Reinforcement, Reward, and Intrinsic Motivation: A Meta-Analysis

Author(s): Judy Cameron and W. David Pierce

Source: Review of Educational Research, Vol. 64, No. 3 (Autumn, 1994), pp. 363-423
Published by: American Educational Research Association
Stable URL: http://www.jstor.org/stable/1170677
Accessed: 21-02-2016 18:22 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected].

Sage Publications, Inc. and American Educational Research Association are collaborating with JSTOR to digitize, preserve
and extend access to Review of Educational Research.

http://www.jstor.org

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Review of Educational Research
Fall 1994, Vol. 64, No. 3, pp. 363-423

Reinforcement,Reward, and Intrinsic Motivation:

A Meta-Analysis

Judy Cameron and W. David Pierce

University of Alberta

This article reviews research on the effects of reinforcement/reward on

intrinsic motivation. The main meta-analysis included 96 experimental stud-
ies that used between-groups designs to compare rewarded subjects to
nonrewarded controls on four measures of intrinsic motivation. Results
indicate that, overall, reward does not decrease intrinsic motivation. When
interaction effects are examined,findings show that verbal praise produces
an increase in intrinsic motivation. The only negative effect appears when
expected tangible rewards are given to individuals simply for doing a task.
Under this condition, there is a minimal negative effect on intrinsic motiva-
tion as measured by time spent on taskfollowing the removal of reward. A
second analysis was conducted on five studies that used within-subject
designs to evaluate the effects of reinforcement on intrinsic motivation;
results suggest that reinforcement does not harm an individual's intrinsic
motivation.

Reinforcementtheory has had a significant impact on education. Education

professors routinely teach the basic elements of behavior theory. As a conse-
quence, most classroom teachers have at least some rudimentaryunderstanding
of the principles of reinforcement.These principles are often used to promote
learning and to motivate students. In recent years, however, there has been a
growing concern over the applicationof rewardsystems in educationalsettings.
Several researchershave presentedevidence and arguedthat incentive systems
based on reinforcementmay have detrimentaleffects. The contention is that
reinforcementmay decrease an individual'sintrinsic motivationto engage in a
particularactivity.To illustrate,if a child who enjoys drawingpicturesis externally
reinforced(e.g., with points or money) for drawing,the child may come to draw
less once the reward is discontinued. In other words, one alleged effect of
reinforcementis that it underminesintrinsicinterest in a task.
The literatureconcernedwith the effects of reinforcementon intrinsicmotiva-
tion draws mainly from experimentalinvestigations. In an article published in
the AmericanPsychologist, Schwartz(1990) cited the intrinsicmotivationexperi-
ment of Lepper,Greene, and Nisbett (1973) and concluded that

This articleis based on JudyCameron'sPhD dissertation.The researchwas sup-

portedby the Social SciencesandHumanitiesResearchCouncil(SSHRC)in Canada.
We thankDr. W. FrankEpling and Dr. Steve Hunkafor their helpful comments
and suggestions.

363

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation
reinforcementhas two effects. First, predictablyit gains control of [an]
activity,increasingits frequency.Second, . .. when reinforcementis later
withdrawn,people engage in the activityeven less than they did before
reinforcementwas introduced.(p. 10)
While several researchersagree with this conclusion (e.g., Kohn, 1993; Suther-
land, 1993), otherscontinueto favorthe use of reinforcementprinciplesin applied
settings (e.g., Hopkins & Mawhinney, 1992). This is, obviously, an important
issue. Incentive systems are often implemented(or not) in schools, industry,
hospitals, and so forth on the basis of researchfindings and conclusions. The
presentarticleevaluatesthe literatureconcernedwith the effects of reinforcement
and rewardon intrinsicmotivationby a meta-analysisof the relevantexperimen-
tal investigations.
Several researchersdraw a distinctionbetween intrinsicand extrinsic motiva-
tion. Intrinsicallymotivatedbehaviors are ones for which there is no apparent
rewardexcept the activity itself (Deci, 1975). Extrinsicallymotivatedbehaviors,
on the other hand, refer to behaviors in which an external controlling variable
can be readilyidentified.Accordingto Deci (1975), intrinsicmotivationis demon-
stratedwhen people engage in an activity for its own sake and not because of
any extrinsic reward.The result of such behavior is an experience of interest
and enjoyment;people feel competent and self-determining,and they perceive
the locus of causality for their behavior to be internal.Intrinsicallymotivated
behavior is seen to be innate and is said to result in creativity,flexibility, and
spontaneity(Deci & Ryan, 1985). In contrast,extrinsicallymotivatedactions are
characterizedby pressureand tension and result in low self-esteem and anxiety
(Deci & Ryan, 1985).
A greatdeal of debatehas surroundedthe intrinsic/extrinsicdistinction.Several
critics (e.g., Guzzo, 1979;Scott, 1975) pointout difficultiesin identifyingintrinsi-
cally motivatedbehaviors.Although many humanbehaviorsappearto occur in
the absenceof any obvious or apparentextrinsicconsequences,they may, in fact,
be due to anticipatedfuturebenefits(Bandura,1977) or intermittentreinforcement
(Dickinson, 1989). From this perspective, intrinsically motivated behavior is
simply behaviorfor which appropriatecontrollingstimulihave yet to be specified.
In spite of these conceptualdifficulties, other social scientists frequentlyaccept
the intrinsic/extrinsicdistinction.In fact, a large body of researchis concerned
with the effects of extrinsicrewardsand reinforcerson behaviorthat is thought
to have been previously maintainedby intrinsicmotivation.The next section of
this article presentsa descriptionof the early studies concernedwith the effects
of rewardandreinforcementon intrinsicmotivation,the variousresearchdesigns
used to furtherinvestigatethe issue, the variablesinvestigated,andmajorfindings.
THE EFFECTSOF REWARDAND REINFORCEMENTON
INTRINSICMOTIVATION
The termsrewardand reinforcementhave frequentlybeen used synonymously.
Althoughthis is the case, behavioralpsychologistsmake an importantdistinction
between the two terms. A reinforceris an event that increases the frequencyof
the behavior it follows. A reward, however, is not defined by its effects on
behavior.Rewards are stimuli that are assumed to be positive events, but they
have not been shown to strengthenbehavior.Incentive systems (e.g., classroom
364

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

token economies) may be based on rewardor reinforcementand are designed to

increasemotivation.Because of these distinctions(between rewardand reinforce-
ment), this review separatesthose studies dealing with effects of rewardfrom
those concernedwith the effects of reinforcementon intrinsic motivation.
The Early Studies
The first laboratoryinvestigations to test the effects of reward on intrinsic
motivationwere conductedby Deci (1971, 1972a, 1972b). In the first experiment
(1971), 24 college students,fulfilling a course requirement,were presentedwith
a puzzle-solvingtask (Soma, a commercialpuzzle, producedby ParkerBrothers,
composed of seven differentshapes thatcan be solved in a varietyof ways). The
Soma puzzle was chosen because it was believed that college studentswould be
intrinsicallyinterestedin the task.The studywas madeup of three 1-hoursessions
over a 3-day period. Twelve subjects were assigned to an experimentalgroup;
the other 12 to a control group. During each session, subjects were individually
taken to a room and asked to work on the Soma puzzles in order to reproduce
variousconfigurationswhich were drawnon a piece of paper.Fourpuzzles were
presentedin a session, and subjects were given 13 minutes to solve each one.
In the second session only, experimentalsubjects were told that they would
receive $1.00 for each puzzle solved. Control subjects were offered no money.
In the middle of each session, the experimentermade an excuse to leave the
room for 8 minutes.Subjectswere told thatthey could do as they pleased.During
these 8-minuteperiods, the experimenterobservedthe subjectsthroughone-way
glass and recordedthe time that each subject spent engaged on the Soma task.
The amount of time spent on the task during the free periods was taken to be
the measureof intrinsic motivation,the dependentvariable.
Deci hypothesizedthatreward(money) would interferewith subsequentintrin-
sic motivationand that subjectsin the experimentalgroup would spendless time
on the task in the thirdsession than they had in the first. He suggested thatthere
would be a significant differencebetween the experimentaland control subjects
on this measure.Using a one-tailedt test, Deci found the differencebetween the
two groups to be significant at p < .10. The rewardedgroup spent less time on
the task thanthe controlgroup.Althoughsocial scientists do not generallyaccept
results at p > .05 as significant, and althoughDeci (1971) noted the marginal
natureof his result, the data have been taken as supportfor the hypothesisthat
If a personis engagedin some activityfor reasonsof intrinsicmotivation,
andif he begins to receive the externalreward,money,for performingthe
activity,the degree to which he is intrinsicallymotivatedto performthe
activitydecreases.(Deci, 1971, p. 108)
Deci's experimentis often cited as groundbreakingevidence for the negative
effects of reinforcementon intrinsic motivation (e.g., Kohn, 1993). Given the
distinctionbetween reward and reinforcement,however, Deci's (1971) experi-
ment, at most, demonstrates that rewards may have a negative impact on a
person's interestin a task. Nonetheless, his study was the first to investigate an
issue thatwas of prime concernto many psychologists. The experimentprovided
researcherswith a way to measureintrinsicmotivation and with a paradigmto
investigate the negative effects of reward.
365

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

In anotherexperiment,Deci (1971; Experiment3) used the same experimental

paradigmto investigate the effects of verbal reward.The reward contingency
introducedin the second session was verbal praise, ratherthan money. During
the second phase, subjects in the experimentalgroup were told after each trial
that their performancewas very good or much better than average. Deci found
that the reinforcedgroup spent significantly more time on the task (difference
scores between Session 3 and Session 1) than those who received no praise
(p < .05). These results suggest that social rewardsmay increasethe motivation
to performan activity.
One of the best known and most cited studies on the detrimentaleffects of
rewardon behavior is the work of Lepper,Greene, and Nisbett (1973). In this
study, nurseryschool childrenwere observed in a free-playperiod to determine
their initial interest on an activity (drawing).Two observers sat behind a one-
way glass and recordedthe amountof time each child was engaged in the activity.
Those childrenwho spentthe most time on the task were selected as subjectsfor
the experiment.Threeexperimentalconditionswere employed.In the "expected-
reward"condition, children were offered a "good-player"award, which they
received for drawingwith magic markers.Childrenin the "unexpected-reward"
group received the awardbut were not promisedit beforehand,and "no-reward"
subjects did not expect or receive an award.
In a subsequentfree-playsession, those childrenwho were promisedan award
(expected-rewardsubjects) spent significantly less time drawingthan the other
two groups. Furthermore,the expected-rewardgroup spent less time drawingin
the postexperimentalsession thanthey had in the initial session (preexperimental
free-play session). The unexpected-rewardand no-rewardsubjectsshowed slight
increases in time on task from preexperimentalto postexperimentalsessions.
Lepper et al. (1973) concluded that their results provided "empiricalevidence
of an undesirableconsequence of the unnecessary use of extrinsic rewards,"
(p. 136).
However, those who received an unexpectedrewardspent more time on the
task duringthe postexperimentalfree-playperiodthaneitherthe expected-reward
or the control group. Because the unexpected- and expected-rewardgroups are
both rewardconditions,the conclusionthatthese resultsdemonstratethe negative
effects of rewardmay not be warranted.This is becauserewardwas held constant
in the unexpected-rewardandexpected-rewardgroups;whatdifferedwas promise
or no promise. That is, the promises made or the instructionsgiven could have
producedthese results.Nonetheless, the findings of Lepperet al.'s (1973) study
are frequentlycited in journalarticles and introductorypsychology textbooksas
evidence that extrinsicrewardsand reinforcementundermineintrinsicinterestin
a task.
The early studiesby Deci (1971) andLepperet al. (1973) have raiseda number
of issues and controversies that have generated considerable research. Some
psychologists have claimed that the original findings provide evidence for the
view that reinforcementdecreases intrinsic motivation (e.g., Schwartz, 1990).
Othersrecognize that not all types of reinforcementundermineintrinsicinterest
(e.g., Deci & Ryan, 1985). Still others argue that one must demonstratethat
rewardsare, in fact, reinforcersbefore any statementsabout the effects of rein-
forcementcan be made (Feingold& Mahoney, 1975; Mawhinney,1990). Several
366

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

researchersare cautious about equating reward with reinforcement;their focus

has been to discover when and under what conditions reward is detrimental
(Bates, 1979; Morgan, 1984). In orderto address these issues, researchershave
employed a variety of researchparadigms.

Between-GroupDesigns
Studies designed to assess the effects of rewardon intrinsicmotivationhave
been conductedusing between-groupdesigns. Typically,one of two methods is
employed. The first method, referredto as a before-afterdesign (Deci & Ryan,
1985), involves a three-session paradigm.In these studies, a baseline measure
of intrinsicmotivationon a particulartask is taken. This entails measuringtime
on task in the absenceof extrinsicreward,usually from a session of shortduration
(e.g., 10 minutes). Subjectsare then assigned to a rewardor no-reward(control)
condition, and an interventionwith extrinsic rewards is carriedout. Following
this, rewardis withdrawn,and time on task is again measured.The procedure
is identical for both groups except that control subjects do not experience the
interventionin the second session. Mean differences in time on task between
pre- and postinterventionare calculated for each group, and the scores for the
experimentaland control subjectsare then statisticallycompared.Any difference
between the two groups is considered evidence of the effects of withdrawal
of reward.
One advantageto the before-afterprocedureis that it allows the researcherto
examine differences within groups from pre- to postexperimentalsessions as
well as differencesbetween groups. In most studies of this type, however, only
differences between groups are investigated. This is because the before-after
procedurehas generally been used to identify individuals who show an initial
interestin a specific task;those people are then selected as subjectsfor the study.
In suchcases, differencesbetweenrewardedandnonrewardedsubjectsareusually
measuredin the after-rewardsession only.
Most researchershave used an after-onlybetween-groupsexperimentaldesign
to assess the effects of rewards on intrinsic motivation. In this approach,no
pretreatmentmeasureof intrinsicinterestis collected. In the typical experiment,
subjectsare presentedwith a task thatis assumedto be intrinsicallymotivating-
solving and assemblingpuzzles, drawingwith felt-tippedpens, word games, and
so on. Experimentalsubjectsare rewardedwith money or grades,candy,praise,
good-player awards, and so forth for performingthe activity. In some studies,
the rewardis delivered contingenton a certainlevel of performanceon the task;
in others, subjects are simply rewardedfor participatingin the task. Control
subjects are not rewarded.The rewardinterventionis usually conductedover a
10-minuteto 1-hour period. All groups are then observed during a nonreward
period.This usually occurs immediatelyafter the experimentalsession, although
some researchershave observed subjects several weeks later. If experimental
subjectsspend less time on the task (duringthe postrewardobservation)thanthe
controls, reinforcement/rewardis said to undermine intrinsic motivation. The
amount of time subjects spend on the task during the nonrewardperiod is one
of the major ways in which intrinsic motivation has been measured,and it is
usually referredto as free time on task.
367

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

Findings from the Group Design Studies

Generally,the results of the group design studies examining the main effects
of rewardsare conflicting.While some researchershave found that rewardslead
to decreasedtime on the task relative to control groups (e.g., Deci, 1971; Fabes,
1987; Morgan, 1981), othersreportthe opposite (e.g., Brennan& Glover, 1980;
Deci, 1972a;Harackiewicz,Manderlink& Sansone,1984). Some studiesreportno
significantdifferences(e.g., Amabile, Hennessey,& Grossman,1986; DeLoach,
Griffith& LaBarba,1983).
Not all studies use the free-timemeasureof intrinsicmotivation.Otherdepen-
dent variableshave includedself-reportsof task enjoyment,interest,and satisfac-
tion;performanceduringthe free time period(numberof puzzles/problemssolved,
number of drawings completed, etc.); and willingness to volunteer for future
projectswithout reward.Overall, the results from studies employing these mea-
sures are conflicting and do not help to clarify the issue of whetherrewardleads
to decreasedintrinsicmotivation.
A number of reviewers (e.g., Bates, 1979; Deci & Ryan, 1985; Dickinson,
1989; Morgan, 1984) have noted the contradictorynatureof the findings and
have attemptedto identify the conditionsunderwhich extrinsicrewardproduces
decrementsin intrinsicmotivation.Some of the conditionsthoughtto be critical
in determiningthe impact of rewards include the type of reward (tangible or
verbal),rewardexpectancy(whetherrewardis expected-i.e., offeredbeforehand
or received unexpectedly),and rewardcontingency(whetherrewardis delivered
simply for performing the task or is contingent on some specified level of
performance).Although this categorizationsystem is useful, an examinationof
the literaturewithin each category reveals conflicting results.

Type of Reward
Whenverballypraisedsubjectsarecomparedto a controlgroup,some research-
ers have found an increasein intrinsicmotivation(e.g. Deci, 1971) while others
reportno significantdifferences(e.g., Orlick & Mosher, 1978). The same holds
true when subjects receiving tangible rewardsare comparedto controls. While
some results provide evidence for a decrease in intrinsic motivation following
the receipt of a tangible reward(e.g., Danner & Lonkey, 1981), others indicate
an increase (e.g., Rosenfield, Folger, & Adelman, 1980).'

Reward Expectancy

Comparisonsbetween subjects who receive an unexpected tangible reward

and subjects who receive no rewardare also not clear cut. Some results indicate
that unexpected-rewardsubjects show a decrease in intrinsic motivation (e.g.,
Orlick & Mosher, 1978); others have found no significant differences (e.g.,
Greene & Lepper, 1974). Experimentsdesigned to investigate the effects of
expected tangible rewardsare also contradictory.Some studies, comparingsub-
jects offered an expectedrewardto nonrewardedcontrols,show a negative effect
of rewardon intrinsicmotivation(e.g., Deci, 1971; Lepper,Greene, & Nisbett,
1973). Others, however, demonstratethat expected-rewardsubjects show an
increasein intrinsicmotivationrelativeto controls(e.g., Brennan& Glover, 1980).
368

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameron and Pierce

Reward Contingency

Morgan (1984) and Deci and Ryan (1985) suggest that rewardcontingency
may play a criticalrole in determiningthe negativeeffects on intrinsicmotivation.
Again, however, results from such studies vary. When rewards are delivered
contingenton some level of performance,some researchershave found a positive
effect (e.g., Karniol & Ross, 1977); others reportnegative findings (e.g., Ryan,
Mims, & Koestner,1983). Whenrewardsaredeliveredcontingenton engagement
in the task regardlessof subjects' level of performance,an underminingeffect
has been found in some studies (e.g., Lepper,Greene & Nisbett, 1973; Morgan,
1983, Experiment 1). Others report no decrease in intrinsic motivation (e.g.,
Pittman,Emery & Boggiano, 1982).

Within-SubjectDesigns
One of the criticismsof the groupdesign researchis thatresearchersemploying
such a design often refer to their rewardmanipulationas a reinforcementproce-
dure. By definition, a reinforceris an event that increases the frequency of the
behaviorit follows. In most studies on intrinsicmotivation,researchershave not
demonstratedthat the events used as rewards increased the frequency of the
behaviorstudied.In addition,critics (e.g., Feingold & Mahoney, 1975; Mawhin-
ney, 1990) suggest thatthe measurementphases in the groupdesign researchare
too brief to detect any temporaltrends and transitionstates. In orderto address
these issues, a few studies have been conducted using a repeated measures,
within-subjectdesign.
In this paradigm,the amount of time subjects spend on a particulartask is
measuredover a numberof sessions. Reinforcementproceduresare then imple-
mentedover a numberof sessions. In the final phase, reinforcementis withdrawn,
and time on task is again repeatedlymeasured.Intrinsicmotivation is indexed
as a differencein time on task between pre- and postreinforcementphases where
differences are attributedto the externalreinforcement.
In general,no substantialdifferenceshave been foundwhen rateof performance
andtime on task in postreinforcementsessions arecomparedto pre-reinforcement
phases (although,see Vasta & Stirpe, 1979).
The advantageof within-subjectsdesigns is that the researchercan determine
whether the rewards used are actual reinforcers-that is, whether behavior
increases during the reinforcementphase. Statements can then be made about
the effects of reinforcement,ratherthan reward. However, only a handful of
studies have been conductedusing this type of design.
Criticsof within-subjectresearch(e.g., Deci & Ryan, 1985) suggest thatresults
from these designs are not generalizablebecause so few subjects are studied in
any one experiment. A furthercriticism has to do with the lack of a control
group. The argumentis that in the within-subjectdesigns there is no group that
performsthe activity without reinforcement;thus, one cannot know if there is
an underminingeffect relative to a control group. Finally, for these studies,
the definition of a reinforceris necessarily circular.That is, reward becomes
reinforcementonly after its effects are shown to increase behavior.
369

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

THEORETICALACCOUNTS OF THE LITERATURE

Although the results of laboratoryinvestigations into the effects of reward
and reinforcementon intrinsicmotivationappearcontradictoryand confusing, a
general contentionin many textbooks andjournal articles is that rewardand/or
reinforcementis detrimentalto an individual'sintrinsicmotivation (e.g., Kohn,
1993; McCullers, 1978; Schwartz, 1990; Zimbardo, 1988). In an attempt to
accountfor the disparateoutcomes, a few psychologists have offered theoretical
explanations.Three majoraccounts are outlinedbelow.
The OverjustificationHypothesis
One explanationthat has been put forth to accountfor the detrimentaleffects
of rewardis termedthe overjustificationeffect (Lepper,Greene,& Nisbett, 1973).
This hypothesisis largely based on attribution(Kelly, 1967) and self-perception
(Bem, 1972) theories. A person's perceptionsabout the causes of behavior are
hypothesizedto influence futuremotivationand performance.In the presenceof
externalcontrols,people attributetheir behaviorto an externalagent; when this
is removed, futuremotivationand performancedecrease. Conversely,behavior
is attributedto internalcauses in the absence of obvious external controls. In
this case, motivationand performanceare not affected.
A decrease in intrinsicmotivation following the withdrawalof a rewardhas
been termed the overjustificationeffect because it is thought that an external
rewardprovidesoverjustificationfor participatingin an alreadyattractiveactivity.
Put another way, when individuals are rewarded for engaging in an already
interestingactivity,their perceptionsshift from accountingfor their behavioras
self-initiatedto accountingfor it in terms of externalrewards.That is, they are
faced with too many reasons (justifications)for performingthe activity, and the
role of intrinsicmotivationis discountedresultingin a decline in intrinsicmotiva-
tion.
Lepper (1981) has suggested that extrinsic rewards lead to a decrease in
intrinsicmotivationwhen they allow perceptualshifts of causality.Accordingto
Lepper, this occurs when there is sufficient initial interest in an activity, when
the extrinsicrewardsaresalient,andwhenrewardsdo not increaseperceivedcom-
petence.
Cognitive EvaluationTheory
Deci and Ryan (1985) suggest that the overjustificationhypothesis should not
be considereda theoryof motivation.They arguethatself-attributionsmay affect
intrinsic motivation,but they do not see them as necessary mediators.Instead,
Deci and Ryan offer cognitive evaluation theory as an explanationfor intrinsic
motivation.
Cognitiveevaluationtheoryis based on the assumptionthatpeople have innate
needs for competence and self-determination.From this perspective,a person's
intrinsic motivationis affected by changes in feelings of competence and self-
determination.According to Deci and Ryan (1985), events facilitate or hinder
feelings of competence and self-determinationdepending on their perceived
informational,controlling,or amotivationalsignificance.Events seen as informa-
tional indicateskill in performinga task;hence, competenceis facilitated,which
leads to increasedintrinsic motivation. A controllingevent is one perceived as
370

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

an attemptto determinebehavior.This type of event diminishes an individual's

self-determinationand intrinsic motivation. An amotivational event provides
negative feedback, indicating a lack of skill, which reduces one's competence
and intrinsic motivation.
Cognitive evaluation theory focuses on a person's experiences of an activity.
For this reason, Deci and Ryan (1985) emphasize the importanceof self-report
measures of task interest, satisfaction, and enjoyment as more indicative of
intrinsicmotivation than the free time-on-taskmeasure.
According to cognitive evaluation theory, rewards are not always harmful.
Verbalrewardsmay be informationaland lead to an increasein intrinsicmotiva-
tion. Tangible rewards, on the other hand, are seen as controlling when their
delivery is stated before the rewardperiod (expected rewards).This is because
the cognitive evaluationprocess is believed to begin while the rewardedactivity
is occurring.Further,rewardspromisedto personsfor engaging in a task without
a performancecriterion(referredto as expectedtaskcontingentrewardsby Deci &
Ryan, 1985) are controlling and decrease intrinsic motivation. Deci and Ryan
suggest that rewards delivered to a person contingent on a specified level of
performanceare more complicated.This type of rewardcan be informationalor
controlling,but the difficulty is that its function can only be determinedby how
well a person performs in relation to the specified standard.If the individual
performswell, the rewardis informational,and, if performanceis poor,it is con-
trolling.
Rummel and Feinberg (1988) conducted a meta-analysisto assess cognitive
evaluation theory. Subjects who received rewardsthat were defined to convey
"controlling"information were compared to groups receiving other types of
rewards or no reward. The dependent measure of intrinsic motivation was a
combinationof both free time-on-taskmeasuresand self-reportsof satisfaction
and task interest.Results providedsupportfor cognitive evaluationtheory.Rum-
mel and Feinbergconcluded thatcontrolling,extrinsic rewardshave detrimental
effects on intrinsic motivation.
In Rummel and Feinberg'smeta-analysis,rewardswere defined as controlling
after the fact. That is, when a rewardwas found to produce a negative effect, it
was seen as controlling,and the studywas selected for the analysis.This exempli-
fies the majordifficulty with cognitive evaluationtheory.Rewardsare defined as
controllingor informationalaftertheireffect on performancehas been measured.
Although cognitive evaluationtheory may account for the diverse findings of
the effects of reward on intrinsic motivation, there are difficulties with this
interpretation.One problemis thatfeelings of competenceand self-determination
are seen as causes of changes in intrinsicmotivation,but they are not measured.
They are assumedto be operatingbecause behaviorchanges. In otherwords, the
existence of competence, self-determination,and intrinsicmotivationis inferred
from the very behaviorit supposedlycauses. Rewardsare defined as controlling
if measuresof intrinsicmotivationdecreaseandinformationalwhen the dependent
variableindexes an increase in motivation.
BehavioralAccounts
An operant analysis of behavior involves consideration of a prior learning
history and the three-termcontingency,the SD:R -> Sr relationship.The three
terms are: (a) discriminativestimulus (SD)or setting event, (b) the response (R)
371

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and Intrinsic Motivation
or behavior,and(c) contingentreinforcement(Sr).Flora(1990) has suggestedthat
all of the empiricalresultsof the intrinsicmotivationresearchcan be accountedfor
by consideringthe promisedrewardprocedures(expectedreward)as discrimina-
tive stimuli. That is, telling a person that he or she will receive a rewardis a
stimulusevent thatprecedesthe operantand, as such, is a discriminativestimulus
ratherthan a reinforcer.From this perspective, if behavior is regulatedby its
consequences (i.e., reinforcement),no loss of intrinsic motivation is expected.
When individualswho are engaged in a task are reinforcedfor doing the task,
they will spend as much time on the activity as they originally did once the
reinforceris withdrawn.A behavioralview suggests thatit is only when rewards
function as discriminitivestimuli that one might expect to observe a decline in
intrinsicmotivation.
Although discriminitive stimuli are part of the three-termcontingency and
affect the probabilityof an operant,they can and do have very differenteffects
fromreinforcers.Taskperformanceevokedby instructionsandpromisesof reward
(SDs)can be influencedby a numberof factorssuch as the subject'shistory with
respectto whetherpromisedrewardswere actuallyreceived, the subject'sverbal
repertoire,the natureof priorexposure to the object being offered as the reward,
and so on (Dickinson, 1989).

SUMMARY
The overjustificationeffect, cognitive evaluationtheory,and recentbehavioral
explanations each attempt to account for the disparateeffects of reward and
reinforcementon intrinsicmotivation.Given the diverse findings reportedin this
literature,however,it is not clear at this point whateffect rewardor reinforcement
has on intrinsicmotivation.Reviewers on all sides of the issue tend to be highly
criticalof researchdesigned outside of theirown paradigm,and, more often than
not, findings from studies in opposing camps are not considered relevant. For
these reasons, the literatureand its interpretationsare still contentious.Because
a substantialnumberof experimentalstudies have been carriedout to assess the
effects of rewardand reinforcementon intrinsicmotivation,one way to evaluate
their effects is to conduct a meta-analysis.

THE PRESENTMETA-ANALYSIS
The primarypurposeof the presentmeta-analysisis to make a causal statement
about the effects of extrinsic rewardsand reinforcementon intrinsicmotivation.
This analysis should be useful in addressinga numberof concerns. Of major
importanceis whetherthe bulk of evidence suggests that extrinsic rewardsand/
or reinforcementproducedecrementsin intrinsicmotivation. If so, what is the
size of the relationshipsbeing uncovered? Also, do different patternsemerge
with differentrewardtypes (e.g., tangible, verbalrewards),rewardexpectancies
(expected, unexpected), or reward contingencies (e.g., rewards delivered for
engaging in a task, competing or solving a task, or meeting a specified level of
performance)?In the following sections of this article, the research questions
addressedin the presentmeta-analysisareoutlined,the steps involved in conduct-
ing the meta-analysisaredescribed,and the findingsare presentedand discussed.
372

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

ResearchQuestions
The following questions have been addressedin this meta-analysis:
1. Overall, what is the effect of reward on intrinsic motivation? In order
to answer this question, a meta-analysisof the group design experimentswas
conducted. Subjects who received a tangible reward and/or an extrinsic verbal
reward were compared to a nonrewardedcontrol group. This analysis should
shed light on the overall effects of rewardon intrinsicmotivation.
2. Whatare the effects of specificfeatures of rewardon intrinsicmotivation?
Several researchersnote that reward interacts with other variables to produce
incrementsor decrementsin intrinsicmotivation.That is, intrinsicmotivationis
affecteddifferentlyby the type of rewardimplemented,the rewardexpectancyand
the rewardcontingency.Specifically,researchershave investigatedthe following:
(a) the effect of rewardtype on intrinsic motivation (i.e., whetherrewardsare
verbal or tangible),
(b) the effect of rewardexpectancyon intrinsicmotivation(i.e., whetherrewards
are expected-promised and delivered to subjects or unexpected-
delivered to subjects but not promised),
(c) the effect of rewardcontingencyon intrinsicmotivation(i.e., whetherrewards
aredeliveredto subjectsfor participatingin an experimentalsession regard-
less of what they do, for engaging in a task, for completing or solving a
task, or for attaininga specific level of performance).
All analyses performed on these features were conducted with group design
studies in which a rewardedgroup was compared to a control group. These
analyses should lead to a greaterunderstandingof the specific conditions under
which rewardaffects intrinsicmotivation.
Although the present analyses present a breakdown of several features of
reward, there are other moderatorvariables mentioned in the literature(e.g.,
salience of reward,task type, rewardattractiveness,goals of individuals,etc.).
These conditionsmay interactwith rewardto affect intrinsicmotivation.Unfortu-
nately, these variables appear in only one or two studies and are, thus, not
amenableto a meta-analysis.At this pointin time, placingemphasison interaction
effects that have few replicationswould not be beneficial to an understanding
of rewardand intrinsic motivation.
3. Overall, what is the effect of reinforcementon intrinsic motivation?One
of the criticisms of the group designs has been that rewardis frequentlycited
as synonymouswith reinforcement,yet no evidence has been providedto indicate
thatthe rewardsused in groupdesigns areactualreinforcers.In the single-subject,
repeatedmeasuresdesigns, researchershavedemonstratedthatthe rewardsadmin-
isteredincreasedbehaviorand can be consideredas reinforcers.For this reason,
a separateanalysis was conductedwith the single-subjectdesigns where subjects
served as their own controls. This analysis should allow a more definitive state-
ment to be made about the effects of reinforcementon intrinsicmotivation.
METHOD
Selection of Studies
A basic list of studies was assembledby conductinga computersearchof the
psychological literature(PSYCH LIT) using intrinsic motivationas the search
373

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and Intrinsic Motivation

term.The meta-analysisstartedwith Deci (1971), and relevantarticlespublished

up to September1991 were identified.Studiesnot listed on the computerdatabase
were identified through the bibliographiesof review articles, chapters, books,
and papers located in the original search.
Two sets of studies were collected (between-groupdesigns and within-subject
designs). The main analysis entailed assessing the overall effects of rewardon
intrinsicmotivationfrom studies involving group designs. Criteriafor including
studies in the sample were:
(a) thatthe study involve an experimentalmanipulationof a rewardcondition
and include a nonrewardedcontrol group;
(b) that any characteristicsof rewardedsubjects be either held constant or
varied but be representedidentically for both rewardedand control groups;and
(c) that studies be published (no unpublisheddocumentswere collected) and
written in English.2
In addition,only studiesthatmeasuredintrinsicmotivationas a dependentvariable
were included.
Intrinsicmotivationhas been measuredas free time on task afterwithdrawalof
reward;self-reportsof task interest,satisfaction,and/orenjoyment;performance
during the free time period (number of puzzles/problems solved, number of
drawingscompleted,etc.); and subjects' willingness to participatein futureproj-
ects withoutreward.One study which met the criteriawas excluded (Boggiano &
Ruble, 1979) because the statisticalcontrastsused in the article were not logical
given the sample size of the study.3Otherstudies were omitted from the sample
if some subjects in a rewardcondition were not actually given a reward (e.g.,
Pritchard,Campbell,& Campbell, 1977). The resultingsample consisted of 83
documents,reporting96 independentstudies.
A major criticism of the meta-analytictechnique has been that researchers
often lump differentmeasurestogether.This has been referredto as the apples-
and-orangesproblemin thatit is arguedthatlogical conclusionscannotbe drawn
from comparisonsof studies using differentmeasuresof the dependentvariable
(see Glass, McGaw, & Smith, 1981). In order to avoid this problem, separate
analyses were conducted on the overall effect of reward for each measure of
intrinsicmotivation.Using this strategy,61 studies compareda rewardedgroup
to a control group on the free-time measure;64 studies investigatedthe attitude
(task interest, enjoyment, and satisfaction) measure; 11 studies assessed the
willingness to volunteerfor futurestudieswithoutrewardmeasure,and 12 studies
measuredperformanceduring the free-time period.
In order to assess the impact of specific featuresof reward,furtheranalyses
were conducted with data from the 96 group design studies. In these analyses,
subjectsassignedto differenttypes of rewards(tangible,verbal),rewardexpectan-
cies (unexpected,expected), and rewardcontingencieswere comparedto nonre-
wardedcontrol groups.
The second meta-analysiswas conducted on studies that employed a within-
subject, multiple-trialsdesign. In this type of design, subjects served as their
own controls. These experimentsare conducted in three phases with a number
of sessions in each phase. Baseline measures of intrinsic motivation are taken
in the first phase;reinforcementproceduresare then implementedover a number
of sessions, and in the third phase reinforcementis withdrawn. Changes in
374

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameron and Pierce

intrinsicmotivation are measuredas differences between the pre- and postrein-

forcementphase.
Single-subject studies were included in this analysis when a reinforcement
effect was demonstrated(i.e., the rewardsused showed an increasein behavior)
and when baseline, reinforcement, and postreinforcementphases involved
repeated measures. One study reporting a reinforcementeffect was excluded
(Vasta,Andrews, McLaughlin,Stirpe,& Comfort, 1978, Experiment1) because
the authorsreportedonly one measureof behaviorduringthe postreinforcement
phase. Two studies used a repeatedmeasures group design to assess the effects
of reinforcementbetween and within groups (Greene,Sternberg& Lepper,1976;
Mynatt, Oakley, Arkkelin, Piccione, Margolis, & Arkkelin, 1978). Although
subjects' performancein these studies was measuredrepeatedlyas in the single-
subjectdesigns, only group effects were reported.In addition,the rewardsused
in these studies were not shown to be reinforcers for some of the rewarded
groups.Thus, these two studies were not included in the meta-analysisof single-
subject designs (Mynattet al., 1978, are included in the meta-analysisof group
designs because their study included a nonrewardedcontrol group). In all, five
studies were selected for the within-subjectmeta-analysis.
A list of studies included in the meta-analysesis presentedin Appendix A.

Coding of Studies
Once all relevantarticles had been collected, each study was read and coded.
The following generalinformationwas extractedfrom each report:(a) author(s),
(b) date of publication,(c) publicationsource, (d) populationsampled (children
or adults), (e) sample size, (f) type of experimentaldesign (before-aftergroups
design, after-onlygroups design, or single-subjectmultiple-trialdesign), and (g)
type of task used in the study.
The following aspects of the independentvariablewere also coded: (a) reward
type (tangible or verbal), (b) rewardexpectancy (expected or unexpected) and
(c) rewardcontingency. Rewardcontingency was coded accordingto Deci and
Ryan's (1985) taxonomy.Task noncontingentrewardsreferredto rewardsdeliv-
ered to subjects for participatingin an experimentalsession regardlessof what
they did in the session. The term task contingent rewardwas used to mean that
a rewardwas given for actuallydoing the task and/orfor completingor solving
the task. Performancecontingentrewardswere defined as rewardsdelivered for
achieving a specified level of performance.In additionto using Deci and Ryan's
classification,contingencywas also codedin accordwith a behavioralperspective.
Using operantdefinitions,rewardswere defined as noncontingentor contingent.
Noncontingentrewardsreferredto rewardsdeliveredfor participatingin the study
or engagingin the taskregardlessof anylevel of performance.Contingentrewards
were defined as rewards dependent on performance (i.e., rewards given for
completing a puzzle, solving a task, and/orattaininga specified level of perfor-
mance).
Other characteristicsof studies that were coded were: (a) type of dependent
measure (e.g., free time on task, task interest, etc.), (b) whether experimenter
was blind to conditions, and (c) whether experimenterwas present or absent
duringthe post-rewardphase. As well, statisticalinformationwas recorded,and
effect sizes were calculated from appropriatecontrasts.
375

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

Descriptivecharacteristicsand effect sizes of the reviewed studies are summa-

rized in AppendixC.
IntercoderReliability
Fromthe pool of relevantstudies, 10 wererandomlyselected andindependently
coded by the secondauthor.A standardizedcoding form4was createdthatallowed
the second coder to extractinformationregardingindependentvariables(reward
type, rewardexpectancy,rewardcontingency),dependentvariables(measuresof
intrinsicmotivation),sample sizes, type of task used in the study,and calculation
of effect sizes for available contrasts. Reliability calculated as percentage
agreementwas 93.4%. For 6 of the 10 studies, agreementwas 100%.Disagree-
ments in the other four studies involved (a) miscommunicationof formulas to
use for calculatingeffect size (for two studies), (b) mislabeling of rewardexpec-
tancy (in one study), and (c) a misreadingof the numberof subjects in a group
(in one study). Disagreements were resolved through discussion and a more
careful readingof the studies and coding criteria.

Computationand Analysis of Effect Sizes

The proceduresused in the meta-analysisof the groupdesign studies followed
those of Hedges and Olkin (1985). Meta-analysisis a statistical technique for
aggregatingthe resultsof many experimentalstudieswhich comparetwo groups
on a common dependentmeasure.Once the studies and groups to be compared
are identified, the statisticalresult of each study is transformedinto a measure
called an effect size. An effect size is found by converting the findings from
each study into a standarddeviation unit. The effect size indicates the extent to
which experimentalandcontrolgroupsdifferin the meansof a dependentvariable
at the end of a treatmentphase. In its simplest form, the effect size calculated,
g, is the differencebetween the means of the rewardedgroup and a nonrewarded
control groupdivided by the pooled standarddeviationof this difference.When
means or standarddeviations were not available from reports, effect size was
calculated from t tests, F statistics, and p-level values (e.g., p < .05) by using
Hedges and Becker's (1986) formulas.Formulasfor calculating effect size are
listed in AppendixB.
One problem that arises in conducting a meta-analysisis determiningeffect
sizes from studies with limited information.In a few studies, for example, con-
trasts are simply reportedas t or F < 1.00. In such cases, effect size estimates
were calculated by making t or F equal to a number between 0.01 and 1.00
chosen froma randomnumberstable.Whenresultsfroma studywere not reported
or were reportedas nonsignificantand when t or F values were not available
but means and/or direction of means were known, a random numberbetween
0.01 and the criticalvalue of t or F atp = .05 was chosen to calculatean estimate
of effect size. When results for an outcome measurewere not reportedor were
reportedas nonsignificantand when means and direction were unknown, the
effect size for that measure was set at 0.00 (indicating exactly no difference
between rewardedand nonrewardedgroups). For each analysis, results were
calculatedwith 0.00 values included and with 0.00 values omitted.
For several studies, more than one effect size was calculated.For example, if
a single study containedtwo measuresof intrinsicmotivation(e.g., free time on
376

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

task, attitude)and two types of rewardgroupsplus a controlgroup(e.g., tangible

reward, verbal reward), a total of four effect sizes was calculated (e.g., free
time-tangiblereward,free time-verbalreward,attitude-tangiblereward,attitude-
verbal reward).
In order to satisfy the independence assumption of meta-analyticstatistics
(Hedges & Olkin, 1985), only one effect size per study was entered into each
analysis. When two or more effect sizes from one study were appropriatefor a
particularanalysis, these effect sizes were averaged.To illustrate,for the estimate
of the overall effect of rewardon the free-time measureof intrinsicmotivation,
some studies assessed the effects of severaltypes of rewards.If a single study,for
example,containedtwo or morerewardgroups(e.g., expectedreward,unexpected
reward)and a control condition, the two effect sizes were averaged so that the
study contributedonly one effect size to the overall analysis of reward.For an
analysis of the effects of expected rewardon intrinsic motivation,only the one
appropriateeffect size from the study would be used. This strategyretainedas
muchdataas possible withoutviolatingthe assumptionof independence.Average
effect sizes were obtainedby weightingeach g index by the numberof participants
on which it was based (see Cooper, 1989).
As was previously mentioned,in the single-subject,repeatedmeasuredesigns,
there is no separate control group; subjects serve as their own controls. An
increase or decrease in intrinsic motivation is indexed by a difference in the
amountof time spenton the taskbetweenbaselineandpostreinforcementsessions.
Effect sizes for these studies were calculated by subtractingthe average time
spent by all subjects in the baseline phase from the average time spent by all
subjects in the postreinforcementphase. This numberwas then divided by the
pooled standarddeviation.
After all effect sizes were calculated,the analyses were run on the computer
programMeta (Schwarzer, 1991). Results reportedin this article are based on
the weighted integrationmethod (Hedges & Olkin, 1985). Using this technique,
effect sizes g areconvertedto ds by correctingthemfor bias (g is anoverestimation
of the populationeffect size, particularlyfor small samples; see Hedges, 1981).
To obtain an overall effect size, each effect size is weighted by the reciprocalof
its variance,and the weightedds are averaged.This proceduregives more weight
to effect sizes that are more reliably estimated. Once mean effect sizes are
calculated,95% confidence intervalsare constructedaroundthe weighted mean.
In order to verify the accuracy of the computer program,one analysis (the
overall effect of rewardon free time) was hand calculated.All obtainedvalues
from the meta-analysisprogramand the hand calculationswere identical within
roundingerror.
To determinewhether each set of effect sizes in a sample shareda common
effect size (i.e., was consistent across studies), a homogeneity statistic, Q, was
calculated. Q has an approximatechi-squaredistributionwith k- 1 degrees of
freedom, where k is the number of effect sizes (Hedges & Olkin, 1985). The
null hypothesis is that the effect sizes are homogeneous (i.e., effect sizes in a
given analysis are viewed as values sampledfrom a single population;variation
in effect sizes among studies is merely due to sampling variation).For purposes
of the present analyses, samples were consideredhomogeneous at p > .01.
377

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

When samples are not homogeneous, studies can be classified by characteris-

tics, such that effect sizes within categories are homogeneous. This strategy
was undertakenby examining the effects of differenttypes of rewards,reward
expectancies, and rewardcontingencies.
As a supplementaryanalysis, homogeneitywas attainedby removingoutliers.
Thatis, studieswere omittedwhen they providedestimatesthatwere inconsistent
with those from otherstudies. Outliersin each dataset were first identifiedusing
Tukey's (1977) procedure.These outliers were then omitted from the analysis.
If homogeneitywas still not attained,otherstudies thatreducedthe homogeneity
statistic by the largest amount were removed. Hedges (1987) has pointed out
that this is a common procedurein both the physical and social sciences. In one
area of physics, for example, Hedges (1987) found that data from 40% of the
availablestudieswere omittedfromcalculations.Formeta-analysesof psycholog-
ical topics, Hedges (1987) notes that removal of up to 20% of the outliers in a
group of heterogeneouseffect sizes usually results in a high degree of homoge-
neity.
In an article in Psychological Bulletin, McGrawand Wong (1992) noted that
one of the problems with effect size statistics (e.g., d) is that many readersof
meta-analyseshave difficulty interpretingthe meaning and generalizabilityof
findings. McGrawand Wong have introducedanotherway to look at effect size,
by a statisticthey call the commonlanguage effect size indicator(CL). CL refers
to the probabilitythat a score sampledfrom one distributionwill be greaterthan
a score sampled from some other distribution.McGraw and Wong suggest that
CL is a useful way to talk about effect size because it is easily interpretable.
They provide an example in which a sample of young adult men is compared
to a sample of young adult women on the variableheight. A CL of .92 indicates
the probabilityof a male being taller than a female. Put another way, in any
randompairingof young adult men and women, the male will be tallerthan the
female 92 out of 100 times.
CL is calculatedfrom means and standarddeviations. Additionally,an effect
size, d, can be convertedto CL by multiplyingd by 1// or 0.707 to obtain a Z
value (K.O. McGraw,personalcommunication,April 24, 1992). The uppertail
probabilityassociated with this value correspondsto CL and can be calculated
using the unit normalcurve.
To test the robustnessof the CL statistic,McGrawand Wong (1992) conducted
a series of 118 tests (simulations)to determinethe implicationsof violating the
assumptionthat sample data come from populationsof values that are normally
distributedwith equal variances. They found small discrepancies between the
estimateof CL underthe normalityassumptionand the estimateof CL when the
normalityassumptionwas violated in termsof skewness and kurtosis.The worst
case discrepancy was 0.1 which occurred with a large violation of the equal
variance assumption,considerablenegative skewness, and a large violation of
kurtosis.Given the robustnessof CL andthe ease with which it can be interpreted,
resultsfrom the presentanalyseshave also been expressed using the CL statistic.
The meta-analyticproceduresused in the presentreview include:(a) the estima-
tion of average effect sizes and 95% confidence intervals, (b) homogeneity
analyses to determinewhethereffect sizes are drawnfrom the same population,
(c) removal of outliers to attain homogeneity, and (d) conversion of average
378

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameron and Pierce

effect sizes to the commonlanguagestatistic(CL). Note thatoutliersare included

and excluded in each analysis.
RESULTSFROM GROUP DESIGNS
The Overall Effect of Reward on IntrinsicMotivation
To assess the overall effect of rewardon intrinsicmotivation,descriptiveand
meta-analyticprocedureswere performedon each of the four differentmeasures
of intrinsicmotivation(free time on task, attitude,performanceduringthe free-
time period;willingness to volunteerfor futurestudies withoutreward).For each
measure,negative effects representa decrementin intrinsicmotivation;positive
effects indicate an increment.
Direction of Effects
The number of studies collected for each analysis of the overall effects of
rewardson intrinsic motivationand the directionof their effects is presentedin
Table 1. On the free-time measure,the majorityof studies showed that reward
decreasedintrinsicmotivation.However,when intrinsicmotivationwas measured
by attitudetowarda task,performanceduringthe free-timeperiod,or willingness
to volunteer for future studies without reward, more studies showed positive
effects.
Distribution of Effect Sizes
Frequencydistributionsof the data are shown in Figure 1. Studies that found
no significant differencesbut did not provide sufficient informationto calculate
effect sizes are not portrayedin the graphs.
When intrinsicmotivationwas measuredas time on taskfollowing the removal
of a reward (free time), effect sizes ranged from -1.94 to 1.06. The bulk of
experiments found effects between -0.59 and 0.19. Using Tukey's (1977)
procedure,one negative outlier was identified in the free-time data. This effect
(g = -1.94) was calculatedfroma studyconductedby Morgan(1983, Experiment
1). In this study, subjectswho received an expected, task contingent(noncontin-
gent), tangible rewardwere comparedto no-rewardcontrol subjects. The large

TABLE 1
Number of studies and direction of effects for reward versus control groups on
four measures of intrinsic motivation
Free Performance Willingness
Numberof studies time Attitude in free time to volunteer
Showinga positive effect
of reward 22 31 6 6
Showinga negativeeffect
of reward 34 15 4 4
Showingno effect 1 1
With lack of sufficientinfor-
mationto calculateeffects 4 17 2 1
Total 61 64 12 11

379

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Reward,andIntrinsicMotivation
Reinforcement,
10- 12 -
FREETIME
U,
aC 1 - I ATTITUDE
9-
0) 10 -
8- 03 9-
V)
7- 0 -
0
6 - 7 -
0 6-
0o
0
5 - L
4 - a) 5 -
t0 4 -
3-

-
E

l1i I
3 -

z
E 2 -
I- . I, l z 2 -

0-
O'.
000- 0 0 0060%

000! i I I o o o o o
r i ir i00 i s! 0000
o s o I o o00
o
000 0 0 0? 00
o
s0000000000
c00
o o o o-

Effect sizes
Effect sizes

0, WILLINGNESSTO VOLUNTEER
()
a)
PERFORMANCE
._V 3-
2 -
u)

0 0 2-
L-
a)
.
E E -

0-
z

o-0.'.
I
..
00000000000000000%00%0%0
. I ..
I .r_Wl
I
i I
O 0 0 0 O

_
0% _ - - oN o' 4Soooo^Mw ooooooooo
0% 6 0% 0 0 (N 0> 0%
0% 0C 0%oC 0C, r j CM
t0 0v. oI
.......... I e.... E Efec t sse 000

Effect sizes
Effect sizes
FIGURE1. Frequencydistributionsof effect sizesfor overall rewardversus con-
trol groups on four measures of intrinsicmotivation

negative effect could be due to the type of reward(tangible), the rewardexpec-

tancy,and/orthe rewardcontingency.All of these featuresareexaminedin further
analyses. In addition, this study was somewhat different from other studies in
that subjects who performedthe activity for a reward were observed by other
subjects.Thatis, subjectswere offereda rewardfor engaging in an activity while
theirperformanceon the task was being watched.Thus, the large negative effect
could be a result of an interactionof reward type, expectancy, contingency,
and surveillance.
The attitudemeasureof intrinsicmotivationrefers to subjects' self-reportsof
task interest,enjoyment,and/or satisfaction.Effect sizes ranged from -0.69 to
+ 1.98 with the majorityof effects fallingbetween -0.19 and +0.59. Two positive
outliers in this data set come from studies conducted by Vallerand(1983) and
Butler(1987). In bothof these studies,extrinsicverbalrewardis comparedto a no-
rewardgroup.The effect of verbalrewardon intrinsicmotivationis investigatedin
a subsequentanalysis.
Effect sizes on the performancemeasure ranged from -3.72 to +0.96; the
median was +0.03. One large negative outlier (-3.72) comes from a study
conductedby Deci (1971, Experiment2). This study differedfrom othersin that
it was a field experimentwhere studentsworking for a college newspaperwere
380

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameron and Pierce

paid to write headlines. Only eight subjects participated,and two subjects in the
control group droppedout and were not included in the analysis.
On the willingness-to-volunteermeasure, effect sizes ranged from -0.63 to
+0.68. There were no outliers in this sample.
To establish whetherthe CL statistic (McGraw& Wong, 1992) could be used
confidentlyin the analyses,the extent to which the free-timedistributionof effect
sizes deviated from normalitywas determined.Obtained values for skewness
and kurtosis were -0.21 and 0.55, respectively (where normal skewness and
kurtosis equal 0.00). McGraw and Wong tested the effect that violations from
normality would have on CL. Based on their findings and the skewness and
kurtosis values obtainedhere, in the meta-analysisof effect sizes for the free-
time measure, one could expect, at worst, an underestimateof 0.02 and an
overestimateof 0.04 for CL. Given this small discrepancy,the implicationis that
the CL statistic can be used and interpretedwithout any serious concern about
violations of normalityand homogeneity of variance.

Meta-Analysisof Effect Sizes

The overall meta-analysis of effect sizes presented in Table 2 allows one
to determine whether rewardedsubjects showed less intrinsic motivation than
nonrewardedsubjects as measured by time on task following the removal of
reward(free time); self-reportsof task interest,satisfaction,and enjoyment (atti-
tude); performanceduringthe free-time period;and willingness to volunteerfor
future studies without reward.
For each measure of intrinsic motivation, an analysis was conducted which
included all studies that providedsufficient informationto calculate effect sizes
(see "All known effects" in Table 2). When samples were not homogeneous,
outlierswere identifiedand removedusing Tukey's (1977) procedure.If samples
were still significantlyheterogeneous,additionaloutlierswere removed.Homoge-
neity was attainedfor the free-time and attitudemeasuresby omitting approxi-
mately20%of the effect sizes, a typical meta-analyticprocedure.An examination
of Table 2 indicates that the procedureof including and excluding outliers does
not drasticallyalter mean effect sizes.
On the free-timemeasure,rewardedsubjects showed less intrinsicmotivation
than nonrewardedcontrols (mean weighted d = -0.04), but this effect was not
significant (i.e., the confidence intervalincluded0.00). When the mean effect of
the homogeneous sample was converted to CL, results indicate that, given a
sample of studies designed to investigate the effects of rewardon time on task,
51 out of 100 studies would show thatoverall, rewardedsubjectsspend less time
on the task than nonrewardedcontrols (assuming that all studies are of equal
importanceand have the same characteristics).
Results from the attitude measure indicate greater intrinsic motivation for
rewardedsubjects.This effect was small at 0.14 (fromthe homogeneoussample)
but differed significantlyfrom the value of 0.00 (i.e., the confidence intervaldid
not include 0.00). The CL statistic was .54 and can be interpretedto mean that,
in comparisons of rewardedto nonrewardedsubjects, rewardedsubjects will
show a more positive attitudetowarda task than nonrewardedsubjectsin 54 out
of 100 studies. Rewardedsubjects also showed a tendency to score higher on
381

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
TABLE 2
Overall effect of reward versus control groups on four measures of intrinsic
motivation
Sample Mean 95% CI
Analysis k size weighted d for d Q CL
Free time on task
All known effects
(zeros excluded) 57 3539 -0.06 -0.13 to 0.01 225.51* .48
Outliers removed
using Tukey's
procedure
(zeros excluded) 56 3459 -0.03 -0.10 to 0.04 177.40* .49
Additional outliers
removed
(no zeros) 44 2634 -0.04 -0.12 to 0.04 66.39 .49
All reports (zeros
and outliers
included) 61 3858 -0.06 -0.12 to 0.01 225.80* .48

Attitude
All known effects
(zeros excluded) 47 3184 +0.21 0.14 to 0.29 167.50* .56
Outliers removed
using Tukey's
procedure
(zeros excluded) 45 3034 +0.17 0.09 to 0.24 110.70* .55
Additional outliers
removed
(no zeros) 39 2680 +0.14 0.06 to 0.22 58.03 .54
All reports (zeros
and outliers
included) 64 4431 +0.15 0.09 to 0.21 177.07* .54

Performance during free time period

All known effects
(zeros excluded) 10 575 +0.08 -0.09 to 0.25 27.90* .52
Outliers removed
using Tukey's
procedure
(zeros excluded) 9 569 +0.09 -0.08 to 0.26 21.63* .52
Additional outliers
removed
(no zeros) 8 509 -0.0004 -0.18 to 0.18 11.73 .50
All reports (zeros
and outliers
included) 12 770 +0.06 -0.09 to 0.21 28.07* .52

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

TABLE 2
Overalleffect of rewardversuscontrolgroupson four measuresof intrinsic
motivation-continued
Sample Mean 95%CI
Analysis k size weightedd for d Q CL
Willingnessto volunteer
All knowneffects
(zeros
excluded) 10 561 +0.05 -0.12 to 0.23 17.38 .52
All reports(zeros
and outliers
included) 11 609 +0.05 -0.12 to 0.22 17.42 .52
Note. Negative effect sizes indicatea decreasein intrinsicmotivationfor reward/
reinforcementgroups;positive effect sizes indicatean increase.k = numberof
effect sizes; samplesize = sum of n in all studies;meanweightedd = meanof
weighted effect sizes (weighted by sample size); CI = confidence interval;
Q = homogeneitystatisticfor meaneffect sizes; CL = commonlanguageeffect
size statistic.
*Significanceindicatesrejectionof the hypothesisof homogeneity.
*p < .01.

performancemeasuresand to volunteerfor the futureprojectsmore than nonre-

warded subjects, but these effects were not significant.
Studies that could not be representedwith effect sizes were given a value of
0.00. When these studies were includedin the overall analyses (see "All reports"
in Table 2), the mean effect size for each measurewas little changed.
Overall, the results show that reward does not significantly affect intrinsic
motivation as measuredby free time on task following removal of reward,by
performanceduringthe free-timeperiod,or by subjects' willingness to volunteer
for future projects without reward. When intrinsic motivation is measuredby
attitudetowarda task, rewardedsubjects reporthigher intrinsicmotivationthan
nonrewardedsubjects. It is importantto point out that these main effect results
should be viewed with caution. This is because many studies show interaction
effects that are obscuredwhen results are aggregated.
Previous reviewers (e.g., Deci & Ryan, 1985; Morgan, 1984) have suggested
that rewardtype, rewardexpectancy,and rewardcontingency may influence the
effect of rewardon intrinsicmotivation.In subsequentanalyses, effect sizes have
been partitionedinto groups based on these characteristicsin an attemptto test
potential moderatorvariablesand to establish homogeneity of variance.

Interactions:Effect Size as a Functionof RewardCharacteristics5

In the following section, type of reward and its impact on effect size are
presented. Studies are included that measured the effects of either verbal or
tangible reward (e.g., money) on intrinsic motivation. The second part of this
section involves an analysis of rewardexpectancy(i.e., expected and unexpected
rewards).Finally,rewardcontingency is assessed. Specifically, the questionhere
is whethereffect size varies as a function of rewarddelivered for engaging in a
383

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

task, completing or solving a task, or achieving a certain level of performance.

Studies that could not be representedas effect sizes due to lack of sufficient
informationare not included in furtheranalyses presentedin this article.6
Type of Reward
The purposeof the present analyses is to assess the effects of differenttypes
of rewards(i.e., tangibleand verbal)on intrinsicmotivation.Because few studies
assessed intrinsicmotivationas a functionof "performanceduringthe free-time
period"and "willingness to volunteer,"no furtheranalyses on these measures
have been conducted.
Effect sizes for both types of rewardon the free-time and attitudemeasures
are presentedin funnel distributionsin Figure 2. Funnel graphsare used to plot
effect size against sample size of the study. The advantageof a funnel display
is that it capitalizes on a well-known statistical principle (Light & Pillemer,
1984). That is, the larger the sample, the closer the effect size will come to
representthe true underlyingpopulationvalue; variabilitydue to samplingerror
decreases. Conversely, smaller samples are more prone to sampling errorand
are likely to deviate considerablyabout the true mean. For these reasons, the
distributionis expected to take the shape of an invertedfunnel.
An inspectionof the funneldistributionof effect sizes for the free-timemeasure
indicates that, overall, larger samples tend to concentratearoundzero; greater
variationis evident with smaller samples. Verbalreward appearsto produce a
positive effect. Results of tangiblerewardsuggest a negative effect. These differ-
ences suggest that, on the free-time measure,the effects of rewarddepend on
the type of reward.On the attitudemeasure,positive effects emerge from both
tangible and verbal rewardstudies; verbal rewardappearsto producea slightly
more positive effect. There is no indicationof a publicationbias because studies
with small sample sizes and near zero effects are representedin the funnel
distribution(for a discussion of this issue, see Light & Pillemer, 1984). Although
it is not possible to rule out experimenterbias (Rosenthal, 1966), the funnel
graphsdemonstratethat samplingvariabilitymay account for the fact that some
researchersfind rewardhas a detrimentaleffect while others do not.
The results from the meta-analysisof the effects of rewardtype presentedin
Table 3 indicate that, when studies compared subjects who received a verbal
reward(i.e., praise or positive feedback)to those who did not receive a reward,
rewardedsubjectsdemonstratedsignificantlyhigher intrinsicmotivationas mea-
suredby both time on task and attitude.On the time measure,homogeneitywas
attained by removing one outlier. This extreme positive value (+1.61) was
obtainedfrom a study conductedin India (Tripathi& Agarwal, 1985). Because
all other studies in this analysis came from North America, the large effect size
may have been due to differences in the population studied.7Three outliers
from studies measuringthe effects of verbal reward were removed to achieve
homogeneityon the attitudemeasure.Inspectionof these outliers suggestedthat
they did not differ in obvious ways from other studies in the sample except for
their tendency to generate extreme values of effect size. From these analyses,
one can estimate that the probabilityof a sample of verballyrewardedsubjects'
being morehighly intrinsicallymotivatedthannonrewardedsubjectsis 0.61 (CL)
as measuredby time on task and attitudetowardtask.
384

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce
180 -
FREETIME 0
160 -
o Tangible
140 - *A Verbal

mD 120 -
N
,s ."

u
C) 100- 8
00.
I ADA
0
1. 0-0
E 0
0 o o co ? o
Un 60 - 0 o ^ ?
^ A
40- 0 o0 ? "
O . , .1 0

0 I I

-2 -1 0 +1 +2
Effect size

160 -
ATTITUDE 0
140- o Tangible
A Verbal
120- a

?
Ca
acn ^ A ^C

0 0

-3 -2 -1 0 +1 +2 +3
Effect size
FIGURE 2. Funnel distributions of effect sizes for tangible and verbal reward
on two measures of intrinsic motivation

Studies assessing the effects of tangible rewardon intrinsicmotivation show

a decrease on the free-time measureas indicatedby a negative mean effect size
thatdiffered significantlyfrom 0.00. The CL statisticof .44 implies that subjects
who receive a tangible rewardwill show a decrease in intrinsic motivation as
measured by time on task in 56 out of 100 studies. The mean effect size on
attitudefor subjectsgiven a tangible rewardwas positive, but once outliers were
removed, the mean did not differ significantly from 0.00.
385

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

TABLE3
Effect size as a function of the type of reward delivered
Mean
Reward Sampleweighted 95% CI
type Analysis k size d for d Q CL
Free time on task
Verbal All knowneffects 15 958 +0.42 0.29 to 0.56 29.37* .62
Verbal Outliersremoved
using Tukey's
procedure 14 918 +0.38 0.25 to 0.52 18.96 .61
Tangible All knowneffects 51 2983 -0.20 -0.28 to -0.12 181.01* .44
Tangible Outliersremoved
using Tukey's
procedure 47 2761 -0.22 -0.30 to -0.14 97.55* .44
Tangible Additional
outliers
removed 43 2591 -0.21 -0.29 to -0.13 63.53 .44

Attitude
Verbal All knowneffects 15 1024 +0.45 0.31 to 0.58 69.71* .63
Verbal Outliersremoved
using Tukey's
procedure 13 874 +0.30 0.15 to 0.43 26.75* .58
Verbal Additional
outliers
removed 12 785 +0.39 0.24 to 0.53 8.73 .61
Tangible All knowneffects 37 2362 +0.09 0.004 to 0.17 143.29* .52
Tangible Outliersremoved
using Tukey's
procedure 33 2149 +0.05 -0.04 to 0.13 50.56 .52
Note. Negative effect sizes indicatea decreasein intrinsicmotivationfor reward/
reinforcementgroups;positive effect sizes indicatean increase.k = numberof
effect sizes; samplesize = sum of n in all studies;mean weightedd = meanof
weighted effect sizes (weighted by sample size); CI = confidence interval;
Q = homogeneitystatisticfor meaneffect sizes; CL = commonlanguageeffect
size statistic.
*Significanceindicatesrejectionof the hypothesisof homogeneity.
*p < .01.

In summary,subjects rewardedwith verbal praise or positive feedback show

significantlygreaterintrinsicmotivationthan nonrewardedsubjects.Those who
receive a tangible reward evidence significantly less intrinsic motivation than
nonrewardedsubjects, as measuredby time on task, but they do not differ in
their reportsof task interest or enjoyment.
The next step in the analysis involves a furtherbreakdownof the effects of
tangible reward.The goal is to identify variablesthat may moderatethe effects
of tangiblerewardon intrinsicmotivationand to establishwithin-grouphomoge-
neity.One factorthat may impacteffect size is whetherthe rewardsimplemented
in the studies were promised to subjects prior to the experimentalsessions or
whetherthey were received unexpectedly.
386

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

RewardExpectancy
Within the intrinsic motivation literature, researchers draw a distinction
between expected and unexpectedreward.Expectedrewardsreferto a procedure
whereby subjects are offered a reward prior to the experimental session and
delivered the rewardfollowing the session. Subjectswho receive an unexpected
rewardhave not been promisedthe rewardbeforehand.These termsare generally
used to describe proceduresinvolving the administrationof tangible rewards.
In most studies on verbal reward,praise was deliveredunexpectedlyand was
not contingenton any specified level of performance.The few studies on verbal
rewardthat did employ expected and/orcontingencyproceduresdid not produce
effect sizes that deviated much from the mean effect size presentedin Table 3.
For this reason, no furthersubdivisionof effect sizes from verbalrewardstudies
was undertaken.The following analyses concern the effects of tangible reward.
Results are displayed in Table 4.
Only six studies assessed the effects of unexpected tangible reward on the
time measureof intrinsicmotivation;five studies investigatedattitude.The aver-
age effect sizes for unexpected tangible reward versus control groups on free
time and attitudewere slightly positive but did not differfrom 0.00. These results
indicate that subjectsreceiving an unexpectedrewarddo not differ significantly
from nonrewardedcontrol subjects on measuresof intrinsicmotivation.
For the expected tangiblerewardversus controlcomparisons,expected reward
subjects demonstratedsignificantly less intrinsic motivation on the free-time
measure. On attitude,when homogeneity was attained,the two groups did not
differ.
In the following section of this article, studies comparingexpected, tangible
rewardgroupsto nonrewardedcontrolswere furthersubdividedinto groupsbased
on rewardcontingency.

Reward Contingency
In some studies, subjectswere promised a tangible rewardthat was delivered
for participatingin the study or for engaging in a specific task. In other studies,
a tangible reward was offered for solving a puzzle, completing a task, and/or
attaininga certainlevel of performance.Rewardsadministeredin these various
ways have been labeledby Deci and Ryan (1985) as task noncontingent(rewards
offered for participatingin the study regardlessof what subjectsdo), task contin-
gent (rewards offered for engaging in a task, and/or completing or solving a
task), and performancecontingent(rewardsoffered for attaininga specified level
of performance).Table5 presentsresultsfrom the meta-analysisof these compari-
sons.
Table 5 indicates that when subjects who are promised a tangible reward
regardless of what they do in the study (task noncontingent)are compared to
nonrewardedcontrols,no significantdifferenceemergeson the free-timemeasure
of intrinsic motivation. No analyses were conducted with this type of reward
contingencyon the attitudemeasurebecause only two studiesof this type assessed
attitude.Subjectswho receive an expected tangiblerewardfor doing, completing,
or solving a task (task contingent) show significantly less intrinsic motivation
thancontrols,as measuredby time on task, once rewardis withdrawn.On attitude,
387

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

TABLE4
Effect size as a function of reward expectancy for tangible reward versus
control comparisons
Mean
Reward Sample weighted 95% CI
expectancy Analysis k size d for d Q CL
Free time on task:Tangiblerewardversuscontrol
Unexpected All known
effects 6 275 +0.01 -0.24 to 0.25 7.38 .50
Expected All known
effects 50 2825 -0.23 -0.30 to -0.15 185.48* .44
Expected Outliers
removed
using
Tukey's
procedure 46 2603 -0.25 -0.33 to -0.17 101.36* .43
Expected Additional
outliers
removed 42 2408 -0.25 -0.33 to -0.16 64.78 .43
Attitude:Tangiblerewardversuscontrol
Unexpected All known
effects 5 311 +0.06 -0.16 to 0.28 12.42 .52
Expected All known
effects 35 2126 +0.10 0.01 to 0.19 135.26* .53
Expected Outliers
removed
using
Tukey's
procedure 32 1961 +0.07 -0.02 to 0.16 50.48 .52
Note. Negative effect sizes indicatea decreasein intrinsicmotivationfor reward/
reinforcementgroups;positive effect sizes indicatean increase.k = numberof
effect sizes; samplesize = sum of n in all studies;meanweightedd = meanof
weighted effect sizes (weighted by sample size); CI = confidence interval;
Q = homogeneitystatisticfor meaneffect sizes; CL = commonlanguageeffect
size statistic.
*Significanceindicatesrejectionof the hypothesisof homogeneity.
*p < .01.

they show less intrinsicmotivation,but this difference is not significant.When

rewardsare delivered contingent on a certain level of performance,there is no
significanteffect on the free-timemeasure;subjectsin this conditiondo, however,
reporta more positive attitudethan controls.
Studies employing various rewardcontingencies were also categorizedusing
behavioral definitions. Rewards delivered for participatingin a study or for
engaging in a task are referredto as noncontingentrewards.Rewardsare called
contingent when they are offered for solving a puzzle, completing a task, or
reachinga specified level of performance.The results of this analysis are shown
in Table 6. The findings indicate that, when reward contingency is defined
behaviorally,subjectsdemonstratea decreasein intrinsicmotivationon the free-
388

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
TABLE 5
Effectsize as a functionof rewardcontingency(as definedby Deci & Ryan,
1985)for expectedtangiblerewardversuscontrolcomparisons
Mean
Reward Sampleweighted 95% CI
contingency Analysis k size d for d Q CL
Free time on task:Expectedtangiblerewardversuscontrol
Tasknon-
contingent All knowneffects 6 225 +0.55 +0.27 to 0.83 20.02* .65
Tasknon-
contingent Outliersremoved 4 124 +0.10 -0.26 to 0.45 1.86 .53
Task
contingent All knowneffects 45 2257 -0.32 -0.41 to -0.24 130.90*.41
Task Outliersremoved
contingent usingTukey's
procedure 44 2177 -0.28 -0.37 to -0.19 94.99* .42
Task Additionaloutliers
contingent removed 40 2015 -0.23 -0.32 to -0.14 62.08* .44
Performance
contingent All knowneffects 10 484 -0.12 -0.31 to 0.06 26.22* .47
PerformanceOutliersremoved
contingent using Tukey's
procedure 8 439 -0.13 -0.34 to 0.06 17.83 .46
Attitude:ExpectedtangiblerewardversusControl
Task
contingent All knowneffects 21 1217 -0.07 -0.18 to 0.05 53.75* .48
Task Outliersremoved
contingent using Tukey's
procedure 20 1157 -0.01 -0.13 to 0.10 36.24* .49
Task Additionaloutliers
contingent removed 19 1058 -0.08 -0.20 to 0.04 21.76 .48
Performance
contingent All knowneffects 14 819 +0.38 0.24 to 0.52 70.03* .61
PerformanceOutliersremoved
contingent using Tukey's
procedure 13 762 +0.29 0.14 to 0.43 27.35* .58
PerformanceAdditionaloutliers
contingent removed 11 682 +0.19 0.04 to 0.35 11.54 .55
Note. Negative effect sizes indicatea decreasein intrinsicmotivationfor reward/
reinforcementgroups;positive effect sizes indicatean increase.k = numberof
effect sizes; samplesize = sum of n in all studies;Meanweightedd = meanof
weighted effect sizes (weighted by sample size); CI = confidence interval;
Q = homogeneitystatisticfor meaneffect sizes; CL = commonlanguageeffect
size statistic.
*Significanceindicatesrejectionof the hypothesisof homogeneity.
*p < .01.
No effect size was calculatedfor the attitudemeasureof tasknoncontingentrewards
becausetherewere only two studiesthatfit in this category.

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

TABLE6
Effect size as a function of reward contingency (as defined behaviorally) for
expected tangible reward versus control comparisons
Mean
Reward Sampleweighted 95% CI
contingency Analysis k size d for d Q CL
Free time on task:Expectedtangiblerewardversus control
Contingent All knowneffects 18 906 -0.12 -0.26 to 0.01 37.44* .47
Contingent Outliersremoved 16 861 -0.13 -0.26 to 0.01 29.06 .46
Non-
contingentAll knowneffects 40 2017 -0.27 -0.35 to -0.18 167.05*.42
Non- Outliersremoved
contingent using Tukey's
procedure 38 1894 -0.26 -0.35 to -0.16 100.86*.43
Non- Additionaloutliers
contingent removed 34 1728 -0.26 -0.36 to -0.16 54.66 .43

Attitude:Expectedtangiblerewardversuscontrol
Contingent All knowneffects 20 1224 +0.24 0.12 to 0.36 88.64* .57
Contingent Outliersremoved
using Tukey's
procedure 17 1087 +0.11 -0.01 to 0.23 22.24 .53
Non-
contingentAll knowneffects 17 913 -0.04 -0.17 to 0.09 50.14* .49
Non- Outliersremoved
contingent using Tukey's
procedure 16 853 +0.03 -0.10 to 0.17 31.52* .49
Non- Additionaloutliers
contingent removed 15 833 +0.05 -0.08 to 0.19 27.91 .48
Note. Negative effect sizes indicatea decreasein intrinsicmotivationfor reward/
reinforcementgroups;positive effect sizes indicatean increase.k = numberof
effect sizes; samplesize = sum of n in all studies;Meanweightedd = meanof
weighted effect sizes (weighted by sample size); CI= confidence interval;
Q = homogeneitystatisticfor meaneffect sizes; CL = commonlanguageeffect
size statistic.
*Significanceindicatesrejectionof the hypothesisof homogeneity.
*p < .01.

time measurewhen expected tangible rewardsare not contingent on successful

performance.On the attitudemeasure,noncontingentrewardsproduceno signifi-
canteffect. Rewardscontingenton successfulperformancedo not producesignifi-
cant effects on either the free-time or attitudemeasures.
The majordifferencebetween a behavioralclassification of contingency and
Deci and Ryan'scategorizationsystem concernsthose studies where subjectsare
given a rewardfor completingor solving a task. The first experimentconducted
by Deci (1971) is an example of a study coded as task contingent using Deci
andRyan'scategoriesand contingentusing a behavioralframework.In this study,
subjectswere paid money for each puzzle they solved. Deci and Ryan classified
such rewardproceduresas task contingentbecause the rewardswere not contin-
gent on how well subjectsperformedrelativeto some standard.Froma behavioral
390

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

perspective,however, completion or solution of a task is seen as dependenton

successfulperformance;these studieswere labeledcontingent.Thus,performance
contingentrewardsas definedby Deci andRyan(1985) includeonly those studies
where subjects are offered a rewardfor attaininga certainlevel of performance;
using a behavioraldefinition, studies coded as contingentinclude both rewards
thatarecontingenton completingor solving a task andrewardsthatarecontingent
on reachinga specified level of performance.Because these two types of reward
contingenciesmay have oppositeeffects on intrinsicmotivation,a separateanaly-
sis was conducted on studies in which rewardwas delivered for completing or
solving a task. Results given in Table7 show no significantdifferencesbetween
rewardedand control groups on the free-time or attitudemeasuresfor this type
of rewardcontingency.These findings suggest that contingentrewards(which
include performancecontingentrewards),as defined behaviorally,do not harm
intrinsicmotivation.

Summaryof Results From GroupDesigns

A summaryof the variousanalyses conductedon the groupdesign studies and
the major findings is given in Figure 3.8
When all types of reward are aggregated, overall, the results indicate that
rewarddoes not negativelyaffectintrinsicmotivationon any of the four measures
(free time on task once rewardis withdrawn,self-reportsof attitude,performance
duringthe free-time measure,willingness to volunteerfor futurestudies without
reward).When rewardsare subdividedinto rewardtype (verbal,tangible),reward
expectancy(expected,unexpected),and rewardcontingency,the findingsdemon-
stratethat people who receive a verbal rewardspend more time on a task once
the reward is withdrawn;they also show more interest and enjoyment than
nonrewardedpersons.
Tangible reward produces no decrement in intrinsic motivation when it is
received unexpectedly. Expected tangible rewards produce differing effects
dependingon the mannerin which they areadministered.Individualswho receive
an expected rewardfor solving or completing a task or for achieving a specific

TABLE 7
Effectsize as a functionof rewardscontingenton task completionor solution
for expectedtangiblerewardversuscontrolcomparisons
Mean
Sample weighted 95%CI
Measure k size d for d Q CL
Free time 8 423 -0.12 -0.32 to 0.08 11.21 .47
Attitude 6 405 -0.05 -0.25 to 0.14 6.89 .48
Note. Negative effect sizes indicatea decreasein intrinsicmotivationfor reward/
reinforcementgroups;positive effect sizes indicatean increase.k = numberof
effect sizes; samplesize = sum of n in all studies;Meanweightedd = meanof
weighted effect sizes (weighted by sample size); CI = confidence interval;
Q = homogeneitystatisticfor meaneffect sizes; CL = commonlanguageeffect
size statistic.
*Significanceindicatesrejectionof the hypothesisof homogeneity.
*p < .01.
391

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

[INTRINSIC MOTIYATIONJ

/ I I\ Willingness to
Ill

Free Time LAttitude Performacej Volunteer

I I Revud
I
RI
Revard Revai
n.s. (dv 0.14) n.s. n.s.

'0/\\
Verbal Tangible Verbal
I Tangible
(dv = -0.21) (d = 0.39) n.s.
(dw -0.38)
/ \ Unexpected Expected
Unexpected Expected
n.s. (dv - -0.25) Ms. nLs.

/
Contingent
\
Nnc ati agent
/
Contingent Noncontingent
n.s. (dv - -0.26) n.s.
nls.
Contingenton
taskcompletion ontingenton task
or sol ution ccempletionor solution
/^~~n..

Performance Task centingent Task noncontinge :nt Performance Taskcontingent

n.s
conti ngent (dv = -0.23) n.s. contingent
n.s. dvw- 0.19).
FIGURE 3. A summary of the meta-analysis of the effects of reward versus control
groups on intrinsic motivation
Note. dw = mean weightedeffect size (basedon homogeneoussamples);n.s. =
not significant;analysesin regulartype indicateno effect; analysesin bold indicate
a negative effect; underlinedanalyses indicate a positive effect. When no dw is
reported,therewas no significanteffect. No analyseswere conductedon the attitude
measurefortasknoncontingent rewardbecauseonlytwo studiesassessedthismeasure.

level of performancedo not spend less time on a task than controls once the
rewardis withdrawn.They do, however, report more interest, satisfaction,and
enjoymentof the task when the rewardis given for a certainlevel of performance.
The detrimentaleffects of rewardappearwhen rewardsare offered to people
simply for engaging in a task, independentof successful performance.Under
these conditions,once the rewardis removed, individualsspend less time on the
task than controls;they do not, however, reporta less favorableattitudetoward
the task.
RESULTSFROM SINGLE-SUBJECTDESIGNS
To determinethe effects of reinforcementon intrinsicmotivation,an analysis
was conducted on effect sizes from single-subject, repeated measures designs
where the rewards used were shown to be reinforcersfor each subject in the
study.That is, rewardswere shown to increase behaviorduringa reinforcement
phase. An increase or decrease in intrinsicmotivationwas measuredas a differ-
ence betweenbehaviorduringthe pre- andpostreinforcementphases.Five studies
392

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

contributedan effect size to this analysis. Fourstudies showed thatsubjectsspent

more time on the task duringthe postreinforcementphasethanthe baseline phase.
One study (Vasta& Stirpe, 1979) showed a decreasein time on task immediately
following the removalof rewardbut an increasein time when intrinsicmotivation
was measured2 weeks later. To make this analysis comparableto the analysis
of groupdesign studies,however,only differencesbetween the immediatepostre-
inforcementphase and baseline were analyzed.
The average effect size and confidence interval for this analysis was +0.34
(-0.28, 0.96) indicatingno significantchange in intrinsicmotivationfrom base-
line to postreinforcementphases. Effect sizes were homogeneous (Q = 2.96,
df = 4). These results suggest that reinforcementdoes not alter people's intrin-
sic motivation.
As noted previously,two studies used a between- and within-grouprepeated
measures design to assess the effects of reinforcementon intrinsic motivation
(Greene, Sternberg,& Lepper, 1976; Mynatt, Oakley, Arkkelin,Piccione, Mar-
golis, & Arkkelin, 1978). Although these studies did not meet the criteria for
inclusion in the meta-analysisof within-subjectdesigns, it is possible to assess
the within-groupeffects for reward conditions that were comparable in both
studies. Both Greene et al. (1976) and Mynattet al. (1978) included a group of
subjectsrewardedfor playingwith activitiesthatthey hadspentthe most time with
duringbaseline phases (high interestcondition)and a groupthatwas rewardedfor
playing with activities they had spent the least time with during baseline (low
interestcondition).In termsof the high interestconditions,Mynattet al. did not
find a reinforcementeffect but reporteda decrease in intrinsicmotivation from
baseline to postrewardphases. Greene et al. reporteda reinforcementeffect for
the high interest group and a decrease in intrinsic motivationbetween baseline
and postreinforcementsessions. It is difficult to draw conclusionsfrom only two
studies. Nonetheless, because a decline in intrinsicmotivationoccurredwith or
without a reinforcementeffect, it may be that reinforcementis not the critical
variable.Both studies reporteda reinforcementeffect for the low interestcondi-
tions, but there was no change in intrinsicmotivationfrom baseline to postrein-
forcement phases. Again, conclusions based on two studies are tenuous. One
interpretation,however, is that the time spent on low interest activities was so
low that a decline in intrinsic motivation could not be detected. Alternatively,
reinforcementdoes not interruptintrinsic motivationfor low interest activities.
DISCUSSION
A majorcontentionin educationand psychology is thatrewardsand reinforce-
ment negatively impacta person'sintrinsicmotivation.The view is that,if people
are reinforcedor rewardedfor activities they already spend time on and enjoy,
they will be less motivatedto engage in the activity than they were prior to the
introductionof reward,once the rewardis no longer forthcoming.In otherwords,
rewardsand reinforcementare said to decrease people's intrinsic motivation.
Over the past 20 years, dozens of studies have been conductedto investigate
this issue. The primaryobjective of this articlewas to assess the researchfindings
by conducting a meta-analysis of results from experiments on the effects of
rewardand reinforcementon intrinsic motivation.What follows is a discussion
of the results obtainedfrom the meta-analysis.
393

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

The vast majorityof studies have assessed the effects of rewardon intrinsic
motivationby using group designs. Rewardedsubjects are comparedto nonre-
wardedcontrols.Intrinsicmotivationis measuredby differencesbetween groups
on attitude,time spent on a task following the removal of reward(free time),
performanceduringthe free-timeperiod, and willingness to volunteerfor future
studies without reward. The main meta-analysis reported in this article was
conducted on results from these studies. This analysis concernedassessing the
overalleffects of rewardon intrinsicmotivationas well as the effects of a number
of rewardcharacteristics.The resultssuggestthatin the laboratory,overall,reward
does not negatively impact intrinsic motivation on any of the four measures
analyzed here.
A separate analysis was conducted using single-subject, repeated measures
designs. A few researchersemployed this type of design to evaluate the effects
of reinforcementon intrinsicmotivation.The rewardsused in these studies were
shown to be reinforcers,and intrinsicmotivationwas indexed as differencesin
subjects' behavior between pre- and postreinforcementsessions. Results from
the meta-analysisindicate no effect of reinforcementon intrinsic motivation.
That is, the evidence suggests that reinforcementdoes not decrease a person's
intrinsicmotivationto engage in an activity.
In terms of rewardsand extrinsic reinforcement,our overall findings suggest
thatthereis no detrimentaleffect on intrinsicmotivation.These findingsarebased
on laboratoryexperiments,but a similarconclusionwas reachedby Workmanand
Williams (1980) in their review of the effects of extrinsic rewardson intrinsic
motivationin the classroom.Generally,on task behavior,WorkmanandWilliams
found that externalreinforcementincreasedand maintainedintrinsicmotivation
for prolongedperiods (up to 12 months). Thus, it no longer seems appropriate
to argue against the use of incentive systems in applied settings.
The findings from both experimentaland appliedresearchrun contraryto the
views expressedby many psychologistsand educators(e.g., Deci & Ryan, 1985;
Kohn, 1993; Levine & Fasnacht,1974; Schwartz, 1990). For example, Deci and
Ryan (1987) state that:
In general [italicsadded],rewardshave been foundto undermineintrinsic
motivation.When people receivedrewardsfor workingon an interesting
activity,they tendedto displayless interestin and willingnessto workon
thatactivityafterthe terminationof the rewardsthandid people who had
workedon the activitywithoutreceivinga reward.(p.1026)
Results from the present meta-analysissuggest that this statementis erroneous.
The findings indicate that, in general, rewardedpeople are not less willing to
work on activities and they do not display a less favorableattitudetowardtasks
than people who do not receive rewards.
Whenrewardsarebrokendown into rewardtype, expectancy,andcontingency,
resultsindicatethat,on the free-timemeasure,verbalrewardproducesan increase
in intrinsicmotivation;tangiblerewardsproduceno effect when they aredelivered
unexpectedly,andthey are not detrimentalwhen they areexpected andcontingent
on level of performanceor completing or solving a task. Expected tangible
rewardsproducea decrease in intrinsicmotivationas measuredby free time on
task when they are given to individualssimply for engaging in an activity. On
394

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

the attitudemeasure,verbal rewardproducesan increasein intrinsicmotivation,

and tangible rewardsdo not lead to a decrease in intrinsicmotivationunderany
conditions. An increasein intrinsicmotivationis shown on the attitudemeasure
when individualsare offered a rewardfor performingto a set of standards.Thus,
the present results suggest that rewards are detrimentalonly under a highly
specified set of circumstances. That is, when subjects are offered a tangible
reward(expected)thatis deliveredregardlessof level of performance,they spend
less time on a task than control subjects once the rewardis removed. The same
condition has no effect on attitude.
Given these results, why is it that one commonly finds general statements
condemning reinforcementand/or reward in journal articles and introductory
textbooks? The present meta-analysis makes it clear how circumscribedthe
negative effect of rewardreally is. One possibility is that terms such as tangible,
expected, unexpected,contingent and noncontingentbecome very confusing to
a readersortingthroughthis literature.Consider,at its simplest,a studyinvestigat-
ing the effects of expected reward on intrinsic motivation. Suppose the results
showed a negative effect for expected reward.When discussing findings, do the
researcherstalk about the negative effects of the promise of rewardor aboutthe
negative effects of reward,in general?Thereis no doubtthatconclusions reached
from such studies are often made about rewardor reinforcementin general, not
promise of reward.This has led to a great deal of misunderstandingabout the
overall effects of reward and reinforcementon intrinsic motivation. Even an
informed readercan have difficulty keeping in mind what a particularstudy is
investigating. It may be for this reason that rewards are often equated with
reinforcersand, overall, have come to be seen as harmful.It is hoped that the
present meta-analysishas helped to clarify the issue.
TheoreticalImplications
How do results from the present meta-analysisfit in with the various theories
that have been formulatedto account for the negative effects of rewards on
intrinsic motivation?
Advocates of cognitive evaluation theory (e.g., Deci & Ryan, 1985) would
probably not have difficulty reconciling results from the free-time measure of
intrinsic motivation.According to cognitive evaluationtheory, competence and
self-determinationunderlieintrinsicmotivation.Rewardscan facilitateor hinder
competence and self-determinationdependingon whetherthey are perceived as
informational,controlling,or amotivational.From this perspective, results from
the meta-analysiswould suggest that verbalrewardsincreasea person's intrinsic
motivationbecause of their informationalvalue. Verbalpraise would be seen to
lead an individualto feel competentin performinga task;hence, intrinsicmotiva-
tion would increase. Because the cognitive evaluation process is said to take
place while the rewardedactivity is occurring,unexpected rewards would not
altera person'sintrinsicmotivation.On the otherhand,rewardsofferedto people
for participatingin a task, in spite of how well they perform,would be perceived
as controlling and would decrease intrinsicmotivation.
The problemfor cognitive evaluationtheory arises when one considersresults
from the attitudemeasureof intrinsicmotivation.Deci and Ryan (1985) suggest
that interest, enjoyment, and satisfaction are central emotions that accompany
395

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

intrinsic motivation. A person's experience of an activity is a focal point of

cognitive evaluationtheory.In otherwords, cognitive evaluationtheorydepends
on an internalattitudechange thatis laterexpressedbehaviorallyas time on task.
Results from the presentmeta-analysisindicate that rewarddoes not negatively
affect attitude.Individualswho receive verbal praise reportgreaterinterestthan
nonrewardedpeople. Tangiblerewardsproduceno change in attitudewhen they
are given for doing, completing, or solving a task; a positive effect is evident
when rewardsare contingenton a specified level of performance.
One way of mitigatingthe findingsfor cognitive evaluationtheoryis to question
the reliabilityof the attitudemeasure.In many studies, the attitudemeasurewas
obtained from a single-item Likert scale. An additional problem is that the
questions designed to assess attitudetoward the task may have been unable to
separatesubjects' liking of the rewardfrom theirliking of the task. If the attitude
measuresareunreliable,they will fail to reflecttruedifferencesbetweenrewarded
and nonrewardedgroups. This may be one way to handle the puzzling results;
however, it also suggests that there has been no test of the major mediator
proposedby the theory.
The problem of operationalizingthe construct of intrinsic motivation was
recently addressedin a meta-analysisby Wiersma(1992).9 Results from Wiers-
ma's study depended on whether intrinsicmotivation was operationalizedas a
free-time measure or as a task performanceduring rewardedperiod measure.
Free-timemeasuresshowed a decline in intrinsicmotivation;performancemea-
sures showed an increase. As noted, in the present analyses, results from the
attitudemeasuredo not coincide with the free-time measure.Additionally,mea-
sures of intrinsic motivation as performanceduring free time or as willingness
to volunteer for future studies do not clarify the issue of operationalizationof
intrinsicmotivation.
Given the lack of covariationamong the measures, it seems appropriateto
devote furtherresearchto clarifying the concept of intrinsic motivationand to
developing suitable measures. A different solution is offered by Rigby, Deci,
Patrick, and Ryan (1992) who suggest that attention be directed toward the
conceptof self-determinationratherthana pursuitof the intrinsic/extrinsicdichot-
omy. Othersconcurbut suggest thatresearchersshould focus on goal definitions
(Sansone& Morgan, 1992). A final alternativewould be to agree thatconstructs
such as self-determination,goal definition,and intrinsicmotivationare scientifi-
cally unclear and that it would be more appropriateto deal with the effects of
rewardand reinforcementon behavior (e.g., Bandura, 1977, 1986; Dickinson,
1989). Such a course of action would mean abandoningcognitive evaluation
theory.
Another theoretical explanation that has been proposed to account for the
effects of rewardson intrinsicmotivationis the overjustificationeffect (Lepper,
Greene,& Nisbett, 1973). The view is thatpeople's perceptionsaboutthe causes
of their behavior influence future motivation. Rewards lead to a decrease in
intrinsic motivation when people's perceptions shift from accounting for their
behavioras self-initiatedto accountingfor it in termsof externalreward.Because
the present analysis did not evaluate subjects' perceptionsabout the causes of
their behavior, it is impossible to determinewhether overjustificationexplains
396

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

the results. Furtherresearchthat measures subjects' attributionsto internaland

external factors is warranted.
Finally, how would the findings of the meta-analysisbe interpretedfrom a
behavioral perspective? The results from single-subject designs indicate that
reinforcementdoes not producedecrementsin intrinsicmotivation.This finding
is compatiblewith a behavioralview. That is, behavioristsmaintainthatbehavior
returnsto baseline after reinforcementis withdrawn.If the rewardsused in the
groups' design studies are reinforcers,one would expect behaviorto eventually
return to baseline. Research designed to investigate the effects of reward on
intrinsic motivation has typically measuredtime on task for a brief 8- to 10-
minute period, immediately following the removal of reward. Thus, if verbal
praise were a reinforcer,one might interpretthe positive effect as a carryoverof
the reinforcementprocedure.Anotherinterpretationis that the positive effect is
the result of an extinctionburst.That is, when reinforcementis first withdrawn,
the immediate,short-termeffect is that rate of responseincreases.After a period
of time, behavior would returnto baseline. In terms of the negative effect of
expected, noncontingent,tangible reward, some writers (e.g., Dickinson, 1989;
Flora, 1990) have suggested that such a rewardproceduredoes not representa
reinforcementcontingency.The promise of a rewardis seen by behavioristsas
a discriminativestimulus(SD),and the negative effect is understoodas the result
of a bribe. A difficulty with this interpretationis that it does not account for
findings from other conditions where promise of reward does not produce a
negative effect. Furtherresearchis necessary to determinewhen and underwhat
conditionspromisesof rewardsfunctionas bribes.Ourdatasuggest thatpromises
linked to noncontingentreward may function as bribes ratherthan as positive
incentives.
PracticalImplications
The presentfindings suggest that verbalpraise and positive feedback enhance
people's intrinsicinterest.This is an importantfinding. Most social interactionin
business,education,andclinical settingsinvolves verbalfeedbackfrommanagers,
teachers, and therapists.When praise and other forms of positive feedback are
given and laterremoved,people continueto show intrinsicinterestin theirwork.
In contrast to recent claims made by Kohn (1993, p. 55), verbal praise is an
extrinsic motivatorthat positively alters attitudesand behavior.
Rewards can have a negative impact on intrinsic motivation when they are
offered to people for engaging in a task without considerationof any standard
of performance.In a classroom, this might occur if a teacherpromised students
tangiblerewardssimply for doing an activity.For example, a teacherwho prom-
ises stars or other awards to students for spending time doing math problems
may undermineintrinsicmotivation.In such a case, one could expect rewarded
individualsto enjoy the task as much as those who are not offered an incentive.
But, they may spend less time on the activity in a study period when the reward
is no longer forthcoming.According to our results, this would not occur if the
teacherused the samerewardsbut madethemcontingenton successful completion
of the problems.
Overall, the present review suggests that teachers have no reason to resist
implementingincentive systems in the classroom. This conclusion is based on
397

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement, Reward, and Intrinsic Motivation

our findings, which show that verbal praise enhances intrinsic motivation and
thatotherrewardsandreinforcementleave intrinsicmotivationlargelyunaffected.
A small negative effect occurs when tangible rewards are promised without
regardto a standardof performance.Under this circumstance,the promise of
rewardmay act as a bribe.Importantly,on a practicallevel, the implicationis that
rewardofferedin educationalandothersettingsshouldbe deliveredcontingenton
performance.

Notes
'Althoughthere was an overall positive effect of tangiblerewardon intrinsic
motivation,Rosenfield et al.(1980) also found that rewardsthat did not indicate
abilityled to less intrinsicinterest.
2In additionto studies reportedin English, five relevantJapaneseexperiments
were identifiedby the CD-ROMsearch.The informationin the abstractswas not
adequateto code the findings. Therefore,these studies are not included in the
meta-analysis.
3Boggianoand Ruble(1979) reportedthat 147 childrenparticipatedin the study.
Therewere two rewardconditions(taskcontingent,performancecontingent)and a
nonrewarded controlgroup.Thecontrastforthecontrolversustask-contingent reward
groupson the free-timemeasureis reportedas t(130) = 2.0, p < .05; the contrast
for the controlversusperformance-contingent rewardgroupsis reportedas t(130) =
1.16, n.s.
4A copy of the coding formis availableon requestfrom the first author.
5A list of the experimentsincludedin each interactionis availableon requestfrom
the first author.
6Furtheranalyseswhichincludestudiesthatindexeffect size as 0.00 areavailable
in Cameron(1992).
7Thepresentreview does not assess culturaldifferencesin the impactof reward
on intrinsicmotivation.However,it is interestingto note that, althoughthe study
fromIndia(Tripathi& Agarwal,1985)showsan extremepositivevaluefor the effect
of verbalpraise on the free-timemeasure,the directionof the resultis consistent
with the NorthAmericanstudies.
8A few researchershave assessed the effects of expected tangiblerewardson
intrinsicmotivationrelative to unexpectedtangible rewards(e.g., Enzle & Ross,
1978;Fazio, 1981;Lepper& Greene,1975).Otherresearchers haveconductedstudies
comparingexpected noncontingentrewardgroups to expected contingentreward
groups(e.g., Farr,1976; Phillips& Lord,1980;Pinder,1976). Such studiesconcern
directcomparisonsbetweenthe two types of rewardexpectancies(expectedversus
unexpected)andthe two typesof rewardcontingencies(noncontingent versuscontin-
gent)withoutreferenceto a nonrewarded controlgroup.Resultsfrommeta-analyses
conductedon these comparisonsand a list of studiesincludedin such analysescan
be obtainedin Cameron(1992). One significanteffect emergedfromtheseanalyses;
subjectswho receivedan expectedtangiblerewardshowedless intrinsicmotivation
on the free-timemeasurethansubjectswho receivedan unexpectedtangiblereward.
The averageeffect size and confidenceintervalfor this comparisonwas -0.26
(-0.45, -0.06).
9Wiersma(1992) reportedresultsof a meta-analysisof 23 experimentson reward
and intrinsicmotivation.These studiesmake up a subset of those analyzedin the
presentarticle. Effect sizes from Wiersma'sstudy were not always based on a
comparisonof a rewardconditionto a no-rewardcondition.This makesit impossible
to directlycompareour findingswith those of Wiersma.
398

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

APPENDIX A
Studies included in the meta-analysis of group designs
Amabile,T. M., Hennessey,B. A., & Grossman,B. S. (1986). Social influenceson
creativity:The effectsof contracted-for reward.Journalof Personalityand Social
Psychology,50, 14-23.
Anderson,R., Manoogian,S. T., & Reznick,J. S. (1976). The underminingand
enhancingof intrinsicmotivationin preschoolchildren.Journalof Personality
and Social Psychology,34, 915-922.
Anderson,S., & Rodin,J. (1989). Is badnews alwaysbad?Cue andfeedbackeffects
on intrinsicmotivation.Journalof AppliedSocial Psychology,19, 449-467.
Arkes, H. R. (1979). Competenceand the overjustificationeffect. Motivationand
Emotion,3, 143-150.
Arnold,H. J. (1976). Effects of performancefeedbackand extrinsicrewardupon
high intrinsicmotivation.OrganizationalBehavior and Human Performance,
17, 275-288.
Arnold,H. J. (1985).Taskperformance,perceivedcompetence,andattributedcauses
of performanceas determinants of intrinsicmotivation.Academyof Management
Journal,28, 876-888.
Blanck,P. D., Reis, H. T., & Jackson,L. (1984). The effects of verbalreinforcement
of intrinsicmotivationfor sex-linkedtasks.Sex Roles, 10, 369-386.
Boal, K. B., & Cummings,L. L. (1981). Cognitiveevaluationtheory:anexperimental
testof processesandoutcomes.Organizational BehaviorandHumanPerformance,
28, 289-310.
Boggiano,A. K., Harackiewicz,J. M., Besette,J. M., Main,D. S. (1985). Increasing
children'sinterestthroughperformancecontingentreward.Social Cognition,3,
400-411.
Boggiano,A. K., & Hertel,P.T. (1983). Bonusesandbribes:moodeffectsin memory.
Social Cognition,2, 49-61.
Boggiano,A. K., Ruble,D. N., & Pittman,T. S. (1982). The masteryhypothesisand
the overjustificationeffect. Social Cognition,1, 38-49.
Brennan,T. P., & Glover,J. A. (1980). An examinationof the effect of extrinsic
reinforcerson intrinsicallymotivatedbehavior:experimentalandtheoretical.Social
Behaviorand Personality,8, 27-32.
Broekner,J., & Vasta,R. (1981).Do causalattributions mediatetheeffectsof extrinsic
rewardson intrinsicinterest?Journalof Researchin Personality,15, 201-209.
Butler,R. (1987). Task-involvingandego-involvingpropertiesof evaluation:Effects
of differentfeedbackconditionson motivationalperceptions,interest,and perfor-
mance.Journalof EducationalPsychology,79, 474-482.
Calder,B. J., & Staw,B. M. (1975). Self-perceptionof intrinsicandextrinsicmotiva-
tion. Journalof Personalityand Social Psychology,31, 599-605.
Crino,M. D., & White,M. C. (1982). Feedbackeffects in intrinsic/extrinsic reward
paradigms.Journalof Management,8, 95-108.
Daniel, T. L., & Esser,J. K. (1980). Intrinsicmotivationas influencedby rewards,
task interest,and task structure.Journalof AppliedPsychology,65, 566-573.
Danner,E W., & Lonkey,E. (1981). A cognitive developmentalapproachto the
effects of rewardson intrinsicmotivation.ChildDevelopment,52, 1043-1052.
Deci, E. L. (1971). Effects of externallymediatedrewardson intrinsicmotivation.
Journalof Personalityand Social Psychology,18, 105-115.
Deci, E. L. (1972a).Intrinsicmotivation,extrinsicreinforcement, andinequity.Jour-
nal of Personalityand Social Psychology,22, 113-120.
399

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

Deci, E. L. (1972b). The effects of contingent and noncontingent rewardsand controls

on intrinsic motivation. Organizational Behavior and Human Performance, 8,
217-229.
DeLoach, L. L., Griffith, K., & LaBarba, R. C. (1983). The relationship of group
context and intelligence to the overjustification effect. Bulletin of the Psychonomic
Society, 21, 291-293.
Dollinger, S. J., & Thelen, M. H. (1978). Overjustification and children's intrinsic
motivation: comparative effects of four rewards. Journal of Personality and Social
Psychology, 36, 1259-1269.
Earn, B. M. (1982). Intrinsic motivation as a function of extrinsic financial rewards
and subjects' locus of control. Journal of Personality, 50, 360-373.
Fabes, R. A. (1987). Effects of reward contexts on young children's task interest.
Journal of Psychology, 121, 5-19.
Fabes, R. A., Eisenberg, N., Fultz, J., & Miller, P. (1988). Reward, affect and young
children's motivational orientation. Motivation and Emotion, 12, 155-169.
Freedman, S. M., & Phillips, J. S. (1985). The effects of situational performance
constraints on intrinsic motivation and satisfaction: the role of perceived compe-
tence and self-determination. Organizational Behavior and Human Decision Pro-
cesses, 35, 397-416.
Greene, D., & Lepper, M. R. (1974). Effects of extrinsic rewards on children's
subsequent intrinsic interest. Child Development, 45, 1141-1145.
Griffith, K. M., DeLoach, L. L., & LaBarba, R. C. (1984). The effects of rewarder
familiarity and differential reward preference in intrinsic motivation. Bulletin of
the Psychonomic Society, 22, 313-316.
Hamner, W. C., & Foster, L. W. (1975). Are intrinsic and extrinsic rewards additive:
A test of Deci's cognitive evaluation theory of task motivation. Organizational
Behavior and Human Performance, 14, 398-415.
Harackiewicz, J. M. (1979). The effects of reward contingency and performance
feedback on intrinsic motivation. Journal of Personality and Social Psychology,
37, 1352-1363.
Harackiewicz, J. M., Abrahams, S., & Wageman, R. (1987). Performance evaluation
and intrinsic motivation: The effects of evaluative focus, rewards, and achievement
orientation. Journal of Personality and Social Psychology, 53, 1015-1023.
Harackiewicz, J. M., & Manderlink, G. (1984). A process analysis of the effects of
performance-contingentrewards on intrinsic motivation. Journal of Experimental
Social Psychology, 20, 531-551.
Harackiewicz, J. M., Manderlink, G., & Sansone, C. (1984). Rewarding pinball
wizardry: effects of evaluation and cue value on intrinsic interest. Journal of
Personality and Social Psychology, 47, 287-300.
Hom, H. L. (1987). A methodological note: time of participationeffects on intrinsic
motivation. Personality and Social Psychology Bulletin, 13, 210-215.
Kariol, R., & Ross, M. (1977). The effect of performance relevant and performance
irrelevant rewards on children's intrinsic motivation. Child Development, 48,
482-487.
Koestner,R., Zuckerman,M., & Koestner,J. (1987). Praise, involvement, and intrinsic
motivation. Journal of Personality and Social Psychology, 53, 383-390.
Kruglanski, A. W., Alon, S., & Lewis, T. (1972). Retrospective misattribution and
task enjoyment. Journal of Experimental Social Psychology, 8, 493-501.
Kruglanski,A. W., Friedman,I., & Zeevi, G. (1971). The effects of extrinsic incentive
on some qualitative aspects of task performance. Journal of Personality, 39,
606-617.

400

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

Kruglanski,A. W., Riter,A., Amitai,A., Margolin,B. S., Shabatai,L., & Zaksh,D.

(1975).Canmoneyenhanceintrinsicmotivation?: A testof thecontent-consequence
hypothesis. Journal of Personality and Social Psychology, 31, 744-750.
Lepper,M. R., Greene,D., & Nisbett,R. E. (1973). Underminingchildren'sintrinsic
interestwithextrinsicreward:A test of the "overjustification"
hypothesis.Journal
of Personality and Social Psychology, 28, 129-137.
Loveland,K. K., & Olley,J. G. (1979). The effectof externalrewardon interestand
qualityof taskperformancein childrenof high andlow intrinsicmotivation.Child
Development, 50, 1207-1210.
Luyten,H., & Lens,W.(1981).Theeffectof earlierexperienceandrewardcontingen-
cies on intrinsic motivation. Motivation and Emotion, 5, 25-36.
McGraw,K. O., & McCullers,J. C. (1979). Evidenceof'a detrimentaleffect of
extrinsic incentiveson breakinga mental set. Journal of ExperimentalSocial
Psychology, 15, 285-294.
McLoyd,V. C. (1979). The effects of extrinsicrewardsof differentialvalue on high
and low intrinsicinterest.ChildDevelopment,50, 1010-1019.
Morgan,M. (1981).The overjustification effect:A developmentaltest of self-percep-
tion interpretations.Journal of Personality and Social Psychology, 40, 809-821.
Morgan,M. (1983). Decrementsin intrinsicinterestamongrewardedand observer
subjects. Child Development, 54, 636-644.
Mynatt,C., Oakley,D., Piccione,A., Margolis,R., & Arkkelin,J. (1978).An examina-
tion of overjustificationunderconditionsof extendedobservationand multiple
reinforcement: Overjustification or boredom? Cognitive Therapy and Research,
2, 171-177.
Ogilvie, L., & Prior,M. (1982). The overjustificationeffect in retardedchildren:
durability and generalizability. Australia and New Zealand Journal of Develop-
mental Disabilities, 8, 213-218.
Orlick,T. D., & Mosher,R. (1978). Extrinsicawardsand participantmotivationin
a sport related task. International Journal of Sport Psychology, 9, 27-39.
Palack,S. R., Costomotis,S., Sroka,S., & Pittman,T. S. (1982). School experience,
reward characteristics,and intrinsic motivation. Child Development, 53,
1382-1391.
Pittman,T. S., Cooper,E. E., & Smith,T. W. (1977). Attributionof causalityand
the overjustification effect. Personality and Social Psychology Bulletin, 3, 280-283.
Pittman,T. S., Davey,M. E., Alafat,K. A., Wetherill,K. V., & Kramer,N. A. (1980).
Informationalversus controlling verbal rewards.Personality and Social Psychology
Bulletin, 6, 228-233.
Pittman,T. S., Emery,J., & Boggiano,A. K. (1982). Intrinsicandextrinsicmotiva-
changesin preferencefor complexity.Journal
tionalorientations:reward-induced
of Personality and Social Psychology, 42, 789-797.
Inducingintrinsic
Porac,J. F, & Meindl,J. (1982). Underminingoverjustification:
and extrinsic task representations. Organizational Behavior and Human Perfor-
mance, 29, 208-226.
Pretty,G. H., & Seligman,C. (1984). Affect andthe overjustification
effect. Journal
of Personality and Social Psychology, 46, 1241-1253.
competingresponses,and
Reiss, S., & Sushinsky,L. W. (1975). Overjustification,
the acquisition of intrinsic interest. Journal of Personality and Social Psychology,
31, 1116-1125.
Rosenfield,D., Folger,R., & Adelman,H. F (1980). Whenrewardsreflectcompe-
tence: A qualificationof the overjustificationeffect. Journalof Personalityand
Social Psychology, 39, 368-376.

401

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

Ross, M. (1975). Salience of reward and intrinsic motivation. Journal of Personality

and Social Psychology, 32, 245-254.
Ross, M., Kario, R., & Rothstein, M. (1976). Reward contingency and intrinsic
motivation in children: a test of the delay of gratification hypothesis. Journal of
Personality and Social Psychology, 33, 442-447.
Ryan, R. M., Mims, B., & Koestner, R. (1983). Relation of reward contingency and
interpersonal context to intrinsic motivation: A review and test using cognitive
evaluation theory. Journal of Personality and Social Psychology, 45, 736-750.
Salinick, G. R. (1975). Interactioneffects of performance and money on self-percep-
tion of intrinsic motivation. Organizational Behavior and Human Performance,
13, 339-351.
Sansone, C. (1986). A question of competence: the effects of competence and task
feedback on intrinsic interest. Journal of Personality and Social Psychology, 51,
918-931.
Sansone, C. (1989). Competence feedback, task feedback, and intrinsic interest: An
examination of process and context. Journal of Experimental Social Psychology,
25, 343-361.
Sansone, C., Sachau, D. A., & Weir, C. (1989). Effects of instruction on intrinsic
interest:The importance of context. Journal of Personality and Social Psychology,
57, 819-829.
Sarafino, E. P. (1984). Intrinsic motivation and delay of gratification in preschoolers:
the variables of reward salience and length of expected delay. British Journal of
Developmental Psychology, 2, 149-156.
Shanab, M. E., Peterson, D., Dargahi, S., & Deroian, P. (1981). The effects of positive
and negative verbal feedback on the intrinsic motivation of male and female
subjects. The Journal of Social Psychology, 115, 195-205.
Shapira, Z. (1976). Expectancy determinants of intrinsically motivated behavior.
Journal of Personality and Social Psychology, 34, 1235-1244.
Smith, T. W., & Pittman, T. S. (1978). Reward, distraction, and the overjustification
effect. Journal of Personality and Social Psychology, 36, 565-573.
Staw, B. M., Calder, B. J., Hess, R. K., & Samdelands, L. E. (1980). Intrinsic
motivation and norms about payment. Journal of Personality, 48, 1-14.
Swann, W. B., Jr., & Pittman, T. S. (1977). Moderating influence of verbal cues on
intrinsic motivation. Child Development, 48, 1128-1132.
Taub, S. I., & Dollinger, S. J. (1975). Reward and purpose as incentives for children
differing in locus of control expectancies. Journal of Personality, 43, 179-195.
Tripathi, K. N., & Agarwal, A. (1985). Effects of verbal and tangible rewards on
intrinsic motivation in males and females. Psychological Studies, 30, 77-84.
Tripathi, K. N., & Agarwal, A. (1988). Effect of reward contingency on intrinsic
motivation. The Journal of General Psychology, 115 (3), 241-246.
Vallerand,R. J. (1983). The effect of differential amounts of positive verbal feedback
on the intrinsic motivation of male hockey players. Journal of Sport Psychology,
5, 100-107.
Vallerand, R. J., & Reid, G. (1984). On the causal effects of perceived competence
on intrinsic motivation: A test of cognitive evaluation theory. Journal of Sport
Psychology, 6, 94-102.
Weinberg, R. S., & Jackson, A. (1979). Competition and extrinsic rewards: Effect
on intrinsic motivation and attribution.Research Quarterly, 50, 494-502.
Weiner,M. J. (1980). The effect of incentive and control over outcomes upon intrinsic
motivation and performance. The Journal of Social Psychology, 112, 247-254.
Weiner, M. J., & Mander, A. M. (1978). The effects of reward and perception of
competency upon intrinsic motivation. Motivation and Emotion, 2, 67-73.
402

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

Wicker,F. W., Brown,G., Wiehe,J. A., & Shim, W. Y. (1990). Moods, goals, and
measures of intrinsic motivation. The Journal of Psychology, 124, 75-86.
behaviorconstraintandthe overjustification
Williams,B. W.,(1980). Reinforcement,
effect. Journal of Personality and Social Psychology, 39, 599-614.
Wimperis,B. R., & Farr,J. L. (1979). The effects of task content and reward
contingencyupon task performanceand satisfaction.Journal of AppliedSocial
Psychology, 9 (3), 229-249.
Zinser,O., Young,J. G., & King, P. E. (1982). The influenceof verbalrewardon
intrinsic motivation in children. The Journal of General Psychology, 106, 85-91.

Studies Included in the Meta-Analysis of Single-Subject Designs

Davidson,P.,& Bucher,B. (1978).Intrinsicinterestandextrinsicreward:The effects
of a continuingtokenprogramon continuingnonconstrained preference.Behavior
Therapy, 9, 222-234.
Feingold,B. D., & Mahoney,M. J. (1975).Reinforcementeffectson intrinsicinterest:
Underminingthe overjustificationhypothesis.BehaviorTherapy,6, 357-377.
Mawhinney,T. C., Dickinson,A. M., & Taylor,L. A. (1989). The use of concurrent
schedulesto evaluatethe effects of extrinsicrewardson "intrinsicmotivation."
Journal of Organizational Behavior Management, 10, 109-129.
Vasta,R., Andrews,D. E., McLaughlin,A. M., Stirpe,L. A., & Comfort,C. (1978).
Reinforcement effects on intrinsicinterest:A classroomanalog.Journalof School
Psychology, 16, 161-168.
Vasta,R., & Stirpe,L. A. (1979).Reinforcement
effectson threemeasuresof children's
interest in math. Behavior Modification, 3, 223-244.

403

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

APPENDIXB
Formulas for calculating effect size, g
1.
XE- Xc
g
where
XE= mean of experimental group
Xc = mean of control group
Sp= pooled standarddeviation
(nE - 1)S2 + (nc - 1)S
nE+ nc-2

where
25= pooled variance
SE = varianceof experimentalgroup
S2c= varianceof controlgroup
nE= sample size of experimental group
nc = sample size of control group

2.
g = t - for equal ns; n = sample size of each group
n

3.
g = t-+ - for unequalns
./nE nc

4.
E+ nC
V ^
nEnc

404

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 APPENDIX C
Characteristics of studies included in the meta-analysis
Reward Expect- Contin-
Author(s) Journal Design Subjects Task type ancy gency
Deci (1971) JPSP B/A Adults Soma T E Cont, TC
Exp. 1
Deci (1971) JPSP B/A Adults Soma T E Cont, TC
Exp. 1
Deci (1971) JPSP Field study Adults Writing T E Cont. TC
Exp. 2 headlines
Deci (1971) JPSP B/A Adults Soma V U
Exp. 3
Deci (1971) JPSP B/A Adults Soma V U
Exp. 3
Kruglanskiet J of Pers. A/O 15-16 yrs Creativity T E Not, TC
al. (1971) & recall
Kruglanskiet J of Pers. A/O 15-16 yrs Creativity T E Not, TC
al. (1971) & recall
Deci (1972a) JPSP A/O Adults Soma V U
Deci (1972a) JPSP A/O Adults Soma T E Cont, TC
Deci (1972a) JPSP A/O Adults Soma T E Cont, TC
Deci (1972b) Org Beh & A/O Adults Soma T E Not, NC
Hum Perf
Kruglanskiet J. Exp. Soc A/O Children 5 games T U
al. (1972) Psych
Lepperet al. JPSP B/A Children Drawing T E Not, TC
(1973)
Lepperet al. JPSP B/A Children Drawing T U
(1973)
Greene, Child dev A/O Children Drawing T E Not, TC
Lepper(1974)
O Greene, Child dev A/O Children Drawing T U
,t Lepper(1974)

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
O

APPENDIXC-continued
Characteristicsof studies included in the meta-analysis
Reward Expect- Contin-
Author(s) Journal Design Subjects Task type ancy gency
Greene, Child dev A/O Children Drawing T U
Lepper(1974)
Ross (1975) JPSP A/O Children Playing T E Not, TC
Exp. 1 drum
Ross (1975) JPSP A/O Children Playing T E Not, TC
Exp. 1 drum
Ross (1975) JPSP A/O Children Drum T E Not, TC
Exp. 2
Taub, J of Pers A/O Children Coding T E Cont, PC
Dollinger
(1975)
Kruglanskiet JPSP A/O 14-15-yr.- 2 tasks T E Cont, PC
al. (1975) olds
Exp. 1
Kruglanskiet JPSP A/O 15-16-yr.- 2 tasks T E Cont, PC
al. (1975) olds
Exp. 2
Reiss, JPSP A/O Children Listening T E Not, TC
Sushinski to songs
(1975)
Salanick(1975) Org Beh & A/O Adults Train E Cont, PC
Hum Perf game
Salanick (1975) Org Beh & A/O Adults Train E Cont, PC
Hum Perf game
Hamner,Foster Org Beh & A/O Adults Scoring E Not, NC
(1975) Hum Perf questions
Hamner,Foster Org Beh & A/O Adults Scoring E Cont, TC
(1975) Hum Perf questions
Calder,Staw JPSP AO/ Adults Puzzles E Not, TC
(1975)

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Calder,Staw JPSP A/O Adults Puzzles T E Not, TC
(1975)
Feingold, Behavior SS Children Dot-to- T E Cont
Mahoney Therapy Repeated dot
(1975) measures connections
Andersonet al. JPSP B/A Children Drawing V U
(1976)
Andersonet al. JPSP B/A Children Drawing T E Not, T
(1976)
Arnold (1976) Org Beh & Multiple Adults Computer T E Not, T
Hum Perf trials game
Arnold (1976) Org Beh & Multiple Adults Computer T E Not, T
Hum Perf trials game
Ross et al. JPSP A/O Children Drawing T E Not, T
(1976)
Ross et al. JPSP A/O Children Drawing T E Not, N
(1976)
Shapira(1976) JPSP A/O Adults Soma T E Cont, P
Swann, Pittman Child Dev A/O Children Drawings T E Not, N
(1977) Exp. 1
Swann, Pittman Child Dev A/O Children Drawing T E Not, N
(1977) Exp. 1
Swann, Pittman Child Dev A/O Children Drawing T E Not, T
(1977) Exp. 2
Karniol,Ross Child Dev A/O Children Slide T E Not, T
(1977) show
Kariol, Ross Child Dev A/O Children Slide T E Cont, P
(1977) show
Pittmannet al. Per & Soc A/O Adults Gravitation T E Cont, P
(1977) Psy Bull
Mynattet al. Cog Ther & B/A mult. Children Educ T E Not, T
(1978) Res trials games
Weiner, Mot & A/O Adults Decoding T E Not, T
Mander(1978) Emotion cartoons
Weiner, Mot & A/O Adults Decoding T E Cont, P
Mander(1978) Emotion cartoons

4:,
0

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
o APPENDIX C-continued
oo
Characteristics of studies included in the meta-analysis
Reward Expect- Contin-
Author(s) Journal Design Subjects Task type ancy gency
Weiner, Mot & A/O Adults Decoding T E Not, TC
Mander(1978) Emotion cartoons
Weiner, Mot & A/O Adults Decoding T E Cont, PC
Mander(1978) Emotion cartoons
Weiner, Mot & A/O Adults Decoding T E Not, TC
Mander(1978) Emotion cartoons
Weiner, Mot & A/O Adults Decoding T E Cont, PC
Mander(1978) Emotion cartoons
Orlick,Mosher Int J. of B/A Children Stabilometer T E Cont, TC
(1978) Sport Psy
Orlick,Mosher Int J. of B/A Children Stabilometer T U
(1978) Sport Psy
Orlick,Mosher Int J. of B/A Children Stabilometer V U
(1978) Sport Psy
Smith, Pittman JPSP A/O Adults Labyrinth T E Cont, TC
(1978)
Smith, Pittman JPSP A/O Adults Labyrinth T E Cont, TC
(1978)
Dollinger, JPSP A/O Children Mazes T&V E Both
Thelan (1978)
Davidson, Behavior SS Children Playing T E Not
Bucher (1978) Therapy Repeated with
measures clown
Vastaet al. J of School SS Children Coloring T&V U
(1978) Psych Repeated
measures
Arkes (1979) Mot & A/O Adults Soma T E Cont, TC
Emotion
Arkes (1979) Mot & A/O Adults Soma T E Cont, TC
Emotion

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Loveland, Child Dev A/O Children Drawing T E Not, T
Olley (1979)
Harackiewicz JPSP B/A 16-yr.-olds Hidden V U
(1979) puzzles
Harackiewicz JPSP B/A 16-yr.-olds Hidden T E Not, T
(1979) puzzles
McLoyd (1979) Child Dev A/O Children Reading T E Cont, T
books
McLoyd (1979) Child Dev A/O Children Reading T E Cont, T
books
Wimperis,Farr J. Applied A/O Adults Erector T E Not, T
(1979) Soc Psych sets
Wimperis,Farr J. Applied A/O Adults Erector T E Cont, P
(1979) Soc Psych sets
Wimperis,Farr J. Applied A/O Adults Erector T E Both
(1979) Soc Psych sets
Weinberg, Research A/O Adults Stabilometer T E Cont, P
Jackson Quarterly
(1979)
McGraw, J Exp Soc A/O Adults Waterjar T E Cont, P
McCullers Psych problem
(1979)
McGraw, J Exp Soc A/O Adults Waterjar T E Cont, P
McCullers Psych problem
(1979)
Vasta,Stirpe Behavior SS Children Math T E Not
(1979) Mod Repeated problems
measures
Brennan, Soc Beh & B/A Adults Soma T E Not, N
Glover (1980) Pers
Weiner(1980) J of Soc A/O Adults Anagrams T E Cont, P
Psych
Weiner(1980) J of Soc A/O Adults Anagrams T E Cont, P
Psych
Weiner(1980) J of Soc A/O Adults Anagrams T E Cont, P
Psych

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
APPENDIX C-continued
Characteristics of studies included in the meta-analysis
Reward Expect- Contin-
Author(s) Journal Design Subjects Task type ancy gency
Rosenfieldet JPSP A/O Adults Ad Lib V E
al. (1980)
Rosenfieldet JPSP A/O Adults Ad Lib V E
al. (1980)
Rosenfieldet JPSP A/O Adults Ad Lib V E
al. (1980)
Rosenfieldet JPSP A/O Adults Ad Lib T E Cont, PC
al. (1980)
Rosenfieldet JPSP A/O Adults Ad Lib T E Cont, PC
al. (1980)
Rosenfieldet JPSP A/O Adults Ad Lib T E Cont, PC
al. (1980)
Staw et al. J of Pers A/O Adults Puzzles T E Not, TC
(1980)
Staw et al. J of Pers A/O Adults Puzzles T E
(1980)
Williams JPSP B/A Children 4 games T E Not, TC
(1980)
Williams JPSP B/A Children 4 games T E Not, TC
(1980)
Daniel, Esser J Applied A/O Adults Puzzles T E Cont, TC
(1980) Psych
Daniel, Esser J Applied A/O Adults Puzzles T E Cont, TC
(1980) Psych
Daniel, Esser J Applied A/O Adults Puzzles T E Cont, TC
(1980) Psych
Morgan(1981) JPSP A/O Children Puzzles T E Not, TC
Exp. 1
Morgan(1981) JPSP A/O Children Puzzles T E Not, TC
Exp. 1

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Morgan (1981) JPSP A/O Children Puzzles T E Not, T
Exp. 2
Morgan (1981) JPSP A/O Children Puzzles T E Not, T
Exp. 2
Brockner, J of Res in A/O Adults Soma T E Cont,
Vasta (1981) Pers
Brockner, J of Res in A/O Adults Soma T E Cont,
Vasta (1981) Pers
Pittmanet al. Pers & Soc A/O Adults Soma V U
(1980) Psych Bull
Shanabet al. J of Soc A/O Adults Soma V U
(1981) Psych
Shanabet al. J of Soc A/O Adults Soma V U
(1981) Psych
Danner, Child Dev A/O Children Class T E Not, T
Lonkey (1981) inclusion
Danner, Child Dev A/O Children Class T E Not, T
Lonkey (1981) inclusion
Danner, Child Dev A/O Children Class V U
Lonkey (1981) inclusion
Danner, Child Dev A/O Children Class V U
Lonkey (1981) inclusion
Boal, Org Beh & Field study Adults Coding T E Not, N
Cummings Hum Perf data
(1981)
Boal, Org Beh & Field study Adults Coding T E Cont,
Cummings Hum Perf data
(1981)
Luyten, Lens Mot & A/O Adults Wood T E Not, T
(1981) Emotion models
Luyten, Lens Mot & A/O Adults Wood T E Not, T
(1981) Emotion models
Luyten, Lens Mot & A/O Adults Wood T E Not, T
(1981) Emotion models
Luyten, Lens Mot & A/O Adults Wood T E Cont,
(1981) Emotion models

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
APPENDIX C-continued
Characteristics of studies
Characteristics of studies included in the
included in the meta-analysis
meta-analysis
Reward Expect- Contin-
Author(s) Journal Design Subjects Task type ancy gency
Luyten,Lens Mot & A/O Adults Wood T E Cont, PC
(1981) Emotion models
Luyten,Lens Mot & A/O Adults Wood T E Cont, PC
(1981) Emotion models
Fabes et al. Am. J Psych A/O Adults Algorithms T E All
(1981) heuristic
tasks
Boggiano et al. Social A/O Children Hidden T E Not, TC
(1982) Cognition pictures
Zinseret al. J General A/O Children Hidden V U
(1982) Psych pictures
Porac, Meindl Org Beh & A/O Adults Soma T E Not, TC
(1982) Hum Perf
Earn(1982) J of Pers A/O Adults Anagrams T E Not, TC
Earn(1982) J of Pers A/O Adults Anagrams T E Not, TC
Pittmanet al. JPSP A/O Children Matching T E Not, NC
(1982) Exp. 1 games
Pittmanet al. JPSP A/O Children Matching T E Not, TC
(1982) Exp. 1 games
Pittmanet al. JPSP A/O Children Matching T E Not, TC
(1982) Exp. 1 games
Pittmanet al. JPSP A/O Children Drawing T E Not, TC
(1982) Exp. 2
Pallacket al. Child Dev A/O Children Drawing V U
(1982)
Pallacket al. Child Dev A/O Children Drawing V E
(1982)
Pallacket al. Child Dev A/O Children Drawing T U
(1982)

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Pallack et al. Child Dev A/O Children Drawing T E
(1982)
Crino, White J A/O Adults Puzzles V U
(1982) Management
Crino, White J A/O Adults Puzzles V U
(1982) Management
Crino, White J A/O Adults Puzzles V U
(1982) Management
Crino, White J A/O Adults Puzzles V U
(1982) Management
Ogilvie, Prior Aust & N.Z. B/A Children Drawing T E Not, T
(1982) J Dev. Dis.
Boggiano, Social A/O Adults Memory T U
Hertel (1983) Cognition task
Ryan et al. JPSP A/O Adults Hidden T E Cont, P
(1983) puzzles
Ryan et al. JPSP A/O Adults Hidden T E Cont, P
(1983) puzzles
Ryan et al. JPSP A/O Adults Hidden T E Not, T
(1983) puzzles
Ryan et al. JPSP A/O Adults Hidden T E Not, T
(1983) puzzles
Ryan et al. JPSP A/O Adults Hidden V E
(1983) puzzles
Ryan et al. JPSP A/O Adults Hidden V E
(1983) puzzles
Morgan(1983) Child Dev A/O Children Puzzles T E Not, T
Exp. 1
Morgan(1983) Child Dev A/O Children Puzzles T E Not, T
Exp. 1
Morgan(1983) Child Dev A/O Children Puzzles T E Not, T
Exp. 2
Morgan(1983) Child Dev A/O Children Puzzles T E Not, T
Exp. 2
Vallerand J Sport Psych A/O Children Slideshow V E
(1983) game

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
APPENDIXC-continued
Characteristicsof studies included in the meta-analysis
Reward Expect- Contin-
Author(s) Journal Design Subjects Task type ancy gency
DeLoach et al. Bull Psych B/A Children Connect T E Not, TC
(1983) Society dots
Blanck et al. Sex Roles A/O Adults Word V U
(1984) game
Blanck et al. Sex Roles A/O Adults Word V U
(1984) game
Sarafino(1984) Br. J Dev A/O Children Riddles T E Not, TC
Psych
Sarafino(1984) Br. J Dev A/O Children Riddles T Not, TC
Psych
Harackiewicz J Exp. Psych A/O 16-yr.-olds Hidden T E Cont, PC
et al. (1984) puzzles
Griffithet al. Bull Psych A/O Children Reading T E Not, TC
(1984) Society books
Griffithet al. Bull Psych A/O Children Reading T E Not, TC
(1984) Society books
Pretty, JPSP B/A Adults Soma T E Not, TC
Seligman
(1984) Exp. 1
Pretty, JPSP B/A Adults Soma T E Not, TC
Seligman
(1984) Exp. 1
Pretty, JPSP B/A Adults Soma T U
Seligman
(1984) Exp. 1
Pretty, JPSP B/A Adults Soma T U
Seligman
(1984) Exp. 1

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Pretty, JPSP B/A Adults Soma V U
Seligman
(1984) Exp. 1
Pretty, JPSP B/A Adults Soma V U
Seligman
(1984) Exp. 1
Pretty, JPSP B/A Adults Soma T E Not, T
Seligman
(1984) Exp. 2
Pretty, JPSP B/A Adults Soma T E Not, T
Seligman
(1984) Exp. 2
Pretty, JPSP B/A Adults Soma T U
Seligman
(1984) Exp. 2
Pretty, JPSP B/A Adults Soma T U
Seligman
(1984) Exp. 2
Harackiewicz JPSP B/A Adults Pinball T E Cont,
et al. (1984)
Exp. 1
Harackiewicz JPSP B/A Adults Pinball T E Cont,
et al. (1984)
Exp. 1
Harackiewicz JPSP B/A Adults Pinball T E Cont,
et al. (1984)
Exp. 1
Harackiewicz JPSP B/A Adults Pinball T E Cont,
et al. (1984)
Exp. 2
Harackiewicz JPSP B/A Adults Pinball T E Cont,
et al. (1984)
Exp. 2

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
4 APPENDIX C-continued
Characteristicsof studies included in the meta-analysis
Reward Expect- Contin-
Author(s) Journal Desiggn Subjects Task type ancy gency
Harackiewicz JPSP B/A Adults Pinball T U
et al. (1984)
Exp. 2
Harackiewicz JPSP B/A Adults Pinball T U
et al. (1984)
Exp. 2
Harackiewicz JPSP B/A Adults Pinball T E Cont, PC
et al. (1984)
Exp. 3
Harackiewicz JPSP B/A Adults Pinball T E Cont, PC
et al. (1984)
Exp. 3
Vallerand, J Sport Psych B/A Adults Stabilometer V E
Reid (1984)
Arnold (1985) Acad. Man. B/A Adults Computer T E Both
J. game
Boggiano et al. Social A/O Children Puzzles T E Not, TC
(1985) Cognition
Boggiano et al. Social A/O Children Puzzles T E Cont, PC
(1985) Cognition
Freedman, Org Beh & A/O Adults Proof T E Not, TC
Phillips (1985) Hum Dec P reading
Freedman, Org Beh & A/O Adults Proof T E Cont, PC
Phillips (1985) Hum Dec P reading
Tripathi, Psych A/O Adults Puzzles T E Not, TC
Agarwal Studies
(1985)
Tripathi, Psych A/O Adults Puzzles T E Not, TC
Agarwal Studies
(1985)

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Tripathi, Psych A/O Adults Puzzles T E Not, T
Agarwal Studies
(1985)
Tripathi, Psych A/O Adults Puzzles V E
Agarwal Studies
(1985)
Tripathi, Psych A/O Adults Puzzles V E
Agarwal Studies
(1985)
Tripathi, Psych A/O Adults Puzzles V E
Agarwal Studies
(1985)
Sansone (1986) JPSP A/O Adults Identify V U
Exp. 1 names
Amabile et al. JPSP A/O Children 3 tasks T E Not, T
(1986) Exp.1
Amabile et al. JPSP A/O Children 3 tasks T E Not, T
(1986) Exp. 1
Amabile et al. JPSP A/O Adults 3 tasks T E Not, T
(1986) Exp.3
Harackiewicz JPSP A/O 16-yr.-olds Puzzles T E Cont, P
et al. (1987)
Hom (1987) Pers & Soc A/O Adults Pursuit T ? Not
Exp. 1 Psych Bull rotor task
Hom (1987) Pers & Soc A/O Adults Pursuit T ? Not
Exp. 1 Psych Bull rotor task
Hor (1987) Pers & Soc A/O Adults Solving V ? ?
Exp. 2 Psych Bull anagrams
Fabes (1987) J of Psych A/O Children Block T E Not, T
Exp. 1 building
Fabes (1987) J of Psych A/O Children Block T E Cont, P
Exp. 1 building
Fabes (1987) J of Psych A/O Children Block T E Not, T
Exp. 2 building

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
4 APPENDIX C-continued
Characteristics of studies included in the meta-analysis
Reward Expect- Contin-
Author(s) Journal Design Subjects Task type ancy gency
Koestneret al. JPSP A/O Adults Hidden V U
(1987) puzzles
Koestneret al. JPSP A/O Adults Hidden V U
(1987) puzzles
Butler (1987) J Ed Psych A/O Children Problem V U
solving
Butler (1987) J Ed Psych A/O Children Problem V U
solving
Tripathi, J Gen Psych A/O Adults Problem T E Not, TC
Agarwal solving
(1988)
Tripathi, J Gen Psych A/O Adults Problem T E Cont, PC
Agarwal solving
(1988)
Tripathi, J Gen Psych A/O Adults Problem T E Both
Agarwal solving
(1988)
Fabes et al. Mot & A/O Children Beanbag T E Not, TC
(1988) Emotion game
Sansone (1989) J Exp Soc A/O Adults Identify V U
Psych names
Sansone et al. JPSP A/O Adults Computer V U
(1989) games
Anderson, J App Soc A/O Adults Brain V U
Rodin (1989) Psych teasers
Mawhinneyet J Org Beh SS Adults Video T E Not
al. (1989) Management Repeated game
measures
Wickeret al. J of Psych A/O Adults Think T E Not, TC
(1990) Tac Toe

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Wickeret al. J of Psych A/O Adults Think T E Not, TC
(1990) Tac Toe
Notes.
Design: B/A = before-aftergroups design, A/O = after-onlygroups design, SS = single-subject design
Rewardtype: T = tangible, V = verbal
Rewardexpectancy:E = expected, U = unexpected
Rewardcontingency:cont = contingent, not = not contingent;NC = nontask contingent, TC = task cont
aindicateseffect sizes given a value of zero (nonsignificantresults with no reportof means or direction of
bindicatesestimatedeffect sizes
JPSP = Journalof Personalityand Social Psychology
J of Pers = Journalof Personality
Org Beh & Hum Perf = OrganizationalBehavior and Human Performance
J Exp Soc Psych = Journalof ExperimentalSocial Psychology
Child Dev = Child Development
Per & Soc Psy Bull = Personalityand Social Psychology Bulletin
Cog Ther & Res = Cognitive Therapyand Research
Mot & Emotion = Motivationand Emotion
Int J of Sport Psy = InternationalJournalof Sport Psychology
J of School Psych = Journalof School Psychology
J Applied Soc Psych = Journalof Applied Social Psychology
Behavior Mod = Behavior Modification
Soc Beh & Pers = Social Behavior and Personality
J of Soc Psych = Journalof Social Psychology
J Applied Psych = Journalof Applied Psychology
J of Res Pers = Journalof Research in Personality
J GeneralPsych = Journalof General Psychology
J Management= Journalof Management
Aust & N.Z. J Dev Dis = Australiaand New ZealandJournalof DevelopmentalDisabilities
J Sport Psych = Journalof Sport Psychology
Bull Psych Society = Bulletin of the Psychonomic Society
Br J Dev Psych = British Journalof Developmental Psychology
J Exp Psych = Journalof ExperimentalPsychology
Acad Man J = Academy of ManagementJournal
Org Beh & Hum Dec P = OrganizationalBehavior and Human Decision Processes
Psych Studies = Psychological Studies
o J Org Beh Management= Journalof OrganizationalBehavior Management

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation
References

Amabile, T. M., Hennessey, B. A., & Grossman, B. S. (1986). Social influences on

creativity: The effects of contracted-for reward. Journal of Personality and Social
Psychology, 50, 14-23.
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive
theory. Englewood Cliffs, NJ: Prentice-Hall.
Bates, J. A. (1979). Extrinsic reward and intrinsic motivation: A review with implica-
tions for the classroom. Review of Educational Research, 49, 557-576.
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in Experi-
mental Social Psychology (Vol. 6, pp. 1-62). New York: Academic.
Boggiano, A. K., & Ruble, D. N. (1979). Competence and the overjustification
effect: A developmental study. Journal of Personality and Social Psychology,
37, 1462-1468.
Brennan, T. P., & Glover, J. A. (1980). An examination of the effect of extrinsic
reinforcerson intrinsically motivated behavior:experimental and theoretical. Social
Behavior and Personality, 8, 27-32.
Butler, R. (1987). Task-involving and ego-involving properties of evaluation: Effects
of different feedback conditions on motivational perceptions, interest, and perfor-
mance. Journal of Educational Psychology, 79, 474-482.
Cameron, J. (1992). Intrinsic motivation revisited. Unpublished doctoral dissertation,
University of Alberta, Canada.
Cooper, H. M. (1989). Integrating research: A guide for literature reviews (2nd ed.).
Beverly Hills: Sage.
Danner, F. W., & Lonkey, E. (1981). A cognitive developmental approach to the
effects of rewards on intrinsic motivation. Child Development, 52, 1043-1052.
Deci, E. L. (1971). Effects of externally mediated rewards on intrinsic motivation.
Journal of Personality and Social Psychology, 18, 105-115.
Deci, E. L. (1972a). Intrinsic motivation, extrinsic reinforcement, and inequity. Jour-
nal of Personality and Social Psychology, 22, 113-120.
Deci, E. L. (1972b). The effects of contingent and noncontingent rewards and controls
on intrinsic motivation. Organizational Behavior and Human Performance, 8,
217-229.
Deci, E. L. (1975). Intrinsic Motivation. New York: Plenum.
Deci, E. L., & Ryan, R. M. (1985). Intrinsic Motivation and Self-Determination in
Human Behavior. New York: Plenum.
Deci, E. L., & Ryan, R. M. (1987). The support of autonomy and the control of
behavior. Journal of Personality and Social Psychology, 53, 1024-1037.
DeLoach, L. L., Griffith, K., & LaBarba, R. C. (1983). The relationship of group
context and intelligence to the overjustification effect. Bulletin of the Psychonomic
Society, 21, 291-293.
Dickinson, A. M. (1989). The detrimentaleffects of extrinsic reinforcement on "intrin-
sic motivation." The Behavior Analyst, 12, 1-15.
Enzle, M. E., & Ross, J. M. (1978). Increasing and decreasing intrinsic interest with
contingent rewards. Journal of Experimental and Social Psychology, 14, 588-597.
Fabes, R. A. (1987). Effects of reward contexts on young children's task interest.
Journal of Psychology, 121, 5-19.
Farr, J. L. (1976). Task characteristics, reward contingency and intrinsic motivation.
Organizational Behavior and Human Performance, 16, 294-307.
Fazio, R. H. (1981). On the self-perception explanation of the overjustification effect:
the role of salience and initial attitude.Journal of Experimental Social Psychology,
17, 417-426.

420

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

Feingold,B. D., & Mahoney,M. J. (1975).Reinforcement effectson intrinsicinterest:

Underminingthe overjustificationhypothesis.BehaviorTherapy,6, 357-377.
intrinsicinterestfromthestandpointof a behaviorist.
Flora,S. R. (1990).Undermining
The Psychological Record, 40, 323-346.
Glass, G. V., McGaw, B., & Smith, M. L. (1981). Meta-Analysis in Social Research.
BeverlyHills: Sage.
Greene, D., & Lepper,M. R. (1974). Effects of extrinsic rewardson children's
subsequentintrinsicinterest.ChildDevelopment,45, 1141-1145.
Greene,D., Sternberg,B., & Lepper,M. R. (1976). Overjustificationin a token
economy. Journal of Personality and Social Psychology, 34, 1219-1234.
Guzzo, R. A. (1979). Typesof rewards,cognitionsand work motivation.Academy
of Management Journal, 22, 75-86.
Harackiewicz,J. K., Manderlink,G., & Sansone, C. (1984). Rewardingpinball
wizardry:effects of evaluationand cue value on intrinsicinterest.Journal of
Personality and Social Psychology, 47, 287-300.
Hedges, L. V. (1981). Distributiontheoryfor Glass's estimatorof effect size and
related estimators. Journal of Educational Statistics, 6, 107-128
Hedges, L. V. (1987). How hard is hard science, how soft is soft science? The
empiricalcumulativenessof research.AmericanPsychologist,42, 443-455.
Hedges, L., & Becker, B. J. (1986). Statisticalmethods in the meta-analysisof
researchon genderdifferences.In J. S. Hyde& M. C. Linn(Eds.),Thepsychology
of gender: Advances through meta-analysis (pp. 14-50). Baltimore: John Hopkins
UniversityPress.
Hedges, L., & Olkin, I. (1985). Statistical Methodsfor Meta-Analysis. Orlando:Aca-
demic.
Hopkins, B. L., & Mawhinney, T. C. (1992). Pay for performance: History, contro-
versy, and evidence. New York: Haworth.
Karniol,R., & Ross, M. (1977). The effect of performancerelevantandperformance
irrelevantrewardson children'sintrinsic motivation.Child Development,48,
482-487.
Kelly, H. H. (1967). Attributiontheory in social psychology.In D. Levine (Ed.),
Nebraska symposium on motivation (Vol. 15, pp. 192-238). Lincoln: University
of NebraskaPress.
Kohn, A. (1993). Why incentive plans cannot work. HarvardBusiness Review,
71(5), 54-63.
Lepper,M. R. (1981). Intrinsicand extrinsicmotivationin children:Detrimental
effects of superfluoussocial controls. In W. A. Collins (Ed.), Aspects of the
development of competence: The Minnesota symposia on child psychology (Vol.
14, pp. 155-214). Hillsdale,NJ: Erlbaum.
Lepper,M. R., & Greene, D. (1975). Turningplay into work: Effects of adult
surveillanceand extrinsicrewardson children'sintrinsicmotivation.Journalof
Personality and Social Psychology, 31, 479-486.
Lepper,M. R., Greene,D., & Nisbett,R. E. (1973). Underminingchildren'sintrinsic
hypothesis.Journal
interestwith extrinsicreward:A test of the "overjustification"
of Personality and Social Psychology, 28, 129-137.
Levine,F. M., & Fasnacht,G. (1974). Tokenrewardsmay lead to token learning.
American Psychologist, 29, 817-820
Light, R. J., & Pillemer, D. B. (1984). Summing Up: The Science of Reviewing
Research.Cambridge,MA: HarvardUniversityPress.
with extrinsicrewards:
Mawhinney,T. C. (1990). Decreasingintrinsic"motivation"
Easier said than done. Journal of Organizational Behavior Management, 11,
175-191.
421

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
Reinforcement,Reward,and IntrinsicMotivation

McCullers, J. C. (1978). Issues in learning and motivation. In M. R. Lepper &

D. Greene (Eds.), The Hidden Costs ofReward: New Perspectives on the Psychology
of Human Motivation (pp. 5-18). Hillsdale, NJ: Erlbaum.
McGraw, K. O., & Wong, S. P. (1992). A common language effect size statistic.
Psychological Bulletin, 111, 361-365.
Morgan, M. (1981). The overjustification effect: A developmental test of self-percep-
tion interpretations.Journal of Personality and Social Psychology, 40, 809-821.
Morgan, M. (1983). Decrements in intrinsic interest among rewarded and observer
subjects. Child Development, 54, 636-644.
Morgan, M. (1984). Reward-induced decrements and increments in intrinsic motiva-
tion. Review of Educational Research, 54, 5-30.
Mynatt, C., Oakley, D., Arkkelin, D., Piccione, A., Margolis, R., & Arkkelin, J. (1978).
An examination of overjustification under conditions of extended observation
and multiple reinforcement: Overjustification or boredom? Cognitive Therapy and
Research, 2, 171-177.
Orlick, T. D., & Mosher, R. (1978). Extrinsic awards and participant motivation in
a sport related task. International Journal of Sport Psychology, 9, 27-39.
Phillips, J. S., & Lord, R. G. (1980). Determinants of intrinsic motivation: locus of
control and competence information as components of Deci's cognitive evaluation
theory. Journal of Applied Psychology, 65, 211-218.
Pinder, C. C. (1976). Additivity versus nonadditivity of intrinsic and extrinsic incen-
tives: Implications for work, motivation, performance, and attitudes. Journal of
Applied Psychology, 61, 693-700.
Pittman, T. S., Emery, J., & Boggiano, A. K. (1982). Intrinsic and extrinsic motiva-
tional orientations: reward-induced changes in preference for complexity. Journal
of Personality and Social Psychology, 42, 789-797.
Pritchard, R. D., Campbell, K. M., & Campbell, D. J. (1977). Effects of extrinsic
financial rewards on intrinsic motivation. Journal of Applied Psychology, 62, 9-15.
Rigby, C. S., Deci, E. L., Patrick, B. C., & Ryan, R. M. (1992). Beyond the intrinsic-
extrinsic dichotomy: Self-determination in motivation and learning. Motivation
and Emotion, 16, 165-185.
Rosenfield, D., Folger, R., & Adelman, H. F. (1980). When rewards reflect compe-
tence: A qualification of the overjustification effect. Journal of Personality and
Social Psychology, 39, 368-376.
Rosenthal, R. (1966). Experimenter effects in behavioral research. New York:Apple-
ton-Century-Crofts.
Rummel, A., & Feinberg, R. (1988). Cognitive evaluation theory: A meta-analytic
review of the literature. Social Behavior and Personality, 16, 147-164.
Ryan, R. M., Mims, B., & Koestner, R. (1983). Relation of reward contingency and
interpersonal context to intrinsic motivation: A review and test using cognitive
evaluation theory. Journal of Personality and Social Psychology, 45, 736-750.
Sansone, C., & Morgan, C. (1992). Intrinsic motivation and education: Competence
in context. Motivation and Emotion, 16, 249-270.
Schwartz, B. (1990). The creation and destruction of value. American Psychologist,
45, 7-15.
Schwarzer, R. (1991). Meta: Programs for secondary data analysis, MS-DOS Version
5.0 [Computer program]. Dubuque, IA: Wm. C. Brown.
Scott, W. E., Jr. (1975). The effects of extrinsic rewards on "intrinsic motivation."
Organizational Behavior and Human Performance, 25, 311-335.
Sutherland, S. (1993). Impoverished minds. Nature, 364, 767.
Tripathi, K. N., & Agarwal, A. (1985). Effects of verbal and tangible rewards on
intrinsic motivation in males and females. Psychological Studies, 30, 77-84.

422

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions
 Cameronand Pierce

Tukey,J. W. (1977). Exploratorydata analysis.Reading,MA: Addison-Wesley.

Vallerand,R. J. (1983). The effectof differentialamountsof postiveverbalfeedback
on the intrinsicmotivationof male hockeyplayers.Journalof SportPsychology,
5, 100-107.
Vasta,R., Andrews,D. E., McLaughlin,A. M., Stirpe,L. A., & Comfort,C. (1978).
Reinforcementeffects on intrinsicinterest:A classroomanalog.Journalof School
Psychology, 16, 161-168.
Vasta,R., & Stirpe,L. A. (1979).Reinforcement
effectson threemeasuresof children's
interest in math. Behavior Modification, 3, 223-244.
Wiersma,U. J. (1992).Theeffectsof extrinsicrewardsin intrinsicmotivation:A meta-
analysis. Journal of Occupational and Organizational Psychology, 65, 101-114.
Workman,E. A., & Williams,R. L. (1980). Effectsof extrinsicrewardson intrinsic
motivation in the classroom. Journal of School Psychology, 18, 141-147.
Zimbardo,P.G. (1988)Psychologyandlife (I11thed.). Glenview,IL:Scott,Foresman.

Authors
JUDY CAMERONis AssistantProfessor,Departmentof EducationalPsychology,
Universityof Alberta,Edmonton,CanadaT6G 2G5. She specializesin learning
and motivation,social psychology,education,and second languageacquisition.
W. DAVIDPIERCEis Professor,Centrefor ExperimentalSociology, 1-48 TORY,
Universityof Alberta,Edmonton,CanadaTG6 2H4. He specializes in social
psychologyand behavioranalysis.

ReceivedJuly 22, 1993

RevisionreceivedJanuary28, 1994
AcceptedFebruary7, 1994

423

This content downloaded from 142.51.1.212 on Sun, 21 Feb 2016 18:22:30 UTC
All use subject to JSTOR Terms and Conditions