The role of self-esteem and anxiety in decision making for self versus others in relationships

Abstract

Previous research has documented a tendency for people to make more risk-seeking decisions for others than for themselves in relationship scenarios. Two experiments investigated whether this self-other difference is moderated by participants' self-esteem and anxiety levels. In Experiment 1, lower self-esteem and higher anxiety levels were associated with more risk-averse choices for personal decisions but not for decisions for others. Therefore, participants with lower self-esteem/higher anxiety showed greater self- other differences in comparison to participants with higher self-esteem/lower anxiety levels. Experiment 2 demonstrated that this effect was largely mediated by participants' expectations of success and feelings about potential negative outcomes. These results are discussed in the context of "threats to the self," with a central role played by anxiety and self-esteem threats in personal decision making but not in decision making for others. (PsycINFO Database Record (c) 2007 APA ) (journal abstract)

Full text

Journal of Behavioral Decision Making
J. Behav. Dec. Making,18: 125–144 (2005)
DOI: 10.1002/bdm.490
The Role of Self-esteem and Anxiety
in Decision Making for Self versus
Others in Relationships
LAURA D. WRAY*and ERIC R. STONE
PsychologyDepartment,Wake ForestUniversity, North Carolina, USA
ABSTRACT
Previous research has documented a tendency for people to make more risk-seeking
decisions for others than for themselves in relationship scenarios. Two experiments
investigated whether this self – other difference is moderated by participants’ self-
esteem and anxiety levels. In Experiment 1, lower self-esteem and higher anxiety levels
were associated with more risk-averse choices for personal decisions but not for deci-
sions for others. Therefore, participants with lower self-esteem/higher anxiety showed
greater self – other differences in comparison to participants with higher self-esteem/
lower anxiety levels. Experiment 2 demonstrated that this effect was largely mediated
by participants’ expectations of success and feelings about potential negative out-
comes. These results are discussed in the context of ‘‘threats to the self,’’ with a central
role played by anxiety and self-esteem threats in personal decision making but not in
decision making for others. Copyright #2005 John Wiley & Sons, Ltd.
key words decision making; self – other differences; relationships; self-esteem;
anxiety; risk taking
Although decision-making research has long tackled the question of how people make decisions for them-
selves (cf. Abelsen & Levi, 1995; Kahneman, Slovic, & Tversky, 1982), only recently have researchers
begun to investigate how people make decisions for other people. Since people make decisions for others
in a wide variety of contexts, ranging from medical decisions regarding treatment (e.g., Raymark, 2000)
to business decisions (e.g., Borresen, 1987) to providing advice or making decisions for others regarding
relationships (e.g., Beisswanger, Stone, Hupp, & Allgaier, 2003), it is important to understand how decision
making for others differs from deciding for the self.
SELF – OTHER DIFFERENCES IN RISK-TAKING BEHAVIOR
The majority of research that has compared personal decision making to decision making for another person
has investigated whether there are self – other differences in people’s tendency to take risks (but see
Copyright #2005 John Wiley & Sons, Ltd.
* Correspondence to: Laura D. Wray, Human Development and Family Studies, 110 Henderson Building South, The Pennsylvania State
University, University Park, PA 16802, USA. E-mail: ldw134@psu.edu

Kray, 2000; Kray & Gonzalez, 1999, for investigations that did not examine risk-taking behavior per se).
In particular, a number of studies have investigated whether people make more risk-averse or risk-seeking
monetary choices for the self versus for another person when presented with risky gambles (e.g., Cvetkovich,
1972; Stone, Yates, & Caruthers, 2002; Teger & Kogan, 1975). Although certain specific conditions (such as
reciprocal decision making) can produce self – other differences (Teger & Kogan, 1975), generally no self –
other differences have been uncovered in risky monetary situations.
Recently, however, Beisswanger et al. (2003) documented a large self – other difference in risk-taking
behavior when making relationship decisions for a friend versus for the self.
1
For example, undergraduate
participants were told to imagine that they noticed someone attractive across the room at a fraternity party.
They were given the choice between a risk-seeking option (e.g., introducing themselves) and a more risk-
averse option (e.g., staying with their friends). Participants were more likely to choose a risk-seeking option
when deciding for or giving advice to a same-sex friend than they were to choose that option for themselves,
although there were no differences in participants’ responses when asked to ‘‘decide for’’ versus ‘‘give
advice to’’ a friend. In addition, Beisswanger et al. differentiated between what they labeled high- and
low-impact decisions, depending on the seriousness of the decision’s consequences. No self – other differ-
ence was found in the high-impact condition (e.g., eloping with someone), but a large self – other discre-
pancy arose in the low-impact condition (e.g., introducing him- or herself at a party), with people being
more risk taking for others than for themselves.
Beisswanger et al. (2003) speculated that this self –other difference was in part due to the value placed on
risk in our culture (see, e.g., Wallach & Wing, 1968). Using Beisswanger et al.’s decision situations, Allgaier
and Stone (2002) confirmed that risk taking is valued in low-impact decision situations but not in high-
impact scenarios. Thus, viewing risk as a cultural value can explain why people make risky decisions for
others in certain situations.
However, to the extent that risk is valued, everyone should want to take more risks, and it seems logical
that this desire would be particularly strong for decisions made for oneself. Indeed, Wallach and Wing (1968)
found that most people feel that their level of risk-taking behavior is greater than that of others. The self –
other differences found by Beisswanger et al. (2003), however, demonstrate that many individuals are not as
risk taking for themselves as they are for others. Thus, risk taking appears to be valued in theory but not
always in practice for oneself. The question then becomes, why do people avoid taking risks for themselves
when they are willing to take those risks for other people?
One possibility, also suggested by Beisswanger et al. (2003), is that people consider potential negative
outcomes to a greater extent when making decisions for themselves than for other people. This possibility
is consistent with work by Kray (Kray, 2000; Kray & Gonzalez, 1999) suggesting that people’s personal
decisions are influenced by more factors than are their decisions for other people. Further, even if people
consider the same information in both situations, they may react to the information differently. For example,
Vorauer and Ratner (1996) found that people thought their actions would be more inhibited by fear of rejec-
tion than would a potential partner’s actions. Additional evidence that people are less affected by the poten-
tial negative outcomes when deciding for others was found by Kray (2000), who showed that individuals
giving advice to others felt less regret and less blamed for a negative outcome compared to personal decision
makers. Similarly, certain visceral emotions such as anxiety are perceived as being stronger for the self than
for another person (see Loewenstein, Weber, Hsee, & Welch, 2001). As most of the low-impact scenarios
investigated by Beisswanger et al. (2003) contained the possibility of negative outcomes such as rejection, it
seems plausible that these potential negative outcomes had a greater influence on deciding for oneself than
on deciding for another person, thereby leading to the observed self – other differences.
1
Although the decision scenarios were arguably not relationship scenarios as the term is often used in social psychology, in that they
involved the potential for forming a romantic relationship rather than decisions made within relationships, we are using the term
relationships for ease of exposition and for consistency with the Beisswanger et al. (2003) study.
126 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

If the preceding argument is correct, it suggests that decision making for the self, and thus self – other
differences, should be influenced by how much people are affected by concerns such as rejection, or more
generally, by a need to protect one’s self-image from threats to the self (see Larrick, 1993). To the extent that
people are only minimally affected by these concerns when deciding for themselves, self – other differences
should disappear or at least be reduced. Although a number of personality characteristics might moderate the
degree to which people are affected by these types of concerns, two that could be expected to be particularly
important are self-esteem and anxiety.
Self-esteem
Self-esteem is generally considered to be the degree to which a person likes, values, and accepts himself or
herself (Rogers, 1951). Self-esteem is an important variable to consider when there is potential for failure or
rejection in personal decisions (Brown & Dutton, 1995). In particular, self-esteem has been linked to rejec-
tion sensitivity (Downey & Feldman, 1996). According to sociometer theory, trait self-esteem develops from
evaluations of one’s experiences with acceptance and rejection over time, and is related to how accepted
people generally feel (Leary & Downs, 1995; Leary & MacDonald, 2003).
Many theorists argue that people are motivated to preserve self-esteem (e.g., Steele, 1988; however, see,
e.g., Swann, Stein-Seroussie & Giesler, 1992, for an opposing view), and some suggest that people attempt to
maintain self-esteem due to a fundamental motive of avoiding social exclusion (Baumeister & Leary, 1995;
Leary & Downs, 1995). To the extent this is the case, one’s self-esteem would reflect the perceived state of
that person’s relationships with other people (Leary & MacDonald, 2003). Further, Leary and Downs (1995)
suggested that concerns with social acceptance are more salient for individuals with lower self-esteem,
which may deter them from actions that would undermine social acceptance. In contrast, those individuals
with higher self-esteem are free to take interpersonal risks because they see themselves as more acceptable to
others. Brockner, Wiesenfeld, and Raskas (1993) made a similar argument, suggesting that individuals with
lower self-esteem are more likely to self-protect by acting in ways that minimize bad things happening while
higher self-esteem people are more apt to self-enhance and thus make more risky choices.
Consistent with the above argument, Josephs, Larrick, Steele, and Nisbett (1992) found a correlation
between risk-seeking tendencies and self-esteem in the domain of monetary decisions, with higher self-
esteem associated with more risks taken in positive gain scenarios. When decision outcomes were withheld
from the individuals in order to remove the ‘‘threat’’ to self-esteem, however, the low self-esteem group was
equally as risk seeking as the high self-esteem group. This research suggests, then, that the potential damage
to self-esteem due to failure motivates those with lower self-esteem to avoid risk. Although the relationship
between self-esteem and risk-taking behavior has not been examined for decision making for others, it seems
likely that certain types of threats should be less salient when deciding for others than for oneself. In parti-
cular, concerns with rejection should pose a greater threat when deciding for the self than for someone else.
To the extent that these threats to the self have a greater influence when deciding for the self than for another
person, then, we would expect that self – other differences should be greater for people with lower self-
esteem than for people with higher self-esteem.
Anxiety
Anxiety arises in response to the perception of threat (Epstein, 1973). Often, the threat experienced is a
threat to self-esteem (Epstein, 1973). Indeed, most studies on self-esteem have found a link between low
self-esteem and anxiety (Mruk, 1999). Although anxiety and self-esteem are highly negatively correlated,
the causal relationship between these two variables remains unclear. Some say that anxiety is a mechanism
that alerts an individual to problems with the self-system and the need to maintain self-esteem (Epstein,
1973; Leary & Downs, 1995). Others argue in favor of a reciprocal relationship between anxiety and
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 127
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

self-esteem (Coopersmith, 1967; Horney, 1950), whereby those who are highly anxious over time will come
to think of themselves in less-positive terms, and those with low self-esteem are hypersensitive to threats,
putting them in a constant state of anxiety. Still others argue that self-esteem serves as an anxiety buffer
(Greenberg, Pyszczynski, & Solomon, 1995; Mruk, 1999). Evidence for this view is found in that those with
higher self-esteem perform better under stress than do those with lower self-esteem (Solomon, Greenberg, &
Pyszczynski, 1991). Furthermore, increased anxiety leads to a greater focus on self-esteem maintenance, and
the temporary bolstering of self-esteem leads to lower reported anxiety (Solomon et al., 1991).
In keeping with the results regarding self-esteem, anxiety has been associated with risk-taking tendencies
but in the opposite direction (Eisenberg, Baron, & Seligman, 1996; Raghunathan & Pham, 1999). For exam-
ple, Raghunathan and Pham (1999) found that inducing anxiety led to more risk-averse decisions in a gam-
bling as a well as a job-selection task. Additionally, research has suggested that, across a range of different
situations, the occurrence of negative events is seen as being more likely by people high in anxiety (Butler &
Mathews, 1983, 1987; Eisenberg et al., 1996).
A small amount of research has examined the role of anxiety in deciding for other people. Using a gam-
bling task, Raghunathan and Pham (1999) found no difference between anxious and sad participants in how
they made decisions for others. However, it is unclear whether their results found with gambling would gen-
eralize to the present scenarios. Further, due to their design, it is unclear whether both anxiety and sadness
had no effect on decision making for others, or whether the two mood states had similar effects. Eisenberg
et al. (1996) found results similar to those of Ragunathan and Pham (1999), across both a wider range of
scenarios and examining anxious versus non-anxious participants. However, as the purpose of their research
was not to explore self – other differences per se, their results were not reported in enough detail to determine
whether the lack of anxiety effects held in all domains or even whether self – other differences occurred at all
in their study.
In sum, the present evidence suggests that, like low self-esteem, high anxiety should be associated with
making risk-averse decisions for the self. Also, the research conducted to date suggests that these effects
should be reduced, if not eliminated, in decision making for other people. To the extent the above arguments
hold for decision making about relationships, then, the previously-documented self – other differences
detected in that domain should disappear, or at least be reduced, in participants with lower levels of anxiety.
EXPERIMENT 1
Our primary hypothesis was that self –other differences would be greater in participants with lower levels
of self-esteem/higher levels of anxiety than in participants with higher levels of self-esteem/lower levels
of anxiety. To test this prediction, we chose five scenarios from the Beisswanger et al. (2003) study that
produced self – other differences, all of which included the potential for interpersonal rejection. We also
measured participants’ levels of self-esteem and anxiety, and examined whether these variables moderated
the influence of self versus other decision making on level of risk-taking behavior.
One other methodological point is worth discussing. We have been assuming that risk is generally valued,
at least in the types of scenarios that we are using. Nonetheless, there are individual differences in the extent
to which risk is valued. In addition, there could be systematic tendencies to assume that other people value
risk either to a greater or lesser extent than oneself. In an attempt to minimize this concern, we followed the
procedure used by Beisswanger et al. (2003) and had participants make decisions for a specific friend who
was similar to the participant. This decision was based in part on research by Hsee and Weber (1997), who
found that participants predicted that others would display the same level of risk taking that they themselves
would, but only if the person was ‘‘concrete’’ (defined, for example, as the person next to them in the room).
By making the participant decide for a particular person, then, it was assumed that the person would be
more concrete to the participant, and hence could be assumed to have more similar values to him or her.
128 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

This procedure is also in keeping with research conducted by Hoch (1987), who found that participants dis-
played a greater tendency to project their own values onto spouses or peers than onto the average American
consumer.
METHOD
Participants
Participants were 233 undergraduate students (80 male, 153 female) from a small southeastern US
university enrolled in introductory level psychology courses. They participated as one means of partially
fulfilling a course requirement.
Materials
Self-esteem scale. The Rosenberg self-esteem scale (Rosenberg, 1965) is an index of one’s global or
trait self-esteem, and both its reliability and validity have been demonstrated in previous research (e.g.,
Baumeister, Tice, & Hutton, 1989; Rosenberg, 1979). This questionnaire consists of ten items with a
four-point Likert scale ranging from strongly agree to strongly disagree, of which five items were
reverse-scored. Each item was coded from 1 to 4 and summed so that the possible score range was 10 to
40, with higher scores indicating higher self-esteem.
Anxiety scale. We used the trait portion of the state-trait-anxiety inventory (STAI-T) (Spielberger,
Gorsuch, & Lushene, 1970). This is a widely used and reliable measure of anxiety (Barnes, Harp, & Jung,
2002). The questionnaire consists of 20 items, which participants answer by choosing (1) almost never, (2)
sometimes, (3) often, or (4) almost always, with nine of the items reverse-scored. The possible range of
scores was from 20 to 80, with higher scores indicating higher anxiety.
Decision scenarios. Five different scenarios, based on those used by Beisswanger et al. (2003), were used
to measure level of risk taking in relationships. These five scenarios were chosen out of the scenarios used by
Beisswanger et al. because they produced self – other differences in their research, and because they con-
tained the potential for rejection. Each scenario presented a situation where the participant needed to decide
whether to engage in a safe (risk-averse) activity, or a more risky but potentially rewarding alternative.
For example, in one scenario, participants were told to imagine that they saw someone cute talking with
some friends at a bar, and needed to decide whether to introduce themselves to that person of the opposite
sex or to stay with their friends. All of the scenarios were designed such that there was a chance for rejection
but that any negative outcomes would likely not have a serious effect on the person.
The precise wording of the questionnaires varied according to decision type (self vs. other) and gender
(male vs. female). In the ‘‘self ’’ condition, participants were asked to imagine that they were not currently in
a relationship, and to answer as they would personally react to the given situation. In the ‘‘other’’ condition,
participants began by writing down the name of a same-sex friend who is similar to them, and were then
asked to respond to the scenarios as if making the decision for that friend. Asking the participant to respond
for a particular friend was done to make the situation more concrete for the participant, as discussed pre-
viously. For all scenarios, participants were asked to answer either as if they were not presently in a relation-
ship or as if the friend was not presently in a relationship, depending on condition.
In addition, we attempted to manipulate the level of social evaluation in the decision situation. Our
original idea was that low self-esteem participants would be more risk averse in situations where there
was a greater potential for social evaluation (e.g., many people in a quiet room so that the situation would
be observed by many people) compared to situations where there was less potential for social evaluation
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 129
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

(e.g., when no one was likely to observe the interaction). However, this manipulation had only a very limited
effect,
2
and we suspect that our manipulation was too subtle. To conserve space, then, we will not consider
this variable further in this paper.
Finally, the wording of the scenarios was manipulated to match gender pronouns with the participant’s
gender. A sample scenario used in the male, decision making for self condition is presented below:
You and your friends are studying at a local coffee shop. You see a girl you are interested in studying with
and her friends at a nearby table. You want to go over and talk to her but last time you saw her it was your
birthday and you feel like you may have made a fool of yourself. She and her friends get in line to get
more coffee, and you consider doing the same thing so you can talk to her. You decide to:
A. Go up and talk to her.
B. Avoid her.
Presented below is the same scenario in the decision making for other condition.
You and your friends are studying at a local coffee shop. Your friend sees a girl that he’s interested in
studying with and her friends at a nearby table. He wants to go over and talk to her but last time he
saw her it was his birthday and he feels like he may have made a fool of himself. She and her friends get
in line to get more coffee, and he considers doing the same thing so he can talk to her. You tell him he should:
A. Go up and talk to her.
B. Avoid her.
Procedure
The primary design was a 2 (decision type: self vs. other) 2 (participant gender: male vs. female) mixed
design, where decision type was manipulated within subjects. The within-subjects nature of the design was
used to increase the power of the experiment by removing error due to individual differences, as well as to
allow for an examination of the consistency of participants’ responses across the two decision types. Due to
concerns about experimenter demand, a time interval of at least two weeks separated the two administrations
of the experiment.
Specifically, the experiment consisted of two separate parts, each lasting approximately 30 minutes, and
the participants took part in each 30-minute procedure in groups of 10 to 20. In each session, participants
were asked to complete their responses to the relationship scenarios in terms of how they would decide for
themselves or for others, with the self and other scenarios counterbalanced and separated by the two-week
time interval. A different experimenter tested the participants during the second session, and nothing about
the previous session was mentioned to participants.
At the completion of the decision scenarios in the first portion of the experiment, participants were given
the Rosenberg self-esteem scale, an exploratory questionnaire, and an unrelated task. The participants com-
pleted the STAI-T and several other exploratory questionnaires at the end of the second session. The self-
esteem and anxiety scales were administered this way due to time constraints in each session, but because
these scales measure stable traits, time of completion should not influence responses (see Marsh, 1986).
2
There were actually trends in the predicted direction for both decision making for the self and decision making for others, but the effects
produced were quite small (about 0.2 – 0.3 points on a 0 – 5 scale). We examined this issue inferentially by running a 2 (self-esteem: low
vs. high) 2 (social evaluation: low vs. high) ANOVA both on decision making for the self and decision making for others. (We ran a
similar analysis using anxiety rather than self-esteem as well.) None of the effects involving social evaluation were significant at the
alpha ¼0.05 level, although the main effect of social evaluation was relatively close to significant in each of the analyses. Given the
small size of the effects, their non-significance at the 0.05 level, and the lack of evidence of any interaction (all interaction Fs<1.0),
we decided it was not worth further pursuing social evaluation effects in this research.
130 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

When participants finished all of the required materials, they were thanked for participating and given an
opportunity to hear an explanation of the experiment’s purpose at the conclusion of the second session.
Results
Nineteen participants were not included in the analyses because they failed to return for the second portion
of the experiment, and two additional participants were eliminated from the analyses involving anxiety for
failure to respond to all of the items on that scale.
Repeated-measures ANOVAs
To determine what role, if any, self-esteem and anxiety play in deciding for the self versus another person,
we began by treating both variables as dichotomous by taking median splits.
3
Mean scores for the low versus
high self-esteem groups were 28.76 and 36.72, respectively, and mean scores for high versus low anxiety
groups were 45.51 and 32.62, respectively. We then conducted two separate repeated-measures ANOVAs,
one with self-esteem and one with anxiety.
Specifically, we conducted a 2 (decision type: self vs. other) 2 (self-esteem: low vs. high) 2 (partici-
pant gender: male vs. female) repeated-measures ANOVA on level of risk taking, where decision type was
the within-subjects variable. There was a main effect of gender, whereby males (M¼3.92) tended to be more
risk seeking than were females (M¼3.34), F(1, 210) ¼17.23, p<0.001. However, there were no interac-
tions with gender and any of the other predictor variables, all ps>0.15. Therefore, we provide means col-
lapsed across gender in Table 1, although all the inferential analyses we report include gender in the analysis.
In keeping with the Beisswanger et al. (2003) study, there was a main effect of decision type, whereby
participants made more risk-seeking decisions for others (M¼4.19) than for themselves (M¼2.88),
F(1, 210) ¼113.61, p<0.001. Additionally, there was a main effect of self-esteem, whereby high self-
esteem participants (M¼3.75) made more risk-seeking decisions than did low self-esteem participants
(M¼3.35), F(1, 210) ¼8.43, p¼0.004. Most importantly, we found the predicted decision type by self-
esteem interaction, F(1, 210) ¼22.17, p<0.001. As shown in Table 1, although both high and low self-
esteem individuals displayed a self – other difference (both ps<0.001), this difference was more pronounced
for people with low self-esteem (Mdifference ¼1.82) than for participants with high self-esteem
(Mdifference ¼0.72). This interaction can also be seen in the fact that there were no differences in deciding
for another person between those participants with low (M¼4.26) and high (M¼4.11) self-esteem,
3
Although we dichotomize self-esteem and anxiety in terms of ‘‘high’’ and ‘‘low,’’ it is important to note that low self-esteem, for
example, is only ‘‘low’’ relative to the other scores. These participants do not rate themselves as low in any absolute sense; rather, they
simply provide lower ratings than do the individuals classified as having high self-esteem (see Baumeister, Tice, & Hutton, 1989).
Table 1. Experiment 1: Mean risk-taking scores by decision type and
self-esteem
Self-esteem level
Decision type Low High
Self 2.44 (1.46) 3.39 (1.36)
Other 4.26 (0.80) 4.11 (1.05)
Note: Standard deviations are in parentheses. Scores range from 0 (all risk-
averse choices) to 5 (all risk-seeking choices). Sample sizes were N¼114 for
the low self-esteem group and N¼100 for the high self-esteem group.
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 131
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

F(1, 210) ¼1.21, p¼0.27, but a large difference in deciding for oneself between those participants with low
(M¼2.44) and high (M¼3.39) self-esteem, F(1, 210) ¼30.90, p<0.001.
A 2 (decision type: self vs. other) 2 (anxiety: low vs. high) 2 (participant gender: male vs. female)
repeated-measures ANOVA produced similar results. As before, there was only a main effect of gender
and no significant interactions, so the means as displayed in Table 2 are collapsed over gender, although
again all inferential tests include gender. Also in keeping with the previous analysis, there was a main effect
of decision type, whereby participants made more risk-seeking decisions for others (M¼4.19) than for
themselves (M¼2.88), F(1, 208) ¼116.36, p<0.001. Additionally, there was a main effect of anxiety,
whereby low-anxiety participants (M¼3.75) made more risk-seeking decisions than did high-anxiety par-
ticipants (M¼3.31), F(1, 208) ¼10.36, p¼0.001. As with self-esteem, we found the predicted decision type
by anxiety interaction, F(1, 208) ¼19.04, p<0.001.
4
As shown in Table 2, although both high and low anxi-
ety individuals displayed a self – other difference (both ps<0.001), this difference was more pronounced in
people with greater anxiety levels (Mdifference ¼1.81) than in participants with lower anxiety levels
(Mdifference ¼0.82). As with self-esteem, there were no differences in deciding for another person between
those participants with high (M¼4.22) and low (M¼4.16) anxiety, F(1, 208) ¼0.42, p¼0.52, but a large
difference in deciding for oneself between those participants with high (M¼2.41) and low (M¼3.34)
anxiety, F(1, 208) ¼30.49, p<0.001.
Correlational analyses
The previous analyses suggest that self-esteem and anxiety influence decision making for the self but not for
other people. To ensure that the lack of relationships regarding decision making for others was not due to
dichotomizing self-esteem and anxiety, we kept self-esteem and anxiety as continuous variables and used
them to predict both risk taking for the self and risk taking for others. In keeping with the previous analysis,
we found that both anxiety and self-esteem were correlated with the number of risky decisions made for the
self (r¼0.39, p<0.001; r¼0.35, p<0.001, respectively), but neither anxiety nor self-esteem were
related to the number of risky decisions made for the other person (r¼0.01, p¼0.88; r¼0.08, p¼
0.26, respectively).
Given that the results with self-esteem and anxiety were so similar, we next examined whether either of
these variables predicted any unique variance in level of risk taking. The two variables were indeed highly
correlated (r¼0.73, p<0.001), in keeping with previous research (e.g., Coopersmith, 1967; Heatherton &
Ambady, 1993). To determine the amount of unique variance associated with the two predictors, we con-
ducted a simultaneous multiple regression with level of risk taking for the self as the criterion variable
4
We also ran these analyses separately for the individual scenarios, and there was a significant main effect of decision, decision anxiety
interaction, and decision self-esteem interaction (all ps<0.06) for each of the five scenarios.
Table 2. Experiment 1: Mean risk-taking scores by decision type and
anxiety
Anxiety level
Decision type Low High
Self 3.34 (1.36) 2.41 (1.49)
Other 4.16 (1.02) 4.22 (0.83)
Note: Standard deviations are in parentheses. Scores range from 0 (all risk-
averse choices) to 5 (all risk-seeking choices). Sample sizes were N¼107 for
the low-anxiety group and N¼105 for the high-anxiety group.
132 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

and anxiety and self-esteem as predictors. This analysis showed that anxiety (partial r ¼0.19, p¼0.006)
predicted a unique amount of variance in decision making for the self, and self-esteem was marginally sig-
nificant as well (partial r ¼0.13, p¼0.06). The unique predictability of self-esteem and anxiety after taking
into account the large correlation between the two demonstrates that self-esteem and anxiety are both impor-
tant in explaining risk-taking behavior.
Consistency in self vs. other decision making
Recall that the primary rationale for using a within-subjects design was to reduce the amount of individual
difference variability in the analyses. To determine whether or not this technique was successful, we exam-
ined the consistency among participants’ judgments across self versus other decision making. Surprisingly,
there was virtually no relationship between how people decided for themselves versus for others, r¼0.10,
p¼0.15. Although not predicted, this finding provides further support to the notion that people make deci-
sions differently when deciding for themselves versus for someone else.
Discussion
As predicted, self-esteem and anxiety both played a moderating role in terms of the observed self – other
differences in risk taking. Nor were these effects small. Indeed, participants who were below average on
self-esteem or above average on anxiety produced a self – other difference (M¼1.82 and M¼1.81, respec-
tively) more than twice as great as did those participants above average on self-esteem or below average on
anxiety (M¼0.72 and M¼0.82, respectively).
Subsequent analyses revealed that these personality variables were related to self – other differences
primarily as a result of their role in deciding for the self. Specifically, participants with lower self-esteem
and higher anxiety demonstrated less personal risk taking than did those individuals with higher self-esteem
and lower anxiety levels. Although the relationships between self-esteem and anxiety with risk taking were
clearly demonstrated, this study provided no evidence as to why these relationships exist. In other words,
how does lower self-esteem/higher anxiety work to influence the level of risk taking in relationships?
Experiment 2 was designed to provide some insight into that issue.
EXPERIMENT 2
Expected utility theory (e.g., von Neumann & Morgenstern, 1944), or more generally, subjective expected
utility theory (e.g., Savage, 1954), provides a useful perspective for seeing how personality variables such as
self-esteem and anxiety could influence decision making. When used as descriptive theories of choice, these
theories suggest that people’s decisions are based on two factors: the uncertainties involved in the decision
and the values that people assign to the outcomes. For example, assume that your good friend Mark is trying
to decide whether to ask someone attractive at a bar to dance or stay with his friends to talk. According to
expected utility models, his decision will depend on how he perceives the two alternatives. That is, he will
consider the likelihood that the person would agree to dance if asked, as well as how good it would make him
feel if his invitation was accepted and how bad it would make him feel if rejected. He would then combine
that information into an overall evaluation of the alternative, according to a formula like the following:
EUðask to danceÞ¼ðpÞðu1Þþð1pÞðu2Þð1Þ
where pis the likelihood that the person would dance with him if asked, u
1
is the utility, or value, that he
would assign to his invitation being accepted, and u
2
is the utility, or value, of being rejected. Once this
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 133
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

information is combined into an overall evaluation of that alternative, it is compared to an overall evaluation
of staying with his friends, and he would choose the alternative with the greater expected utility. Although
these types of expectation models are limited in multiple ways (see Larrick, 1993; Plous, 1993), they none-
theless can serve as useful approximations of how people make decisions (see, e.g., Yates, 1990).
According to this theoretical approach, then, self-esteem and anxiety could relate to risk taking by affect-
ing either: (1) the probability that the risk-seeking choice would produce a positive outcome (e.g., the person
agrees to dance); (2) the value placed on the positive outcome of the risk-seeking alternative (e.g., how good
it would feel to dance with the person); (3) the value placed on the negative outcome of the risk-seeking
alternative (e.g., how bad it would feel to be rejected); and/or (4) the value placed on the risk-averse alter-
native (i.e., how good it would feel to stay with friends and not ask the person to dance). The primary goal of
our second study was to determine which, if any, of these components of the decision process were related to
self-esteem and anxiety.
Using a similar theoretical framework, Brockner et al. (1993) defined behavioral and psychological stra-
tegies for low versus high self-esteem individuals in terms of expectancy-value discrepancy. Those with
higher self-esteem feel as if they can achieve the things that they desire and engage in self-enhancement,
which refers to acting or thinking in ways that provide the potential for good things to happen (e.g., by mak-
ing risk-seeking choices). Individuals with lower self-esteem, however, have a greater discrepancy between
their desires and perceived level of competence, which leads them to engage in self-protection, responding to
situations in ways that minimize negative outcomes (e.g., by making risk-averse choices). Much of the rea-
son for these different strategies comes from different expectations of success, which tend to vary more
between low and high self-esteem individuals than do values of outcomes (see Brockner et al., 1993). Thus,
it seems plausible that the effect of self-esteem on risk taking for the self is mediated in part by different
expectations of success, i.e., decision component (1) above.
Although this hypothesis has not been directly tested within the relationship domain or with self-esteem,
Eisenberg et al. (1996) tested a similar issue by examining the role of anxiety. Specifically, they provided
people with a variety of risky decision scenarios across many different domains and measured participants’
levels of risk seeking for themselves as well as their probabilities and utilities. They found that greater levels
of anxiety were associated with greater levels of risk aversion and some speculative evidence that this effect
was mediated by participants’ probability estimates. Similar results were found by Butler and Mathews
(1983, 1987), although there was some evidence that the relationship between anxiety and perceived like-
lihood held for negative but not positive occurrences (Butler & Mathews, 1983).
In sum, it is plausible that the relationship between both self-esteem and anxiety with risk taking may be
mediated by perceived probability of success. In addition, the research by Butler and Mathews (1983) sug-
gests that highly anxious people may perceive the negative outcome as being worse than would less anxious
people. More specifically, they defined threat as a function of the probability of a negative event multiplied
by the cost of the bad event, similar to the idea behind expected utility theory. Butler and Mathews then
found differences between anxious and less anxious participants in terms of both the probability of the nega-
tive event as well as the cost of the negative event, suggesting that both components may be related to anxiety
level. Similarly, Eisenberg et al. (1996) found that anxiety was associated with the perceived badness of an
event, though to a lesser degree than probability of its occurrence. Further, they found no relationship
between anxiety and the perceived goodness of an event.
Based on these previous findings, then, we predicted that self-esteem and anxiety would be related to
perceived probability of success (i.e., decision component (1) above) as well as feelings about negative
outcomes (i.e., decision component (2) above). Those with lower self-esteem/higher anxiety should think
that a successful outcome is less likely, and feel worse about a potential bad outcome, than would
those participants with lower self-esteem/higher anxiety. Given the lack of research demonstrating
any influence on either of the other two components, we made no predictions regarding either of those
components.
134 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

On the other hand, since we found no influence of self-esteem or anxiety on decision making for others
in Experiment 1, it seems reasonable to expect that neither of these personality variables would be related
to any of the four components of the decision situation. In support of this possibility, Eisenberg et al.
(1996) found, in regard to other decision making, no difference in terms of ratings of probability of
success, feelings toward the positive outcome, or feelings toward the negative outcome depending on
anxiety level. Similarly, Butler and Mathews (1983, 1987) found no difference between anxious and less
anxious individuals in terms of predicting probability of success for others. Thus, we expected that for
decision making for others, there would be no effect of anxiety (or self-esteem) on any of the decision
components.
To test these hypotheses, we constructed questions to measure each of the four components of the deci-
sion process discussed previously for each of the scenarios. We then measured the effects of self-esteem
and anxiety on each of the four components both for decision making for the self and for others, as well
as the relationship between each of those components and risk aversion. Finally, we tested whether
the relationships between the personality variables and risk seeking are mediated by the particular
components of the decision process, specifically predicting that perceived probability of success and
feelings about rejection would be most likely to mediate the effects of self-esteem and anxiety on risk
taking for the self.
Method
Participants
Participants were 270 introductory psychology students (158 female, 112 male) from a small southeastern
US university enrolled in introductory level psychology courses. They participated as one means of partially
fulfilling a course requirement.
Procedure
We employed a 2 (decision type: self vs. other) 2 (participant gender: male vs. female) between-subjects
design. The self versus other variable was manipulated between subjects given the lack of relationship
between deciding for the self and the friend in Experiment 1. The primary dependent variables were the
probability of having a positive outcome if the risk-seeking option was taken, the participant’s reaction to
both the positive and negative outcome of the risk-seeking option as well as a reaction to the risk-averse
option, and the actual decision made. Participants received either the self-decide or the other-decide form
of the questionnaire, block-randomized for each gender.
As in Experiment 1, participants were provided with decision scenarios relevant to themselves and were
asked to answer either as if they were not presently in a relationship, or for a specific friend as if he or she was
not presently in a relationship. These scenarios were identical to the ones used in Experiment 1, except that
we made minor wording clarifications and only used the high social evaluation conditions in Experiment 2.
Rather than making decisions about each scenario, however, participants were asked to answer four ques-
tions for each scenario regarding: the probability of a successful outcome (e.g., asking someone to dance and
having them accept); one’s feeling about the negative outcome (e.g., being turned down); one’s feeling about
the positive outcome (e.g., having the person agree to dance); and one’s feeling about the risk-averse out-
come (e.g., not asking the person to dance). The probability judgment was on a 0 – 100% scale, although
participants were provided with response options corresponding to each 10% (e.g., 0%, 10%, 20%, etc.)
rather than on a continuous scale. The other three questions were on a nine-point scale ranging from very
bad to very good. The specific wording was varied appropriately for both the self and other conditions, as
well as according to gender.
After responding to the questions regarding the four decision components for each scenario, participants
were given copies of the same scenarios to make decisions for either themselves or the other person, depending
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 135
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

on their condition. After this task, participants were given the Rosenberg self-esteem scale (Rosenberg, 1965)
and the trait portion of the STAI (Spielberger et al., 1970).
Results
Two of the participants did not respond to all items on the self-esteem questionnaire, and two participants did
not respond to all items on the anxiety questionnaire. They were excluded from the analyses regarding the
particular construct.
Replication of Experiment 1 results
Experiment 2 utilized a between-subjects design rather than the within-subjects design used in Experiment 1.
The use of the between-subjects design in the current study made it more straightforward to analyze the
results via regression analysis, enabling us to take advantage of the power of keeping self-esteem and anxiety
as continuous variables.
We began by centering each of our predictors and performing a regression analysis on level of risk-taking
behavior with decision-type (self vs. other), participant gender (male vs. female), self-esteem, and all two-
way and three-way interaction terms as predictor variables. Unlike in Experiment 1, there was no main effect
of gender. Specifically, females (M¼3.59) were almost equally as risk taking as were males (M¼3.62),
t(262) ¼0.23, p¼0.82. In keeping with Experiment 1, however, the interactions including gender were
all non-significant, all p>0.10. Also in keeping with Experiment 1, main effects of condition and self-
esteem revealed that participants were more risk seeking for others (M¼4.18) than for themselves
(M¼3.02), t(262) ¼8.70, p<0.001, and self-esteem level was positively correlated with risk taking (partial
r¼0.22), t(262) ¼3.68, p<0.001. Most importantly, the self-esteem by decision-type interaction was again
significant, though not as strongly as in Experiment 1 (partial r ¼0.13), t(262) ¼2.12, p¼0.04. Note that
the negative correlation indicates that the self – other difference became weaker as participants increased in
self-esteem level, similar to the results we obtained in Experiment 1.
We then performed the same analyses with anxiety, decision type, gender, and the interaction terms as
predictor variables in a regression equation predicting risk-taking behavior. Again, there were no significant
effects with gender, all ts<1. As in Experiment 1, there were main effects of condition and anxiety, whereby
participants were more risk seeking for others than for themselves, and anxiety level was negatively corre-
lated with risk taking (both ps<0.001). The anxiety by decision type interaction also mirrored the results
from Experiment 1 (partial r ¼0.17), t(262) ¼2.74, p¼0.007, whereby self – other differences were stron-
ger in participants with higher anxiety than in participants with lower anxiety levels.
Given the significant interactions between self-esteem/anxiety and decision type, we computed the simple
slope of each of the personality constructs with risk taking separately for decision making for the self and
for another person (see Aiken & West, 1991, p. 18).
5
In the decision making for self condition, risk taking
was significantly related to self-esteem (B ¼0.371) as well as to anxiety (B ¼0.482), t(266) ¼3.77,
p<0.001, t(266) ¼4.89, p<0.001, respectively. Thus, for every change of one standard deviation in the
personality construct, the amount of risk taking increased by a little less than half a point (out of a five-point
scale). Conversely, in the decision making for other condition, risk taking was not significantly related to
either self-esteem (B ¼0.115) or anxiety (B ¼0.110), both ps>0.20. The same general relationships
can be seen with the zero-order correlations, which were substantial in the decision making for self condition
(r¼0.28 for self-esteem and r¼0.38 for anxiety) but not in the decision making for other condition
5
Given the lack of significance of gender or any of the interactions involving gender, we did not consider gender when constructing these
regression models.
136 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

(r¼0.11 for self-esteem and r¼0.10 for anxiety). In addition, self-esteem and anxiety were again highly
correlated (r¼0.77; p<0.001).
Relationships of self-esteem and anxiety to the decision components
After establishing that the main results of Experiment 1 were replicated in Experiment 2, we examined the
relationships of self-esteem and anxiety to each of the components of the decision process to determine
more precisely just what effects self-esteem and anxiety had. The correlations are given in the top portion
of Table 3 for decision making for the self. As predicted, self-esteem and anxiety were correlated with the
belief that a successful outcome is probable for the self, with higher self-esteem and lower anxiety indivi-
duals tending to view success as more likely than those with lower self-esteem or higher anxiety. Also in line
with our predictions, individuals with higher self-esteem and lower anxiety viewed rejection as less negative
than did individuals with lower self-esteem or higher anxiety. There were no significant relationships
between these personality variables and either the experience of the positive outcome or the risk-averse alter-
native, however.
The bottom portion of Table 3 provides the correlations between the personality variables and the decision
components for deciding for the other person. None of these correlations were significant (all ps>0.05).
Relationships between the decision components and choice
We next examined the relationships between the decision components and the amount of risk-taking beha-
vior, and whether these relationships differed for deciding for the self vs. for another person. Table 4 provides
the correlations between the four components and the level of risk-taking behavior. When deciding for one-
self, risk taking was related to the perceived probability of a successful outcome, feelings about the negative
outcome, and feelings about the risk-averse alternative. In particular, the results indicate that, to the extent
that one finds success more likely, reacts less negatively to rejection, and views risk aversion as more nega-
tive, one will tend to make riskier personal decisions.
As in the self-decision condition, risk taking for others was significantly correlated with the perceived
probability of success and feelings about the risk-averse outcome. However, in contrast to the results in
the self-decision condition, risk taking for others was not related to feeling negatively about bad outcomes,
but was related to feeling positively about good outcomes. This greater focus on the potential negative out-
come when deciding for the self supports the idea that in the present scenarios rejection concerns (or what-
ever negative outcomes constituted threats to the self) play a greater role in decision making for the self than
for others.
Table 3. Experiment 2: Correlations of self-esteem and anxiety with the decision components
Decision components
Probability Neg. outcome Pos. outcome Risk-averse
Risk taking for self
Self-esteem 0.30** 0.29** 0.03 0.11
Anxiety 0.31** 0.38** 0.00 0.07
Risk taking for other
Self-esteem 0.09 0.15 0.08 0.06
Anxiety 0.11 0.08 0.04 0.02
Note: Probability ¼probability of successful outcome. Neg. outcome ¼feelings with regard to the negative outcome.
Pos. outcome ¼feelings with regard to the positive outcome. Risk-averse ¼feelings with regard to the risk-averse
alternative.
**p<0.01.
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 137
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

Mediational analysis regarding decision making for the self
The findings that self-esteem and anxiety are correlated with perceived probability of a successful outcome
and feelings about negative outcomes, and that these variables are in turn correlated with the level of risk-
taking behavior, suggest that these decision components might mediate the effects of self-esteem and anxiety
on risk taking for personal decisions. According to Baron and Kenny (1986), mediation occurs when: (a) the
predictor variable affects the mediating variable; (b) the predictor variable affects the criterion variable; (c)
the mediator predicts the criterion variable after controlling for the effect of the predictor variable; and (d)
the effect of the predictor on the criterion variable is significantly reduced when the mediator is included in
the model.
We proceeded to test each of our four possible mediational models individually. That is, we tested whether
perceived probability of success mediated the relationship between self-esteem (anxiety) and risk taking, and
whether feelings about negative outcomes mediated the relationship between self-esteem (anxiety) and risk
taking. The results with self-esteem and anxiety were very similar, so we present the results only with self-
esteem below.
6
Our correlational results show that the first two requirements have been satisfied, in that self-
esteem is correlated with perceived probability of success and feelings toward the negative outcome, as well
as with risk-taking behavior for the self. To test the third requirement, we regressed level of risk-taking beha-
vior on each of the mediation variables individually. Both potential mediators did in fact remain significant,
partial r ¼0.33, p<0.001, and partial r ¼0.29, p¼0.001, respectively. Finally, the Sobel test (see Preacher
& Leonardelli, 2003) showed that the indirect path from self-esteem to risk-taking behavior via perceived
probability of success was significant, z¼2.66, p¼0.01, as was the indirect path via feelings about negative
outcomes, z¼2.39, p¼0.02. Again, qualitatively similar results were found using anxiety rather than self-
esteem as the predictor variable.
These findings suggest that perceived probability of success and feelings about negative outcomes do
mediate the relationship between self-esteem/anxiety and risk-taking behavior. To determine whether the
mediation was complete or only partial, we next examined whether the relationship between self-esteem/
anxiety and risk-taking behavior remained after controlling for perceived probability of success and feelings
about negative outcomes (i.e., with both mediators in the regression model simultaneously). The relationship
with self-esteem remained marginally significant, partial r ¼0.15, p¼0.09, and the relationship with anxi-
ety remained significant, partial r ¼0.22, p¼0.01. All told, it therefore appears that the mediation is not
complete, and anxiety and perhaps self-esteem do still predict some unique variance even after controlling
for the mediating variables.
Table 4. Experiment 2: Correlations of the decision components
with risk-taking behavior
Risk taking
Self Other
Probability 0.39** 0.39**
Neg. outcome 0.35** 0.03
Pos. outcome 0.04 0.32**
Risk-averse 0.25** 0.36**
Note: Probability ¼probability of successful outcome. Neg. outco-
me ¼feelings with regard to negative outcome. Pos. outcome ¼feelings
feelings with regard to positive outcome. Risk-averse ¼feelings with
regard to risk-averse decision.
**p<0.01.
6
Specifically, all of the analogous tests using anxiety rather than self-esteem were significant at p¼0.01 or less.
138 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

Discussion
Replicating the results of Experiment 1, those with higher self-esteem/lower anxiety produce smaller self –
other differences in their level of risky behavior than do those with lower self-esteem/higher anxiety. Also in
keeping with Experiment 1, self-esteem and anxiety levels were related to risk taking for the self but not for
others. In the process of deciding for oneself, self-esteem and anxiety play a role in judgments of probability
of success and feelings toward negative outcomes. Furthermore, the decision components of perceived prob-
ability of success and feelings regarding negative outcomes partially account for the effects of self-esteem
and anxiety on risk taking for the self. These results are consistent with the argument by Butler and Mathews
(1983) that the costs and probability of a potentially negative outcome constitute a threat to the self.
One puzzling result of Experiment 2 was the lack of gender differences. Previous research has consistently
found that males are significantly more risk seeking than are females in a range of situations (cf. Powell &
Johnson, 1995), and both Beisswanger et al. (2003) and our Experiment 1 found a large gender difference.
We have no ready explanation for the lack of gender difference in this study, and it is plausible that this lack
of effect is just due to chance.
GENERAL DISCUSSION
The primary goal of our research was to determine what role, if any, the personality factors of self-esteem
and anxiety play in deciding for oneself and for others and consequently in self – other differences. In both
our experiments, we found that self-esteem and anxiety levels were highly related to decision making for the
self, but unrelated to decision making for another person. Therefore, both these factors are in part responsible
for self – other differences. These results are consistent with the work of Kray and Gonzalez (1999; Kray,
2000), who found that personal decisions were influenced by more factors than were decisions made for
other people. In Experiment 2, we also observed that the effects of self-esteem and anxiety on personal deci-
sion making are partially mediated by expectations of success and feelings toward the negative outcome. We
shall proceed by discussing the theoretical implications of the above findings for how decisions are made for
the self and for another person. Then, we will briefly speculate about the implications of our results for the-
ories on self-esteem and anxiety. Finally, we shall end with a brief discussion of the extent to which we would
expect our results to generalize to other situations.
Processes involved in deciding for the self and others
Given that self-esteem and anxiety are related to decision making for the self and others in different ways,
and that the different decision components played separate roles in the two types of decision making, these
two types of decisions appear to be made, at least in part, via different processes. In terms of personal deci-
sion making, our results suggest that self-esteem and anxiety levels are related to both the perceived prob-
abilities of favorable outcomes occurring, as well as how bad people would feel about negative outcomes.
These findings thus indicate that empirical results found more generally with self-esteem (see Brockner et al.,
1993) and anxiety (Butler & Mathews, 1983; Eisenberg et al., 1996) hold with our relationship scenarios as
well. Moreover, we then demonstrated that these two decision components — perceived probabilities of
favorable outcomes and feelings toward negative outcomes — in turn affected the level of risk-taking beha-
vior. Thus, the relationship between self-esteem/anxiety and risk-taking behavior is partially mediated by
these factors.
However, the fact that anxiety and self-esteem continue to be associated with risk-taking behavior, even
when controlling for the different decision components, suggests that our expected utility framework pre-
sented in the introduction to Experiment 2, although useful, does not fully capture how these personality
variables influence decision making. One limitation to the EU approach is that it assumes that any influence
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 139
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

of either self-esteem or anxiety functions through a rational, cognitive analysis of the situation. As recently
argued by Loewenstein et al. (2001), people’s decisions are only based on this type of cognitive analysis to a
certain degree and are additionally influenced by a more affective reaction to the situation. For example,
people with higher levels of anxiety may react to avoid the risk-seeking option not out of any particular cog-
nitive concern, but instead as a result of a more visceral reaction created by the anxiety-provoking situation.
The fact that the decision components did largely mediate the effects of self-esteem and anxiety on risk-
taking behavior suggests that the cognitive analysis does partially account for personal decision making,
but other explanations such as Loewenstein et al.’s risk-as-feelings theory need to be invoked as well.
Additional support for Loewenstein et al.’s (2001) ideas can be seen in the decision making for others
results. To the extent that anxiety is experienced more viscerally when deciding for the self versus for some-
one else, as argued by Loewenstein et al., anxiety should not have much of an effect on decision making for
others. Our finding that anxiety and self-esteem levels are unrelated to deciding for a friend is thus consistent
with Loewenstein et al.’s theorizing. Given, then, that decision making for others was not the result of an
affective reaction to the situation, how were these decisions made? The present research was not primarily
designed to answer that question, but some brief speculation is worthwhile.
Allgaier and Stone (2002) suggested that decision making for others is based largely on the extent to
which risk is valued by one’s society in the domain under investigation. Although other factors clearly do
matter as well, such as one’s assumptions regarding whether risk-taking behavior will produce the desired
results (i.e., the perceived probabilities of success in our study), the basis of Allgaier and Stone’s theory is
that the value placed on risk has a greater influence on decision making for others than on personal deci-
sion making. One implication of that argument (see Allgaier, 2002) is that, since decision making for
others is largely dominated by how risk is valued by society, other factors — such as individual differ-
ences—will play less of a role when deciding for others than for the self. Similarly, Kray (2000) provided
evidence that advisors made choices for others based on the attributes of the decision that most people in
general would find important, while individuals made decisions by weighting attributes differently based
on personal preferences. Since neither self-esteem nor anxiety is related to deciding for another person,
our findings support Allgaier and Stone’s claim that everyone, regardless of their personality traits, will
have the same tendency to decide for others in a risk-taking fashion since that option is generally more
desirable in our society.
Self-esteem, anxiety, and threat of rejection
In both experiments, self-esteem and anxiety were highly related, with more than 50% of the variance in
one measure explained by the other measure. Although the focus of this study was not on explaining the
relationship between the constructs, a brief discussion of how our findings fit with present theorizing about
self-esteem and anxiety is worthwhile.
Due to the strong relationship between self-esteem and anxiety, it seems plausible that both these person-
ality variables reflect some underlying factor that influences risk-taking behavior. Indeed, there is some
research that suggests that both self-esteem and anxiety are lower-order factors of the Big-5 personality
traits, and that neuroticism in particular may be able to explain much of the variance of self-esteem and
anxiety (Judge, Erez, Bono, & Thoresen, 2002; Watson, Suls, & Haig, 2002). Specifically, Watson et al.
found that anxiety was a facet of neuroticism and that neuroticism was highly negatively related to self-
esteem, and Judge et al. (2002) reported that a single factor could account for self-esteem, neuroticism,
and locus of control. However, although the general factor of neuroticism may account for the shared
variance between self-esteem and anxiety, we still found that each of these variables predicted some unique
variance in risk-taking behavior. Likewise, Judge et al. (2002) advise that the individual study of these
variables should not be abandoned due to their unique roles in predicting a variety of factors, such as
subjective well-being.
140 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

Instead, we tentatively suggest that self-esteem and anxiety should be considered as distinct but related
personality factors that interact to influence risk-taking behavior. This idea is in keeping with sociometer
theory, which suggests that negative affective reactions, such as anxiety, act in a way that alerts the individual
to the potential for social rejection, disapproval, or exclusion so that the person can act accordingly to avoid
the negative outcome and thus maintain self-esteem (Leary & Downs, 1995). Of particular relevance to these
ideas, we found strong relationships between self-esteem and especially anxiety with feelings toward the
potential negative outcome, but not toward the potential positive outcome. Thus, it does appear that parti-
cipants with higher levels of anxiety were particularly focused on the potential negative outcomes that
involved the possibility of social rejection.
In addition, those with lower self-esteem not only tend to have higher anxiety, as we found, but also react
to threats of rejection by lowering their opinion of themselves and giving up faster than usual on tasks
(Sommer & Baumeister, 2002). In the context of romantic relationships, those with lower self-esteem
respond to threats by distancing themselves from their romantic partner by finding fault and reporting less
closeness (Murray, Rose, Bellavia, Holmes, & Kusche, 2002). As our work demonstrates, individuals with
lower self-esteem take fewer risks in forming relationships, and are most likely anxious because social re-
jection is a potential result of risky action. Thus, it seems plausible that the anxiety experienced in these
situations serves as a signal to alert the individual to potential negative social outcomes. Further, if in-
dividuals with lower self-esteem do react to anxiety-producing social situations by avoidance (i.e., risk
aversion), shying away from approaching others may serve to reinforce a lack of social connectedness,
feelings of low self-worth, and anxiety about approaching these situations in the future. Thus, a pattern is
created, whereby low self-esteem leads to a lack of willingness to take risks, which in turn decreases the
potential for positive events that might otherwise serve to raise the level of self-esteem.
Extensions and boundary conditions
As argued previously, we believe that decision making for the self and for other people operate at least in
part via different processes. Of interest, then, is why this research and other research on relationships (e.g.,
Beisswanger et al., 2003) found large self –other differences in risk-taking behavior, but research on mone-
tary decision making has generally found small or no differences in risk-taking behavior (e.g., Cvetkovich,
1972; Stone et al., 2002; Teger & Kogan, 1975). Are self – other differences the rule or the exception?
Although ultimately this is an empirical issue that requires investigation, some preliminary speculation is
warranted.
One important aspect of our theoretical analysis has been the role played by ‘‘threat’’ to the self (Butler &
Mathews, 1983; Epstein, 1973) in those with lower self-esteem or higher anxiety (see also Josephs et al.,
1992, and Larrick, 1993, for a related but slightly different approach to this issue). Essentially, those with
lower self-esteem or higher anxiety are particularly vulnerable to various threats to the self and thus will
often behave in ways that minimize these threats. One method for minimizing threats to the self is to behave
in a risk-averse manner, passing up the possibility of a more positive outcome to guarantee that a negative
outcome does not occur. Therefore, it should not be surprising that people with lower self-esteem or higher
anxiety levels are more apt to behave in a risk-averse manner than are those with higher self-esteem or lower
anxiety levels. Moreover, these types of self-esteem effects have been found both in the present research on
relationships, as well as in monetary decision making (Josephs et al., 1992).
Thus, it appears that people act to reduce threats to the self in a wide range of situations (see also Larrick,
1993). Moreover, Stone et al. (2002) found that people act to reduce threats to the self when deciding for
others as well as for themselves. In particular, using scenarios similar to those used by Josephs et al. (1992),
they found that a concern with potential regret affected decision making for others to the same extent that it
affected personal decision making. However, in the present research, we found no effects of self-esteem or
anxiety on decision making for a friend. It is plausible that part of the reason for this lack of effect was that
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 141
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

the self was not threatened when making decisions for another person. However, we see no reason to believe
that people would not feel threatened in the present situation, yet would when making hypothetical gambling
decisions for another person.
Instead, we suggest that the threat to the self is different when making decisions for the self and for another
person. In particular, when deciding for another person, the concern is that one makes the right decision for
the other person. This threat can be alleviated, then, by making the decision that is consistent with what
society values. This idea is consistent with research on advice-giving, where advice is seen in part as sym-
bolic social support that is given and evaluated based on the broader cultural contexts (Goldsmith & Fitch,
1997). In simulated business decisions for others, for instance, Jonas and Frey (2003) found that participants
were motivated by a desire for accuracy, leading them to present relatively balanced information to others. In
our relationship scenarios, however, the societal concerns would be different. In particular, acting in accord
with what society values would entail choosing the risk-taking option. If this hypothesis is correct, then par-
ticipants should generally make risk-seeking decisions for others, regardless of their level of self-esteem or
anxiety, which is consistent with our results.
The preceding argument suggests that in decision making for others, threats to the self can be alleviated by
making decisions in keeping with what society values, but only when there is a clearly valued behavior. As
the threat resolution strategy would likely be different in decision making for the self, self – other differences
are likely to emerge when there is a clearly valued behavior. In situations where there is no clearly valued
behavior, however, there is less reason for decision making for others to deviate from decision making for the
self. Although this hypothesis needs to be verified in future work, it is consistent with data suggesting that
risk is valued in relationship decision making but not in monetary decision making (Allgaier & Stone, 2002).
Thus, we tentatively propose that self – other differences will emerge in situations in which the threat to the
self can be resolved by different behaviors in the two situations.
ACKNOWLEDGEMENTS
We would like to thank Mark R. Leary, Mike Ricci, and Liz Allgaier for helpful conversations regarding the
manuscript, and several anonymous reviewers for useful comments. We are also grateful to Emily Richelson
and Travis Vesel for their help with data collection.
REFERENCES
Abelsen, R. P., & Levi, A. (1995). Decision making and decision theory. In G. Lindzey, & E. Aronson (Eds.), The
handbook of social psychology (3rd ed., pp. 231 – 309). New York: Random House.
Aiken, L. S., & West, S. G. (1991). Multiple regression: testing and interpreting interactions. Thousand Oaks, CA: Sage
Publications, Inc.
Allgaier, L. (2002). The effects of cultural values on decision making for others versus for oneself in relationships.
Unpublished master’s thesis, Wake Forest University, Winston-Salem, NC.
Allgaier, L., & Stone, E. R. (2002, November). The influence of cultural values on decision making for others versus for
oneself. Poster presented at the annual meeting of the Society for Judgment and Decision Making, Kansas City, MO.
Barnes, L. B., Harp, D., & Jung, W. S. (2002). Reliability generalization of scores on the Spielberger state trait anxiety
inventory. Educational and Psychological Measurement, 62, 603 – 618.
Baron, R. M., & Kenny, D. A. (1986). The moderator –mediator variable distinction in social psychological research:
conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173 – 1182.
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: desire for interpersonal attachments as a fundamental
human motivation. Psychological Bulletin, 117, 497 – 529.
Baumeister, R. F., Tice, D. M., & Hutton, D. G. (1989). Self-presentational motivations and personality differences in
self-esteem. Journal of Personality, 57, 547 – 579.
142 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

Beisswanger, A. H., Stone, E. R., Hupp, J. M., & Allgaier, L. (2003). Risk-taking in relationships: differences in deciding
for oneself versus for a friend. Basic and Applied Social Psychology, 25, 121 – 135.
Borresen, C. R. (1987). Decision making as a function of self and others. Perceptual and Motor Skills, 64, 1301 – 1302.
Brockner, J., Wiesenfeld, B. M., & Raskas, D. F. (1993). Self-esteem and expectancy-value discrepancy: the effects of
believing that you can (or can’t) get what you want. In R. F. Baumeister (Ed.), Self-esteem: The puzzle of low
self-regard (pp. 219 – 240). New York: Plenum Press.
Brown, J. D., & Dutton, K. A. (1995). The thrill of victory, the complexity of defeat: self-esteem and people’s emotional
reactions to success and failure. Journal of Personality and Social Psychology, 68, 712 – 722.
Butler, G., & Mathews, A. (1983). Cognitive processes in anxiety. Advances in Behaviour Research and Therapy, 5,
51 – 62.
Butler, G., & Mathews, A. (1987). Anticipatory anxiety and risk perception. Cognitive Therapy and Research, 11,
551 – 565.
Coopersmith, S. (1967). The antecedents of self-esteem. San Francisco, CA: Freeman.
Cvetkovich, G. (1972). Effects of sex on decision policies used for self and decision policies used for other persons.
Psychonomic Science, 26, 319 – 320.
Downey, G., & Feldman, S. I. (1996). Implications of rejection sensitivity for intimate relationships. Journal of
Personality and Social Psychology, 70, 1327 – 1343.
Eisenberg, A. E., Baron, J., & Seligman, M. E. P. (1996). Individual differences in risk-aversion and anxiety. Retrieved 23
October 2001 from http://www.sas.upenn.edu/baron/amyold.html.
Epstein, S. (1973). Anxiety, arousal, and the self-concept. In I. G. Sarason, & C. D. Spielberger (Eds.), Stress and anxiety
(Vol. 3, pp. 185 – 224). New York: Hemisphere Publishing Corporation.
Goldsmith, D., & Fitch, K. (1997). The normative context of advice as social support. Human Communication Research,
23, 454 – 476.
Greenberg, J., Pyszczynski, T., & Solomon, S. (1995). Toward a dual-motive depth psychology of self and social
behavior. In M. H. Kernis (Ed.), Efficacy, agency, and self-esteem (pp. 73 – 99). New York: Plenum Press.
Heatherton, T., & Ambady, N. (1993). Self-esteem, self-prediction, and living up to commitments. In R. Baumeister
(Ed.), Self-esteem: The puzzle of low self-regard (pp. 131 – 145). New York: Plenum.
Hoch, S. J. (1987). Perceived consensus and predictive accuracy: the pros and cons of projection. Journal of Personality
and Social Psychology, 49, 22 – 32.
Horney, K. (1950). Neuroses and human growth. New York: Norton.
Hsee, C. K., & Weber, E. U. (1997). A fundamental prediction error: self – other discrepancies in risk preference. Journal
of Experimental Psychology: General, 126, 45 – 53.
Jonas, E., & Frey, D. (2003). Information search and presentation in advisor – client interactions. Organizational
Behavior and Human Decision Processes, 91, 154 – 168.
Josephs, R. A., Larrick, R. P., Steele, C. M., & Nisbett, R. E. (1992). Protecting the self from the negative consequences
of risky decisions. Journal of Personality and Social Psychology, 62, 26 – 37.
Judge, T. A., Erez, A., Bono, J. E., & Thoresen, C. J. (2002). Are measures of self-esteem, neuroticism, locus of control,
and generalized self-efficacy indicators of a common core construct? Journal of Personality and Social Psychology,
83, 693 – 710.
Kahneman, D., Slovic, P., & Tversky, A (Eds.). (1982). Judgment under uncertainty: Heuristics and biases. New York:
Cambridge University Press.
Kray, L. J. (2000). Contingent weighting in self – other decision making. Organizational Behavior and Human Decision
Processes, 83, 82 – 106.
Kray, L., & Gonzalez, R. (1999). Differential weighting in choice versus advice: I’ll do this, you do that. Journal of
Behavioral Decision Making, 12, 207 – 217.
Larrick, R. P. (1993). Motivational factors in decision theories: the role of self-protection. Psychological Bulletin, 113,
440 – 450.
Leary, M. R., & Downs, D. L. (1995). Interpersonal functions of the self-esteem motive. In M. H. Kernis (Ed.), Efficacy,
agency, and self-esteem (pp. 123 – 144). New York: Plenum Press.
Leary, M. R., & MacDonald, G. (2003). Individual differences in self-esteem: a review and theoretical integration. In
M. R. Leary, & J. P. Tangney (Eds.), Handbook of self and identity (pp. 401 – 420). New York: Guilford.
Loewenstein, G. F., Weber, E. U., Hsee, C. K., & Welch, N. (2001). Risk as feelings. Psychological Bulletin, 127,
267 – 286.
Marsh, H. W. (1986). Global self-esteem: its relation to specific facets of self-concept and their importance. Journal of
Personality and Social Psychology, 51, 1224 – 1236.
Mruk, C. (1999). Self-esteem. Boulder, CO: Westview Press Co.
L. D. Wray and E. R. Stone Self – Other Decision Making in Relationships 143
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)

Murray, S. L., Rose, P., Bellavia, G. M., Holmes, J. G., & Kusche, A. G. (2002). When rejection stings: how self-esteem
constrains relationship-enhancement processes. Journal of Personality and Social Psychology, 83, 556 – 573.
Plous, S. (1993). The psychology of judgment and decision making. New York: McGraw-Hill.
Powell, P. L., & Johnson, J. E. V. (1995). Gender and DSS design: the research implications. Decision Support Systems,
14, 27 – 58.
Preacher, K. J., & Leonardelli, G. J. (2003). Calculation for the Sobel test: An interactive calculation tool for mediation
tests. Retrieved 1 July 2004 from University of North Carolina at Chapel Hill website: http://www.unc.edu/
preacher/sobel/sobel.htm.
Raghunathan, R., & Pham, M. T. (1999). All negative moods are not equal: motivational influences of anxiety and
sadness on decision making. Organizational Behavior and Human Decision Processes, 79, 56 – 77.
Raymark, P. H. (2000). Accepting and rejecting medical treatment: a comparison of decisions made for self versus those
made for a significant other. Journal of Applied Social Psychology, 30, 2409 – 2436.
Rogers, C. R. (1951). Client-centered therapy. New York: Houghton-Mifflin.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.
Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.
Savage, L. J. (1954). The foundations of statistics. New York: Wiley.
Solomon, S., Greenberg, J., & Pyszczynski, T. (1991). A terror management theory of social behavior: the psychological
functions of self-esteem and cultural worldviews. In M. P. Zanna (Ed.), Advances in experimental social psychology
(pp. 91 – 159). San Diego, CA: Academic Press.
Sommer, K. L., & Baumeister, R. F. (2002). Self-evaluation, persistence, and performance following implicit rejection:
the role of trait self-esteem. Personality and Social Psychology Bulletin, 28, 926 – 938.
Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the state-trait anxiety inventory. Palo Alto, CA:
Consulting Psychologists Press.
Steele, C. M. (1988). The psychology of self-affirmation: sustaining the integrity of the self. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 21, pp. 261 – 302). New York: Academic Press.
Stone, E. R., Yates, A., & Caruthers, A. S. (2002). Risk taking in decision making for others versus the self. Journal of
Applied Social Psychology, 32, 1797 – 1824.
Swann, W. B., Stein-Seroussi, A., & Giesler, R. B. (1992). Why people self-verify. Journal of Personality and Social
Psychology, 62, 392 – 401.
Teger, A. I., & Kogan, N. (1975). Decision-making for others under reciprocal and non-reciprocal conditions. British
Journal of Clinical Psychology, 14, 215 – 222.
von Neumann, J., & Morgenstern, O. (1944). Theory of games and economic behavior. Princeton, NJ: Princeton
University Press.
Vorauer, J. D., & Ratner, R. K. (1996). Who’s going to make the first move? Pluralistic ignorance as an impediment to
relationship formation. Journal of Social and Personal Relationships, 13, 483 – 506.
Wallach, M. A., & Wing, C. W. (1968). Is risk a value? Journal of Personality and Social Psychology, 9, 101 – 106.
Watson, D., Suls, J., & Haig, J. (2002). Global self-esteem in relation to structural models of personality and affectivity.
Journal of Personality and Social Psychology, 83, 185 – 197.
Yates, J. F. (1990). Judgment and decision making. Englewood Cliffs, NJ: Prentice-Hall.
Authors’ biographies:
Laura D. Wray began this paper as an honor’s project at Wake Forest University, and later received her MS in experi-
mental psychology from Bucknell University. She is currently working on a PhD in human development and family
studies at Penn State University.
Eric R. Stone is an associate professor of psychology at Wake Forest University. His research interests include decision
making for others, risk communication, the development of expertise, and the effects of overconfidence.
Authors’ addresses:
Laura D. Wray, Human Development and Family Studies, The Pennsylvania State University, 110 Henderson Building
South, University Park, PA 16802, USA.
Eric R. Stone, Department of Psychology, Wake Forest University, PO Box 7778, Reynolda Station, Winston-Salem, NC
27109, USA.
144 Journal of Behavioral Decision Making
Copyright #2005 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 18, 125–144 (2005)