academic intrinsic motivation in elementary and junior

15
Journal of Educational Psychology 1985, Vol. 77, No. 6, 631-645 Copyright 1985 by the American Psychological Association, Inc. 0022-0663/85/$00.75 Academic Intrinsic Motivation in Elementary and Junior High School Students Adele Eskeles Gottfried California State University, Northridge Results of three studies are presented that demonstrate the significance of aca- demic intrinsic motivation for children's education. As predicted, academic intrin- sic motivation was found to be significantly and positively correlated with chil- dren's school achievement and perceptions of academic competence and negatively correlated with academic anxiety. Evidence supported the view that academic intrinsic motivation is differentiated into school subject areas (reading, math, so- cial studies, science) and is also a general orientation toward school learning. Rela- tions between motivation and perception of competence and anxiety were differen- tiated by subject area, whereas achievement was more pervasively related to general motivation. Math motivation, however, emerged as a unique predictor of math achievement. The significance of academic intrinsic motivation as differen- tiated into subjects and as a general orientation is discussed. Intrinsic motivation concerns the per- formance of activities for their own sake in which pleasure is inherent in the activity itself (Berlyne, 1965; Deci, 1975). Educa- tional implications and significance of in- trinsic motivation have been widely ad- vanced. Intrinsic motivation should be associated with pleasure derived from the learning process itself {Berlyne, 1971), curi- osity (Berlyne, 1971; Maw, 1971), the learn- ing of challenging and difficult tasks (Lep- per, 1983; Pittman, Boggiano, & Ruble, 1983), persistence and mastery orientation (Harter, 1981; Lepper, 1983}, and a high de- gree of task involvement (Brophy, 1983; Ni- cholls, 1983). A great deal of experimental research has been oriented to understand- ing the development of intrinsic motivation in children, and extrapolations of these results to children's education have been This research was supported in part by a grant-in- aid from the Society for the Psychological Study of Social Issues and faculty research grants from Cali- fornia State University, Northridge. Gratitude is extended to Allen Gottfried for his thoughtful comments and to Jim Fleming, Allen Webb, Ida Guillermo, Tom Clendennen, Dorothy Fi- lipowicz, Barnabus Hughes, M. Kevin, and Maryel- len Ambrose for their helpfulness in facilitating var- ious phases of the research. Requests for reprints and a copy of the Children's Academic Intrinsic Motivation Inventory should be sent to Adele Eskeles Gottfried, Department of Ed- ucational Psychology, Monterey Hall, California State University, Northridge, California 91330. advanced (e.g., Pittman et aL, 1983). How- ever, the actual role of intrinsic motivation in children's schooling has received little at- tention. The present studies focused on children's intrinsic motivation specifically for school learning (i.e M academic intrinsic motivation) and provided a new perspective by examining children's academic intrinsic motivation across school subject domains as well as a general motivational orienta- tion. Although some research regarding intrin- sic motivation in the school context has been conducted (Harter & Connell, 1984; Lloyd & Barenblatt, 1984), the role of sub- ject domains in academic intrinsic motiva- tion has been unexplored. Further, the re- lation of academic intrinsic motivation to school achievement and school-related non- cognitive factors remains to be extensively investigated. The purposes of the present research were to investigate the relation be- tween academic intrinsic motivation and achievement as well as noncognitive factors and to examine the relative importance of academic intrinsic motivation as differenti- ated into school subjects in comparison with a general orientation for these rela- tions. The results of three studies are pre- sented in which children of different grades, sexes, and races were included to determine the generality of findings across varying populations. There is a need to develop a generalizable, empirical base to determine 631

Upload: others

Post on 15-Oct-2021

2 views

Category:

Documents


0 download

TRANSCRIPT

Journal of Educational Psychology1985, Vol. 77, No. 6, 631-645

Copyright 1985 by the American Psychological Association, Inc.0022-0663/85/$00.75

Academic Intrinsic Motivation in Elementary andJunior High School Students

Adele Eskeles GottfriedCalifornia State University, Northridge

Results of three studies are presented that demonstrate the significance of aca-demic intrinsic motivation for children's education. As predicted, academic intrin-sic motivation was found to be significantly and positively correlated with chil-dren's school achievement and perceptions of academic competence and negativelycorrelated with academic anxiety. Evidence supported the view that academicintrinsic motivation is differentiated into school subject areas (reading, math, so-cial studies, science) and is also a general orientation toward school learning. Rela-tions between motivation and perception of competence and anxiety were differen-tiated by subject area, whereas achievement was more pervasively related togeneral motivation. Math motivation, however, emerged as a unique predictor ofmath achievement. The significance of academic intrinsic motivation as differen-tiated into subjects and as a general orientation is discussed.

Intrinsic motivation concerns the per-formance of activities for their own sake inwhich pleasure is inherent in the activityitself (Berlyne, 1965; Deci, 1975). Educa-tional implications and significance of in-trinsic motivation have been widely ad-vanced. Intrinsic motivation should beassociated with pleasure derived from thelearning process itself {Berlyne, 1971), curi-osity (Berlyne, 1971; Maw, 1971), the learn-ing of challenging and difficult tasks (Lep-per, 1983; Pittman, Boggiano, & Ruble,1983), persistence and mastery orientation(Harter, 1981; Lepper, 1983}, and a high de-gree of task involvement (Brophy, 1983; Ni-cholls, 1983). A great deal of experimentalresearch has been oriented to understand-ing the development of intrinsic motivationin children, and extrapolations of theseresults to children's education have been

This research was supported in part by a grant-in-aid from the Society for the Psychological Study ofSocial Issues and faculty research grants from Cali-fornia State University, Northridge.

Gratitude is extended to Allen Gottfried for histhoughtful comments and to Jim Fleming, AllenWebb, Ida Guillermo, Tom Clendennen, Dorothy Fi-lipowicz, Barnabus Hughes, M. Kevin, and Maryel-len Ambrose for their helpfulness in facilitating var-ious phases of the research.

Requests for reprints and a copy of the Children'sAcademic Intrinsic Motivation Inventory should besent to Adele Eskeles Gottfried, Department of Ed-ucational Psychology, Monterey Hall, CaliforniaState University, Northridge, California 91330.

advanced (e.g., Pittman et aL, 1983). How-ever, the actual role of intrinsic motivationin children's schooling has received little at-tention. The present studies focused onchildren's intrinsic motivation specificallyfor school learning (i.eM academic intrinsicmotivation) and provided a new perspectiveby examining children's academic intrinsicmotivation across school subject domainsas well as a general motivational orienta-tion.

Although some research regarding intrin-sic motivation in the school context hasbeen conducted (Harter & Connell, 1984;Lloyd & Barenblatt, 1984), the role of sub-ject domains in academic intrinsic motiva-tion has been unexplored. Further, the re-lation of academic intrinsic motivation toschool achievement and school-related non-cognitive factors remains to be extensivelyinvestigated. The purposes of the presentresearch were to investigate the relation be-tween academic intrinsic motivation andachievement as well as noncognitive factorsand to examine the relative importance ofacademic intrinsic motivation as differenti-ated into school subjects in comparisonwith a general orientation for these rela-tions. The results of three studies are pre-sented in which children of different grades,sexes, and races were included to determinethe generality of findings across varyingpopulations. There is a need to develop ageneralizable, empirical base to determine

631

632 ADELE ESKELES GOTTFRIED

the significance of academic intrinsic moti-vation for children's schooling.

In the present research, academic intrin-sic motivation was conceptualized as bothdifferentiated into academic subject areasand a general orientation toward schoollearning. The pendulum regarding speci-ficity and generality in intrinsic motivationhas swung from a position supporting atrait-oriented approach (Haywood & Burke,1977; Maw, 1971) to one supporting intrin-sic motivation as existing in components(Deci, 1975; Harter, 1981). Recently,Brophy (1983) proposed that student moti-vation to learn is both general (a dispositiontoward learning for its own sake), and situa-tionally specific, depending on such factorsas learning and experience. Therefore, theinclusion in this research of both specificand general aspects of academic intrinsicmotivation provided an opportunity to de-termine the relative contributions of bothapproaches for different outcome measures.Subject areas were chosen as the differenti-ated dimension of academic intrinsic moti-vation because curriculum is generally or-ganized into subject areas and childrenmay develop varying competencies and ex-perience differential success across differ-ent subject areas. Successes in particularareas have been identified as important in-fluences on intrinsic motivation (Brophy,1983; Harter, 1978).

7b investigate academic intrinsic motiva-tion as conceptualized in this research, anew inventory was developed, because noinstruments were available that measureacademic intrinsic motivation specificallywithin subject areas, as well as generallyfor school learning. The new inventory iscalled the Children's Academic IntrinsicMotivation Inventory (CAIMI). It mea-sures the construct of academic intrinsicmotivation defined as enjoyment of schoollearning characterized by an orientation to-ward mastery; curiosity; persistence, task-endogeny; and the learning of challenging,difficult, and novel tasks (Berlyne, 1971;Brophy, 1983; Deci, 1978; Harter, 1981;Maw, 1971; Nicholls, 1983; Pittman et al,1983; White, 1959). The subject areas in-cluded are reading, math, social sciences,and science.

In the present studies, academic intrinsic

motivation, as measured by the CAIMI, wasrelated to the following achievement andnoncognitive variables: standardizedachievement test scores and teachergrades, students' academic anxiety, andperception of academic competence. Thesewere selected on the basis of theoreticalconsiderations detailed in hypotheses pre-sented below. These achievement and non-cognitive variables were distinguished intoreading, math, social studies, and science todetermine their relation to academic intrin-sic motivation in corresponding and non-corresponding subjects as well as to generalintrinsic motivation. In addition, teachers*perceptions of students' academic intrinsicmotivation were measured to determinewhether such motivation was observable toteachers and if teachers' perceptions wouldrelate to students' own reports of academicintrinsic motivation. Finally, the relationof academic intrinsic motivation to chil-dren's intrinsic versus extrinsic motiva-tional classroom orientations was exam-ined.

On the basis of theoretical expectation,the following hypotheses were advanced:(a) Academic intrinsic motivation is posi-tively related to school achievement Be-cause children who experience a great dealof academic intrinsic motivation should en-joy learning and show task persistence anda mastery orientation, it is reasonable toexpect them to strive to learn more andshow higher achievement than would chil-dren who experience relatively less intrinsicmotivation, (b) Academic intrinsic motiva-tion is negatively related to academic anxi-ety. Academic intrinsic motivation andanxiety involve opposing learning orienta-tions (Gottfried, 1982): Whereas intrinsicmotivation involves approach toward learn-ing and mastering challenging and difficulttasks, academic anxiety involves with-drawal from and threat regarding learningnew and challenging tasks because the out-come could be a negative evaluation. Aca-demic anxiety is also a negative correlate ofschool achievement (Gottfried, 1982). (c)Academic intrinsic motivation is positivelyrelated to children's perceptions of their ac-ademic competence. Children with higherlevels of academic intrinsic motivationshould experience task mastery and should

ACADEMIC INTRINSIC MOTIVATION 633

therefore perceive that they are more com-petent in school learning than those withlower levels (Dweck & Elliott, 1983; Harter,1981). (d) Students' academic intrinsic mo-tivation is positively related to teachers'perceptions of students' academic intrinsicmotivation. Children with higher aca-demic intrinsic motivation may evidencebehaviors (ag., investigation, concentra-tion, persistence, challenge orientation)that teachers would recognize, resulting ina positive correlation between student andteacher ratings. And finally, (e| the CAIMIis positively related to' the motivationalscales of Harter's (1981) Scale of IntrinsicVersus Extrinsic Orientation in the Class-room. Harter's scale comprises an intrin-sic and an extrinsic pole, with high scoresrepresenting the intrinsic pole. Positivecorrelations between the CAIMI and this in-trinsic-extrinsic orientation scale would es-tablish that higher academic intrinsic moti-vation is associated with higher intrinsicand lower extrinsic orientation and at thesame time show that the CAIMI is positivelyrelated to another measure of intrinsic mo-tivation.

These hypotheses address issues con-cerning the construct of academic intrinsicmotivation. Across all relations studied,the role of subject areas and a general orien-tation in academic intrinsic motivation wasinvestigated.

MethodSubjects

Study 1

Participants were 141 white middle-class childrenattending fourth and seventh grades in a suburbanpublic school district. There were 77 fourth graders(33 girls and 44 boys} and 64 seventh graders {32girls and 32 boys). The fourth graders attended asingle elementary school. The seventh graders at-tended a single junior high school and had all previ-ously attended the same elementary school that thefourth graders were currently attending. The sam-ple's mean percentiles in reading and math on theStanford Achievement Test (SAT) were 72.4 (SD =21) and 75.8 {SD = 20), respectively.

Study 2

Participants were 260 black and white middle-class children in Grades 4 through 7 of an integratedpublic school. The fourth graders were located in a

primary school, and the fifth through seventhgraders were attending a middle school located inthe same school district. The total sample com-prised 132 girls (68 black and 64 white) and 128 boys(71 black and 57 white). The ns were 39, 67, 77, and77 for fourth, fifth, sixth and seventh grades, re-spectively, with approximately half of the childrendivided between the races and sexes at each age.The mean percentile of the sample on the Total Bat-tery of the Comprehensive Tests of Basic Skills(CTBS) was 64 (SD = 27).

Study 3

Participants were 166 white middle-class boys (n= 82) and girls (n = 84) in Grades 5 through 8 at aprivate school. The ns were 43,44, 36, and 43 at thefifth, sixth, seventh, and eighth grades, respectively,and were approximately evenly divided between thesexes. The mean percentile score of the sample onthe CTBS was 77 (SD = 19).

Measures

Children's Academic Intrinsic MotivationInventory

The CAIMI is a 122-item self-report inventory mea-suring children's intrinsic motivation for schoollearning. It contains five subscales, four of whichmeasure intrinsic motivation in the subject areas ofreading, math, social studies, and science, with thefifth measuring intrinsic motivation as a general ori-entation toward school learning (not differentiatedby subject area). Each of the subject area sub-scales contains 26 items; the General subscale con-tains 18. Items in all four subject areas are identi-cal, except for reference to the particular subject.Items in the General subscale are similar in contentto those in the subject area subscales. Items weredesigned to measure enjoyment of learning; an ori-entation toward mastery; curiosity; persistence;task endogeny; and the learning of challenging, dif-ficult, and novel tasks. The high end of the sub-scales corresponds to high academic intrinsic moti-vation as just defined. The low end of the subscalescorresponds to low academic intrinsic motivationcharacterized by Uttle enjoyment of learning; an ori-entation toward accomplishing easy rather than dif-ficult, challenging, or novel tasks; little curiosity forschool learning; little interest in task mastery; andlow persistence and task endogenous orientation.Items were designed to measure both high and lowacademic intrinsic motivation, and extrinsic motiva-tion was not considered the opposite pole of intrinsicmotivation.

Of the 26 items in each of the subject area sub-scales, 24 are responded to on the basis of a 5-pointLikert scale ranging from strongly agree (1) tostrongly disagree (5). The remaining two items re-quire a forced choice between an intrinsic and nonin-trinsic alternative. In the General subscale, allitems are responded to on the basis of a 5-point Li-kert scale ranging from strongly agree (1) to

634 ADELE ESKELES GOTTFRIED

strongly disagree (5). The items were balanced sothat for approximately half, high intrinsic motiva-tion is indicated by agreement, and for the otherhalf, high intrinsic motivation is indicated by dis-agreement. For scoring, items are appropriately re-versed so that high CAIMI scores correspond to highintrinsic motivation.

In Study 1 the initial version of the CAIMI wasdeveloped, which contained 38 items, of which 28were differentiated into the four subject areas and10 were general. On the basis of the positive resultsof Study 1, the CAIMI was expanded to the 122-iteminventory used in Studies 2 and 3. Representativeexamples of items in the CAIMI include the following.

Subject area items. The same stem is used for theitems, with a separate response for each of the foursubjects. Items for which high motivation is indi-cated by agreement include

I enjoy learning new things in: reading, math,social studies, science.

I feel good inside when I know I have learnedsomething new in:

I enjoy understanding my work in:I like to do as much work as I can in:I would like to learn more about:When I get bored, I look for new things to learn

in:I think it is interesting to do work in:I don't give up on an assignment until I under-

stand it in:I enjoy doing hard assignments in:I like to find answers to questions in:

Items for which high motivation is indicated by dis-agreement include

I give up easily when I don't understand an as-signment in:

New ideas are not interesting to me in:I enjoy doing easy assignments in:

A forced choice subject area item is

Is it more important to you to do a school assign-ment to learn more or get a good grade in:

General items. Those for which high motivation isindicated by agreement include

I keep working on a problem until I understand it.I try to learn more about something that I don't

understand right away so that I will understandit.

I enjoy doing new work in school.When I don't have new things to do in school, I

get bored.When I know I have learned something new, I feel

good inside.I like to learn.When I get bored, I look for new things to do.

Items for which high motivation is indicated by dis-agreement include:

I like to do easy assignments.

I don't like to do more school work than I have to.When I don't understand a problem, I give up

right away.

Reliability of the CAIMI is quite substantial. Bothinternal consistency and test-retest reliability wereestablished. To assess internal consistency, a coeffi-cient alpha was computed for each of the subscales.For the Reading, Math, Social Studies, Science, andGeneral subscales, respectively, coefficient alphaswere .71, .71, .73, .69, and .67 in Study 1 (n = 141);.90, .89, .91, .90, and .80 in Study 2 {n = 260); and.92, .93, .93, .91, and .83 in Study 3 (n = 166).' Inter-nal consistency was quite high, particularly in Stud-ies 2 and 3, indicating substantial item homogeneitywithin the subscales. Test-retest reliability over a2-month interval was established on a randomsample of subjects in Studies 1 and 2.These coefficients ranged from .66 to .76 {df = 83, p< .01) in Study land .69 to .75 [df = 136, p < .01) inStudy 2. These coefficients indicate moderatelyhigh stability over a 2-month interval. For bothinternal consistency and test-retest reliability, coef-ficients were consistent across grade, sex, and race.

In all 3 studies, the subscales of the CAIMI wereintercorrelated (see Table 1). The correlations indi-cate that the CAIMI subscales measured varianceunique to each separate area, although there wassome common variance between them. The propor-tion of shared variance between the subscalesranged from .00 to .42. The average correlation was.39, indicating that the average proportion of vari-ance shared between the subscales was .15. Princi-pal-components analyses with varimax rotation con-ducted in Studies 2 and 3 supported distinctionbetween CAIMI scales.

The CAIMI was developed, worded, and adminis-tered to eliminate the effects of social desirabilityand response biases and acquiescence. It is gener-ally agreed that the most effective method of reduc-ing errors due to these latter variables is throughthe construction of the test itself, such as inclusionof items presenting both positive and negative in-stances, reversals of items, clear and unambiguouswording and directions, and administrative proce-dures emphasizing examiner and subject rapport tomaximize truthful responding (Jackson, 1967; Nun-nally, 1978; Wylie, 1974). The CAIMI was developedin accordance with these recommendations. For ex-ample, the wording of the items was reviewed by apanel of judges (elementary and junior high schoolteachers) to ensure the appropriateness of the vo-cabulary and syntactic constructions for fourth andseventh graders; response set and acquiescencewere minimized by varying wording and changingthe content of items in contiguous positions; severalitems were included that were reversals of eachother; and both positive and negative instances ofacademic intrinsic motivation were included.

In Studies 1 and 2, response acquiescence tenden-cies were tested by correlating pairs of items that

lIn all studies, in computing coefficient alpha, thelength of the General subscale was adjusted to beequivalent to that of the subject area subscales forcomparison with the latter.

ACADEMIC INTRINSIC MOTIVATION 635

Table 1Intercorrelations Between CAIMI Subscales:Studies 1, 2, and 3

Subscales

ReadingStudy 1Study 2Study 3

MathStudy 1Study 2Study 3

SocialStudies

Study 1Study 2Study 3

ScienceStudy 1Study 2Study 3

Math

.02

.40***

.29***

SocialStudies

.22**

.52***

.35***

.16*

.52***

.33***

Science

.20**

.57***

.36***

.13

.45***

.33***

.50***

.54***

.48***

General

.26«*»

.58***

.53«*»

.36**»

.65***

.60***

.23**

.58***

.44***

.15*

.58***

.49***

Note, CAIMI = Children's Academic Intrinsic Moti-vation Inventory. In Study 1, n = 139; in Study 2,n = 260; in Study 3, ns = 139-148.*p < .05. **p < .01. ***p < .001.

were reversals of each other. Across both studies,these correlations were consistently negative andsignificant, ranging from -.69 to - .21 (p < .01 top< .05), indicating that children responded to theitems on the basis of content as they responded ap-propriately on opposite sides of the scale for itemsworded as reversals of each other. Hence responseacquiescence did not characterize the children's re-sponses.

Five items were administered to measure socialdesirability (e.g., "I like everyone I know"), modeledafter other such scales (Castaneda, McCandless, &Palermo, 1956; Crandall, Crandall, & Katkovsky,1965; Lunneborg & Lunneborg, 1964). In all stud-ies, a total social desirability score was obtained andcorrelated with each of the CAIMI subscales. Thesecorrelations were nonsignificant, indicating that so-cial desirability was not related to the CAIMI. Fur-ther, children were informed of the confidentiality oftheir responses prior to administering the CAIMI tomaximize truthful responding.2

Achievement Measures

Standardized tests. In all three studies, the stand-ardized tests used in the schools' testing programswere administered. The achievement tests were ad-ministered concurrently within the same time periodas the CAIMI. In Study 1, the SATs in math, read-ing, and auditory comprehension skills were admin-istered to all children. Social studies and scienceachievement scores were available for the seventhgraders. In Study 2, all children were administered

the CTBS, which yielded scores in reading, lan-guage, and math. In Study 3, the CTBS was ad-ministered to all children, yielding scores in reading,language, math, social studies, and science. TheCTBS also provided a score for knowledge of refer-ence skills (e.g., library resources).

Teacher-assigned grades. In Study 2, final year-end report card grades were available for the currentschool year in reading, math, social studies, and sci-ence. In Study 3, teacher report card grades for thecurrent semester were available in reading, math,social studies, and science. A letter-grade system|A+ to F) was used in both studies.

Academic Anxiety

In all three studies, anxiety was measured withthe reliable and valid Children's Academic AnxietyInventory (CAAI), developed by the present author(Gottfried, 1982) to distinguish anxiety into subjectareas. This inventory measures academic anxiety(worry about classroom learning, tests, and peercomparison with regard to school performance).This inventory is differentiated into the four subjectarea subscales of Reading, Math, Social Studies,and Science and contains a total of 12 items that areresponded to on a 5-point Likert scale from stronglyagree (1) to strongly disagree (5). An example of anitem is "I usually worry about taking a test inmath." Details on reliability and validity, as well asother scale properties, can be found in Gottfried(1982). For data analysis, scores were reversed sothat higher scores corresponded to higher anxiety.

Perception of Academic Competence

In Studies 2 and 3 children were asked to ratetheir perceptions of their academic competence byresponding to four items: "I do well in reading"; "Ido well in math"; "I do well in social studies"; and"I do well in science." Each item was rated on a 5-point Likert scale from strongly agree 11} to stronglydisagree (5). For data analysis, scores were re-versed so that higher scores corresponded to higherperception of competence.

Teachers'Perceptions of Students'Intrinsic Motivation

In Study 3 teachers were asked to rate the degreeof academic intrinsic motivation they perceivedeach of their students possessed in reading, math,social studies, science, and in general. The teacherswere first presented with the definition of academicintrinsic motivation used in this research and werethen asked to rate each student's intrinsic motiva-tion in each of the subject areas and in general on a5-point Likert scale from very low (I) to very high(5).

2Further information about the CAIMI is availablefrom the author.

636 ADELE ESKELES GOTTFRIED

Intrinsic Versus Extrinsic Orientation andConvergent Validity

In Study 3, the CAIMI was concurrently adminis-tered and correlated with an instrument developedby Harter (1981) purporting to measure children'sintrinsic versus extrinsic motivational orientation inthe classroom. Only three of the subscales in thisinstrument are motivational in nature (Harter, 1981)and were hence chosen for analysis. These were (a)Preference for Challenge versus Preference for EasyWork Assigned; (b| Curiosity/Interest versus Pleas-ing the Teacher/Getting Good Grades, and (c> Inde-pendent Mastery versus Dependence on theTeacher. For each item on Harter's subscales, thechild must make a forced choice between an intrinsicversus extrinsic alternative. Higher scores repre-sent higher intrinsic motivation and lower scoresrepresent higher extrinsic motivation.

Procedure

In Studies 1 and 2, the CAIMI and CAAI were indi-vidually administered to each child concurrently in asingle session, with the CAIMI preceding the CAAI.In Study 2, the perception of competence items wereadministered in the same session after the CAIMIand CAAI. An administrator read the instructionsand items of each inventory (and perception of com-petence items in Study 2) to each child. The childhad a copy of the instructions and inventories andfollowed along silently as each was read aloud.Children recorded their own responses on the inven-tories. Instructions had been pilot tested( and allchildren were successful in responding to the inven-tories.

Children spent approximately 30 min completingthe inventories, which were administered in a pri-vate room, untimed, away from the class. Subse-quent to completing the inventories, children wererequested not to discuss the items or their answerswith their schoolmates.

In Study 3 the CAIMI, the CAAI, and the Harterscale were group administered. All children suc-cessfully completed the inventories. Two adminis-tration periods were used on 2 consecutive days,with the CAIMI and CAAI administered on one day,and the Harter scale administered on the other day.The order in which the scales were administered wascounterbalanced so that the CAIMI was administeredfirst for Grades 5 and 7 and the Harter scale wasadministered first for Grades 6 and 8. There wereno significant order effects, so data were collapsedacross order in all analyses. The perception of com-petence items were administered last.

Results

Analyses were conducted on the scores ofeach of the five CAIMI subscales, which wereobtained by summing the scores of theitems within each subscale. Higher scoresrepresented higher intrinsic motivation.

The means and standard deviations for thefive subscales of the CAIMI are presented inTable 2 for all three studies. In each study,multivariate analyses of variance were per-formed on the CAIMI scales (Grade x Sex inStudies 1 and 3; Grade x Sex x Race inStudy 2). Significant multivariate effectswere followed by univariate analyses of var-iance. The only consistent significantgroup differences that occurred across thestudies were (a) intrinsic motivation in read-ing significantly decreased with advancinggrade in Studies 1 and 2 [univariate F(l,132) = 6.40,p = .Ol.AfSp = 15.89, in Study1; F(3, 244) = 12.04, p < .001 MSe =191.81, in Study 2], and (b) intrinsic motiva-tion in social studies significantly increasedwith advancing grade in Studies 1 and 3[univariate F(l, 132) = 3.96, p < .05, MSe- 19.99, in Study 1; F (3,127) = 3.42, p =.02, MSe - 283.54, in Study 3],

In all correlations to be reported, therewere no significant differences in the matri-ces between the grades, sexes, and races.Therefore data were combined across thesevariables. Pearson product-moment cor-relations were used throughout.

Table 2Means and Standard Deviations for Scores onthe CAIMI Subscales: Studies 1, 2, and 3

Subscale

ReadingMSD

MathMSD

Social StudiesMSD

ScienceMSD

GeneralMSD

1{N= 140)

21.24.0

21.24.4

20.14.5

22.34.0

40.15.2

Study

2\N= 260)

96.614.6

101.314.2

94.515.9

96.915.5

72.67.8

3{N= 166)

84.615.5

94.817.7

86.217.7

91.415.1

66.78.2

Note, CAIMI = Children's Academic Intrinsic Moti-vation Inventory. For subject area subscales, num-ber of items and maximum score, respectively, are 7and 29 in Study 1 and 26 and 124 in Studies 2 and 3.For the General subscale, number of items andmaximum score, respectively, are 10 and 50 in Study1 and 18 and 90 in Studies 2 and 3.

ACADEMIC INTRINSIC MOTIVATION 637

Table 3CAIMI Subscales as Related to Perception of Competence and Anxiety: Studies 1, 2, and 3

Subscales

ReadingStudy 1Study 2Study 3

MathStudy 1Study 2Study 3

Social StudiesStudy 1Study 2Study 3

ScienceStudy 1Study 2Study 3

GeneralStudy 1Study 2Study 3

Perception of competence

Reading

.57***

.49***

.16**-.01

.29***

.15*

.25***

.18*

.34***

.25***

Math

.10

.06

.50***

.58***

.10.19**

.13*.21**

.27***

.30***

Socialstudies

.26***

.15*

.20***

.14*

.55***

.60***

.31***

.31***

.29***

.24**

Science

.20***

.16*

.14*

.16*

.25***

.38***

.53***

.62***

.23***

.23**

Reading

-.46***-.44***_ .47***

.11-.16**-.07

-.06-.32***-.14*

-.09_,24***-.17*

-.01-.36***-.35***

Anxiety

Math

.02-.15**-.22**

_ .46***-.41***- .52***

-.08-.19***-.10

-.01-.17**-.19**

-.05-.29***- .39***

Socialstudies

-.18*-.30***-.20**

-.02-.21***-.08

-.49***- .49***-.50***

-.14-.32***-.22**

-.10-.31***- .29***

Science

-.08-.29***-.22**

-.05-.19**»-.10

_ 23**-.30***-.18*

-.38**-.49***-.50***

-.10-.37***- .29***

Note, CAIMI = Children's Academic Intrinsic Motivation Inventory. Sample size iV8 for perception ofcompetence: Study 2, 260; Study 3, 149-156. Sample size Na for anxiety: Study 1, 129; Study 2, 260;Study 3,149-156.Study 3, 14y-15t>.*p < .05. **p < .01. ***p < .001.

Intrinsic Motivation as Related toPerception of Academic Competenceand Anxiety

Perception of competence. The findingssupported the hypotheses of a positive rela-tion between academic intrinsic motivationand perception of academic competence.Specific subject areas played an impor-tant role in differentiating these relations.Correlations were computed between eachof the CAIMI subscales and perception ofcompetence items (see Table 3). There wasa strong consistent trend within each studyfor correlations between corresponding sub-ject areas to be of higher magnitudes thanthe correlations between noncorrespondingsubject areas and between the General in-trinsic motivation subscale and perceptionof competence items. Significant positivecorrelations within corresponding subjectareas ranged from .49 to .62 (p < .001) andindicated that children with higher intrinsicmotivation in a specific subject area per-ceived themselves as more competentwithin that subject area compared withchildren with lower intrinsic motivation.

These results showed that children distin-guished both their academic intrinsic moti-vation and perception of competence bysubject area and revealed the importance ofmeasuring academic intrinsic motivationseparately in subject areas.

Anxiety. The data clearly supported thehypothesis of a negative relation betweenacademic intrinsic motivation and anxiety.Correlations were computed between eachof the CAIMI and CAAI scales (see Table 3).Like results for perception of competence,the correlations between motivation andanxiety were differentiated by subject area.Within each study, correlations betweencorresponding motivation and anxiety sub-ject areas were of the highest magnitudes,ranging from -.38 to -.52 (p < .001), com-pared with correlations between noncorre-sponding subject area subscales and theGeneral intrinsic motivation scale and anxi-ety. Children with higher academic intrin-sic motivation in a specific subject area hadlower academic anxiety in that subject areathan did children with lower motivation.Hence, children differentiated both their ac-ademic intrinsic motivation and anxiety bysubject area, and those with higher motiva-

638 ADELE ESKELES GOTTFRIED

Table 4Correlations Between CAIMI Subscales and Achievement Test Scores: Studies 1, 2 and 3

SubscalesReading

Study 1Study 2Study 3

MathStudy 1Study 2Study 3

Social StudiesStudy 1Study 2Study 3

ScienceStudy 1Study 2Study 3

GeneralStudy 1Study 2Study 3

R'Study 1Study 2Study 3

Reading

.15*

.02

.33***

.07

.14*

.13

.23**

.12*

.21**

.19*

.12*

.20**

.08

.20***

.28***

.28*

.24**

.36**

Language

.30***

.07

.24**

.13

.26***

.20**

.21**

.16**

.08

.17*

.14*

.06

.12

.27***

.19*

.40***

.32**

.28*

Achievement test

Math

.06

.07

.10

.22**

.34***

.16*

.17*

.08

.09

.13

.20***

.09

.01

.29***

.05

,29*.40***.20

Socialstudies

.13

.19*

.07

.15*

.13

.05

.02

.17*

.07

.21**

.21

.23

Science

.32**

.17*

.15

.14

.25*

.07

.08

.18*

.08

.18*

.44*

.23

Referenceskills

.19*

.17*

.15*

.13

.20*

Note, CAIMI = Children's Academic Intrinsic Motivation Inventory. In Study 1, N =55 (Grade 7 socialstudies, science) to 130 {total sample); in Study 2,N = 247 to 249; in Study 3, N = 125 to 136.'Multiple correlation between all CAIMI subscales and achievement.*p < .05. **p < .01. ***p < .001.

tion had lower anxiety within specific sub-ject areas.

Correlations Between Academic IntrinsicMotivation and Achievement

Bivariate and multiple correlations be-tween the CAIMI subscales and standard-ized achievement scores and teacher gradesare presented in Tables 4 and 5, respec-tively.3

The results supported the hypothesisthat academic intrinsic motivation is posi-tively and significantly related to children'sschool achievement as measured by bothstandardized achievement tests andteacher grades. The multiple correlations(with all CAIMI subscales) showed thatachievement in every subject area (exceptfor social studies standardized achievementscores) was significantly correlated withthe CAIMI, with significant correlationsranging from .24 to .44. Hence, academic

intrinsic motivation accounted for up to ap-proximately 20% of the variance in schoolachievement.

The bivariate correlations were examinedto determine if any patterns appeared re-garding the correlation of the CAIMI sub-scales with achievement. For standardizedscores and teacher grades, 55% and 62.5%of the correlations were significant, respec-tively. Two patterns emerged. Forteacher grades, the General motivationsubscale was significantly and positivelycorrelated across all measures. This wasnot true for any other subscale and was rep-licated in Studies 2 and 3. For standard-

3Correlations between the CAIMI subscales and na-tional percentiles are presented for the standardizedachievement data because this measure provided fora common metric across the three studies. In Stud-ies 1 and 2, scaled scores were available. Correla-tions between the CAIMI and scaled scores proved tobe virtually identical to correlations with percen-tiles. Correlations between percentile and scaledscores were all above .9.

ACADEMIC INTRINSIC MOTIVATION 639

ized test scores, results of Studies 2 and 3revealed a similar trend for the Generalsubscale to be correlated with all but onemeasure of achievement. The second trendthat emerged was that within each study,math achievement showed the highest cor-relation with the Math CAIMI subscale com-pared with its correlation with any othermotivation subscale, and the Math sub-scale was significantly related to everymeasure of math achievement across allthree studies. Hence, whereas the GeneralCAIMI subscale was the most consistent in-trinsic motivation correlate of academicachievement, for math achievement, themagnitudes of the correlations with theMath CAIMI subscale exceeded those of thecorrelations with the General subscale.

Hierarchical multiple regression analyseswere conducted to determine if a motiva-tion subscale independently predictedachievement in its corresponding subjectarea, controlling for the other motivationalsubscales. Across all three studies, it was

Table 5Correlations Between CAIMI Subscales andTeachers' Grades: Studies 2 and 3

Sub-scales

ReadingStudy 2Study 3

MathStudy 2Study 3

SocialstudiesStudy 2Study 3

ScienceStudy 2Study 3

GeneralStudy 2Study 3

R*Study 2Study 3

Reading

.08

.20**

.25***

.07

.14*

.07

.09

.11

.24***

.21**

.29**

.24*

Teachers'

Math

.01

.10

.32***

.42***

.06

.06

.07

.15*

.20***

.29***

>37***.44*»*

grades

Socialstudies

.06

.26***

.25***

.25***

.18**

.20**

.13*

.23 ••

.19**

.31***

.27**

.35*

Science

.03

.25***

.16*

.13

.09

.15*

.08

.25***

.17*

.26***

.24*

.32*

Note, CAIMI = Children's Academic Intrinsic Moti-vation Inventory. In Study 2,-JV = 247 to 248; inStudy 3, N = 143 to 154.aMultiple correlation between all CAIMI subscalesand teachers' grades.*p < .05. **p < .01. ***p < .001.

consistently found that the Math CAIMIsubscale was a significant, independent,and unique predictor of math achievement,controlling for all other motivation sub-scales including the General subscale.Math motivation was the only significantpredictor of standardized math achieve-ment in Study 1, F(l, 115)= 6.02, p = .016,MSe = 450.43, and math teacher grades inStudy 2^(1,161) = 18.63, p < .001, MSe =7.25 and Study 3, F<1, 112) = 24.39 p <.001, MSe = 6.64. In Study 2, the Math

. subscale was the primary independent pre-dictor, accounting for the most variance ofstandardized math achievement, F(3, 244)= 13.82, p < .001, MSe = 626.75. TheMath CAIMI subscale accounted for up to18% of variance in math achievement.Math proved to be the only CAIMI subscalethat was a consistent independent predic-tor of its corresponding achievement areabeyond the contribution of any other moti-vation subscale, across all 3 studies. Theseresults were consistent across grade, sex,and race.4

Correlations between the CAIMI subscalesand the Reference Skills subtest of theCTBS were computed in Study 3. It isinteresting to note that this area was posi-tively correlated with intrinsic motivation,indicating that children with higher moti-vation had more knowledge of use of li-brary procedures and selection of referencematerials. Correlations are presented inTable 4. This finding is consistent withtheory because children with higher intrin-sic motivation should seek knowledge morethan should children with lower motivationand should therefore have better knowledgeof reference skills.

Correlations Between CAIMI Subscales andIntrinsic Versus Extrinsic MotivationalOrientation

In Study 3, correlations were computedbetween the CAIMI subscales and the moti-vational subscales of Harter's (1981) Scaleof Intrinsic versus Extrinsic Orientation inthe Classroom (see Table 6). The results

'Preliminary regression analyses showed thatwhen grade, aex, and race were included, they werenonsignificant terms and hence were dropped fromthe model.

640 ADELE ESKELES GOTTFRIED

Table 6CAIMI Subscale Scores as Related to Harter's(1981) Scale of Intrinsic Versus ExtrinsicOrientation in the Classroom: Study 3

Subscale

ReadingMathSocial

StudiesScienceGeneral

Intrinsic versus Extrinsic scale

Challengevs.

Easy Work

.32***

.64***

.33***

.38***

.62***

Curiosityvs.

Grades

.25»«

.41***

20**.11.41***

Masteryvs.

TeacherDependence

.18*

.42***

.06

.17*

.35***

Note, CAIMI = Children's Academic Intrinsic Moti-vation Inventory. N = 141 to 151.*p < .05. **p < .01 . ***p < .001.

showed pervasive, positive, significant cor-relations with significant rs ranging from.17 to .64 (p< .05 t o p < .001), which sup-ported the hypothesis. The CAIMI showedthe strongest correlations with the intrin-sic-extrinsic subscale measuring the child'sinterest in challenge (intrinsic) versus pref-erence for easy work {extrinsic}, with rsranging from .32 to .64, p < .001, althoughconsistently positive correlations werefound across the three subscales. The nat-ure of the Challenge versus Easy Work sub-scale appears to be the most consistentwith the conception of intrinsic motivationin the CAIMI, resulting in the strongest cor-relations. The Math and General CAIMIsubscales showed the strongest correla-tions across the three intrinsic-extrinsicsubscales, with rs ranging from .35 to .64 (p< .001). These findings revealed that theCAIMI subscales were positively correlatedwith, and showed convergent validity with,another measure of intrinsic motivation.Higher CAIMI scores corresponded to higherorientation toward challenge, curiosity, andmastery, and lower extrinsic orientation to-ward easy work, grades, and teacher depen-dence. It is considered desirable for twomeasures of a construct not to be identicalor provide too much correlational overlap,so that each can contribute its own perspec-tive to the measurement of that construct(Anastasi, 1982). Hence, the correlationsbetween the CAIMI and the Harter scale ap-propriately demonstrated convergence butnot duplication.

Correlations Between CAIMI Scales andTeachers'1 Perceptions of Students'1

Intrinsic Motivation

Correlations between the CAIMI subscalesand teachers' perceptions of students' in-trinsic motivation were computed. Of 25correlations, 12 (48%) were positive and sig-nificant, which provided support for the hy-pothesis of a positive relation betweenteachers' perceptions and students' reportsof intrinsic motivation. Teachers' ratingsof students' general intrinsic motivationwere significantly correlated with the Read-ing, Math, and General CAIMI subscales,r{df = 146 to 153) = .27, .22, and .25 ip <.01), respectively. Teachers' ratings of stu-dents' intrinsic motivation in each of thesubject areas were significantly correlatedwith students' CAIMI scores on the Mathand General subscales, with rs rangingfrom .20 to .40 ip < .05). The highest cor-relation occurred between the Math CAIMIsubscale and teachers' ratings of students'math motivation (r = .40, p < .001).

Overall, these correlations indicated thatteachers' perceptions of students' intrinsicmotivation were related to students' re-ports of their own intrinsic motivation,with the Math and General CAIMI subscalesshowing the most consistent correlationwith teacher ratings. Except for Math, thecorrelations were not distinguished by spe-cific subject area Teachers appeared torate students' general and subject area mo-tivation similarly. This is not surprising inview of teachers' statements that ratingstudents' motivation in the subject areaswas more difficult than was rating generalmotivation. Because teachers taught all ofthe subject areas in self-contained class-rooms, they may have generalized their per-ceptions of students' intrinsic motivationacross areas. It is noteworthy that despitethe overall generality of the correlations,the Math CAIMI subscale again emerged asa specific component of intrinsic motiva-tion.

Although the hypothesis was supported,correlations tended to be low. Because thechildren's intrinsic motivation scores corre-lated more strongly, and as predicted, withother measures, it is likely that teachers'ratings were not as accurate as the stu-

ACADEMIC INTRINSIC MOTIVATION 641

dents'. Each teacher in Study 3 had morethan 35 children in the classroom. With somany students, it is probably difficult to beknowledgeable about the intrinsic motivesof each student within and between subjectareas. In addition, teachers are not privyto the internal states of students but canonly make inferences about intrinsic moti-vation on the basis of overt behaviors, stu-dents' verbalizations, or indirect sources(such as parents or other teachers). There-fore, students are expected to be moreaware of their own intrinsic motivationthan their teachers. Despite these difficul-ties, the correlations show that teacherswere able to assess students' intrinsic moti-vation with some degree of accuracy, partic-ularly in math.

Partial Correlations

To demonstrate that the correlations be-tween intrinsic motivation (CAIMI) and anxi-ety (CAAI) are independent of achievementand perception of competence, partial corre-lations were conducted controlling for theselatter variables (Studies 2 and 3). Bothfirst- and second-order partial correlationswere computed (i.e., achievement and per-ception of competence were controlled sepa-rately and together). For achievement,both CTBS scores and teachers' gradeswere controlled in separate analyses. Par-tial correlations were obtained within cor-responding subject areas (e.g., readingachievement and perception of competencein reading were partialed from the relationbetween intrinsic motivation and anxiety inreading), between noncorresponding sub-ject areas (e.g., reading achievement andperception of competence in reading werepartialed from the relations between intrin-sic motivation in reading and anxiety inmath, social studies, and science), and be-tween the General CAIMI and each CAAIsubscale.

Results showed that the CAIMI and theCAAI continued to be significantly andnegatively correlated beyond the varianceattributable to achievement and perceptionof competence in both the first- and second-order analyses. These partial correlationswere highly similar to the Pearson correla-tions reported in Table 3 with regard to sig-

nificance, magnitudes, and subject area dif-ferentiation. Hence, the relations betweenthe CAIMI and the CAAI are unique and notdue to achievement and perception of com-petence.

To farther demonstrate the uniqueness ofthe construct of intrinsic motivation be-yond the variance attributable to achieve-ment and perception of competence, first-and second-order partial correlations werealso computed between the CAIMI and Har-ter subscales controlling for these variables(Study 3). Achievement (CTBS scores andteacher grades) and perception of compe-tence in each subject area were partialedfrom all correlations between the CAIMI andHarter subscales. These partial correla-tions were virtually the same as the Pear-son correlations reported in Table 6 withregard to significance, magnitudes, andpatterns. Therefore, the CAIMI and Hartersubscales are significantly and positivelycorrelated beyond the variance attributableto achievement and perception of compe-tence. These analyses provide an impor-tant validation of the intrinsic motivationconstruct by demonstrating the uniquenessof the correlations between these twoscales.

Partial correlations were also computedthat clearly showed that the CAIMI correl-ated with achievement, anxiety, perceptionof competence, and teachers' perceptions ofstudents' intrinsic motivation independentof the Harter subscales. First-order par-tial correlations between the CAIMI andeach of these criterion variables were con-ducted in which each of the three Hartersubscales was controlled. The CAIMI con-tinued to show significant correlations withanxiety and perception of competence afterpartialing out the Harter subscales, withvirtually the same magnitudes, directions,and patterns of significance (including sub-ject area differentiation) as the Pearson cor-relations reported in Table 3. Likewise,partial correlations between the CAIMI andachievement (CTBS scores and teachergrades) continued to be positive and signifi-cant beyond the Harter subscales andshowed highly similar magnitudes to thePearson correlations reported in Tables 4and 5. Math intrinsic motivation contin-ued to show unique and specific relations

642 ADELE ESKELES GOTTFRIED

with math achievement after partialing outthe Harter subscales. Partial correlationsbetween the CAIMI and teacher perceptionsof students' intrinsic motivation showedthat teachers' ratings of students' mathmotivation were significantly, positively,and specifically related to CAIMI Math sub-scale scores, and teachers' ratings of stu-dents' general motivation were positivelyand significantly related to General andReading CAIMI scores, beyond the Hartersubscales. Magnitudes of these partialcorrelations were similar to the Pearson cor-relations reported.

Finally, to show that intrinsic motivationis correlated with achievement beyond chil-dren's ability, partial correlations were con-ducted (Study 1) in which children's IQ(measured by the Otis-Lennon Mental Abil-ity Test [Otis & Lennon, 1967] adminis-tered concurrently with the SAT) was par-tialed from the correlations between theCAIMI and SAT scores (IQ scores were avail-able from the school district only in Study1). When IQ was partialed from these cor-relations, intrinsic motivation in math waspositively and significantly correlated withmath achievement, and intrinsic motiva-tion in reading was positively and signifi-cantly related to language and scienceachievement Hence, intrinsic motivationand achievement were correlated indepen-dent of ability.

Overall, these partial correlations indi-cate that academic intrinsic motivation, asmeasured by the CAIMI, is a unique con-struct and not simply a byproduct of chil-dren's achievement, perception of compe-tence, or ability.

Discussion

The present research has elucidated theconstruct of academic intrinsic motivationwith regard to children's schooling. Theresults support all of the hypotheses, show-ing that children who reported higher aca-demic intrinsic motivation had signifi-cantly higher school achievement, morefavorable perceptions of their academiccompetence, lower academic anxiety, andlower extrinsic classroom orientation, andwere rated by their teachers as being gener-ally more intrinsically motivated. More-

over, generality of findings was demon-strated across studies as well as withchildren varying in grade, sex, and race.

The reliability and validity of the CAIMIwere established across the three presentstudies. Concurrent criterion-related valid-ity with anxiety, perception of competence,and achievement was demonstrated. Fur-ther, the construct validity of the CAIMI hasbeen established through the confirmationof the hypotheses based on theories. Bothpositive and negative correlations were pre-dicted and obtained, indicating convergentand discriminant validity of the CAIMI.The partial correlations further indicatethat the CAIMI is an independent andunique measure of intrinsic motivation.

A major issue examined in this researchconcerned the relative contributions of aca-demic intrinsic motivation as differentiatedinto school subject areas and as a generalorientation toward school learning. Thefindings of this research indicate that aca-demic intrinsic motivation is characterizedboth by differentiation as well as by a gen-eral orientation. The degree of differentia-tion demonstrated by academic intrinsicmotivation depends on the criterion towhich it is related. The most distinct dif-ferentiation occurred for relations betweenacademic intrinsic motivation and aca-demic anxiety as well as perception of aca-demic competence. For both of these lat-ter variables, correlations with the CAIMIsubscales were of much higher magnitudeswithin corresponding subject areas than be-tween noncorresponding subject areas orwith the General CAIMI subscale. There-fore, support for the differentiation of aca-demic intrinsic motivation into subjectareas was demonstrated. For theachievement variables, however, a differentpattern emerged. Multiple regressionanalyses showed that the Math CAIMI sub-scale was specifically and uniquely relatedto math achievement when the other CAIMIsubject area subscales were held constant.This was the only school subject for whichthe corresponding motivation subject areaprovided a unique relation with achieve-ment. For the other subject areas, theGeneral CAIMI subscale provided the mostconsistent correlations with achievement.Therefore, the role of subject areas in intrin-

ACADEMIC INTRINSIC MOTIVATION 643

sic motivation is different for these achieve-ment and noncognitive variables.

Some explanations are possible for thesefindings. First, children may distinguishtheir anxiety and perception of competenceaccording to subject area on the basis ofclassroom experiential factors. Forachievement, however, there may be an en-vironmental press (e.g., from parents,peers, teachers) to do well in all subject ar-eas regardless of preference. Hence,achievement may be more consistentacross subject areas, whereas for anxietyand perception of competence, everydaysuccesses and.setbacks may play more of arole in making subject areas distinctions.Second, anxiety and competence percep-tions may be more differentiated con-structs than is achievement and may there-fore result in more differentiatedcorrelations with the CAIMI scales. Inter-correlations between the subject areas forachievement, anxiety, and perception ofcompetence showed that subject areas hadlower intercorrelations for the two lattervariables than they had for achievement.Across the three studies, the range of corre-lations and the mean correlation betweenthe subject areas were as follows: percep-tion of competence, .14-.61, M = .33; anxi-ety, .27-.65, M = .51; achievement, .49-.84,M = .69. Even higher intercorrelations be-tween subject areas for achievement werereported in the CTBS and SAT manuals onthe basis of national populations. Further,there is evidence that self-concept, a varia-ble similar to perception of academic com-petence, is differentiated into school sub-ject areas (Marsh, Relich, & Smith, 1983;Shavelson & Bolus, 1982}, and literature onacademic anxiety has been supportive ofsituational specificity (see Gottfried, 1982).Third, it is possible that academic intrinsicmotivation, anxiety, and perception of com-petence tap a consistent network of differ-entiated self-perceptions with regard tosubject area, although these self-percep-tions may not necessarily be as veridical inrelation to achievement in subject areas.

The present data provide a new perspec-tive to the issue of specific versus generalintrinsic motivation. The findings of thisresearch support Brophy's (1983) positionthat a general orientation, as well as situa-

tional specificity, exists in student motiva-tion, but goes beyond by delineating thecriterion variables for which academic in-trinsic motivation plays a more differenti-ated role and those for which it plays amore general role. Therefore, neither a po-sition espousing a totally differentiatedview (e.g., Harter, 1981) nor one espousinga trait view (e.g., Haywood & Burke, 1977)is sufficient to account for the range of rela-tions of academic intrinsic motivation toschool-related variables. Both views areclearly needed. Further, differentiation ofintrinsic motivation occurs along differentdimenstions, as indicated by the present re-search, which differentiates between sub-ject areas, and Harter's research, which dis-tinguishes challenge, interest, and mastery.The type of differentiation used needs to beclearly specified.

The singularity of math intrinsic motiva-tion as a specific component of mathachievement is intriguing. It is possiblethat math is perceived as a more difficultand challenging subject area than others.Children with higher intrinsic motivation inmath may therefore be better able to mas-ter challenging and difficult math tasks andshow higher academic achievement in thissubject. This explanation is supported byfindings of Eccles, Adler, and Meece (1984),in which math was rated as more difficultthan English by 8 through 10th graders,and by findings of licht and Dweck (1984),in which fifth-grade children who were mas-tery-oriented performed better on a confus-ing, difficult task than did children who dis-played helplessness with respect toachievement. licht and Dweck suggestedthat the difficulty of specific content areas,such as math, is quite important for differ-ences in achievement between mastery-ori-ented versus helpless children.

It is also interesting that teachers weremost accurate in assessing students' mathintrinsic motivation, as shown by the high-est correlation having been obtained be-tween teachers' perceptions of students'math intrinsic motivation and students'ratings of their own math motivation, com-pared with other correlations betweenteacher and student ratings. Perhaps stu-dents with higher math motivation displaycertain behaviors, such as persistence, con-

644 ADELE ESKELES GOTTFRIED

centration, or direct verbalizations aboutmath and are thereby identified by theirteachers. Teachers may encourage highmath motivation in students with initiallyhigh math motivation.

The magnitudes of the correlations be-tween academic intrinsic motivation andanxiety and perception of competence (incorresponding subject areas) were higherthan correlations between motivation andachievement. The correlations betweencorresponding subject areas between theCAIMI and perception of competence andanxiety ranged from .49 to .62 and - .38 to- .52, respectively. On the other hand, thesignificant multiple correlations betweenacademic intrinsic motivation and achieve-ment ranged from .24 to .44. The magni-tudes of these correlations with achieve-ment are comparable to those reported byUguroglu and Walberg (1979) in a meta-analysis of the relation between motivationand achievement and to those reported byHarter and Connell (1984) and Lloyd andBarenblatt (1984). Motivation measuresmay reach a limit in their relation toachievement, after which achievement isaccounted for by other factors, includingability, time spent on instruction, quality ofinstruction, classroom climate, and homestimulation (Walberg & Uguroglu, 1980).Further, motives other than intrinsic moti-vation may play a role in school learning.In some classrooms, intrinsic motivationmay actually divert children's attentionfrom expected tasks, whereas in others,high emphasis on extrinsic incentives maybe the route to achievement. It has re-cently been suggested that children inter-nalize extrinsic reward systems to self-regu-late their achievement activities (Connell &Ryan, 1984). Hence, although academic in-trinsic motivation is clearly related toachievement, as the present study shows, itplays a role within a complex set of otherfactors.

Subject areas appear to be involved inincreases and decreases in academic intrin-sic motivation across the grades. Readingmotivation decreased whereas social stud-ies motivation increased with advancinggrade. Curriculum emphases or changes ininterests at different grade levels may con-tribute to grade differences in intrinsic mo-

tivation in particular subject areas. Read-ing may be emphasized more and be agreater challenge in the lower grades,whereas social studies may be given moreattention and be of greater interest in theupper grades. Whereas Harter (1981) pro-posed that the school culture causes an in-creasingly extrinsic orientation and hence adevelopmental decline in intrinsic motiva-tion, the present study indicates that sucha decline is not necessarily a general trend,but varies with subject area.

The present research provides several di-rections for educators. First, the CAIMImay be useful for program evaluations inwhich students' motivational developmentis a goal, guiding children into areas of in-terest, and for identification of students'high or low intrinsic motivation for specialprograms or intervention (e.g., Messick,1979). Second, because academic intrinsicmotivation, anxiety, and perception of com-petence are related in specific subject areas,it is necessary for children to be assessedfor intrinsic motivation, anxiety, and per-ception of competence separately in thesesubject areas. Educators must be awarethat for these noncognitive areas, intrinsicmotivation in one subject does not neces-sarily indicate similar trends across all sub-ject areas. Third, for math achievement,math intrinsic motivation should be as-sessed separately. General intrinsic moti-vation assessment would be suitable forreading, social studies, and science achieve-ment.

Finally, intrinsic motivation has beencited as an important educational goaLThis research provides empirical supportfor this goal in achievement and noncogni-tive areas. Because intrinsic motivation isa significant construct in children's educa-tion, the school learning environmentshould foster this important motive.

References

Anastasi, A. (1982). Psychological testing (5thed.). New York: Macmillan.

Berlyne, D. E. (1965). Curiosity and education.In J. D. Krumboltz (Ed.), Learning and the educa-tional process (pp. 67-89). Chicago: RandMcNally.

Berlyne, D. E. (1971). What next? Concluding8ummary. In H. I. Day, D. E. Berlyne, & D. E.

ACADEMIC INTRINSIC MOTIVATION 645

Hunt (Eds.), Intrinsic motivation: A new direc-tion in education (pp. 186-196). Toronto: Holt,Rinehart & Winston of Canada.

Brophy, J. (1983). Conceptualizing student moti-vation. Educational Psychologist, 18, 200-215.

Castaneda, A., McCandless, B. R., & Palermo, D. C.(1956). The children's form of the Manifest Anxi-ety Scale. Child Development, 27, 317-326.

Connell, J. P., & Ryan, R. M. (1984). A develop-mental theory of motivation in the classroom.Teacher Education Quarterly, 11, 64-77.

Crandall, V. C., CrandaU, V. J., & Katkovsky, W. A.(1965). A children's social desirability question-naire. Journal of Consulting Psychology, 29, 27-36.

Deci, E. L. (1975). Intrinsic motivation. NewYork: Plenum Press.

Deci, E. L. (1978). Applications of research on theeffects of rewards. In M. R. Lepper & D. Greene(Eds.), The hidden costs of reward (pp. 193-203).HillsdaleNJ: Erlbaum.

Dweck, C. S., & Elliot, E. S. (1983). Achievementmotivation. In P. H. Mussen (Ed.). Handbook ofchild psychology (4th ed.): Vol 4. Socialization,personality and social development (pp. 643-691).New York: Wiley.

Ecdes, J., Adler, T., & Meece, J. L. (1984). Sexdifferences in achievement: A test of alternatetheories. Journal of Personality and Social Psy-chology, 46, 26-43.

Gottfried, A. E. (1982). Relationships between ac-ademic intrinsic motivation and anxiety in chil-dren and young adolescents. Journal of SchoolPsychology, 20, 205-215.

Harter, S. (1978). Effectance motivation reconsid-ered: Toward a developmental model. HumanDevelopment, 21, 34-64.

Harter, S. (1981). A new self-report scale of intrin-sic versus extrinsic orientation in the classroom:Motivational and informational components. De-velopmental Psychology, 17, 300-312.

Harter, S., & Connell, J. P. (1984). A model of chil-dren's achievement and related self-perceptions ofcompetence, control and motivational orientation.In J. Nicholls (Ed.), Advances in motivation andachievement (pp. 219-250). Greenwich, Ct: JAIPress.

Haywood, H. C, & Burke, W. P. (1977). Develop-ment of individual differences in intrinsic motiva-tion. In I. C. Uzgiris & E Weizmann (Eds.), Thestructuring of experience (pp. 235-263). NewYork: Plenum Press.

Jackson, D. N. (1967). Acquiescence responsestyles: Problems of identification and control.In. I. A. Berg (Ed.), Response set in personalityassessment Chicago: Aldine.

Lepper, M. R. (1983). Extrinsic reward and intrin-sic motivation: Implications for the classroom.In J. M. Levine and M. C. Wang (Eds.), Teacherand student perceptions: Implications for learn-ing (pp. 281-317). Hillsdale, NJ: Erlbaum.

Licht, B. G., & Dweck, C. S. (1984). Determinantsof academic achievement: The interaction of chil-

dren's achievement orientations with skill area.Developmental Psychology, 20, 628-636.

Lloyd, J., & Barenblatt, L. (1984). Intrinsic intel-lectuality: Its relations to social class, intelli-gence, and achievement. Journal of Personalityand Social Psychology, 46, 646-654.

Lunneborg, P. W., & Lunneborg, C. E. (1964). Therelationship of social desirability to other test-tak-ing attitudes in children. Journal of Clinical Psy-chology, 20, 473-477.

Marsh, H. W., Relich, J. D., & Smith, I. D. (1983).Self-concept: The construct validity of interpre-tations based upon the SDQ. Journal of Personal-ity and Social Psychology, 45, 173-187.

Maw, W. H. (1971). Differences in the personali-ties of children differing in curiosity. In H. I.Day, D. E. Berlyne, & D. E. Hunt {Eds.), Intrinsicmotivation: A new direction in education. To-ronto: Holt, Rinehart & Winston of Canada.

Messick, S. (1979). Potential uses of noncognitivemeasurement in education. Journal of Educa-tional psychology, 71, 281-292.

Nicholls, J. G. (1983). Conceptions of ability andachievement motivation: A theory and its impli-cations for education. In S. G. Paris, G. M.Olson, and H. W. Stevenson (Eds.), Learning andmotivation in the classroom (pp. 211-237). Hills-dale, NJ: Erlbaum.

Nunnally, J. C. (1978). Psychometric theory (2nded.). New York: McGraw-Hill.

Otis, A. S., & Lennon, R. T. (1967). Otis-LennonMental Ability Test New York: PsychologicalCorporation.

Pittman, T. S., Boggiano, A. K., & Ruble, D. N.(1983). Intrinsic and extrinsic motivational ori-entations: Limiting conditions on the undermin-ing and enhancing effects of reward on intrinsicmotivation. In J. M. Levine & M. C. Wang (Eds.),Teacher and student perceptions: Implicationsfor learning (pp. 319-340). Hillsdale, NJ:Erlbaum.

Shavelson, R. J., & Bolus, R. (1982). Self-concept:The interplay of theory and methods. Journal ofEducational Psychology, 74, 3-17.

Uguroglu, M. E., & Walberg, H. J. (1979). Motiva-tion and achievement: A quantitative synthesis.American Educational Research journal, 16, 375-389.

Walberg, H. J., & Uguroglu, M. (1980). Motiva-tional and educational productivity: Theories,results, and implications. In L. J. Fyans, Jr.(Ed.), Achievement motivation: Recent trends intheory and research (pp. 114-134). New York:Plenum Press.

White, R. W. (1959). Motivation reconsidered:The concept of competence Psychological Review,66, 297-333.

Wylie, R.C. (1974). The self-concept: Vol 1(Rev. ed.). Lincoln, NE: University of NebraskaPress.

Received April 17,1984Revision received August 12,1985 •