Self-Esteem, Self-Serving Cognitions, and Health Risk Behavior

Self-Esteem, Self-Serving Cognitions,
and Health Risk Behavior
Meg Gerrard
Frederick X. Gibbons
Monica Reis-Bergan
Daniel W. Russell
Iowa State University
ABSTRACT The current article reviews prospective and experimental research on the relation between self-esteem and perceptions of vulnerability.
These studies demonstrate that individuals with high self-esteem who engage
in risk behavior often utilize a variety of self-serving cognitive strategies that
protect them from fully acknowledging their vulnerability to the potential
negative consequences of their behavior; e.g., they minimize their estimates of
personal risk and overestimate the prevalence of the risk behavior among their
peers. The article also provides data on an additional self-serving cognitive
strategy employed by adolescents with high self-esteem—alteration of perceptions of others’ reactions to their own risk behavior. Finally, the article reviews
the emerging literature on the relation between these cognitive strategies and
Meg Gerrard, Frederick X. Gibbons, Monica Reis-Bergan, and Daniel Russell, Department
of Psychology and Institute for Social and Behavioral Research, Iowa State University;
Monica Reis-Bergan now at School of Psychology, James Madison University.
This study was supported by National Institute of Mental Health Grant MH48165-08
and National Institute for Alcohol Abuse and Alcoholism Grant AA102808. We thank
the adolescents and parents for their cooperation, and the staff of the Department of
Psychology and Institute for Social and Behavioral Research for their assistance with the
research.
Correspondence concerning this article should be addressed to Meg Gerrard, Department of Psychology, Iowa State University, Ames, IA 50010.
Journal of Personality 68:6, December 2000.
Copyright © 2000 by Blackwell Publishers, 350 Main Street, Malden, MA 02148,
USA, and 108 Cowley Road, Oxford, OX4 1JF, UK.
1178
Gerrard et al.
maladaptive health behavior, and proposes that whether these strategies are
maladaptive depends on the nature of the threat and the availability of opportunities to engage in compensatory self-enhancement.
The purpose of this article is to review a line of research that has
demonstrated that people with high self-esteem (HSE) who engage in
unwise or less than optimal health behavior employ a variety of selfserving cognitive strategies to cope with the inconsistency between their
behavior and their positive self-perceptions. These strategies facilitate
justification of risk behaviors such as drinking, smoking, and unprotected
sex, by minimizing perceptions of the risks associated with these behaviors and overestimating prevalence of the behaviors among peers. In the
current article, we will review previous research demonstrating that
people with HSE who engage in risky behavior utilize a variety of
esteem-maintaining health cognitions. Then we will present data on a
new cognitive strategy, overestimation of others’ approval of an individual’s risky behavior, and discuss the implications of these strategies for
HSE risk-takers’ subsequent behavior.
Justifying Risky Behavior
The idea that people use various cognitive strategies to justify their risk
behavior is not new. The historical precedent for the argument dates at
least back to Festinger’s (1957) hypothesis that one way to deal with the
dissonance produced by engaging in an unwise behavior is to change
one’s thinking about that behavior. Festinger described a hypothetical
smoker (in the 1950s) who had recently learned that smoking is harmful.
He suggested that the smoker could employ various cognitive strategies
to deal with the discomfort produced by this information and, in so doing,
justify his behavior. For example, he could decide that the benefits of
smoking outweigh the dangers, or that the risk is negligible in comparison
to that of other activities. Twenty years later, Ross, Greene, and House
(1977) demonstrated a specific type of self-justification, called the “false
consensus effect”—the tendency for people who engage in a given
behavior to believe that the behavior is more common than do people
who do not engage in the behavior. Presumably, this normalization
process helps justify the behavior.
Subsequent research has demonstrated false consensus among people
engaging in a number of health risk behaviors. For example, smokers
Self-Esteem and Self-Serving Health Cognitions
1179
make higher estimates of the prevalence of smoking than do nonsmokers
(Sherman, Presson, Chassin, Corty, & Olshavsky, 1983; Sussman et al.,
1988); adolescents who have a history of drunk driving report higher
estimates of the prevalence of drunk driving than do those without such
a history (Arnett, 1990); drivers who commit various driving violations
report higher estimates of the prevalence of these violations (Manstead,
Parker, Stradling, Reason, & Baxter, 1992); and adolescent drinkers
overestimate the prevalence of drinking more than do their nondrinking
peers (Baer, Stacy, & Larimer, 1991; Marks, Graham, & Hansen, 1992;
also see Suls, Wan, & Sanders, 1988).
Recent research has expanded the study of self-justification of health
risk behavior by demonstrating additional cognitive strategies. For example, Gerrard, Gibbons, Benthin, and Hessling (1996) conducted a
prospective study in which adolescents reported their risk behaviors and
the extent to which concerns about health and safety would influence
those behaviors in the future. Results indicated that these concerns
predicted subsequent risk behavior—the more influential the adolescents
reported these concerns were, the less likely they were to begin reckless
driving, drinking, and smoking. Once the adolescents initiated these risk
behaviors, however, they appeared to engage in what Lazarus (1983)
called a “denial-like” process of avoiding thinking about the danger.
More specifically, as the adolescents increased their reckless driving,
drinking, and smoking, they reported that the influence of health and safety
concerns on their behavior diminished. Furthermore, avoiding thinking
about the consequences of a risk behavior and overestimating the prevalence
of that behavior both predicted changes in the behavior—adolescents whose
increases in risk behavior were accompanied by changes in these health
cognitions were more likely than others to report increased risk behavior 1
year later. Thus, these studies appear to provide ample evidence of the kinds
of strategies Festinger described over 40 years ago.
Motivational Basis for Bias
Although purely cognitive explanations of these biases have been offered
(e.g., selective exposure, availability), a fair amount of research suggests
that they are at least partially motivational in nature (Gerrard, Gibbons,
& Warner, 1991; Mullen & Hu, 1988). The basic motivational argument
is that people inflate or magnify consensus and minimize perceptions of
risk and severity, because these biases can foster and justify feelings that
1180
Gerrard et al.
one is appropriate and rational. Thus, these biases serve to legitimize
one’s attitudes or behavior (Goethals, 1986; Sherman et al., 1983; 1984;
Suls et al., 1988). Data consistent with this explanation have been
provided by a number of different lines of research. For example, it has
been shown that people evidence more biased perceptions when they
have a vested interest in a given opinion than when they do not (Crano,
1983), and when their opinions or behaviors represent a minority position
(Sanders & Mullen, 1983; Sherman et al., 1983). Similarly, bias has been
shown to be greater when an individual’s self-esteem has been threatened
by a failure experience (Sherman, Presson, & Chassin, 1984). The
assumption in these studies is that estimates of social consensus and risk
are biased to maintain and/or enhance self-esteem.
Self-Esteem and Defensive Cognitive Strategies
If normalizing risk behavior and minimizing risk estimates are esteem
maintenance strategies, then it would seem logical to assume that these
biases would be exhibited more frequently by individuals with high
self-esteem (HSE) than those with low self-esteem (LSE). Once again,
there is historical precedent for this argument. E. Aronson (1969, 1994)
suggested that individuals with HSE are particularly likely to react to
dissonance by altering their discrepant cognitions. His reasoning was that
inconsistent and/or inappropriate behavior is incongruent with HSE
individuals’ self-perceptions, whereas individuals with LSE are more
tolerant of, and may even expect, inconsistencies in their behavior.
According to this reasoning, HSE smokers should be more likely than
LSE smokers to be bothered by the dissonance associated with the
realization of the dangers of smoking, and therefore should be more likely
to engage in defensive cognitive strategies to deal with this dissonance.
Similarly, Tennen and Affleck (1993) have argued that HSE individuals
can’t accept critical feedback without being defensive and employing
counterappraisals, and Baumeister, Heatherton, and Tice have proposed
that when threatened, people with HSE “develop an overriding concern
with maximizing their esteem” (1993, p. 143). These authors, then, would
suggest that individuals with HSE should be more likely than those with
LSE to reject criticisms of their behavior.
The empirical evidence supports this contention. When faced with
feedback that threatens their self-concept, people with HSE appear to be
more capable than those with LSE of employing a variety of cognitive
Self-Esteem and Self-Serving Health Cognitions
1181
strategies that promote and maintain psychological well-being. More
specifically, relative to LSE individuals, those with HSE are more likely
to seek feedback only when they think it will be self-affirming (Swann
& Read, 1981), and to accept and remember information that is consistent
with their (positive) self-concept than information that is not (Brockner,
1979; Cohen, 1959; Knight & Nadel, 1986; Shrauger & Kelly, 1988;
Swann, 1984). They also see negative information as less credible than
positive information (Shrauger, 1975; Shrauger & Rosenberg, 1970), and
minimize the importance of situations and arenas in which they perceive
they are not competent (Campbell, 1986). Once again, it would appear
that HSE individuals are more facile than those with LSE in utilizing
self-serving and self-protective strategies to maintain their self-esteem.
Awareness of risk. An important (and in some sense preliminary) question to be addressed when examining the relations among risk behavior,
self-esteem (SE), and self-justification is whether people understand the
potential risks associated with their health relevant behavior. Recently, a
number of studies have provided evidence that people do have a relatively
good understanding of the relation between their behavior and their
vulnerability to potential negative consequences. For example, in an
experimental study of perceptions of vulnerability to unwanted pregnancy, college women were asked to make judgments about the likelihood of pregnancy in a series of hypothetical situations in which
descriptions of both frequency of intercourse and contraceptive method
were manipulated (Gerrard & Luus, 1995). The results indicated that
women recognize that the probability of pregnancy is a function of
frequency of intercourse and method of contraception, and that they make
relatively accurate estimates of pregnancy risk. Another study revealed
similar results regarding men’s understanding of the likelihood of contracting sexually transmitted diseases (Blanton & Gerrard, 1997). Similarly, Rothman, Klein, and Weinstein (1996) reported that college
students’ estimates of their risk for 11 different health hazards were
significantly correlated with the actual probability of occurrence of these
events. Finally, in the longitudinal study by Gerrard et al. (1996, described above), increases in risk behavior were accompanied by increases
in perceptions of vulnerability.
Together, these studies suggest that even adolescents understand the
relation between risk behaviors and the probability of negative outcomes
associated with these behavior. Furthermore, they can apply this knowledge
1182
Gerrard et al.
to themselves. What this means is that individuals who engage in risk
behavior do not respond to the apparent contradiction between their
behavior and their awareness of risk with absolute denial. Instead, they
perform risky behaviors in spite of their awareness of potential negative
consequences. Apparently, people with HSE are more likely than those
with LSE to cope with this awareness by minimizing their estimates of
the magnitude of the risk.
Self-esteem and risk estimates. A recent series of studies has demonstrated this effect, that is, moderation of the relation between risk estimates and health risk behaviors by SE. The first of these studies employed
an experimental paradigm in which HSE and LSE college women who
were sexually active were asked to review their past and current sexual
and contraceptive risk behavior, that is, frequency of intercourse, and
birth control methods (Smith, Gerrard, & Gibbons, 1997, Study 1). The
purpose of this review was to make the discrepancy between the women’s
risk behavior and their knowledge of effective behavior salient. Following the review, the women were asked to rate their vulnerability to
unplanned pregnancy. As expected, in spite of the fact that HSE and LSE
women reported essentially the same level of risk behavior, reviewing
that behavior increased the vulnerability estimates of the LSE women,
but did not change the estimates of the HSE women. A second study
employed a longitudinal design to demonstrate a similar relation between
SE, naturally occurring changes in sexual behavior, and risk perception
(Smith et al., 1997, Study 2). In this case, in spite of equal increases in
actual risk behavior during their college years, women with HSE evidenced smaller increases in vulnerability estimates than did women with
LSE (also see Gerrard, Gibbons, & Warner, 1991).
Another study demonstrated this effect among adults who had made
a public commitment to cease a specific risk behavior—smoking (Gibbons, Eggleston, & Benthin, 1997). In this study, HSE and LSE members
of a smoking cessation group provided estimates of the health risks
associated with smoking when they initially joined the group (before
attempting to quit), and then again 6 months following the group’s “quit
date.” Most of these smokers indicated they had joined the group because
they were aware of and concerned about the dangers of smoking. Those
who were able to abstain from smoking for 6 months maintained the
perception that they would endanger their health if they relapsed. The
risk estimates of people who were unsuccessful in abstaining revealed a
Self-Esteem and Self-Serving Health Cognitions
1183
different pattern, however. In spite of their initially high estimates of the
risks associated with smoking, relapsers with HSE, but not those with
LSE, lowered their estimates. In other words, the HSE relapsers apparently were able to convince themselves that smoking is less dangerous
than they had originally thought. In addition, this study provided evidence that these justifications were motivated by esteem maintenance:
Relapsers who did not reduce their estimates experienced a decline in
SE; relapsers who reduced their risk estimates did not demonstrate such
a decline. Thus, these studies support the hypothesis that individuals with
HSE are more likely than those with LSE to minimize the potential
consequences of their health risk behaviors. In addition, the smoking
study suggests that this process operates primarily when HSE individuals
feel threatened.
Attention to and Retention
of Health Information, and Motivation
to Change Behavior
An early study of the relation between threat and self-serving strategies
examined retention of information about AIDS and contraception among
female college students who found sexual material threatening (i.e.,
erotophobics; Gerrard, Kurylo, & Reis, 1991). It was hypothesized that
HSE erotophobic women in a course on human reproduction would learn
less of the material pertaining to sexual topics than would women with
LSE, or those who were less threatened by the material. Although HSE
erotophobic women did not have difficulty retaining other course material (e.g., information about hormonal processes), it appears that they
chose to “tune out” or ignore the threatening material. More specifically,
HSE erotophobic women, as expected, were less likely than LSE women
or HSE nonerotophobic women to retain birth control and AIDS information presented in class.
A study of optimists’ attention to information about risk and precautionary behavior provides additional evidence relevant to this issue
(Aspinwall & Brunhart, 1996). The primary finding in this study was that
optimists attended to and recalled more information about vitamin use
and exposure to ultraviolet (UV) rays than did nonoptimists. This relation, however, was moderated by prior risk behavior in a manner similar
to that found in the studies we have already reviewed. Like the HSE
erotophobic women in Gerrard et al. (1991) who failed to learn about
1184
Gerrard et al.
birth control and AIDS, optimists who did not typically take vitamins or
protect themselves from UV exposure paid less attention to, and subsequently recalled less, information than those who were regularly
engaging in precautionary behavior. Thus, both of these studies suggest
that when threatened by the material, individuals with positive views of
the future or of the self have difficulty retaining health information.
Two of our studies have offered more direct evidence that the threat
associated with knowing that one is engaging in risky behavior is associated with cognitive defensiveness. First, a dissertation conducted in our
laboratory provided sexually active college women with accurate information regarding the prevalence of condom use and casual sex among
their peers (Eggleston, 1997). Given that college students typically think
that unprotected sex is more prevalent than it actually is (Baer, Stacy &
Larimer, 1991), the hypothesis was that HSE women who were themselves engaging in risky sexual behavior would respond to this information more defensively than would LSE or low risk women. More
specifically, we hypothesized that after receiving the prevalence information, HSE risk-takers would report a greater willingness to have
unprotected intercourse if an opportunity arose (Gibbons, Gerrard, Blanton, & Russell, 1998). Results supported this hypothesis: Although other
participants lowered their reported willingness to engage in unsafe sex
after receiving the information, the HSE, high risk group responded by
increasing their willingness to have unprotected and casual sex.
Data relevant to motivation to change behavior was also reported in
the study of members of smoking cessation groups described above
(Gibbons et al., 1997). This study revealed that the decline in risk
perceptions reported by the relapsers was associated with a decrease in
their commitment to future smoking cessation. In other words, the more
successful a HSE relapser’s efforts to minimize the risks of smoking, the
less likely he or she was to report a desire to make another quit attempt.
Thus, this study provided evidence of the association between selfesteem and decreased motivation to change behavior with a different
intervention and a different risk behavior.
Self-Esteem and Self-Serving Health Cognitions
1185
Adolescent Self-Esteem, Alcohol Consumption,
and Distortion of Parents’ Reactions
OVERVIEW OF CURRENT STUDY
Given this accumulating evidence that HSE individuals have self-serving
perceptions of their vulnerability to potential negative consequences of
their risk behavior, it is logical to assume that they would also be more
likely than those with LSE to have distorted perceptions of others’
responses to this behavior. Thus, we hypothesized that when HSE adolescents increase their alcohol consumption, they will engage in a process
of justification that includes altering their perceptions of their parents’
approval of their drinking. To test this hypothesis, we analyzed two waves
of data from our ongoing panel study of adolescent risk behavior.
METHOD
Participants and procedure. The original sample in the longitudinal study included 245 male and 255 female adolescents from 50 rural counties in Iowa,
and their parents. The current analyses were conducted on measures from wave
3 (T3), when approximately half of the participants were 15 years old and half
were 17, and 1 year later (T4). Data collection occurred in the participants’
homes, and families were paid $55.00 per year for participating. The subset of
constructs from the larger study that were chosen for these analyses is described
below.
Surveys were completed by 399 adolescents at T3 and T4 (84% of the
original sample). In order to be included in the analyses, the adolescent’s
mother or father also had to complete the interview at T3 and T4. This
resulted in an N of 259, or 65% of the adolescents enrolled in the study at
T4. Because more mothers than fathers completed the surveys at both waves,
the analyses reported here include only those 251 adolescents whose mothers
completed the survey at both times. Analyses conducted with the fathers’
reports revealed identical results. For additional information about the sample and procedures see Gerrard et al. (1996) and Gibbons, Gerrard, and
Boney-McCoy (1995).
Measures. Self-esteem was assessed using a 7-point Likert-type version of the
Rosenberg Self-Esteem Scale (Rosenberg, 1965). Sample items include “I feel
that I have a number of good qualities,” and “All in all I am inclined to feel that
I am a failure” (1 = strongly disagree, 7 = strongly agree). The internal consistency
and test-retest reliabilities of the scale in this sample were comparable to those
1186
Gerrard et al.
reported elsewhere in the literature; T3 and T4 alpha = .87 and .85, respectively;
r T3 to T4 = .61 (cf. Gray-Little, Williams, & Hancock, 1997).
Adolescents responded to two questions about their recent drinking behavior: “How many times in the last 3 months have you had a whole drink
of alcohol (for example, a bottle of beer, a glass of wine, or a whole mixed
drink)?” followed by a 5-point scale that ranged from never to regularly (at
least two or three times a week) and How often in the last 3 months . . . too
much to drink?” followed by a 4-point scale from never to 4 or more times.
In addition, the mother and one sibling in each family were asked the same
question about the adolescent. The mothers’, siblings’, and adolescents’
reports of the adolescents’ consumption loaded on a single factor (loadings
of .71, .77, and .95), indicating reasonable concordance, and providing
evidence of validity of the adolescents’ self-reports (Gerrard, Gibbons, Zhao,
Russell, & Reis-Bergan, 1999.)
Adolescents’ perceptions of their parents’ reactions to drinking were measured by a single item, “Whether you have drunk alcohol or not, how do you
think your parents would react if they thought you had drunk alcohol?” with
responses ranging from 1 = have a strong negative reaction and tell you to stop
to 5 = they would approve. Mothers were asked a parallel question regarding
their reaction to their adolescents’ consumption of alcohol, “How would you
react if you thought your son/daughter [in this study] had drunk alcohol?” with
a response scale comparable to the 5-point scale for adolescents’ perceptions of
their parents’ reactions. The inclusion of this measure permitted us to test the
primary hypothesis, that is, HSE, but not LSE, adolescents adjust their perceptions of their parents’ approval of their alcohol consumption, bringing it in line
with their own behavior, controlling for the mothers’ actual self-reports of their
reactions.
RESULTS
Descriptive analyses. At T3, 40% of the adolescents reported having
consumed a whole glass of wine or beer in the last 3 months, and 27%
reported drinking to excess at least once in that period. As expected, these
figures had risen significantly by T4 (63% reported drinking and 47%
reported excessive drinking; ts > 7.60, p < .001). As can be seen in Table
1, HSE adolescents reported less frequent and less excessive drinking
than did LSE adolescents at both T3 (ts = 3.00 and 1.95, respectively)
and T4 (ts = 3.70 and 3.38; all ps < .05). In addition, LSE adolescents
reported a significantly greater increase in consumption between T3 and
T4 than did HSE adolescents, t = 2.12, p < .05.
Self-Esteem and Self-Serving Health Cognitions
1187
Table 1
Means and Standard Deviations for the Measures
Measure
HSE
Mean (SD)
LSE
Mean (SD)
Total
Mean (SD)
Frequency of drinking behavior
Time 3
1.60a (1.08)
Time 4
2.17ab (1.35)
2.04(1.24)
2.80b (1.33)
1.82 (1.18)
2.45b (1.38)
Frequency of excessive drinking behavior
Time 3
1.37a (0.81)
Time 4
1.74ab (1.13)
1.59 (0.97)
2.24b (1.22)
1.48 (0.90)
1.99b (1.20)
Total drinking behavior (latent variable)
Time 3
1.48a (0.91)
Time 4
1.95ab (1.18)
1.81 (1.06)
2.52b (1.21)
1.65 (1.00)
2.24b (1.23)
Adolescents’ perceptions of parents’ reactions
Time 3
1.29a (0.65)
Time 4
1.51ab (0.86)
1.49 (0.87)
1.90b (1.08)
1.39 (0.77)
1.71b (0.99)
Mothers’ reaction to adolescents’ drinking behavior
Time 3
1.30 (0.71)
1.29 (0.73)
Time 4
1.37 (0.76)
1.33 (0.70)
1.29 (0.71)
1.35 (0.73)
N = 125 HSE; N = 126 LSE.
a Indicates significant mean differences between HSE and LSE groups, p < .05.
b Indicates significant mean differences over time, p < .05.
Not surprisingly, mothers’ reactions to alcohol consumption by their
adolescents remained quite negative over this period of time, and there
were no differences between HSE and LSE adolescents’mothers’reports
at T3 or T4, ps > .60. The zero-order correlations between mothers’
reactions and adolescent consumption were nonsignificant at both T3 and
T4, with rs ranging between .01 and .08, ps > .20 (cf. Kandel, Kessler,
& Margulies, 1978; Peterson, Hawkins, Abbott, & Catalano, 1994).
Adolescents’ perceptions of their parents’ approval of consumption
were comparable to their mothers’ reports of their own reactions at T3
(M = 1.39 vs. 1.29, t = 1.72, p = .09), but they were significantly less
negative than their mothers’ reports at T4 (M = 1.71 vs. 1.35, t = 5.30,
p < .001). The zero-order correlations between consumption and perception of parental approval at both T3 and T4 were moderate and positive,
ranging from .20 to .37. The differences in these correlations over time
was significant, ∆χ2 (1, N = 251) = 3.88, p < .05, indicating that
1188
Gerrard et al.
adolescents who drank were more likely to perceive that their parents did
not disapprove, and that this association increased over time.1 In sum,
these descriptive analyses suggest that HSE adolescents drank less than
did LSE adolescents and were less likely to increase their consumption
between T3 and T4. In general, increases in consumption were associated
with increased perceptions of parental approval, in spite of the fact that
this approval did not actually change over time.
Structural equation models. Next, a stacked (multiple group) structural
equation model was employed to test the hypothesis that HSE adolescents who increase consumption are more likely than LSE adolescents
who increase consumption to adjust their perceptions of parental approval in a self-serving direction.2 More specifically, this model was
designed to compare the relation between T4 consumption and T4
perceptions of parents’ approval of drinking for LSE versus HSE adolescents, controlling for prior consumption, prior perceptions of parents’
approval, and mothers’ actual approval, current and prior. Thus, crosslagged structural pathways were employed to control for the influence
of T3 consumption and perceptions on T4 consumption and perceptions
(see Figure 1). In this model, frequency of drinking and frequency of
excessive drinking were used as indicators of a latent alcohol consumption construct (standardized factor loadings ranged from .87 to .98).
Adolescents’ perception of parents’ approval and the mothers’ reports of
their own reactions were represented by single indicators of their respective constructs.3
1. The equivalence of the correlations between these variables at T3 and T4 was tested
by examining the measurement model for the entire sample. In a first version of the
measurement model, we constrained these two correlations to be equal, and in a second
model, we allowed them to vary (see Brooke, Russell, & Price, 1988, for an example of
this procedure).
2. Another approach to testing this hypothesis would be to use ordinary least squares
regression. We chose to use structural equation modeling because it permits us to treat
perceptions of parental reactions and consumption as dependent variables simultaneously
(see Figure 2), and because it incorporates interrelations between all variables into the
final solution.
3. The loading of the individual indicator variables on their respective factors was fixed
at 1.0, and the error terms for these measures were fixed at zero. Since respondents were
assessed using identical items at two points in time, the possibility of correlated measurement error between the parallel measures could increase the apparent stability of
the latent variables, thereby lessening the potential causal effects of other variables
Self-Esteem and Self-Serving Health Cognitions
Adolescent
Alcohol
Consumption
.75*
.02
.03
.48*
1189
Adolescent
Alcohol
Consumption
.20*
.24*
.28*
.06
Perception
of Parents’
Reactions
-.08
.14
.05
*
.23
.37*
.56*
Perception
of Parents’
Reactions
.05
.21*
.44*
.08
Mother’s
Reaction
.42*
.31*
Mother’s
Reaction
HSE
LSE
TIME 3
TIME 4
χ2(26) = 32.67(p = .17)
GFI = .96
CFI = .99
RMSEA = .03
Figure 1
Structural Equation Model I.
A confirmatory factor analysis, conducted using the maximum likelihood estimation algorithm from LISREL VIII (Joreskog & Sorbom,
1996) was found to provide a good fit to the data, χ2 (16, N = 251) =
16.55, p = .42, GFI = .98, CFI = .99, RMSEA = .01.4 The correlation
matrix for measured variables used in testing this model is presented in
Table 2. The structural model also fit the data well, χ2(26, N = 251) =
(Ulrick-Jakubowski, Russell, & O’Hara, 1988). Therefore, we included this possibility
by correlating the error terms for the measured variables related to drinking behavior. We
also constrained the factor loadings of these indicators to be constant over time.
4. A Heywood case occurred in the initial test of this model, reflected by the negative
error variance for the T3 measure of alcohol frequency. Constraining the correlated error
term between the T3 and T4 measures of alcohol consumption at zero eliminated the
problem, although the fit of the model decreased significantly, ∆χ2 (2, N = 251) = 11.94,
p < .001. The relationships among the factors did not change greatly due to fixing this
correlated error term to zero. The change in t values ranged from 0 to .93 and none of the
correlations among the factors changed from significant to nonsignificant or vice versa.
1190
Gerrard et al.
Table 2
Correlation Matrix for the Structural Equation Models
1
2
1. T3 alcohol consumption
—
.25
2. T3 adolescents’ perceptions
of parents’ reactions
.21*
3. T3 mothers’ approval of
adolescents’ drinking
3
*
4
5
*
6
*
.02
–.07
.72
.30
—
.06
.20*
.45*
.15
.12
.38*
—
–.12
.03
.41*
4. T4 alcohol consumption
.52*
.38*
.04
—
.35*
.02
5. T4 adolescents’ perceptions
of parents’ reactions
.16*
.61*
.18*
.32*
—
.30*
6. T4 mothers’ approval of
adolescents’ drinking
.01
.30*
.32*
.19*
.24*
—
Note: HSE group above the diagonal; LSE group below the diagonal.
* p < .05.
32.67, p = .17, GFI = .96, CFI = .99, RMSEA = .03.5 As can be seen in
Figure 1, the hypothesized path from consumption at T4 to perceptions
at T4 was significant for HSE adolescents, β = .24, p < .01, but not for
LSE adolescents, β = .06, ns. In addition, this model revealed a
significant path from T3 perceptions to T4 consumption for the LSE
adolescents, β = .23, p < .01, but not for the HSE adolescents, β = .05,
ns. The overall difference between the models for the two SE groups was
5. It is worth noting that HSE adolescents’reports of their consumption were more stable
over time than were those of the LSE adolescents. This difference is consistent with
Campbell’s research (e.g., 1990), indicating that HSE individuals are more likely than
those with LSE to demonstrate temporal stability in self-descriptions. It should also be
noted that, consistent with the current hypothesis, HSE adolescents’ perceptions of their
parents’ approval of drinking were less stable over time than were LSE adolescents’
perceptions. These differences in stability, however, could have an impact on the coefficients for the HSE and LSE paths from T4 consumption to T4 perceptions of parents’
reactions. Thus, we also tested a model that constrained the HSE and LSE groups’
longitudinal pathways between T3 and T4 to be equal to each other. This model
demonstrated essentially the same pattern as that reported above, with a significant β =
.23, t = 2.19, p = .03, for the HSE group and a nonsignificant β = .11, t = 1.13, p = .26,
for the LSE group.
Self-Esteem and Self-Serving Health Cognitions
1191
evaluated by constraining the structural pathways to be equal and comparing the fit of the constrained model with that for the model that allowed
the paths to vary. This omnibus test revealed a significant difference in
the fit of the HSE and LSE models, ∆χ2 (6, N = 251) = 12.96, p = .04. A
test of whether the pathway of primary interest—from T4 consumption
to T4 perceptions of parental approval—was the source of the difference
between the HSE and LSE models was conducted in the same manner,
that is, the path was constrained for the first model and then allowed to
vary for the second model. This test revealed that the paths for the HSE
and LSE adolescents were not significantly different from each other,
∆χ2 (1, N = 251) = 1.72, p = .19, indicating that the difference between
the two models could not be accounted for by this path alone.
Because previous research has demonstrated that the relation between
health cognitions (e.g., prevalence estimates) and risk behavior is reciprocal (Gerrard et al., 1996), we were also interested in evaluating reciprocal paths between alcohol consumption and perceptions of parent
reactions at T4. The model with both the reciprocal paths between T4
behavior and T4 perception, and the cross-lagged paths was not identified, however. Therefore, we ran a series of models to determine the
best-fitting model for the two self-esteem groups. First, the reciprocal
model was tested without specifying the cross-lagged paths. The data fit
this model well, χ2 (28, N = 251) = 36.37, p = .13, GFI = .95, CFI = .99,
RMSEA = .04, and produced a significant overall difference in the paths
between the two self-esteem groups, ∆χ2 (5, N = 251) = 14.56, p = .01.
The path from T4 perception to T4 behavior was significant for the LSE
group, β = .25, p < .01, but not for the HSE group, β = .02. The difference
between these two paths was not significant, ∆χ2 (1, N = 251) = .30, p =
.58. Once again, however, the path from T4 behavior to T4 perceptions
was significant for the HSE group, β = .24, p < .01, but not for the LSE
group, β = –.11. The difference between the HSE and LSE adolescents’
paths was marginally significant, ∆χ2 (1, N = 251) = 3.33, p = .07. 6
Additional models were tested modifying the reciprocal and the crosslagged pathways, including only one lagged or only one T4 path, all of
which revealed significant differences between the SE groups. Finally,
6. A model was also run constraining the longitudinal paths from T3 to T4 to be equal.
The pattern was identical; however, the path from perception to consumption for the LSE
group became marginally significant, β = .17, t = 1.93, p = .05. The path from behavior
to perception for the HSE group remained significant, β = .21, t = 2.27, p < .05.
1192
Gerrard et al.
the best-fitting model was determined by trimming nonsignificant reciprocal and cross-lagged paths for each SE group (see Figure 2). As
expected, this model provided an excellent fit to the data, χ2 (28, N = 251)
= 32.08, p = .27, GFI = .96, CFI = 1.00, RMSEA = .02. This trimmed
model includes significant paths from T3 perceptions of parental approval to T4 consumption for LSE adolescents, β = .23, p < .001, and
from T4 behavior to T4 perception for HSE adolescents, β = .24, p < .01.
Thus, consistent with the stacked models, this best-fit model revealed
different patterns for the two SE groups: LSE individuals’ T4 behavior
was influenced by their perceptions of parental reactions at an earlier
point in time, whereas HSE adolescents’ perceptions of their parents’
current reactions were influenced by their own current behavior. Because
the more conservative test (the stacked models above) yielded only
marginal support for the primary hypothesis, i.e., HSE, but not LSE,
adolescents’ alcohol consumption shapes their perceptions of their parents’ approval of drinking, it should be noted that these findings can only
be interpreted as suggestive until they are replicated.
Adolescent
Alcohol
Consumption
.82*
.44*
.19*
.25*
Perception
of Parents’
Reactions
.14
.24* HSE
*L
.23
-.08
SE
Adolescent
Alcohol
Consumption
.38*
.58*
Perception
of Parents’
Reactions
.05
.21*
.08
.08
Mother’s
Reaction
.42*
.31*
Mother’s
Reaction
HSE
LSE
TIME 3
TIME 4
Figure 2
Structural Equation Model II (trimmed).
χ2(28) = 32.08(p = .27)
GFI = .96
CFI = 1.00
RMSEA = .02
Self-Esteem and Self-Serving Health Cognitions
1193
DISCUSSION
It is common for adolescents to increase their alcohol consumption over
the course of their teenage years (cf. Chassin & DeLucia, 1996; Keefe,
1994); and it is not surprising that parental approval of this behavior does
not show a corresponding increase during this period of time. It appears
that HSE adolescents are more likely than LSE adolescents to deal with
this discrepancy between their drinking and their parents’ reactions to it
by adjusting their perceptions in a self-justifying manner. In other words,
although the HSE adolescents in this study actually reported less alcohol
consumption at both waves, and smaller increases in consumption over
time, than the LSE adolescents, those HSE adolescents who did increase
their drinking adjusted their perceptions of their parents’ reactions more
than did LSE adolescents who increased their drinking.
The current analyses also suggest that the reciprocity between risk
behavior and risk cognitions demonstrated in previous studies (Gerrard
et al., 1996; cf. Weinstein & Nicolich, 1993; Weinstein, Rothman, &
Nicolich, 1998) may operate differently for HSE and LSE individuals.
Among those with HSE, naturally occurring changes in risk behavior
influence risk cognitions such as estimates of prevalence and vulnerability, and perceptions of others’ approval. These perceptions, then, become
reflections of the new risk behavior rather than predictors of risk behavior. Among those with LSE, however, the causal direction of this relation
may be reversed with LSE adolescents’ perceptions of parental disapproval of drinking shaping their behavior more than being shaped by it.
Esteem-Maintaining Strategies
and Subsequent Behavior
The differences in consumption between the HSE and LSE adolescents
raises an obvious and important question: Does the use of these cognitive
strategies result in subsequent increases in risk behavior among HSE
adolescents? This question is at the heart of an ongoing vigorous debate
regarding the extent to which positive views of the self are adaptive or
maladaptive in terms of behavioral outcomes (Armor & Taylor, 1998;
Colvin & Block, 1994; Colvin, Block, & Funder, 1995; Taylor & Brown,
1988). On the one hand, it has been argued that HSE promotes mental
health and psychological well-being. Consistent with this, Taylor and
Brown (1988) cite evidence that people with HSE are happier, more
1194
Gerrard et al.
capable of caring relationships, more persistent and effective, and have
a greater capacity for intellectual and creative work than do those with
LSE. On the other hand, it has been argued that the benefits of positive
views of the self may be somewhat circumscribed and, in particular, may
not extend to health behaviors (Baumeister, 1989; Tennen & Affleck,
1993; Weinstein, 1980). Thus, two competing hypotheses regarding the
relation between self-esteem and health behavior have emerged. The first
is an intuitive argument that people with HSE value themselves and their
well-being, and so are motivated to take care of themselves (i.e., avoid
risk behaviors and engage in precautionary behaviors; cf. Taylor &
Brown, 1988; Torres, Fernandez, & Maciera, 1995). According to this
view, although they may engage in esteem maintaining strategies like
minimization of risk estimates and false consensus, they do attend to
health information and ultimately “do the right thing” by performing
precautionary behaviors. The second hypothesis is that HSE is associated
with defensiveness regarding the appropriateness of current and past
behavior that can lead to resistance to ameliorative behaviors.
An Apparent Paradox
On the face of it, the evidence appears to support both of these competing
hypotheses. HSE risk-takers do engage in cognitive strategies that allow
them to minimize their perceptions of risks associated with their behavior, and these strategies are associated with avoidance of health information and decreased motivation to change behavior. Very few studies,
however, have reported that people with HSE are more likely than those
with LSE to engage in risky behaviors (e.g., Sharp & Getz, 1995)—most
have found no differences or the opposite (e.g., Abernathy, Massad, &
Romano-Dwyer, 1995; Dielman, Campanelli, Shope, & Butchart, 1987).
Thus, the question becomes, why aren’t these strategies associated with
increases in risky behavior, and are there circumstances under which they
are more likely to foster future risk behavior?
We propose that the association between the psychological defensiveness exhibited by HSE risk-taking individuals and their subsequent
behavior is dependent on the nature of the threat. More specifically, we
suggest that whether HSE defensiveness inhibits preventive or ameliorative behavior depends on two factors: (a) whether reactance is aroused
by the implication that one’s current or previous behavior has been
inappropriate, and (b) the availability of additional opportunities to reduce
Self-Esteem and Self-Serving Health Cognitions
1195
the threat. When HSE risk-takers are confronted with evidence that their
health behavior is less than optimal and they perceive this evidence as a
threat to the self, their immediate response is the kind of self-justification
we have demonstrated in our previous studies. Over time, however, they
may be successful in reducing the threat by focusing on strengths in other
domains, that is, affirm their competence or worth on dimensions that are
not threatened.
Self-affirmation and compensatory self-enhancement. B a u m e i s te r
and his colleagues have suggested that rather than deny the validity of
negative information about themselves, HSE people balance this information by focusing on their positive characteristics, a process
called compensatory self-enhancement (Baumeister, 1982). Similarly,
Steele proposed that HSE people employ self-affirmation to reduce
the discomfort produced by realizing their behavior is incongruent
with their (positive) self-images. Hence, an HSE ex-smoker who
relapses might cope with the threat to his sense of self-control by
thinking about his accomplishments in some other arena (Steele,
Spencer, & Lynch, 1993). In a test of this hypothesis using a traditional
free choice dissonance paradigm, Steele et al. (1993) reported that
self-affirmation is effective in significantly reducing HSE individuals’
rationalization of decisions.
The results of a study conducted in our laboratory were significantly
different, however. In Boney-McCoy, Gibbons, and Gerrard (1999),
HSE and LSE college students reviewed their sexual histories and then
listed past and present behaviors that increased their likelihood of
contracting a sexually transmitted disease (STD). Following this review, the students were given an opportunity to self-affirm. The data
supported the hypothesis that HSE individuals are more likely than
those with LSE to engage in self-affirmation when their dissonant
behavior is made salient. More important, the more HSE students
self-enhanced, the more they subsequently lowered their estimates of
the likelihood they would contract STDs. Thus, contrary to Steele et
al. (1993), these results indicate that self-affirmation can be associated
with increased rationalization of dissonant behavior, suggesting that
self-affirmation and rationalization may be complementary protective
strategies that occur simultaneously (see also Blanton, Cooper,
Skurnick, & Aronson, 1997).
1196
Gerrard et al.
Psychological reactance. Although there are a number of differences
between the Steele et al. (1993) and Boney-McCoy et al. (1999) studies,
the primary one appears to be the nature of the threat. Steele et al. induced
threat by forcing participants to choose a record album that they had
previously rated as only moderately desirable. Boney-McCoy et al.
induced threat by reminding participants that their volitional behavior
(failure to use condoms) put them at risk. Because it raises questions
about an individual’s ability to regulate his or her behavior, acknowledging risky actions poses more of a threat to the self than does choosing
a record album. In addition, because it implies the need for behavior
change, it should arouse more psychological reactance (Brehm, 1966).
Thus, we suggest that if health information is interpreted as accusatory,
individuals with HSE will not simply engage in self-serving cognitive
adjustments and demonstrate reactance. Instead, they may also increase
their willingness to engage in risk behavior, reduce their commitment to
ameliorative behavior, or tune out threatening preventive information. In
general, then, HSE serves a protective function by inhibiting risky
behavior. When people with HSE do engage in risky behavior, however,
their motivation to protect their self-esteem can turn into defensiveness
that may interfere with precautionary behavior.
Conclusion
Individuals with HSE engage in a variety of self-serving cognitive
strategies regarding their health behavior. Specifically, they are more
likely than those with LSE to minimize the potential for negative consequences of their risky behavior, are more likely to ignore information that
threatens them, and are more likely to make defensive adjustments to
their estimates of others’ approval of that behavior. In addition, information that suggests that their behavior has been foolish or less than optimal,
and therefore should be altered, elicits defensive self-justification and
reactance. Although these responses may not lead to escalation of risk
behavior or reduced precautionary behavior in the long-run, it is possible
that even short-lived defensiveness can have detrimental effects if it
results in missed opportunities to either acquire information about health
hazards or reluctance to alter behavior.
Self-Esteem and Self-Serving Health Cognitions
1197
REFERENCES
Abernathy, T. J., Massad, L., & Romano-Dwyer, L. (1995). The relationship between
smoking and self-esteem. Adolescence, 30, 899–907.
Armor, D. A., & Taylor, S. E. (1998). Situated optimism: Specific outcome expectancies
and self-regulation. In Mark Zanna (Ed.), Advances in experimental psychology
(Vol. 30, pp. 309–379). New York: Academic Press.
Arnett, J. J. (1990). Drunk driving, sensation seeking, and egocentricism among adolescents. Personality and Individual Differences, 11, 541–546.
Aronson, E. (1969). The theory of cognitive dissonance: A current perspective. In L.
Berkowitz (Ed.), Advances in experimental social psychology (Vol. 4, pp. 1–34). San
Diego, CA: Academic Press.
Aronson, E. (1994). The social animal. New York: W.H. Freeman.
Aronson, J., Blanton, H., & Cooper, J. (1995). From dissonance to disidentification:
Selectivity in the self-affirmation process. Journal of Personality and Social Psychology, 86, 986–996.
Aspinwall, L. G., & Brunhart, S. M. (1996). Distinguishing optimism from denial:
Optimistic beliefs predict attention to health threats. Personality and Social Psychology Bulletin, 22, 993–1003.
Baer, J. S., Stacy, A., & Larimer, M. (1991). Biases in the perception of drinking norms
among college students. Journal of Studies on Alcohol, 52, 580–586.
Baumeister, R. F. (1982). Self-esteem, self-presentation, and future interaction: A dilemma of reputation. Journal of Personality, 50, 29–45.
Baumeister, R. F. (1989). The optimal margin of illusion. Journal of Clinical and Social
Psychology, 8, 176–189.
Baumeister, R. F., Heatherton, T. F., & Tice, D. M. (1993). When ego threat leads to
self-regulation failure: Negative consequences of high self-esteem. Journal of Personality and Social Psychology, 64, 141–156.
Blanton, H., Cooper, J., Skurnik, I., & Aronson, J. (1997). When bad things happen to
good feedback: Exacerbating the need for self-justification with self-affirmations.
Personality and Social Psychology Bulletin, 23, 684–692.
Blanton, H. & Gerrard, M. (1997). Effect of sexual motivation on the perceived riskiness
of a sexual encounter: There must be fifty ways to justify a lover. Health Psychology,
16, 374–379.
Boney-McCoy, S., Gibbons, F. X., & Gerrard, M. (1999). Self-esteem, compensatory
self-enhancement, and the consideration of health risk. Personality and Social Psychology Bulletin, 25, 954–965.
Brehm, J. W. (1966). A theory of psychological reactance. New York: Wiley.
Brockner, J. (1979). The effects of self-esteem, success-failure, and self-consciousness
on task performance. Journal of Personality and Social Psychology, 37, 1732–1741.
Brooke, P. P., Russell, D. W., & Price, J. L. (1988). Discriminant validation of measures
of job satisfaction, job involvement, and organizational commitment. Journal of
Applied Psychology, 73, 139–145.
Campbell, J. D. (1986). Similarity and uniqueness: The effects of attribute type, relevance, and individual differences in self-esteem and depression. Journal of Personality and Social Psychology, 50, 281–294.
1198
Gerrard et al.
Campbell, J. D. (1990). Self-esteem and the clarity of the self-concept. Journal of
Personality and Social Psychology, 59, 538–549.
Chassin, L., & DeLucia, C. (1996). Drinking during adolescence. Alcohol Health and
Research World, 20, 175–180.
Cohen, A. R. (1959). Some implications of self-esteem for social influence. In I. Janis,
C. Hovland, P. B. Field, H. Linton, E. Graham, A. R. Cohen, D. Rife, R. Abelson, G.
Lesser, & B. King (Eds.), Personality and persuasibility (pp. 102–120). New Haven,
CT: Yale University Press.
Colvin, C. R., & Block, J. (1994). Do positive illusions foster mental health? An
examination of the Taylor and Brown formulation. Psychology Bulletin, 116, 3–20.
Colvin, C. R., Block, J., & Funder, D. C. (1995). Overly positive self-evaluations and
personality: Negative implications for mental health. Journal of Personality and
Social Psychology, 68, 1152–1162.
Crano, W. D. (1983). Assumed consensus of attitudes: The effect of vested interest.
Personality and Social Psychology Bulletin, 9, 597–608.
Dielman, T. E., Campanelli, P. C., Shope, J. T., & Butchart, A. T. (1987). Susceptibility
to peer pressure, self-esteem, and health locus of control as correlates of adolescent
substance abuse. Health Education Quarterly, 14, 207–221.
Eggleston, T. J. (1997). Self-esteem, risk behavior and reactance: The impact of prototype
modification on willingness to engage in unprotected sex. Unpublished doctoral
dissertation, Iowa State University.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University
Press.
Gerrard, M., Gibbons, F. X., Benthin, A. C., & Hessling, R. M. (1996). The reciprocal
nature of risk behaviors and cognitions: What you do shapes what you think, and vice
versa. Health Psychology, 15, 344–354.
Gerrard, M., Gibbons, F. X., & Warner, T. D. (1991). Effects of reviewing risk-relevant
behavior on perceived vulnerability among women Marines. Health Psychology, 10,
173–179.
Gerrard, M., Gibbons, F. X., Zhao, L., Russell, D. W., & Reis-Bergan, M. (1999). The
effects of peers’ alcohol consumption on parental influence: A cognitive mediational
model. Journal of Studies on Alcohol, 13, 32–44.
Gerrard, M., Kurylo, M., & Reis, T. (1991). Self-esteem, erotophobia, and retention of
contraceptive and AIDS information in the classroom. Journal of Applied Social
Psychology, 21, 368–379.
Gerrard, M., & Luus, C. A. E. (1995). Judgments of vulnerability to pregnancy: The role
of risk factors and individual differences. Personality and Social Psychology Bulletin,
21, 160–171.
Gibbons, F. X., Eggleston, T., & Benthin, A. C. (1997). Cognitive reactions to smoking
relapse: The reciprocal relation between dissonance and self-esteem. Journal of
Personality and Social Psychology, 72, 184–195.
Gibbons, F. X., Gerrard, M., Blanton, H., & Russell, D. W. (1998). Reasoned action and
social reaction: Willingness and intention as independent predictors of health risk.
Journal of Personality and Social Psychology, 74, 1164–1180.
Self-Esteem and Self-Serving Health Cognitions
1199
Gibbons, F. X., Gerrard, M., & Boney-McCoy, S. (1995). Prototype perception predicts
(lack of) pregnancy prevention. Personality and Social Psychology Bulletin, 21,
85–93.
Goethals, G. R. (1986). Fabricating and ignoring social reality: Self-serving estimate of
consensus. In Olson, J. M., Herman, C. P., and Zanna, M. P. (Eds.), Relative
deprivation and social comparison. The Ontario Symposium (Vol. 4, pp.135–157).
Hillsdale, NJ: Lawrence Erlbaum.
Gray-Little, B., Williams, V., & Hancock, T. (1997). An item-response theory analysis
of the Rosenberg Self-Esteem Scale. Personality and Social Psychology Bulletin, 23,
443–451.
Joreskog, K. G., & Sorbom, D. (1996). LISREL 8: User’s Reference Guide, Mooresville,
IN: Scientific Software.
Kandel, D. B., Kessler, R. C., & Margulies, R. Z. (1978). Antecedents of adolescent
initiation into stages of drug use: A developmental analysis. Journal of Youth and
Adolescence, 7, 13–40.
Keefe, K. (1994). Perceptions of normative social pressure and attitudes toward alcohol
use: Changes during adolescence. Journal of Studies on Alcohol, 55, 46–54.
Knight, P. A., & Nadel, J. I. (1986). Humility revisited: Self-esteem, information search,
and policy consistency. Organizational Behavior and Human Decision Processes, 38,
196–206.
Lazarus, R. S. (1983). The costs and benefits of denial. In S. Breznitz (Ed.), The denial
of stress (pp. 1–30). New York: International Universities Press.
Manstead, A. S. R., Parker, D., Stradling, G. G., Reason, J. T., & Baxter, J. S. (1992).
Perceived consensus in estimates of the prevalence of driving errors and violations.
Journal of Applied Social Psychology, 22, 509–530.
Marks, G., Graham, J., & Hansen, W. (1992). Social projection and social conformity in
adolescent alcohol use: A longitudinal analysis. Personality and Social Psychology
Bulletin, 18, 92–101.
Mullen, B., & Hu, L. (1988). Social projection as a function of cognitive mechanisms: Two meta-analytic integrations. British Journal of Social Psychology, 27,
333–356.
Peterson, P. L., Hawkins, J. D., Abbott, R. D., & Catalano, R. F. (1994). Disentangling
the effects of parental drinking, family management, and parental alcohol norms on
current drinking by Black and White adolescents. Journal of Research on Adolescence, 4, 203–228.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton
University Press.
Ross, L., Greene, D., & House, P. (1977). The “false consensus effects”: An egocentric
bias in the social perception and attribution process. Journal of Experimental Social
Psychology, 13, 279–301.
Rothman, A. J., Klein, W. M., & Weinstein, N. D. (1996). Absolute and relative biases
in estimation of personal risk. Journal of Applied Social Psychology, 26,
1213–1236.
Sanders, G. S., & Mullen, B. (1983). Accuracy in perceptions of consensus: Differential
tendencies of people with majority and minority positions. European Journal of Social
Psychology, 13, 57–70.
Gerrard, et al.
Sharp, M. J., & Getz, J. G. (1995). Substance use as impression management. Personality
and Social Psychology Bulletin, 22, 60–67.
Sherman, S. J., Presson, C. C., Chassin, L., Corty, E., & Olshavsky, R. (1983). The false
consensus effect in estimates of smoking prevalence: Underlying mechanisms. Personality and Social Psychology Bulletin, 9, 197–207.
Sherman, S. J., Presson, C. C., & Chassin, L. (1984). Mechanisms underlying the false
consensus effect: The special role of threats to the self. Personality and Social
Psychology Bulletin, 10, 127–138.
Shrauger, J. S. (1975). Responses to evaluation as a function of initial self-perception.
Psychological Bulletin, 82, 581–596.
Shrauger, J. S., & Kelly, R. J. (1988). Global self-evaluation and changes in self-description as a function of information discrepancy and favorability. Journal of Personality,
56, 709–728.
Shrauger, J. S. & Rosenberg, S. E. (1970). Self-esteem and the effects of success and
failure feedback on performance. Journal of Personality, 38, 404–417.
Smith, G. E., Gerrard, M., & Gibbons, F. X. (1997). Self-esteem and the relation between
risk behavior and perceived vulnerability. Health Psychology, 16, 137–146.
Steele, C. M., Spencer, S. J., & Lynch, M. (1993). Self-image resilience and dissonance:
The role of affirmational resources. Journal of Personality and Social Psychology, 64,
885–896.
Suls, J. M., Wan, C. K., & Sanders, G. S. (1988). False consensus and false uniqueness
in estimating the prevalence of health-protective behaviors. Journal of Applied &
Social Psychology, 18, 66–79.
Sussman, S., Dent, C. W., Mestol-Rauch, J., Johnson, C. A., Hansen, W. B., & Flay, B. R.
(1988). Adolescent non-smokers, triers, and regular smokers’ estimates of cigarette
smoking prevalence: When do overestimates occur and by whom? Journal of Applied
Social Psychology, 18, 537–551.
Swann, W. B. (1984). Quest for accuracy in person perception: A matter of pragmatics.
Psychological Review, 91, 457–477.
Swann, W. B., & Read, S. J. (1981). Self-verification processes: How we sustain our
self-conceptions. Journal of Experimental Social Psychology, 17, 351–370.
Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological
perspective on mental health. Psychological Bulletin, 103, 193–210.
Tennen, H., & Affleck, G. (1993). The puzzles of self-esteem: A clinical perspective. New
York: Plenum.
Torres, R., Fernandez, F., & Maciera, D. (1995). Self-esteem and value of health as
correlates of adolescent health behavior. Adolescence, 30, 403–412.
Ulrick-Jakubowski, D., Russell, D. W., & O’Hara, M. W. (1988). Marital adjustment
difficulties: Cause of consequence of depressive symptomatology. Journal of Social
and Clinical Psychology, 7, 312–318.
Weinstein, N. D. (1980). Unrealistic optimism about future life events. Journal of
Personality and Social Psychology, 39, 431–457.
Weinstein, N. D., & Nicolich, M. M. (1993). Correct and incorrect interpretations of
correlations between risk perceptions and risk behaviors. Health Psychology, 12,
235–245.
Self-esteem and Self-serving Health Cognitions
Weinstein, N. D., Rothman, A. J., & Nicolich, M. M. (1998). Using correlations to study
relationships between risk perceptions and preventive behavior. Psychology and
Health, 13, 479–501.