+ All Categories
Home > Documents > Harter 1981 SElf Report Scale of Intrinsic

Harter 1981 SElf Report Scale of Intrinsic

Date post: 18-Oct-2015
Category:
Upload: christiane-arrivillaga
View: 239 times
Download: 1 times
Share this document with a friend
Popular Tags:

of 13

Transcript
  • Developmental Psychology1981, Vol. 17, No. 3, 300-312

    Copyright 1981 by the American Psychological Association, Inc.0012-1649/81 /1703-0300S00.75

    A New Self-Report Scale of Intrinsic VersusExtrinsic Orientation in the Classroom:

    Motivational and Informational ComponentsSusan Harter

    University of Denver

    This article reports on a new self-report scale that taps a child's intrinsic versusextrinsic orientation toward learning and mastery in the classroom. Five separatedimensions are defined by an intrinsic and an extrinsic pole: preference for chal-lenge versus preference for easy work, curiosity/interest versus teacher approval,independent mastery attempts versus dependence on the teacher, independentjudgment versus reliance on the teacher's judgment, and internal versus externalcriteria for success/failure. The reliability and factorial validity of the scale havebeen adequately demonstrated. Additional validity studies are reported. Higherorder factoring reveals two distinct clusters of subscales: The first three dimen-sions form one factor and are interpreted as more motivational in nature; theremaining two are viewed as more cognitive-informational in nature. Develop-mental data reveal that across Grades 3-9 there is a shift from intrinsic toextrinsic on the first motivational cluster. Conversely, there is a dramatic de-velopmental shift from extrinsic to intrinsic on the cognitive-informational clus-ter. Interpretations for these developmental differences are advanced, and theeducational implications are explored. The discussion focuses on the need to beprecise in our conceptualization and operationalization of the term intrinsicmotivation.

    Recent trends within the domain of mo-tivation have led to an increasing emphasis

    An earlier version of this article was presented at themeeting of the Society for Research in Child Develop-ment, San Francisco, March 1979. This research wassupported by Grant HD-09613 from the National In-stitute of Child Health and Human Development, De-partment of Health, Education and Welfare, U.S. Pub-lic Health Service.

    The author would like to acknowledge the extensivecooperation of both the school personnel and pupils fromthe following school systems, without whose assistancethis scale could not have been constructed: the CherryCreek Public School System, the Denver Public SchoolSystem, and the Jefferson County Public School System,all in Colorado; the Ventura School District, Ventura,California; the Killingworth Public School System andthe Wightwood School, in Connecticut; and the Roch-ester Public School System, Rochester, New York. Thecontributions of Jim Connell and Bob Engstrom wereinvaluable in bringing conceptual clarity to the scaleduring the various phases of its evolution. I would alsolike to thank Ed Deci, from the University of Rochester,for providing the data from the New York sample.

    A manual for the scale is now available for $5.95from the author.

    Requests for reprints should be sent to Susan Harter,Department of Psychology, University of Denver, Uni-versity Park, Denver, Colorado 80208.

    on the construct of intrinsic motivation (seeDeci, 1975, for a review of many of the the-oretical models that address this construct).One approach can be seen in the efforts ofexperimental social psychologists, notablyLepper (1980), Deci (1975), and their col-leagues, who are empirically examining at-tributional models that specify the con-ditions under which extrinsic rewardsundermine intrinsic motivation. One alsofinds the concept of intrinsic motivation inthose broad theoretical formulations thathave focused on mastery and competence.Our own approach has taken White's (1959)model of effectance motivation as a point ofdeparture. In his challenge to traditionaldrive theory, White proposed a new moti-vational construct that impels the organismto engage in mastery attempts. He viewedthis need to deal effectively with the envi-ronment as intrinsic, postulating that itsgratification produced inherent pleasure.

    The effectance motive construct has ob-vious heuristic appeal, particularly for thestudy of the developing child in whom striv-ings toward mastery and competence are

    300

  • INTRINSIC VERSUS EXTRINSIC ORIENTATION 301

    universally evident. The global nature of thisconstruct, however, has made it difficult tooperationalize. White's formulation does notreadily lend itself to an empirical test. Thusmy own efforts have focused on refining andextending this formulation, casting it withina developmental framework that could beexamined empirically (see Harter, 1978,1980, for a complete discussion of thismodel).

    The present article describes one empiri-cal effort in this programmatic research,namely, the construction of a self-reportmeasure to assess intrinsic motivation in theelementary school child. A major purposewas to devise an instrument that would makeit possible to test certain predictions postu-lated in the model. These involved hy-potheses concerning the antecedents of in-trinsic motivation as well as correlates (e.g.,perceived competence). One central hypoth-esis was that motivational orientation andperceived competence should be related suchthat children with an intrinsic orientation ina given domain would have higher perceivedcompetence in that domain. Conversely,children with an extrinsic orientation wouldhave lower feelings of competence.

    In constructing a measure sensitive tothese predictions, I was guided by the fol-lowing general considerations:

    1. Rather than view effectance motiva-tion as a global or unitary construct, weshould attempt to delineate the possible com-ponents of this motive system.

    2. Although White's (1959) major em-phasis was on the intrinsic properties of ef-fectance motivation, we should also addressthe issue of extrinsic motivation and examinethe relative strength of intrinsic versus ex-trinsic motivational orientations.

    3. We should examine the components ofthese motive systems within a developmentalframework, charting ontogenetic change.

    4. We should also be sensitive to individ-ual differences in motivational orientationwithin a given developmental level.

    5. In our attempts to devise measures thatare psychologically meaningful as well aspsychometrically sound, we should give care-ful attention to the ecological validity of ourconstruct.

    Classroom learning was chosen as a sit-uation in which the motivational orientationof the child would be particularly relevant.As a starting point, I addressed the followingquestion: To what degree is a child's moti-vation for classroom learning determined byher or his intrinsic interest in learning andmastery, curiosity, and preference for chal-lenge, in contrast to a more extrinsic ori-entation in which the child is motivated toobtain teacher approval and/or grades andis dependent on the teacher for guidance?

    With this as a framework, I delineatedfive dimensions of classroom learning thatcould be characterized as having both anintrinsic and an extrinsic motivational pole:(a) learning motivated by curiosity versuslearning in order to please the teacher, (b)incentive to work for one's own satisfactionversus working to please the teacher and getgood grades, (c) preference for challengingwork versus preference for easy work, (d)desire to work independently versus depen-dence on the teacher for help, and (e) in-ternal criteria for success or failure versusexternal criteria (e.g., grades, teacher feed-back) to determine success or failure.

    My psychometric efforts were guided bythe following criteria. Although I had iso-lated components that seemed meaningful,it was critical to determine whether the scalestructure that I imposed actually emergedin the children's responses such that subscalescores could be meaningfully interpreted.Thus I relied heavily on factor-analytic pro-cedures in examining the structure of thescale. Furthermore, to make meaningful de-velopmental comparisons, it was importantthat the factor structure remain stable acrossthe grade levels examined. I also placed em-phasis on the internal consistency of givensubscales as a primary index of reliability.My sensitivity to the tendency for many self-report measures to pull for socially desirableresponses caused me to devote considerableenergy to the design of a new question for-mat that would offset this tendency.

    The age levels initially selected were thirdto sixth grade, although I subsequently ex-tended the sample to include junior highschool pupils. The initial approach to theexamination of developmental differences

  • 302 SUSAN HARTER

    was to use a cross-sectional design across thisgrade span.

    MethodSubjects

    Over 3,000 pupils have participated in various phasesof my scale construction studies. Data have been col-lected in four states; Connecticut, New York, Colorado,and California.

    Data from six samples are described in this article.Sample A consisted of 130 third through sixth gradersfrom Colorado to whom early versions of the scale wereindividually administered. The initial scale resultingfrom these efforts was then group administered to Sam-ple B, 780 third through sixth graders from New York,and to Sample C, 341 third through sixth graders fromCalifornia. The revised scale was then group adminis-tered to Sample D, 761 third through sixth graders fromNew York, and to Sample E, 793 third through ninthgraders from California. Sample F, 120 third throughsixth graders from Colorado, was group administeredthe final version. In each sample, the socioeconomic levelranged from lower-middle to upper-middle class. Therewere approximately the same number of children at eachgrade level, with about the same number of males andfemales within each grade.

    Original Scale StructureThe content of the initial five subscales, each defined

    by an intrinsic and an extrinsic pole, was as follows:(a) Learning for curiosity versus learning to please theteacher contrasted asking questions to satisfy one's owncuriosity versus learning because the teacher wanted oneto master the material, (b) Incentive to work for one'sown satisfaction versus working to get grades and pleasethe teacher pitted a desire to do schpolwork out of per-sonal interest in learning against learning in order tomeet the teacher's expectations and to insure goodgrades, (c) Preference for challenge versus preferencefor easy work contrasted a desire to engage in chal-lenging schoolwork with a preference for merely doingthe easier work assigned by a teacher, (d) Desire to workindependently versus dependence on the teacher pittedan orientation in which the child chose to work on as-signments and projects on his or her own against anorientation in which the child preferred help and assis-tance from the teacher, (e) Internal criteria of successand failure versus external criteria contrasted the child'sown judgment about how well she or he did on a testor an assignment with a reliance on teacher feedback,marks, or grades as an indication of how well she or heperformed.

    Question FormatThe question format was one initially devised in the

    construction of the Perceived Competence Scale forChildren. The rationale for this format has been de-scribed in detail in the presentation of that scale (see

    Harter, in press). My previous experience with true-false formats has revealed several problems, the mostcritical of which has been their susceptibility to sociallydesirable responses. These problems attenuate both thereliability and the validity of such scales. After consid-erable pilot work, much of which involved the individualinterviewing of children, I devised a structured alter-native format in which the child is presented with thetype of question shown in Figure 1.

    The child is first asked to decide which kind of kidis most like him or her and then asked whether this isonly sort of true or really true for him or her. The ef-fectiveness of this question format lies in the implicationthat half of the children in the world (or in one's ref-erence group) view themselves in one way, whereas theother half view themselves in the opposite manner. Thatis, this type of question legitimizes either choice. Theoption of checking "sort of true for me" or "really truefor me" broadens the range of choices over the typicaltwo-choice format. In addition, none of the choices in-volve the response false. Rather, the child must decidewhich of the options is most true for him or her.

    My confidence in this format was first bolstered bythe fact that the verbal elaborations given for theirchoice indicated that children were giving accurate per-ceptions of their motivational orientations rather thansocially desirable responses. Subsequently, we admin-istered the Children's Social Desirability Scale (Cran-dall, Crandall, & Katkovsky, 1965) to Sample F, whichrevealed correlations between .09 and .15 across thevarious subscales. Three of the values were negative,reflecting a slight tendency for socially desirable re-sponses to be correlated with an extrinsic orientation.

    Earlier versions of the scale contained from six toeight items per subscale. Each item was scored on anordinal scale from 1 to 4, where a score of 1 indicatedthe maximum extrinsic orientation and a score of 4 in-dicated the maximum intrinsic orientation. Thus, in theexample given above, children who indicate that theyknow if they have made a mistake without checkingwith the teacher and describe this as really true forthemselves would receive a 4. Children for whom thispart of the statement is only sort of true would receivea 3. Children who indicate that they need to check withthe teacher and describe this as sort of true would receivea 2, and children for whom this second part of the state-ment is really true would receive a 1. Items were coun-terbalanced within each subscale such that half of thembegan with a statement reflecting an intrinsic orientationand half with an extrinsic orientation. Subscale meanswere then obtained by averaging the item scores.

    ResultsPilot Data on the Original Version

    As emphasized above, it was important todemonstrate the factorial validity of thescale to interpret the dimensions of class-room motivation as meaningful subscales.Considerable exploratory work with individ-ual children (Sample A) resulted in a pool

  • INTRINSIC VERSUS EXTRINSIC ORIENTATION 303

    Really Sort ofTrue True

    for Me for Me

    Sort of ReallyTrue True

    for Me for Me

    Some kids know whenthey've made mistakeswithout checking with theteacher

    Other kids need to checkBUT with the teacher to know

    if they've made a mistake

    Figure I . Sample item.

    of items that were comprehensible to ele-mentary school children and appeared to tapthe orientations in question at the item level.The scale was then group administeredto two separate samples of third-, fourth-,fifth-, and sixth-grade children, 780 in NewYork (Sample B) and 341 in California(Sample C). The resulting factor structureswere virtually identical and revealed thatthree of the initial five dimensions describedhere in terms of their intrinsic pole emergedas clear and interpretable factors, althoughnot perfectly defined by all of the particularitems included to tap these dimensions.These three were preference for challenge,independent mastery, and internal criteriafor success failure. Two of the initial di-mensions, incentive to work for one's ownsatisfaction and curiosity, merged as a singlefactor. Finally, one new factor emerged.Examination of the particular items defining-that factor indicated that it could be mean-ingfully interpreted as independent judg-ment versus reliance on the teacher.

    Thus items were reorganized to define thisnew subscale structure and additional itemswere written where necessary to determinewhether an interpretable five-factor solution,reflecting these dimensions, would emerge.

    Revised Scale StructureThe five subscales, described in terms of

    their intrinsic and extrinsic poles, are pre-sented in Table 1. Sample items are givenin Table 2. The revised scale contained 30items, 6 for each subscale. Within each sub-scale, 3 of the items begin with the intrinsicorientation and 3 with the extrinsic orien-tation. With regard to item order, there weretwo constraints: No two consecutive itemsare from the same subscale, and no more

    than two consecutive items are keyed in thesame direction. Each item could receive ascore of from 1 to 4. Item scores are summedfor a given subscale, and a mean is calcu-lated for each separate subscale. Given thefocus on isolating the components of moti-vational orientation, based on identifiablefactors, no total scale score is calculated be-cause such a score would mask subscale dif-ferences manifest in the profiles of individualchildren.

    Factorial Validity of the Revised ScaleFactor analyses were performed on data

    collected on the revised scale from a NewYork sample of 761 third, fourth, fifth, andsixth graders (Sample D) and a Californiasample of 793 pupils, third through ninthgrades (Sample E). Although both orthog-onal and oblique solutions reveal the samebasic factor structure, the oblique rotationsare presented here. Although it was assumedthat individuals would show differencesacross the five subscales, it was also antici-pated that there would be a moderate rela-tionship among subscale scores. Thus anoblique solution, which allows the factors tointercorrelate, was considered to be the mostappropriate. Cattell's (1962) "scree" test,which uses criteria based on the magnitudeof the eigenvalues, indicated that five factorsshould be extracted.

    The factor pattern for the third throughsixth graders in Sample D is presented inTable 3. Item means and standard deviationsare also included. There it can be observedthat the five factors emerge, with items load-ing on their designated factors and with vir-tually no cross-loadings. The factor structurefor the third through ninth graders in Sam-ple E was virtually identical. Average load-

  • 304 SUSAN HARTER

    Table 1Intrinsic Versus Extrinsic Orientation in the Classroom

    Subscale dimension Intrinsic pole Extrinsic pole

    Preference for challangevs

    Preference for easy work

    Curiosity/interestvs

    Pleasing teacher/gettinggrades

    Independent masteryvs

    Dependence on teacher

    Independent judgmentvs.

    Reliance on teacher'sjudgment

    Internal criteriavs.

    External criteria

    Does child like hard,challenging work?

    Does child work to satisfy owninterest and curiosity?

    Does child prefer to work,figure out problems on his/her own?

    Does child feel capable ofmaking judgments aboutwhat to do?

    Does child know when she/hehas succeeded/failed onassignments or tests?

    Does child like easierassignments and subjects?

    Does child do schoolwork tosatisfy teacher, get marksand grades?

    Does child rely on teacherfor help and guidance,particularly when figuringout problems andassignments?

    Is child primarily dependenton teacher's opinion andjudgment about what todo?

    Is child dependent onexternal sources ofevaluation (e.g., teacherfeedback, grades, marks)?

    ings for each of the five subscales in thissample were .53, .50, .46, .50, and .54, in theorder in which the scales are presented inTable 3. In this sample, only two items hadmoderate cross-loadings on other factors.The similarity in factor patterns across sam-ples has also been examined by calculatingcongruence coefficients representing the cor-relation between factor loadings in differentsamples. These correlations range from .67to .84.

    Item Means and Standard DeviationsAs can be seen in Table 3, item means fall

    slightly above the midpoint of 2.5, with stan-dard deviations of slightly less than 1.0.These values are highly replicable acrosssamples. Item means indicate that there areno ceiling or floor effects for particularitems. The standard deviations reveal thevariability for each item, indicating that sub-jects are utilizing the entire range of scores.Reliability

    The reliability of each subscale was firstassessed by using a reliability coefficient

    (Formula 20 from Kuder & Richardson,1937) that provides an index of internal con-sistency. Across samples from New York,California, and Colorado, reliabilities rangefrom .78 to .84, .68 to .82, .54 to .78, .72 to.81, and .75 to .83 for the challenge, mastery,curiosity, judgment, and criteria subscales,respectively. (The relatively low reliabilityof .54 for the curiosity subscale was basedon only three items, omitting those that weresubsequently revised. The .78 value repre-sents a reliability estimate based on sixitems, which includes the three revised items.)Test-retest reliability data were collectedafter a 9-month period for Sample D andafter 1 year for Sample E, with values rang-ing from .48 to .63 across the various sub-scales. Test-retest data from Sample F, col-lected after a 5-month period, yieldedsomewhat higher values, in the .58 to .76range.

    Developmental Trends in Subscale MeansFigure 2 presents the means for each sub-

    scale for Grades 3-9 of California SampleE. This same pattern was replicated in a fol-

  • INTRINSIC VERSUS EXTRINSIC ORIENTATION 305

    Table 2Sample Items and Scoring for Each of the Five Subscales

    Score

    Really Sort oftrue true

    for me for me Subscale dimension

    Score

    Sort of Reallytrue true

    for me for me

    Preference for challenge vs.Some kids like to go on tonew work that's at a more butdifficult level

    Preference for easy workOther kids would rather stick tothe assignments that are prettyeasy to do

    Curiosity/interestSome kids do extraprojects so they can getbetter grades

    vs. Pleasing teacher/getting gradesOther kids do extra projects

    but because they learn about thingsthat interest them

    Independent masteryWhen some kids get stuckon a problem they ask theteacher for help

    but

    Dependence on teacherOther kids keep trying to figureout the problem on their own

    Independent judgment vs. Reliance on teacher's judgmentSome kids think theteacher should decidewhat work to do

    Other kids think they shouldbut have a say in what work they do

    Internal criteria

    Some kids know whetheror not they're doing wellin school without grades

    but

    External criteria

    Other kids need to have grades toknow how well they are doing inschool

    low-up study 1 year later. The trends forthird through sixth grades were replicatedboth in New York (Sample D) and in Col-orado (Sample F).

    As can be seen in the figure, three of thesubscales, preference for challenge versuspreference for easy work, curiosity/interestversus teacher approval, and independentmastery versus dependence on the teacher,begin with relatively high intrinsic scores inthe third grade and show a systematic shifttoward the extrinsic pole across the gradelevels. The opposite linear trend was ob-tained for the remaining two subscales, in-dependent judgment versus reliance onteacher's judgment and internal versus ex-ternal criteria for success/failure. For thelatter two subscales, third graders are rela-tively extrinsic, and with higher grade levels,

    pupils become increasingly intrinsic in theirorientation. Linear trend analyses indicatedthat for all five subscales these trends weresignificant at p < .001.

    Intercorrelations Among SubscalesAlthough the emphasis has been on the

    identification of the components of motiva-tional orientation, moderate correlations wereanticipated among subscales. These intercor-relations are presented in Table 4 for subjectsfrom New York and California Samples Cand D. For both samples, the intercorrelationsamong curiosity, challenge, and independentmastery are moderate to high. Independentjudgment and internal criteria bear a mod-erate relationship to each other but do notcorrelate as highly with the other three sub-

  • 306 SUSAN HARTER

    Table 3Factor Structure From Oblique Rotation, Based on Data from 761 Third Through Sixth Graders(Sample D)

    Subscale/item description 1 2 3 4 5 M S D

    1. ChallengeLike hard, challenging work .73 2.8 .88Like difficult problems .73 2.8 .88Learn as much as I can .42 .37 2.9 .94Like new, difficult work .74 2.9 .83Like hard school subjects .54 2.9 .81Find difficult work interesting .58 2.8 .84

    2. CuriosityWork to learn how to solve3Find out things I want to know .38 2.7 .94Read out of interest3Ask questions to learn3Want extra projects to learn .60 2.5 .97Work to learn new things .60 2.3 1.03

    3. MasteryFigure out things myself .43 3.1 .73Figure out mistakes myself .55 2.9 .86Do hard problems on my own .59 2.7 .88Make my own plans"Figure out assignments on my own .29 2.7 .88Do schoolwork without help .57 2.9 .78

    4. JudgmentLike my own ideas better .68 2.5 .86Stick to my own opinion .70 2.5 .95Learn things that interest me .50 2.8 .94Think my opinions are important .54 2.7 .93Think I should have a say .59 2.4 .92Feel it's best when I decide when to work .62 2.6 .93

    5. CriteriaKnow mistakes without the teacher .28 2.7 .78Know how I'm doing without grades .72 2.8 .97Know without report card .81 2.7 .93Know before I get paper back .62 2.6 .94Know before the teacher tells me .65 2.7 .83Know when I turn work in .48 2.7 .84

    Note, Loadings smaller than .28 not included for the sake of clarity of presentation.3 Wording of new items subsequently revised.

    scales. Higher order factoring revealed that validity of the scale. In one study, differencesa two-factor solution best described this sub- between two groups of fourth, fifth, and sixthscale pattern, with curiosity, challenge, and graders were predicted. These groups variedmastery defining one factor and judgment and in several respects, each of which I expectedcriteria defining the second. would influence motivational orientation.

    The first group consisted of 26 pupils in aValiditv private "open" school from upper-middle-

    class families that strongly supported theThe initial goal, as noted above, was to educational philosophy espoused by the

    establish the factorial validity of the scale, school. The school was chosen on the basisSubsequently, I addressed the discriminant of its emphasis on precisely those principles

  • INTRINSIC VERSUS EXTRINSIC ORIENTATION 307

    | Preference for Challenge Curiosity/Interest Independent Mastery

    2 2-1Preference for Easy Work Assigned ^ Teacher Approval, Grades Dependence on Teacher

    3 4 5 6 7 8 9 3 4 5 6 7

    Independent judgment Internal Criteria

    2 -

    Reliance on Teacher's Judgment

    4 5 6 7 8 9

    Grade Level

    External Criteria for Success/FailureI

    3 4 5 6 7 8 9

    Grade Level

    Figure 2. Mean score by grade level for each subscale.

    captured by the intrinsic pole of our sub-scales. School personnel reinforced childrenfor their intrinsic interest in learning, forcuriosity, for working on their own, and forsetting their own classroom goals. In thecomparison group were pupils matched forage and sex in a "traditional" public school,drawn from lower-middle-class families.These pupils attended the neighborhoodschool to which they had been assigned.Respective mean scores for the two groupswere 2.98 versus 1.81 for challenge (p


Recommended