You are on page 1of 20

Psychological Bulletin Copyright 1989 by the American Psychological Association, Inc.

1989, Vol. 105, No. 1, 31-50 Q033-2909/89/S00.75

Effects of Personal Experience on Self-Protective Behavior

Neil D. Weinstein
Departments of Human Ecology and Psychology
Rutgers—The State University of New Jersey

This article seeks to further our understanding of self-protective behavior by examining the effects
of a particularly powerful stimulus to action: personal experience. It reviews the effects of automobile
accidents on seat belt use, criminal victimization other than rape on individual crime prevention
efforts, natural hazards experience on both natural hazards preparedness and compliance with evac-
uation warnings, and myocardial infarction on smoking. Theories suggesting mechanisms that could
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

link personal experience to behavior are described, and data concerning the effects of experience on
This document is copyrighted by the American Psychological Association or one of its allied publishers.

some key variables in these theories are discussed. Tentative propositions are offered to resolve the
many apparent discrepancies in this literature. These propositions concern the effects of experience
on risk perceptions, the influence of experience on risk salience, the specificity of responses to vic-
timization, and the duration of experience effects.

"How do you know the sky has fallen?" asked Henny Penny. "I saw of view, this analysis may ultimately improve campaigns to en-
it with my eyes," said Chicken Little. "I heard it with my ears. And
courage the adoption of precautions. Although we cannot vic-
a piece of it fell on my poor little head." (from "Chicken Little," a
folktale) timize people in order to get them to act, an understanding of
personal experience may direct us to ingredients that can make
Direct confrontation with the threat seems to be extraordinarily prevention programs more effective.
effective in breaking through the defensive facade that normally In the past, psychological research on victimization has fo-
enables a person to maintain an unwarranted but highly cherished
attitude of complacency. (Janis, 1967, p. 219)
cused mainly on problems of emotional adjustment (e.g., Ja-
noff-Bulman, 1985; Siegel, 1983; Titchener, Kapp, & Winget,
A major limitation to human ability to use improved flood hazard 1976). Much of the research has utilized special samples, such
information is the basic reliance on experience. Men on flood as victims who have sought out counseling or have chosen to
plains appear to be very much prisoners of their experience, and
report crimes to police' (e.g., Friedman, Bischoff, Davis, & Per-
the effect of such experience is not consistently in the direction of
taking. . .action to reduce flood damage. (Kates, 1962, p. 140) son, 1982;Shapland, 1984). These investigations have generally
looked at extremely traumatic events, including rape and para-
As these quotations suggest, personal experience is widely be- lyzing injuries (e.g., Burt & Katz, 1985; Bulman & Wortman,
lieved to have a powerful impact on the recognition of risk and 1977; Frey, Rogner, Schuler, & Korte, 1985). The conclusions
the willingness to take precautions. The flurry of interest in pre- derived from such topics and samples may not apply to the
vention that seems to follow disasters is viewed as evidence of many incidents of victimization that do not shatter victims'
the effects of experience. And when people fail to take precau- views of themselves or their world.
tions, this, too, is attributed to experience, because the lesson The hazard experiences selected for review in this article—
we learn about most hazards is that they do not happen to us. automobile accidents, criminal victimization other than rape,
Nevertheless, many studies report that victimization has no natural disasters, and myocardial infarction—all have substan-
effect on behavior (Burton & Kates, 1964; Drabek, 1986; Du- tial literatures. It is important to examine several different haz-
Bow, McCabe, & Kaplan, 1979; Perry & Mushkatel, 1984; ards because each has unique qualities, and focusing on any
Quarantelli, 1980;Saarinen, 1982). single hazard would yield misleading impressions. The first part
The goal of this article is to study the effects of experience in of this article reviews the effects of these hazard experiences on
detail, asking what effects do occur and when they appear. Such self-protective behavior. The next section examines the serious
an inquiry can benefit both theory and practice. An understand- methodological problems in this literature. Third, the paper
ing of the powerful but apparently complex effects of personal discusses possible mechanisms for the effects of experience.
experience may shed light on the determinants of self-protective This section links personal experience to key variables in the
behavior and sharpen pertinent theories. From a practical point principle theories of self-protective behavior and points out re-
search not concerned with hazards that suggests a special im-
pact for personally experienced events. This third section also
reviews some of the empirical evidence linking hazard experi-
The author would like to acknowledge the hospitality of the Depart- ence to these intervening variables. Finally, some tentative con-
ment of Psychology, University of Arizona, where this article was writ- clusions are offered about the effects of experience on precau-
ten, and would like to thank Lee Sechrest and an anonymous reviewer tions and about the mechanisms responsible for these effects.
for helpful comments.
Correspondence concerning this article should be addressed to Neil
D. Weinstein, Department of Human Ecology, Cook College, Rutgers ' Roughly half of all crimes, including crimes of all types, are never
University, P.O. Box 231, New Brunswick, New Jersey 08903. reported to police (U.S. Department of Justice, 1986).
32 NEIL D. WEINSTEIN

Review of Empirical Research Weir, 1967;Miransky&Langer, 1978; Tyler, 1980, pp. 18-19)
conclude that experience has no effect. Large-sample surveys
This section summarizes data from five research areas con- often find significant but modest effects on behavior (e.g., Garo-
cerning the influence of hazard experience on precautionary falo, 1977; Hough, 1985; Skogan, 1977, 1987), the actual size
behavior. Four reviews are presented: effects of automobile acci- of the effect varying with the type of crime examined. In-depth
dents on seatbelt use, effects of criminal victimization (other studies of the victimization experience (e.g., Berg & Johnson,
than rape) on crime prevention efforts, effects of natural hazard 1979; Friedman et al., 1982; LeJeune & Alex, 1973; Maguire,
experience on preparedness for future disasters, and effects of 1980) report large changes in precautionary behavior, but the
natural hazard experience on responses to warnings of immi- studies often lack control groups. Household protection mea-
nent natural disaster. In addition, the effects of myocardial in- sures are usually reported separately from behavior changes
farction on smoking behavior, recently reviewed by Burling, that are intended to reduce the risk of street crime, but no con-
Singelton, Bigelow, Baile, & Gottlieb (1984), are briefly dis- sistent differences in the relationships between these two catego-
cussed. ries of precautions and experience are apparent.
An attempt has been made to locate all pertinent empirical
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

investigations, though some have undoubtedly been overlooked.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Natural Disasters and Natural Hazard Preparedness


Studies are excluded from this review if their evidence is solely
anecdotal; if they do not present the data upon which their con- Nearly all studies show preparedness increasing with the se-
clusions are based; if it is not clear what comparisons (i.e., verity of past damage. Exceptions (Kunreuther, 1978, no effect
groups or variables) are used to reach stated conclusions; or if of earthquake experience; Perry & Lindell, 1986, no effect in
measures of experience or behavior appear too confounded to the less-affected of two communities studied) seem to reflect
provide useful information. This review focuses on personal ex- situations in which past disaster experience produced relatively
perience, a reasonably well-defined concept. In contrast, studies little damage. Some evidence from insurance policy purchases
of indirect experience have examined a range of heterogeneous (Baumann & Sims, 1978; Institute of Behavioral Science, 1973;
phenomena, including effects of the mass media, beliefs about Kunreuther, 1978) indicates that sales increase for only a few
neighbors' experiences, victimizations incurred by other house- months after hazard impact. However, there are also claims that
hold members, and the impact of observing another person vic- attempts to generate interest in natural hazards insurance after
timized. No attempt is made here to review studies dealing with a disaster have been unsuccessful (Syfert, 1972).
indirect experience (see Tyler, 1983).
Tables 1-4 present the results of the literature review. Unless
Natural Hazard Experience and Response to
otherwise indicated, (a) all data have been obtained through
Warnings of Imminent Disaster
self-report; (b) there is a no-experience control condition (in
some cases, no-experience refers only to a lack of experience Most studies report no effect of prior experience (primarily
during a specified time period); and (c) all results listed are sta- from hurricanes and floods) on response to evacuation recom-
tistically significant beyond the .05 level. It should be noted that mendations. However, anecdotal evidence (Baker, Brigham,
several of the original reports did not provide statistical tests, Paredes, & Smith, 1976; Bates, Fogelman, Parenton, & Tracy,
and calculations carried out on the data presented in those re- 1963; Instituut voor Sociaal Onderzoek van het Nederlandse
ports sometimes contradicted the conclusions originally Volk, 1955; Wilkinson & Ross, 1970) suggests that people as-
reached by the authors. sume the approaching storm or flood will have the same magni-
Because of the many methodological problems in these inves- tude as the one previously experienced. People then take what-
tigations, it seems unwise to attempt a detailed analysis of indi- ever actions would have been appropriate for the prior event.
vidual studies or of discrepancies between studies. Instead, sub- They will be reluctant to evacuate if the previous storm was not
sequent paragraphs will briefly summarize the main findings in severe. The unusually high evacuation rate noted in Cameron
each hazard area, emphasizing those that seem relatively con- Parish (Moore, Bates, Layman, & Parenton, 1963) is not ade-
sistent. The apparent effects of hazard experience on precau- quately explained by past experience at the individual level. Al-
tions will need to be examined again after methodological issues though this community did have recent contact with a damag-
have been discussed, and we will return to the relationship be- ing hurricane, the number of individuals who had suffered per-
tween experience and behavior in the final section of this article. sonal damage in Cameron was not much greater than the
number in other sites who had suffered damage. Cameron's
Automobile Accidents and Seatbelt Use nearly complete evacuation probably reflects its many com-
munity-level preparedness activities and well-publicized evacu-
With few exceptions, investigators find no relationship be- ation plan.
tween personal accident or injury experience and seatbelt use.
Although self-reports are not a very reliable measure of seatbelt
Myocardial Infarction and Smoking Behavior
use, the studies based on observed belt use (Robertson, 1975;
Robertson, O'Neill, & Wixom, 1972) give the same results. The findings from 14 studies on this topic have been reviewed
recently by Burling et al. (1984). These authors report that
Criminal Victimization and Crime Prevention Measures roughly one third to one half of smokers who suffer a myocar-
dial infarction (MI) quit or reduce smoking. The quitting rate
Findings vary greatly from study to study. A substantial num- increases with the severity of the infarction. Follow-up periods
ber of investigations (e.g., Biderman, Johnson, Mclntyre, & varied from a few days post-Mi to at least 8 years. Recent stud-
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 33

Table 1
Effects of Automobile Accident Experience on Seat Belt Use

Study Sample Experience measure Behavior measure Result Comment

Fhaner & Hane 73 Swedish drivers Hypothetical driving 1 (almost never) to 5 Would use belts more No actual experience
(1973) conditions (almost always) use for city driving but involved. Accuracy of
including seat belts not for highway belt use under
"recently seen or driving hypothetical conditions
heard about an questionable
accident"
Manheimer, 5 1 8 owners of cars Close friend or Divided into regular Greater experience in Special sample since cars'
Mellinger, with seat belts relative seriously users, moderate regular users (68%) owners had to purchase
& Crossley injured in auto users, or light users than in moderate belts at time of study
(1966) accident on basis of users (64%) or light
responses to five users (58%)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

questions
This document is copyrighted by the American Psychological Association or one of its allied publishers.

concerning belt use


in different
conditions
Robertson 394 drivers Personal injury; Observed belt use on No effect Overall high use owing to
(1975) close friend or one occasion starter-interlock system
relative seriously in cars at time of study
injured; close
friend or relative
killed
Robertson, 548 drivers Personal injury; Observed belt use on Greater belt use if
O'Neill, & close friend or one occasion close friend or
Wixom relative seriously relative injured.
(1972) injured; close No effect of
friend or relative personal injury or
killed friend killed.
Runyan(1983) 3 1 8 graduate students Involvement in crash Percentage of trips No effect
posing possibility during which belts
of serious injury were used
or death to
respondent, family
member, or close
friend
Svenson, 1 69 American and Involvement in auto I (never) to 13 No effect Result provided by personal
Fischhoff, & 169 Swedish accident as (always) in various communication
MacGregor college students passenger or driver situations (e.g., as
(1985) driver, as passenger)
Weinstein 85 adults Serious auto injury Percentage of trips No effect Result not reported in
(1987b) assessed on 1 during which belts original publication
(don 't know were used
anyone this has
happened to) to 5
(has happened to
me more than
once) scale

ies not included in that review also show a high rate of smoking The discussion that follows may help readers identify the studies
cessation after infarction (e.g., Aberg et al., 1983; Perkins & whose conclusions are problematic and may encourage re-
Dick, 1985; Sivarajan et al., 1983; Vermeulen, Lie, & Durrer, searchers to avoid these problems in the future (see also DuBow
1983). Few of these investigations have non-Mi control groups, etal., 1979).
and one article (Scott & Lamparski, 1985) questioned whether
the long-term quitting rate is much different from that of people Preexisting Differences Between Victimized and
who did not suffer a heart attack. Still, the short-term quitting Nonvictimized Groups
rate is much higher than one would expect in a normal popula-
tion. Implicit in most of this research is the assumption that prior
to victimization, the preventive behaviors of the victimized and
nonvictimized groups were the same. In many instances, how-
Methodological Limitations
ever, the absence of protective measures may have contributed
There are serious methodological problems in this literature, to victimization. For example, Scarr's (1973) data suggest that
some related to the impossibility of random assignment to vic- burglary victims had fewer household security measures in
tim and nonvictim groups, others to measurement difficulties. place at the time of the burglary than did non victims. Similarly,
34 NEIL D. WEINSTEIN

Table 2
Effects of Crime Victimization on Crime Prevention Measures

Experience
Study Sample measure Behavior measure Result Comment

Berg & Johnson 298 people known Victimization Multiple questions 60- 70% (varying with Not clear how specific
(1979) from police (uncategorized) ranging from sample subgroup) precaution
files as victims home security reported as having questions were
of juvenile measures to behavior changes combined to yield
offenders street crime yes/no outcome for
avoidance behavior change.
measures No control group
Bider man, Johnson, 5 1 1 adults in Times ever Street crime No association
Mclntyre, & Weir three police victimized avoidance
(1967) precincts (uncategorized). measures (four
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Victimizations items); carrying


This document is copyrighted by the American Psychological Association or one of its allied publishers.

weighted by weapon
seriousness (self;
friend/relative
injured or
killed; observing
crime)
Clarke, Ekblom, 9, 150 Great Street crime Number of Victims went out Does not establish
Hough, & Britain victim in evenings out in more frequently whether victims
Mayhew(1985) residents 16 preceding year week preceding thn did nonvictims changed their
years and older interview behavior
Friedman, Bischoff, 274 New York Victims of assaults Questions about Precautions in homes No control group.
Davis, & Person City residents (8%), robberies precautions in added by 60%; 38% Wording of
(1982) who had (29%), or home and going said they went out questions not
reported crimes burglaries out at night less at night reported
to police (63%). Time
period unstated
Garofalo(1977). Approximately Type of Percentage yes Increase in percentage Nonspecific
Also reported by 22,000 people victimization responses to "In yes for victimized behavioral response
Hindelang, 16 years and in preceding general, have groups: assault, 8%; measure
Gottfredson, & older in each of year assessed by you limited or robbery without
Garofalo(l978) eight large detailed changed your injury, 12%;
American cities National Crime activities in the personal larceny
Survey protocol past few years with/without
because of contact, 8%;
crime?" household crime,
4%; all victims, 5%
Hough (1985) 17,000 people Burglary Leaving doors Percentage unlocked:
over age 1 5 in experience: in unlocked when recent victim, 4%;
England and previous year; home is empty 1-5 years ago, 8%;
Wales 1-5 years ago; during the day not in 5 years, 8%
not in last 5
years
Lavrakas(1980) 1,803 people in Victim of actual Street crime Correlation of Not clear if "crime
metropolitan or attempted avoidance avoidance scale victim" included
Chicago area break-in in past actions when with "crime break-in
couple of years. in own victim" significant experiences or just
"Crime neighborhood (r = .13); assault and robbery
victim" in past (three items, correlation of
couple of years alpha = .73). household scale
Household with break-in
protection scale experience not
(six items, significant (r =
alpha = .51) .025)
Maguire(l980) 322 people who Burglary within Purchases of: 43% of uninsured No control group
had reported a last 4- 10 weeks insurance or bought policies;
burglary more insurance; 42% of
new locks or underinsured
alarms. Less increased policies;
careless about 50% purchased new
locking doors locks or alarms;
and shutting 80% less careless
windows about locking doors
and windows
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 35

Table 2— Continued

Experience
Study Sample measure Behavior measure Result Comment

Midwest Research 1,399 burglary Offense Change in actions Increase in: locking No control group
Institute (1977) and personal approximately 2 from before to doors/windows
crime victims weeks after crime (15 when away (54%)
over 60 years of previously topics) and at home (7 3%),
age who (59% burglary better locks (53%),
reported the victims; 24% screens/bare on
offense to robbery) windows or doors
police (23%). Decrease in
walking alone
(26%) and going
out in evening
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(16%)
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Miransky & Langer 1 26 residents of Ever burglarized Reported and No effects Burglaries at previous
(1978) apartments observed door- homes included.
lock use Many burglars did
not enter through
door
Reppetto(l974) 685 randomly Burglary Question about More victims Burglaries occurred
selected changes in reported security over 2-'/>year
Boston security changes (43% vs. period. Length of
residents plus practices in past 19%). More special time after burglary
125 residents year. Short list lock use by victims interviews took
who had of home (63% vs. 48%) and place is unclear
reported security victims less likely to
burglary to equipment have nothing on list
police (18% vs. 36%).
Few had alarms
or window
bars
Rifai(1976) 430 randomly Ever been a victim Nonvictims asked More victims (39%) Very broad victim
selected of crime (58% about actions to than non victims category. Different
community "victims"; two- protect (23%) had taken time span for
residents over thirds property themselves. some action. victim and
60 years of age crimes) Victims asked Significant nonvictim actions
about actions differences in
since property marking
victimization. and installing better
Nine topics. locks. Victims
slightly higher
in all
categories
Scarr(1973) 90 victims of Burglary victim 2- List of 12 home At time of interview, Appears that victims
burglaries '/; to 3-1/! years security victims report started with fewer
reported to before measures greater use of dead precautions and
police matched interviews present at time bolt locks and bars/ ended with more,
with 90 ofburglary mesh on windows. but no statistical
non victims (victim group) Victims higher on test conducted and
living near or at time of 11/1 2 items, but no measure of
victims interview (both many differences change in control
groups) small group over this
time period
Skogan(1987) 1,738 residents of Personal crime Personal (street) Regression analyses Effects small: .05
seven Newark victim (robbery, crime avoidance controlled for units on personal
and Houston rape, assault, actions (four- precautions and action scale for
neighborhoods purse snatching) item, 4-point victimization personal crime
interviewed two or property scale). experience at first victimization; . 1 7
times, 1 year crime victim Household interview. units on household
apart (burglary, theft, security Victimization security scale for
vandalism, auto measures (six- significantly property
theft) item, 6-point increased the victimization
during scale) corresponding
interval precaution scale
between
interviews
36 NEIL D. WEINSTEIN

Table 2— Continued

Experience
Study Sample measure Behavior measure Result Comment

Skogan(1977) 50% subsample of Type of Response (yes/no) Correlation (gamma)


23,022-person victimization to question between
1973 Census in previous year about changing victimization and
Bureau survey or limiting your activity change:
of five largest activities rape, .51; robbery,
American cities because of .32; personal theft,
crime in past .40; assault, .09;
few years burglary, .14; auto
theft, .06
Skogan & Maxfield 540 residents Burglarized in Personal (street) Burglary experience
(1981) each in San past 2 years; crime avoidance correlated with
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Francisco, know local scale (three locks/bars on


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Chicago, and burglary victim; items). Locks/ windows and


Philadelphia number of local bars on surveillance
crime victims windows. (gammas = .14 and
known Marked .09). Number of
property. Theft local victims
insurance. Four- known correlated
item household with personal
surveillance avoidance scale
scale (e.g., lights ( r = .11). Knowing
at night; ask local burglary
neighbors to victim correlated
watch house) with window
protection, theft
insurance, and
surveillance
(gammas = .18,
.20.&.15)
Tyler (1980) Study 1:244 Los Number of actual, Household Beta coefficients: .17, Unclear how
Angeles attempted, and security p < .05, in Study 1; experience indices
residents. Study observed measures win Study 2 created.
2: 1618 burglaries, car assessed with Correlations among
residents of San thefts, thefts same seven individual
Francisco, from cars, items as in precaution items
Chicago, and robberies, Skogan & appear small
Philadelphia assaults, and Maxfield (1981).
personal larceny Six items to
both in and assess personal
outside protection. All
respondent's items combined
neighborhood
in past 5 years.
Separate
burglary and
forcible crime
indices were
eventually
combined

the data of Clarke, Ekblom, Hough, & Mayhew (1985) suggest confound disaster studies because increasing damage would
that street crime victims are more likely to be people who go mean not only a more severe experience but also fewer initial
out often. Failure to control for preexperience differences in precautions. However, in the flood or earthquake studies that
precautions will lead researchers to underestimate the conse- have assessed the amount of damage, few people had protective
quences of victimization. For example, if victims of auto acci- measures in place at the time of the impact that would have
dents use seat belts no more than do nonvictims, it could be significantly reduced their losses.
because they tended to use seat belts less often before the acci- Longitudinal research designs can control for preexisting
dent. It does seem plausible that high-risk drivers, ones who differences between groups, but longitudinal studies of unpre-
drive recklessly or drive while intoxicated, would be less likely dictable events such as crime, illness, accidents, and natural
to wear seat belts than would other drivers. hazards are difficult to carry out. In cross-sectional research, it
Lack of preparedness cannot produce a flood or an earth- seems preferable to ask respondents directly about changes in
quake, but it can increase the damage experienced. This could behavior or to ask when precautions were adopted rather than
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 37

just to ask about current behavior, but answers to these ques- you limited or changed your activities in the past few years be-
tions are subject to greater recall errors. Researchers should not cause of crime?"
assume that any changes reported were caused by the victimiza- The open-ended questions used in a number of natural haz-
tion experience; nonvictims may have changed just as much ards studies (e.g., Jackson, 1974; Perry & Lindell, 1986) seem
during the time interval examined (a weakness in the design ill-advised because the answers people give reflect both the ac-
used by Scarr, 1973). A control group is needed unless the pe- tions they have taken and also their recall of these actions. It
riod being studied is so short that it would be unreasonable to should not be assumed that the only respondents who have
expect any significant change in the nonvictim group. For ex- taken a particular measure are those who mention this mea-
ample, a control group does not seem essential for studies that sure.
look at smoking cessation in the month following a heart attack. Observations are not infallible. They, too, can be unreliable
Nevertheless, a study of insurance purchases in the weeks fol- if preventive behaviors are not stable over time. A single obser-
lowing a flood would need an unflooded control group from the vation of seat belt use, for example, is not a dependable indica-
same community to distinguish the effects of personal experi- tor of use on other occasions (Weinstein, Grubb, & Vautier,
ence from the effects of secondhand experience derived from
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

1986).
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the media and from victimized acquaintances.

Ambiguities in Time
Assessment of Past Experience Most seat belt and natural hazards studies fail to report the
length of time between victimization and the assessment of pre-
Accurate classification of victimization is surprisingly
cautions. Because there is evidence that the impact of experi-
difficult (Skogan, 1982). A large proportion of auto accidents
ence may be relatively brief (Baumann & Sims, 1978; Hough,
and crimes are forgotten within a year or two (Biderman et al.,
1985; Oliver, 1975), this is a serious omission. If effects of vic-
1967; Nilsson, 1980, summarized in Svenson, Fischhoff, &
timization are transient, delayed measurement of precautions
MacGregor, 1985). People may report having lived through a
requiring repetition will give the misleading impression that ex-
hurricane, when all they actually experienced were modestly
perience has no effect at all. Thus, a delay in assessment is likely
elevated winds from the edge of the storm (Levy & Smith, circa
to be more serious for a study of seat belt use than for a study
1979).
of fire extinguisher purchases, because the latter action need be
Although the severity of experience seems to be crucial in
taken only once.
determining future behavior, the degree of damage (e.g., dollar
Being too specific about time may also cause problems. In
cost) or harm (e.g., time in hospital, time lost from work) has
many studies, crime experience refers only to the previous 12
usually been ignored or only crudely measured, and experience
months. A household burglarized 14 months prior to the inter-
is commonly treated as an undifferentiated, all-or-none vari-
view is classified as inexperienced; one burglarized 10 months
able. For example, very different crime experiences are often
before the interview is labeled experienced. It seems unreason-
lumped together in a victim category.
able to expect substantial differences between the two. Further
Finally, experience is sometimes quite complex. Many resi-
clouding the relationship between experience and action, sev-
dents of communities along the Gulf of Mexico have lived
eral surveys conducted by the U.S. Bureau of the Census (Garo-
through half a dozen hurricanes of varying intensities. Re-
falo, 1977; Skogan, 1977) assessed criminal victimization dur-
searchers have not suggested how to summarize such experi-
ing the previous 12 months but asked about changes in behavior
ence.
"in the past few years."

Assessment of Precautions Confounding Variables

All but four of the studies in Tables 1-4 (i.e., Kunreuther, Most crime victimization studies examine a substantial geo-
1978; Miransky & Langer, 1978; Robertson, 1975; Robertson graphical region, such as a whole metropolitan area, that con-
et al., 1972), assessed precautions by self-report. In some in- tains neighborhoods with very different crime rates (exceptions
stances, the precaution is a clear-cut behavior, and self-reports are Hough, 1985, and Scarr, 1973). Positive correlations be-
seem reasonably trustworthy. Evacuation is an example. For tween experience and precautions might simply indicate
other actions—such as stopping smoking, wearing a seatbelt, or greater crime problems in some areas than others, not effects of
planning what to do in the event of a flood—there is a real dan- experience at the individual level. Income is another potentially
ger that respondents will exaggerate their adherence to recom- confounding variable. High income levels are usually associ-
mendations. Some disaster preparedness actions—such as ated with lower rates of victimization (the wealthy are less likely
keeping a flashlight, a portable radio, and extra food in the to live in floodplains or high-crime areas) and with an increased
home—may be taken for other reasons, but respondents may ability to pay for precautions (Skogan & Maxfield, 1981). This
label them as safety measures in order to impress the inter- could make it appear that low levels of victimization cause an
viewer. Some self-report errors are simple misunderstandings: increase in precautions.
homeowners may be mistaken about whether or not their insur- Another problem in the crime literature is the fact that there
ance policy covers flood or earthquake damage. The crime pre- are specific sex and age profiles for some types of victimization.
vention behavior questions used in many studies are vague (Du- Victims of "personal larceny with contact" (e.g., purse snatch-
Bow et al., 1979). For example, there can be honest differences ing) tend to be older; victims of weapons crimes tend to be
of opinion over what warrants a yes answer to the query, "Have young men. The apparently large impact of personal larceny
38 NEIL D. WEINSTE1N

Table 3
Effects of Natural Disaster Experience on Natural Hazards Preparedness

Study Sample Experience measure Behavior measure Result Comment

Baumann & Sims 144 homeowners Degree of damage in Purchase of flood insurance Increase in percentage Before flood, insurance
(1978) in flood plain recent flood purchasing coverage near zero
that recently insurance with though available 2
experienced damage: no damage, years. Coverage rose
serious flood 7%; not very to 1 7% within 6
serious, 43%; fairly months after flood,
serious, 53%; very but grew only
serious, 86% slightly thereafter
Jackson (1974, 302 homeowners No earthquake Adjustments mentioned in Increasing preparatory Open-ended behavioral
1977, 1981) in three cities experience or response to open-ended actions (for measure. 93% of
of high minor tremors; question about things immediate impact sample had taken no
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

seismic risk major quake with that can be done to period, immediate or one preparatory
This document is copyrighted by the American Psychological Association or one of its allied publishers.

loss < $ 1 ,000; prepare for earthquake recovery period, and action
major quake with plus specific question total preparatory
loss >$ 1,000 about eathquake actions) with
insurance severity of prior
experience
Kunreuther 1 ,006 residents Value of damage Special insurance for Insurance purchases Substantially higher
(1978) of earthquake caused by appropriate hazard increased with prior damage in
risk areas; previous hazard determined from degree of flood floods (17. 1%>
1,413 experience (five insurers files. Open- damage but not with $500) than in
residents of flood damage ended question about degree of earthquakes (6.6% >
coastal categories; four protective actions taken earthquake $500). If losses are
hurricane earthquake damage. Flood small, insurance may
flood risk damage experience led to not pay because of
areas; 642 categories) more protective deductible clauses
residents of measures (40% vs.
riverine risk 23%) but little effect
areas. About of earthquake
half of each experience
group
possessed
special
insurance for
the hazard
involved
Perry & Lindell 202 residents of Degree of damage: Open-ended question asked Positive correlation Damage moderate to
(1986) two towns that none, slight, for measures taken to between damage severe in the
experienced moderate, or protect themselves from and number of community with
damage from severe future threats from Mt. adjustments in one significant
Mt. St. Helens St. Helens community, not correlation, slight to
eruption significant in the moderate in the
other other
Rydant(1979) 125 farmers Value of crop loss Having crop-hail insurance No association Only 16 respondents
from prior hail did not have
storms (none to insurance, so
over $2,000) statistical tests weak
Schiff(l977) 328 Ontario Experience (self = 3, Total scores on 22-item Positive correlation No evidence that scales
residents spouse, children = natural hazard between total composed of such
2, other = 1, adjustment list and 25- experience and total diverse items have
none = 0) on 30 item nonhazard adjustments for internal consistency.
hazards. adjustment list (ranging both natural and Some so-called
including natural from windshield scraper nonnatural hazards adjustments may not
hazards, accidents, in car to annual medical reflect hazard
financial and checkups) preparedness
health problems)
Turner, Nigg, Oaz, 1,450 Los Prior earthquake 1 6-item checklist of Preparedness Strength of association
& Young Angeles experience: none; earthquake preparedness increased with same for both
(1981a, 1981b) County damaging quakes kept on hand "because of earthquake experience
residents but no harm; any future quake." experience, measures. Overall
personal harm Respondents divided into earthquake low level of damage.
from earthquake. four groups damage, and other Many
Degree of damage hazard experience "preparedness"
indicated on five items (transistor
yes/no questions radio, extra food)
might be available
for other reasons
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 39

Table 3—Continued

Study Sample Experience measure Behavior measure Result Comment

Waterstone (1978) 1 74 residents of One point each for: Presence of: emergency Significant increase in Specific experience
100-year personally experi- evacuation plan, flood preparedness with items associated
floodplains in enced a flood, insurance, any flood- experience at both with preparedness
two areas. damage in flood, proofing measure, had sites were different at the
Parts of areas other natural taken or would be willing two sites
flooded 4 disaster experi- to take public action to
years earlier ence, knew deal with floods
someone who
experienced a
damaging flood,
knew someone
with damage in
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

other natural
This document is copyrighted by the American Psychological Association or one of its allied publishers.

disaster, heard
about the possi-
bility of flooding
in home area
through second-
hand experience

with contact on preventive measures (Garofalo, 1977) reflects crimes that did not take place inside the victim's residence or
the sex and age characteristics of the victims more than it re- workplace, that had involved actual contact between criminal
flects the impact of victimization. When these demographic and victim, that were not merely threats, and that were commit-
differences are controlled, the apparent effects of this crime ted by individuals who were either strangers to the victims or
change dramatically (Skogan & Maxfield, 1981). known only by sight. To identify such incidents obviously re-
quires a large and detailed data set.
Experience-Precaution Correspondence Precision like this is rare. Studies of natural hazards and
crime, in which the location of victimization seems particularly
The final problem in this literature is more a matter of con-
important, frequently do not ascertain whether the respon-
ceptualization than methodology. It concerns the correspon-
dent's current residence is the same as the one in which the
dence between the particular victimization experience and the
experience occurred. A person with flood experience may have
type of precaution chosen for study (see Fishbein and Ajzen,
derived that experience in another part of the country. Investi-
1974, and Skogan, 1981, for related discussions). Would we ex-
gations of responses to criminal assault usually ignore the fact
pect a woman who is assaulted inside her place of work to re-
that assaults are often committed by acquaintances or relatives.
frain from walking outside her home at night? If a flood pro-
Most studies of household locks and window bars forget that
duced property damage, but no personal injury, should the in-
thefts from homes are frequently committed by visitors and do
fluence of this experience be evaluated solely by whether or not
not involve forcible entry.
people evacuate when another flood approaches?
A generalized response to victimization, in which case preci-
There is a range of plausible hypotheses concerning the effects
sion would not be needed, would be indicated by the adoption
of experience. These hypotheses can be arranged in a contin-
of a variety of precautions. Natural hazard researchers some-
uum reflecting the specificity of the effects expected. The non-
times combine questions about protective measures to produce
specific end of this range of hypotheses reflects the notion that
victimization leads to generalized feelings of vulnerability and an index of preparedness, but they have not published the corre-
prompts a broad array of self-protective behaviors, including lations among the items that go into their scales. Crime re-
ones not directly linked to the victimization experience. If this searchers sometimes do report such statistics. They find that
notion were correct, auto theft victims would adopt security interitem correlations are often small, even within a limited do-
measures to protect all of their property and might take mea- main of action. For example, questions about household protec-
sures to avoid street crime as well. tive measures show correlations averaging around 0.1 (Lavra-
The specific end of the continuum suggests that an incident of kas, 1980; Tyler, 1981; Wilson & Schneider, 1978). Questions
victimization leads people to adopt those particular precautions referring to caution when traveling in one's neighborhood at
they believe will prevent a repetition of that type of victimiza- night show greater consistency, with correlations around .45
tion. According to this hypothesis, if an automobile is stolen, (Lavrakas, 1980; Skogan, 1987; Tyler, 1981), but such items
the owner would be inclined to install an auto theft alarm but correlate poorly with questions about daytime caution. Across
not a home burglar alarm. Tests of such specific hypotheses domains—that is, home security devices, home security behav-
need to be very precise about the nature of the victimization iors, street crime avoidance behaviors, carrying of weapons,
incident being studied. For example, in their investigation of the self-defense skills, and involvement in community crime-pre-
effects of street crime experiences on the willingness to travel vention activities—correlations are quite small (Furstenburg,
in the community, Clarke et al. (1985) restricted themselves to 1972; Tyler, 1981).
40 NEIL D. WEINSTE1N

Table 4
Effects of Natural Disaster Experience on Responses to Warnings of Imminent Natural Hazards

Study Sample Experience measure Behavior measure Result Comment

Baker, Brigbam, 205 residents of town Experienced previous Followed No effect Apparently, previous
Paredes, & hit by hurricane hurricane evacuation hurricanes were mild
Smith Eloise recommendation
(1976)
Danzig, Thayer, & 146 residents of town Living in area just Evacuation in Just-flooded-area Rumor may have been
Galanter(1958) partially flooded in flooded response to residents more heard by more residents
previous week rumor of dam likely to evacuate of the just-flooded area
breaking or heard earlier (i.e.,
prior to official denials)
Hanson, Vitek, & 555 people living Experienced previous Take shelter in No effect
response to
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Hanson (1979) along route of tornado


This document is copyrighted by the American Psychological Association or one of its allied publishers.

destructive 1953 tornado warning


tornado broadcast
Hutton(1976) 189 people who had Severe prior flood Evacuation in Greater evacuation No definition of "severe"
been living in the experience vs. response to 1972 rate in severe prior- given
flood plain at time nonsevere or no flash flood experience group
of 1972 Rapid prior flood warnings
City, SD, flash experience
flood
Moore, Bates, 1 ,500 residents in Experienced previous Evacuated home to Percentage leaving Evacuation rate of people
Layman, & five localities along hurricane; other location in community: no with nonhurricane
Parenton(1963) Gulf Coast warned experienced other community; experience, 35%; hazard experience, 53%.
to evacuate natural hazard evacuated hurricane Many community-level
because of community; experience, 48%. hurricane preparedness
approaching stayed home Nearly total activities in Cameron
hurricane Carla evacuation from Parish
one locality,
Cameron Parish,
that had recent,
severe hurricane
experience
Perry, Lindell, & 553 residents of four Experienced previous Evacuated; No effect when data
Green (1981) communities with flood protected from towns
separate recent property, no combined
flooding events evacuation;
normal routine
Smith & Tobin 262 residents living Experienced previous Hypothetical Inexperienced more
(1979) in flood plains of flood response to likely to say they
two communities official 6-hr would use sandbags
warning in the or evacuate;
future experienced more
likely to say they
would move
valuables to safer
location and keep
watch
Wilkinson & Ross 384 coastal residents Experienced previous Evacuation No effect Previous hurricanes in
(1970) living in areas hurricane. 1947,1956, 1957, 1961,
warned to Number of 1964, 1965, 1967. Only
evacuate from hurricanes two produced
hurricane Camille experienced. Had appreciable damage.
lost "anything" in Only 20% of
previous respondents had any
hurricanes damage in prior storms
Wind ham, Posey, 378 residents of two Experience of Evacuation from No effect of hurricane One area had no hurricane
& Spencer communities previous approach of experience or in past 40 years. The
(1977) threatened by hurricane. Injury Eloise damage other had carried out an
hurricane Eloise or significant evacuation 3 years
damage in previous earlier that proved
hurricane unnecessary. Only 10%
reported moderate or
severe property damage
in a prior hurricane
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 41

Such low associations indicate that generalized responses to benefits associated with the available precautions (e.g., Ajzen
crime are uncommon. Nevertheless, many studies still use a & Fishbein, 1980; Janz & Becker, 1984; Rogers, 1983; Sutton,
simple victim-nonvictim dichotomy, totally disregarding the 1982). In the context of this theory, experience is primarily in-
nature of the victimization. In fact, DuBow et al. (1979) com- formational. Key issues are as follows.
ment in their review that all of the studies reporting no effect of Perceived vulnerability: The likelihood of harm if there is no
victimization on behavior had combined victims into a single change in behavior. Experience may provide information that
category. One reason why studies interested in the trauma of affects the perceived likelihood of victimization, especially per-
victimization (e.g., Berg & Johnson, 1979; LeJeune & Alex, ceptions of one's personal likelihood of victimization. Osberg
1973) may find greater impacts on behavior than do large-scale and Shrauger (1986) found that people use past experience to
victimization surveys is because they are more likely to ask predict the future more than they use such other factors as cir-
questions about precautions related to the specific crime expe- cumstances, personal dispositions, and population base rates.
rienced. The data are equivocal as to whether people fail to gen- Because accessibility from memory influences probability judg-
eralize even across a narrow range of risks (e.g., they become ments (Kahneman & Tversky, 1973), personal experience may
concerned about door locks after a burglar enters through the elevate the perceived likelihood of personal victimization (Per-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

door but ignore the ease of window access) or whether they do loff, 1983). However, if people believe in the gambler's fallacy or
become concerned about a small range of related hazards but believe that risk is cyclical (Slovic, Kunreuther, & White, 1974),
are very selective about the precautions they choose to adopt they may conclude that risk is lower after victimization.
(e.g., they purchase a burglar alarm but are not convinced that Perceived severity: The amount of harm that would occur if
timers or window bars are beneficial). Both patterns would lead one experienced the problem. Experience provides information
to low correlations among protective measures. about the severity of victimization.
Overall, the data do not support the nonspecific form of the Perceived effectiveness of existing preventive measures. Expe-
experience-precautions hypothesis; people do not appear to rience provides information about the effectiveness of those
adopt a wide range of new precautions. Existing results suggest preventive measures already adopted.
that attempts to understand the factors that prompt self-protec- Awareness of preventive measures: Perceived availability and
tive behavior should focus on relationships between specific cost. People may learn of precautions that had been unknown
hazard experiences and the specific precautions relevant to before their victimization. If many people suffer harm and com-
those experiences. The data suggest further that adoption even panies or government agencies respond with services and assis-
of relevant measures is not automatic and that researchers tance, precautions may become more available (i.e., from more
should study victims' views of the effectiveness, cost, and avail- sources) or available at lower cost.
ability of precautions. Before we conclude from the studies in The beliefs entering this decision-making model may be in-
Table 1 that injury in an automobile accident does not create fluenced by the victim's own experience, by communications
any desire to protect oneself from future accidents, we should from acquaintances, by campaigns organized by profit-making
find out whether the victims think their injuries would have or nonprofit organizations, and by the mass media. Victimiza-
been prevented or minimized by the use of a seat belt. At pres- tion usually leads to increased interpersonal communication
ent, we do not have such information. about the hazard, and large-scale victimization leads to in-
creased media coverage.
Mechanisms for the Effects of Personal Experience
Theories Relevant to Precautionary Behavior Cognitive Limitations and Other Cognitive Issues
A variety of approaches have been used to explain precau- According to this point of view, people's actions reflect their
tionary behavior (Weinstein, 1987a). The following pages sum- beliefs about risks and benefits, but the correspondence be-
marize these approaches under four headings: Decision-Mak- tween beliefs and action is far from perfect. There are unmoti-
ing Perspective, Fear as a Direct or Indirect Motivator, Unrealis- vated errors in the ways people store, recall, and use informa-
tic Optimism, and Social Influence Perspective. These tion about hazards (e.g., Kahneman & Tversky, 1973; Nisbett
categories prove convenient for organizing the most frequently & Ross, 1980; Slovic, Fischhoff, & Lichtenstein, 1987; Slovic et
cited variables but certainly do not exhaust all possible determi- al., 1974), and there are other cognitive variables that influence
nants of precautionary behavior. the degree of belief-behavior consistency. Key issues, some of
Because they emphasize different phenomena, these four ap- which may overlap, are as follows.
proaches are not mutually exclusive. Each describes mecha- Vividness and detail of hazard information. Research has
nisms that might help us understand the impact of personal ex- shown that concrete information (Borgida & Nisbett, 1977;
perience on preventive action. In addition, a fifth heading la- Kahneman & Tversky, 1973; Nisbett, Borgida, Crandall, &
beled "Cognitive Limitations and Other Cognitive Issues" Reed, 1976) and vivid information (Nisbett & Ross, 1980) have
describes potentially relevant research that has not been ap- a greater influence on judgments than does abstract statistical
plied to hazard issues, primarily research on social cognition information. Thus, personal experience should have stronger
and on attitude-behavior correspondence (Fazio & Zanna, effects on beliefs about hazards and precautions than would in-
1981). formation obtained secondhand.
Recall and availability. Personal experiences may be pro-
Decision-Making Perspective cessed differently or more thoroughly (Fazio, Zanna, & Cooper,
According to this point of view, the need for action is decided 1978), making them easier to recall and more likely to be re-
by weighing the magnitude of the threat against the costs and called at appropriate times to stimulate action. Memory for
42 NEIL D. WEINSTEIN

Table 5
Effects of Experience on Hazard Perceptions

Finding" Hazard perception Study

Frequency/future likelihood for people in general

+ Burglary victims rated burglary as more frequent Hoffman & Brewer (1978)
+ College students and physicians with experience of particular illnesses gave them higher Jemott, Croyle, & Ditto
prevalence ratings (1988)
+/0 Experience increased expectations of future flooding for commercial but not residential Kates(1962)
respondents
+ Earthquake and flood experience increase the view that these are serious hazards for the Kunreuther(1978)
community
0 No consistent effect of number of drought months experienced on chances of drought next year Saarinen(1966)
+ Burglary victims more likely to report that burglary occurs in their neighborhoods Scarr(1973)
+ Residents of often-flooded community more likely to expect future flooding. People with severe Smith&Tobin(1979)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

prior damage more likely to expect future flooding


+/- Burglary victims more likely to say streets are unsafe; violent crime victims less likely Sparks, Genn, & Dodd
(1977)
+ Earthquake experience increased perceived probability of quake next year Turner, Nigg, Oaz, &
Young(1981a, 1981b)
0 Crime victims no different in estimates of number of recent local crimes Tyler (1980)
+ Illness experience increases perceived prevalence of this illness in the population Weinstein (1982)
+ Illness experience increases perceived prevalence of this illness in the population Weinstein (1987b)

Personal likelihood

Recently flooded residents perceived greater likelihood of danger Danzig, Thayer, &
Galanter(1958)
Victims more likely to think their chances of being attacked or robbed have gone up in past few Garofalo(1977)
years
Major loss in quake associated with both more people expecting future damage and more Jackson (1981)
people doubting future quake will occur
Prior hurricane experience did not increase belief that hurricane Carla would hit their town Moore, Bates, Layman, &
Parenton(1963)
Perceived likelihood of future auto accident injury not affected by past accident experience Runyan(1983)
Crime victims higher on measure of personal likelihood/worry Tyler (1980)
Illness experience decreases tendency to claim below-average risk Weinstein (1982)
Illness experience decreases tendency to claim below-average risk Weinstein (1987b)

Preventability

- Burglary victims rated burglary as less controllable Hoffman* Brewer (1978)


0 Burglary victims rated burglary as no less preventable Miransky & Langer
(1978)
- Burglary victims rated burglary slightly less controllable Scarr(1973)
0 Number of victimization experiences uncorrelated with beliefs about efficacy of self-protective Tyler (1981)
action
- Owners of homes destroyed by brush fire were more likely to attribute destruction to luck than Parker, Brewer, & Spencer
people whose homes were not destroyed (1980)
0 No effect of illness experience Weinstein (1982)
— Small decrease in perceived preventability with experience of this illness Weinstein 1987b)

Seriousness

- Adolescents having experience with 10 risk behaviors rated them as less harmful Dolcinietal. (1987)
+/- Rankings of desire to avoid 12 problems/hazards went up with experience in some cases and Golant & Burton (1969)
down in others
0 Burglary victims and nonvictims rated burglary the same in seriousness Hoffman & Brewer (1978)
- Burglary victims rated burglary as less serious Hough(1985)
—/O College students who had suffered particular illnesses gave the illnesses lower seriousness Jemott, Croyle, & Ditto
ratings. Experience had no effect on physicians' seriousness ratings (1988)
+ Experienced flood victims more likely to expect loss in a future flood, but little effect of Kunreuther(1978)
earthquake experience
+ Greater future damage expected from volcano by those who experienced severe damage in Perry & Lindell (1986)
recent eruption
0 Prior flood experience did not increase belief that damage would be greater in subsequent flood Perry, Lindell, & Green
(1981)
+ Greater future flood damage expected by those who experienced severe past damage Roder(1961)
- Illness experience decreased perceived seriousness Weinstein (1982)
+ Illness experience increased perceived seriousness Weinstein (1987b)
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 43

Table 5—Continued

Finding" Hazard perception Study

Worry

+ Victims of juvenile offenders report increased anxiety Berg & Johnson (1979)
+ Crime victims report substantial emotional distress and decreased feelings of safety in their Friedman, Bischoff,
neighborhoods Davis, & Person (1982)
+/— Crime victims feel more or less unsafe than do nonvictims depending on type of crime Garofalo(1977)
+/0 Reports that fear increases during the flooding season associated with past flooding experience. Hansson, Noulles, &
Those flooded in past not higher on "desperation" scale Bellovich(I982)
+ Burglary victims more worried about burglary Hough(1985)
+ Fear-related words used more often in describing quakes by those with greater past damage Jackson (1981)
+ Robbery victims report many different psychosomatic symptoms Leyman(1985)
+ Burglary victims report feelings of unease and insecurity Maguire(1980)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

+ Burglary victims more worried about house being broken into Reppetto(1974)
This document is copyrighted by the American Psychological Association or one of its allied publishers.

+ Violent crime victims' affective and psychosomatic reactions increased over time Shapland(1984)
+ Property crime and personal crime experience increased worry and concern for corresponding Skogan(1987)
type of victimization
+/0 Victims of rape, robbery, personal theft, and burglary victims, but not assault, feel less safe Skogan(l977)
when out alone in neighborhood
+ Victims of various crimes feel less safe out alone at night in neighborhood Skogan&Maxfield(1981)
+ Fear of flooding increases with previous experience and severity of previous experience Smith &Tobin( 1979)
+ Crime victims higher on measure of personal vulnerability/worry Tyler (1980)
+/0 No effect or small positive effects of prior earthquake experience on reported fear during Turner etal.(198Ia,
quakes and fear of future quakes I981b)
+ Experienced have greater concern/knowledge concerning flood risk Waterstone(1978)
+ Illness victims worry more about the illness Weinstein(1982)
+ Illness victims think people in general worry more about the illness Weinstein(1987b)

Salience

+ Experienced more likely to recognize date of tornado 22 years earlier, though media reminders Hanson, Vitek, & Hanson
occurred yearly (1979)
+ People mentioning earthquakes as disadvantages of city lived in more likely to have suffered Jackson (1981)
severe damage
-t- Burglary victims report a tendency to keep thinking about the crime Maguire(l980)
+ Flood-experienced more likely to mention flooding as one of most serious community Smith &Tobin( 1979)
problems
-I- Illness experience associated with clearer mental image of typical victim Weinstein(1982)

" A plus sign indicates that experience led to an increase in the dependent variable; a zero indicates that experience led to no change in the dependent
variable; and a minus sign indicates that experience led to a decrease in the dependent variable.

events having reference to oneself has been shown to be superior minders that keep the risk and the possible precautions in the
to memory for events lacking self-reference (Bellaza, 1984; minds of everyone.
Brown, Keenan, & Potts, 1986). There is also evidence that the Avoidance of complex decisions. Decisions about probabilis-
increase in attitude-behavior agreement observed with atti- tic situations are difficult, and people are often unable to weigh
tudes based upon direct experience is due to the greater accessi- the evidence for and against taking precautions. Victims may
bility of such attitudes (Fazio, Chen, McDonel, & Sherman, find that associates offer recommendations or testimonials that
1982). enable them to bypass the decision process.
Uncertainty. There is less uncertainty when information is Total available information. With beliefs held constant, the
gained from personal experience than when information is greater the total amount of information about an issue, the
gained in other ways, so such information may be more con- greater the attitude-behavior consistency (Davidson et al.,
vincing. Researchers have reported greater certainty for atti- 1985). Experience is likely to lead to increased information
tudes based upon personal experience (Fazio & Zanna, 1978) about the hazard and the available precautions.
and greater attitude-behavior agreement when attitudes are Personal involvement. There is evidence that attitude change
more certain (Davidson, Yantis, Norwood, & Montana, 1985; is more permanent and that there is a greater correspondence
Fazio & Zanna, 1978; Osberg & Schrauger, 1986; Sample & between attitudes and behavior when personal involvement is
Warland, 1973). high (Cialdini, Petty, & Cacioppo, 1981; Petty & Cacioppo,
Attentional limitations. People have limited attentional ca- 1986). Personal experience of harm is a perfect example of a
pacity. Victims may be more likely to seek out or attend to infor- high-involvement condition.
mation about risk and precautions. When harm is experienced Victimization scripts. People may hold idiosyncratic theories
on a large scale, society may focus on the issue, providing re- about the ways in which victimization occurs, and experience
44 NEIL D. WEINSTEIN

may lead them to reevaluate these theories. For example, people ings are presented for frequency, personal likelihood, prevent-
who have avoided harmful experiences may develop theories to ability, seriousness, worry, and salience. A plus, zero, or minus
explain why they are invulnerable (see the discussion of "ab- in the left-hand column indicates whether experience was asso-
sent/exempt" thinking in Weinstein, 1987b). Experience may ciated with an increase, no change, or decrease, respectively, in
lead them to abandon these theories. the variable indicated.
Table 5 and Figure 1 include data gathered in other investiga-
Fear as a Direct or Indirect Motivator tions (Weinstein, 1982, 1987b) but not previously presented in
terms of experience effects. These studies provide information
This perspective takes the position that reduction in fear about an unusually large set of risk perceptions and hazards
rather than avoidance of harm is the goal of preventive behavior. (mostly health-related). The subjects were community adults in
Avoidance ofaversive emotions. Personal experience of harm the 1987 study and college freshmen in the 1982 study. The
may lead to fear of recurrence, and people may act to reduce mean ratings for each hazard dimension at each level of subject
the unpleasant sensations of fear (see Janis, 1967; Leventhal, experience are displayed in Figure 1.2
Safer, & Panagis, 1983). Researchers have generally found that victims see the hazard
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Emotion-mediated salience and severity effects. Personal ex-


This document is copyrighted by the American Psychological Association or one of its allied publishers.

as more frequent and their own risk of victimization as higher


perience of harm may be associated with persistent, vivid, un- than do nonvictims. The exceptions reported in Table 5 do not
pleasant images or even physical sensations that increase the seem to threaten this general conclusion. For example, experi-
salience of the threat and the perceived severity of victimization ence in Runyan's (1983) study did not necessarily involve ei-
(Averill, 1987; Horowitz, 1980). ther personal experience or actual injury; the report by Sparks,
Genn, & Dodd (1977) did not control for the fact that young
Unrealistic Optimism men (who are less concerned about crime) are more likely to be
violent-crime victims; and all of Saarinen's (1966) farmers had
People tend to believe that they are not vulnerable or that
been in drought areas long enough to have considerable hazard
they are less vulnerable than others around them (Perloff &
experience. Furthermore, experience could increase the per-
Fetzer, 1986; Weinstein, 1980, 1987b). This optimistic bias
ceived likelihood of damaging hurricanes, without necessarily
about personal likelihood constitutes a barrier to preventive ac-
leading people to believe that a particular hurricane is going to
tion. Victimization seems to reduce unrealistic beliefs about
hit their town (Moore et al., 1963). Finally, as was noted earlier,
personal invulnerability (Janoff-Bulman, 1985; Perloff, 1983;
people may sometimes possess theories of hazard causation that
Weinstein, 1980, 1982, 1987b).
lead them to expect decreased risk after victimization (e.g.,
floods are cyclical; earthquakes reduce the stress along fault
Social Influence Perspective lines). This might explain the lack of association between per-
People are strongly motivated to gain praise and avoid cen- sonal likelihood and experience reported by Kates (1962) and
sure (McAlister, 1987). Victims are urged by society, family, and also Jackson's (1981) finding that earthquake losses led some
acquaintances to take precautions in order to avoid recur- respondents to expect more damage in the future but others to
rences. They expect criticism if they do not take these actions doubt that a quake would reoccur.
and expect blame rather than sympathy if they become victims Four out of seven studies (two concerning burglary and one
again (Janoff-Bulman, 1985; Ryan, 1971). Victims are espe- each concerning brush fires and illnesses) found victims less
cially likely to be blamed if harm is viewed as preventable, convinced that the hazard is preventable. Of course, decreased
which is at least partly true when precautions are available. feelings of controllability, if they occur, would inhibit action, so
this variable cannot explain positive effects of experience on
precaution-taking. It could, however, explain why increases in
Empirical Investigations Linking Hazard Experience to
Theories of Protective Behavior
2
Only problems with at least 2 subjects at each of the four levels of
The preceding theoretical perspectives may seem to offer an
experience were included in Figure 1 and in the statistical analyses. For
overwhelming number of mechanisms linking personal experi-
the 1987 study this left 27 problems; for the 1982 study it left 17 prob-
ence to precautions. Yet, some of these mechanisms are un-
lems. Typical of the topics included were arthritis, asthma, suicide at-
doubtedly much more important than others. Most hazards re-
tempts, broken bones, cancer, cold sores, deafness, diabetes, drinking
search, unfortunately, has been atheoretical, so information problems, and gum disease. Differences among the means in Figure 1
about intervening variables has not been gathered in any sys- were examined by an analysis of variance, with the hazard as the unit
tematic manner. The variables reported most often reflect in- of analysis and with experience level and hazard as the independent
vestigators' interest in decision-making models (hazard likeli- variables. For the 1987 data, effects of experience were as follows: per-
hood, personal likelihood, severity, and preventability) and in sonal risk, prevalence in the population, absent/exempt, and serious-
the notion of fear as motivator (worry). Findings are occasion- ness, p < .001; preventability and others' worry, p < .05. For the 1982
data, the results were personal risk, prevalence in the population, seri-
ally reported that relate to the salience of the hazard. Little or
ousness, and own worry, p < .001; mental image, p < .01; and prevent-
no information is available from past hazard research concern-
ability, ns. Note that experience decreased perceived seriousness for the
ing most of the other possible mechanisms.
college freshmen and increased perceived seriousness for the commu-
Table 5 presents the data on mediating variables that were nity sample. All the significant effects were monotonic with increasing
encountered by the author while searching for data to include degree of experience, but the effects of direct experience on personal
in Tables 1-4. Table 5 does not purport to be a complete review risk judgments were much larger than the effects of secondhand experi-
of all studies linking hazard experience to these variables. Find- ence.
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 45
40-r
above
average

30-•

Population
prevalence 20- Personal Risk

10-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

below
This document is copyrighted by the American Psychological Association or one of its allied publishers.

average

completely very

Preventable Serious

not at all not at all

moderate agree

Absent/
Worry
exempt

not at all ^M , WtM , WtM , disagree


2 3 1 2 3
Experience Experience
Figure 1. Effects of hazard experience on risk perceptions. (Shaded and hatched bars represent data from a
community survey [Weinstein, 1987b] and from college students [Weinstein, 1982], respectively. Levels of
experience are as follows: i = do not know anyone this has happened to; 2- has happened to acquaintances;
3 = has happened to close friends or relatives; and 4 = has happened to me. In the 1982 data, worry refers
to the respondent's own worry; in the 1987 data, worry refers to ratings of how much others worry. The
variable absent/exempt denotes responses to the statement, "If this hasn't happened to someone by the
time they're my age, it isn't likely to happen")
46 NEIL D. WEINSTEIN

preventive action after victimization are sometimes smaller sponse. These routes will be described briefly, with emphasis on
than expected. the last of the three.
Both Figure 1 and the section on Seriousness in Table 5 reveal When a hazard strikes a large number of people in a short
no consistent relationship between experience of harm and the period of time, as in an epidemic, natural disaster, or crime
perceived severity of a hazard. It seems likely that perceptions wave, society takes cognizance of the risk in a variety of ways.
of seriousness would be influenced more by the severity of the The mass media carry information about the nature of the haz-
experience than by experience per se. For example, experience ard, about the suffering of victims, and about risk reduction.
was associated with significantly lower severity ratings for the Government agencies make information and sometimes mitiga-
college students in Figure 1 (Weinstein, 1982); they probably tion measures available to the public, and public and private
recovered quickly from the mild illnesses that they contracted. agencies may provide special services for victims. Within social
In contrast, the older respondents in Figure 1 judged the ill- networks there is increased communication about the hazard
nesses that they had experienced as more severe (Weinstein, and about ways of reducing risk.
1987b). All of these factors facilitate precautionary behavior, yet they
Table 5 shows that victims feel more worried, fearful, or un- apply to the entire affected community, not just to those who
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

safe than do nonvictims. The exceptions reported by Garofalo have suffered individual harm. In other words, variables sum-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(1977) for certain types of crime are probably due to uncon- marizing community-level hazard experience may add explan-
trolled demographic variables (Skogan & Maxfield, 1981). Sko- atory power to models of individual behavior distinct from that
gan's (1977) failure to find an effect for assault on victims' re- provided by variables indicating the experience of the individu-
ported feelings of safety when out in the neighborhood at night als themselves. For example, the natural hazard studies de-
may simply reflect the fact that many assaults are not commit- scribed by Moore et al. (1963), Perry, Lindell, and Green
ted by strangers and do not occur on the streets. (1981), Perry and Lindell (1986), and Smith and Tobin (1979)
A small group of studies suggest that experience increases the all seem to show the impact of societal attention. The differ-
salience of the hazard, leading victims to think about the hazard ences they found between the preparedness of experienced and
more often, have it more available in memory, and have a inexperienced residents was a function of the community: the
clearer image of the person likely to experience harm. However, effects of individual experience were greater in the communities
the indicators of salience in the studies of Jackson (1981) and that had more hazard experience.
Smith and Tobin (1979) are not clearly different from perceived Although the effects of mass experience on community-level
severity or likelihood. One could also say that Maguire's (1980) hazard awareness are clear, there are usually significant differ-
burglary victims are simply worrying more. In fact, when peo- ences within communities between those who have and those
ple say they worry about a hazard, part of what they seem to be who do not have personal experience. Thus, individual victim-
telling us is that they think frequently about the risk. This is ization must set additional forces into motion. For heart attack
certainly an indication of salience. Thus it may be appropriate patients, some of these forces are quite apparent. Family,
to treat all reports of increased worry as signs of greater hazard friends, and medical staff urge or even order the patient to stop
salience. smoking. During the hospital stay, smoking is forbidden, pro-
Finally, Figure 1 reveals that people without hazard experi- viding a start to the smoking cessation process. Education
ence are more likely to believe that those who have avoided vic- about smoking and support groups for quitting are often pro-
timization in the past are free from future risk (absent/exempt). vided in special hospital-based programs. In other words, the
People with more experience are less likely to endorse such a second route between experience and precautions is a conse-
view. Acceptance of an absent/exempt perspective should de- quence of social pressure and education specifically directed to-
crease interest in preventive measures. ward the victim.
Together, data concerning a wide range of threats demon- The third route to action refers to effects of personal experi-
strate that experience has a large impact on how people view ence that are primarily intraindividual. As was demonstrated
these risks. Personal experience substantially increases feelings earlier, victims' perceptions of risk are likely to change in a
of susceptibility and worry, as seen in its effects on personal risk number of ways. Victims also have new information about the
judgments, perceived frequency, worry, and absent/exempt be- adequacy of their own prior protective efforts. Because method-
liefs. Experience can increase or decrease perceived seriousness, ological limitations make interpretations of data uncertain and
apparently depending on the severity of the problem and the age because information about potential, mediating variables is
of the individual; experience increases the tendency of people to usually missing, the available studies do not indicate clearly the
have a mental image of the likely victim; and experience has nature of these intraindividual phenomena. Nevertheless, the
no consistent effect on perceived preventability. These data are following discussion offers a parsimonious explanation of previ-
correlational and do not indicate the direction of causality, but ous findings. Focusing on situations not adequately explained
it seems unlikely that worry or feelings of susceptibility can by the first two routes to action, it summarizes the intraindivid-
cause victimization. ual effects of personal experience in terms of five tentative prop-
ositions.
1. Persona] experience generally leads people to see hazards
Discussion as more frequent and to view themselves as potential future vic-
tims. As a consequence, interest in prevention is increased. Ex-
Three major routes from experience to protective behavior periences differ, however, in their diagnostic value. Flood losses
can be identified in the studies that have been reviewed: societal clearly identify the regions that are vulnerable to high water, but
attention, victim-directed influence, and intraindividual re- earthquake damage is often scattered irregularly over a region,
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 47

and its implications for the next earthquake are more ambigu- locks (Miransky & Longer, 1978). The limited duration of in-
ous. Sometimes, a harmful event is viewed as a freak occur- creased insurance purchases (Baumann & Sims, 1978; Institute
rence, or is even seen as decreasing future risk (e.g., earthquake of Behavioral Science, 1973; Kunreuther, 1978) and door-lock
strain is relieved), but these are not the typical reactions to per- use (Hough, 1985) after victimization suggests that the in-
sonal experience. creases in self-protection motivation are usually short-lived.
2. Experience leads people to think about the risk more often The relatively small changes in behavior often observed after
and with greater clarity. This conclusion is supported by the hazard experience may reflect a combination of factors: modest
data linking salience and worry to past victimization. Attitude experienced harm, lack of confidence in suggested precautions,
researchers have shown that thinking about an issue frequently and limited duration of impact. The failure to observe greater
has a tendency to increase the consistency between beliefs and seat belt use after significant automobile injuries, the most sur-
behavior. The vividness, concreteness, and certainty of thoughts prising finding in this review, may reflect the length of time be-
originating in personal experience should also increase their tween the accidents and the assessment of belt use or it may
impact on behavior. indicate doubts among victims that a seat belt would have pre-
3. Effects of experience on perceptions of seriousness and vented injury. There are no data, however, to support or refute
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

controllability are specific to the type of experience and the type these speculations.
of situation encountered. Experience is often mild, even milder If the five preceding propositions are correct, existing models
than expected. Many crimes are only "attempted," and losses of preventive behavior may not need major modifications.
from most thefts are small (Skogan, 1987). Most "hurricane- From the data available, it seems that a decision-making model
experienced" people have only been exposed to the fringe of of preventive action provides an adequate starting point for ex-
the storm. Obviously, experience does not necessarily lead to plaining when experience will lead to precaution taking. Al-
increased perceptions of severity. Furthermore, if little effort though the victim versus nonvictim distinction is clearly inade-
had been expended to prevent victimization (a situation true quate, it appears that knowing the severity of victimization, the
for many hazards), experience may not change beliefs about situation in which harm occurred, and the time since victimiza-
controllability. In cases in which people have taken preventive tion may be sufficient in many instances to predict increases in
measures (to deter home burglaries, for example), victimization prevention activities. To determine whether any specific mea-
experiences may produce feelings of uncontrollability. sure will be adopted, however, one still needs to look at an indi-
4. People take the precautions that they believe are appropri- vidual's beliefs about its costs and benefits.
ate for the particular hazard experiences encountered in the The data do not suggest that experience needs to be analyzed
past. The limited data reviewed here suggest that victimization along many new dimensions in order to anticipate its impact on
does not normally create a generalized feeling of vulnerability. the motivation to act. However, it would be a mistake to con-
Instead, people are inclined to adopt precautions suited to the clude that experience merely provides additional information
specific type of harm and degree of harm previously experi- that is inserted into a decision equation. The mass media and
enced. (See also Kates, 1962, and the discussion of experience- public educators attempt to encourage precautionary behavior,
precaution correspondence in the methodological section of but their information is seldom very influential. To explain why
this article). A heightened sense of vulnerability appears to be the experience of harm is a more powerful stimulus to action,
limited to situations very similar to the one in which victimiza- we need to explain why its lessons are uniquely persuasive and
tion occurred, but clear evidence about the range of generaliza- salient. Research mentioned earlier suggests that the salience
tion is lacking. If the harm experienced is minor, people are not may arise from the greater detail and sensations associated with
inclined to adopt additional preventive measures or to respond personally experienced events. The persuasiveness is consistent
more quickly to future warnings. A great deal of hazard experi- with previous studies demonstrating that case history informa-
ence teaches this no-extra-response-required lesson. tion is much more powerful than pallid summaries and with
If the harm experienced is serious, people have increased mo- the fact that victimization experience is personalized, providing
tivation to reduce their risk, but they are not necessarily con- greater certainty because there is no need to extrapolate from
vinced that the actions recommended by authorities are effec- another's encounters to one's own situation.
tive. For example, flood victims may decide that the correct re- There is already considerable evidence that beliefs about sus-
sponse in a future flood would be to stay home and protect their ceptibility and severity influence preventive behavior (e.g., Janz
possessions, not to follow the advice of government officials and & Becker, 1984), but our conjectures about the time course of
evacuate (Smith &Tobin, 1979). Wealsoknowthatthepublicis self-protection motivation after victimization and about the
skeptical about the effectiveness of crime prevention measures roles of salience and certainty in the adoption of precautions
(Mendelsohn et al., 1981). To predict the adoption of precau- need verification. Research already conducted on other haz-
tions, one needs to know the nature of the harm experienced ard-precaution relationships—the effects of pregnancy on con-
and the perceived efficacy of preventive measures. Failure to traceptive use, the effects of malpractice suits on the practice of
consider these two issues is perhaps the most important reason medicine, or the effects of arrests for drunk driving on subse-
for the apparent inconsistencies in the experience-behavior lit- quent driving behavior—might help to clarify some of these is-
erature. sues. Future investigations of personal experience can greatly
5. The duration of the increased inclination to act may be improve our understanding of preventive actions, but to make
short. Effects of experience seem to show up more clearly for a significant contribution, they will have to avoid the method-
precautions that entail action on a single occasion, such as the ological problems described earlier. They will be most informa-
purchase of new locks or insurance, than for precautions that tive if they include measurements of the variables thought to
require repeated actions, as in the use of seat belts and door- mediate the effects of experience (see McCaul & Glasgow, 1985,
48 NEIL D. WE1NSTEIN

for a clear discussion of the research designs appropriate for S. H. (1984). Smoking following myocardial infarction: A critical re-
this purpose). Studies of the time course of precautions after view of the literature. Health Psychology, 3, 83-96.
victimization and of the salience, certainty, and vividness of Burt, M. R., & Katz, B. L. (1985). Rape, robbery, and burglary: Re-
sponses to actual and feared criminal victimization with special focus
hazard cognitions produced by victimization would be particu-
on women and the elderly. Victimology, 10, 325-358.
larly valuable.
Burton, I., & Kates, R. W. (1964). The perception of natural hazards in
The present analysis of personal experience has revealed a resource management. Natural Resources Journal, .3,41 2-441.
number of reasons why precautions do and do not occur. The Cialdini, R. B., Petty, R. E., & Cacioppo, J. T. (1981). Attitude and
three routes from experience to precautionary action that have attitude change. Annual Review of Psychology, 32, 357-404.
been outlined—societal attention, victim-directed influence, Clarke, R., Ekblom, P., Hough, M., & Mayhew, P. (1985). Elderly vic-
and intraindividual response—may prove helpful for identify- tims of crime and exposure to risk. The Howard Journal of Criminal
ing the ingredients that are present or missing in a particular Justice, 24, 1-9.
prevention campaign. This review has not uncovered any new Danzig, E. R., Thayer, P. W., & Galanter, L. R. (1958). The effects of a
secrets. Nevertheless, it does suggest that programs emphasiz- threatening rumor on a disaster-stricken community (Disaster Study
No. 10). Washington, DC: National Academy of Sciences, National
ing concrete, personalized information about likelihood, sever-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Research Council. (NAS-NRC Publication No. 517)


This document is copyrighted by the American Psychological Association or one of its allied publishers.

ity, and precautions; programs attacking unrealistic optimism;


Davidson, A. R., Yantis, S., Norwood, M., & Montana, D. E. (1985).
and programs finding ways to increase hazard salience will be Amount of information about the attitude object and attitude-behav-
more successful than will traditional attempts to disseminate ior consistency. Journal of Personality and Social Psychology, 49,
general hazard information to the public. 1184-1198.
Dolcini, P., Conn, L., Adler, N., Irwin, C., Jr., Kegeles, S., Millstein, S.,
& Stone, G. (1987, March). Adolescent egocentrism and risk percep-
References tion in young adults. Paper presented at the Convention of the Society
for Adolescent Medicine, Seattle, WA.
Aberg, A., Bergstrand, R., Johansson, S., Ulvenstam, G., Vedin, A., We- Drabek, T. E. (1986). Human system responses to disaster. New York:
del, H., Wilhelmsson, C, & Wilhelmsson, L. (1983). Cessation of Springer-Verlag.
smoking after myocardial infarction: Effects on mortality after 10 DuBow, E, McCabe, E., & Kaplan, G. (1979). Reactions to crime: A
years. British Heart Journal, 49, 416-422. critical review of the literature. Washington, DC: U.S. Government
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predict- Printing Office.
ing behavior. Englewood Cliffs, NJ: Prentice Hall. Fazio, R. H., Chen, J., McDonel, E. C, & Sherman, S. J. (1982). Atti-
Averill, J. R. (1987). The role of emotions and psychological defense tude accessibility, attitude-behavior consistency, and the strength of
in self-protective behavior. In N. D. Weinstein (Ed.), Taking care: the object-evaluation association. Journal of Experimental Social
Understanding and encouraging self-protective behavior (pp. 54-78). Psychology, 18, 339-357.
New York: Cambridge University Press. Fazio, R. H., & Zanna, M. P. (1978). Attitudinal qualities relating to
Baker, E. J., Brigham, J., Paredes, J. A., & Smith, D. D. (1976). The the strength of the attitude-behavior relationship. Journal of Experi-
social impact of hurricane Eloise on Panama City, Florida (Technical mental Social Psychology, 14, 398-408.
Paper No. 1). Tallahassee, FL: Florida State University, Florida Re- Fazio, R. H., & Zanna, M. P. (1981). Direct experience and attitude-
sources and Environmental Analysis Center. behavior consistency. In L. Berkowitz (Ed.), Advances in experimen-
Bates, F. L., Fogelman, C. W., Parenton, V. J., & Tracy, G. S. (1963). tal social psychology (Vol. 14, pp. 161-202). New York: Academic
The social and psychological consequences of a natural disaster: A Press.
longitudinal study of hurricane Audrey (Disaster Study No. 18). Fazio, R. H., Zanna, M. P., & Cooper, 3. (1978). Direct experience and
Washington, DC: National Academy of Sciences, National Research attitude behavior consistency: An information processing analysis.
Council, Disaster Research Group. (NAS-NRC Publication No. Personality and Social Psychology Bulletin, 4, 48-51.
1081) Fhaner, G., & Hane, M. (1973) Seat belts: The importance of situational
Baumann, D. D., & Sims, J. H. (1978). Rood insurance: Some determi- factors. Accident Analysis and Prevention, 5, 267-285.
nants of adoption. Economic Geography, 54, 189-196. Fishbein, M., & Ajzen, I. (1974). Attitudes toward objects as predictors
Bellaza, F. S. (1984). The self as a mnemonic device: The role of internal of single and multiple behavioral criteria. Psychological Review, 81,
cues. Journal of Personality and Social Psychology, 47, 506-516. 59-74.
Berg, W. E., & Johnson, R. (1979). Assessing the impact of victimiza- Frey, D., Rogner, O., Schuler, M., & Korte, C. (1985). Psychological
tion. In W. H. Parsonage (Ed.), Perspectives on victimology (pp. 58- determinants in the convalescence of accident patients. Basic andAp-
71). Beverly Hills, CA: Sage. plied Social Psychology, 6, 317-328.
Biderman, A. D., Johnson, L. A., Mclntyre, J., & Weir, A. W. (1967). Friedman, K., Bischoff, H., Davis, R., & Person, A. (1982). Victims
Report on a pilot study in the District of Columbia on victimization and helpers: Reactions to crime (Report to the National Institute of
and attitudes toward law enforcement (President's Commission on Justice, U.S. Department of Justice). New York: New York City Vic-
Law Enforcement and Administration of Justice, Field Surveys No. tim Services Agency.
1). Washington, DC: U.S. Government Printing Office. Furstenberg, F. F. (1972). Fear of crime and its effects on citizen behav-
Borgida, E., & Nisbett, R. E. (1977). The differential impact of abstract ior. In A. Biderman (Ed.), Crime and justice: A symposium. New
versus concrete information on decisions. Journal of Applied Social York: Nailburg.
Psychology, 7,258-271. Garofalo, J. (1977). Public opinion about crime: The attitudes of victims
Brown, P., Keenan, J. M., & Potts, G. R. (1986). The self-reference and nonvictims in selected cities (Report No. SD-VAV-1). Washing-
effect with imagery encoding. Journal of Personality and Social Psy- ton, DC: U.S. Department of Justice, Law Enforcement Assistance
chology, 51, 897-906. Administration.
Bulman, R. J., & Wortman, C. B. (1977). Attributions of blame and Golant, S., & Burton, I. (1969). Avoidance response to the risk environ-
coping in the "real world": Severe accident victims react to their lot. ment (Natural Hazards Working Paper No. 6). Toronto, Ontario,
Journal of Personality and Social Psychology, 35, 351-363. Canada: University of Toronto, Department of Geography.
Burling, T. A., Singelton, E.G., Bigelow, G. E., Baile, W. F., & Gottlieb, Hanson, S., Vitek, J. D., & Hanson, P. O. (1979). Natural disaster: Long
EXPERIENCE AND SELF-PROTECTIVE BEHAVIOR 49

range impact on human response to future disaster threats. Environ- perience of life threatening events: A profile analysis. Victimology, 10,
ment and Behavior, 11, 268-284. 512-538.
Hansson, R. Q, Noulles, D., & Bellovich, S. J. (1982). Knowledge, Maguire, M. (1980). The impact of burglary on victims. British Journal
warning, and stress: A study of comparative roles in an urban flood- of Criminology, 20,261-275.
plain. Environment and Behavior, 14, 171-185. Manheimer, D. I., Mellinger, G. D., & Crossley, H. M. (1966). A follow-
Hindelang, M. J., Gottfredson, M. R., & Garofalo, J. (1978). Victims up study of seat belt usage. Traffic Safety Research Review, 70,2-13.
of personal crime: An empirical foundation for a theory of personal McAlister, A. (1987). Social learning theory and preventive behavior. In
victimization. Cambridge, MA: Ballinger. N. D. Weinstein (Ed.), Taking care: Understanding and encouraging
Hoffman, D. M., & Brewer, M. B. (1978, August). Perception of rape: self-protective behavior (pp. 42-53). New \brk: Cambridge University
A subjective probability analysis. Paper presented at the American Press.
Psychological Association Convention, Toronto, Ontario, Canada. McCaul, K. D., & Glasgow, R. E. (1985). Preventing adolescent smok-
Horowitz, M. J. (1980). Psychological reactions to serious life events. In ing: What have we learned about treatment construct validity? Health
V. Hamilton & D. M. Warburton (Eds.), Human stress and cognition Psychology, 4, 361-387.
(pp. 237-264). New York: Wiley. Mendelsohn, H., O'Keefe, G. J., Lin, J., Spetnagel, H. T., Vengler, C,
Hough, M. (1985). The impact of victimisation: Findings from the Brit- Wilson, D., Wirth, M. Q, & Nash, K. (1981). Public communications
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ish Crime Survey. Victimology, 10, 488-497. and the prevention of crime: Strategies for control. Vol. 1: Narrative
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Hutton, J. R. (1976). The differential distribution of death in disaster. report. Denver, CO: University of Denver, Center for Mass Communi-
A test of theoretical propositions. Mass Emergencies, 1, 261-266. cations.
Institute of Behavioral Science. (1973). Hail-lightning-tomado draft re- Midwest Research Institute. (1977). Crimes against the aging: Patterns
port. Boulder: University of Colorado. and prevention. Kansas City, MO: Author.
Instituut voor Sociaal Onderzoek van het Nederlandse Volk. (1955). Miransky, J., & Langer, E. J. (1978). Burglary (non)prevention: An in-
Studies in Holland Flood Disaster 1953 (Vols. 1-4). Washington, DC: stance of relinquishing control. Personality and Social Psychology
National Academy of Sciences, National Research Council, Commit- Bulletin, 4, 399-405.
tee on Disaster Studies. Moore, H. E., Bates, F. L., Layman, M. V., & Parenton, V. J. (1963)
Jackson, E. L. (1974). Response to earthquake hazard: Factors related Before the wind: A study of the response to hurricane Carla (Disaster
to the adoption of adjustments by residents of three earthquake areas Study No. 19). Washington, DC: National Academy of Sciences, Na-
of the west coast of North America. Unpublished Ph.D. dissertation. tional Research Council, Disaster Research Group. (NAS-NRC Pub-
Toronto, Ontario, Canada: University of Toronto, Department of Ge- lication No. 1095)
ography. Nisbett, R. E., Borgida, E., Crandall, R., & Reed, H. (1976). Popular
Jackson, E. L. (1977). Public response to earthquake hazard. California induction: Information is not always informative. In J. S. Carroll &
J. W. Payne (Eds.), Cognition and social behavior. Hillsdale, NJ: Erl-
Geology, 30, 278-280.
baum.
Jackson, E. L. (1981). Response to earthquake hazard. Environment
Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and
and Behavior, 13, 387-416.
shortcomings. Englewood Cliffs, NJ: Prentice Hall.
Janis, I. L. (1967). Effects of fear arousal on attitude change: Recent
Oliver, J. (1975). The significance of natural hazards in a developing
developments in theory and experimental research. In L. Berkowitz
area: A case study from North Queensland. Geography, 60. 99-110.
(Ed.), Advances in experimental social psychology (Vol. 4, pp. 166-
Osberg, T. M., & Shrauger, J. S. (1986). Self-prediction: Exploring the
224). New York: Academic Press.
parameters of accuracy. Journal of Personality and Social Psychology,
Janoff-Bulman, R. (1985). Criminal vs. noncriminal victimization:
SI, 1044-1057.
Victims' reactions. Victimology, 10, 498-511.
Parker, S. D., Brewer, M. B., & Spencer, J. R. (1980). Natural disaster,
Janz, N. K., & Becker, M. H. (1984). The Health Belief Model: A decade
perceived control, and attributions to fate. Personality and Social
later. Health Education Quarterly, 11, 1-47.
Psychology Bulletin, 6, 454-459.
Jemott, J. B. Ill, Croyle, R. T., & Ditto, P. H. (1988). Commonsense
Perkins, J., & Dick, T. B. S. (1985). Smoking and myocardial infarction:
epidemiology: Self-based judgments from laypersons and physicians.
Secondary prevention. Postgraduate Medical Journal, 67(714), 295-
Health Psychology, 7(1), 55-73.
300.
Kahneman, D., &Tversky, A. (1973). On the psychology of prediction.
Perloff, L. S. (1983). Perceptions of vulnerability to victimization. Jour-
Psychological Review, 80, 237-251.
nal of Social Issues, 39, 41-61.
Kates, R. W. (1962). Hazard and choice perception infloodplain man- Perloff, L. S., & Fetzer, B. K. (1986). Self-other judgments and perceived
agement (Research Paper No. 78). Chicago: University of Chicago, vulnerability to victimization. Journal of Personality and Social Psy-
Department of Geography. chology, 50, 502-511.
Kunreuther, H. (1978). Disaster insurance protection: Public policy les- Perry, R. W, & Lindell, M. K. (1986). Twentieth century volcanicity at
sons. New York: Wiley. Mt. St. Helens: The routinization of life near an active volcano (Final
Lavrakas, P. J. (1980). Factors related to citizen involvement in personal, Report of National Science Foundation Grant CEE-8322868).
household, and neighborhood anti-crime measures. Evanston, IL: Tempe: Arizona State University, School of Public Afiairs.
Northwestern University, Center for Public Affairs, Citizen Participa- Perry, R. W, Lindell, M. K., & Green, M. R. (1981). Evacuation plan-
tion and Community Crime Prevention Project. ning in emergency management. Lexington, MA: Heath.
LeJeune, R., & Alex, N. (1973). On being mugged: The event and its Perry, R. W., & Mushkatel, A. H. (1984). Disaster management: Warn-
aftermath. Urban Life and Culture, 2, 259-287. ing response and community relocation. Westport, CT: Quorom
Leventhal, H., Safer, M., & Panagis, F. D. (1983). The impact of com- Books.
munications on the self-regulation of health beliefs, decisions, and Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion:
behavior. Health Education Quarterly, 10, 3-29. Central and peripheral routes to attitude change. New "Vbrk: Springer-
Levy, C. C, & Smith, J. K. (circa 1979). Apathy and inexperience: A Verlag.
study of hurricane perception in Sarasola County. Tampa, FL: New Quarantelli, E. L. (1980). Evacuation behavior and problems: Findings
College of the University of South Florida, Environmental Studies and implications from the research literature (Report No. 27). Co-
Program. lumbus, OH: Disaster Research Center, Ohio State University.
Leyman, H. (1985). Somatic and psychological symptoms after the ex- Reppetto, T. A. (1974). Residential crime. Cambridge, MA: Balinger.
50 NEIL D. WEINSTEJN

Rifai, M. A. Y. (1976). Older Americans'crime prevention research proj- Smith, K.., & Tobin, G. A. (1979). Human adjustment to the flood haz-
ect. Portland, OR: Multnomah County Division of Public Safety, ard. London: Longman.
Community Affairs/Crime Prevention Unit. Sparks, R. F, Genn, H., & Dodd, D.}. (1977). Surveying victims. New
Robertson, L. S. (1975). Factors associated with safety belt use in 1974 York: Wiley.
starter-interlock equipped cars. Journal of Health and Social Behav- Sutton, S. R. (1982). Fear arousing communications: A critical exami-
ior, 16. 173-177. nation of theory and research. In J. R. Eiser (Ed.), Social psychology
Robertson, L. S., O'Neill, fl., & Wixom, C. W. (1972). Factors associ- and behavioral medicine (pp. 303-338). New York: Wiley.
ated with observed seat belt use. Journal oj'Health and Social Behav- Svenson, Q, Fischhoff, B, & MacGregor, D. (1985). Perceived driving safety
ior, 13, 18-24. and seatbelt usage. Accident Analysis & Prevention, 17(2), 119-133.
Roder, W. (1961). Attitudes and knowledge on the Topeka flood plain. In Syfert, R. (1972). The unwilling market for earthquake insurance.
G. F. White (Ed.), Papers on flood problems (Research Paper No. 70, pp. Best's Review, 73, 14-18.
62-83). Chicago: University of Chicago, Department of Geography. Titchener, J. L., Kapp, F. T., & Winget, C. (1976). The Buffalo Creek
Rogers, R. W. (1983). Cognitive and physiological processes in fear ap- syndrome: Symptoms and character change after a major disaster. In
peals and attitude change: A revised theory of protection motivation. H. J. Parad, H. L. P. Resnick, & L. G. Parad (Eds.), Emergency and
In J. T. Cacioppo & R. E. Petty (Eds.), Social psychophysiology (pp. disaster management. Bowie, MD: Charles Press.
Turner, R. H., Nigg, J, M., Oaz, D. H., & Young, B. S. (198 la). Commu-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

153-176). New York: Guildford Press.


nity response to earthquake threat in Southern California. Part 4:
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Runyan, C. W. (1983). Public health policy making: Theroleofepidemi-


ologic data in decisions about motor vehicle safety. Unpublished doc- Awareness and concern in the public (NTIS Accession No. PB82-
toral dissertation. Chapel Hill: University of North Carolina. 111261). Los Angeles, CA: Institute for Social Research, University
Ryan, W. (1971). Blaming the victim. New York: Basic Books. of California.
Turner, R. H., Nigg, J. M., Oaz, D. H., & Young, B. S. (198 Ib). Commu-
Rydant, A. L. (1979). Adjustments to natural hazards: Factors affecting the
nity response to earthquake threat in Southern California. Part 5: Ac-
adoption of crop-hail insurance. Professional Geographer, 31, 312-320.
tion response in the public (NTIS Accession No. PB82-111279). Los
Saarinen, T. F. (1966). Perception of drought on the great plains (Re-
Angeles, CA: Institute for Social Research, University of California.
search Paper No. 106). Chicago: University of Chicago, Department
Tyler, T. R. (1980). Impact of directly and indirectly experienced events:
of Geography.
The origin of crime-related judgments and behaviors. Journal oj'Per-
Saarinen, T. F. (1982). The relation of hazard awareness to adoption of
sonality and Social Psychology, 39, 13-28.
mitigation measures. In T. F. Saarinen (Ed.), Perspectives on increas-
Tyler, T. R. (1981). Perceived control and behavioral reactions to crime.
ing hazard awareness (Program on Environment and Behavior
Personality and Social Psychology Bulletin, 7, 212-217.
Monograph No. 35, pp. 1-35). Boulder: University of Colorado, In-
Tyler, T. R. (1983). Assessing the risk of crime victimization: The inte-
stitute of Behavioral Science.
gration of personal victimization experience and socially transmitted
Sample, J., & Warland, R. (1973). Attitude and the prediction of behav-
information. Journal of Social Issues, 40, 27-38.
ior. Social Forces. 51. 292-304.
U.S. Department of Justice. (1986, October). Criminal victimization.
Scarr, H. A. (1973). Patterns of burglary (2nd ed.). Washington, DC: 19S5 (Bureau of Justice Statistics Bulletin No. NCJ-102534). Wash-
Law Enforcement Assistance Administration, National Institute of
ington, DC: U.S. Department of Justice, Bureau of Justice Statistics.
Law Enforcement and Criminal Justice. Vermeulen, A., Lie, K. I., & Durrer, D. (1983). Rehabilitation after
Schiff, M. (1977). Hazard adjustment, locus of control, and sensation myocardial infarction: Changes in coronary risk factors and long-
seeking: Some null findings. Environment and Behavior, 9, 233-254. term prognosis. American Heart Journal, 105, 798-801.
Scott, R. R., & Lamparski, D. (1985). Variables related to long-term smok- Waterstone, M. (1978). The role of hazard experience and hazard
ing status following cardiac events. Addictive Behaviors, 10, 257-264. awareness in mitigation behavior of urban floodplain residents. Un-
Shapland, J. (1984). Victims, the criminal justice system, and compen- published Masters thesis. Boulder University of Colorado, Depart-
sation. British Journal of Victimology, 24, 131-149. ment of Geography.
Siegel, M. (1983). Crime and violence in America: The victim. Ameri- Weinstein, N. D. (1980). Unrealistic optimism about future life events.
can Psychologist, 38. 1267-1273. Journal of Personality and Social Psychology, 39, 806-820.
Sivarajan, E. S., Newton, K. M., Almes, M. J., Kempf, T. M., Mansfield, Weinstein, N. D. (1982). Unrealistic optimism about susceptibility to
L. W., & Bruce, R. A. (1983). The patient after myocardial infarction. health problems. Journal of Behavioral Medicine, 5, 441 -460.
Limited effects of outpatient teaching and counseling after myocar- Weinstein, N. D. (1987a). Takingcare: Understanding and encouraging
dial infarction: A controlled study. Heart and Lung, 12, 65-73. self-protective behavior. New "fork: Cambridge University Press.
Skogan, W. G. (1977). Public policy and fear of crime in large American Weinstein, N. D. (I987b). Unrealistic optimism about illness suscepti-
cities. In J. A. Gardiner (Ed.), Public law and public policy (pp. 1- bility: Conclusions from a community-wide sample. Journal of Be-
18).NewYork:Praeger. havioral Medicine. 10(5), 481-500.
Skogan, W. G. (1981). Assessing the behavioral context of victimiza- Weinstein, N. D., Grubb, P. D., & Vautier, J. (1986). Increasing automo-
tion. Journal of Criminal Law and Criminology, 72. 727-742. bile seat belt use: An intervention emphasizing perceived susceptibil-
Skogan, W. G. (1982). Methodological issues in the measurement of ity. Journal of Applied Psychology, 71, 285-290.
crime. In H. J. Schneider (Ed.), The victim in international perspec- Wilkinson, K. P., & Ross, P. J. (1970). Citizens' responses to warnings
tive (pp. 203-208). Berlin: Walter de Gruyter. of hurricane Camille (Social Science Research Center Report No.
Skogan, W. G. (1987). The impact of victimization on fear. Crime and 35). State College: Mississippi State University.
Delinquency, 33, 135-154. Wilson, L. A., & Schneider, A. (1978). Investigating the efficacy and
equity of public initiatives in the provision of public safety. Eugene,
Skogan, W G., & Maxfield, M. G. (1981). Coping with crime. Beverly
OR: Institute for Policy Analysis.
Hills, CA: Sage.
Windham, G. O., Posey, E. L., & Spencer, B. G. (1977). Reactions to
Slovic, P., Hschhoff, B., & Lichtenstein, S. (1987). Behavioral decision
storm threat during hurricane Ehise (Social Science Research Center
theory perspectives on self-protective behavior. In N. D. Weinstein
Report No. 51). State College: Mississippi State University.
(Ed.), Takingcare: Understanding and encouraging self-protective be-
havior (pp. 14-42). New York: Cambridge.
Slovic, P., Kunreuther, H., & White, G. F. (1974). Decision processes, ratio- Received May 7, 1987
nality, and adjustment to natural hazards. In G. F. White (Ed.), Natural Revision received January 18,1988
hazards: Local, national, global (pp. 187-205). New York: Oxford. Accepted February 12,1988 •

You might also like