The Differential Effects of Intrinsic and Identified Motivation on Well-Being and Performance: Prospective, Experimental, and Implicit Approaches to Self-Determination Theory

Journal of Personality and Social Psychology
Copyright 2006 by the American Psychological Association
2006, Vol. 91, No. 4, 750 –762
0022-3514/06/$12.00
DOI: 10.1037/0022-3514.91.4.750
The Differential Effects of Intrinsic and Identified Motivation on
Well-Being and Performance: Prospective, Experimental,
and Implicit Approaches to Self-Determination Theory
Kimberly D. Burton, John E. Lydon, David U. D’Alessandro, and Richard Koestner
McGill University
Self-determination theory research has demonstrated that intrinsic and identified self-regulations are
associated with successful adaptation. However, few distinctions are typically made between these
regulations and their outcomes. In the present studies, the associations between intrinsic and identified
motivations and outcomes of psychological well-being and academic performance are compared in
educational settings. In Study 1, intrinsic self-regulation predicted psychological well-being, independent
of academic performance. In contrast, identified regulation predicted academic performance. Addition-
ally, the more that students demonstrated an identified academic regulation, the more that their
psychological well-being was contingent on performance. In Study 2a, priming intrinsic self-regulation
led to greater psychological well-being 10 days later. In Study 2b, an implicit measure of identified
regulation predicted academic performance 6 weeks later. Results indicate the need to address important
distinctions between intrinsic and identified regulations.
Keywords: motivation, self-determination, well-being, performance
Most things that matter in life are not easy to achieve. How do
Although intrinsic and identified regulations are correlated, these
we manage to attain our goals while being happy in the process?
constructs are theoretically distinct. Despite this, little research has
Does a certain type of motivation lead us to work toward doing
examined intrinsic and identified regulatory styles separately, and
well, and another type lead us to feel good as we work? Research-
there is not much empirical differentiation between the outcomes
ers have identified styles of self-regulation that may help to answer
associated with each type of motivation.
these questions.
We propose that, in addition to the benefits associated with
Self-determination theory built on the classic distinction be-
autonomous motivation generally, theory and research may be
tween extrinsic and intrinsic motivation by developing a contin-
advanced by examining the relative contributions of intrinsic and
uum model of motivation (Deci & Ryan, 2000), with points along
identified regulatory styles to the prediction of positive outcomes.
the continuum representing distinct self-regulatory styles for be-
Briefly, in the academic domain, among both elementary school
havior. Past research demonstrated that the autonomous end of the
and university students, we will show that intrinsic self-regulation
continuum, comprising intrinsic and identified self-regulations,1 is
is an important predictor of psychological well-being outcomes,
associated with positive outcomes, such as psychological well-
controlling for identified self-regulation, and that identified regu-
being (e.g., Reis, Sheldon, Gable, Roscoe, & Ryan, 2000) and
lation is an important predictor of performance outcomes, control-
academic performance (e.g., Grolnick & Ryan, 1987). Intrinsic
regulation is the most autonomous of the regulatory styles and
ling for intrinsic self-regulation.
exists when people freely choose to perform an activity out of a
sense of interest. In contrast, the identified regulatory style in-
Self-Regulatory Style and Outcomes
volves an individual’s recognition and acceptance of the value and
importance of a behavior and the integration of this into the self.
Autonomous reasons for engaging in a particular behavior are
associated with beneficial psychological outcomes, such as feeling
good about an activity, goal progress, and psychological well-
Kimberly D. Burton, John E. Lydon, David U. D’Alessandro, and
being, as well as with positive behavioral outcomes, such as school
Richard Koestner, Department of Psychology, McGill University,
performance. Autonomous regulation is positively correlated with
Montreal, Que´bec, Canada.
children’s enjoyment of elementary school (Ryan & Connell,
This work was supported by research grants from the Social Sciences
1989), with university students’ enjoyment of an organic chemistry
and Humanities Research Council of Canada and the Fonds pour la
class (Black & Deci, 2000), as well as with positive mood (Reis et
Formation de Chercheurs et l’Aide a` la Recherche du Quebec. We thank
Miri Rozenek for her help with the programming of the lexical decision
al., 2000), vitality (e.g., Nix, Ryan, Manly, & Deci, 1999; Reis et
task and Stephanie Blum for her assistance with data entry.
Correspondence concerning this article should be addressed to John
1
E. Lydon, Department of Psychology, McGill University, 1205 Docteur
Subsequent theory and research described an integrated self-regulation
Penfield Avenue, Montreal, Que´bec H3A 1B1, Canada. E-mail:
that is highly correlated with identified regulation but that is thought to be
jlydon@hebb.psych.mcgill.ca
relatively more autonomous.
750

---

EFFECTS OF INTRINSIC AND IDENTIFIED MOTIVATION
751
al., 2000; Ryan & Frederick, 1997), and positive coping strategies
mediate experience (i.e., intrinsic motivation) and on long-term
(Ryan & Connell, 1989). In addition, autonomous regulation is
goals (i.e., identification).
associated with greater conceptual learning of material (Grolnick
Educational settings provide a fertile environment in which to
& Ryan, 1987), with academic performance (Grolnick, Ryan, &
study distinct regulatory styles because students often report hav-
Deci, 1991), and with grades (Miserandino, 1996) in elementary
ing high levels of both intrinsic and identified motivation for their
school children. Likewise, junior college students who persisted in
academic involvement (Vallerand, Fortier, & Guay, 1997) and
a course were significantly more autonomous at the beginning of
because academic involvement is central to the identities of many
the semester than were students who later dropped out (Vallerand
young people (Blais, Vallerand, Brie`re, Gagnon, & Pelletier,
& Bissonnette, 1992). Overall, research has indicated that having
1990). In the present studies, we examine the correlates of intrinsic
an autonomous self-regulatory style is associated with psycholog-
and identified regulation among elementary school and university
ical well-being and positive behavioral outcomes.
undergraduate students. Because the intrinsic self-regulatory style
The majority of the above research treats self-regulation in
reflects the positive experience that individuals have regarding an
terms of the self-determination continuum. For example, a number
activity, such as feelings of enjoyment and interest, we hypothe-
of researchers report using autonomy summary scores that are
sized that intrinsic regulation would predict positive affect and
often computed from the Self-Regulation Scale (Ryan & Connell,
satisfaction with life, two of the primary indices of psychological
1989) by subtracting the sum of the nonautonomous regulations
well-being (Diener & Seligman, 2004). Moreover, we expected
(i.e., introjected and extrinsic) from the sum of intrinsic and
that the link between intrinsic motivation and psychological well-
identified regulations, thereby creating a Relative Autonomy Index
being would not be contingent on performance. That is, the posi-
(e.g., Black & Deci, 2000; Grolnick et al., 1991; Miserandino,
tive psychological benefits of pursuing intrinsically regulated
1996). Others compute the same index by attaching weights to the
goals should be based on the activity itself and not dependent on
self-regulation subscales (i.e., extrinsic
2, introjection
1,
associated outcomes.
identified
1, intrinsic
2) and use this composite to predict
However, because success at school involves performing delib-
outcome variables (e.g., Grolnick & Ryan, 1987; Reis et al., 2000;
erate, effortful, and challenging exercises, such as studying and
Vallerand & Bissonnette, 1992). As a result of these indexing
doing homework, we hypothesized that an identified regulation
methods, distinctions between regulations that are close together
would be predictive of performance, as assessed by students’
on the self-determination continuum, such as the distinctions be-
grades. Internalizing a goal into the self, as is done when an
tween intrinsic and identified regulations, may be overlooked
individual has an identified self-regulation, is necessary for estab-
(Koestner & Losier, 2002).
lishing the importance of the goal and also for developing and
Implicit in the combining of intrinsic and identified self-
maintaining commitment to, and persistence at working toward,
regulation scores is that the promotion of high levels of both
the goal (Austin & Vancouver, 1996). Research has indicated that
regulations is an appropriate objective toward which socializing
the extent to which individuals identify with their goals is predic-
agents should strive. Indeed, self-determination theory suggests
tive of their goal commitment and progress, even in the face of
that intrinsic motivation and internalization work in a complemen-
adversity (e.g., Lydon, Burton, & Menzies-Toman, 2005; Lydon &
tary fashion to encourage vitality, growth, and adaptation (Deci &
Zanna, 1990). Therefore, we believed that students’ academic
Ryan, 2000). Intrinsic self-regulation promotes a focus on the task
performance would be best predicted by their levels of identified
itself and yields energizing emotions such as interest and excite-
self-regulation.
ment, whereas identification keeps one oriented toward the long-
term significance of one’s current pursuits and may foster persis-
Including Experimental and Implicit Cognitive
tence at uninteresting, but important, activities. Possessing high
Methodologies
levels of both intrinsic motivation and identification would seem to
allow one the flexibility to adapt to a wide array of situations.
Whereas our primary objective was to examine the distinct
Researchers who do not use self-determination theory as their
contributions of intrinsic and identified motives in accounting for
starting point also established the adaptive value of possessing
psychological well-being and performance outcomes, we also
both intrinsic and more instrumental goals. For example, in a
sought to expand on the methodologies typically used in self-
longitudinal study of talented teenagers, Wong and Csikszentmi-
determination theory research by incorporating social– cognitive
halyi (1991) distinguished between two forms of academic moti-
theory and its methods. To date, few studies have directly used
vation: intrinsic motivation and work orientation. They argued that
cognitive experimental methods in the examination of motiva-
intrinsic motivation is based on the rewards of ongoing experience,
tional processes. Levesque and Pelletier (2003) demonstrated that
whereas work orientation reflects an investment in long-term goals
priming intrinsic motivation, or temporarily increasing its cogni-
such as fulfilling one’s career expectations and meeting one’s
tive accessibility, led participants to report greater enjoyment of a
psychological needs. Their results indicated that work orientation,
laboratory activity and to show better performance than those
which we would liken to identified regulation, was significantly
primed with extrinsic motivation. This interesting work appears to
associated with the amount of time that students spent studying but
be one of the first published reports of the incorporation of cog-
was unrelated to their experience while studying. By contrast,
nitive principles into the study of self-determination theory. How-
intrinsic motivation was related to enjoyable studying experiences
ever, similar to others, Levesque and Pelletier (2003) compared
but not to the amount of time spent studying. To achieve good
only intrinsic and extrinsic motivation and so did not examine
grades and simultaneously enjoy the process would seem to re-
performance
differences
between
intrinsic
and
identified
quire combining self-regulation strategies that focus both on im-
regulations.

---

752
BURTON, LYDON, D’ALESSANDRO, AND KOESTNER
In addition to the benefits of experimental control that priming
form poorly, they may feel a strong sense of disappointment and
affords, we viewed social– cognitive theory as an approach that
therefore experience a decrease in well-being. For all analyses in
would expand our understanding of how autonomous motives may
Study 1, we deemed it important to include both the intrinsic and
guide behavior. Although a relatively large amount of research
identified self-regulation styles to examine the possible distinct
exists that examines explicit self-reported autonomous motives
contributions that one style may make to a particular outcome,
and their correlates, to our knowledge there is little research that
relative to the other style.
uses implicit methods of assessing individual differences in auton-
In Study 2a, we created an experimental test of the first two
omous motivation in the self-determination framework (cf. Mc-
hypotheses from Study 1, with a sample of university students who
Clelland, Koestner, & Weinberger, 1989). Social– cognitive theory
were initially assessed 10 days prior to, and then only mere hours
suggests that, over time, an explicit, conscious motive can come to
after, writing a difficult midterm examination. We hypothesized
operate in an implicit, efficient, and effortless fashion (Bargh &
that experimentally priming intrinsic motivation would increase
Chartrand, 1999; 2000). In fact, recent research indicated that
students’ levels of psychological well-being, measured immedi-
some types of motivation can be activated merely by associated
ately after the exam. We also predicted that this effect would not
cues (Ratelle, Baldwin, & Vallerand, 2005). In the context of our
be contingent on how well students believed that they had per-
educational paradigm, this might mean that a student who con-
formed on the exam. Whereas most students might be expected to
sciously identifies with an academic goal may learn to associate
experience a change in well-being dependent on their perceived
environmental cues, such as a textbook, with his or her motivation.
exam performance, we hypothesized that those in the intrinsic
Consequently, the student’s identified motivation may become
priming condition would experience an increase in their psycho-
activated simply by the sight of a textbook, and this may occur
logical well-being regardless of their perceived performance. In
even when he or she is tired, distracted, or facing other demands on
addition, we sought to examine the relation between active in-
conscious attention, for little effort is required for such activation.
volvement in goal pursuit, as measured by the amount of time
Thus, a measure of motivation that operates at an implicit level
spent studying, and changes in well-being. We hypothesized that
would add explanatory power to research based in self-
among those in the intrinsic regulation condition, the amount of
determination theory.
time that students spent studying for the exam would represent the
frequency of environmental cues associated with an intrinsically
Present Studies
motivated goal and thereby would predict increases in psycholog-
ical well-being.
Our goal in this set of studies was to examine intrinsic and
Finally, in Study 2b, we tested our third hypothesis from Study
identified self-regulations in the academic domain, with respect to
1, with the addition of an implicit measure of motivation. We
performance and psychological well-being outcomes.2 Study 1
predicted that identified self-regulation, assessed both implicitly
explored three primary hypotheses with a sample of elementary
and explicitly, would predict performance on the course’s final
school children. First, we predicted that pursuing an intrinsically
examination, when controlling for intrinsic regulation and previous
motivating goal would be associated with higher psychological
course grades. Lastly, an exploratory goal was to examine the
well-being, as measured by positive affect. Second, we hypothe-
possibility of interactive effects between explicit and implicit
sized that the association between intrinsic regulation and psycho-
self-regulations. We believed that if it is identified regulation that
logical well-being would not be contingent on students’ perfor-
is most critical for performance, then having high levels of explicit
mance, as measured by their report card grades. Specifically, we
or implicit identification should contribute to better final exami-
believed that intrinsic motivation, assessed 7 days prior to the
nation grades. In contrast, we expected that those with low levels
receipt of report card grades, would be a significant predictor of
on both measures would perform more poorly than others.
positive affect 1 day after the receipt of grades, when controlling
for baseline affect and report card grades. Third, we hypothesized
that identified regulation would be predictive of academic perfor-
Study 1
mance. We statistically controlled for students’ grade expectations
because students who did well in the past and expected to do well
With a sample of elementary school children, our goal in this
in the future may have developed identified motives. We sought to
study was to demonstrate the importance of distinguishing be-
demonstrate that identified motives assessed 7 days prior to the
tween intrinsic and identified self-regulatory styles. Briefly, we
receipt of report cards would be predictive of report card grades,
hypothesized that intrinsic self-regulation would best predict psy-
over and above students’ grade expectations.
chological well-being outcomes, such as positive affect, when
In addition to our three primary hypotheses, we examined one
controlling for baseline levels, and that this association would not
exploratory hypothesis. Given that progress toward autonomous
be contingent on students’ performance as assessed by their report
goals is associated with greater well-being (Sheldon & Elliot,
card grades. In addition, we believed that such performance out-
1999; Sheldon & Kasser, 1998) and our prediction that, specifi-
comes would be best predicted by identified self-regulation. Fi-
cally, identified regulations would be positively associated with
nally, we sought to examine the interaction between identified
grades, we speculated that identified motives would interact with
grades to predict positive affect. Our rationale was that when
2 Ryan and Deci (2003) noted that “identifications can be thought of as
students recognize the importance of the goal and incorporate it
more versus less compartmentalized, and only those that are well integrated
into the self, that is, when they have an identified self-regulation,
within the psyche represent the full endorsement of the self” (p. 258). The
performing well will be rewarding and result in an increase in
measures that we use to assess identification do not currently capture this
psychological well-being. Conversely, when such students per-
important distinction.

---

EFFECTS OF INTRINSIC AND IDENTIFIED MOTIVATION
753
regulation scores and academic performance on students’ psycho-
positive affect, rs(239)
.15–.17, ps
.02, and with Time 2
logical well-being.
positive affect, rs(239)
.23–.26, ps
.001. However, when
examining grades, only identified regulation was significantly
Method
correlated with actual report card grades, r(239)
.26, p
.001,
and expected grades, r(239)
.24, p
.001. Finally, expected
Participants
grades correlated with actual grades, r(239)
.42, p
.01.
Participants were 241 elementary school children ranging in age from 8
years to 13 years (127 girls, M for age
11 years, 4 months), attending
Mean Differences
schools in Toronto, Ontario, Canada. Parents and children consented to the
We made a comparison to examine whether mean positive affect
children’s participation as well as to the school’s disclosure of report card
grades to the researchers. Participants completed scales 7 days before their
scores changed from Time 1 to Time 2; however, no change was
report cards were distributed (Time 1) and 1 day after the distribution
observed, t(240)
1, ns. We then made a similar comparison
(Time 2).
between children’s expected grades (M
9.11) and their actual
report card grades (M
8.05). Although these two grade indices
Procedure
were significantly correlated, r(239)
.42, p
.001, children
significantly overestimated what their grades would be, t(240)
Researchers visited the classrooms of participants at two different time
6.87, SD
2.40, p
.001.
points, during which children completed paper and pencil questionnaires.
To aid comprehension, the researchers read each item of the questionnaires
aloud to the class, and children then recorded their own responses on
Predicting Positive Affect
individual questionnaire packets. Although the measures were completed
To test our first two hypotheses that intrinsic self-regulation
in group settings, participants responded individually, and privately, at
would be independently predictive of psychological well-being
their own desks.
At Time 1, children completed 24 items from Ryan and Connell’s (1989)
and that this association would not be contingent on performance,
Self-Regulation Scale, representing three of the four domains from the
we conducted a hierarchical multiple regression analysis in which
original scale (reasons for class work, homework, and trying to do well in
we sought to use intrinsic self-regulation scores to predict positive
school). On a 4-point scale, participants endorsed various statements re-
affect at Time 2. In the first step of the regression, we entered
flecting the different styles of self-regulation and the extent that they
positive affect at Time 1 and students’ actual report card grades as
explained the reasons for their own behavior with regard to school. Five
control variables. Then, in Step 2, we entered scores on both the
items assessed intrinsic regulation (
.90), and five items assessed
intrinsic and identified subscales of the Self-Regulation Scale
identified regulation (
.75). An example of an identified self-regulation
(Ryan & Connell, 1989), and finally in Step 3, we entered the
reason is “because I want to learn new things,” whereas an example of an
two-way interactions between report card grades and each self-
intrinsic regulation reason is “because it’s fun.”
regulation. As one might expect, Time 1 positive affect was a
Children also completed the Positive and Negative Affect Scale for
Children (PANAS-C; Laurent et al., 1999), which is a child-oriented
significant predictor of Time 2 positive affect,
.54, t(234)
adaptation of the PANAS (Watson, Clark, & Tellegen, 1988). On a 5-point
10.00, p
.001, as were actual report card grades,
.12,
scale, they rated words describing their feelings and emotions (e.g., happy,
t(234)
2.33, p
.03. Central to our hypothesis, intrinsic self-
calm, sad, gloomy) in the past few weeks. The 12 items from the positive
regulation scores significantly predicted greater positive affect at
affect subscale were used to measure subjective well-being (
.88).3
Time 2,
.14, t(234)
2.22, p
.03, but identified regulation
Finally, participants indicated their grade expectations for their upcoming
scores did not,
.04, t(234)
1, ns. This result indicated that
report cards. Because this was the first reporting period of the school year,
students with a highly intrinsic orientation toward school reported
there was no baseline measure of students’ academic performance avail-
greater well-being after the receipt of their report card grades.
able for that year. As an alternative, we used students’ grade expectations
Then, in the final step, we examined interactions between the
as the baseline measure.
predictors to address our second and exploratory hypotheses that
One week later, after report cards had been distributed, the researchers
returned to the classrooms of participating children. At this Time 2 session,
the relationship between intrinsic self-regulation and positive af-
participants again completed the PANAS-C (Laurent et al., 1999) as the
fect would not be contingent on performance, as assessed by report
measure of their psychological well-being, reflecting their current feelings
cards, but that the relationship between identified self-regulation
and emotions. As at Time 1, researchers read each item aloud, and
and positive affect would be contingent on performance. To ad-
participants indicated their responses in their own questionnaires. Follow-
dress the first, we examined the interaction between intrinsic
ing this session, the researchers received copies of children’s report cards
self-regulation and report card grades. This did not prove signifi-
from school administration and then converted students’ expected grades
cant,
.10, t(234)
1.45, p
.15, and therefore indicated
and report card grades to a 13-point scale ranging from 13 (A ) to 1 (F).
that the relationship between higher intrinsic self-regulation and
greater positive affect is not contingent on performance.
Results
Correlations Among Predictor Variables
3 Consistent with the notion that positive and negative affect are distinct
and not two ends of the same continuum, positive and negative affect were
We first performed simple correlations between intrinsic regu-
correlated but not highly so, r(239)
.30 at Time 1 and r(239)
.40
lation, identified regulation, positive affect at Time 1, positive
at Time 2. Neither intrinsic nor identified self-regulation, nor any interac-
affect at Time 2, expected grades, and actual report card grades.
tions, were significant in predicting changes in negative affect, although as
Intrinsic and identified self-regulation scores were significantly
one might expect, lower report card grades were associated with increases
correlated with each other, r(239)
.56, p
.001, with Time 1
in negative affect,
.24, t(240)
4.56, p
.01.

---

754
BURTON, LYDON, D’ALESSANDRO, AND KOESTNER
To address our exploratory hypothesis, we examined the inter-
Summary of Study 1 Findings
action between identified self-regulation scores and report card
grades, which proved significant,
.15, t(234)
2.04, p
.04.
Results suggested that intrinsic self-regulation predicts increases
The interaction is depicted in Figure 1. Among children who had
in indices of students’ psychological well-being, such as positive
lower levels of identification, there was little difference in Time 2
affect, and that changes in well-being are not contingent on per-
positive affect between those with higher report card grades and
formance, as measured by the grades that students receive. The
those with lower grades. However, when only children who had
more that students had an intrinsic academic self-regulation, the
higher levels of identification were examined, a marked difference
greater were the increases in their well-being, regardless of their
was observed between those with low and those with high report
performance. As hypothesized, identified self-regulation was a
card grades. Students who were more identified with school but
significant positive predictor of students’ academic performance
who received low grades on their report cards showed lower levels
such that the more identified students were with their education,
of Time 2 positive affect than did the others. By contrast, those
the higher their grades. Identification also interacted with students’
children who were more identified with school and who received
grades to predict changes in positive affect. Although intrinsic
higher grades on their report cards showed greater Time 2 positive
regulation was associated with psychological well-being in a non-
affect than did the other groups of children.
contingent fashion, the relationship between identified self-
Overall, intrinsic regulation positively predicted students’ well-
regulation and students’ well-being was contingent on their per-
being after receiving their report cards, independent of identified
formance: The more that students had an identified regulation, the
regulation and of grades received. However, among those who
more their well-being was positively associated with their report
were more identified with school, the extent to which they expe-
card grades.
rienced greater or lesser positive affect was contingent on the
grades that they received.
Study 2a
To extend the findings of Study 1 on the differing roles of
Predicting Report Card Grades
identified and intrinsic self-regulations, we conducted a second
study concerning psychological well-being and performance in the
To address our third primary hypothesis that identified self-
context of a specific academic goal, this time with university
regulation would be predictive of performance, we conducted a
students. On the basis of the finding that intrinsic self-regulation
second hierarchical multiple regression analysis. We entered chil-
was positively associated with psychological well-being in Study
dren’s grade expectations, assessed 1 week before report cards
1, our objective in Study 2a was to experimentally obtain a similar
were distributed (Time 1), along with each self-regulation score,
effect. We sought to induce intrinsic and identified self-regulation
followed by the interactions between grade expectations and self-
orientations and to test whether the inductions of these different
regulation scores. As might be anticipated, expected grades were a
regulatory styles would influence students’ well-being and aca-
strong predictor of actual grades. However, identification also
demic performance when writing a difficult examination. Specif-
proved to be a significant positive predictor,
.24, t(235)
ically, we predicted that a manipulation involving intrinsic regu-
3.32, p
.001. The more that students identified with school, the
lation
would
result
in
greater
psychological
well-being
higher were their report card grades. When controlling for identi-
immediately following the examination, 10 days after the experi-
fied self-regulation, there was a nonsignificant trend for intrinsic
mental session. Furthermore, on the basis of results of Study 1, we
self-regulation to be negatively associated with grades,
–.12,
hypothesized that this effect would not be contingent on how well
t(235)
1.67, p
.10, although the zero-order correlation with
students thought that they had performed on the exam that day. In
grades was not significant, r(239)
.06, ns.
addition, we predicted that a manipulation of an identified regu-
lation focus would result in better performance on the midterm
examination.
We believed that if priming an association between intrinsic
0.5
motivation and the particular academic goal can influence psycho-
logical well-being, then activities related to the course, such as
studying, should reactivate the link to intrinsic motivation and
0.25
thereby affect individuals’ psychological well-being. Therefore, an
Low Identified Self-
regulation
exploratory hypothesis for Study 2a was that for those primed with
0
High Identified Self-
intrinsic motivation, the amount of time spent studying would be
regulation
related to changes in subjective well-being, whereas for those not
primed with intrinsic motivation, the amount of time spent study-
-0.25
ing would be unrelated to changes in well-being.
Time 2 Positive Affect (z score)
-0.5
Method
Low High
Report Card Grades
Participants
Figure 1.
Study 1: Time 2 positive affect (z score) as predicted by
Participants were 60 (59 female; M for age
21 years) undergraduate
students’ report card grades and identified self-regulation, controlling for
students enrolled in an upper level undergraduate psychology course at
Time 1 positive affect.
McGill University in Montreal, Que´bec, Canada. At Time 1, 65 individuals

---

EFFECTS OF INTRINSIC AND IDENTIFIED MOTIVATION
755
participated, with 60 returning for the Time 2 session. There were no
location had not yet been set. Under the guise of providing the Time 2
differences in Time 1 psychological well-being between those who re-
information, participants were asked to record their e-mail addresses for the
turned and those who did not, F(1, 63)
0.20, MSE
1.01, ns. Of the 5
experimenters if they wished to be contacted. All participants complied.
individuals who did not return, 1 was in the control condition, 2 were in the
Approximately 4 days prior to the midterm exam, when students were
intrinsic condition, and 2 were in the identified condition.
presumably in the process of studying, the experimenters sent an e-mail
message to each participant. These messages varied according to condition.
Procedure
In the control condition, the message simply contained the necessary
information for attending the next session of the study. However, in the
Recruitment.
Approximately 2 weeks prior to a midterm examination,
experimental conditions, the message served to reinforce the Time 1
participants were recruited from a psychology class for a study examining
manipulation. In the intrinsic condition, part of the e-mail message read
academic goals. This class was chosen because it was not a required
“Thank you for your continued participation in our study on the ways in
course, and much of the material is typically reported to be interesting to
which students find their academic goal of mastering course material to be
students, something that would suggest the possibility of students having at
interesting and enjoyable,” whereas in the identified condition, this same
least some intrinsic motivation. The experimenters explained that those
sentence finished with “the ways in which students identify with their
who participated in this multipart study would be remunerated $10
academic goal of mastering course material and find this goal to be
(U.S.$8.89) for their time.
important.”
Participants completed a questionnaire 10 days prior to the exam. At the
Participants returned to the lab for the second session shortly after they
beginning of the session, the experimenters explained that, for ease of
finished writing their midterm examination, for which the mean grade was
investigation, they were interested in an academic goal that was possibly
66%. In this second component, they completed a questionnaire containing
common to many of the students taking part in the study: mastering the
measures similar to those assessed in the first session, including their
course material. All participants endorsed having this goal to some degree.4
current satisfaction with life, and indicated the grades that they anticipated
Part 1 measures.
Because participants were aware of a variety of
receiving on the midterm, based on their experience of having just written
social psychological phenomena and techniques, the questionnaire in-
the test as well as the amount of time they spent studying for the test.
cluded numerous measures to maintain the appearance of it being simply a
survey. In addition to a variety of individual differences measures, students
were asked questions about their study habits, about their impressions of
Results
course difficulty, and about their commitment. Of importance, participants
also completed a baseline measure of their current psychological well-
Effect of the Manipulation on Psychological Well-Being
being, the Satisfaction with Life Scale (Diener, Emmons, Larsen, & Grif-
fin, 1985). In addition, participants completed a measure of perceived locus
To address our hypothesis that the induction of an intrinsic
of causality (Sheldon & Kasser, 1995, 1998) to assess intrinsic and iden-
self-regulation focus would positively affect psychological well-
tified motives for the goal of mastering the course material.
being, we conducted an analysis of covariance, controlling for
Self-regulatory manipulation.
The last two components of the ques-
initial well-being scores. Results indicated a significant main effect
tionnaire composed the manipulation of regulatory style. Participants were
of condition on satisfaction with life scores after the midterm test,
randomly assigned to one of three conditions: control, intrinsic regulation,
F(2, 56)
5.60, MSE
0.37, p
.001, whereby the well-being
or identified regulation. For those in the control condition, the question-
of those in the intrinsic condition (M
5.50) was significantly
naire ended following the completion of the last scale, whereas for those in
higher than that of those in the control condition (M
4.90),
either of the two experimental conditions, the questionnaire continued for
t(56)
3.17, p
.01, and than that of those in the identified
one extra page. The first part of the manipulation consisted of a list of
statements with which participants were asked to indicate their agreement
condition (M
5.01), t(56)
2.55, p
.01. In addition, a
by writing the word “Yes” (to denote “Yes, I agree at least somewhat.”) or
repeated measures analysis of variance revealed a significant
the word “No” (to denote “No, I disagree completely.”) next to each
Time
Condition effect, F(2, 57)
4.14, MSE
0.27, p
.02,
statement. These response options were constructed to increase the likeli-
such that there was a significant increase in satisfaction with life
hood that participants would endorse the statements on the list (Salancik,
from Time 1 to Time 2 for those in the intrinsic condition, t(57)
1974).
3.12, p
.01, but not in the control condition, t(57)
0.93, ns, or
With items included in the identified and intrinsic subscales of Ryan and
in the identified condition, t(57)
0.87, ns. The means from this
Connell’s (1989) Self-Regulation Scale, two lists of seven statements were
analysis are presented in Table 1. These results indicated that the
created, one for each condition. For example, in the intrinsic condition,
intrinsic manipulation led to significant improvements in individ-
participants were presented with items such as “I find the course material
uals’
psychological
well-being
after
writing
the
midterm
interesting” and “I enjoy the course material,” whereas those in the iden-
tified condition endorsed items such as “Mastering the course material is
examination.
important to me” and “I value being able to learn from the course material.”
All participants endorsed the vast majority of the statements, and there
Effect of the Manipulation on Midterm Examination
were no differences in endorsement between the priming conditions, F(1,
41)
1.74, MSE
0.04, ns. Following the completion of the statement
Grades
component, participants were asked to write about their goal of mastering
the course material in terms of it being fun, enjoyable, and interesting in the
Contrary to our hypothesis, an analysis of covariance in which
intrinsic condition or in terms of values, identity, and meaning in the
we controlled for previously obtained course grades revealed no
identified condition. These words were again chosen on the basis of Ryan
significant effects of the manipulations on students’ midterm ex-
and Connell’s (1989) work. Participants in both conditions were able to
amination grades, F(2, 50)
0.08, MSE
1.04, ns.
write about their goals and wrote passages that were approximately equal
in length.
E-mail manipulation booster.
At the end of the experimental session,
4 Of the participants, 63% endorsed the goal by indicating “Yes, I have
participants were informed that the second component of the study would
this goal.” The remaining 37% of participants indicated that “Yes, I sort of
take place approximately 10 days later, but that the exact date, time, and
have this goal.”

---

756
BURTON, LYDON, D’ALESSANDRO, AND KOESTNER
Table 1
when examining the psychological well-being of only those par-
Study 2a: Mean Satisfaction With Life Scores as a Function of
ticipants in the identified regulation condition, the partial correla-
Experimental Condition
tion between changes in psychological well-being and midterm
exam grade perceptions, controlling for previously obtained course
Point of assessment
grades, did not reach statistical significance, pr(14)
.27, p
.26.
Condition
Time 1
Time 2
Exploratory Analyses: Days Spent Studying and the
Control
5.17
5.01
Intrinsic Manipulation
Identified
4.73
4.87
Intrinsica
5.00
5.53
To investigate a possible mechanism for how the intrinsic ma-
a
nipulation led to greater psychological well-being, we conducted
Significant increase from Time 1 to Time 2, p
.01.
exploratory analyses. In theory, the intrinsic manipulation should
have led to a cognitive association between the engagement in
course activities, such as studying, and feelings of enjoyment. As
Effect of Perceived Midterm Grades on Psychological
a result, the more that students in this condition spent time study-
Well-Being
ing for the midterm examination, the more that they should have
We conducted further analyses to examine the role of students’
experienced greater well-being. Therefore, we investigated differ-
midterm exam grade perceptions in the prediction of subjective
ences in the relationship between the amount of time spent study-
well-being after the midterm examination had taken place. On the
ing and changes in psychological well-being from Time 1 to Time
basis of the results of Study 1 that indicated that the psychological
2. In both the control condition, r(17)
.11, ns, and the identified
well-being of those who have highly intrinsic regulation is not
condition, r(20)
.05, ns, correlations were not significant. How-
contingent on their performance, we believed that the well-being
ever, in the intrinsic condition there was a significant relationship
of students in the intrinsic induction condition would not be related
between changes in satisfaction with life and the amount of time
to their perceived performance on the test. In contrast, we believed
spent studying, r(15)
.56, p
.02. The more that participants in
that among students not in the intrinsic condition, those who felt
the intrinsic condition studied, the greater was the increase in their
that they had performed well on the exam would experience
psychological well-being from Time 1 to Time 2. A comparison of
greater subjective well-being than those who felt that they had
effect sizes revealed that this correlation was significantly greater
performed poorly.
than those in the identified and control conditions, z
1.75, p
In a hierarchical multiple regression, Time 1 satisfaction with
.05, one-tailed. It should be noted that the manipulation did not
life scores were entered to control for baseline levels of well-being.
lead participants in the intrinsic condition (M
4.62) to spend
Then, students’ perceptions of their midterm exam grades and
more time studying than those in the identified condition (M
dummy coded variables reflecting the manipulation were entered.
4.55) or in the control condition (M
4.97), F(2, 55)
1, ns. In
Finally, the interactions between students’ grade perceptions and
addition, the manipulation did not interact with any measure of
the manipulation variables were entered. To test for three levels of
individual differences to predict the number of days that students
the experimental manipulation, it was necessary to create two
spent studying, Fs(2, 55)
1, ns.
dummy codes, one contrasting the intrinsic condition with the
other groups and the second contrasting the identified condition
Summary of Study 2a Findings
with the other two groups. Results revealed that the intrinsic
Results indicate that the manipulation of intrinsic self-regulation
condition was a highly significant predictor of well-being,
.28,
in students’ approach to an academic goal significantly improved
t(53)
3.08, p
.003, and that grade perceptions approached
significance in prediction of well-being,
.25, t(53)
1.72,
p
.09, following the midterm. In addition, the Perception of
Midterm Grades
Condition interaction approached significance,
1
.24, t(53)
1.96, p
.055. This interaction is depicted
in Figure 2, whereby the psychological well-being of those stu-
dents in the intrinsic manipulation condition was not affected by
0.5
how well they felt that they did on the midterm exam ( pr
.15),
Intrinsic Condition
but the well-being of all other participants was contingent on their
0
Other Conditions
perception of having performed well versus having performed
poorly ( pr
.30). Among those not in the intrinsic condition,
students experienced greater well-being if they felt that they had
-0.5
performed well on the exam but experienced lower levels of
well-being if they felt they had performed poorly. The intrinsic
Time 2 Satisfaction with Life (z score)
-1
manipulation appeared to serve as a protective factor against these
Low High
contingencies, with those in the intrinsic condition experiencing
Perceived Midterm Exam Grade
greater psychological well-being regardless of their perceptions of
achievement.
Figure 2.
Study 2a: Time 2 satisfaction with life (z score) as predicted by
There was no significant main effect or interaction with the
condition and students’ perceived midterm examination grades, controlling
dummy code for the identified condition, ts(53)
1, ns. Similarly,
for Time 1 satisfaction with life.

---

EFFECTS OF INTRINSIC AND IDENTIFIED MOTIVATION
757
their subsequent psychological well-being. This intrinsic regula-
strings were nonwords, which resembled actual words but had letters
tion manipulation appeared to serve as a protective factor against
missing or out of order, and the other half were words. Some words were
the grade-contingent changes in well-being that were experienced
determined to be of neutral valence (Anderson, 1968) and would later be
by the other participants in the study. Finally, exploratory analyses
used to control for individual differences in baseline responding. Words
representing the two different types of self-regulation were selected by
suggested a possible mechanism for how students in the intrinsic
again borrowing from Ryan and Connell’s (1989) Self-Regulation Scale.
manipulation condition experienced greater well-being following
Intrinsic items included the words interesting, enjoyable, fun, exciting, and
the midterm exam. In that condition, the more time that students
fascinating, whereas identification items included important, meaning,
spent studying, the greater was the increase in psychological
value, identity, and worthwhile. All strings were randomly presented and
well-being from Time 1 to Time 2. Presumably, working toward
response latencies to each were measured in milliseconds.
their academic goal with an intrinsic focus allowed them to expe-
Permission to obtain grades.
At the conclusion of the session, the
rience
the
positive
outcomes
associated
with
intrinsic
experimenters explained to participants that they were also interested in
self-regulation.
students’ academic performance and how it may be related to variables
assessed during the course of the study. Participants were then provided
with the option of giving the experimenters consent to obtain their final
Study 2b
examination grades. All but 4 individuals consented.
Study 1 revealed that identified self-regulation was associated
with better academic performance, as indicated by students’ report
Results
card grades. In Study 2b, we sought to expand on this finding in
Preliminary Analyses
four ways. First, we extended the time frame between the assess-
ment of self-regulation and academic performance to 6 weeks.
To verify that the experimental manipulations in Study 2a did
Second, we assessed students’ academic self-regulatory styles two
not influence the implicit measures of self-regulation in Study 2b
thirds of the way through the course to ensure that regulation
or students’ final examination grades, we conducted analyses of
scores were based on a substantial amount of actual experience in
variance. Results indicated that implicit measures (i.e., response
the goal pursuit (Fazio & Zanna, 1981). Third, we were able to use
latencies) did not differ between the experimental conditions out-
previously obtained course grades as a baseline for academic
lined in Study 2a, Fs(2, 50)
0.16 – 0.71, MSEs
1.05–1.10, ns.
performance. Last, we created a cognitive measure of self-
Similarly, final examination grades were not affected by the earlier
regulation that enabled us to examine the independent, and possi-
Study 2a manipulations, F(2, 50)
0.78, MSE
1.01, ns. In
bly interactive, effects of an explicit self-report measure and an
addition, explicit measures of self-regulation, assessed prior to the
implicit cognitive measure in predicting academic performance.
manipulation in Study 2a, were not affected by experimental
condition, Fs(2, 62)
0.32– 0.74, MSEs
1.01–1.02, ns. Thus,
Method
the manipulations in Study 2a did not influence implicit or explicit
self-regulation scores nor did they influence final examination
Participants
scores.
Of the 60 participants in Study 2a, 53 remained in the sample for Study
2b. Of the participants, 4 did not consent to having their course grades used
Predicting Final Exam Grades: Identified Versus Intrinsic
in the research, and 3 did not have a full set of grades for the semester,
Regulations
leaving a total of 53 participants for analyses involving academic
performance.
To create indices of implicit intrinsic and identified regulation,
we first performed a logarithmic transformation, as suggested by
Procedure
Fazio (1990), to normalize the positively skewed distribution of
reaction time data (Bargh & Chartrand, 2000). Then, to create
Approximately 6 weeks prior to the final examination in the course,
indices for both of the self-regulation styles in question, we ag-
participants completed a lexical decision task to ascertain the cognitive
gregated response latencies to the intrinsic and identified words,
accessibility of intrinsic and identified self-regulation words. This task was
respectively. Following this, we removed the variance associated
included to provide an implicit measure of intrinsic and identified regula-
with response latencies to the words of neutral valence from each
tions. In addition, participants’ scores on the intrinsic and identified items
of Sheldon and Kasser’s (1998) measure of perceived locus of causality,
index to control for individual differences in general speed of
collected at the beginning of Study 2a, were used as direct indices of
responding. Analyses involving the implicit measures of self-
intrinsic and identified regulation. On a 9-point scale, participants rated
regulation reflected these computed values. Lower scores on these
their agreement with reasons for pursuing the academic goal of mastering
indices denote faster response latencies and therefore higher levels
course material. On average, students reported high levels of intrinsic (M
of intrinsic and identified motivation.
6.19, Mdn
7) and identified (M
7.03, Mdn
7) self-regulation.
The explicit (self-report) and implicit (response latencies) mea-
Lexical decision task.
Using E-Prime software (Psychology Software
sures of identification were significantly correlated with each
Tools, n.d.), we presented participants with letter strings on a computer
other, r(51)
.27, p
.05, and with this same pattern, the
monitor and asked them to indicate whether each string constituted a word
correlation between explicit and implicit measures of intrinsic
or a nonword by pressing particular keys on the computer keyboard. In the
regulation approached significance, r(51)
.25, p
.07. In
initial instructions, participants were reminded of the course name and
number. To establish the appropriate context, prior to the presentation of
addition, explicit measures of identified and intrinsic self-
each letter string, the course number 333 was subliminally flashed for 20
regulations were correlated with each other, r(51)
.31, p
.02,
ms, followed by a mask of XXX, similar to the methodology used by
as were the implicit measures of identified and intrinsic self-
Mikulincer, Birnbaum, Woddis, and Nachmias (2000). Half of the letter
regulations, r(51)
.29, p
.03.

---

758
BURTON, LYDON, D’ALESSANDRO, AND KOESTNER
In addressing our hypothesis that measures of identified self-
partial correlation between response latencies to intrinsic words
regulation, both explicit and implicit, would significantly predict
and final exam grades did not reach significance, pr(50)
.23,
academic performance, we conducted two hierarchical multiple
p
.11.
regression analyses in which we sought to predict final examina-
Then, in a hierarchical multiple regression analysis, we entered
tion grades. In the first regression, we entered students’ previous
previous course grades to again control for individual differences
course grades in Step 1 to control for general differences in
in academic achievement in Step 1. As might be expected, previ-
students’ academic achievement. Then, in Step 2, we entered all
ous course grades were a highly significant predictor of final exam
variables representing intrinsic self-regulation: explicit (self-
grades,
.65, t(51)
6.04, p
.001. In Step 2, we entered the
report) intrinsic regulation scores, implicit (response latencies)
terms for each of the explicit measures and each of the implicit
intrinsic regulation scores, and the Explicit
Implicit Intrinsic
measures. The improvement to the model approached significance,
Regulation interaction. Finally, in Step 3 we entered all variables
R2
.08, F
(2, 48)
2.41, p
.10. The sole main effect
change
change
representing identified self-regulation scores (explicit identified,
accounting for unique variance, controlling for the contributions of
implicit identified, and the Explicit
Implicit Identified Regula-
the other measures, was the implicit measure of identified regula-
tion interaction). By examining the R2
from Step 2 to Step 3,
tion that also approached significance,
.26, t(48)
2.02,
change
we sought to determine if the addition of identified self-regulation
p
.07. The explicit measures of identification and intrinsic
significantly contributed to the prediction of students’ final exam-
self-regulation and the implicit measure of intrinsic self-regulation
ination grades above and beyond what was accounted for by
did not explain any unique variance in final exam grades, ts(48)
intrinsic regulation scores. Results indicated that intrinsic regula-
1.1, ns. Finally, in Step 3, we entered the interaction terms between
tion, entered in Step 2, did not significantly predict final exam
explicit and implicit measures of identified self-regulation and
grades, R2
.06, F
(3, 48)
1.88, p
.15, but that
between explicit and implicit measures of intrinsic self-regulation.
change
change
identified regulation, entered in Step 3, did add significantly to the
This provided a significant contribution to the model, R2change
prediction of exam grades, R2
.13, F
(3, 45)
5.04,
.14, F
(1, 46)
4.89, p
.02, with the interaction between
change
change
change
p
.004.
explicit and implicit measures of identification proving to be a
We then conducted a similar second hierarchical multiple re-
significant predictor of final examination grades,
.34, t(46)
gression, this time reversing the order of entry of intrinsic and
3.15, p
.01. The interaction term for intrinsic self-regulation,
identified self-regulation scores. As in the first regression, we
however, was not significant, t(46)
1.
entered previous course grades in Step 1. Then, in Step 2, we
entered the identified regulation variables (explicit, implicit, and
Decomposing the Explicit
Implicit Identified Regulation
Explicit
Implicit), and in Step 3, we entered the intrinsic
Interaction
regulation variables (explicit, implicit, and Explicit
Implicit).
Results revealed that identified regulation significantly predicted
We then conducted an analysis of covariance (controlling for
final exam grades, R2
.15, F
(3, 48)
5.36, p
.003,
previous course grades) to further decompose the Explicit
change
change
but that the addition of intrinsic regulation did not significantly
Implicit Identified Regulation interaction and make specific cell
contribute to the prediction of grades, R2
.05, F
(3,
comparisons. To this end, we used median splits of the explicit and
change
change
45)
1.77, p
.17. Together, the results of these two regression
implicit measures of identified self-regulation. As can be seen in
analyses indicated that identified self-regulation significantly, and
Table 2, this analysis revealed a significant interaction, F(1, 48)
independently of intrinsic self-regulation, predicted students’ aca-
3.98, MSE
72.41, p
.05, such that those with low scores on
demic performance, as measured by their final examination grades,
both the explicit and implicit measures of identification had sig-
but that intrinsic self-regulation itself was not a significant predic-
nificantly lower final exam grades than did those in the other three
tor of performance. The more that students had an identified
groups, ts(48)
2.25, ps
.05. For example, having low self-
self-regulation, the higher were their grades on the final
reported identification did not result in lower examination grades
examination.
unless one also had low implicit identified regulation scores. Those
students who had high levels of either explicit or implicit identi-
fication achieved higher grades on the final exam than did those
Individual Predictors of Final Examination Grades