriousness with which he views the
health hazard.” J@

Ogionwo attempted to influence response to a cholera

immunization and educational program in Nigeria and showed health
concern (motivation), perceived vulnerability to cholera, and
knowledge of preventive measures highly associated with response, as
were also several attitudes to prevention and correct knowledge of
cholera .
Gochman 41 has shown that where health-motivation is high
children’s perceptions of vulnerability to dental disease and perceived
benefits of dental treatment predict their intentions to make a future
dental visit better than where health motivation is low. Health
motivation is thus seen to be a n important organizing factor in health
beliefs and intentions.
I n a n experimental attempt to change beliefs and behavior of
women in a n urban ghetto concerning screening for cervical cancer,
Kegeles 42 demonstrated that women with relatively high beliefs in
their vulnerability to cervical cancer and in the effectiveness of
cytology made more visits than their counterparts.
Antonovsky and Kats 4J have developed a n integrated model of the
determinants of health behavior which is acknowledged to be in many
respects similar to the Health Belief Model and which explains much
of the variation in behavior of a sample of more than 500 employees of
the Hadassah Medical Organization in Jerusalem. The authors raise,
however, three points of difference between their model and the
Health Belief Model, two of which are of considerable significance.
These should be considered in some detail.
The first criticism is of our concept of “cues.” For Antonovsky and
Kats “cue” is a superfluous concept - the individual otherwise
prepared to take action will “create his own cues.”4J There
may be no more than a semantic difference here; nevertheless, it seems
useful to retain the concept of cues which serve as triggers to initiate a
train of events. It has already been suggested that Zola’s “critical
incidents” may be identical with cues, and the concept seems to fit
well with general psychological theory.
Of greater moment is the criticism that the early Health Belief
Model failed to include any frank motivational concept. This criticism
is quite well taken. For Antonovsky and Kats, motivation is goal
oriented behavior and the relevant goal is that of maintaining health.
Other relevant goals might include achieving approval by significant
others or achievement of self-approval.
As indicated in several chapters, we have independently come to
recognize the need to include a motivational variable, and recent

370 Health Education Monographs VOL. 2, NO. 4

studies using the Health Belief Model have incorporated such a
component.
Unfortunately, we cannot go as far as the authors do in regarding
the principal roles of the perception of susceptibility and severity as
that of increasing the salience of the motive. We believe that the
cognitive factors play a somewhat independent role in influencing
behavior, though motivation just as clearly is important (see papers by
Recker and Kirscht).
One final important point is the introduction by Antonovsky and
Kats of the concept of a threshold level rather than a linear,
monotonic relationship between a given variable and preventive
health behavior. There is some reason to believe they are correct.
Difficulties in measurement to date, to be described subsequently,
have precluded testing the hypothesis that the motive and the
perceptions of susceptibility, severity, benefits, and costs may each
have cut-off points for a given individual below which the variable will
have no effect and above which it will have effect, with additional
increments making no difference. While this is a most useful concept,
data already presented, especially as regards perceived severity,
suggest that the threshold concept may need to be applied cautiously,
since a t very high levels, anxiety and maladaptive behavior may
result. Even with this caution the notion of threshold levels, rather
than that of monotonic relationships, seems well worth pursuing.
CRITIQUE OF THE MODEL AND
NEEDED ADDITIONAL RESEARCH
The Place of the Model in the
Hea lth Decision-Mak ing Process
Health decision-making, including health behavior, illness behavior,
and sick-role behavior, is a process in which the individual moves
through a series of stages or phases in each of which he interacts with
individuals and events. The nature of the interactions a t any one of
these stages may increase or decrease the probability that a particular
subsequent response will be made. Freidson ZIJ and Zola 19 have
illustrated some of these stages. The individual’s relevant health
beliefs as described in this paper are presumed to serve a s a setting for
his subsequent responses at other stages in the decision process. For
example, individuals who accept their susceptibility to a particular
condition and are aware of actions that might be beneficial in
reducing their susceptibility may well be the same persons who exhibit
what Freidson terms “cosmopolitan” rather than “parochial”
orientations toward health services. They may be more prone to learn
about and seek out professional diagnosis rather than using the ”lay
referral” system. In such a case the initial set of beliefs would itself
determine subsequent choices in the decision-making process.

Health Education Monographs Winter 19 74 371

 W h a t have here been termed “cues” a r e probably identical with
Zola’s “critical incidents.” 18 One cannot b u t agree heartily with his
recommendation t h a t the role of such triggers to action be much more
thoroughly investigated t h a n h a s previously been done. T h i s is urged
despite the forbidding difficulties in identifying cues t h a t have already
been described.

Operational Definitions of the Variables

No two studies of the model’s variables have used identical questions
for determining the presence o r absence of each belief. T h i s raises the
possibility t h a t t h e concepts being measured m a y also vary from study
to study.
It m a y be helpful to quote the questions actually asked (where
available) in the first five studies guided by the H e a l t h Belief Model.

PERCEIVED SUSCEPTIBILITY
Hochbaum (1958)
(1) If you were some day to get very seriously sick what do you think it
would be? (a) What else could it be?
(2) Did you ever think you might some day get TB? ( a ) Why do you feel that
way?
(3) Please look a t this card. (SHOW CARD A ) Which choice comes closest to
your feeling about how likely it is that you may get TB some day? ( a ) What is
your reason for picking that one?
( 4 ) (USING SAME CARD) How likely do you think it is that you may get
TB in the near future? ( a ) Why do you pick that one?
Kegeles (1963)
(1)How likely do you think it will be that (worst dental problem previously
experienced by respondent) will happen to you again? ( a ) Why do you think
that?
(2) How likely do you think it will be that (worst dental problem he could
imagine) will ever happen to you?
Fhch (1960)
The specific questions used are not available but the report notes that
respondents were asked:
(1) Whether single women, married women with children, or married
women without children “are most likely to get uterine cancer.” Women
choosing their own group were coded as “susceptible.”
(2) Each woman was also asked whether she had ever thought it possible
that she could have cancer.
Heinzelmann (1962)
Susceptibility was an index score derived from the following two questions:
(1) Taking all possible factors into consideration, what do you think your

312 Health Education Monographs VOL. 2, NO. 4

own chances are of getting rheumatic fever again? ( a ) Very Likely, ( b ) Likely,
(c) Unlikely, ( d ) Very Unlikely?
( 2 ) Taking all possible factors into consideration, what do you think your
chances of getting it again are in comparison to otherpeople who have had a n
attack of rheumatic fever? ( a ) Much More, ( b ) More, (c) Same, ( d ) Less, (e)
Much Less?
Leventhal ( 1 960) a9

Questions are not available but the report indicates t h a t each respondent
was asked whether he thought he might contract influenza.

PERCEIVED SEVERITY
Hoch baum
No measures
Kegeles
( 1 ) If any of the dental problems (previously experienced) happened, how
serious would i t be?
( 2 ) If any of the dental problems (imagined but not experienced) happened,
how serious would i t be?
I n addition, a respondent was categorized as believing in severity if he
indicated anywhere in the interview t h a t such problems could detract from
one’s aesthetic appearance.
Flach
No measures. It was presumed that all or virtually all women believed
cancer to be serious.
Heinzelmann
Severity was a n index score derived from the following four items:
(1) How would you estimate the seriousness of your attack of rheumatic
fever? ( a ) Serious, (b) Fairly serious, (c) Fairly mild, ( d ) Very mild.
( 2 ) Please check all of the following situations which apply to your attack:
(a) Missed more than two months of schooling. (b) After the first few days,
didn’t feel bad a t all. (c) Didn’t have any long-lasting after effects. ( d ) Was not
permitted to take part in school gym or athletics. (e) Was not in bed more
than a week. (f) Compared to others I know, my attack was mild. (g) Had a
heart condition.
(3) Rank the following illnesses in order of seriousness from most serious to
least serious. Use “1” for most serious. (a) Measles, (b) Pneumonia, (c)
Rheumatic fever, ( d ) Polio, (e) Tuberculosis, (f) Chicken pox.
( 4 ) Have you had any other diseases or illnesses, which you think were more
serious than your attack of rheumatic fever? (a) No, (b) Yes.
Leventhal
(1) “Do you think it would be worse or not as bad if you had the Asian flu
compared to how i t usually is when you have the flu (grippe, bad cold)?”
(2) “How would the Asian flu be different from the usual flu or grippe?”

Health Education Monographs Winter I974 313

 PERCEIVED BENEFITS
Hochbaum
(1)If you should happen to get T B some day, how do you think you would
be most likely to find out you had it? ( a )Why do you think you would find out
that way (those ways)? (b) What other ways are there for discovering that a
person has T B ?
(2) You’ve just mentioned X-rays as one of the ways of finding out that a
person has TB: In your opinion could an X-ray discover T B before or after the
person himself could notice that something is wrong? (a) Why do you say that?
(3)Suppose X-rays show that a person has TB, would it make any difference
whether he starts treatment immediately or waits about six months to a year?
( 4 ) If “YES”: What difference would it make?
(5) Some people think it is a good idea to have chest X-rays and some may
not think so. Why do some think it is a good idea? (Else?) ( a ) How do you feel
about these reasons?
(6)Some may not think it is a good idea to have chest X-rays. Why do they
feel t h a t way? (Else?) (a) How do you feel about these reasons?

Kegeles
The following questions were asked separately for ( a ) the worst problem
experienced and (b) the worst problem anticipated:
(1) Do you know of anything a person could do that would make it less
likely t h a t he would get (worst dental problem)? What could he do?
(2) If “NO” - “Do you mean that there is absolutely nothing a person can
do to make it less likely t h a t (worst dental problem) would happen?

Flach
Questions are not available but the report indicates that respondents were
asked about (a) their beliefs about whether early detection makes a great deal
of difference so far as chances of cure are concerned, and (b) their degree of
optimism concerning effects of cancer surgery upon marital relations and on
physical energy.

Heinze lmann
Benefits is an index score derived from following four items of knowledge
and beliefs:
(1) Please check below what you believe is the most important cause of
rheumatic fever. (a) Overweight condition, (b) Heart weakness, (c) Bacteria,
viruses, or other infections, ( d ) Typhoid fever, (e) Runs in the family, (f) Sore
throats, (g) R u n down physical condition, (h) Improper diet.
(2) A person who has a lot of colds and sore throats is more likely to get
rheumatic fever again than a person who never has a cold or sore throat. ( a )
Agree very much (b) Agree, (c) Disagree, ( d ) Disagree very much.
(3) Rheumatic Fever: (a) Can lead to rheumatoid arthritis, (b) Has no after
effects, (c) Can lead to heart disease, ( d ) Usually weakens a person for a t least
10 years after, (e) None of these.
( 4 ) Which of the following do you think would be most important in keeping
you from getting rheumatic fever again? ( a ) Keep from getting tired, (b) Take

374 Health Education Monographs VOL. 2, NO. 4

care of sore throats, (c) Eat proper foods, (d) Get regular checkups from a
doctor, (e) Take medicine, (f) Stay in good physical condition.
Leventhal
Questions are not available but the report indicates that respondents were
asked ( a ) whether vaccination provides protection against Asian influenza;
and, (b) what actions families could take to prepare for an epidemic of Asian
flu.
I n a n effort to bring some stability into the area of measures of
health belief variables, the National Study of Health Beliefs described
earlier was undertaken.Ju7Jl I n t h a t study alternative methods were
used to identify beliefs about the severity of a n d susceptibility to four
diseases: dental decay, gum trouble, tuberculosis, a n d cancer. Four
different question formats were developed, differing simultaneously on
two dimensions: ( 1) "self-reference" versus "reference to men-women
your age" a n d ; ( 2 ) fixed-alternative versus more open items. A two-by-
two design was used with approximately one-quarter of the total
sample randomly assigned to each of the four question formats.
T h e findings demonstrated t h a t the question types obtained different
distributions of responses.Ju However, since in t h a t study no clear
relationship was demonstrated between possession of the beliefs,
however measured, and health behavior, no decision can be made on
which method of questioning is most valid. Clearly, more standardized
ways of asking questions will have to be developed.

Quantification
T h e model implies t h a t certain levels of readiness a r e optimal in
stimulating behavior but neither theory nor research have disclosed
what the levels are. I n most of the studies limitations in sample size
have necessitated dichotomizing scores on the variables into categories
of "high" a n d "low." Until d a t a can be collected on a t least a n ordinal
scale the problem of determining optimal quantities will not be solved
and the dispute between threshold effects a n d linear relationships as
predictors of behavior will not be resolved.

Stability and Reliability

of the Beliefs
Little is known about the stability of the beliefs, although they may
well vary from time-to-time as a function of situational changes.
Learning t h a t a friend or a president h a s suffered a serious illness may
raise personal levels of motivation a n d readiness to act. Research is
needed to determine how stable the beliefs are.
Similarly, little information is available on the reliability of the
measures of beliefs. More work is also needed in this area. The
appropriate approach to testing reliability depends on the stability of
the beliefs. If the beliefs do change from time to time, test-retest

Health Education Monographs Winter I 9 74 315

measures of reliability would not be as appropriate as split-half
measures of reliability.

Perceived Seriousness
Hochbaum,zd Kegeles,’u and Kirscht et al,Y” failed to demonstrate the
importance of perceived severity in determining behavior. Flach z 7 did
not measure severity since she assumed that cancer was universally
seen as severe in the group of women studied. On the other hand. the
studies by Heinzelmann,z6 the first Kegeles study,’Z and Leventhal et
al,” did support the importance of perceived severity. Subsequently,
Haefner and Kirscht 33 and Becker et a1 34 again showed the
importance of perceived severity although, in the latter case, the
relationship was complex. Greater doubt must be maintained about
the importance of perceived severity as a n explanatory factor in health
behavior than about the other variables.
I t should be noted that while doubt remains concerning the role of
perceived severity in stimulating preventive health behavior, its role,
while complex, is clearer in studies of illness and sick-role behavior
(see the chapters by Kirscht and Becker in this issue). Severity would
appear to be related to behavior in a curvilinear manner. Where it is
very low or very high, maladaptive behavior would seem to result.12,44
Where it is “moderate,” more adaptive behavior results. Clearly, more
research on this variable is needed.
Genesis of the Beliefs
Not enough is known about the genesis of the beliefs, nor of the
conditions under which they are acquired. Moreover, no research has
been done on how a n individual’s position on the three health beliefs
is related to other comparable beliefs he holds. The potential value of
the model would be greatly enhanced if the origins and development
of the health beliefs were specified and if the beliefs were placed
within a broader theoretical framework that would account for
fesponses to a wide variety of stimuli.
Gochman 4 1 9 4 6 9 4 7 has initiated efforts to study the genesis of the
determinants of the health beliefs and their correlates. Haefner et al,J1
have made a beginning toward identifying the correlations among
preventive orientations toward three health conditions. More work in
both these areas is needed.
The Need for Additional
Experimental Studies
Convincing demonstrations of cause and effect can rarely, if ever, be
provided through cross-sectional surveys of the kind typically
employed to study the model. This is true because of the survey’s
susceptibility to errors in judging which of two associated factors

316 Health Education Monographs VOL. 2, NO.4

preceded the other in time and because the possibility is great that
apparent relationships may be spurious. For these reasons,
experimental studies must be undertaken to determine the causal role
of the relevant health beliefs. For example, a n effort could be made to
modify the health beliefs of a randomly assigned experimental group
while holding constant the beliefs of a comparable control group. Both
groups would then be offered a particular health service and
observations taken of the relative responsiveness of the groups to the
health appeal. A variety of specific experiments could be devised to
assess the contribution of the health beliefs to behavior.
As indicated earlier, Haefner and Kirscht 9J did conduct such a n
experiment and showed positive findings. But such research must be
repeated.

Susceptibility of the
Beliefs to Modification
Even if the model did predict behavior, its ultimate usefulness would
depend upon the extent to which the health beliefs can be modified in
a planned way. Three efforts to attempt such change have been re-
ported. Guskin,'* through the use of a film succeeded in modifying the
reported beliefs of fifth and sixth grade students relative to their
perceived susceptibility to and severity of tuberculosis, although no
changes in perceived benefits took place. In a study of fear arousal
and persuasion, which will be discussed in some detail in a subsequent
chapter, Haefner '9 obtained data which showed that the health
beliefs of ninth graders could be modified. High fear messages tended
to have more favorable effects on beliefs about severity and
preventibility (benefits) than did low fear messages. One of two effects
was observed: (1) high fear messages led to a greater increase in each
of the two beliefs than low fear messages, or (2) high fear meassages
led to a smaller reduction in the beliefs than did low fear messages.
Results for perceived susceptibility were not clear; in one
experimental treatment a high fear message led to a greater increase
in perceived susceptibility than did a low fear message while in a
second experimental treatment, a high fear message resulted in no
change or even led to a reduction in perceived susceptibility.
Finally, a s indicated, Haefner and Kirscht 93 were able to modify
beliefs and concomitant preventive health behavior.

UNIVERSALITY OF MODEL
Voluntary, Symptom-free Health Behavior
The model has been applied largely in situations in which the
behavior in question is purely voluntary and the individuals studied
do not believe themselves to have disabling symptoms. These criteria

Health Education Monographs Winter 19 74 311

are not met in a variety of situations in which people obtain health
services. For instance, social pressures may be effective in stimulating
action. Legal compulsion and job requirements also account for much
“health behavior.” Finally, the appearance of clear symptoms is a
most frequent instigator to health action. The likelihood is, therefore,
that only a minority of the population currently takes voluntary
preventive action or action to detect disease in the absence of distinct
symptoms. Despite these facts, continued work with the model may
have great ultimate benefit. The aim in public health education is to
increase the proportion of people who consistently, rationally and
freely take preventive actions or actions to check on the presence of
disease while free of symptoms. Careful analysis of the health
decision processes in what is currently a small group of people may
well be useful in subsequent planning efforts to modify the behavior of
very large groups of people. Studying the exceptional case may have
vast practical implications for working with the more typical.
Moreover, work cited in the following chapters by Becker and by
Kirscht suggest the Health Belief Model may have direct applicability
to illness behavior and to sick-role behavior.
Health Beliefs and Social Class
The Health Belief Model would seem to have greater applicability
to middle class groups than to lower status groups since possession of
the health beliefs implies a n orientation toward the future, toward
deliberate planning, toward deferment of immediate gratification in
the interest of long-run goals. The fact has frequently been
debated 6 O ~ 6 1 whether lower status people accord greater priority to
immediate rewards than to long-range goals. This difference in the
time orientation of the different social classes may well have
implications for the planning of preventive health programs. But these
implications are far from obvious ones. Hochbaum and Kegeles, in
earlier cited studies, have indeed shown that social classes differ in the
frequency with which the beliefs are held. But they have also shown
that where the proper constellation of beliefs exists, the probability is
greater that the recommended behavior will occur irrespective of
social class. Thus, public health workers must recognize that members
of the lower social classes are not as prone to accept health beliefs of
the kind described as are members of the higher classes. But they must
also recognize that many members of the lower classes do accept such
beliefs, indicating their ability to adopt a long-range perspective.
Subjective time horizons are thus not immutable.
Health Habits
Another possible limitation in the ultimate applicability of the
model is in the case of habitual behaviors and in styles of behavior.

318 Health Education Monographs VOL. 2, NO. 4

Patterns of behavior that are developed in early life most likely are
not motivated by the kinds of health concerns that may guide the
adult’s behavior. During the socialization process, children learn to
adopt many health related habits and practices which will
permanently influence their adult behavior, e.g., brushing teeth,
visiting the physician or dentist regularly, and adopting unique
nutritional practices. Yet, these patterns of adjustment cannot be
explained by applying the Health Belief Model. Clearly, the entire
area of the determinants of health related habits is worthy of detailed
investigation.

THE RELATIONSHIP BETWEEN HEALTH

BELIEFS AND DEMOGRAPHIC FACTORS
Typical demographic analysis of utilization rates was previously
criticized, partially on the grounds that few attempts have been made
to show the mechanisms that link behavior with fixed, personal
characteristics. However, two published studies are relevant in this
connection. Kegeles e t al, investigated relationships among the use of
Papanicolaou tests, demographic factors and beliefs in the benefits of
early detection of cancer. Beliefs in benefits were measured by
responses to questions on the perceived importance of early versus
delayed treatment for cancer and on opinions a s to whether medical
check-ups or tests could detect cancer before the appearance of
symptoms. An analysis of the findings discloses that personal
characteristics and beliefs each make independent contributions to the
understanding of behavior. Tests were much more likely to have been
taken by women who were relatively young, age 35-44,white, of higher
income, married, relatively well educated, and who reported higher
occupational levels (using husband’s occupation in the case of married
women).
The study also showed that accepting the benefits of early
professional detection and treatment was highly associated with
having taken the test. However, the joint analysis is of most interest.
Within every demographic grouping those who held a belief in benefits
were much more likely to have taken the test than those not holding
that belief. Similarly, within each of the belief categories those with
the appropriate demographic characteristics were much more likely to
have taken the action than those who did not. Clearly, the joint effects
of the beliefs and the personal characteristics is much greater than the
effects of either alone.
In Hochbaum’s earlier study 94 a similar finding was obtained.
Socioeconomic status (education and income) and the combination of
beliefs in susceptibility and benefits were independently associated
with having taken voluntary chest X-rays in the absence of symptoms.
Within each socioeconomic status category, however, those who scored

Health Education Monographs Winter I974 379

high on the combination of beliefs were much more likely to have
taken the X-ray than those scoring medium or low.
An interpretation of the findings of the two studies suggests that
certain of the beliefs may be necessary for taking preventive or
screening tests, but that they are distributed unevenly in the
population, tending to be more prevalent among whites, females, those
of higher socioeconomic status, and the relatively young. Why this is so
is not yet known.
INDUCING BEHAVIORAL CHANGE
The major focus in this paper has been on identifying factors that
help to explain why people use health services. Since, however, the
ultimate aim of understanding behavior in the health area is a n
applied one, the problem of persuading people to use health services
may appropriately be considered.
Material presented earlier indicates that a decision to take a health
action is influenced by the individual’s motivation, his perceived
susceptibility to illness, the perceived severity of illness, socially and
individually determined beliefs about the efficacy of alternative
actions, psychological barriers to action, interpersonal influences, and
one or more cues or critical incidents which serve to trigger a response.
No a priori reason may be found to indicate that action directed
toward any one of these will in the long run prove more effective than
action directed toward the others. Therefore, action programs to
modify behavior could legitimately focus on any one or more of the
determinants. Only systematic investigation will demonstrate the
conditions under which one or another of the determinants is most
susceptible to effective manipulation.
Despite the lack of definitive research findings, a few practical
considerations may clarify the problem. Ordinarily, to change people
is much more difficult than to change their environment (though the
latter may itself represent no simple task). Therefore efforts to
increase public response should always aim a t minimizing the barriers
to action, increasing the opportunities to act (which will increase
perceived benefits), and providing cues to trigger responses. Some
simple but important environmental features may be modified with
good effect, e.g., minimizing inconvenience by reducing financial costs
of services and distances that have to be traveled to obtain them, and
setting hours for service that are convenient. Moreover, cues may
frequently be arranged to trigger respones, e.g., reminders from
dentists and physicians, spot announcements in the mass media.
Fairly simple situational changes of the kinds described may well
increase the rate of preventive and diagnostic behavior. However, their
effect is probably limited, if current views of the determinants of
health behavior are a t all correct. Probably, after all situational

380 Health Education Monographs VOL. 2, NO. 4

improvements are made, a large number of individuals remain who
are not in a state of readiness to act, and, other things being equal,
Nil1 not act. Concerning such people, one must ask whether a direct
effort to increase their readiness can be successful and efficient or
whether success is more likely through a n indirect effort to stimulate
the behavior as, for example, through the use of social pressures.
Again, the question is empirical; research cited is only suggestive.
Definitive research has not been performed.
As a concluding section we may outline a program of research that
would seem to have great potential pay-off in answering many
unresolved questions a bout individual beliefs and behavior affecting
the public's health, and how beliefs and behavior may be modified.
The Study Population
A population group of family units in a selected geographic area
should be studied for a period of a t least ten or fifteen years. While no
"typical" population exists, the group selected should exhibit
heterogeneity with respect to economic and educational status and
racial and ethnic background. The group should be large enough to
permit definitive conclusions to be drawn about the three study topics
described below, after due allowance is made for inevitable sample
losses due to mobility, mortality and dropouts. At least 15,000 to
20,000 families should constitute the population of concern. This, of
course, means a community of 40,000 to 75,000 individuals.
The studies would have three major sets of objectives and activities,
each to be phased in gradually.

Descriptive Studies
Descriptive studies would focus on the nature of health motives,
beliefs, and practices of a population over the study period with
attention to changes that occur in beliefs and behavior as a
consequence of purely natural events including changes in age,
changes in health status, and changes in fads and fashions in
medicine.
( a ) The distribution and changes over time in preventive practices
should be studied, including such personal practices as exercise,
smoking, diet, toothbrushing, etc., as well as professional practices,
including frequency of visits to various professionals for check-ups in
the absence of symptoms. The importance of habits vs conscious
decision processes should also be investigated.
( b ) In addition, behavior in response t'o symptoms should be investi-
gated to identify pathways through which various groups travel
between initial experience of a symptom and the ultimate seeking of
help from others. Of special interest would be the use of self
medication - vitamins, tonics, tranquilizers, etc. - and the use of

Health Education Monographs Winter I 9 74 381

“others,” including non-professional, para-professional, a n d
professional.
(c) Finally, studies are needed of behavior undertaken in response to
perceived sickness. Included here might be the components of the ”sick
role” (avoidance of certain social role responsibilities, acceptance of
the idea that one needs help, a desire to get well, and a willingness to
seek and follow the advice of competent help). Of particular
importance in this connection would be studies of compliance with
health recommendations.
We would also wish to describe how individuals are currently
involved in planning for and operating health delivery systems and
how such involvement changes over time.
The descriptive studies are necessary for two reasons: (1) to provide
a baseline of data on what various sub-groups of people currently
believe and how they behave, and ( 2 ) to provide a beginning for the
study of the natural history of health beliefs and practices, that is,
toward a n epidemiology of health behavior.

Explanatory Studies
It is important not merely to describe but to explain health behavior.
We need to learn first how much of behavior that affects health and
health practice is attributable to long standing habits which may have
been acquired for reasons unrelated to health. And we need to learn
when and under what circumstances these habits begin to be “stamped
in.” Concerning behavior that cannot be classified as purely or
primarily habitual we need to test hypotheses to account for practices.
I n the area of health behavior there is no absence of hypotheses which
attempt to link health practices to certain predisposing socio-
psychological conditions. The following eclectic list of classes of
variables included theories proposed by a wide variety of social
psychologists and sociologists.
The general hypothesis to be tested is that people will undertake
action to prevent illness, will respond to symptoms in specified ways
and/or will follow professional recommendations if ( a ) they exhibit a t
least moderate health motivation, ( b ) they believe they are vulnerable
to or currently suffering from serious problems, ( c ) they believe the
potential or existing problem to be preventable or controllable, ( d )
their subjective time horizons are long enough for them to justify
taking immediate action to ward off future health threats or impacts,
(e) their social groups sanction the use of particular health treatments
or providers, (f) they are willing to overcome the barriers involved in
taking certain personal health actions, entering the professional
delivery system and remaining in it, and (g) they are willing to follow
professional advice.
We know that people vary widely in their positions on each of the

382 Health Education Monographs VOL. 2, NO. 4

foregoing seven classes of predisposing psychological factors and more
often than not their behavior will reflect a particular position on one
or several of the factors. We do not know, however, how these
psychological factors change over time and how such naturally
occurring changes in the factors are reflected in subsequent behavior.
I t is also necessary to learn how these social psychological factors are
influenced by cues coming from the social or physical environment,
such as symptoms, the occurrence of disease in significant others, the
mass media, and personal influence and group dynamics.
Longitudinal studies, using families a s units, would permit a test of
the hypothesis that these seven classes of belief or orientation are
associated with the kinds of health practices people engage in and, if
so, how naturally occurring events influence those beliefs and
subsequent practices.
The design would permit retesting of segments of the same sample
and of testing equivalent (randomly drawn) samples of the population.
This feature will permit the development of reliable measures of socio-
psychological factors associated with health behavior, using such
approaches as test-retest, split-half and comparable forms.

Experimental Studies
While descriptive and explanatory studies of health behavior are
important they fail to be completely persuasive since the causal
association between beliefs and behavior is not fully tested.
Furthermore, even if we were persuaded that these beliefs do partly
cause subsequent behavior it would still be important to know
whether they can be modified by outside intervention and whether
habits and practices can be taught or modified directly without
resorting to efforts to change people's beliefs. Therefore cooperative
field experiments with service providers should be conducted on
selected sub-groups in the population directed toward teaching or
modifying selected health practices, for example, dietary practices,
smoking practices, volunteering for multiphasic screening, increasing
compliance with health regimens and the like. Two distinct
approaches should be tried. The first would attempt to teach or
modify practice in children and adults by altering the relevant socio-
psychological factors. Persuasive techniques would be used, delivered
both through the mass media and personal and group influence.
'

Among the persuasive techniques that seem worthy of experimentation

are a number which have from time to time been shown to have some
impact on opinions and behavior. These could include such variables
as credibility of the communicator, the presentation of two sides vs one
side of a message, the selective use of print and broadcast media for
different topics and audiences, and the degree of specificity of the

Health Education Monographs Winter I9 74 383

action recommended. In addition, we need to experiment with
cumulative effects of exposure to a number of messages.
A second effort would concern direct attacks on the practices
themselves, both through the use of behavior modification techniques,
where appropriate and ethical, and through social engineering, that is,
modifying the social or physical environment to increase the
probability of desired behavior. Behavior modification techniques may
be particularly useful for those who are motivated to acquire or give
up a particular behavior but who lack the ability to do it themselves,
for example, smokers who wish to quit, obese persons who wish to lose
weight, sedentary individuals who wish to exercise. Such techniques
would also seem appropriate for use with school children where the
cooperation of the parent and the school can be obtained in the
interests of teaching or eliminating some behavior in the child.
Such experimentation would also permit the test of the hypothesis
that learning new complex practices will be enhanced if the desired
outcome behavior is subdivided into a series of small, manageable
steps, each of which is reinforced as it occurs. This, of course, reflects a
fairly basic learning principle most systematically used in
programmed learning but it appears not to have been systematically
employed in health education programs.

REFERENCES
1. Kasl SV, Cobb S: Health behavior, illness behavior and sick-role behavior. Arch
Environ Health I 12:246-266, February 1966 and I1 12:534-541, April 1966.
2. Lerner M, Anderson OW: Health Progress in the United States, 1900-1960: A Report
of the Health Information Foundation. Chicago, University of Chicago Press, 1963.
3. Somers HM, Somers AR: Doctors, Patients, and Health Insurance: The Organization
a n d Financing of Medical Care. Washington, T h e Brookings Institution, 1961.
4. United States Department of Health, Education and Welfare, Office of the Secretary:
Health, Education, and Welfare Trends, 1963 Edition. Washington, United States
Government Printing Office, 1963.
5. Borsky PN, Sagen OK: Motivations toward health examinations. Am J Public
Health 49:514-527, April 1959.
6. Kegeles S S , Kirscht JP, Haefner DP, Rosenstock IM: Survey of beliefs about cancer
detection and taking Papanicoloau tests. Public Health Rep 80:815-824, September
1965.
7. Rosenstock IM, Derryberry M, Carriger BK: Why people fail to seek poliomyelitis
vaccination. Public Health Rep 74:98-103, February 1959.
8. United States Department of Health, Education and Welfare, Public Health Service:
Health Statistics From the U S . National Health Survey: Dental Care, Interval and
Frequency of Visits, United States, July 1957-June 1959, Public Health Service
Publication Number 584-B14. Washington, United States Government Printing
Office, 1960.
9. Kegeles SS, Lotzkar S , Andrews LW: Dental care for the chronically ill and aged. 11.
Some factors relevant for predicting the acceptance of dental care by nursing home
residents. (Unpublished manuscript), United States Department of Health,
Education and Welfare, Public Health Service, 1959.
10. United States Department of Health, Education and Welfare, Public Health Service:

3 84 Health Education Monographs VOL. 2, NO. 4

 Health Statistics From the U S . National Health Survey: Volume of Physician Visits,
United States, July 1957-June 1959, Public Health Service Publication Number 584-
B19. Washington, United States Government Printing Office, 1960.
11. United States Department of Health, Education and Welfare, Public Health Service:
Health Statistics From the US. National Health Survey, Hospital Discharges and
Length of Stay: Short-Stay Hospitals, United States, 1958-1960, Public Health
Service Publication Number 584-B32. Washington, United States Government
Printing Office, 1962.
12. Antonovsky A, Hartman H: Delay in the detection of cancer: A review of the
literature. Health Educ Mongr 2:98-128, Summer 1974.
13. Herman MW: The poor: Their medical needs and the health services available to
them. Ann Am Acad Political SOCSci 399:12-21, January 1972.
14. Alpert J J , Kosa J , Haggerty RJ, e t al: Attitudes and satisfactions of low-income
families receiving comprehensive pediatric care. Am J Public Health 60:499-506,
March 1970.
15. Colombo T J , Saward EW, Greenlick MR: T h e integration of a n OEO health
program into a prepaid comprehensive group practice plan. Am J Public Health
59:641-650, April 1969.
16. Monteiro L: Expense is no object Income and physician visits reconsidered. J
Health SOCBehav 14:99-115, June 1973.
17. Koos EL: T h e Health of Regionville - What the People Thought and Did About It.
New York, Columbia University Press, 1954.
18. Stoeckle JD, Zola IK, Davidson GE: On going to see the doctor, the contributions of
the patient to the decision to seek medical aid, a selected review. J Chronic Dis
16:975-989, 1963.
19. Zola IK: Illness behavior of the working class: Implications a n d recommendations.
In Shostak A, Gomberg W (eds): Blue Collar World. Englewood Cliffs, New Jersey,
Prentice-Hall, Inc., 1964, pp 351-361.
20. Freidson E : Patients’ Views of Medical Practice. New York, Russell Sage
Foundation, 1961.
21. Suchman EA: Social patterns of illness and medical care. J Health Hum Behav 6:2-
16, Spring 1965.
22. Rosenstock IM: What research in motivation suggests for public health. Am J Public
Health 50:295-302, March 1960.
23. Rosenstock IM: Why people use health services. Milbank Mem Fund Q 44:94-127,
July 1966.
24. Hochbaum GM: Public Participation in Medical Screening Programs: A
Sociopsychological Study, Public Health Service, Public Health Service Publication
Number 572. Washington, United States Government Printing Office, 1958.
25. Kegeles S S : Some motives for seeking preventive dental care. J Am Dent Assoc
67:90-98, July 1963.
26. Heinzelmann F: Determinants of prophylaxis behavior with respect to rheumatic
fever. J Health Hum Behav 3:73-81, 1962.
27. Flach E : Participation in case finding program for cervical cancer. Administrative
Report, Cancer Control Program, United States Public Health Service, Washington,
D.C., 1960.
28. Leventhal H, Hochbaum G, Rosenstock I: Epidemic impact on the general
population in two cities. In T h e Impact of Asian Influenza on Community Life: A
Study in Five Cities. United States Department of Health, Education and Welfare,
Public Health Service, Pub No 766, 1960.
29. Kegeles SS: Why people seek dental care: A test of a conceptual formulation. J
Health Hum Behav 4:166-173, Fall 1963.
30. Kirscht JP, Haefner D, Kegeles S , Rosenstock I : A national study of health beliefs. J
Health Hum Behav 7:242-254. 1966.

Health Education Monographs Winter I 9 74 385

31. Haefner DP, Kegeles S, Kirscht J, Rosenstock I: Preventive actions in dental disease,
tuberculosis and cancer. Public Health Rep 82:451-459, 1967.
32. Festinger L: A Theory of Cognitive Dissonance. Evanston, Illinois, Row Peterson,
1957.
33. Haefner DP, Kirscht JP: Motivational and behavioral effects of modifying health
beliefs. Public Health Rep 85:478-484, June 1970.
34. Becker M, Kaback M, Rosenstock I, Ruth M: Some influences on public
participation in a genetic screening program. ( I n press) Community Health 1, 1975.
35. Heinzelmann F, Bagley RW: Response to physical activity programs and their effects
on health behavior. Public Health Rep 85:905-911, October 1970.
36. Fink R, Shapiro S, Roester R : Impact of efforts to increase participation in repetitive
screenings for early breast cancer detection. Am J Public Health 62:328-336, March
1972.
37. Battistella RM: Limitations in use of the concept of psychological readiness to
initiate health care. Med Care 6:308-319, July-August 1968.
38. Battistella RM: Factors associated with delay in the initiation of physicians’ care
among late adulthood persons. Am J Public Health 61:1348-1361, July 1971.
39. Suchman EA: Health attitudes and behavior. Arch Environ Health 20:105-110,
January 1970.
40. Ogionwo W: Socio-psychological factors in health behavior: a n experimental study
on methods and attitude change. I n t J Health Educ, Supplement to Vol. 16, No. 2,
April-June 1973.
41. Gochman DS: Some correlates of children’s health beliefs and potential health
behavior. J Health SOCBehav 12:148-154, June 1971.
42. Kegeles SS: A field experimental attempt to change beliefs and behavior of women
in a n urban ghetto. J Health SOCBehav 10:115-124, June 1969.
43. Aiitonovsky A, Kats R: T h e model dental patient: An empirical study of preventive
health behavior. SOCSci Med 4:367-380, 1970.
44. Birch HG: T h e role of motivational factors in insightful problem-solving. J Comp
Psych 38:295-317, 1945.
45. Gochman DS: Children’s perceptions of vulnerability to illness a n d accidents. Public
Health Rep 85:69-73, January 1970.
46. Gochman DS: Children’s perceptions of vulnerability to illness and accidents: A
replication, extension, and refinement. HSMHA Health Rep 86:247-252, March
1971.
47. Gochman DS: Children’s perceptions of vulnerability to illness and accidents: a
vulnerability to health problems. Health Serv Rep 87:282-288, March 1972.
48. Guskin SL: T h e measurement of change in beliefs about tuberculosis. In Proceedings
of the First Annual Conference on Behavioral Aspects of Tuberculosis Control,
Atlanta, Georgia. Tuberculosis Branch, Communicable Disease Center, Public
Health Service, 1965.
49. Haefner DP: Arousing fear in dental health education. J Public Health Dent 25:140-
146, 1965.
50. Simmons OG: Social Status and Public Health. Social Science Research Council,
Pamphlet number 13, New York, 1958.
51. Green LW: Status Identity and Preventive Health Behavior. University of
California, Berkeley, Pacific Health Education Report No. 1, 1970.

386 Health Education Monographs VOL. 2, NO. 4