Journal of Personality and Social Psychology Copyright 1996 by the American Psychological Association, Inc.

1996, Vol. 70, No. 3,461-475 0022-3514/96/S3.00

Approach and Avoidance Achievement Goals and Intrinsic Motivation:

A Mediational Analysis
Andrew J. Elliot Judith M. Harackiewicz
University of Rochester University of Wisconsin—Madison

Most contemporary achievement goal conceptualizations consist of a performance goal versus mas-
tery goal dichotomy. The present research offers an alternative framework by partitioning the per-
formance goal orientation into independent approach and avoidance motivational orientations. Two
experiments investigated the predictive utility of the proposed approach-avoidance achievement
goal conceptualization in the intrinsic motivation domain. Results from both experiments supported
the proposed framework; only performance goals grounded in the avoidance of failure undermined
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

intrinsic motivation. Task involvement was validated as a mediator of the observed effects on intrin-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

sic motivation. Ramifications for the achievement goal approach to achievement motivation and
future research avenues are discussed.

Achievement motivation theorists focus their research atten- Atkinson (1957) drew heavily on the work of both McClelland
tion on a particular class of behaviors, those involving compe- and Lewin in formulating need achievement theory, a mathe-
tence. Individuals may aspire to attain competence or may matical framework that designated the desire to attain success
strive to avoid incompetence, and this approach-avoidance dis- and the desire to avoid failure (construed as both motive dispo-
tinction was explicitly incorporated into the earliest achieve- sitions and resultant achievement tendencies) as important de-
ment motivation conceptualizations. Two independent motiva- terminants of achievement behavior.
tional orientations, the desire for success and the desire to avoid In the late 1970s and early 1980s, Carol Dweck, John Nicholls,
failure, were identified by Lewin and colleagues as critical and others introduced an achievement goal approach to achieve-
determinants of aspiration behavior (Hoppe, cited in Lewin, ment motivation (Dweck & Bempechat, 1983; Maehr & Nicholls,
Dembo, Festinger, & Sears, 1944). In his nascent achievement 1980; Nicholls, 1979,1984). These theorists defined achievement
motivation theory, McClelland (1951) proposed that "there are goal as the reason for or purpose of competence-relevant activity
at least two kinds of achievement motivation, one of which ap- (cf. Maehr, 1989). Initially, achievement goal theorists followed
pears to be oriented around avoiding failure and the other the lead of Lewin, McClelland, and Atkinson in incorporating the
around the more positive goal of attaining success" (p. 202). distinction between approach and avoidance motivation into their
frameworks. Three types of achievement goals were posited
(Dweck & Elliott, 1983; Nicholls, 1984): a learning or task in-
volvement goal focused on the development of competence and
Andrew J. Elliot, Department of Psychology, University of Rochester;
task mastery (an approach orientation), a performance or ego in-
Judith M. Harackiewicz, Department of Psychology, University of Wis-
consin—Madison. volvement goal directed toward attaining favorable judgments of
This research was supported by a Dissertation Research Award from competence (also an approach orientation), and a performance
the American Psychological Association and a Berkowitz Dissertation or ego involvement goal aimed at avoiding unfavorable judgments
Award granted by the Department of Psychology at the University of of competence (an avoidance orientation). Although these initial
Wisconsin—Madison. Thanks are extended to Patricia Devine, Ran- models utilized the approach-avoidance distinction, the concept
dall Dunham, Dacher Keltner, and Carolin Showers for their service on of independent approach and avoidance goal orientations received
the dissertation committee. This research was also supported by a grant little theoretical and no empirical attention and was soon over-
from the Vilas Associates program at the University of Wisconsin— looked. Dweck (1986), for instance, shifted to a performance-
Madison. learning goal dichotomy with the approach and avoidance com-
Additional gratitude is expressed to those who assisted in the process ponents of the performance goal collapsed together into a unitary
of data collection: Kenneth Barron, Suzanne Carter, Marcy Church,
orientation. Nicholls (Nicholls, Patashnick, Cheung, Thorkildsen,
Rachel Dean, Roberta Deppe, Brooke Gallagher, Leonard Gicas,
Heather Gore, Elizabeth Grille, Kristen Kolodzik, Michael Krause, & Lauer, 1989) appeared to abandon the approach-avoidance dis-
Phyllis Lee, Allyson Mease, Laura Moeller, Katie Mrzack, Christine Ng, tinction altogether, characterizing his ego and task orientations as
Brian Needle, Jodi Ritter, Jeannine Sayer, Eric Sobel, Mikko Sperber, "two forms of approach motivation" (p. 188).
Jessica Stein, and Randall Young. Kenneth Baron, Keith Campbell, Achievement goal theory is, at present, the predominant ap-
Marcy Church, Edward Deci, Richard Ryan, Kennon Sheldon, and
proach to the analysis of achievement motivation, and most
Miron Zuckerman provided helpful comments on earlier versions of
this article. contemporary achievement theorists proffer achievement goal
Correspondence concerning this article should be addressed to An- frameworks that are similar to the revised models of Dweck and
drew J. Elliot, Department of Psychology, University of Rochester, Nicholls in two important ways. First, most theorists posit two
Rochester, New York 14620. primary orientations toward competence. For instance, Ames
461
 462 ELLIOT AND HARACKIEWICZ

(1984) differentiated mastery and ability goals, Roberts (1992) sideration of possible differences between the performance-ap-
distinguished between mastery and competitive goals, and Deci proach and mastery goal orientations). The performance-avoid-
and Ryan (1985) contrasted task and ego involvement. Ames ance goal is conceptualized as an avoidance orientation grounded
and Archer (1987) contended that these achievement goal in self-regulation according to potential negative outcomes. This
frameworks are conceptually similar enough to justify con- form of regulation evokes self-protective processes that interfere
vergence in the form of a mastery goal (learning, mastery, task with or preclude optimal task engagement (e.g., threat construal,
involvement) versus performance goal (performance, ability, sensitivity to failure-relevant information, anxiety-based preoccu-
ego involvement, competitive) distinction (see also Ryan & pation with self-presentational rather than task concerns) and lead
Stiller, 1991). Second, all of the aforementioned theorists either to the helpless set of motivational responses. Thus, the framework
explicitly or implicitly characterized both mastery and perfor- we endorse is fundamentally process oriented in nature; approach
mance goals as approach forms of motivation (Ames, 1992; and avoidance goals are viewed as exerting their differential effects
Meece, Blumenfeld, & Hoyle, 1988; Nicholls et al., 1989), or on achievement behavior by activating divergent sets of motiva-
they failed to consider approach and avoidance as independent tional processes.
motivational tendencies within the performance goal orienta- In essence, the approach-avoidance achievement goal con-
tion (Deci & Ryan, 1985;Dweck, 1986). ceptualization represents an integration of classic and contem-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Achievement goal theorists posit that the type of orientation porary approaches to achievement motivation: Approach mo-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

adopted at the outset of an activity creates a framework for how tivation is partitioned into independent mastery and perfor-
individuals interpret, evaluate, and act on achievement-relevant mance components (as in contemporary achievement goal
information and experience achievement settings (Ames & frameworks), and avoidance is incorporated as an orthogonal
Archer, 1987; Dweck, 1986). The adoption of a mastery goal is motivational tendency (as in the classic formulations of Lewin,
hypothesized to produce a "mastery" motivational pattern char- McClelland, and Atkinson). A comprehensive achievement
acterized by a preference for moderately challenging tasks, persis- goal model will undoubtedly necessitate the incorporation of
tence in the face of failure, a positive affective stance toward learn- additional elements of the classic formulations, such as the con-
ing, and enhanced task enjoyment. A constellation of "helpless" sideration of dispositional variables (hope for success and fear
motivational responses is posited to result from the adoption of a of failure) and the dynamic interplay between approach and
performance goal orientation, as evidenced by a preference for avoidance motivational tendencies; the proposed framework
easy or difficult tasks, effort withdrawal in the face of failure, attri- represents an initial step in the integration process. In the pres-
bution of failure to lack of ability, and decreased task enjoyment. ent research, we sought to investigate the predictive utility of the
Some achievement goal theorists espouse a more complex hypoth- proposed achievement goal conceptualization in the intrinsic
esis, designating perceived competence as a critical moderator of motivation domain.
achievement goal effects (Elliott & Dweck, 1988). Mastery goals
are expected to have a uniform effect across levels of perceived
Achievement Goals and Intrinsic Motivation
competence (they lead to the mastery motivational pattern); per-
formance goals are posited to lead to the mastery pattern for indi- Intrinsic motivation is manifest in the enjoyment of and interest
viduals with high perceived competence, but this same orientation in an activity for its own sake (Lepper, 1981; Ryan, 1992), and
is expected to produce the helpless pattern for those low in per- this fundamentally approach form of motivation (Berlyne, 1960;
ceived competence. Deci, 1975;Harlow, 1953; White, 1959) is identified as an impor-
In the present article, we propose that the conventional achieve- tant component of the achievement goal nomological network
ment goal dichotomy be expanded to incorporate independent ap- (Ames, 1992; Dweck, 1985,1986; Nicholls, 1989). Most achieve-
proach and avoidance components within the performance goal ment goal and intrinsic motivation theorists contend that mastery
orientation. Specifically, we call for a reconsideration of the tri- goals are facilitative of intrinsic motivation and constituent pro-
chotomous variant of achievement goal framework initially prof- cesses, whereas performance goals are posited to have negative
fered (in nascent form) by achievement goal theorists, composed effects (Deci & Ryan, 1990; Heyman & Dweck, 1992; Nicholls,
of a mastery goal and two performance goals, one directed toward 1989). That is, mastery goals are said to promote intrinsic moti-
the demonstration of competence and the other aimed at avoiding vation by fostering perceptions of challenge, encouraging task in-
the demonstration of incompetence. In the following, these three volvement, generating excitement, and supporting self-determina-
goal types are referred to as mastery, performance-approach, and tion, whereas performance goals are portrayed as undermining in-
performance-avoidance, respectively. trinsic motivation by instilling perceptions of threat, disrupting
We posit that performance-approach and mastery goals both task involvement, and eliciting anxiety and evaluative pressure.
represent approach orientations grounded in self-regulation ac- Other theorists, however, contend that the deleterious effects of
cording to potential positive outcomes (the attainment of norma- performance goals on intrinsic motivation should be manifest only
tive competence and task mastery, respectively). These approach at low levels of perceived competence (cf. Butler, 1992).
forms of regulation commonly engender a host of affective and An alternative set of predictions may be derived from the
perceptual-cognitive processes that facilitate optimal task engage- approach-avoidance achievement goal framework. Both per-
ment (e.g., challenge construal, sensitivity to success-relevant in- formance-approach and mastery goals are focused on attain-
formation, cognitive and affective immersion in the activity; De- ing competence, and these approach orientations commonly
preeuw, 1992; Folkman& Lazarus, 1985;Wegner, 1994) and lead engender a functionally equivalent set of processes that facili-
to the mastery set of motivational responses delineated by achieve- tate optimal task engagement and foster intrinsic motivation.
ment goal theorists (see the General Discussion section for a con- Specifically, in a performance-approach or mastery orienta-
 APPROACH AND AVOIDANCE 463
tion, individuals perceive the achievement setting as a chal- only; thus, participants could demonstrate high but not low
lenge, and this construal is likely to generate excitement, ability in the former case and low but not high ability in the
encourage affective and cognitive investment, facilitate con- latter. In Experiment 2, a more subtle manipulation simply
centration and task absorption, and orient the individual to- focused participants' attention on the possibility of perform-
ward the presence of success-relevant and mastery-relevant in- ing well (performance-approach) or poorly (performance-
formation, processes hypothesized to facilitate intrinsic moti- avoidance) on an achievement task. Both experiments also in-
vation. The performance-avoidance goal, on the other hand, cluded a mastery goal condition that focused participants'
is focused on avoiding incompetence, and this avoidance ori- attention on the task itself and established task-based perfor-
entation is viewed as evoking processes that are antithetical to mance referents.
the very nature of the intrinsic motivation construct. Specifi- We predicted that a comparison of the omnibus perfor-
cally, in a performance-avoidance orientation, individuals mance goal (collapsed across approach and avoidance
construe the achievement setting as a threat and may therefore conditions) and the mastery goal would yield null results on
try to escape the situation if such an option is readily available. intrinsic motivation; only the performance-avoidance goal
Barring physical escape, the prospect of potential failure is was predicted to produce an undermining effect. Given the
likely to elicit anxiety, encourage self-protective withdrawal of constraints of the experimental setting (e.g., participation
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

affective and cognitive resources, disrupt concentration and norms and knowledge of imminent performance feedback),
This document is copyrighted by the American Psychological Association or one of its allied publishers.

task involvement, and orient the individual toward the pres- the possibility of "passive avoidance" in the form of physical
ence of failure-relevant information, processes hypothesized or mental withdrawal from the session was greatly reduced in
to undermine intrinsic motivation. Thus, we posit that ap- the present experiments. Thus, performance-avoidance par-
proach and avoidance forms of motivational regulation acti- ticipants were expected to exert as much effort as their coun-
vate diametrically divergent sets of intrinsic motivation pro- terparts in the other conditions and to perform equally well
cesses, and it is on this basis that we view the approach-avoid- while trying to avoid failure (cf. Atkinson, 1957; Birney, Bur-
ance distinction as a more powerful discriminator of intrinsic dick, & Teevan, 1969; Geen, 1987). This "active" failure
motivation effects than the performance-mastery distinction avoidance was predicted to exact its cost on intrinsic motiva-
per se. tion, with performance-avoidance participants displaying less
In a recent meta-analytic review of the achievement goal-in- free-choice persistence and subsequent enjoyment of the target
trinsic motivation literature, Elliot (1994, 1995) obtained evi- activity than those in either of the approach conditions. The
dence supporting the predictive utility of the proposed trichoto- approach forms of motivation were predicted to have an equiv-
mous framework relative to the prevailing dichotomous frame- alent effect on intrinsic motivation. Experiment 1 also in-
work. Elliot found that less than half of the published experiments cluded a neutral performance goal manipulation in which nei-
investigating the achievement goal-intrinsic motivation relation- ther success nor failure was highlighted. Participants in this
ship yielded data congruent with the prevailing hypothesis that condition were presumed to adopt an approach or avoidance
performance goals are deleterious to intrinsic motivation. In con- orientation as a function of a multitude of (unmeasured) char-
trast, when each of the performance goal manipulations was clas- acteristic propensities (e.g., achievement orientation pessi-
sified according to the approach-avoidance distinction, more than mism) and to therefore manifest intrinsic motivation midway
90% of the experiments yielded data congruent with hypotheses between that evidenced in the approach and avoidance condi-
generated from the approach-avoidance achievement goal frame- tions. Perceived competence was not expected to moderate
work. Experimental manipulations presumed to elicit a perfor- any of the predicted effects.
mance-approach orientation led to intrinsic motivation compa- In addition to examining the utility of the approach-avoid-
rable to that of mastery goal or neutral control conditions, whereas ance distinction, the present experiments sought to test task in-
manipulations classified as performance-avoidance undermined volvement—the degree to which an individual concentrates on
intrinsic motivation. or becomes cognitively immersed in an activity—as a mediator
of the proposed direct effects. Of the numerous hypothesized
The Present Research mediators of the achievement goal-intrinsic motivation rela-
tionship, task involvement would seem an optimal candidate to
The approach-avoidance achievement goal conceptualiza- account for the proposed inimical effect of the performance-
tion appears to fare quite nicely as a post hoc interpretive avoidance orientation on intrinsic motivation. Test anxiety and
framework for the extant achievement goal-intrinsic motiva- fear of failure researchers alike have documented the propensity
tion data, but none of the experiments in the existing corpus for test anxious and failure avoidant individuals to experience
represent a direct test of the proposed model. We conducted various forms of task distraction (cognitive interference, task-
two intrinsic motivation experiments with the aim of di- irrelevant thinking, mind wandering, self-preoccupation) in
rectly testing the predictive utility of the approach-avoidance evaluative achievement settings (Jerusalem, Liepmann, & Her-
achievement goal conceptualization. Performance-approach mann, 1985; Sarason, 1984; Wine, 1982). Intrinsic motivation
and performance-avoidance motivational orientations were theorists posit a strong positive relationship between task in-
manipulated in the two experiments by establishing a norma- volvement and intrinsic motivation (Csikszentmihalyi, 1990;
tive referent for performance evaluation and differentially Harackiewicz & Sansone, 1991), and some have proceeded to
highlighting potential achievement outcomes. In Experi- validate task involvement as a mediator of the effects of external
ment 1, the target task was described as diagnostic of success constraints on intrinsic motivation (Elliot & Harackiewicz,
(performance-approach) or failure (performance-avoidance) 1994). We predicted that the instantiation of a performance-
 464 ELLIOT AND HARACKIEWICZ

avoidance orientation would disrupt task involvement, which, Participants in the performance-avoidance condition were informed:
in turn, would reduce intrinsic motivation. In contrast, the ma-
In our previous work, we have found that most UW students are
nipulation of either approach orientation was predicted to fos- fairly comparable in their ability to solve Nina puzzles, but some
ter task involvement and subsequent intrinsic motivation. students stand out because they do quite poorly on the puzzles.
This session will give you the opportunity to demonstrate that you
Experiment 1 are not a poor puzzle solver.

Method Thus, performance-approach participants were protected from the

possibility of failure, whereas failure was the only diagnostic outcome
Overview possible for performance-avoidance participants. Performance-neutral
participants were simply informed, "This session will give you the op-
Experiment 1 was designed to investigate the effects of performance portunity to demonstrate your level of puzzle solving." After the initial
goal diagnosticity on intrinsic motivation for hidden word puzzles paragraph, participants in each of the three performance goal condi-
called "Nina puzzles." The object of the activity is tofindand circle the tions received the following information:
word Nina, which is hidden a number of times throughout a drawing.
Nina puzzles have been used in previous intrinsic motivation research, In today's session, you will be solving four different Nina puzzles.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and pilot testing has revealed that university undergraduates desire to You will be given 90 seconds per puzzle to find the hidden Ninas.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

perform competently at the activity andfindit enjoyable (Harackiewicz The number of hidden Ninas is different for each puzzle, but none
& Elliot, 1993). Participants solved four Nina puzzles in one of four of the puzzles has more than 15 Ninas hidden in it. When you have
experimental contexts: a performance goal with success diagnostic completed the four puzzles, you will be provided with information
(performance-approach), a performance goal with failure diagnostic regarding how you did compared to other UW students.
(performance-avoidance), a performance goal with no diagnosticity
information provided (performance neutral), or a mastery goal. On Participants in the mastery condition read the following (see Koestner
completion of the puzzles, participants in all four conditions received & Zuckerman, 1994, for a similar mastery goal manipulation):
positive task-based and normative feedback. Process measures were col-
lected before, at the midpoint of, and at the conclusion of the puzzle Our research group studies game playing and puzzle solving and
solving session. A behavioral indicator of intrinsic motivation was ob- the focus of today's session is on hiddenfigurepuzzles. The purpose
tained during a free-choice period; a self-report measure of puzzle en- of this project is to collect data on college students' reactions to
joyment was collected immediately thereafter. hidden figure puzzles—specifically, our Nina puzzles. In today's
session, you will be solving four different Nina puzzles. You will be
given 90 seconds per puzzle tofindthe hidden Ninas. The number
Participants of hidden Ninas is different for each puzzle, but none of the puzzles
has more than 15 Ninas hidden in it. When you have completed
Thirty male and 54 female university undergraduates were randomly the four puzzles, you will be provided with information regarding
assigned to one of the four experimental conditions. The participants the percentage of the total hidden Ninas that you found in today's
received extra course credit for their participation. session.

On receipt of the manipulation, participants were asked to indicate

Procedure
the purpose of the experiment (as a manipulation check).1 Participants
On arrival at the experimental laboratory, participants were intro- also completed a question regarding their perception of the difficulty of
duced to the concept of Nina puzzle solving and were provided with a the Nina task to assess their perceptions of competence relative to task
sample puzzle to familiarize themselves with the activity. When partic- demands, and they reported their affect as they anticipated the puzzle
ipants completed the sample puzzle, they were given a folder that con- solving session. After completing these items, participants proceeded to
tained the manipulation. The experimenter was unaware of experimen- solve the four Nina puzzles. As a means of minimizing participant-
tal condition throughout the session and was not knowledgeable of the experimenter interaction, tape-recorded instructions guided partici-
hypotheses being tested. pants through the puzzle solving period. Between Puzzles 2 and 3, the
Participants in the three performance goal conditions read the follow- tape instructed participants to complete a questionnaire that contained
ing (see Harackiewicz & Elliot, 1993, for a similar performance goal items relevant to task involvement, as well as a question concerning
manipulation): participants' perceptions of competence at the midpoint of task engage-

Our research group studies game playing and puzzle solving and
1
the focus of today's session is on hiddenfigurepuzzles. The purpose Coding of these open-ended responses revealed that 94% of the par-
of this project is to compare college students to one another in their ticipants correctly stated the purpose of the experiment (i.e., perfor-
ability to solve hidden figure puzzles—specifically, our Nina mance-approach and performance-avoidance goal participants made
puzzles. explicit mention of normative comparison, whereas mastery goal par-
ticipants made reference to finding Ninas and puzzle solving without
The remaining information in the first paragraph of the manipulation referring to normative comparison). At the end of the experiment, par-
varied as a function of performance goal condition. Performance-ap- ticipants in the performance-approach and performance-avoidance
proach participants were informed: conditions were also asked, "What were you told about our previous
work?" as a further check on the diagnosticity element of the manipu-
In our previous work, we have found that most UW [University lation (participants indicated their response by checking "Some stu-
of Wisconsin—Madison] students are fairly comparable in their dents stand out because they do quite poorly on Nina puzzles" or "Some
ability to solve Nina puzzles, but some students stand out because students stand out because they do quite well on the puzzles"). All but 2
they do quite well on the puzzles. This session will give you the participants, both in the performance-avoidance condition, responded
opportunity to demonstrate that you are a good puzzle solver. correctly to this query.
 APPROACH AND AVOIDANCE 465
ment. On completion of the four puzzles, participants were given a filler Actual Performance was obtained by summing the number of hidden
task while the experimenter ostensibly scored their puzzles. All partici- Ninas found in each of the four puzzles. Participants were also asked to
pants then received an "information form" indicating that they had rate their degree of Effort Expenditure ("I put a lot of effort into solving
found 80% of all the Ninas hidden in the puzzles and that this percent- the Nina puzzles") on a 7-point scale ranging from strongly disagree
age represented "good puzzle solving compared to other UW students" (1) to strongly agree (7).
(80 was the average percentage that 56 pilot participants designated as
"good compared to other U W students").
Results
After participants read the information form, the experimenter in-
formed them that they were finished with the Nina puzzle activity. The Overview
experimenter then glanced at his or her watch and hurriedly announced
that he or she would have to leave the room to check on another partic- Sequential simultaneous regression analyses were conducted
ipant. Participants were also told that they could do whatever they to investigate the effect of the predictor variables on the behav-
wanted during the experimenter's absence, including solve extra puz- ioral (Time) and self-report (Enjoy) indicators of intrinsic mo-
zles, play with any of the toys in the room, or read magazines. During tivation and to test Task Involvement as a mediator of the direct
this free-choice period, participants' behavior was monitored, via a con- effects observed. Following the guidelines of Judd and Kenny
cealed video camera, by an observer situated in an adjoining room. The
(1981), each outcome measure was first regressed on the pre-
experimenter returned after 5 min and presented participants with a
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

final questionnaire consisting of puzzle enjoyment, perceived compe- dictor variables to test for direct effects. Once direct effects had
This document is copyrighted by the American Psychological Association or one of its allied publishers.

tence, and effort expenditure items. On completion of the final ques- been documented, the mediator variable was regressed on the
tionnaire, participants were debriefed and dismissed. predictor variables to examine thefirstlink in the proposed me-
diational sequence. Finally, the link between the mediator vari-
able and each outcome measure was tested by regressing each
Measures outcome measure on the mediator variable with the predictor
variables controlled.
Intrinsic motivation. Both behavioral and self-report indicators of A set of orthogonal contrasts (henceforth referred to as the
intrinsic motivation were obtained in the experiment. An assortment of
Basic model) tested the primary hypotheses: The Performance
extra Nina puzzles and Highlights puzzles (hidden figure puzzles sim-
ilar to Nina puzzles; see Koestner, Zuckerman, & Koestner, 1987) was Goal contrast tested whether the establishment of a perfor-
available on the subject desk during the free-choice period, and the ob- mance goal context had a negative effect relative to a mastery
server recorded the amount of time that participants chose to puzzle orientation (performance goal conditions — 1; mastery +3),
solve during the experimenter's absence (Time). Participants' ratings and the Approach-avoidance contrast compared the perfor-
on three items on thefinalquestionnaire ("I enjoy doing Nina puzzles mance-approach condition (+1) and the performance-avoid-
very much," "I think that doing Nina puzzles is boring" [reversed], and ance condition (— 1). When analyses with the Basic model re-
"Nina puzzles are fun") were averaged to form the self-report index of vealed a significant Approach-avoidance effect, ancillary
enjoyment (Enjoy). Ratings were made on 7-point scales ranging from planned comparisons sought to anchor the performance-ap-
strongly disagree (1) to strongly agree (1). Similar indexes have been proach and performance-avoidance conditions to the mastery
used in previous intrinsic motivation research (Elliot & Harackiewicz, group: Approach-mastery compared the performance-ap-
1994; Harackiewicz & Elliot, 1993).
proach condition ( + 1) and the mastery condition ( - 1 ) , and
Process and performance measures. Before the puzzle solving ses-
sion, participants were provided an Anticipated Difficulty item: "I Avoidance-mastery compared the performance-avoidance
think that solving these Nina puzzles will be difficult." Participants re- (— 1) and mastery groups (+1). These supplementary contrasts
sponded to this item on a 7-point scale ranging from strongly disagree are nonorthogonal to the Basic model but represent protected
(1) to strongly agree (7). Anticipatory affect was assessed with a stem, planned comparisons (conceptually analogous to Fisher's least
"As I anticipate doing these Nina puzzles I feel. . . ," and the following significant difference tests; Howell, 1987).2
items: eager, challenged, nervous, and worried. Participants' responses
(1 = strongly disagree, 7 = strongly agree) on the first two items were
averaged to form a Challenge Appraisal index; their responses to the last Preliminary Analyses
two items were averaged to form a Threat Appraisal index (see Folkman The regression of Challenge Appraisal on the Basic model
& Lazarus, 1985, for similar appraisal indexes). Between Puzzles 2 and revealed a significant Approach-avoidance effect, F(l, 81) =
3, participants completed a questionnaire that contained a Mid-puzzles
5.80, p < .05 (/3 = .25). Participants in the performance-ap-
Perceived Competence (Mid PC) item—"How do you think you did on
the first two Nina puzzles?" (1 = very poorly, 7 = very well)—and a proach condition were more likely to appraise the puzzle solv-
Task Involvement index. The Task Involvement index began with the ing session as a positive challenge (M = 5.12) than participants
following stem: "While solving thefirsttwo Nina puzzles, I. . .." Par-
ticipants Task Involvement score consisted of the average of the follow-
2
ing items: "was totally absorbed in the puzzles," "lost track of time," In both experiments, gender and all possible Gender X Contrast in-
and "concentrated onfindingthe hidden Ninas." Participants indicated teractions were included in preliminary analyses with all outcome mea-
their responses on 7-point scales ranging from strongly disagree (1) to sures. Gender significantly interacted with Performance Goal, Ap-
strongly agree (7). Similar task involvement indexes have been used in proach-mastery, and Avoidance-mastery in predicting Time in Exper-
previous research on intrinsic motivation processes (Elliot & Harackie- iment 1, and there were significant main effects of gender on Mid and
wicz, 1994; Harackiewicz & Elliot, 1993). Post PC. In Experiment 2, there was a main effect of gender on Enjoy.
After the puzzle solving session, participants were presented with a Inclusion of these gender variables into their respective regression
Post-puzzles Perceived Competence (Post PC) item: "How do you equations did not change any of the effects reported (other than making
think you did on the four Nina puzzles today?" They responded on a 7- them stronger). Readers interested in more detailed information on
point scale ranging from very poorly (1) to very well (7). An index of these gender effects are encouraged to contact Andrew J. Elliot.
 466 ELLIOT AND HARACKJEWICZ

in the Performance-avoidance condition (M = 4.38). Ancillary proach condition reported greater enjoyment of the puzzles
planned comparisons with Challenge Appraisal yielded a sig- than participants in the performance-avoidance condition. An-
nificant Avoidance-mastery effect, F(l, 82) = 8.79, p < .01 cillary planned comparisons revealed a significant Avoidance-
(/3 = .31), indicating that performance-avoidance participants mastery effect for Enjoy, F( 1, 82) = 6.67, p < .05 (0 = .27),
were less likely to appraise the puzzle solving session as a posi- indicating that participants in the performance-avoidance con-
tive challenge than their mastery condition counterparts (M = dition reported enjoying Nina puzzles less than those in the
5.29). No other significant effects were obtained with the ap- mastery condition. No other effects were significant for Enjoy.
praisal indexes. Regressing Actual Performance on the Basic Similar to the results with the behavioral measure, the pattern
model failed to yield any significant effects. These null results of Enjoy means across experimental conditions (see Table 1)
were expected (and desired) given that the puzzles selected for suggests that the provision of a performance goal undermined
use in the present experiments were those that produced the enjoyment of the puzzles only when the performance goal was
least degree of performance variability in pilot testing (with 20 diagnostic of failure.
university undergraduates) with 11 different Nina puzzles.
Effort Expenditure also failed to yield significant effects when Direct Effects From the Predictor Variables to the
regressed on the Basic model. The Pearson product-moment Mediator Variable
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

correlation between Time and Enjoy was .49 (p < .001), and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

this relationship did not significantly differ between experimen- Regressing Task Involvement on the Basic model yielded a
tal conditions. significant effect of Approach-avoidance, F( 1,81) = 5.04, p <
.05 (/3 = .24), indicating that performance-approach partici-
Direct Effects From the Predictor Variables to the pants reported greater task involvement than their perfor-
mance-avoidance counterparts. Ancillary planned compari-
Outcome Measures
sons with Task Involvement revealed a significant Avoidance-
Table 1 displays the means for Time and Enjoy by experi- mastery effect, F(l, 82) = 4.50, p < .05 (/8 = .23), such that
mental condition. Regressing Time on the Basic model revealed performance-avoidance participants reported less Task In-
a significant effect for Approach-avoidance, F(l, 81) = 4.24, volvement than those in the mastery condition. No other effects
p < .05 (0 = .22). Participants in the performance-approach were significant for Task Involvement. The pattern of means in
condition engaged in more puzzle solving during the free-choice Table 1 is highly comparable to that obtained for the two out-
period than their performance-avoidance counterparts. Ancil- come measures and indicates that only when a performance
lary planned comparisons with Time revealed a significant goal was diagnostic of failure did it reduce task involvement.
effect for Avoidance-mastery, F( 1, 82) = 5.52, p < .05 (0 =
.25), indicating that the provision of a performance-avoidance Mediation Analyses: From the Mediator Variable to the
goal undermined intrinsic motivation relative to the mastery
condition. No other effects were significant for Time. Consider-
Outcome Measures
ation of the overall pattern of means in Table 1 suggests that the Mediation was tested by regressing each intrinsic motivation
provision of a performance goal maintained intrinsic motiva- measure on the Basic model with Task Involvement inserted
tion for all participants, except when the performance goal was into the equation (henceforth referred to as the Basic mediation
diagnostic of failure, in which case the performance goal un- model). Regressing Time on the Basic mediation model yielded
dermined intrinsic motivation. a significant effect for Task Involvement, F(l,80) = 8.91,p<
The regression of Enjoy on the Basic model yielded a signifi- .005 (/? = .32), indicating that greater task involvement led to
cant effect of Approach-avoidance, F( 1,81) = 10.77, p < .005 more free-choice puzzle solving. The direct effect for Ap-
(/3 = .34), indicating that participants in the performance-ap- proach-avoidance no longer attained significance with Task In-

Table 1
Means for Time, Enjoy, and Task Involvement by Experimental Condition: Experiment 1
Experimental condition
Outcome Performance- Performance- Performance-
measure approach avoidance neutral Mastery
Time 169.86. 89.86b 139.19ab 181.14.
Enjoy 5.83. 4.75b 5.17ab 5.62,
Task involvement 5.24. 4.63b 5.10.b 5.21.
Note. Within each dependent measure, means not sharing common subscripts are significantly different
from each other (p < .05 at minimum; Fisher's least significant difference test). Time values ranged from 0
s (did not engage in any puzzle solving during the free-choice period) to 300 s (solved puzzles during the
entire free-choice period). Scores on Enjoy had a possible range of 1 (low intrinsic motivation) to 7 (high
intrinsic motivation). Task Involvement values had a possible range of 1 (low involvement) to 7 (high
involvement). Standard deviations were 129.32, 1.13, and 0.89 for Time, Enjoy, and Task Involvement,
respectively.
 APPROACH AND AVOIDANCE 467

Time
.21

Task
Approach-Avoidance
Involvement

Enjoy

Figure 1. Path coefficients for mediation: Experiment 1. All paths indicate significant effects (p < .05 at
minimum).

volvement in the equation, and the decrease in the beta coeffi- ization. Performance goals in general did not undermine intrin-
cient for this effect (from .22 to. 14) provides evidence that Task sic motivation relative to the mastery goal; only the perfor-
Involvement partially mediated the direct effect of Approach- mance goal directed toward avoiding incompetence produced
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

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

negative effects. Performance-avoidance participants tried as

The regression of Enjoyment on the Basic mediational model hard and performed as well as those in the other three condi-
yielded a significant effect for Task Involvement, F(l, 80) = tions and were provided with the same positive feedback, but
21.75, p<.0001 ((8 = .45) .indicating that greater task involve- their experience of striving to avoid failure undermined their
ment led to greater enjoyment of the puzzles. The direct effect intrinsic motivation nonetheless. Performance goal partici-
of Approach-avoidance on Enjoy remained significant with pants attempting to achieve a successful outcome, on the other
Task Involvement in the equation, F( 1, 80) = 5.95, p < .05 (/3 hand, displayed intrinsic motivation equivalent to that of par-
= .23); however, the diminution in the beta coefficient for this ticipants in the mastery goal condition and significantly higher
effect (from .34 to .23) suggests that Task Involvement was a than that of participants in the performance-avoidance condi-
partial mediator of the Approach-avoidance direct effect.3 Fig- tion. The provision of a performance goal without diagnosticity
ure 1 illustrates the mediational processes validated in the pre- information did not reduce intrinsic motivation; these partici-
ceding analyses.
3
Mediation of the Avoidance-mastery effects was also investigated
Supplementary Analyses With the Perceived by regressing each outcome measure on the Avoidance-mastery con-
Competence Variables trast with Task Involvement included in the equation (the Avoidance
mediation model). The regression of Time on this model revealed a
No significant effects were obtained when each of the per- significant effect for Task Involvement, F( 1, 81) = 9.60, p < .005 (0 =
ceived competence measures was regressed on the Basic model. .32), and the Avoidance-mastery direct effect no longer attained sig-
A series of regression analyses was conducted to test perceived nificance (the beta coefficient for this effect decreased from .27 to. 16).
competence as a moderator of the achievement goal-intrinsic Regressing Enjoy on this model also yielded a significant Task Involve-
motivation relationship. Each regression tested the effect of one ment effect, F( 1,81) = 24.20, p < .0001 (0 = .47), and the Avoidance-
of the three perceived competence variables and its correspond- mastery direct effect no longer attained significance (the beta coefficient
ing interaction with performance goal, approach-avoidance, or decreased from .27 to. 16). These results validate Task Involvement as
Avoidance-mastery on each measure of intrinsic motivation. a partial mediator of the Avoidance-mastery effects on intrinsic
For instance, Mid PC was tested as a moderator of the effect of motivation.
the Approach-avoidance contrast on free-choice puzzle solving We also tested Challenge Appraisal as an additional mediator of the
observed effects by regressing each outcome measure on both the Basic
by regressing Time on Approach-avoidance, Mid PC, and the and Avoidance mediation models but substituting Challenge Appraisal
Approach-Avoidance X Mid PC interaction product term. Of in the equation for Task Involvement. The regression of Time on the
the 18 interactions tested, only the Avoidance-Mastery X An- Basic mediation model revealed a significant effect for Challenge Ap-
ticipated Difficulty interaction on Enjoy attained significance, praisal, F(\, 80) = 6.59,p < .05 (0 = .28), and the Approach-avoid-
F(l, 80) = 4.68,/><. 05 (j8=-.23). All of the significant effects ance effect no longer attained significance (the beta coefficient de-
reported in the preceding analyses remained significant when creased from .22 to . 15). Regressing Enjoy on this model also yielded a
the perceived competence variables were inserted into the rele- significant effect of Challenge Appraisal, F( 1, 80) = 9.97, p < .01 (0 =
vant regression equations. Thus, these supplementary analyses .33). The direct effect of Avoidance-mastery on Enjoy remained sig-
yielded little evidence in support of perceived competence as a nificant, but the beta coefficient for this effect decreased from .34 to .26.
moderator of the effects of performance goals on intrinsic mo- The regression of Time on the Avoidance mediation model yielded a
significant effect of Challenge Appraisal, F( 1, 81) = 6.99, p < .01(0 =
tivation and demonstrate that the direct effects obtained in the
.29), and the Avoidance-mastery effect no longer attained significance
present experiment are independent of perceived competence (the beta coefficient decreased from .25 to .16). Regressing Enjoy on
processes. this model revealed a significant Challenge Appraisal effect, F( 1, 81) =
11.41, p < .01 (0 = .36), and the direct effect of Avoidance-mastery
Discussion on Enjoy no longer attained significance (the beta coefficient decreased
from .27 to .16). These results clearly establish Challenge Appraisal
Results from Experiment 1 clearly attest to the predictive uti- as a partial mediator of the direct effects of Approach-avoidance and
lity of the approach-avoidance achievement goal conceptual- Avoidance-mastery on intrinsic motivation, (footnote continues)
 468 ELLIOT AND HARACKIEWICZ

pants displayed levels of interest midway between those of the form of goal manipulation (Prentice & Miller, 1992). Several inves-
performance-approach and performance-avoidance groups. tigators have demonstrated that motivational orientations can be
Thus, these results countervail the prevailing position that per- manipulated by procedures that differentially focus participants' at-
formance goals per se are deleterious to intrinsic motivation; tention on potential success or failure outcomes. These procedures
only performance-avoidance goals are so implicated by the vary in complexity from the cognitive rehearsal of hypothetical suc-
present data. cess or failure scenarios (e.g., Goodhart, 1986) to the mere framing
Task involvement was validated as a mediator of the direct of a single question in terms of positive or negative possibilities (e.g.,
effects observed on intrinsic motivation. Performance-avoid- Higgins, Roney, Crowe, & Hymes, 1994; Tversky & Kahneman,
ance participants reported reduced task involvement relative to 1981). The manipulation used in Experiment 2 resembled the lat-
participants in the performance-approach condition, and this ter more than the former Performance-approach and perfor-
distraction resulted in less free-choice puzzle solving and re- mance-avoidance orientations were differentially instantiated sim-
duced enjoyment of the activity. Participants in the perfor- ply through the provision of a brief sentence highlighting the possi-
mance-avoidance condition also reported less task involvement bility of a success or failure outcome, respectively.
than those in the mastery condition, and this task disruption, in
turn, led to a reduction in intrinsic motivation. Experiment 2
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

A series of analyses provided little support for perceived compe-

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

tence as a moderator of the observed effects; the goals instantiated Method

apparently exerted their influence on intrinsic motivation in a uni-
form manner across levels of perceived competence. Optimistically,
Overview
this suggests that some types of performance goals (approach) can Experiment 2 was designed to replicate the direct and mediational
facilitate intrinsic motivation even for individuals presumably most effects demonstrated in Experiment 1 using a more subtle manipula-
susceptible to their inimical effects (those with low perceived tion of the performance-approach and performance-avoidance ori-
competence); pessimistically, it implies that avoidance strivings re- entations. Participants solved Nina puzzles under one of three experi-
duce intrinsic motivation for those with high as well as low percep- mental contexts: a performance goal with a positive outcome focus
tions of competence. (performance-approach), a performance goal with a negative out-
come focus (performance-avoidance), or a mastery goal. Participants
Performance-approach and performance-avoidance orienta- in all three conditions received positive task-based and normative feed-
tions were instantiated in Experiment 1 by delimiting the availabil- back on completion of the puzzles, and process measures were col-
ity of success-relevant and failure-relevant information. The exper- lected three times over the course of the experimental session. Both
imental control afforded by this diagnosticity manipulation resulted behavioral and self-report indicators of intrinsic motivation were
in a powerful test of our predictions, yet it simultaneously limited used.
the generalizability of the observed effects. Motivational orienta-
tions are commonly elicited by subtle situational cues as well as Participants
strong evaluative constraints, and it remains to be seen whether the
Experiment 1 results can be replicated with a more "minimalist" Forty-seven male and 45 female university undergraduates were ran-
domly assigned to one of the three experimental conditions. The partic-
ipants received extra course credit for their participation.
We conducted afinalset of mediational analyses to test the following
sequential model: independent variables - • Challenge Appraisal -»• Procedure and Measures
Task Involvement -+ Intrinsic Motivation. These analyses were the The procedure used in Experiment 2 was the same as that used in
same as those reported in the preceding paragraph, except that Task Experiment 1, excepting the new performance goal manipulations and
Involvement was included in each regression equation. Regressing Time some minor additions and revisions in the measures. Before the manip-
on the Basic mediation model yielded a significant Task Involvement ulation, pretest enjoyment (Preenjoy) was assessed with the question
effect, F( 1, 79) = 5.03, p < .05 (0 = .25), and the effects of Approach- "At this time, how enjoyable do you think this Nina puzzle activity is?"
avoidance and Challenge Appraisal no longer attained significance (the and participants responded on a 7-point scale ranging from not at all
beta coefficients decreased from .22 to .11 and from .28 to .19, enjoyable (1) to very enjoyable (7). Participants then proceeded to
respectively). The regression of Enjoy on this model revealed a signifi- solve four Nina puzzles in one of three experimental conditions. Those
cant Task Involvement effect, F(l, 79) = 14.31, p < .01 (0 = .38). in both performance goal conditions were informed:
The Challenge Appraisal effect no longer attained significance (the beta
coefficient decreased from .33 to. 19); the Approach-mastery effect re- Our research group studies game playing and puzzle solving and
mained significant, but the beta coefficient decreased from .34 to .20. the focus of today's session is on hiddenfigurepuzzles. The purpose
The regression of Time on the Avoidance mediation model revealed a of this project is to compare college students to one another in their
significant effect for Task Involvement, F(\, 80) = 5.37, p < .05 (0 = ability to solve hidden figure puzzles—specifically, our Nina puz-
.26), and neither the Avoidance-mastery nor the Challenge Appraisal zles. \bur performance in today's session will show your level of
effects remained significant (the beta coefficients decreased from .25 to puzzle solving ability.
.13 and from .29 to .19, respectively). Regressing Enjoy on this model
revealed a significant Task Involvement effect, F( 1,80) = 15.69, p < .01 In addition, performance-approach participants read the following:
(0 = .40). The Avoidance-mastery effect no longer attained significance "For instance, if you find more Ninas than a majority of UW students,
(the beta coefficient decreased from .27 to. 12); the Challenge Appraisal you will demonstrate that you have good puzzle solving ability." Partic-
effect remained significant, but the beta coefficient decreased from .36 ipants in the performance-avoidance condition read: "For instance, if
to .21. These results represent strong evidence in support of the hypoth- you find fewer Ninas than a majority of UW students, you will demon-
esized sequential model. strate that you have poor puzzle solving ability." Thus, the two perfor-
 APPROACH AND AVOIDANCE 469

mance goal orientations focused participants' attention on a potential the Avoidance Orientation measure than those in the mastery condition
performance outcome, and two words distinguished the performance- (M = 2.00). No other effects were significant. These analyses suggest
approach (more and good) from the performance-avoidance (fewer that the Experiment 2 manipulations do indeed prompt differential mo-
and poor) condition. After the initial paragraph, participants in both tivational orientations along the approach-avoidance dimension.
performance goal groups were also informed:

In today's session, you will be solving four different Nina puzzles.

Results
•You will be given 90 seconds per puzzle tofindthe hidden Ninas. Overview
The number of hidden Ninas is different for each puzzle, but none
of the puzzles has more than 15 Ninas hidden in it. When you have The data-analytic approach used in Experiment 2 was the same
completed the four puzzles, you will be provided with information as that used in Experiment 1, with two exceptions. First, the per-
regarding how you did compared to other UW students. formance-neutral group was dropped from each of the contrasts
because this goal orientation was not included in the Experiment
The mastery goal condition was identical to that used in Experiment 1.
2 design. Second, Preenjoy (measured continuously) and the
On receipt of the manipulation, all participants were asked to indi-
cate the purpose of the experiment (as a manipulation check).4 Imme- Preenjoy X Contrast interaction product terms were included in
diately before task engagement, participants were asked, "How impor- preliminary analyses with all outcome measures (these variables
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tant to you is your performance on the Nina puzzles?" (1 = not at all were retained in regression models only when they attained
This document is copyrighted by the American Psychological Association or one of its allied publishers.

important, 7 = very important). They also responded to the following significance).

item: "I care very much about how I do on these Nina puzzles" (1 =
strongly disagree, 7 = strongly agree). The responses to these items were Preliminary Analyses
averaged to form a competence valuation index. As in Experiment 1,
perceived competence was assessed before, at the midpoint of, and at Regressing Actual Performance and Effort Expenditure on
the conclusion of task engagement. However, in addition to Anticipated the Basic model failed to yield any significant effects. In addi-
Difficulty, participants' Anticipated Performance was also assessed by tion, the regression with Competence Valuation revealed no sig-
the question "How do you think you will do on the Nina puzzles today?"
This item was designed to correspond more closely to the Mid and Post
PC measures used in Experiment 1, and participants responded on a 7- 4
Coding of these open-ended responses revealed that 95% of the par-
point scale ranging from very poorly (1) to very well (7). ticipants correctly stated the purpose of the experiment. At the end of
Task involvement is a construct that encompasses both absorption in the experiment, participants were also asked, "Which example were
the activity and a lack of distraction from the task. However, the Exper- you given at the beginning of the experiment?" Participants indicated
iment 1 Task Involvement index assessed only the presence of absorp- their response by checking one of the following: "If youfindmore Ninas
tion; thus, we added items that directly probed for the presence of task than a majority of UW students, you will demonstrate that you have
distraction as well. Accordingly, the Experiment 2 Task Involvement good puzzle solving ability"; "If you find fewer Ninas than a majority
index comprised both absorption (e.g., "I concentrated onfindingthe of UW students, you will demonstrate that you have poor puzzle solving
hidden Ninas") and distraction ("I had trouble focusing my attention ability"; or "neither of the above." Reflecting the relative subtlety of the
on the puzzles," "I felt self-conscious," and "I thought about things un- experimental manipulation, 73% of participants checked the correct
related to the puzzles or the experiment") items. Participants re- response (in comparison with 98% in Experiment 1). Eight participants
sponded on 7-point scales ranging from strongly disagree (1) to strongly in the performance-approach condition checked "neither" rather than
agree (I).5 the "good" option; 7 participants in the performance-avoidance condi-
tion checked the "good" option, and 1 checked "neither" rather than
Pilot Study the "poor" option; and 6 participants in the mastery goal condition
checked the "good" option, and 3 checked the "poor" option rather
We conducted a pilot study for Experiment 2 to ensure that the subtle than the "neither" option.
5
outcome focus manipulations would indeed have their intended effects Following the lead of Reeve and Deci (in press; see also Wild, Enzle,
(the Experiment 1 appraisal items could not be used in this role because & Hawkins, 1992), we also partitioned the behavioral measure of in-
the directness of the questions would have compromised the subtle na- trinsic motivation into two distinct categories: persistence with the ac-
ture of the manipulations). Twenty-one male and 30 female partici- tual target activity used in the experimental session (Nina Puzzles) and
pants underwent the Experiment 2 procedure to the point at which they persistence with a slight variant of the target activity (Highlights
received one of the three manipulations; then they completed a brief Puzzles). Nina and Highlights puzzles were placed in separate piles on
questionnaire and were dismissed. An Approach Orientation measure the subject desk, and observers were trained to independently record
was created by averaging participants' responses (1 = not at all true of extra Nina and Highlights puzzle solving. Reeve and Deci argued that
me, 7 = very true of me) on two items: "My goal for the puzzle solving time spent engaging in a slight variant of the target activity is a more
session is to demonstrate that I have good puzzle solving ability" and "I pure indicator of intrinsic motivation because it is unlikely to be associ-
view the puzzle solving session as a positive challenge." An avoidance ated with internal pressure or frustration engendered by the experimen-
orientation measure consisted of "My goal for the puzzle solving session tal procedure. Both Nina and Highlights puzzles were significantly cor-
is to avoid demonstrating that I have poor puzzle solving ability" and "I related with Enjoyment (p < .05), and these relationships were of ap-
view the puzzle solving session as a threat." proximately equal magnitude (.21 and .23, respectively). Neither
Regression analyses revealed a significant Approach-avoidance effect correlation differed significantly between experimental conditions. This
for Avoidance Orientation, F( 1,48) = 5.00,p < .05 (0 = .30), indicat- pattern of correlations suggests that both Nina and Highlights Puzzles
ing that Performance-avoidance participants scored higher on the assessed intrinsically motivated free-choice behavior. Highlights Puz-
Avoidance Orientation measure (M - 2.1 \) than their performance- zles did prove to be more sensitive to the experimental manipulations
approach counterparts (M = 1.94). The Avoidance-mastery effect was than Nina Puzzles: The results for Highlights Puzzles were virtually
also significant for Avoidance Orientation, F( 1,49) = 4.20, p < .05 (/3 identical to those reported in the text for the omnibus measure (time);
= .28), such that performance-avoidance participants scored higher on null results were obtained for Nina Puzzles.
 470 ELLIOT AND HARACKIEWICZ

nificant differences between experimental conditions. These Direct Effects From the Predictor Variables to the
null results suggest that participants in all conditions were Mediator Variable
equally committed to competent performance on the Nina puz-
zles. The correlation between Time and Enjoy was .30 (p < The regression of Task Involvement on the Basic model re-
.005), and this relationship did not significantly differ between vealed a significant Approach-avoidance effect, F(l, 89) =
experimental conditions. 4.14, p < .05 (/3 = .21). Participants reported greater task in-
volvement in the performance-approach condition than in the
performance-avoidance condition (see Table 2 for means). A
Direct Effects From the Predictor Variables to the significant effect for the Avoidance-mastery planned compari-
Outcome Measures son, F( 1, 90) = 5.99, p< .05 (0 = .25), indicated that perfor-
mance-avoidance participants were less task involved during
Table 2 displays the means for Time and Enjoy by experi- the puzzle solving session than those in the mastery group. No
mental condition. The regression of Time on the Basic model other significant effects were obtained for Task Involvement.
revealed a significant effect for Approach-avoidance, F( 1, 89)
= 4.57, p < .05 (/? = .22). Participants in the performance- Mediation Analyses: From the Mediator Variable to the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

approach condition engaged in more puzzle solving during the Outcome Measures
This document is copyrighted by the American Psychological Association or one of its allied publishers.

free-choice period than participants in the performance-avoid-

ance condition. Ancillary planned comparisons revealed a sig- Mediation was tested by regressing each intrinsic motivation
nificant Avoidance-mastery effect, F(l, 90) = 6.12, p < .05 measure on the Basic model with Task Involvement included
(j3 = .25), indicating that performance-avoidance participants in the regression equation (the Basic mediation model). The
spent less time solving the extra puzzles than their mastery regression of Time on the Basic mediation model yielded a sig-
counterparts. No other significant effects were obtained for nificant Task Involvement effect, F( 1, 88) = 4.02, p < .05 (0 =
Time. Thus, it appears that the provision of a performance goal .21), indicating that greater task involvement led to more free-
undermined intrinsic motivation when the possibility of failure choice puzzle solving. From the decrease in the beta coefficient
was made salient, but intrinsic motivation was maintained for the Approach-avoidance direct effect (from .22 to . 17 with
when the performance goal focused on the possibility of success. Task Involvement in the equation) and the fact that this effect
The regression of Enjoy on the basic model revealed a sig- no longer attained significance, it may be concluded that Task
nificant Preenjoy main effect, F( 1,88) = 32.26,/? < .0001 (0 = Involvement partially mediated the direct effect of Approach-
.51), indicating that those who reported higher enjoyment of avoidance on Time.
the puzzles before the experimental manipulation reported Regressing Enjoy on the Basic mediation model yielded a sig-
greater enjoyment at the conclusion of the experimental ses- nificant effect for Task Involvement, F(l, 87) = 16.15, p <
sion. The Approach-avoidance effect was marginally signifi- .0005 (/3 = .35). Participants reporting greater levels of Task
cant, F( 1,88) = 3.31, p < .07 (0 = . 17), suggesting that perfor- Involvement also reported greater enjoyment of the puzzles.
mance-approach participants enjoyed the puzzles more than The Preenjoy effect remained significant with Task Involve-
those in the performance-avoidance condition. No other effects ment in the regression equation, F( 1, 87) = 37.59, p < .0001
were significant for Enjoy. (/? = .51), but the Approach-avoidance direct effect no longer
approached significance (p > .25). From the decrease in the
beta coefficient for the Approach-avoidance effect (from. 17 to
.09 with Task Involvement in the equation), it is evident that
Table 2 Task Involvement served to partially mediate the direct effect
Means for Time, Enjoy, and Task Involvement by of Approach-avoidance on Enjoyment.6 Figure 2 illustrates the
Experimental Condition: Experiment 2 pattern of mediation demonstrated in the preceding analyses.

Experimental condition
Supplementary Analyses With the Perceived
Performance- Performance- Competence Variables
Outcome measure approach avoidance Mastery
Regressing Post PC on the basic model revealed a significant
Time 135.53. 69.42b 145.39. Preenjoy main effect, F( 1, 88) = 7.81, p < .01 (/S = .28), indi-
Enjoy 5.40, 4.99b 5.05.b cating that participants who enjoyed the puzzles before the ex-
Task involvement 5.43. 5.03b 5.50. perimental session reported greater perceived competence after
Note. For Time and Task Involvement, means not sharing common task engagement. However, no significant experimental effects
subscripts are significantly different from each other (p < .05 at mini-
mum; Fisher's least significant difference test). Scores for Time ranged
from 0 s (did not engage in any puzzle solving during the free-choice 6
Mediation of the Avoidance-mastery effect for Time also was inves-
period) to 300 s (solved puzzles during the entire free-choice period). tigated. Regressing Time on the Avoidance mediation model yielded a
Task Involvement values had a possible range of 1 (low involvement) to
7 (high involvement). For Enjoy, means not sharing common subscripts significant effect for Task Involvement, F( 1, 89) = 4.43, p < .05 (0 =
differ at p < .07. Enjoy values had a possible range of 1 (low intrinsic .22). The Avoidance-mastery contrast no longer attained significance,
motivation) to 7 (high intrinsic motivation), although the values shown and the beta coefficient decreased from .25 to .20, thereby establishing
were adjusted for Preenjoy. Standard deviations were 124.24,0.78, and Task Involvement as a partial mediator of the direct effect of Avoid-
1.02 for Time, Task Involvement, and Enjoy, respectively. ance-mastery on Time.
 APPROACH AND AVOIDANCE 471

Time

.24
Approach-Avoidance
.23

Enjoy

Figure 2. Path coefficients for mediation: Experiment 2. All paths indicate significant effects (p < .05 at
minimum). The direct effect of Preenjoy (0 = .50) was also significant on Enjoy (p < .05), but this effect
was not included in the path model for the sake of presentation clarity.

were obtained when each of the perceived competence variables tion and provide further substantiation of the predictive utility
was regressed on the Basic model. As in the Experiment 1 anal-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

of the approach-avoidance achievement goal conceptualization.

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

yses, a series of regression analyses was conducted to test per- Analyses with the behavioral measure of intrinsic motivation
ceived competence as a moderator of the performance goal- yielded a pattern of effects virtually identical to that demon-
intrinsic motivation relationship. Each regression tested the strated in Experiment 1. Only performance-avoidance partici-
effect of one of the four perceived competence variables and its pants displayed decreasedfree-choicepuzzle solving relative to
corresponding interaction with Performance Goal, Approach- those in the mastery condition; performance-approach partici-
avoidance, or Avoidance-mastery on one measure of intrinsic pants' free-choice puzzle solving was equivalent to that of mas-
motivation. No significant effects were obtained in any of the tery participants and significantly greater than that of their per-
24 analyses conducted. In addition, all of the significant effects formance-avoidance counterparts. Results for self-reported en-
reported in the preceding analyses remained significant, and the joyment partially replicated those obtained in Experiment 1.
marginal effect for Enjoy remained marginally significant. Performance-avoidance participants tended to report less enjoy-
Clearly, the effects documented in the present experiment are ment of the Nina puzzles than those in the performance-ap-
independent of perceived competence processes. proach condition, but they did not report less enjoyment than
mastery goal participants. However, when results from the two
Meta-Analysis ofExperiments 1 and 2 experiments were combined meta-analytically, this Avoidance-
mastery effect did appear to be a reliablefinding(at the p = .06
Given that Experiment 2 included a conceptual replication level), and meta-analyses of the other aforementioned effects at-
of Experiment 1, results from the two experiments were com- test to the robustness of the results across experiments (for all
bined by means of the Stoufer method of adding Z scores effects, p < .005 at minimum). Thus, the data from the two ex-
(Rosenthal, 1978). Any effect that surpassed the conventional periments strongly support the proposition that performance
level of significance (p < .05) in either experiment was analyzed goals aimed at avoiding the demonstration of incompetence un-
via this meta-analytic procedure. Both the Approach-avoid- dermine intrinsic motivation, whereas performance goals di-
ance and Avoidance-mastery effects on Time became highly rected toward the attainment of competence do not.
significant when combined meta-analytically across experi- As in Experiment 1, task involvement was validated as a me-
ments (Z = 2.94, p < .005, and Z = 3.48, p < .0005, diator of the observed effects on intrinsic motivation. Perfor-
respectively). Combining the Enjoy results revealed a highly mance-avoidance participants reported a reduction in task in-
significant effect for Approach-avoidance (Z = 3.57, p < .0005) volvement relative to those in the performance-approach condi-
and a marginally significant effect for Avoidance-mastery (Z = tion, and this distraction subsequently resulted in less free-choice
1.87, p = .06). The meta-analytic effect of Approach-avoid- puzzle solving and reduced enjoyment of the activity. Perfor-
ance on Task Involvement was highly significant (Z = 2.99, p < mance-avoidance participants' reports of task involvement were
.005), as was the effect for Avoidance-mastery (Z = 3.18, p < also lower than those of participants in the mastery condition,
.005). Thus, the direct effects of Approach-avoidance and and this reduction, in turn, led to less free-choice puzzle solving.
Avoidance-mastery proved highly robust across the two exper- These mediational processes correspond nicely to those docu-
iments. Likewise, the relationship between Task Involvement mented in Experiment 1. Also congruent with Experiment 1, all
and intrinsic motivation (the final link in the mediational of the observed effects were independent of perceived compe-
sequence) was highly reliable, as evidenced by the Z scores for tence processes.
the Time (3.53, p < .0005) and Enjoy (5.47, p < .0001)
variables.
General Discussion
Discussion Most achievement goal and intrinsic motivation theorists
Despite the use of a relatively subtle outcome focus manipula- posit the existence of two primary goal orientations—mastery
tion, the results of Experiment 2 largely replicated those obtained and performance—and contend that mastery goals facilitate in-
in Experiment 1. Thesefindingsverify the generalizability of the trinsic motivation, whereas performance goals are inimical in
Experiment 1 effects beyond the strong diagnosticity manipula- their effects. In the present research, we proffered an alternative
 472 ELLIOT AND HARACKIEWICZ

achievement goal conceptualization by partitioning the perfor- unable to "lose themselves" in the task in the same fashion.
mance goal into independent approach and avoidance orienta- Immersion in the activity relatively unencumbered by higher
tions, and we proposed that the deleterious effect of perfor- order concerns facilitated enjoyment of the task; inability to be-
mance goals on intrinsic motivation should be witnessed only come absorbed in the activity resulted in a less enjoyable expe-
for the performance-avoidance goal state. Results from two ex- rience (see Hembree, 1988, and Spielberger & Vagg, 1995, for
periments attested to the predictive utility of the alternative rel- discussions relevent to the more complex relationship between
ative to the prevailing achievement goal framework. Perfor- task involvement and performance). This mediation via task
mance goals focused on avoiding incompetence undermined in- involvement was documented in both experiments, and these
trinsic motivation relative to both a mastery goal and a results represent the first within manuscript replication of me-
performance goal directed toward the attainment of compe- diational processes in the achievement and intrinsic motivation
tence. These latter, approach orientations manifested equiva- literatures (indeed, they represent one of the few existing dem-
lent levels of intrinsic motivation. Results from the two experi- onstrations of either mediation with a behavioral measure of
ments displayed an impressive degree of convergence, and meta- intrinsic motivation or mediation of an achievement goal effect;
analyses attested to the robustness of the observed effects across see also Elliot & Harackiewicz, 1994). It is likely that a number
variants of the approach-avoidance manipulation and indica- of processes in addition to task involvement are relevant to the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tors of intrinsic motivation. relationship between approach-avoidance goal and intrinsic

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

Performance-avoidance participants in the present experi- motivation (e.g., self-determination [Deci & Ryan, 1990] and
ments valued competence as much, exerted as much effort, and perceptual-cognitive sensitivity to success- and failure-relevant
performed as well as their performance-approach and mastery information [Higgins et al., 1994; Wegner, 1994]); subsequent
goal counterparts, but they evidenced a decrease in subsequent research efforts are needed to explore the role of these other
intrinsic motivation nonetheless. Striving to avoid failure and potential mediational mechanisms.
striving to attain success (be it task or normatively referenced) Since the early 1970s, a number of achievement motivation
apparently engendered the same quantity of motivation; it was the theorists have displayed a tendency to use high-low perceived
qualitative nature of the motivation that differed, with important competence as a surrogate for approach-avoidance motivation
intrinsic motivation ramifications. Thus, an avoidance goal may (Kukla, 1972; Meyer, 1987), and those in the achievement mo-
be a "great motivator" in the sense that it can elicit affective invest- tivation tradition have clearly followed this trend (cf. Nicholls,
ment and vigorous action resulting in successful accomplish- 1984). Given this context, the results of the present experi-
ments, but this process of "active avoidance" apparently exacts a ments are noteworthy because they demonstrate effects of ap-
phenomenological cost. The negative effects of avoidance motiva- proach and avoidance motivational orientations that are inde-
tion are not necessarily constrained to the phenomenological pendent of perceived competence processes. An extensive series
realm, however; deleterious consequences have also been docu- of analyses failed to validate perceived competence as a moder-
mented on a variety of achievement behaviors such as persistence ator of the observed effects, and the effects remained significant
in the face of failure, task choice, and patterns of attributions for when the direct and interactive influence of perceived compe-
success and failure (Atkinson & Litwin, 1960; Feather, 1963; tence was statistically controlled. This is not to say that percep-
Heckhausen, Schmalt, & Schneider, 1985). Performance decre- tions of competence never play a role in approach-avoidance
ments as a function of failure avoidance are also sometimes wit- motivational regulation. On the contrary, we believe that per-
nessed, particularly in investigations that (unlike the present ceived competence effects are likely to be revealed in investiga-
experiments) allow maximum performance variability (Anderson tions that conceptualize and test perceived competence as an
& Sauser, 1995; Hembree, 1988), use cognitively demanding tasks antecedent of goal adoption, manipulate success and failure, or
(Birney et al., 1969; Sieber, O'Neill, & Tobias, 1977), and utilize explore the development, rather than the maintenance, of in-
longitudinal designs (Norem & Cantor, 1990; O'Conner, Atkin- trinsic interest (Elliot & Harackiewicz, 1994; Harackiewicz,
son, & Homer, 1966). Sansone, & Manderlink, 1985). Thus, results from the present
Process analyses validated task involvement as a mediator of experiments support the conclusion that approach-avoidance
the deleterious effects of performance-avoidance goals on in- orientations can affect motivational processes independent of
trinsic motivation. Specifically, performance-avoidance goals perceptions of competence; in no way do they call into question
led to reduced task involvement relative to performance-ap- the general relevance or utility of the perceived competence
proach and mastery goals, and this distraction, in turn, resulted construct.
in decreased intrinsic motivation. Performance-approach and One of the most important issues related to perceived com-
mastery goal participants evidenced similar levels of task in- petence that awaits empirical exploration is the stability of the
volvement and subsequent intrinsic motivation. In essence, this performance-approach orientation on receipt of negative feed-
pattern of mediation suggests that the approach forms of moti- back. Feedback was held constant in the present experiments
vation enabled individuals to "drop down" to the activity level (all participants received positive task-based and normative
and become absorbed in the process of task engagement, feedback); variation in success and failure feedback would have
whereas the avoidance orientation disrupted task focus and been necessary to determine whether individuals can maintain
promoted perseveration at the "higher" level of self-concerns a performance-approach goal in the face of failure or whether
(Kuhl, 1985; Vallacher & Wegner, 1987). In colloquial termi- failure would inevitably elicit a performance-avoidance orien-
nology, approach-motivated participants were able to "forget tation (and, presumably, decrease intrinsic motivation). Al-
about themselves" and their reasons for task engagement and though some types of failure experiences would undoubtedly
"dive into the activity"; avoidance-motivated individuals were elicit a performance-avoidance orientation (e.g., repeated fail-
 APPROACH AND AVOIDANCE 473

ure that unequivocally signifies incompetence), the very adop- effect of this goal adoption on competence-relevant affect, cogni-
tion of a performance-approach goal evokes a set of "direc- tion, and behavior. Thesefindingsnot only further the progress of
tional" or biased perceptual-cognitive processes (Dweck & basic psychological research on achievement motivation pro-
Leggett, 1988; see Kunda, 1990) that may serve as bulwarks, cesses, they also have important applied ramifications for the class-
protecting and consequently perpetuating the approach orien- room, the office, and the ball field. However, a limitation of the
tation. These goal-induced processes would probably be bol- research conducted to date is the nearly exclusive consideration
stered and expanded by myriad "normal" social information- of only two distinct motivational orientations: performance and
processing biases (Brown, 1991; Taylor & Brown, 1988) and mastery goals. Dweck and colleagues (Bergen & Dweck, 1989;
the host of post hoc self-protective strategies available to the Henderson & Dweck, 1990) have argued that the prevailing per-
individual (Snyder & Higgins, 1988; Wills, 1981). To the extent formance-mastery distinction is a rudimentary, simplified repre-
that these processes and strategies successfully evade, minimize sentation of achievement motivation and that achievement goal
the impact of, or reconstrue negative information, the perfor- frameworks will eventually need to evolve toward a greater degree
mance-approach orientation is likely to remain intact. of complexity. One evolutionary option was proffered in the pres-
In introducing the approach-avoidance achievement goal ent research: the partitioning of the performance goal orientation
conceptualization, we have focused on the fundamental sim- into independent approach and avoidance components. Two in-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ilarity of performance-approach and mastery goals, comparing trinsic motivation experiments clearly attested to the predictive
This document is copyrighted by the American Psychological Association or one of its allied publishers.

these approach orientations with the performance-avoidance utility of this approach-avoidance distinction, and a high priority
goal state. In so doing, we do not intend to infer that the two for future research is to test the generalizability of this framework
approach orientations always elicit the same motivational pro- to other important variables in the achievement goal nomological
cesses or that they always produce identical achievement out- network. Ironically, approach and avoidance were recognized as
comes. The external evaluation inherent in performance-ap- independent motivational tendencies in the classic need achieve-
proach goals may, in some contexts (e.g., when normative feed- ment theory (Atkinson, 1957), one of the primary theoretical
back is dispensed in a controlling manner), reduce feelings of frameworks that the contemporary achievement goal approach
self-determination and undermine subsequent intrinsic moti- usurped in establishing its current sovereignty. It is our contention
vation (Deci & Ryan, 1985). In other contexts (e.g., perfor- that incorporation of this approach-avoidance distinction would
mance of a monotonous or overlearned activity), however, the afford the achievement goal approach greater theoretical and em-
provision of a performance-approach goal may make compe- pirical precision, thereby further solidifying its present reign.
tence more salient or valued and, consequently, may enhance
intrinsic motivation through the competence valuation process
(Harackiewicz, 1989). Thus, we believe that the effect of References
achievement goals varies as a function of achievement contexts, Ames, C. (1984). Competitive, cooperative, and individualistic goal
and it seems likely that performance-approach and mastery structures: A cognitive-motivational analysis. In C. Ames & R. Ames
goals will reveal a comparable pattern of results in some (Eds.), Research on motivation in education (Vol. 3, pp. 177-207).
contexts and disparate effects in others. New \brk: Academic Press.
Ames, C. (1992). Classrooms: Goals, structures, and student motiva-
On a related note, consideration of the procedures used in tion. Journal ofEducational Psychology, 84, 261-271.
the present experiments suggests that the performance goal Ames, C, & Archer, J. (1987). Mothers' beliefs about the role of ability
contexts established were relatively benign in nature: Evaluative and effort in school learning. Journal ofEducational Psychology, 79,
pressure was minimized (e.g., by limiting the amount of exper- 409-414.
imenter-participant interaction), positive competence pro- Anderson, S., & Sauser, W. (1995). Measurement of test anxiety: An
cesses were maximized (through the provision of positive task- overview. In C. Spielberger & P. Vagg (Eds.), Test anxiety: Theory,
based and normative feedback at task completion), and the per- assessment, and treatment (pp. 15-34). Washington, DC: Taylor &
formance goal manipulations made no reference to the testing Francis.
of valued attributes (e.g., intelligence; see Ryan, 1982). The Atkinson, J. W. (1957). Motivational determinants of risk-taking be-
fact that the performance-avoidance orientation undermined havior. Psychological Review, 6, 359-372.
Atkinson, J. W., & Litwin, G. H. (1960). Achievement motive and test
intrinsic motivation even under such benign conditions seems
anxiety conceived as motive to approach success and motive to avoid
to warrant the unequivocal conclusion that avoidance motiva- failure. In D. McClelland & R. Steele (Eds.), Human motivation: A
tion is deleterious to intrinsic motivation. Conversely, the pro- book of readings (pp. 145-163). Morristown, NJ: General Learning
cedures used did not afford as stringent a test of the compara- Press.
bility of the two approach orientations; thus, the null effects Bergen, R. S., & Dweck, C. S. (1989). The functions of personality the-
obtained in the present experiments warrant less definitive in- ories. In R. Wyer & T. Srull (Eds.), Advances in social cognition: So-
terpretive statements. Further research is needed to probe the cial intelligence and cognitive assessments of personality (pp. 81 -92).
functional comparability of performance-approach and mas- Hillsdale, NJ: Erlbaum.
tery goals and to delineate the effect that various achievement Berlyne, D. N. (1960). Conflict, arousal, and curiosity. New York:
contexts have on this comparability; we suspect that these em- McGraw-Hill.
pirical efforts will reveal the need to incorporate a greater de- Birney, R., Burdick, H., &Teevan, R. (1969). Fear of failure. New York:
Van Nostrand Reinhold.
gree of differentiation between the two approach orientations.
Brown, J. (1991). Accuracy and bias in self-knowledge. In C. Snyder &
Contemporary achievement goal research has yielded a wealth D. Forsyth (Eds.), Handbook of social and clinical psychology: The
of knowledge regarding the qualitatively distinct motivational ori- health perspective(pp. 158-178). New \ork: Pergamon Press.
entations that individuals adopt in achievement settings and the Butler, R. (1992). What young people want to know when: Effects of
 474 ELLIOT AND HARACKIEWICZ

mastery and ability goals on interest in different kinds of social com- (Eds.), Advances in motivation and achievement (Vol. 7, pp. 21-49).
parisons. Journal ofPersonality and Social Psychology, 62, 934-943. Greenwich, CT: JAI Press.
Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experi- Harackiewicz, J., Sansone, C , & Manderlink, G. (1985). Competence,
ence. New Yfork: Harper & Row. achievement orientation, and intrinsic motivation: A process analy-
Deci, E. L. (1975). Intrinsic motivation. New York: Plenum. sis. Journal ofPersonality and Social Psychology, 48,493-508.
Deri, E. L.( & Ryan, R. M. (1985). Intrinsic motivation and self-deter- Harlow, H. F. (1953). Motivation as a factor in the acquisition of new
mination in human behavior. New \brk: Plenum. responses. In Current theory and research in motivation: A sympo-
Deci, E. L., & Ryan, R. M. (1990). A motivational approach to self: sium (pp. 24-49). Lincoln: University of Nebraska Press.
Integration in personality. In R. Dientsbier (Ed.), Nebraska Sympo- Heckhausen, H., Schmalt, H., & Schneider, K. (1985). Achievement
sium on Motivation (Vol. 38, pp. 237-288). Lincoln: University of motivation in perspective (M. Woodruff & R. Wicklund, Trans.). New
Nebraska Press. "York: Academic Press.
Depreeuw, E. (1992). On the fear of failure construct: Active and pas- Hembree, R. (1988). Correlates, causes, effects, and treatment of test
sive test anxious students behave differently. In K. Hagtvet & T. John- anxiety. Review ofEducational Research, 58, in-11.
sen (Eds.), Advances in test anxiety research (Vol. 7, pp. 32-46). Henderson, V. L., & Dweck, C. S. (1990). Motivation and achievement.
Amsterdam: Swets & Zeitlinger. In S. Feldman & G. Elliot (Eds.), At the threshold: The developing
Dweck, C. S. (1985). Intrinsic motivation, perceived control, and self- adolescent (pp. 308-329). Cambridge, MA: Harvard University
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

evaluation maintenance: An achievement goal analysis. In C. Ames & Press.

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

R. Ames (Eds.), Research on motivation in education: The classroom Heyman, G. D., & Dweck, C. S. (1992). Achievement goals and intrin-
milieu (pp. 289-305). New York: Academic Press. sic motivation: Their relation and their role in adaptive motivation.
Dweck, C. S. (1986). Motivational processes affecting learning. Ameri- Motivation and Emotion, 16, 231-247.
can Psychologist, 4], 1040-1048. Higgins, E. T, Roney, C , Crowe, E., & Hymes, C. (1994). Ideal versus
Dweck, C. S., & Bempechat, J. (1983). Children's theories of intelli- ought predilections for approach and avoidance: Distinct self-regula-
gence: Consequences for learning. In S. Paris, G. Olsen, & H. Steven- tory systems. Journal ofPersonality and Social Psychology, 66, 276-
son (Eds.), Learning and motivation in the classroom (pp. 239-256). 286.
Hillsdale, NJ: Erlbaum. Howell, D. C. (1987). Statistical methods for psychology. Boston: Dux-
bury Press.
Dweck, C. S., & Elliott, E. S. (1983). Achievement motivation. In E. M.
Hetherington (Ed.), Socialization, personality, and social develop- Jerusalem, M., Liepmann, D., & Hermann, C. (1985). Test anxiety and
ment (pp. 643-691). New York: Wiley. achievement motivation: An analysis of causal relationships. In H.
Van der Ploeg, R. Schwarzer, & C. Spielberger (Eds.), Advances in
Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to
test anxiety (Vol. 4, pp. 135-138). Lisse, The Netherlands: Swets &
personality. Psychological Review, 95, 256-273.
Zeitlinger.
Elliot, A. J. (1994). Approach and avoidance achievement goals: An
Judd, C. M., & Kenny, D. A. (1981). Process analysis: Estimating me-
intrinsic motivation analysis. Unpublished doctoral dissertation,
diation in treatment evaluations. Evaluation Review, 5, 602-619.
University of Wisconsin-Madison.
Koestner, R., & Zuckerman, M. (1994). Causality orientations, failure
Elliot, A. J. (1995). Reconceptualizing the achievement goal-intrinsic
and achievement. Journal ofPersonality, 62, 321-346.
motivation relationship: An approach/avoidance analysis. Manu-
Koestner, R., Zuckerman, M., & Koestner, J. (1987). Praise, involve-
script in preparation.
ment, and intrinsic motivation. Journal of Personality and Social
Elliot, A. J., & Harackiewicz, J. M. (1994). Goal setting, achievement Psychology, 53, 383-390.
orientation, and intrinsic motivation: A mediational analysis.
Kuhl, J. (1985). Volitional mediators of cognition-behavior consis-
Journal ofPersonality and Social Psychology, 66, 968-980. tency: Self-regulatory processes and action versus state orientation.
Elliott, E. S., & Dweck, C. S. (1988). Goals: An approach to motivation In J. Kuhl & J. Beckmann (Eds.), Action control: From cognition to
and achievement. Journal of Personality and Social Psychology, 54, behavior (pp. 101 -128). New York: Springer-Verlag.
5-12. Kukla, A. (1972). Attributional determinants of achievement-related
Feather, N. T. (1963). Persistence at a difficult task with alternative task behavior. JournalofPersonality and Social Psychology, 21, 166-174.
of intermediate difficulty. Journal ofAbnormal and Social Psychol- Kunda, Z. (1990). The case for motivated reasoning. Psychological Bul-
ogy, 66, 604-609. letin, 108, 480-498.
Folkman, S., & Lazarus, R. (1985). If it changes it must be process: A Lepper, M. R. (1981). Intrinsic and extrinsic motivation in children:
study of emotion and coping during three stages of a college exami- Detrimental effects of superfluous social controls. In W. A. Collins
nation. Journal ofPersonality and Social Psychology, 48, 150-170. (Ed.), Aspects of the development of competence: The Minnesota
Geen, R. G. (1987). Test anxiety and behavioral avoidance. Journal of Symposium on Child Psychology (Vol. 14, pp. 155-214). Hillsdale,
Research in Personality, 21,481-488. NJ: Erlbaum.
Goodhart, D. E. (1986). The effects of positive and negative thinking Lewin, K., Dembo, T., Festinger, L., & Sears, P. S. (1944). Level of
on performance in an achievement situation. Journal of Personality aspiration. In J. McHunt (Ed.), Personality and the behavior disor-
and Social Psychology, 51, 117-124. ders (Vol. 1, pp. 333-378). New York: Ronald Press.
Harackiewicz, J. M. (1989). Performance evaluation and intrinsic mo- Maehr, M. L. (1989). Thoughts about motivation. In C. Ames & R.
tivation processes: The effects of achievement orientation and re- Ames (Eds.), Research on motivation in education: Goals and cogni-
wards. In D. Buss & N. Cantor (Eds.), Personality psychology: Recent tions (Vol. 3, pp. 299-315). New York: Academic Press.
trends and emerging directions (pp. 128-137). New York: Springer- Maehr, M. L., & Nicholls, J. G. (1980). Culture and achievement mo-
Verlag. tivation: A second look. In N. Warren (Ed.), Studies in cross-cultural
Harackiewicz, J. M., & Elliot, A. J. (1993). Achievement goals and psychology (Vol. 2, pp. 221 -267). New York: Academic Press.
intrinsic motivation. Journal of Personality and Social Psychology, McClelland, D. C. (1951). Measuring motivation in phantasy: The
65,904-915. achievement motive. In H. Guetzkow (Ed.), Groups, leadership, and
Harackiewicz, J. M., & Sansone, C. (1991). Goals and intrinsic moti- men (pp. 191 -205). Pittsburgh, PA: Carnegie Press.
vation: You can get there from here. In M. L. Maehr & P. R. Pintrich Meece, J. L., Blumenfeld, P. C, & Hoyle, R. H. (1988). Students' goal
 APPROACH AND AVOIDANCE 475
orientations and cognitive engagement in classroom activities. (Ed.), Nebraska Symposium on Motivation (pp. 1-56). Lincoln:
Journal ofEducational Psychology, 80, 514-523. University of Nebraska Press.
Meyer, W. (1987). Perceived ability and achievement-related behavior. Ryan, R. M., & Stiller, J. (1991). The social contexts of internalization:
In F. Halisch & J. Kuhl (Eds.), Motivation, intention, and volition Parent and teacher influences on autonomy, motivation, and learn-
(pp. 73-86). New York: Springer-Verlag. ing. In M. L. Maehr & P. R. Pintrich (Eds.), Advances in motivation
Nicholls, J. G. (1979). Quality and equality in intellectual develop- and achievement (Vol. 7, pp. 115-149). Greenwich, CT: JAI Press.
ment. American Psychologist, 34, 1071-1084. Sarason, I. G. (1984). Stress, anxiety, and cognitive interference: Reac-
Nicholls, J. G. (1984). Achievement motivation: Conceptions of ability, tions to tests. Journal ofPersonality and Social Psychology, 46, 929-
subjective experience, task choice, and performance. Psychological 938.
Review, 91, 328-346. Sieber, J. E., O'Neill, H. F., & Tobias, S. (1977). Anxiety, learning and
Nicholls, J. G. (1989). The competitive ethos and democratic education. instruction. Hillsdale, NJ: Erlbaum.
Cambridge, MA: Harvard University Press. Snyder, C, & Higgins, C. (1988). Excuses: Their effective role in the
negotiation of reality. Psychological Bulletin, 104, 23-35.
Nicholls, J. G., Patashnick, M., Cheung, P., Thorkildsen, T. A., & Lauer,
Spielberger, C, & Vagg, P. (1995). Test anxiety: A transactional process
J. M. (1989). Can achievement motivation theory succeed with only
model. In C. Spielberger & P. Vagg (Eds.), Test anxiety: Theory, as-
one conception of success? In F. Halisch & J. Van den Beroken (Eds.),
sessment, and treatment (pp. 3-14). Washington, DC: Taylor &
International perspective on achievement motivation (pp. 187-208). Francis.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Lisse, The Netherlands: Swets & Zeitlinger. Taylor, S. E., & Brown, J. (1988). Illusion and well-being: A social psy-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Norem, J. K., & Cantor, N. (1990). Cognitive strategies, coping, and chological perspective on mental health. Psychological Bulletin, 103,
perceptions of competence. In R. Steinberg & J. Kolligian, Jr. (Eds.), 193-210.
Competence considered (pp. 190-204). New Haven, CT: Yale Uni- Tversky, A., & Kahneman, D. (1981). The framing of decisions and the
versity Press. psychology of choice. Science, 211,453-458.
O'Conner, P., Atkinson, J. W., & Horner, M. S. (1966). Motivational Vallacher, R. R., & Wegner, D. M. (1987). What do people think they're
implications of ability grouping in schools. In J. W. Atkinson & N. T. doing? Action identification and human behavior. Psychological Re-
Feather (Eds.), A theory of achievement motivation (pp. 231-248). view, 94, 3-15.
New York: Wiley. Wegner, D. M. (1994). Ironic process of mental control. Psychological
Prentice, D. A., & Miller, D. T. (1992). When small effects are impres- Review, 101, 34-52.
sive. Psychological Bulletin, 112, 160-164. White, R. W. (1959). Motivation reconsidered: The concept of compe-
Reeve, J., & Deci, E. (in press). Elements of the competitive situation tence. Psychological Review, 66, 297-333.
that affect intrinsic motivation. Personality and Social Psychology Wild, T. C, Enzle, M. E., & Hawkins, W. L. (1992). Effects of perceived
Bulletin. extrinsic versus intrinsic teacher motivation on student reactions to
Roberts, G. C. (1992). Motivation in sport and exercise: Conceptual skill acquisition. Personality and Social Psychology Bulletin, 18,245-
constraints and convergence. In G. Roberts (Ed.), Motivation in 251.
sports and exercise (pp. 3-29). Champaign, IL: Human Kinetics. Wills, T. (1981). Downward comparison principles in social psychol-
Rosenthal, R. (1978). Combining the results of independent studies. ogy. Psychological Bulletin, 90, 245-271.
Psychological Bulletin, 85, 185-193. Wine, J. D. (1982). Evaluation anxiety: A cognitive-attentional con-
struct. In H. W. Krohne & L. Laux (Eds.), Achievement, stress, and
Ryan, R. (1982). Control and information in the intrapersonal sphere: anxiety (pp. 207-219). New York: Hemisphere.
An extension of cognitive evaluation theory. Journal of Personality
and Social Psychology, 43, 450-461. Received November 29, 1994
Ryan, R. M. (1992). Agency and organization: Intrinsic motivation, Revision received September 27, 1995
autonomy, and the self in psychological development. In J. Jacobs Accepted October 7,1995 •