1. finance
2. personal finance
3. lending
4. personal loans

Journal of Personality and Social Psychology
1998, Vol. 75, No. 5, 1282-1299
Copyright 1998 by the American Psychological Association, Inc.
0022-3514/98/$3.00
Avoidance Personal Goals and the Personality- Illness Relationship
A n d r e w J. E l l i o t
University of Rochester
Kennon M. Sheldon
University of Missouri--Columbia
Three studies demonstrated that avoidance personal goals are positively related to physical symptom
reports. These results were obtained (a) using both longitudinal and retrospective methodologies and
(b) controlling for neuroticism and other alternative predictor variables. In 2 of the studies, a process
model was validated in which perceived competence and perceived controlledness were shown to
mediate the observed relationship between avoidance goals and symptomatology. Specifically, avoidance goals predicted perceived competence and perceived controlledness, and these variables in turn
predicted longitudinal and retrospective symptom reports. Ancillary results help clarify the unique
roles of neuroticism and avoidance goals as predictors of physical symptomatology.
Since the Roman physician Galen proposed an association
between humoral imbalance and disease, thinkers from a broad
array of disciplines have contemplated the link between personality and physical illness. Within the domain of (modern) personality psychology, interest in this issue dates back to Freud's
work on the psychodynamics underlying hysterical paralysis
(and related conditions). Although interest has waxed and
waned over the years, researchers and theorists are currently
allocating a great deal of attention to the nature of the personality-illness relationship (Suls, David, & Henry, 1996; Wiebe &
Smith, 1997).
Contemporary research on the personality-illness relationship takes many forms (for reviews, see Michaels, Michaels, &
Peterson, 1997; Wiebe & Smith, 1997), but a majority of the
work highlights the explanatory role of dispositions or coping
strategies. Dispositional approaches focus on broad, nomothetic
propensities and articulate how these constructs affect illness,
either directly, in interaction with stress, or through their influence on situation selection or health behavior (Bolger & Schilling, 1991; Maddi, 1990; Pennebaker & Watson, 1991). Numerous dispositional constructs have been investigated, including
neuroticism, Type A personality, hardiness, optimism, hope, inhibited power motivation, and pessimistic explanatory style (for
reviews, see Davidson & Pennebaker, 1996; Friedman& BoothKewley, 1987). Coping approaches focus primarily on situationspecific responses to stressful events and how these personenvironment transactions affect illness (Contrada, Leventhal, &
Andrew J. Elliot, Department of Psychology, University of Rochester;
Kennon M. Sheldon, Department of Psychology, University of Missouri-Columbia.
This research was supported in part by National Institute of Mental
Health Grant MH57054-01. The research could not have been conducted
without the assistance of the following individuals: Geeta Gadgil, Rubyit
Kibria, Holly McGregor, Tim Servoss, Derek Stewart, and Kristopher
Stevens. We thank the following individuals for their helpful comments
on earlier versions of this article: Shelly Gable, Laird Rawsthorne, Richard Ryan, Todd Thrash, and Geoff Williams.
Correspondence concerning this article should be addressed to Andrew J. Elliot, Department of Psychology--Human Motivation Program,
Meliora Hall, University of Rochester, Rochester, New York 14627. Electronic mail may be sent to andye@scp.rochester.edu.
O'Leary, 1990; Lazarus, 1990; cf. Carver, Scheier, & Weintraub,
1989). Several coping strategies or sets of strategies have been
investigated, including self-blame; wishful thinking; problemand emotion-focused coping; active and avoidant coping; and
task-, emotion-, and avoidance-oriented coping (for reviews,
see Carver & Scheier, 1994; Suls et al., 1996).
Personal Goals and Physical Illness
Although the study of dispositions and coping strategies has
yielded many insights into the personality-illness relationship,
several theorists have acknowledged the need for more personcentered (i.e., idiographic) and future-oriented approaches to
complement existing formulations (Coyne & Gottlieb, 1996;
Michaels et al., 1997; Ouellette Kobasa, 1990). One approach
that has been specifically identified as a promising candidate is
that of personal goals (Karoly, 1991; Wiebe & Smith, 1997).
Personal goals may be defined as the consciously articulated,
personally meaningful objectives that individuals pursue in their
dally fives. These goals have been operationalized in terms of
personal projects (Little, 1983), personal strivings (Emmons,
1986), life tasks (Cantor, Markus, Niedenthal, & Nurius, 1986),
and current concerns (Klinger, 1977; see Cantor & Zirkel, 1990,
for details). Personal goals represent self-investments that provide individuals with a sense of purp.ose and identity, and create
a framework for interpreting everyday experience (Brunstein,
Dangelmayer, & Schultheiss, 1996; Lavalle & Campbell, 1995).
As such, one's personal goals are likely to be linked to one's
quality of functioning and overall well-being, including physical
health and illness (Emmons, 1989; Little, 1989). Despite the
promise of a personal goal analysis of the personality-illness
relationship, relatively little research has been conducted; the
following is a review of the published data.
Emmons and King (1988) used two samples to examine
whether personal striving conflict and ambivalence were related
to physical symptom reports and health center visits. Results
indicated that conflict was positively related to symptoms and
visits; the data were unclear for ambivalence. In a third sample,
Emmons and King (1988) found a positive association between
striving rumination and symptom reports. Cantor, Acker, and
Cook-Flannagan (1992) failed to replicate the conflict findings
of Emmons and King (1988), as their data revealed a null rela1282
AVOIDANCE PERSONAL GOALS
tionship between life task conflict and symptom reports. Zirkel
and Cantor (1990) categorized participants as absorbed or unabsorbed in the independence life task and found that absorbed
participants reported more symptoms than their counterparts.
Emmons ( 1991 ) investigated whether personal striving content
(i.e., achievement, affiliation, intimacy, power) was related to
symptom reports, either directly or in interaction with positive or
negative striving-relevant events. Results revealed no significant
relationships (although power strivings were positively associated with an omnibus indicator of distress that included symptom reports). In three samples, Emmons (1992) correlated the
level of abstraction of participants' personal strivings (lowhigh) with their symptom reports and health center visits. Results indicated that striving level was unrelated to symptoms
but negatively related to visits. Lecci, Karoly, Ruehlman, and
Lanyon (1996) examined the relationship between health-relevant personal projects and symptom reports and found that neither the number nor dimensional ratings (e.g., importance) of
such projects were associated with symptoms.
In this research we focused on a basic conceptual distinction
with a rich history in the study of human behavior: that of
approach-avoidance (see Atldnson, 1957; Freud, 1915/1957;
Homey, 1945; C. Hull, 1943; Lewin, 1935; McClelland, 1951;
Miller, 1944; Mowrer, 1960; Murray, 1938). Three studies were
conducted to investigate the relationship between avoidance
(relative to approach) personal goals and physical symptom
reports) With respect to personal goals, the approach-avoidance distinction is grounded in the regulatory focus of the goal.
Approach goals are focused on a positive outcome or state, and
regulation entails trying to move toward or maintain the desired
outcome or state (e.g., communicate better with my roommate,
stay fit and trim). Avoidance goals, on the other hand, are focused on a negative outcome or state, and regulation entails
trying to move or stay away from the undesired outcome or
state (e.g., not argue with my roommate, avoid gaining any
weight). Using negative outcomes or events as the hub of selfregulatory activity (i.e., the directing, monitoring, and evaluating of one's behavior) is likely to evoke and be associated with
a variety of nonoptimal cognitive and affective processes (see
the next section for details) that are detrimental to psychological
well-being (Elliot & Sheldon, 1997; Elliot, Sheldon, & Church,
1997) and physical health. Thus, our work was guided by the
hypothesis that avoidance personal goals are positively related to
physical symptom reports. Data congruent with this hypothesis
would link a fundamental dimension of personality (Elliot,
1997). to an important indicator of physical well-being (Davidson & Permebaker, 1996) and would do so in the context of an
underutilized approach to the personality-illness relationship
that could nicely complement existing approaches.
1283
are addressed (Krantz & Hedges, 1987; Ouellette Kobasa,
1990). It is only in the past few years that mediational data
have begun to appear in the literature in general (Aspinwall &
Taylor, 1992; Cohen et al., 1995; Epstein & Katz, 1992; Kemeny,
Cohen, Zegans, & Conant, 1989; Strauman, Lemieux, & Coe,
1993), and personal goal researchers have yet to address this
question in their work on the personality-illness link. In the
present research, two psychological constructs were investigated
as mediators of the hypothesized relationship between avoidance
personal goals and physical symptomatology: perceived competence and perceived self-determination, which was represented
by two variables--perceived autonomy and perceived controlledness (see Deci & Ryan, 1991). Perceived competence
refers to the degree to which an individual feels effective or
competent in his or her goal pursuits (Brunstein, 1993; Elliot
et al., 1997). Perceived autonomy refers to the degree to which
goal pursuit is experienced as freely chosen, emerging from
one's core values or inherent interests, whereas perceived controUedness refers to the degree to which goal pursuit is experienced as coerced, or pressured by external forces or intrapsychic
demands (Ryan & Connell, 1989; Sheldon & Elliot, 1998).
The focus on negative outcomes or events inherent in avoidance regulation is likely to elicit and sustain threat appraisals,
anxiety, and self-protection processes, as the individual is incessantly reminded of aversive possibilities (Elliot & Church, 1997;
Elliot & Sheldon, 1997). Research from a variety of literatures
(e.g., test anxiety, achievement motivation, intrinsic motivation)
has clearly demonstrated that threat, anxiety, and self-protection
have deleterious consequences for both the effectiveness of selfregulatory efforts (Covington, 1992; Hembree, 1988) and the
phenomenological experience of goal pursuit (Csikszentmihalyi,
1990; Deci & Ryan, 1991). In addition, avoidance regulation
is likely to be associated with perceptual, attentional, and mental
control processes such as a perceptual-cognitive sensitivity to
negative information (Derryberry & Reed, 1994), the heightened accessibility of negative information (Higgins, 1996), and
a biased search or"hypothesis test" for the presence of negative
information (Wegner, 1994; see also Kunda, 1990) that would
also undoubtedly have negative effects on goal attainment and
phenomenal experience. As such, avoidance (relative to approach) goals are presumed to be less effective forms of regulation and to produce lower perceptions of competence (Elliot et
al., 1997). Furthermore, the pursuit of avoidance goals should
have little inherent or intrinsic appeal; it seems more likely that
individuals will feel compelled to force or pressure themselves
to engage in these phenomenologically aversive forms of regulation (Elliot & Harackiewicz, 1996). Thus, in our research,
avoidance personal goals were hypothesized to be negatively
related to perceived competence and autonomy and positively
related to perceived .controlledness. These hypotheses represent
Mediational Mechanisms
Research on the personality-illness relationship has been focused predominantly on documenting direct links between personality variables and indicators of illness, with little consideration of the psychological and physiological mechanisms responsible for such relationships (Cohen & WiUiamson, 1991;
Wiebe & Smith, 1997). Noting this oversight, several theorists
have argued that the personality-illness literature will remain
at a nascent stage of development until these " h o w " questions
1Although research on the relationship between avoidance personal
goals and symptom reports has yet to be published (i.e., disseminated
in a peer-reviewed journal), Emmons and Kaiser (1996) did present
relevant data in a recent book chapter. Specifically,using several existing
data sets, Emmons and Kaiser coded personal strivings for approach
and avoidance and computed the zero-order correlation between the
number of avoidance strivings and symptom reports. The results were
ambiguous: A positive correlation was obtained in half of the samples
considered, and a null relationship was obtained in the others.
1284
ELLIOT AND SHELDON
the first path in the proposed model linking avoidance personal
goals to physical symptom reports.
Competence-relevant constructs such asperceived competence are at the core of most conceptualizations of motivation
and health (e.g., self-efficacy theory, the health belief model; see
Bandura, 1977; Rosenstock, 1974). Self-determination theorists
have recently espoused the need to also attend to constructs
such as perceived autonomy and controlledness in the health
domain (see Williams, Deci, & Ryan, 1998). Specifically, selfdetermination theorists contend that both competence and selfdetermination (i.e., the presence of autonomy and absence of
controlledness) represent basic psychological needs, nutriments
essential for optimal functioning (Deci & Ryan, 1991; Sheldon,
Ryan, & Reis, 1996). Failure to satisfy these fundamental organismic requirements is posited to lead to decrements in physical
and psychological well-being (Ryan, 1995), presumably because this lack of "psychological nourishment" produces stress
and its inimical physiological concomitants (see Ratliff-Crain &
Baum, 1990; Wiebe & Smith, 1997). A substantial amount of
empirical work has linked competence and (to a lesser degree)
autonomy or controlledness to health-relevant behavior (see
Janz & Becker, 1984; Williams et al., 1998); a modicum of
studies have linked these variables directly to indicators of physical health and illness (Bandura, Cioffi, Taylor, & Brouillard,
1988; Grow Kasser & Ryan, in press; Williams, Freedman, &
Deci, 1996; see also Kasser & Ryan, 1996). In our research,
perceived competence and autonomy were predicted to be negatively related, and perceived controlledness positively, related,
to physical symptom reports. These hypotheses represent the
second path in the proposed model linking avoidance personal
goals to symptom reports. Perceived competence, perceived autonomy, and perceived controlledness were additionally hypothesized to account for the proposed relationship between avoidance goals and symptoms reports, thereby establishing the former's status as mediator variables (Judd & Kenny, 1981).
Methodological Considerations
Physical symptom reports are by far the most frequently used
dependent measure in research on the personality-illness relationship (Larsen, 1992; Wiebe & Smith, 1997). These measures
are popular not only because they are face valid, easy to administer, and "psychometrically elegant" (Davidson & Pennebaker,
1996), but also because symptom perceptions and reports influence health-relevant behavior (e.g,, self-medication, medical
care utilization; Larsen & Kasimatis, 1991); have implications
for daily functioning and self-regulation (e.g., activity level,
work absenteeism; Robbins, Spence, & Clark, 1991); are a primary factor in many physician diagnoses (Brown & Moskowitz,
1997); and are, to some degree, accurate indicators of organic
condition (Watson & Pennebaker, 1989). Like all measures of
illness, however, symptom reports have weaknesses and vulnerabilities that pose methodological challenges for researchers (Davidson & Pennebaker, 1996). Unfortunately, these challenges
have often been ignored in the extant research, leaving much
of the literature susceptible to criticism.
One important methodological issue concerns the role of neuroticism as a potential nuisance variable in research linking
personality measures to symptom reports. Neuroticism may be
construed as a broad disposition representing individual differ-
ences in autonomic nervous system lability and in the tendency
to experience negative affect (H. Eysenck & Eysenck, 1985).
Researchers have repeatedly documented a moderate-to-strong
correlation between neuroticism (or, alternatively, trait negative
affectivity) and symptom reports (Costa & McCrae, 1987; Watson & Pennebaker, 1989). As such, to the extent that a personality measure is associated with neuroticism, any observed relationship between that personality measure and symptom reports
could be spurious, simply reflecting the underlying influence of
neuroticism. For example, Smith, Pope, Rhodewalt, and Poulton
(1989) have shown neuroticism to be a confounding variable
in the relationship between optimism and symptom reports, in
that the observed association between optimism and symptoms
is eliminated when neuroticism is controlled (see also Funk &
Houston, 1987; J. Hull, Tedlie, & Lehn, 1995; Rhodewalt &
Zone, 1989). FOr years researchers have been encouraged to
measure and statistically control neuroticism in their studies to
address this issue (Costa & McCrae, 1987; Holroyd & Coyne,
1987), but few have heeded the call; personal goal researchers
have yet to control for neuroticism in their work on the personality-illness relationship (see Emmons & King, 1988, Study 1,
for the closest analog). In all three studies of the present research, neuroticism was measured and statistically controlled in
testing the relationship between avoidance personal goals and
symptom reports, Furthermore, several other personality variables that have also been empirically or theoretically linked to
symptom reports (Type A personality, optimism, hope, behavioral inhibition system [BIS ] sensitivity, extraversion, and Nenroticism x Extraversion; see Eagleston et al., 1986; Fowles,
1988; Scheier & Carver, 1987; Snyder et al., 1991; Zautra,
Finch, Reich, & Guarnaccia, 1991) were also included to more
rigorously test the discriminant validity of the avoidance goals
construct.
A second important methodological issue concerns the degree
to which response tendencies or biases inflate or even create
correlations between personality and physical symptom measures. The vast majority of research in this area has examined
the link between personality and symptom reports either concurrently or prospectively using a one-time retrospective measure of
symptoms (Brown & Moskowitz, 1997; Larsen, 1992).2 Results
from such studies are difficult to interpret because any association observed between personality and symptom measures may
simply be an artifact of demand characteristics, self-presentational concerns, or (in concurrent designs) mood or context
effects (Costa & McCrae, 1987; Greenberg, Wortman, & Stone,
1996). An optimal way of addressing this issue is t o u s e a
longitudinal design in which symptoms are assessed on more
than one occasion and prior symptom reports (with their attendant response tendencies and biases) are controlled in statistical
analyses. Such designs are highly uncommon in the personalityillness literature (for examples, see Kobasa, Maddi, & Kahn,
1982; Scheier & Carver, 1985) and have yet to be used by
personal goal researchers working in this domain. In the present
research, we used both longitudinal and retrospective designs
: Although any assessment involving recall could technically be labeled retrospective, in practice the term is usually reserved for onetime assessments involving recall over an extended period (typically the
length of an entire study; see Elliot & Sheldon, 1997; Larsen, 1992). It
is this latter use of the term that we adopted in our research.
AVOIDANCE PERSONAL GOALS
to test the relationship between avoidance personal goals and
symptom reports.
O v e r v i e w o f the P r e s e n t R e s e a r c h
In summary, the present research is composed of three studies
that tested, with methodological rigor, the hypothesis that avoidance personal goals are positively related to physical symptom
reports and examined the proposition that perceived progress,
perceived autonomy, and perceived controlledness are the psychological mechanisms responsible for the hypothesized relationship. In Study 1 we tested the relationship between avoidance goals and symptom reports over a I-month period using
both longitudinal and retrospective .methodologies. In Studies 2
and 3 we examined this relationship over the course of an entire
semester-long period (4 months); in Study 2 we used a retrospective approach, whereas in Study 3 we used both longitudinal
and retrospective methodologies. All three studies tested the
relationship between avoidance goals and symptom reports controlling for neuroticism, and each study also included at least
two additional "third-variable" candidates. In Studies 2 and 3
we addressed the issue of mediation by assessing participants'
perceived competence, perceived autonomy, and perceived controlledness with regard to their goals several times throughout
the semester-long period. Path-analytic techniques were used to
test the two components of the proposed mediational model:
(a) that avoidance goals would predict perceived competence,
perceived autonomy, and perceived controlledness and (b) that
these process variables would independently predict symptom
reports and account for (i.e., explain) the hypothesized relationship between avoidance goals and symptoms. Two ancillary
issues were also explored: the degree to which avoidance goals
predict specific physical symptoms and the precise roles played
by neuroticism and avoidance goals as joint predictors of physical symptomatology.
Study 1
Me~od
Participants and Overview o f Procedure
A total of 82 (46 male and 36 female) university undergraduates in
a psychology class participated in the study for extra credit. During the
2nd month of the semester, individuals were contacted by phone and
those who chose to participate in the study were asked to generate a list
of at least five goals that they would be pursuing during the next month.
Participants then brought these candidate goals to an individual session
in which they selected their final set of five goals for the study and
reported their physical symptoms during the past 2 weeks (baseline
symptoms). Approximately 2 weeks later, participants reported their
physical symptoms during the past 2 weeks (the first component of an
aggregate symptoms measure) and completed indicator s of neuroticism,
Type A personality, and optimism. After approximately 2 more weeks,
participants attended a final individual session in which they again reported their physical symptoms during the past 2 weeks (the second
component of the aggregate symptoms measure), completed a physical
symptoms measure for the entire study-long period (retrospective symptoms), and were provided with information about the purpose of the
research.
1285
Measures
Definition of personal goals and the avoidance goals measure. Personal goals were defined as the "individual, specific goals" that participants would be pursuing during the next month. No other information
about personal goals was provided. "l~vo trained individuals independently coded each goal as approach or avoidance (interjudge agreement
was 99.76%, with disagreements resolved through discussion). Exampies of approach goals listed by participants are "practice recycling
more often" and "wake up at 8:30 every morning"; examples of avoidance goals are "not judge new friends" and "eat less fried foods." An
avoidance goals index was created for each participant by summing the
number of avoidance goals listed in his or her set of five goals. Recent
research has demonstrated the validity of this type of measure by linking
the adoption of avoidance personal goals to dispositional indicators of
avoidance motivation (e.g., projective and self-reported fear of failure)
as well as assessments of psychological well-being (Elliot & Sheldon,
1997; Elliot et al., 1997; see also Coats, Janoff-Bulman, & Alpert, 1996).
Individual-differences measures. Neuroticism was assessed with the
Neuroticism subscaie of Costa and McCrae's (1992) NEO Five Factor
Inventory (NEO-FFI). Participants indicated their responses to the 12
items (sample item: "I often feel tense and jittery") on a 1 (strongly
disagree) to 5 (strongly agree) scale, and their responses were summed
to form a neuroticism index. Type A personality was assessed with the
21-item Jenkins Activity Survey, Form T (Krantz, Glass, & Snydez;
1974). This measure presents respondents with questions or statements
(e.g., "How would your spouse [or best friend] rate your general level
of activity?" ) and instructs them to select one of several response alternatives, some of which are presumed to be indicative of Type A behavior
(e.g., "Too active, needs to slow down"). The number of times that
participants selected a Type A response was summed to form a Type A
personality index. The Life Orientation Test-Revised (Scheier, Carver, &
Bridges, 1994) was used to assess optimism. Participants indicated their
responses to the six Life Orientation Test items (sample item: "In uncertain times, I usually expect the best.") on a 1 (strongly disagree) to 5
(strongly agree) scale, and their responses were summed to form an
optimism index.
Physical symptoms measures. All assessments of physical symptoms used a measure designed by Emmons's (1992) comprised of the
following symptoms: headaches, coughing or sore throat, shortness of
breath, stiff or sore muscles, chest or heart pain, faintness or dizziness,
acne or pimples, stomach ache or pain, and runny or congested nose.
This measure is essentially a brief version of the Permebaker Inventory
of Limbic Languidness (Emmons & King, 1988), which is a wellvalidated assessment of physical symptoms and sensations (see Pennebaker, 1982), and this nine-item scale itself evidences good validity
(e.g., scores are positively related to health center visits for illness; Ryan,
1996). Several of the nine symptoms have been specifically identified as
precursors to serious diseases such as ulcers, heart disease, asthma, and
arthritis (Friedman & Booth-Kewley, 1987). For the longitudinally based
assessments, participants indicated how often they had experienced each
of the nine symptoms during the past 2 weeks on a 1 (not at all) to 7
(very frequently) scale. A baseline symptoms measure was created by
summing the nine items on the first assessment that referred to the 2
weeks prior to the study; an aggregate symptoms measure was created
by summing participants' scores on each of the biweekly assessments
and then summing across the two periods. A retrospective symptoms
measure was formed by summing the nine items on the final assessment
that used the entire study-long period as the temporal referent.
Results
Participant Attrition, Descriptive Statistics, and Gender
All but 1 of the initial 82 participants completed the study,
resulting in an attrition rate of 1.2%. Attrition analyses were
1286
ELLIOT AND SHELDON
not conducted given the absence of variance in the study dropout
group. The means, standard deviations, ranges, and reliabilities
of the variables for the final sample of 81 (45 male and 36
female) participants are presented in Table 1. Intercorrelations
among the primary study variables are presented in Table 2. In
all analyses (in this and the remaining studies), we included
gender with the other "main effect" predictors in initial multiple
regression analyses; gender was retained in subsequent or final
analyses only when it was significant.
The Relationship Between Avoidance Goals and
Physical Symptoms
Change in physical symptoms. To investigate the relationship between the pursuit of avoidance goals and longitudinal
change in physical symptoms, we regressed aggregate symptoms on avoidance goals and baseline symptoms. The regression
yielded a significant effect for baseline symptoms, F ( 1, 78) =
57.15, p < .0001,/~ = .63. More important, avoidance goals
also attained significance, F(1, 78) = 5.07, p < .05, /~ =
.19, indicating that participants pursuing a greater number of
avoidance goals evidenced an increase in physical symptoms
during the month. Three additional analyses were then conducted in which the regression was repeated with an alternative
predictor variable (neuroticism, Type A personality, or optimism) also in the equation. Adding the alternative predictors
had a negligible impact on the relationship between avoidance
goals and change in physical symptoms: The beta coefficient
remained the same (and was significant) in all analyses.
Retrospective physical symptoms. To investigate the relationship between the pursuit of avoidance goals and retrospective physical symptoms, we computed a Pearson product-moment correlation between avoidance goals and retrospective
symptoms. Although the relationship was not statistically significant (r = .16, p = .15), a trend was clearly evident in
the same direction as the longitudinal result. The same set of
"alternative predictor" analyses described earlier was conducted for retrospective symptoms. Adding the alternative predictors had a minimal impact on the observed relationship: The
beta coefficient dropped .02, .01, and remained the same with
neuroticism, optimism, and Type A personality in the equation,
respectively (all p s < .19).
Discussion
The results o f this study support the hypothesis that the pursuit of avoidance personal goals is linked to physical symptom-
atology. Avoidance goal adoption at the beginning of the monthlong period was positively related to a longitudinal increase in
symptom reports during the month. This relationship remained
significant when controlling for neuroticism, as well as other
alternative predictor variables. The retrospective results also provided support, albeit somewhat weaker, for the hypothesized
relationship.
In Study 2 we attempted to conceptually replicate and extend
Study 1. The time course of Study 1 was relatively brief (1
month), and therefore, the goals that were generated by participants were short-term or "molecular" forms of regulation. In
Study 2 we examined the ramifications of avoidance regulation
over the course of a semester-long (4-month) period, thereby
affording a test of the generalizability of the Study 1 findings
to a more "molar" set of personal goals (see Little, 1989, for a
discussion of the molecular-molar distinction). Two additional
changes in Study 2 were a sole focus on retrospective symptom
reports and the inclusion of different third-variable candidates
to supplement neuroticism: extraversion (which also enabled
Neuroticism × Extraversion to be tested) and hope. Obtaining
significant results in a semester-long study, with a retrospective
symptoms measure, and controlling for additional alternative
predictors would further attest to the robustness of the relationship between avoidance goals and symptom reports. Most importantly, in Study 2 we addressed the question of mediation.
Participants' perceived competence, perceived autonomy, and
perceived controlledness with regard to their goals was assessed
three times during the semester, and these variables were tested
as mediators of the relationship between avoidance goals and
symptom reports.
Study 2
Method
Participants and Overview of Procedure
A total of 177 (75 male and 102 female) university undergraduates
in a psychology class participated in the study for extra credit. The study
was conducted in multiple sessions over the course of a semester-long
period. In large group sessions at the beginning of the semester, participants identified 10 personal goals that they would be pursuing during
the semester and completed measures of neuroticism, extraversion, and
hope. In a series of individual sessions throughout the semester (every
3-4 weeks ), participants completed a questionnaire assessing their perceptions of competence, autonomy, and controlledness regarding their
goals. In a final individual session at the end of the semester, participants
completed a retrospective physical symptoms measure, identified each
Table 1
Descriptive Statisticsfor Study 1
Variable
M
SD
Possible range
Observed range
Reliability
Avoidance goals
Baseline symptoms
Aggregate symptoms
Retrospective symptoms
Neuroticism
Type A personality
Optimism
0.58
22.37
41.42
21.10
31.88
9.09
20.59
0.84
7.84
14.32
7.94
8.06
3.66
4.66
0-5
9-63
18-126
9-63
12-60
0-21
6-30
0-3
10-55
20-80
9-50
15-55
2-17
9-30
.70
.85
.79
.85
.71
.83
Note. n = 81.
AVOIDANCE PERSONAL GOALS
Table 2
Intercorrelations Among the Primary Variables in Study 1
Variable
1.
2.
3.
4.
5.
Avoidance goals
Baseline symptoms
Aggregate symptoms
Retrospective symptoms
Neuroticism
1
2
3
4
.08
.24*
.16
.06
-.65**
.61"*
.29**
-.85**
.28**
-.36**
5
Note. n = 81.
*p<.05.
**p<.01.
1287
Physical symptoms measure. Emmons's (1992) physical symptoms
measure includes three items from Derogatis and Spencer's (1982) Brief
Symptom Inventory. In this study, Emmons's nine-item measure was
supplemented with the four remaining items from Derogatis and Spencer's scale (i.e., hot or cold spells, numbness or tingling in parts of your
body, nausea or upset stomach, feeling weak in parts of your body; see
Derogatis & Spencer, 1982, for validation information) to form a more
extensive (and perhaps more reliable) 13-item index. Participants indicated how often they had experienced each of the physical symptoms
during the past semester on a 1 (not at all) to 9 (very frequently) scale,
and their responses were summed to form a retrospective symptoms
index.
Resul~
of their goals as approach or avoidance, and were provided with information about the purpose of the research?
Measures
Definition of personal goals and the avoidance goals measure. Personal goals were defined as "things that you typically or characteristically are trying to do in your dally l i f e . . , the purposes or goals that
you seek in your everyday behavior" (see Emmons, 1986). Various
dimensions of personal goals were described (including approachavoidance), and participants were encouraged to consider both approach
and avoidance goals as they generated their representative list.4 Examples
of approach goals listed by participants are "to be more organized"
and "to excel in my workplace"; examples of avoidance goals are
"avoid being lonely" and "avoid excessive partying." An avoidance
goals index was created for each participant by summing the number
of avoidance goals listed in his or her set of 10 goals.
Individual-differences measures. The 12-item versions of the Neuroticism and Extraversion subscales from the Eysenck Personality Questionnaire-Revised (S. Eysenck, Eysenck, & Barrett, 1985) were used
to assess neuroticism (sample item: "Would you call yourself tense or
'high strung'?") and extraversion (sample item: "Do others think of
you as being very lively?"), respectively. Participants indicated their
response to each item on a yes-no scale, and their yes responses were
summed to form the neuroticism and extraversion indexes. Hope was
assessed with the Snyder et al. (1991) eight-item hope scale (sample
item: "Even when others get discouraged, I know I can find a way to
solve the problem" ). Participants indicated their response to each item
on a 1 (definitely false) to 4 (definitely true) scale, and their responses
were summed to form a hope index.
Process measures. For each assessment, participants rated how well
they were doing at the present time on each goal on a I (not at all) to
9 (very) scale. These ratings were averaged across the 10 goals and then
averaged across the three assessments to form a perceived competence
index. For each assessment, participants also rated the degree to which
they felt they were pursuing each goal at the present time for each of
four different reasons (see Ryan & Connell, 1989): external ("Because
somebody else wants you to or because the situation demands it"),
introjected ("Because you would feel ashamed, guilty, or anxious if
you didn't"), identified ("Because you really believe that it's an important goal to have" ), and intrinsic ( "Because of the fun and enjoyment that it provides you" ). Participants indicated their ratings on a 1
(not at all) to 9 (completely) scale. Participants' intrinsic ratings were
multiplied by 2, added to their identified ratings, averaged across the 10
goals, and then averaged across the three assessments to form a perceived
autonomy index. Participants' external ratings were multiplied by 2,
added to their introjected ratings, averaged across the 10 goals, and then
averaged across the three assessments to form a perceived controlledness
index (see Ryan & Connell, 1989, for the rationale for doubling the
intrinsic and external scores; see also Sheldon & Elliot, 1998; Sheldon &
Kasser, 1995).
Participant Attrition and Descriptive Statistics
Eleven of the initial 177 participants did not complete the
study, resulting in an attrition rate of 6.2%. Attrition analyses
revealed no significant differences between the study completion
and study dropout groups. The means, standard deviations,
ranges, and reliabilities o f the variables for the final sample of
71 male and 95 female participants are presented in Table 3.
Intercorrelations among the primary study variables are presented in Table 4.
The Direct Relationship Between Avoidance Goals and
Physical Symptoms
To investigate the relationship between the pursuit of 'avoidance goals and retrospective physical symptoms, we regressed
retrospective symptoms on avoidance goals ( a n d gender). The
regression yielded a significant effect for r e t r o s p e c t i v e symptoms, F ( 1 , 163) = 5.83, p < .05, /3 = .18, indicating that
participants pursuing a greater n u m b e r of avoidance goals reported experiencing more physical symptoms over the semesterlong period. Gender also attained significance, F ( 1 , 163) =
9.09, p < .005,/3 = .23, indicating that w o m e n reported experiencing more symptoms than men. Two additional analyses were
then conducted in which the above analysis was repeated with
an alternative predictor variable ( h o p e ) or set of variables (neuroticism, extraversion, and Neuroticism × Extraversion) also in
3 The data for Study 2 were collected in the context of a larger project,
and a portion of the data examined in this study were also used by
Elliot, Sheldon, and Church (1997) to investigate a conceptually distinct
set of issues.
4 Explicitly instructing participants to consider both approach and
avoidance goals as they generated their representative list had the unintended consequence of prompting them to list individual goals that ineluded both approach and avoidance components. Thus, instead of coding the goals for approach and avoidance ourselves, we had participants
classify each of their goals as approach or avoidance by selecting the
category that they thought best described the goal ("focus on attaining
a positive outcome" vs. "focus on avoiding a negative outcome," respectively). Parenthetically, the variability in the number of avoidance
goals adopted by participants across studies was attributable to the
differential time course of the studies (i.e., 1 month vs. a semesterlong period) and the differential instruction sets employed in the goal
elicitation procedures (i.e., whether the approach-avoidance distinction
was mentioned in introducing the concept of personal goals).
1288
ELLIOT AND SHELDON
Table 3
Descriptive Statistics f o r Study 2
Variable
M
SD
Possible range
Observed range
Reliability
Avoidance goals
Retrospective symptoms
Perceived competence
Perceived autonomy
Perceived controUedness
Neuroticism
Extraversion
Hope
2.86
24.70
5.34
16.32
10.20
5.16
8.50
25.60
1.60
9.34
1.18
3.60
3.47
3.43
3.44
3.36
0-10
13-91
1-9
1-27
1-27
0-12
0-12
8-32
0-8
13-57
2.33-7.97
8.97-25.40
3-19.27
0-12
0-12
15-32
.89
.90
.92
.95
.87
.83
.82
Note. n = 166.
the equation, s Adding the alternative predictors had a relatively
small effect on the relationship between avoidance goals and
retrospective symptoms: With neuroticism, extraversion, and
Nenroticism x Extraversion in the equation, the coefficient decreased .04 standardized units (p = .06), whereas with hope in
the equation, the coefficient decreased .02 standardized units
and the relationship remained (p < .05).
Mediation
The relationship between avoidance goals and the process
variables. Pearson correlations were computed to determine
the relationship between avoidance goals and both perceived
competence and perceived autonomy; a multiple regression analysis was conducted with perceived controlledness (to include
gender). Both correlation coefficients were significant, indicating that participants pursuing a greater number of avoidance
goals reported lower perceptions of competence ( r = -.19, p
< .05) and autonomy ( r = ".27, p < .0005) during the semester. The regression analysis revealed a significant relationship
between avoidance goals and perceived controlledness, F(1,
163) = 21.72, p < .0001,/5 = .34, indicating that participants
pursuing a greater number of avoidance goals reported more
controlledness during the semester. Gender also attained significance, F ( 1, 163) = 6.27, p < .05,/5 = .18, indicating that
women reported more controlledness than men.
The relationship between the process measures and physical
symptoms. To test the final link in the mediational model, we
conducted a multiple regression analysis in which retrospective
symptoms was regressed on perceived competence, perceived
autonomy, and perceived controlledness with avoidance goals
(and gender) also in the equation (see Judd & Kenny, 1981, for
the steps necessary to document mediation). The relationship
between perceived autonomy and retrospective symptoms was
not significant; therefore, perceived autonomy was trimmed
from the final model (Judd & Kenny, 1981), The regression
revealed that both perceived competence, F ( 1, 161 ) = 5.20, p
< .05, /5 = -.16, and perceived controlledness, F ( 1 , 161)
= 26.27, p < .0001, /5 = .38, were significant predictors of
retrospective symptoms, indicating that lower perceptions of
competence and higher perceptions of controlledness were associated with higher reports of symptoms. Gender also attained
significance, F ( 1 , 160) = 5.54, p < .05,/3 = .16, indicating
that women reported more symptoms than men. The direct relationship between avoidance goals and retrospective symptoms
was no longer significant with the mediator variables in the
equation (p > .77), and the beta coefficient for this relationship
dropped from .18 to .02.
Structural equation modeling analyses. Structural equation
modeling (SEM) was used to test the paths in the hypothesized
mediational model simultaneously and to assess the fit of the
model to the observed data. Conceptually peripheral but significant relationships such as that between gender and retrospective
symptoms were also incorporated into the model. The correlation matrix was used as input, and LISP,EL VIII (Jtireskog &
Sbrbom, 1990) generated standardized parameter estimates
based on maximum-likelihood estimation. Each latent variable
was represented by a single observed indicator. Following Hoyle
and Panter's (1995) recommendation, both absolute (goodnessof-fit index [GFI] and adjusted goodness-of-fit index [AGFI])
and incremental (comparative fit index [CFI] and incremental
fit index [ IFI ]) indexes were used to evaluate the fit of the model
to the data. The standardized parameter estimates obtained in
the analysis were essentially identical to (within .01 standardized unit of) those obtained with the regression procedures.
In addition, the analysis revealed that the model provided a
satisfactory fit to the data, X2(7, N = 166) = 13.77, p > .05;
GFI = .97, AGFI = .92, CFI = .93, and IFI = .93. Each
of these values is in the range generally interpreted as being
representative of a good-fitting model (see Tanaka, 1987).6 Repeating this SEM analysis and including an alternative predictor
variable (hope) or set of variables (neuroticism and extraversion) and their significant or marginally significant paths (as
evidenced in regression analyses) yielded highly comparable
results. Each of the hypothesized relationships remained p -<
.05, except the avoidance goals to perceived competence path
controlling for hope (p < .07) and the perceived competence
to retrospective symptoms path (p < . 17) controlling for neurotIn Studies 2 and 3, the neuroticism and extrav~rsion variables were
centered before computing the Neuroticism × Extraversion interaction
product term used in the analyses (see Aiken & West, 1991).
6 Although all the goodness-of-fit values indicated a good-fitting
model, the modification indexes did suggest allowing correlated errors
of measurement between perceived competence and perceived autonomy.
Incorporating this modification had no effect on the parameter estimates
and simply improved the fit of the model; no additional modifications
were suggested in this follow-up analysis. No modifications were suggested in any of the SEM models tested throughout the remainder of
this article.
1289
AVOIDANCE PERSONAL GOALS
Table 4
Intercorrelations Among the Primary Variables in Study 2
Variable
1.
2.
3.
4.
5.
6.
Avoidance goals
Retrospective physical symptoms
Perceived competence
Perceived autonomy
Perceived controlledness
Neuroticism
1
2
3
4
5
.19"
-.19"
-.28**
.36**
.18"
--.~2"*
-.08
.45**
.38**
-.28**
-.17"
-.36**
--.06
-.22**
.34**
6
Note. n = 166.
* p <.05. **p < .01.
icism and extraversion. All goodness-of-fit statistics in both
analyses indicated that the model provided a satisfactory fit to
the data. These results are particularly impressive given the
strong likelihood of multicolinearity and related problems in
highly controlled analyses of this natt}re (see Scheier & Carver,
1987). In summary, the results from both the regression and
SEM analyses clearly established perceived competence and perceived controlledness as mediators of the relationship between
avoidance goals and retrospective symptoms (see Figure 1 for
a pictorial summary of the prirnary results).
symptom assessments were employed in the longitudinal aspect
of the study, one at the beginning of the semester (Time 1 [T1])
and one at the end of the semester (Time 2 [T2]). This change,
in conjunction with the first, enabled an appropriate test of the
mediational hypotheses with the longitudinal and retrospective
data (see Kenny, 1979). Third, BIS sensitivity replaced the hope
variable in Study 2 as an alternative predictor variable.
Study 3
Method
Discussion
The results of this study conceptually replicated and extended
those obtained in Study 1. Avoidance goal adoption at the beginning of the semester was positively related to retrospective reports of physical symptoms experienced during the semester.
This relationship remained significant or marginally significant
when controlling for neuroticism or the other alternative predictor variables. A series of analyses validated perceived competence and perceived controlledness as mediators of the observed
relationship, as avoidance goals predicted perceived competence
and controlledness and these variables, in turn, predicted symptom reports. Interestingly, it was the (perceived) presence of
controlledness and not the absence of autonomy that was the
operative self-determination mediator; avoidance goals predicted perceived autonomy, but perceived autonomy did not predict symptoms.
In Study 3 we sought to simultaneously (conceptually) replicate Studies 1 and 2. The study was conducted over a semesterlong period, both longitudinal and retrospective methodologies
were employed, and perceived competence and perceived controlledness were tested as mediators of the anticipated longitudinal and retrospective relationships. Perceived autonomy was also
examined as a mediator variable to determine whether the null
results obtained in Study 1 would replicate. Three changes from
the previous studies should be noted. First, the time frame for
the longitudinal symptom variables was changed from "the past
2 weeks" in Study 1 to "the past few days." Research has
demonstrated that recall-based symptom reports can be subject
to memorial biases that increase with length of time frame
( B r o w n & Moskowitz, 1997; Larsen, 1992) and that concurrent
symptom reports can be affected by context-specific variables
such as mood (Salovey & Birnbaum, 1989). Using the past few
days as a temporal referent was deemed an optimal balance
between these two forms of potential bias. Second, only two
Participants and Overview o f Procedure
A total of 159 (59 male and 100 female) university undergraduates
in a psychology class participated in the study for exlra credit. The study
was conducted in a series of sessions over the course of a semester-long
period. In group sessions at the beginning of the semester; participants
identified eight personal goals that they would be pursuing during the
semester, completed a T1 physical symptoms measure, designated each
of their goals as approach or avoidance, and completed measures of
neuroticism, extraversion, and BIS sensitivity. In several individual sessions throughout the semester (every 4 weeks), participants completed
a questionnaire assessing their perceptions of competence, autonomy,
and controUedness regarding their goals. In a final individual session at
the end of the semester, participants completed a T2 physical symptoms
measure, filled out a re~ospective physical symptoms measure, and were
given an explanation of the purpose of the study.
Measures
Definition of personal goals and the avoidance goals measure. Personal goals were defined as "things that you typically or characteristically are trying to do in your dally life." Various dimensions of personal
goals were described (including approach-avoidance), and participants
were instructed to select "the eight personal goals that you think best
describe what you will typically be trying to do in your daily life during
this semester." Two trained individuals independently coded each goal
as approach or avoidance (interjudge agreement was 99.8%, with disagreements resolved through discussion). Examples of approach goals
listed by participants are "get in good shape" and "be more gentle and
humble"; examples of avoidance goals are "avoid procrastination" and
"not be as lazy." An avoidance goals index was created for each participant by summing the number of avoidance goals listed in his or her set
of eight goals (given their high intercorrelation, r = .95, the participant
and coder judgments were combined in creating this measure).
Individual-differences measures. As in Study 1, neuroticism was
assessed with the Neuroticism subscale of Costa and McCrae's (1992)
NEO-FFI; extraversion was assessed with the Extraversion subscale of
the NEO-FFI (sample item: "I am a cheerful, high-spirited person").
1290
ELLIOT AND SHELDON
PemeivedCompetence~:.16~
Retrospective
Symptoms
PerceivedAutonomy
PerceivedControlledness ~
Figure 1. The mediational model for retrospectivesymptoms. Path values are standardized regression
coefficients, and only theoreticallycentral relationships are included in the diagram for presentationclarity.
*p <.05. **p <.01.
Participants indicated their responses to the 12 items on each measure
using a 1 (strongly disagree) to 5 (strongly agree) scale and their
responses were summed to form the neuroticism and extraversion indexes. BIS sensitivity was assessed with Carver and White's (1994)
seven-itemBIS scale (sample item: "I worry about making mistakes" ).
Participants indicated their response to each item on a 1 (strongly disagree) to 4 (strongly agree) scale, and their responses were summed
to form the BIS index.
Process measures. The same process measuresused in Study 2 were
used in this study.
Physical symptomsmeasures. The samephysical symptomsmeasure
used in Study 2 was used in this study t6 assess symptoms at T1, T2,
and retrospectively. For the T1 and T2 measures, participants indicated
how often they had experienced each of the symptoms during the past
few days. For the retrospectivemeasure,participants indicatedhow often
they had experienced each of the symptoms during the semester.
Results
Participant Attrition and Descriptive Statistics
Six of the initial 159 participants did not complete the study,
resulting in an attrition rate of 3.8%. Attrition analyses revealed
no significant differences between the study completion and
study dropout groups. Three additional participants were excluded from analyses because of missing data. The means, standard deviations, ranges, and reliabilities of the variables for the
final sample of 56 male and 94 female participants are presented
in Table 5. Intercorrelations among the primary study variables
are presented in Table 6.
The Direct Relationship Between Avoidance Goals and
Physical Symptoms
Change in physical symptoms. To investigate the relationship between the pursuit of avoidance goals and change in physical symptoms, we regressed T2 symptoms on avoidance goals
and T1 symptoms. The regression yielded a significant effect
forT1 symptoms, F ( 1 , 1 4 7 ) = 56.42,p < .0001,/~ = .52. More
important, avoidance goals also attained significance, F ( 1 , 1 4 7 )
= 6.37, p < .05,/3 = .17, indicating that participants pursuing
a greater number of avoidance goals evidenced an increase in
symptoms from the beginning to the end of the semester, Two
additional analyses were then conducted in which the aforementioned analysis was repeated with an alternative predictor variable (BIS) or set of variables (neuroticism, extraversion, and
Neuroticism × Extraversion) also in the equation. Adding the
alternative predictors had a minimal effect on the relationship
between avoidance goals and change in symptoms: With either
neuroticisml extraversion, and Neuroticism x Extraversion or
BIS in the equation, the coefficient decreased .01 standardized
unit (and remained significant).
Retrospective physical symptoms. To investigate the relationship between the pursuit of avoidance goals and retrospective
physical symptoms, we regresged retrospective symptoms on
avoidance goals (and gender). The regression yielded a significant
effect for avoidance goals, F(1, 147) = 4.01, p < .05, 3 = .16,
indicating that participants pursuing a greater number of avoidance
. goals reported experiencing more symptoms over the semester-
Table 5
Descriptive Statistics for Study 3
Variable
M
SD
Possibler a n g e
Observedrange
Reliability
Avoidance goals
T1 symptoms
T2 symptoms
Retrospective symptoms
Perceived competence
Perceived autonomy
Perceived controlledness
Neuroticism
Extraversion
BIS
1.43
27.67
30.10
32.41
5.44
16.60
10.71
35.59
42.07
21.50
1.26
11.33
13.35
11.95
1.24
3.43
3.63
8.49
6.68
3.56
0-8
13-91
13-91
13-91
1-9
1-27
1-27
12-60
12-60
7-28
0-6
13-59
13-72
13-64
1.79-7.96
7.38-25.83
3.17-22.92
15-58
23-56
7-28
.86
.82
.85
.89
.90
.94
.86
.81
.78
Note. n = 81. T1 = Time 1; T2 = Time 2; BIS = behavioral inhibition system.
AVOIDANCE PERSONAL GOALS
long period. Gender also attained significance, F(1, 147) = 6.63,
p < .05,/5 = .21, indicating that women reported experiencing
more symptoms than men. The same set of alternative predictor
analyses described earlier were conducted for retrospective symptoms. Adding the alternative predictors had a negligible effect
on the relationship between avoidance goals and retrospective
symptoms: With neuroticism, extraversion, and Neuroticism ×
Extraversion in the equation, the beta coefficient remained the
same, whereas with BIS in the equation, the coefficient increased
.01 standardized unit (both betas remained significant).
Mediation
The relationship between avoidance goals and the process
variables. Pearson correlations were computed to determine
the relationships among avoidance goals and perceived competence, perceived autonomy, and perceived controlledness. All
three correlation coefficients were significant. Participants pursuing a greater number of avoidance goals reported lower perceptions of progress ( r = - . 2 9 , p < .0005), lower perceptions
of autonomy ( r = -.27, p < .001), and higher perceptions of
controlledness ( r = .23, p < .01 ) during the semester.7
The relationship between the process measures and change
in physical s~mptoms. To test the final link in the mediational
model for change in physical symptoms, we conducted a multiple regression analysis in which T2 symptoms was regressed
on perceived competence, perceived autonomy, and perceived
controlledness with T1 symptoms and avoidance goals (and
gender) also in the equation. The relationship between perceived
autonomy and T2 symptoms was not significant; therefore, perceived autonomy was trimmed from the final model. The regression yielded a significant effect for T1 symptoms, F ( 1 , 1 4 4 ) =
37.45, p < .0001,/~ = .43. More important, perceived competence, F ( 1 , 144) = 4.81, p < .05, fl = -.16, and perceived
controlledness, F ( 1, 144) = 6.15, p < .05,/5 = .18, were also
significant predictors of T2 symptoms, indicating that lower
perceptions of competence and higher perceptions of controlledness were associated with an increase in symptoms. Gender also attained significance, F(1, 144) = 6.71, p < .05, /~
= .18, indicating that women evidenced a greater increase in
symptoms than men. The direct relationship between avoidance
goals and T2 symptoms was no longer significant with the mediator variables in the equation (p > .40), and the beta coefficient
for this relationship dropped f r o m . 17 to .06.
SEM analyses of the mediational model for change in physical symptoms. SEM was used to test the paths in the hypothesized mediational model sirrmltaneously and to assess the fit of
the model to the observed data (see Study 2 for details on
the SEM procedures). The standardized parameter estimates
obtained in this analysis were essentially identical to (within
.01 standardized unit of) those obtained with the regression
procedures. In addition, the analyses revealed that the model
provided a satisfactory fit to the data, X2(8, N = 150) = 13.65,
p > .05; GFI = .97, AGFI = .91, CFI = .95, and IFI = .96.
Repeating this SEM analysis and including an alternative predictor variable (BIS) or set of variables (neuroticism and extraversion) and their significant or marginally significant paths (as
evidenced in regression analyses) yielded the same results. Each
of the hypothesized relationships remained p < .05, and all
goodness-of-fit statistics in both analyses indicated that the
1291
model provided a satisfactory fit to the data. In summary, the
results from both the regression and SEM analyses clearly established perceived competence and perceived controlledness as
mediators of the relationship between avoidance goals and
change in physical symptoms.
The relationship between the process measures and retrospective physical symptoms. To test the final link in the mediational model for retrospective physical symptoms, we conducted
a multiple regression analysis in which retrospective symptoms
was regressed on perceived competence, perceived autonomy,
and perceived controlledness with avoidance goals (and gender)
also in the equation. The relationship between perceived autonomy and retrospective symptoms was not significant; therefore,
perceived autonomy was trimmed from the final model. The
regression revealed that both perceived competence, F ( 1, 145 )
= 10.19, p < .005, /~ --- -.25, and perceived controlledness,
F ( 1 , 1 4 5 ) = 8.59,p < .005,/~ = .23, were significant predictors
of retrospective symptoms, indicating that lower perceptions
of competence and higher perceptions of controlledness were
associated with higher reports of symptoms. Gender alSO atmined significance, F(1, 145) = 11.35, p < .005, /~ = .26,
indicating that women reported more symptoms than men. The
direct relationship between avoidance goals and retrospective
symptoms was no longer significant with the mediator variables
in the equation (p > .74), and the beta coefficient for this
relationship dropped from .16 to .03.
SEM analyses of the mediational model for retrospective
physical symptoms. As with the longitudinally based measures, we also conducted an SEM analysis, and the standardized
parameter estimates obtained in this analysis were essentially
identical to (within .01 standardized unit of) those obtained
with the regression procedures. In addition, the analysis revealed
that the model provided a satisfactory fit to the data, X 2(8, N
= 150) = 12.74, p > .05; GFI = .97, AGFI = .92, CFI = .93,
and IFI = .94. Repeating this SEM analysis and including an
alternative predictor variable (BIS) or set of variables (neuroticism and extraversion) and their significant or marginally significant paths (as evidenced in regression analyses) yielded the
same results. Each of the hypothesized relationships remained
p < .05, and all goodness-of-fit statistics in both analyses indicated that the model provided a satisfactory fit to the data. In
summary, the results from both the regression and SEM analyses
clearly established perceived competence and perceived controlledness as mediators of the relationship between avoidance
goals and retrospective symptoms (see Figure 2 for a pictorial
summary of the primary results).
Discussion
The results of this study conceptually replicate Studies 1 and
2. Avoidance goal pursuit was positively related to a longitudinal
In both Studies 2 and 3, ancillary analyses were conducted in which
the relationship between avoidance goals and the mediational variables
was tested at the goal (within-subjects) level of analysis, controlling for
the fact that goals were nested within subjects (see Emmons, 1991, for
a discussion of the distinction between the goal and person [betweensubjects] levels of analysis). The hypothesized relationships were supported in these analyses as well, as five of the six effects were significant
(only the relationship between avoidance goals and perceived controlledness in Study 3 failed to attain significance in these analyses).
1292
ELLIOT AND SHELDON
Table 6
Intercorrelations Among the Primary Variables in Study 3
Variable
1.
2.
3.
4.
5.
6.
7.
8.
Avoidance goals
T1 symptoms
T2 symptoms
Retrospective symptoms
Perceived competence
Perceived autonomy
Perceived controlledness
Neuroticism
1
2
3
4
5
6
7
.05
.20*
.20"
-.29**
-.28**
.23**
.17"
-.53**
.67"*
-.21"*
-.17"
.26**
.30**
-.82**
-.29**
-.24**
.33**
.33**
--.30**
-.16
.28**
.32**
.24"*
-.22"*
-.39"*
-.09
-.29"*
-.26**
8
Note. n = 150. T1 = Time 1; T2 = Time 2.
*p<.05.
**p<.01.
increase in physical symptom reports from the beginning to the
end of the semester. This relationship remained significant when
controlling for neuroticism, as well as the other alternative predictor variables. The same results were obtained with the retrospective indicator of symptoms. The mediational results documented in Study 2 were again obtained in this study. Avoidance
goals predicted perceived competence and controlledness and
these variables, in turn, predicted longitudinal and retrospective
symptom reports. Avoidance goals were negatively related to
perceived autonomy, but this variable was not a significant predictor of symptoms.
Meta-Analyses of Studies 1, 2, and 3
Primary Issues
To determine the robustness of the hypothesized relationships,
we combined results from the three studies meta-analytically
with the Stouffer method (Rosenthal, 1978). In conducting these
analyses, we averaged the longitudinal and retrospective results
from Studies 1 and 3 to maintain the orthogonality of each
contributing data point. As shown in Table 7, each of the hypothesized relationships proved highly robust when combined across
studies.
Ancillary Issues
The Stouffer method was also used to explore two ancillary
issues: the relationship between avoidance goals and specific
symptoms and the precise roles played by neuroticism and
avoidance goals as joint predictors of physical symptomatology.
Results from the specific symptom analyses clearly attest to the
pervasiveness of the inimical impact of avoidance goal pursuit.
As shown in Table 7, avoidance goals were significantly positively related to 10 of the 13 symptom variables and the other
three relationships displayed a comparable trend.
Our research has established avoidance goals as predictors
of physical symptom reports, and previous researchers have
linked neuroticism to both avoidance goal adoption (Elliot et al.,
1997) and symptom reports (Costa & McCrae, 1987; Watson &
Pennebakei; 1989). Considered together, these findings suggest
two possibilities regarding neuroticism and avoidance goals as
predictors of symptoms: (a) Neuroticism prompts the adoption
of avoidance goals and then both variables independently predict
symptoms or (b) neuroticism prompts the adoption of avoidance
goals, which then serve as mediators of the relationship between
neuroticism and symptoms. Our data afforded a test of these
possibilities, and results from across the three studies provide
support primarily for the first model. Congruent with previous
research, neuroticism was a significant predictor of both avoidance goals (z = 2.77, p < .006) and symptom reports (z =
5.86, p < .00001 ). When considered as simultaneous predictors
of symptoms, both neuroticism (z = 4.91, p < .00001) and
avoidance goals (see above results) were significant and the
relationship between neuroticism and symptoms decreased by
.02 standardized units on average, indicating that avoidance
goals only mediated a small portion of this relationship.
PerceivedCompetence ~
-.16"
..29.)~/Y
AvoidanceGoals
Sym~oms
PerceivedAutonomy ~
y
",. RetrospectiveSymptoms
PerceivedControlledness~
Figure 2. The mediational model for change in symptoms and retrospective symptoms. Path values are
standardized regression coefficients and only theoretically central relationships are included in the diagram
for presentation clarity. *p < .05. **p < .01.
AVOIDANCE PERSONAL GOALS
1293
Table 7
Meta-Analytic Results f o r Studies 1 - 3
Relationship
Avoidance goals --* A physical Symptoms
Avoidance goals --, retrospective physical symptoms
Avoidance goals ---,physical sYmptoms
Acne or pimples
Chest or heart pain
Coughing or sore throat
Faintness or dizziness
Feeling weak in parts of your body
Headaches
Hot or cold spells
Nausea or upset stomach
Numbness or tingling in parts of your body
Runny or congested nose
Shortness of breath
Stiff or sore muscles
Stomach ache or pain
Avoidance goals --, A physical symptoms
(controlling for neuroticism)
Avoidance goals ---,retrospective physical symptoms
(controlling for neuroticism)
Avoidance goals ---,physical symptoms
(controlling for neuroticism)
Avoidance goals --, perceived competence
Avoidance goals ---,perceived autonomy
Avoidance goals ---,perceived controlledness
Perceived competence --*physical symptoms
Perceived autonomy ---,physical symptoms
Perceived controlledness ---,physical symptoms
Avoidance goals --, physical symptoms
(controlling for mediators)
Meta-analytic
result (z)
3.34
3.37
3.69
1.24
2.35
2.47
p
(two-tailed)
No. of
studies
3.19
2.70
2.56
2.14
2.04
2.18
1.60
2.75
<.0009
<.0008
<.0003
<.22
<.019
<.014
< .047
<.17
< .002
<.008
<.011
<.033
< .042
<.030
<.11
< .007
2
3
3
3
3
3
3
2
3
2
2
2
3
3
3
3
3.12
<.002
2
2.99
< .003
3
<.002
<.00001
<.00001
<.00001
< .002
>.32
<.130001
3
2
2
2
2
2
2
> .54
2
1.99
1.38
3.28
-4.28
-4.92
4.80
-3.21
-0.98
4.76
0.60
Note. The z values were obtained with the Stouffer method. The term physical symptoms indicates a
combination of longitudinal and retrospective measures.
General Discussion
In three studies we investigated the hypothesis that avoidance
personal goals are positively related to physical symptom reports. In two of these studies we also tested a process model
that posited perceived competence and perceived self-determination as mediators of the hypothesized relationship. The results
provide clear and strong support for both the direct and mediational hypotheses. Avoidance goals predicted symptom reports
in each of the three studies, and across studies, this relationship
was documented (a) using both longitudinal (Studies 1 and 3)
and retrospective (Studies 1 - 3 ) methodologies; (b) with two
types of longitudinal designs, one emphasizing full coverage of
the study-long period (Study 1 ) and the other sampling brief
time periods at the beginning and end of the study (Study 3);
(c) over two temporal periods--1 month (Study 1) and a 4month semester (Studies 2 and 3 ) - - t h a t presumably evoked
different levels of goals (molecular and molar, respectively);
and (d) controlling for neuroticism and several additional thirdvariable candidates. In both studies in which the hypothesized
process model was tested, perceived competence and perceived
controUedness were validated as mediators of the observed relationship. These mediational processes were documented in the
context of both longitudinal and retrospective methodologies,
with SEM as well as multiple regression procedures, and both
with and without controlling for neuroticism and other alternative predictor variables.
Davidson and Pennebaker (1996) have delineated three different categories of illness indicators: self-report measures (e.g.,
symptom reports ); biological measures (e.g., objective evidence
of a cold or upper respiratory infection); and behavioral measures (e.g., absenteeism, physician visits). Each of these categories, which are typically at least moderately intercorrelated
(Brown & Moskowitz, 1997; Pennebaker, 1982), is considered
by Davidson and Pennebaker (1996) to be a "valid and powerful
reflection of illness" (p. 107), but each is also presumed to
have distinct weaknesses that require careful methodological
attention in empirical work. In the present research, we focused
exclusively on the self-report category and attempted to design
studies that carefully addressed the methodological challenges
posed by this particular class of measure. As such, in Studies
1 and 3 we examined the relationship between avoidance goals
and symptom reports while simultaneously controlling for baseline levels of symptom reports and the primary nuisance variable
identified in the personality-illness literature: neuroticism.
Symptom report research has been (justifiably) critiqued as
lacking in methodological rigor and sophistication (Wiebe &
Smith, 1997), and we strongly recommend that researchers implement these or comparably rigorous procedures (see Brown &
Moskowitz, 1997; Larsen, 1992) when conducting empirical
1294
ELLIOT AND SHELDON
investigations involving symptom reports or any other self-report measure of illness.
Although the methodological rigor of our studies makes it
tempting to conclude that avoidance goal pursuit predicts actual
physical illness, the degree to which the symptom outcome
measures reflected true organically based variance is not known
with certainty. Definitive conclusions about organic condition
require the use of biological measures, but we did not use such
assessments in our research. Another issue that we did not address was the relationship between the pursuit of avoidance
goals and illness-related behaviors. Future empirical work is
needed to examine the link between avoidance goals and these
additional indicators of illness, attending to the unique methodological challenges presented by each class of measure (see
Davidson & Pennebaker, 1996; Larsen & Kasimatis, 1991;
Maddi, Bartone, & Puccetti, 1987). One intriguing possibility
for future research is to investigate symptom perceptions and
reports as a mediator of the relationship between avoidance
goals and illness-related behaviors such as physician visits. Our
research demonstrated that avoidance goal adoption leads to
symptom perceptions and reports, and clearly symptom perception is a prerequisite to proceeding to a physician's office for
medical attention. From this perspective, avoidance goals represent distal predictors of health care utilization, and avoidance
goal adoption would appear to have highly pragmatic implications for the health care system regardless of whether it leads
to actual organically based pathology or mere somatization (see
Karoly & Lecci, 1993; Larsen, 1992).
Our research demonstrates that avoidance (relative to approach) goal pursuit has negative implications for physical
health, but it does not necessarily follow that all types of avoidance goals are inimical or that all types of approach goals are
optimal. To the extent that avoidance goals are linked to concrete, approach-oriented subgoals, they may prove somewhat
adequate forms of regulation, and avoidance goals focused on
actively eradicating an existing negative situation may be less
deleterious than those that involve passive rumination about a
future undesirable possibility. Likewise, approach goals may be
divided into different categories such as extrinsic and intrinsic
(Kasser & Ryan, 1996), performance and learning based
(Dweck, 1986), and state and action oriented (Kuhl, 1984),
and it is possible that extrinsic, performance-based, and stateoriented forms of approach regulation have a negative impact
on physical health. Note that if any of these speculations are
accurate, our research represents a conservative test of the relationship between avoidance goals and symptoms because avoidance and approach goals were conceptualized and operationalized in an undifferentiated manner.
A central aspect of our research, and one that is typically
neglected in empirical work in the personality-illness literature,
was the documentation of mediation. As predicted, perceived
competence and perceived self-determination were the psychologically based mediational mechanisms responsible for the relationship between avoidance goals and physical symptomatology.
Interestingly, the self-determination data indicated that the presence of controlledness undermined physical health, not that the
presence of autonomy served as a prophylactic. This enhanced
predictive utility of controlledness relative to autonomy is congruent with recent work by David, Green, Martin, and Suls
(1997), who documented the greater utility of negatively versus
positively valenced constructs in predicting negative well-being
outcomes.
The simultaneous validation of perceived competence and
perceived controlledness as "joint mediators" in the present
studies is noteworthy for two reasons: First, joint mediational
models represent unexplored territory in the personality-illness
literature (and are quiterare in g0al and self-regulation research
more generally; for examples, see Elliot & Harackiewicz, 1994;
Sansone, 1986). Second, most contemporary accounts of motivation are grounded in competence and either ignore the selfdetermination construct or denigrate it as conceptually vacuous
(see Bandura, 1989). The present data, however, indicate that
perceptions of competence and self-determination have significant and independent influences on physical health. Congruent
with the modal conceptualization of motivation, perceived competence was linked to a decrease in symptom reports regardless
of perceived controlledness, but, importantly, the data also
linked perceived controlledness to an increase in symptom reports regardless of perceived competence. Thus, it appears that
competence and self-determination account for unique variance
in the health domain, suggesting that a thorough, process-oriented analysis of the personality-illness relationship will need
to include both of these variables. Of course, other mediational
variables also warrant consideration in subsequent research on
avoidance goals and illness, and the personality-illness relationship more generally. Promising candidates include additional
psychologically based variables such as subjective stress (see
Glaser et al., 1993) as well as more objective, physiologically
based variables such as natural killer cell activity (see Strauman
et al., 1993), immunoglobulin A concentration (see McClelland,
Alexander, & Marks, 1982), and cortisol level (see Cohen-Cole
et al., 1981). We assume that variables of this nature serve the
role of proximal mediator of the relationship between avoidance
goals and physical symptomatology, accounting for the link between perceived competence and controlledness and symptoms
documented in our research.
The mediational results obtained in this research fit nicely
with self-determination theory's portrayal of competence and
self-determination as fundamental psychological needs (Deci &
Ryan, 1991). Competence and self-determination indeed appeared to function as essential requirements for well-being, as
individuals who felt incompetent or controlled in their goal
pursuits evidenced an increase in physical symptomatology. In
light of these data, avoidance goals must be considered a psychological vulnerability because they place individuals at risk of
not receiving the basic psychological nutrients necessary for
physical health (Ryan, 1995). Personal goals represent, in essence, the vehicles through which individuals negotiate their
daily lives, and these data suggest that some vehicles (approach
goals) are better suited for the terrain of everyday life than
others (avoidance goals). Parenthetically, we should note that
personal goals, by their very nature as idiographic self-regulatory vehicles, are not amenable to experimental manipulation.
Therefore, despite the rigor of the methodological procedures
and data-analytic techniques employed in this research, the direct and mediational results are correlational in nature and definitive causal statements are not warranted, s
8This statement is most relevant to the mediational results because
the nature of the independent variable (avoidance goals adopted at the
AVOIDANCE PERSONAL GOALS
From a broader perspective, our research illustrates the utility
of a personal goals approach to the personaiity-illness relationship. We contend that the personal goals approach holds great
promise for the personality-illness literature, in that it possesses
attributes that are distinct from, and nicely complement, those
of the disposition and coping strategy constructs that dominate
the contemporary scene. For instance, relative to dispositional
constructs, personal goals are more idiographic (Emmons,
1989); more dynamic (representing the "doing" as opposed to
the "having" aspect of personality; Cantor, 1990); more directly applicable to specific behavior in specific situations (Littie, 1989); and, some might argue, better equipped to explain
(rather than merely describe) behavior (see Pervin, 1994).
Compared with coping strategies, personal goals are more idiographic (Coyne & Gottlieb, 1996; cf. Aspinwall & Taylor,
1997); focus on positive as well as stressful outcomes and events
(see Lazarus & Folkman, 1984); represent anticipatory, proactive strategies rather than post hoc, reactive strategies (Coyne &
Gottlieb, 1996); and are more amenable to the broad, representative sampling of self-regulatory experience (i.e., personal goal
assessments commonly include 8 - 1 0 goals, whereas coping assessments often involve reactions to a single stressful event;
Suls et al., 1996). It is interesting to note that the approachavoidance distinction (or its conceptual equivalent) occupies a
prominent place in all three of the approaches under consideration. Several dispositional theorists distinguish between broad
appetitive, approach-based (behavioral activation system, extraversion, positive affectivity) and aversive, avoidance-based
(BIS., neuroticism, negative affectivity) regulatory or experiential systems (Gray, 1987; Tellegen, 1985; Watson & Clark,
1984); some coping frameworks explicitly utilize the approachavoidance distinction, and most structural analyses of coping
yield an avoidance-based factor (Billings & Moos, 1981; see
Suls etal., 1996); and, of course, our research highlights the
need to attend to the approach-avoidance distinction with regard to personal goals.
It is important to bear in mind that the aim of personalityillness research is not simply to develop a taxonomy of constructs or a laundry list of variables within each type of construct
but to formulate overarching, integrative models of the personality-illness relationship. Nearly all of the research in the personality-illness literature, and the personality- well-being literature
more generally, has been conducted "within construct," and it
is only in the past few years that attempts have been made to
forge links between constructs. Specifically, researchers have
begun to examine dispositions and coping strategies as joint
beginning of the semester-long period) did not afford a baseline assessment of ongoing perceived competence, autonomy, and controlledness.
In response to a reviewer's concern that general perceptions of incompetence and controlledness or low self-esteem may actually be responsible
for our mediational results (see Baumeister, Heatherton, & Tice, 1993),
we repeated the primary direct and mediational regression analyses and
the structural equation modeling analyses in Study 3 controlling for selfesteem as measured by Rosenberg's (1965) scale (a measure we had
included for use in a separate investigation). All these analyses produced
results that were the same as those reported in the text (i.e., all the
significant relationships and paths remained significant, and all the fit
statistics continued to indicate that the model provided a satisfactory fit
to the data).
1295
predictors of illness/well-being outcomes, and results to date
have revealed that (a) dispositions lead to coping strategies
and (b) coping strategies, in turn, predict illness/well-being
outcomes and sometimes mediate the direct effects of dispositions (Aspinwall & Taylor, 1992; Bolger, 1990; Bolger & Zuckerman, 1995; Carver et al, 1993; Florian, Mikulincer, & Taubman, 1995; Holahan & Moos, 1990; Scheier et al., 1989; Stanton & Snider, 1993). The inclusion of neuroticism in the present
studies enabled us to examine, in ancillary analyses, a different
type of integrative link: that between dispositions and personal
goals. Results indicated that neuroticism led to avoidance goal
adoption and that avoidance goals in turn predicted symptom
reports. Neuroticism remained a significant predictor of symptoms (avoidance goals mediated only a small portion of the
direct effect), suggesting the need to identify an additional mediator variable, that N is simply associated with certain forms
of "pathophysiological" activity that leads directly to illness,
or both (Wiebe & Smith, 1997).
Putting the disposition-personal goal and disposition-coping
strategy research together suggests a framework of the type
presented in Figure 3a. In this integrative model, global nomothetic dispositions proactively recruit specific idiographic personal goals and influence the coping strategies adopted in response to specific stressful events. These personal goals and
coping strategies in turn influence illness/welt-being outcomes.
Some dispositions, particularly those like neuroticism and extraversion or BIS and beh~tvioral activation system that are presumably undergirded by neuroanatomical substrates (H. Eysenck,
1967; Gray, 1987), are likely to maintain a direct unmediated
relationship with illness/well-being outcomes. Figure 3b shows
a different variant of the integrative model, one that incorporates
the possibility that personal goals predict the coping strategies
that are implemented in response to stressful events. Of course,
other variants of integrative model are possible (e.g., personal
goals may completely mediate the direct relationship between
dispositions and coping strategies); there are other complexities
that pose additional conceptual challenges (e.g., dispositions,
personal goals, or both may influence the degree to which stressful events are encountered, and, therefore, coping strategies are
necessary; see Bolger & Zuckerman, 1995); and a Complete,
detailed analysis would need to incorporate more proximal process variables such as perceptions of competence and self-determination. As such, the proposed models are provided primarily
for their heuristic value.
In closing, our research demonstrates the utility of both the
approach-avoidance distinction and the personal goals construct in accounting for variation in physical symptomatology.
The approach-avoidance distinction has a long and rich history
in scientific psychology and is present, in implicit or explicit
form, in a number of variables in the personality-illness/wellbeing literature. The personal goals unit of analysis nicely complements the disposition and coping strategy constructs that are
utilized in most personality-illness/well-being research, and
the personal goals approach seems relatively easy to implement,
either alone or in conjunction with the more established approaches. Accordingly, we believe that both the approachavoidance distinction and the personal goals approach have the
potential to make a substantial contribution to the personalityillness/well-being literature in general, and we suspect that both
1296
ELLIOT AND SHELDON
a)
Personal
Dispositions
Illness/ Well-being
Coping ~
Outcomes
Strategies
b)
Dispositions
=
Goals
Illness/ Well-being
Coping
~ Strategies
Figure 3. Integrative models of the relationship between personality and illness/well-being.
will indeed figure prominently in the development of integrative
theoretical models in this domain.
References
Aiken, L., & West, S. (1991). Multiple regression.. Testing and interpreting interactions. Newbury Park, NJ: Sage.
Aspinwall, L., & Taylor, S. (1992). Modeling cognitive adaption: A
longitudinal investigation of the impact of individual differences and
coping on college adjustment and performance. Journal of Personality
and Social Psychology, 63, 989-1003.
Aspinwall, L., & Taylor, S. (1997). A stitch in time: Self-regulation and
proactive coping. Psychological Bulletin, 121, 417-436.
Atkinson, J. (1957). Motivational determinants of risk-taking behavior.
Psychological Review, 64, 359-372.
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.
Bandura, A. (1989). Human agency in social cognitive theory. American
Psychologist, 44, 1175-1184.
Bandura, A., Cioffi, D., Taylor, C., & Brouillard, M. (1988). Perceived
self-efficacy in coping with cognitive stressors and opioid activation.
Journal of Personality and Social Psychology, 55, 479-488.
Baumeister, R., Heaterton, T., & Tice, D. (1993). When ego threats lead
to self-regulation failure: Negative consequences of high self-esteem.
Journal of Personality and Social Psychology, 64, 141-156.
Billings, A. G., & Moos, R. H. (1981). The roles of coping responses
and social resources in attenuating the stress of life events. Journal
of Behavioral Medicine, 4, 139-157.
Bolger, N. (1990). Coping as a personality process: A prospective study.
Journal of Personality and Social Psychology, 59, 525-537.
Bolger, N., & Schilling, E. (1991). Personality and the problems of
everyday life: The role of neuroticism in exposure and reactivity to
daily stressors. Journal of Personality, 59, 355-386.
Bolger, N., & Zuckerman, A. (1995). A framework for studying person-
ality in the stress process. Journal of Personality and Social Psychology, 69, 890-902.
Brown, K., & Moskowitz, D. (1997). Does unhappiness make you sick?
The role of affect and neuroticism in the experience of common
physical symptoms. Journal of Personality and Social Psychology,
72, 907-917.
Brunstein, J. (1993). Personal goals and subjective well-being: A longitudinal study. Journal of Personality and Social Psychology, 65,
1061-1070.
Brunstein, J., Dangelmayer, G., & Schultheiss, O. (1996). Personal goals
and social support in close relationships: Effects on relationship mood
and marital satisfaction. Journal of Personality and Social Psychology, 71, 1006-1018.
Cantor, N. (1990). From thought to behavior: "Having" and "doing"
in the study of personality and cognition. American Psychologist, 45,
735-750.
Cantor, N., Acker, M., & Cook-Flannagan, C. (1992). Conflict and
preoccupation in the intimacy life task. Journal of Personality and
Social Psychology, 63, 644-655.
Cantor, N., Markus, H., Niedenthal, P., & Nurius, P. (1986). On motivation and the self-concept. In R. Sorrentino & E. T. Higgins (Eds.),
Motivation and cognition: Foundations of social behavior (pp. 96121 ). New York: Guilford Press.
Cantor, N., & Zirkel, S. (1990). Personality, cognition, and purposive
behavior. In L. Pervin (Ed.), Handbook of personality: Theory and
research (pp. 135-164). New York: Guilford Press.
Carver, C., Pozo, C., Harris, S., Noriega, V., Scheier, M., Robinson, D.,
Ketcham, A., Moffat, E, & Clark, K. (1993). How coping mediates
the effect of optimism on distress: A study of women with early stage
breast cancer. Journal of Personality and Social Psychology, 65, 375390.
Carver, C., & Scheier, M. (1994). Situational coping and coping dispositions in a stressful transaction. Journal of Personality and Social
Psychology, 66, 184-195.
AVOIDANCE PERSONAL GOALS
Carver, C., Scheier, M., & Weintraub, J. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56, 267-283.
Carver, C., & White, T. (1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment:
The BIS/BAS scales. Journal of Personality and Social Psychology,
67, 319-333.
Coats, E., Janoff-Bulman, R., & Alpert, N. (1996). Approach versus
avoidance goals: Differences in self-evaluation and well-being. Personality and Social Psychology Bulletin, 22, 1057-1067.
Cohen, S., Doyle, W., Skonet; D., Fireman, P., Gwaltney, J., & Newsome,
J. (1995). State and trait negative affect as predictors of objective
and subjective symptoms of respiratory viral infections. Journal of
Personality and Social Psychology, 68, 159-169.
Cohen, S., & Williamson, G. (1991). Stress and infectious disease in
humans. Psychological Bulletin, 109, 5-24.
Cohen-Cole, S., Cogen, R., Stevens, A., Kirk, K., Gaitanl E., Hain, J., &
Freeman, A. ( 1981 ). Psychosocial, endocrine, and immune factors in
acute necrotizing ulcerative gingivitis ("trenchmouth"). Psychosomatic Medicine, 43, 91.
Contrada, R., Leventhal, H., & O'Leary, A. (1990). Personality and
health. In L. Pervin (Ed.), Handbook of personality: Theory and
research (pp. 638-669). New York: Guilford Press.
Costa, P., & McCrae, R. (1987). Neuroticism, somatic complaints, and
disease: Is the bark worse than the bite? Journal of Personality, 55,
299-316.
Costa, P., & McCrae, R. (1992). Revised NEO Personality Inventory
(NEO-PI-R) and NEO Five Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources.
Covington, M. (1992). Making the grade: A self-worth perspective on
motivation and school reform. New York: Cambridge University Press.
Coyne, J., & Gottlieb, B. (1996). The mismeasure of coping by checklist. Journal of Personality, 64, 959-991.
Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. New York: Harper & Row.
David, J., Green, P., Martin, R., & Suls, J. (1997). Differential roles of
neuroticism, extraversion, and event desirability for mood in daily
life: An integrative model of top-down and bottom-up influences.
Journal of Personality and Social Psychology, 73, 149-159.
Davidson, K., & Pennebaker, J. (1996). Social psychosomatics. In E. T.
Higgins & A. Kruglanski (Eds.), Social psychology: Handbook of
basic principles (pp. 102-130). New York: Guilford Press.
Deci, E., & Ryan, R. ( 1991 ). A motivational approach to self: Integration
in personality. In R. Dienstbier (Ed.), Nebraska symposium on motivation (Vol. 38, pp. 237-288). Lincoln: University of Nebraska Press.
Derogatis, L., & Spencer, P. (!982). The Brief Symptom Inventory
(BSI). Baltimore: Psychometric Research.
Derryberry, D., & Reed, M. (1994). Temperament and attention: Orienting toward and away from positive and negative signals. Journal of
Personality and Social Psychology, 66, 1128-1139.
Dweck, C. (1986). Motivational processes affecting learning. American
Psychologist, 41, 1040-1048.
Eagleston, J., Kirmil-Gray, K., Thoresen, C., Wiedenfeld, S., Bracke,
P., Heft, L., & Arnow, B. (1986). Physical health correlates of Type
A behavior in children and adolescents. Journal of Behavioral Medicine, 9, 341-362.
Elliot, A. (1997). Integrating the "classic" and "contemporary" approaches to achievement motivation: A hierarchical model of approach
and avoidance achievement motivation. Advances in Motivation and
Achievement, 10, 243-279.
Elliot, A., & Church, M. (1997). A hierarchical model of approach and
avoidance achievement motivation. Journal of Personality and Social
Psychology, 72, 218-232.
Elliot, A., & Harackiewicz, J. (1994). Goal setting, achievement orientation, and intrinsic motivation: A mediational analysis. Journal of Personality and Social Psychology, 66, 968-980.
1297
Elliot, A., & Harackiewicz, J. (1996). Approach and avoidance achievement goals and intrinsic motivation: A mediational analysis. Journal
of Personality and Social Psychology, 70, 461-475.
Elliot, A., & Sheldon, K. (1997). Avoidance achievement motivation:
A personal goals analysis. Journal of Personality and Social Psychology, 73, 171-185.
Elliot, A., Sheldon, K., & Church, M. (1997). Avoidance personal goals
and subjective well-being. Personality and Social Psychology Bulletin, 23, 915-927.
Emmons, R. (1986). Personal strivings: An approach to personality and
subjective well-being. Journal of Personality and Social Psychology,
51, 1058-1068.
Emmons, R. (1989). The personal striving approach to personality. In
L. Pervin (Ed.), Goal concepts in personality and social psychology
(pp. 87-126). HiUsdale, NJ: Erlbaum.
Emmons, R. ( 1991 ). Personal strivings, daily life events, and psychological and physical well-being. Journal of Personality, 59, 453-472.
Emmons, R. (1992). Abstract versus concrete goals: Personal striving
level, physical illness, and psychological well-being. Journal of Personality and Social Psychology, 62, 292-300.
Emmons, R., & Kaiser, H. (1996). Goal orientation and emotional Wellbeing: Linking goals and affect through the self. In L. Martin & A.
Tesser (Eds.), Striving and feeling: Interactions among goals, affect,
and self-regulation (pp. 79-98). Mahwah, NJ: Erlbaum.
Emmons, R., & King, L. (1988). Conflict among personal strivings:
Immediate and long-term implications for psychological and physical
well-being. Journal of Personality and Social Psychology, 54, 10401048.
Epstein, S., & Katz, L. (1992). Coping ability, stress, productive load,
and symptoms. Journal of Personality and Social Psychology, 62,
813-825.
Eysenck, H. (1967). The biological basis of personality, Springfield,
IL: Charles C. Thomas.
Eysenck, H., & Eysenck, M. (1985). Personality and individual differences. New York: Plenum.
Eysenck, S., Eysenck, H., & Barrett, P. (1985). A revised version of
the psYChoticism scale. Personality and Individual Differences, 6, 21 29.
Florian, V., Mikulincer, M., & Taubman, O. (1995). Does hardiness
contribute to mental health during a stressful real-life situation? The
roles of appraisal and coping. Journal of Personality and Social Psychology, 68, 687-695.
Fowles, D. (1988). Psychophysiology and psychopathology: A motivational approach. Psychophysiology, 25, 373-391.
Freud, S. (1957). Repression. In J. Strachey (Ed. and Trans.), The
standard edition of the complete psychological works of Sigmund
Freud (Vol. 14). London: Hogarth Press. (Original work published
1915)
Friedman, H., & Booth-Kewley, S. (1987). The "disease-prone personality": A meta-analytic view of the construct. American Psychologist,
42, 539-555.
Funk, S., & Houston, B. (1987),. A critical analysis of the Hardiness
Scale's validity and utility. Journal of Personality and Social Psychology, 53, 572-578..
Glaser, R., Pearson, G., Bonneau, R., Esterling, B., Atldnson, C., &
Kiecolt-Glaser, J. (1993). Stress and the memory T-cell response to
the Epstein-Barr virus in healthy medical students. Health Psychology,
12, 435-442.
Gray, J. ( 1987 ). The psychology offear and stress (2nd ed.). New York:
Cambridge University Press.
Greenberg, M., Wortman, C., & Stone, A. (1996). Emotional expression
and physical health: Revising traumatic memories or fostering selfregulation? Journal of Personality and Social Psychology, 71, 588602.
Grow Kasser, V., & Ryan, R. (in press). The relation of psychological
1298
ELLIOT AND SHELDON
needs for autonomy and relatedness to vitality, well-being, and mortality in a nursing home. Journal of Applied Social Psychology.
Hembree, R. (1988). Correlates, causes, effects, and treatment of test
anxiety. Review of Educational Research, 58, 47-77.
Higgins, E. T. (1996). The "self digest": Self-knowledge serving selfregulatory functions. Journal of Personality and Social Psychology,
71, 1062-1083.
Holahan, C., & Moos, R. (1990). Life stressors, resistance factors,
and improved psychological functioning: An extension of the stress
resistance paradigm. Journal of Personality and Social Psychology,
58, 909-917.
Holroyd, K.,& Coyne, J. (1987). Personality and health in the 1980s:
Psychosomatic medicine revisited? Journal of Personality, 55, 359375.
Homey, K. (1945). Our inner conflicts. New York: Norton.
Hoyle, R., & Panter, A. (1995). Writing about structural equation models. In R. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 100-119). Thousand Oaks, CA: Sage.
Hull, C. (1943). Principles of behavior. New York: Appleton-CenturyCrofts.
Hull, J., Tedlie, J., & Lehn, D. (1995). Modeling the relation of personality variables to symptom complaints: The unique role of negative
affectivity. In R. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 217-235). New York: Sage.
Janz, N., & Becker, M. (1984). The health belief model: A decade later.
Health Education Quarterly, 11, 1-47.
J6reskog, K., & S6rbom, D. (1993). L1SREL 8 user's guide. Chicago:
Scientific Software International.
Judd, C., & Kenny, D. (1981). Process analysis: Estimating mediation
in treatment evaluations. Evaluation Review, 5, 602-619.
Karoly, P. (1991). Goal systems and health outcomes across the life
span: A proposal. In H. Schroeder (Ed.), New directions in health
psychology assessment (pp. 65-92). New York: Hemisphere.
Karoly, P., & Lecci, L. (1993). Hypochondriasis and Somatization in
college women: A personal projects analysis. Health Psychology, 12,
103-109.
Kasser, T., & Ryan, R. (1996). Further examining the American Dream:
Differential correlates of intrinsic and extrinsic goals. Personality and
Social Psychology Bulletin, 22, 280-287.
Kemeny, M., Cohen, F., Zegens, L., & Conant, M. (1989). Psychological
and immunological predictions of genital herpes recurrence. Psychosomatic Medicine, 51, 195-208.
Kenny, D. (1979). Correlation and causality. New York: Wiley.
Klinger, E. (1977). Meaning and void: Inner experience and the incentives in people's lives. Minneapolis: University of Minnesota Press.
Kobasa, S., Maddi, S., & Kahn, S. (1982). Hardiness and health: A
prospective study. Journal of Personality and Social Psychology, 42,
168-177.
Krantz, D., Glass, D., & Snyder, M. (1974). Helplessness, stress, level
and the coronary-prone behavior pattern. Journal of Experimental
Psychology, 10, 284-300.
Krantz, D., & Hedges, S. (1987). Some cautions for research on personality and health. Journal of Personulity, 55, 351-357.
Kuhl, J. (1984). Volitional aspects of achievement motivation and
learned helplessness: Toward a comprehensive theory of action control. Progress in Experimental Personality Research, 13, 99-171.
Kunda, Z. (1990). The case for motivated reasoning. Psychological
Bulletin, 3, 480-498.
Larsen, R. (1992). Neuroticism and selective encoding and recall of
symptoms: Evidence from a combined concurrent-retrospective study.
Journal of Personality and Social Psychology, 62, 480-488.
Larsen, R., & Kasimatis, M. (1991). Day-to-day physical symptoms:
Individual differences in the occurrence, duration, and emotional concomitants of minor daily illnesses. Journal of Personality, 59, 387423.
Lavallee, L., & Campbell, J. (1995). Impact of personal goals on self-
regulation processes elicited by daily negative events. Journal of Personality and Social Psychology, 69, 341-352.
Lazarus, R. (1990). Stress, coping, and illness, In H. Friedman (Ed,),
Personality and disease (pp. 97-120). New York: Wiley.
Lazarus, R., & Folkman, S. (1984). Stress, appraisal, and coping. New
York: Springer.
Lecci, L., Karoly, P., Ruehlman, L., & Lanyon, R. (1996). Goal-relevant
dimensions of hypochondriacal tendencies and their relation to symptom manifestation and psychological distress. Journal of Abnormal
Psychology, 105, 42-52.
Lewin, K. (1935). A dynamic theory of personality. New York: McGrawHill.
Little, B. (1983). Personal projects: A rationale and method for investigation. Environment and Behavior, 15, 273-309.
Littler B. (1989). Personal projects analysis: Trivial pursuits, magnificent
obsessions, and the search for coherence. In D. Buss & N. Cantor
(Eds.), Personality psychology: Recent trends and emerging directions (pp. 15-31 ). New York: Springer-Verlag.
Maddi, S. (1990). Issues and interventions in stress mastery. In H.
Friedman (Ed.), Personality and disease (pp. 121-154). New York:
Wiley.
Maddi, S., Bartone, B., & Puccetti, M. (1987). Stressful events are
indeed a factor in physical illness: Reply to Schroeder and Costa
(1984). Journal of Personality and Social Psychology, 52, 833-843.
McClelland, D. (1951). Personality. New York: Dryden Press.
McClelland, D., Alexander, C., & Marks, E. (1982). The need for power,
stress, immune function, and illness among male prisoners. Journal
of Abnormal Psychology, 91, 61-70.
Michaels, C., Michaels, A., & Peterson, C. (1997). Motivation and
health. Advances in Motivation and Achievement, 10, 339-374.
Millet; N. (1944). Experimental studies of conflict. In J. McV. Hunt
( Ed.), Personality and the behavior disorders (Vol. 1, pp. 431-465).
New York: Ronald Press.
Mowrer, O. (1960). Learning theory and behavior. New York: Wiley.
Murray, H. (1938). Explorations in personality. New York: Oxford University Press.
Ouellette Kobasa, S. (1990). Lessons from history: How to find the
person in health psychology. In H. Friedman (Ed.), Personality and
disease (pp. 14-37). New York: Wiley.
Pennebaker, J. (1982). The psychology ofphysical symptoms. New York:
Springer-Verlag.
Pennebaker, J., & Watson, D. ( 1991 ). The psychology of somatic symptoms. Current concepts of somatization: Research and clinical perspectives. Washington, DC: American Psychiatric Press.
Pervin, L. (1994). A critical analysis of current trait theory. Psychological Inquiry, 5, 103-113.
Ratliff-Crain, J., & Banm, A. (1990). Individual differences and health:
Gender, coping, and stress. In H. Friedman (Ed.), Personality and
disease (pp. 226-253). New York: Wiley.
Rhodewalt, R., & Zone, J. (1989). Appraisal of life change, depression,
and illness in hardy and nonhardy women. Journal of Personality
and Social Psychology, 56, 81-88.
Robbins, A., Spence, J., & Clark, H. ( 1991 ). Psychological determinants
of health and performance: The tangled web of desirable and undesirable characteristics. Journal of Personality and Social Psychology,
61, 755-765.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton,
NJ: Princeton University Press.
Rosenstock, I. (1974). The health belief model and preventive health
behavior. Health Education Monographs, 1, 354-=386.
Rosenthal, R. (1978). Combining the results of independent studies.
Psychological Bulletin, 85, 185-193.
Ryan, R. (1995). Psychological needs and the facilitation of integrative
processes. Journal of Personality, 63, 397-427.
Ryan, R. (1996). [Subjective well-being and health-related behaviors].
Unpublished raw data.
AVOIDANCE PERSONAL GOALS
Ryan, R., & Connell, J. (1989). Perceived locus of causality and internalization: Examining reasons for acting in two domains. Journal of
Personality and Social Psychology, 57, 749-761.
Salovey, P., & Birnbaum, D. (1989). Influence of mood on healthrelevant cognitions. Journal of Personality and Social Psychology,
57, 539-551.
Sansone, C. (1986). A question of competence: The effects of competence and task feedback on intrinsic interest. Journal of Personality
and Social Psychology, 51, 918-931.
Scheier, M., & Carver, C. (1985). Optimism, coping, and health: Assessment and implications of generalized outcome expectancies. Health
Psychology, 4, 219-247.
Scheier, M., & Carver, C. (1987), Dispositional optimism and physical
well-being: The influence of generalized outcome expectancies on
health. Journal of Personality, 55, 169-210.
Scheier, M., Carver, C., & Bridges, M. (1994). Distinguishing optimism
from neuroticism (and trait anxiety, self-mastery, and self-esteem): A
reevaluation of the Life Orientation Test. Journal of Personality and
Social Psychology, 67, 1063-1078.
Scbeier, M., Matthews, K., Owens, J., Magovern, G., Lefebvre, R., Abbott, R., & Carver, C. (1989). Dispositional optimism and recovery
from coronary artery bypass surgery: The beneficial effects on physical and psychological well-being. Journal of Personality and Social
Psychology, 57, 1024-1040.
Sheldon, K., & Elliot, A. (1998). Not all personal goals are personal:
Comparing autonomous and controlled reasons for goals as predictors
of effort and attainment. Personality and Social Psychology Bulletin,
24, 546-557.
Sheldon, K., & Kasser, T. (1995). Coherence and congruence: Two
aspects of personality integration. Journal of Personality and Social
Psychology, 68, 531-543.
Sheldon, K., Ryan, R., & Reis, H. (1996). What makes for a good day?
Competence and autonomy in the day and in the person. Personality
and Social Psychology Bulletin, 22, 1270-1279.
Smith, T., Pope, M., Rhodewalt, E, & Poulton, J. (1989). Optimism,
neuroticism, coping, and symptom reports: An alternative interpretation of the Life Orientation Test. Journal of Personality and Social
Psychology, 56, 640-648.
Snyder, C., Harris, C., Anderson, J., Holleran, S., Irving, L., Sogmon,
S., Yoshinobv, L., Gibb, J., Langelle, C., & Harney, P. ( 1991 ). The will
and the ways: Development and validation of an individual-differences
measure of hope. Journal of Personality and Social Psychology, 60,
570-585.
1299
Stanton, A., & Snider, E (1993). Coping with breast cancer diagnosis:
A prospective study. Health Psychology, 12, 16-23.
Strauman, T., Lemieux, A., & Coe, C. (1993). Self-discrepancy and
natural killer cell activity: Immunological consequences of negative
self-evaluation. Journal of Personality and Social Psychology, 64,
1042-1052.
Suls, J., David, J., & Harvey, J. (1996). Personality and coping: Three
generations of research. Journal of Personality, 64, 711-735.
Tanaka, J. (1987). How big is big enough? Sample size and goodnessof-fit in structural equation models with latent variables. "ChildDevelopment, 58, 134-146.
Tellegen, A. (1985). Structures of mood and personality and their relationship to assessing anxiety, with an emphasis on self-report. In A.
Tuma & J. Maser (Eds.), Anxiety and the anxiety disorders (pp. 681 706). Hillsdale, NJ: Erlbaum.
Watson, D., & Clark, L. (1984). Positive and negative affectivity: The
disposition to experience aversive emotional states. Psychological
Bulletin, 96, 465-490.
Watson, D., & Permebaker, J. (1989). Health complaints, stress, and
distress: Exploring the central role of negative affectivity. Psychological Review, 96, 234-254.
Wegner, D. (1994). Ironic process of mental control. Psychological
Review, 101, 34-52.
Wiebe, D., & Smith, T. (1997). Personality and health progress and
problems in psychosomatics. Handbook of personality psychology
(pp. 891-918). San Diego, CA: Academic Press.
Williams, G., Deci, E. L., & Ryan, R. M. (1998). Building health-care
partnerships by supporting autonomy: Promoting maintained behavior
change and positive health outcomes. In P. Hinton-Walker, A. Suchman, & R. Botehlo (Eds.), Partnerships, power and process: Transforming health care delivery (pp. 67-88). Rochester, NY: University
of Rochester Press.
Williams, G., Freedman, Z., & Deci, E. (1996). Promoting motivation
for diabetics' self-regulation of HgbAlc. Diabetes, 45, 13.
Zautra, A., Finch, J., Reich, J., & Guarnaccia, C. (1991). Predicting the
everyday life events of older adults. Journal of Personality, 59, 507538.
Zirkel, S., & Cantor, N. (1990). Personal construal of life tasks: Those
who struggle for independence. Journal of Personality and Social
Psychology, 58, 172-185.
Received September 2, 1997
Revision received January 18, 1998
Accepted June 15, 1998 •