A self-determination theory approach to predicting school ... et al...Title A self-determination theory approach to predicting school achievement over time: the unique role of intrinsic

A self-determination theory approach to predicting schoolachievement over time the unique role of intrinsic motivationGeneviegraveve Taylor a Tomas Jungert a1 Geneviegraveve A Mageau b Kaspar Schattke cHelena Dedic d Steven Rosenfield e Richard Koestner a

a Department of Psychology McGill University 1205 Dr-Penfield ave Montreacuteal QC H3A 1B1 Canadab Department of Psychology Universiteacute de Montreacuteal Pavillon Marie-Victorin 90 avenue Vincent drsquoIndy Montreacuteal QC H2V 2S9T Canadac Department of Management John Molson School of Business Concordia University 1455 de Maisonneuve Blvd W MB 13107 Montreacuteal QC H3G 1M8Canadad Department of Mathematics Vanier College 821 Ste-Croix ave Montreacuteal QC H4L 3X9 Canadae Department of Physics Vanier College 821 Ste-Croix ave Montreacuteal QC H4L 3X9 Canada

A R T I C L E I N F O

Article historyAvailable online 27 August 2014

KeywordsIntrinsic motivationAcademic achievementSchool performanceSelf-determinationAutonomous motivationLongitudinal

A B S T R A C T

Although many studies have examined the relation of academic motivation to school achievement usingthe Self-Determination Theory perspective the results have been inconsistent The present investiga-tion represents the first systematic attempt to use a meta-analysis and controlled longitudinal studiesto examine the relations of specific types of motivation to overall academic achievement The meta-analysis (Study 1) pointed toward a potentially important role of intrinsic motivation in predicting schoolachievement Three empirical studies of high school and college students in Canada (Studies 2 and 3)and in Sweden (Study 4) showed that intrinsic motivation was the only motivation type to be consis-tently positively associated with academic achievement over a one-year period controlling for baselineachievement Amotivation was significantly associated with lower academic achievement in Studies 3and 4 Interestingly intrinsic motivation was also associated with reduced amotivation in two of our studiesand it was reciprocally associated with higher school achievement in another study Overall our find-ings highlight the unique importance of intrinsic motivation for the future academic success of high schooland college students

1 Introduction

Teachers and parents all know that school motivation is crucialfor academic success which has been long known as a determi-nant for a host of adaptive outcomes such as school completioncareer success mental and physical health (Archambault JanoszMorizot amp Pagani 2009 Guay Ratelle amp Chanal 2008) Howeverthere is little agreement regarding which one should be pro-moted While some researchers focus on intrinsic motivation as themost important (Deci amp Ryan 2000) others emphasize either ex-trinsic motivation (Wigfield amp Eccles 2000) or a combination of bothintrinsic and extrinsic motivation (Elliot amp Moller 2003 LepperCorpus amp Iyengar 2005) The present investigation examines whichtypes of motivation are most beneficial for academic achieve-

ment over time in different school contexts and cultures It alsoassesses whether there are reciprocal relations among academicachievement and different motivation types

11 Self-determination theory in education

Self-Determination Theory (SDT Deci amp Ryan 2000) adopts amultidimensional approach to motivation It specifies different typesof autonomous and controlled forms of intentional action Auton-omous actions are initiated by a sense of choice and personal volitionwhereas controlled actions are regulated by external or internal pres-sures Individuals who are controlled in their actions have an externallocus of causality whereas those who are autonomous have an in-ternal locus of causality (DeCharms 1968) Intrinsic motivation isviewed as the prototype of autonomy (Deci amp Ryan 1985 2000Lepper Greene amp Nisbett 1973) When intrinsically motivatedindividuals freely engage in an interesting activity simply for theenjoyment and excitement it brings rather than to get a reward orto satisfy a constraint (Deci amp Ryan 1985) They perceive them-selves as the causal agent of their own behaviour (DeCharms 1968)By contrast extrinsic motivation is instrumental in nature Behaviour

Corresponding author Deacutepartement drsquoeacuteducation et peacutedagogie Universiteacute duQueacutebec agrave Montreacuteal CP 8888 Succursale Centre-Ville Montreacuteal Queacutebec H3C 3P8Canada Fax +1 514 987 4608

E-mail address taylorgenevieveuqamca (G Taylor)1 Present address Lund University Department of Psychology SE-221 00 Lund

Sweden

Contemporary Educational Psychology 39 (2014) 342ndash358

Contents lists available at ScienceDirect

Contemporary Educational Psychology

journal homepage wwwelseviercom locate cedpsych

that is extrinsically motivated is not performed out of interest butfor the consequence it is thought to be instrumentally linked to(Wrzesniewski et al 2014) Extrinsic motivation is thought to beimportant for socially prescribed activities such as doing home-work because they are often not inherently interesting Unlike manyconceptualizations of motivation (eg Harter 1981) SDT does notview extrinsic motivation as one-dimensional and opposed to in-trinsic motivation Instead it specifies different types of extrinsicmotivation which vary in the extent to which they are autono-mous (Deci amp Ryan 1985)

These types from the least to the most autonomous consist ofexternal introjected and identified regulation External regulationrefers to behaviours that are initiated by an external contingencyfor example being offered a reward to do onersquos homeworkIntrojected regulation refers to internalizing a regulation withoutfully accepting it into onersquos sense of self It involves feelings of in-ternal coercion and pressure and refers to attempts to avoid feelingunworthy guilty or ashamed or to prove onersquos worth (AssorVansteenkiste amp Kaplan 2009) An example of introjected regula-tion would be a student who studies long hours to prove to herselfthat she is worthy Identified regulation takes place when the valueof an instrumental behaviour has come to be identified with onersquossense of self This type of regulation is considered to be more au-tonomous than the other types of extrinsic motivation because itis initiated from a sense of personal meaning and volition (Deci ampRyan 2000 Koestner amp Losier 2002) A student who does extra ex-ercises at the end of a history chapter because she believes it willhelp her fully understand the subject matter is regulated byidentification1 SDT also considers amotivation the absence of mo-tivation that happens when an individual does not experienceintentionality or a sense of personal causation These different formsof motivation have been proposed to lie along a continuum of rel-ative autonomy starting with the form that exhibits the lowest levelto the one that represents the highest level of autonomy (Deci ampRyan 1985) SDT (Ryan amp Connell 1989) also mentions that adja-cent motivations on the continuum (eg intrinsic motivation andidentified regulation) should relate more strongly to each other thandistal ones (eg intrinsic motivation and external regulation)However evidence for the continuum is inconsistent While somefindings corroborate this pattern others deviate from it in variousways (eg intrinsic motivation being more strongly related tointrojected than to identified regulation) (for examples see BoicheacuteSarrazin Grouzet Pelletier amp Chanal 2008 Ntoumanis 2002 OtisGrouzet amp Pelletier 2005 Ratelle Guay Vallerand Larose amp Seneacutecal2007)

12 Academic motivation and educational achievement

Although many studies have examined the relation of academ-ic motivation to school achievement from the SDT perspective (egChatzisarantis Hagger Biddle Smith amp Wang 2003 Deci VallerandPelletier amp Ryan 1991) the majority have been cross-sectional andhave yielded inconsistent results (Cokley Bernard Cunningham ampMotoike 2001 DrsquoAilly 2003 Fortier Vallerand amp Guay 1995

Grolnick Ryan amp Deci 1991 Hardre amp Reeve 2003 Noels Clementamp Pelletier 1999 Otis et al 2005 Petersen Louw amp Dumont 2009Soenens amp Vansteenkiste 2005 Walls amp Little 2005) A careful ex-amination of past research shows that only a few studies haveadopted a prospective design while also controlling for previousachievement (Baker 2003 Black amp Deci 2000 Burton LydonDrsquoAlessandro amp Koestner 2006 Study 2b) The details of these con-trolled prospective studies merit review Burton et al (2006)conducted a 6-week prospective study of university students toexamine the relations of intrinsic motivation and identified regu-lation to final exam performance in a single psychology courseResults showed that controlling for previous grades identified reg-ulation significantly positively predicted final examination gradeswhereas intrinsic motivation was unrelated to the final grades Blackand Deci (2000) examined the relation of relative autonomy in asample of college chemistry students over a one-semester periodThey found that relative autonomy did not significantly predict finalcourse grade after controlling for previous ability and grade pointaverage (GPA) Results for specific types of motivation were not re-ported Finally Baker (2003) examined the relations of academicmotivation types to total GPA in a sample of university students andcontrolled for academic achievement as measured by entry quali-fications upon entering university Her results showed that intrinsicmotivation assessed during the second semester of the first yearof university was the only type of motivation to significantly predictoverall academic performance measured one year later control-ling for entry qualifications

An example of another longitudinal study that has controlled forbaseline achievement is one by (Guay Ratelle Roy amp Litalien 2010)Using a cross-lagged model to examine the reciprocal relations ofacademic motivation and achievement in a population of high schoolstudents they found that autonomous motivation as defined by arelative autonomy score was positively associated with academicachievement over the course of one year even after controlling forbaseline achievement However they did not estimate the contri-bution of each type of motivation to later achievement making itdifficult to understand which type of motivation was driving thisrelation

Given the inconsistent results of past cross-sectional studies andthe paucity of longitudinal studies that have controlled for base-line levels of achievement a more systematic review of the researchis needed in order to fully understand the effect of each differentmotivation type on school achievement Moreover as Ratelle et al(2007) have suggested more longitudinal studies are necessary toprovide some information about the causal mechanisms betweenmotivation and achievement Finally no studies have examined theselongitudinal relations in samples of high school students Since failureto achieve is a prevalent problem in high school and leads to unde-sirable consequences such as dropout it is imperative to conductcarefully controlled studies in such a pre-university population

13 Overview of studies

In an attempt to overcome the limitations of past research weconducted a meta-analysis and a series of three empirical studiesto systematically examine the contribution of the different moti-vation types to school achievement The meta-analysis (Study 1)reviewed cross-sectional and longitudinal studies that have as-sessed the relation of motivation types to school achievementaccording to SDT using the Academic Motivation Scale (AMS) de-signed by Vallerand et al (1992) This is the most widely used scaleof school motivation from the SDT framework We also designedthree controlled longitudinal studies that used the AMS to measurefive different types of academic motivation and to examinetheir relation to school achievement over time To ensure that ourfindings were robust and generalizable we varied the school context

1 Integrated regulation the most autonomous form of extrinsic motivation occurswhen the value of the instrumental behaviour has come to be in harmony with othervarious aspects of a personrsquos values and identity to form a coherent sense of self Astudent who does not like math but understands the importance and benefits oftaking a statistics class and does so because he wants to eventually become a psy-chologist displays integrated regulation It must be noted that integrated regulationrequires much effort self-awareness and reflection (Vansteenkiste Niemiec ampSoenens 2010) Moreover this type of motivation has not typically been includedin measures of academic motivation because some early studies showed that stu-dents could not differentiate it from identified regulation on self-report scales (RobertJ Vallerand et al 1992)

343G Taylor et alContemporary Educational Psychology 39 (2014) 342ndash358

and cultural context across the three studies Studies 2 and 4 in-cluded high school students while Study 3 included college studentsStudies 2 and 3 included Canadian students while Study 4 in-cluded students from Sweden In each study we controlled forbaseline levels of achievement SDT also predicts that different mo-tivation types relate to each other but past research examining theserelations longitudinally is scarce and does not provide a clear picture(Guay et al 2010 Otis et al 2005) We thus tested the relationsamong different types of motivation in an exploratory way Finallywe examined the reciprocal relations between academic motiva-tion and achievement since some studies that have tested cross-lagged models have shown that prior academic achievement predictsubsequent academic motivation (Garon-Carrier et al 2014Goldberg amp Cornell 1998)

Study 2 examined the relations of motivation types to schoolachievement over a one-year period in a large Canadian high schoolsample Study 3 examined the relations of different types of mo-tivation to achievement after the transition from high school tocollege which is compulsory after high school in the Canadian prov-ince of Queacutebec for those wanting to pursue university studies Study4 examined the relations of motivation types to school achieve-ment in a sample of high school students attending their final yearof the science stream in Sweden

2 Study 1

In this meta-analysis we compiled results of cross-sectional andprospective studies employing the Academic Motivation Scale (AMSVallerand et al 1992) An investigation of each motivation subtypewas conducted to assess its relations to academic achievement Ourhypotheses were as follows First in line with SDT we predicted thatintrinsic and identified regulation would each have a positive as-sociation with school achievement and thus a positive effect sizeSecond we expected a negative effect size for the relation ofintrojected regulation external regulation and amotivation to schoolachievement Lastly we predicted that the relations of the auton-omous motivation types (intrinsic motivation and identifiedregulation) to achievement would be stronger than those of the con-trolling motivation types (introjected and external regulations)Furthermore amotivation was expected to have the strongest neg-ative association with school achievement

Since past studies differ in design (cross-sectional vs longitu-dinal) and have been conducted in different school contexts we

explored whether the study design and the school context (ie el-ementary high school or collegeuniversity) of the studies obtainedin the meta-analysis would moderate the relations of the differentmotivation types on achievement

21 Method

211 Selection of studiesAn electronic search was conducted using Social Sciences Cita-

tion Index (1956ndash2013) Science Citation Index Expanded (1900ndash2013) and Arts amp Humanities Citation Index (1975ndash2013) to identifyall articles that have cited Ryan and Connell (1989) andor Vallerandet al (1992) The review by Deci and Ryan (2000) and meta-analyses conducted by Deci et al (1999) and Chatzisarantis et al(2003) were also used to locate any articles that might not have beenincluded in the database The resulting list of articles was then re-viewed to identify the studies that had tested self-determinationtheory in a schooleducation context

From this pool of articles studies were rejected on the basis ofthe following criteria (1) studies that did not include correlationsor multiple regression coefficients between motivation and achieve-ment (2) studies that did not use the AMS to assess academicmotivation and (3) studies that did not include a measure of aca-demic achievement

Based on these three criteria 18 studies that assessed the rela-tion of motivation types according to SDT to school achievementwere obtained Of these 6 studies had a cross-sectional designwhereas 12 studies reported prospective data However of the 12prospective studies only 3 studies controlled for baseline achieve-ment (Baker 2003 Black amp Deci 2000 Burton et al 2006) In otherwords few studies measured the relation of academic motivationto changes in achievement over time All studies were questionnaire-based field studies Table 1 shows the list of studies included in themeta-analysis

212 Dependent measure and computation of effect sizesThe following dependent variables reflecting academic achieve-

ment in the set of selected studies were included GPA (actual orself-reported) performance as rated by a teacher grades obtaineddirectly from the school administration and national achievementtest scores

In this meta-analysis correlations and regression coefficients foreach study were converted to Cohenrsquos d Composite mean weighted

Table 1List of all studies included in the meta-analysis

Authors and date N Design School context Dependent measure of achievement

Assor et al (2009) Study 2 141 Prospective High school Teacher-rated general achievementBaker (2003) 91 Prospective (c) University GPABlack and Deci (2000) 137 Prospective (c) University Final chemistry gradeBurton et al (2006) Study 1 241 Prospective Elementary school Report card gradesBurton et al (2006) Study 2b 53 Prospective (c) University Final exam gradeCokley et al (2001) 263 Cross-sectional University GPADrsquoAilly (2003) 806 Prospective Elementary school Final test scoresFortier et al (1995) 263 Prospective High school National test scoresGrolnick et al (1991) 456 Prospective Elementary school Math and reading gradesHardre and Reeve (2003) 483 Cross-sectional High school Self-reported GPANoels et al (1999) 78 Prospective University Final grade in language coursePetersen et al (2009) 194 Prospective University Final average scoreRatelle et al (2007) Study 2 942 Prospective High school Report card gradesRatelle et al (2007) Study 3 410 Prospective College Final semester gradesSoenens and Vansteenkiste (2005) Study 1 328 Cross-sectional High school Self-reported GPASoenens and Vansteenkiste (2005) Study 2 285 Cross-sectional High school Self-reported GPAVallerand et al (1993) 217 Cross-sectional College Self-reported gradesWalls and Little (2005) 786 Cross-sectional High school Teacher-assigned grades

Note The notation (c) = prospective studies that controlled for academic achievement GPA = grade point average Here college refers to the schooling system from the prov-ince of Queacutebec Canada In this system high school ends in grade 11 and students move on to college for grades 12 and 13 to complete their pre-university education whichis mandatory to move on to university College students can also choose to enter a 3-year technical program which allows one to work directly after graduation

344 G Taylor et alContemporary Educational Psychology 39 (2014) 342ndash358

effect-size estimates (d+) were obtained from the average of the in-dividual effects (d) weighted by the reciprocal of their varianceAlthough the correlation coefficient r is often recommended as aneffect size measure (egRosenthal amp DiMatteo 2001) meta-analysis experts such as Field and Gillett (2010) indicate that Cohenrsquosd may be privileged in situations when group sizes are very dis-crepant because unlike r it accounts for base rates and is less biasedthereby giving superior weight to the more reliable effect-size es-timates (Hedges amp Olkin 1985) All effect-size computations andsummary analyses were done according to procedures suggestedby Hedges and Olkin (1985) using a meta-analytic software calledDSTAT (Johnson 1993) The calculations of composite d values pro-vided both a significance test and a 95 confidence interval (CI) Tointerpret effect sizes the benchmarks of d = 10 30 and 50 havebeen proposed by Cohen (1992) as representing small medium andlarge effects respectively The homogeneity of each set of effectssizes was tested by the within-class goodness-of-fit statistic (Qw)which has an approximate chi-square distribution with k ndash 1 degreesof freedom where k equals the number of effect sizes (Johnson1993) A significant Qw value indicates systemic variation withina set of effect sizes which suggests the presence of moderatorvariables

213 Moderator analysesStudy design and school context were examined as modera-

tors when the Qw value was found to be significant (ie when theset of effect sizes was heterogeneous) These moderators weredivided into categories First we focused on studies that had a cross-sectional design and compared them with studies that had aprospective design but that had controlled for baseline achieve-ment and with studies that had a prospective design and had notcontrolled for baseline achievement Second we categorized studiesaccording to the school context in which they were conducted Thestudies were either conducted in elementary school in high schoolor in collegeuniversity The moderating effects of these categori-cal variables were calculated by classifying each study accordingto these moderator categories and by testing for homogeneity ofeffect sizes across categories using a between-class goodness-of-fit statistic (QB) A significant QB value indicates systemic variationacross moderator categories It is comparable to a significantmain effect in ANOVA (Sitzmann Kraiger Steward amp Wisher2006)

22 Results

221 Effect sizes for individual types of motivationTable 2 shows the results of the meta-analysis Overall a signif-

icant effect size emerged for intrinsic motivation d+ = 27 (CI = 2332) as well as for identified regulation d+ = 35 (CI = 31 39)showing that both intrinsic motivation and identified regulation weremoderately positively related to achievement in school Howeverthese two sets of effect sizes were not homogeneous Qw(9) = 3302

p lt 001 and Qw(12) = 6749 p lt 0001 respectively Significant effectsizes were found for introjected regulation d+ = minus12 (CI = minus16 minus08)and external regulation d+ = minus22 (CI = minus26 minus17) This indicated thatintrojected regulation had a weak significant negative relation toschool achievement while external regulation had a moderate neg-ative relation to school achievement Once again both of these setsof effect sizes were not homogeneous (Qw(9) = 5752 p lt 001 forintrojected regulation and Qw(10) = 9305 p lt 001 for external reg-ulation) Finally a large effect size was obtained for amotivationd+ = minus61 (CI = minus67 minus55) suggesting that amotivation had astrong significant negative relation to school achievementThis set of effect sizes that was also not homogeneous Qw(6) = 7146p lt 001

222 Moderator analysesNo significant differences were found between different types

of study designs for intrinsic motivation identified regulation andexternal regulation However a difference emerged for introjectedregulation and amotivation More specifically introjected regula-tion had a modest negative relation to achievement in cross-sectional studies d = minus26 (CI = minus33 minus19) but no relation innon-controlled longitudinal studies d = minus02 (CI = minus08 03) More-over non-controlled longitudinal studies had a significantly largernegative composite effect size for amotivation (d = minus71 (CI = minus78minus63)) than cross-sectional studies (d = minus37 (CI = minus49 minus24)) Finallynon-controlled prospective studies had a significantly larger andstronger negative composite effect size for amotivation and achieve-ment (d = minus71 (CI = minus78 minus63)) than the controlled prospective study(d = minus22 (CI = minus51 07))

A significant difference between age groups emerged for all in-dividual motivation types except for introjected regulation Firstintrinsic motivation had a significantly stronger positive relation toschool achievement for high school and college students than forelementary school pupils QB(2) = 653 p lt 05 Second identified reg-ulation had a larger positive effect size on school achievement forelementary school pupils and high school students than for olderstudents QB(2) = 580 p = 05 In contrast regarding identified reg-ulation the effect size was almost twice as large for collegeuniversity students d = 62 (CI = 51-72) than for high schoolstudents d = 36 (CI = 32-39) On the other hand age did not seemto moderate the relation of introjected regulation to achievementWhile external regulation had a moderate negative relation to schoolachievement for high school d = minus29 (CI = minus34 minus23) and collegeuniversity students only d = minus21 (CI = minus28 minus13) it was found thatamotivation had a larger negative effect on school achievement forhigh school students d = minus77 (CI = minus86 minus68) than for collegeuniversity students d = minus49 (CI = minus57 minus41) This suggests thatamotivation is potentially more harmful to younger studentsrsquo schoolachievement

23 Brief discussion

Overall this meta-analysis shows that intrinsic motivation andidentified regulation have a moderately strong positive relation withschool achievement Introjected and external regulation had aweaker but significant negative relation with school achieve-ment Finally we found that amotivation had a strong negativerelation to school achievement Moderator analyses also demon-strated that intrinsic motivation had a stronger relation toachievement for high school and college students but that identi-fied regulation presented a stronger relation to achievement forelementary school pupils However these results must be inter-preted with caution given the small number of studies as well assome methodological problems that were uncovered during thisreview

Table 2Meta-analysis results effect sizes of all individual types of motivation on schoolachievement presented as composite d corrected for sample size

k N ofstudies

d 95 CI

From To

Intrinsic motivation 10 4270 27 23 32Identified regulation 11 4705 35 31 39Introjected regulation 10 4411 minus12 minus16 minus08External regulation 11 4411 minus22 minus26 minus17Amotivation 7 2195 minus61 minus67 minus55

Note The N of studies represents the total number of participants of all the rele-vant studies in a specific analysis CI = confidence interval

3 Study 2

The meta-analysis presented above highlighted that a large pro-portion of past studies comparing the effects of different motivationtypes have been cross-sectional or have assessed motivation andachievement over time without taking baseline levels into accountFurthermore all of these controlled prospective studies were con-ducted solely with college or university students and two of themincluded relatively short-term follow-ups or did not report resultsfor all individual types of motivation as proposed by SDT Stated dif-ferently this review points to the need for more studies to undertakea careful empirical analysis of the relations of different forms of mo-tivation on academic achievement across the high school and earlycollege years

In order to address these issues Study 2 focused on high schoolstudents and examined self-reported academic achievement as anoutcome variable Cross-lagged structural equation modeling wasused to test which of the different motivation types was moststrongly related to changes in achievement one year later when con-trolling for earlier academic achievement as well as for the reciprocalrelation of prior achievement to subsequent motivation types

Our predictions for this study as well as for Studies 3 and 4 wereas follows First we hypothesized that prior intrinsic motivation andidentified regulation would be positively related to subsequent ac-ademic achievement whereas introjected and external regulationas well as amotivation would be negatively related to it In addi-tion based on the evidence provided by our meta-analysis and onthe only well-controlled prospective study that included a broadmeasure of school achievement (Baker 2003) we expected that thepositive relation of prior intrinsic motivation to academic achieve-ment would be stronger than the relations of other types ofmotivation to the same outcome In other words we expected thatintrinsic motivation would be the best positive predictor of schoolachievement Finally we predicted that this relation should be sig-nificantly positive even after controlling for baseline achievementas well as for the reciprocal relation between prior achievement andsubsequent academic motivation

31 Method

311 ParticipantsStudents attending a French-speaking high school in suburban

Montreal completed a questionnaire twice over a school year Theparticipants were in grades 7 to 11 and aged between 12 and 17years old From the initial sample of students who were invited toparticipate in the study (N = 524) a total of 319 students (159 boys160 girls) completed all measures of interest This represents a 608response rate The mean age for the sample was 1432 years Thevast majority of students were French Canadian (993) Accord-ing to the socioeconomic index used by the Quebec Ministry ofEducation (2013) this school was located in an upper-middleclass area

312 ProcedureThe students completed a survey on a voluntary basis with the

authorization of the schoolrsquos principal and teachers Parental consentwas obtained through letters distributed to the students at schoolIn the winter of each year three trained research assistants (oneper group) administered the questionnaire during class time andstayed present to answer studentsrsquo questions Students were toldthat the questionnaire concerned adolescentsrsquo attitudes towardschool and student relationships in an educational setting They werealso informed that their participation was voluntary that they wereallowed to skip items of the questionnaire and that theirresponses would remain anonymous and confidential

313 MeasuresThis study focused on baseline demographic variables (Time 1)

academic motivation (Time 1 and Time 2) and academic achieve-ment (Time 1 and Time 2) Other variables were included in thesurveys but were not the focus of the present study One article basedon this data set has been published focusing on the similarity inlife aspirations between teenagers and their parents (LekesJoussemet Koestner Taylor Hope amp Gingras 2011)

3131 Academic motivation To measure the different types of ac-ademic motivation items from the French version of the AcademicMotivation Scale developed by Vallerand Blais Briegravere and Pelletier(1989) were used The scale was composed of twenty items fromthe five subscales proposed by Self-Determination Theory (Deci ampRyan 2002) In order to create one intrinsic motivation scale weused items from each of the original three intrinsic motivation scales(to know towards accomplishment and to experience stimula-tion) This procedure has been followed in other studies (NtoumanisBarkoukis amp Thoslashgersen-Ntoumani 2009 Otis et al 2005)

The items provide possible answers to the question ldquoWhy do yougo to schoolrdquo An example for each subscale is described as followsintrinsic motivation (eg ldquoBecause I experience pleasure and sat-isfaction while learning new thingsrdquo) identified regulation (egldquoBecause I think that education will help me better prepare for thecareer I have chosenrdquo) introjected regulation (eg ldquoTo show myselfthat I am an intelligent personrdquo) external regulation (eg ldquoTo havea better salary laterrdquo) and amotivation (eg ldquoI cannot see why I goto school and frankly I could not care lessrdquo) Respondents rated theiragreement with each reason for going to school on a 7-point Likerttype scale ranging from 1 (totally disagree) to 7 (totally agree) witha higher score indicating a higher level of endorsement of the par-ticular regulatory style The reliability and predictive validity for thesescales has been consistently established in previous research (egRatelle et al 2007 Vallerand Fortier amp Guay 1997 Vallerand et al1989 1993) In this study internal consistency coefficients ob-tained were as follows intrinsic motivation (84 at Time 1 87at Time 2) identified regulation (72 at Time 1 77 at Time 2)introjected regulation (85 at Time 1 89 at Time 2) external reg-ulation (58 at Time 1 73 at Time 2) and amotivation (82 at Time1 89 at Time 2)

3132 Perceived academic achievement Participants were asked toreport their most recent general grade percent average () Thereis evidence showing that self-reported school grades are stronglycorrelated with actual school grades (Dornbusch Ritter LeidermanRoberts amp Fraleigh 1987 Hennan Dornbusch Herron amp Herting1997 Soenens amp Vansteenkiste 2005)

314 Statistical analysesAll structural equation modeling analyses were performed using

Amos 7 with the maximum likelihood estimation procedure Fol-lowing the guidelines of Marsh and colleagues outlined by (Guayet al 2010) and Retelsdorf Koumlller and Moumlller (2014) we used theldquofull-forwardrdquo SEM approach to allow for a rigorous test of recip-rocal effects In this type of model stability coefficients as well aswithin-time correlations and cross-lagged relations are esti-mated where each variable has paths leading to all other variablesat the other wave of assessment (see Fig 1) Moreover in line withMarsh and Hau (1996) we estimated correlated uniqueness ie cor-relations between the residuals of the same constructs measuredon two different occasions within the same person to control formethod effects In doing so we avoid positively biased stabilitycoefficients

We first performed a confirmatory factor analysis (CFA) to verifythe adequacy of the measurement model and the extent to whichour indicators satisfactorily related to their associated latent

variable (Model 1) To create the measurement model we used thefour individual items of each subscale of the Academic MotivationScale as indicators for their respective type of motivation (intrin-sic identified introjected external and amotivation) We used theaverage course grade as the single indicator for achievement If weobtained acceptable fit of the measurement model we then testedthe invariance of factor loadings to ensure that the meaning of theconstructs was the same across measurement times Finally wetested the structural model to evaluate its ability to explain stu-dentsrsquo academic motivation and achievement over time

To evaluate model fit we first used the χ2 test statistic An ac-ceptable model should have a nonsignificant χ2 value However giventhat this test is known to be overly sensitive to sample size and smalldeviations from multivariate normality (Morin Madore MorizotBoudrias amp Tremblay 2009) three additional criteria were used toevaluate model fit the comparative fit index (CFI) the TuckerndashLewis index (TLI) the root mean square error of approximation(RMSEA) CFI and TLI values of 90 or more was used to guide de-cisions regarding acceptable model fit (see Marsh Hau amp Grayson2005) For the RMSEA which is a summary statistic for the residu-als (ie the lower the number the better) we followed Klinersquos (2011)recommendation to use values of 06 or less as indicative ofgood fit

32 Results

321 Preliminary analyses3211 Missing data On average the proportion of missing dataacross all variables included in the model at both measurement timeswas 191 Missing data resulted from usual factors in longitudi-nal research with adolescents such as absence on the day ofassessment or refusal to complete the questionnaire Sample attri-tion from T1 to T2 was moderate (39) with dropout students joiningnon-participating classrooms in special education classes gradu-ating or changing schools after the first year of the studyComparisons for all main variables were performed to examinewhether students who completed both waves of assessment wereequivalent to those who provided data at T1 only A MANOVA wasperformed to test the main effect of participant group (1 wave vs2 waves) on the 21 indicators of latent constructs at T1 Using Wilksrsquolambda results revealed a significant difference between the twogroups (Λ = 85 F[21 308] = 251 p lt 001) Of the 21 indicators sixpresented a significant effect (29) Of these significant effects oneexplained 6 of the variance while others explained less than 18Specifically students who completed both assessments (M = 7751SD = 798) had a higher grade average than students who only par-ticipated at T1 (M = 7272 SD = 943)

Although very common in the school engagement literature thesedifferences can reduce statistical power andor bias the results Tocorrect for these potential problems we followed the guidelines byBuhi Goodson and Neilands (2008) and Schlomer Bauman and Card(2010) and used full information maximum likelihood (FIML) to es-timate missing observations FIML has been shown to perform wellwith missing datamdashwhether data are missing at random or not andwhen there is a moderate amount of missing data (Buhi et al 2008)This technique has also been shown to outperform older ad-hoc pro-cedures such as listwise deletion or mean substitution (Schlomeret al 2010) and several studies have indicated that it yields theleast biased and most efficient parameter estimates (Peugh amp Enders2004)

3212 Descriptive statistics Means standard deviations and inter-correlations between latent variables across both assessment timesare presented in Table 3

Overall students reported primarily identified regulation andexternal regulation reasons for going to school Students reportedmoderate levels of introjected regulation and intrinsic motivationas well as relatively low levels of amotivation Correlations seemedto follow the pattern expected based on our meta-analysis of thecurrent literature The five autocorrelations between T1 and T2 werestrong (rs gt 50) suggesting that constructs were stable over timeFinally the correlations among motivational subscales were exam-ined We found that motivation types mostly related to each otherin a continuum-like way with adjacent motivations (eg intrinsicmotivation and identified regulation) correlated more stronglythan distal ones (eg intrinsic motivation and external regula-tion) However some correlations did not fit this pattern(a) The correlations of intrinsic motivation with introjected regu-lation were higher (57 at T1 and 54 at T2) than with identifiedregulation (b) at both assessment waves the correlationsof identified regulation with introjected and external regulationwere very similar (c) the correlations between identified and ex-ternal regulations was higher than the correlation betweenintrojected and external regulations This issue is addressed in thediscussion

322 Measurement model and factor loadings invarianceThe goodness of fit statistics for all models are presented in

Table 4 First we tested the adequacy of the measurement modelat T1 and T2 (Model 1) This model yielded adequate fit indiceswhich provides good evidence for the factorial validity of scoresie the fact that indicators relate to their respective factor in theways proposed by the measurement model Factor loadings wereall acceptable across both measurement waves except for one

Table 3Study 2 descriptive statistics and intercorrelations matrix for all latent constructs

1 2 3 4 5 6 7 8 9 10 M SD

1 T1 ACH 7488 8722 T2 ACH 80 7489 9333 T1 IM 33 35 285 974 T1 ID 15 07 36 439 645 T1 INTROJ 07 06 57 44 331 1086 T1 EXT minus00 minus06 11 42 28 437 627 T1 AMOT minus36 minus31 minus46 minus33 minus24 minus05 181 908 T2 IM 37 41 71 28 44 02 minus37 271 809 T2 ID 13 18 37 54 29 20 minus23 50 415 6310 T2 INTROJ 02 12 44 39 68 26 17 54 48 303 9511 T2 EXT minus07 minus10 minus11 26 10 55 17 minus01 41 32 421 6212 T2 AMOT minus32 minus38 minus42 minus22 minus20 03 65 minus53 minus41 minus25 04 176 77

Note ACH = academic achievement IM = intrinsic motivation ID = identified regulation INTROJ = introjected regulation EXT = external regulation AMOT = amotivation p lt 05

p lt 01

indicator of external regulation whose factor loading was relative-ly low (32 at T1 and 38 at T2) After establishing the adequacy ofthe measurement model we tested the invariance of factor load-ings across measurement times (Model 2) The fit indices for Model2 were adequate and the chi-square difference test showed no sig-nificant difference with Model 1 indicating that the meaning of theconstructs did not change over time Therefore the factor load-ings were fixed to equality for subsequent analyses

323 Stability and reciprocal effects modelAccording to the guidelines presented above the reciprocal effects

model (Model 3) yielded a good fit to the data Moreover as de-picted in Fig 1 the results show stability across time of bothacademic achievement and academic motivation types since alldirect path coefficients within one construct between the two waveswere strong and positive ranging from 59 to 69 The results alsoshowed the following positive significant paths between differentmotivation types the path connecting T1 intrinsic motivation to T2identified regulation (β = 27) and the path connecting T1 amotivationto T2 external regulation (β = 22) No other significant pathsbetween different motivation types across measurement wavesemerged

Regarding the relationship between motivation types and aca-demic achievement results showed that T1 intrinsic motivationsignificantly predicted increases in T2 achievement and that no otherT1 motivation type was significantly related to T2 achievementFinally T1 achievement significantly predicted an increase in T2 in-trinsic motivation Overall the model explained a considerableproportion of variance in all six outcomes at T2 including achieve-ment (R2 = 56) intrinsic motivation (R2 = 51) identified regulation(R2 = 33) introjected regulation (R2 = 45) external regulation(R2 = 40) and amotivation (R2 = 41)

33 Brief discussion

Study 2 showed that when different motivation types were ana-lyzed concurrently within the same model intrinsic motivation wasthe only motivation type to be significantly positively related to anincrease in achievement over time This result matches the find-ings in our meta-analysis showing that the relation of intrinsicmotivation to school achievement had a substantial effect size More-over it supports the predictions of SDT as well as Bakerrsquos (2003)controlled longitudinal study showing that intrinsic motivation pre-dicted an increase in the academic achievement of universitystudents over time controlling for baseline achievement This studyextends these findings to a younger population indicating that in-trinsic motivation plays an important role for the future academicsuccess of high school students Another interesting finding was thatacademic achievement predicted later intrinsic motivation In otherwords our results provide support for a reciprocal relation betweenintrinsic motivation and academic achievement over time

4 Study 3

We conducted Study 3 to test the whether the findings ob-tained in Study 2 could extend to a population of college studentsenrolled in a science program One advantage of this study was thatobjective grades were obtained as a measure of academic achieve-ment Another advantage was the focus on students in science anarea that is known for higher base rates of dropout than other fieldsof study (Lavigne Vallerand amp Miquelon 2007) In fact almost 30of Canadian and American college students registered in science pro-grams leave the field before they have graduated (Duchesne RatelleLarose amp Guay 2007) One way to address why some students ex-perience difficulties in science is to clarify how the different typesof motivation contribute to their achievement over time which isassociated with persistence Few studies have attempted to inves-tigate the retention problem of science students by examiningdifferent motivations during a school transition and using a con-trolled prospective design Following the results obtained in Study2 we predicted that intrinsic motivation would be the most stronglypositively related to academic science achievement among all themotivation types

41 Method

411 ParticipantsParticipants were students who graduated from high school

(grade 11) in June 2003 and entered the science program in one ofthe four public English-speaking colleges in Montreal Canada in thefall of that year In the Queacutebec education system students com-plete high school in grade 11 and then make the transition to collegeA general college diploma is usually obtained after two years andleads to university Students can also enrol in a three-year techni-cal program which leads directly to the job market In Quebec acollege diploma is required for students to enter university2 Fromthe initial sample of 1135 students (510 males 625 females) whowere eligible to take part in the study a total of 638 students (296men 342 women) aged between 17 and 30 years (mean age = 17years 11 months) completed Time 1 and Time 2 measures In thisfinal sample 225 of the students spoke French at home 465spoke English at home and 31 spoke another language at home(eg Mandarin Arabic Italian Greek or Vietnamese)

412 ProcedureData for this study were collected as part of a larger research

project on academic success and perseverance in science led by ateam of researchers some of which were also instructors at the

2 In the province of Quebec high school ends in grade 11 and students have tomove on to college for grades 12 and 13 to complete their pre-university educa-tion or to enter a 3-year technical program which allows one to work after graduationfrom college College students registered in a pre-university program must com-plete this before attending university

Table 4Study 2 goodness of fit statistics for confirmatory factor analyses and structural equation modeling analyses

Model χ2 Df CFI TLI RMSEA 90 CI RMSEA Δ χ2 CM

CFA modelsModel 1 (measurement model) 133573 675 91 90 042 [038 045]Model 2 (factor loadings invariance) 135164 690 91 90 041 [038 045] 1591a M1

SEM modelModel 3 (full model disturbances and uniquenesses correlated) 145121 750 91 89 041 [038 044]

Note CFI = comparative fit index TLI = TuckerndashLewis Index RMSEA = root mean square error of approximation CI = confidence interval CM = comparison modela Non significant

colleges involved Students participated on a voluntary basis withthe authorization of their instructors and the college administra-tions Studentsrsquo grades in all science and mathematics courses takenin grade 10 and 11 were obtained from the Ministegravere de lrsquoEacuteducation

du loisir et des sports du Queacutebec (MELSmdashMinistry of Education ofQuebec) records with participantsrsquo consent (Time 1) These in-cluded Physical Science courses Mathematics courses andChemistry Similarly the raw data for studentsrsquo grades in all

Fig 1 Cross-lagged panel modelling of academic achievement and types of motivation in Study 2 Note Correlations between each itemrsquos error term at each assessmentbetween factors at Time 1 and between disturbances were estimated Equality constraints were imposed on same itemsrsquo loading across time Only significant paths arepresented but all were allowed to covary

Mathematics and Science courses taken during the first semesterof college studies were obtained from MELS records during thewinter of 2004 (Time 2) All college Mathematics and Science courseswere similar for students within the same institution

Students filled out two questionnaires assessing academic mo-tivation in class or on the web depending on the college they wereattending The first questionnaire was administered in class duringthe beginning of studentsrsquo first semester in the fall of 2003 (Time1) The second questionnaire was administered during the middleof the second semester of the Science program in winter 2004(Time 2)

413 Measures4131 Academic motivation We assessed academic motivation withthe Academic Motivation Scale (AMS Vallerand et al 1992 1993)described in Study 2 The scale was composed of 10 items from thefive subscales proposed by Self-Determination Theory (Deci amp Ryan2002) Students rated their agreement with each reason for goingto college on a 5-point Likert type scale ranging from 1 (totally dis-agree) to 5 (totally agree) with a higher score indicating a higherlevel of endorsement of the particular regulatory style The inter-nal reliability for each of the subscales was acceptable for two-item scales (Eisinga Grotenhuis amp Pelzer 2013) with Cronbachalphas as follows intrinsic motivation (73 at Time 1 and Time 2)identified regulation (68 at Time 1 100 at Time 2) introjected reg-ulation (54 at Time 1 67 at Time 2) external regulation (61 at Time1 68 at Time 2) and amotivation (75 at Time 1 and Time 2)

4132 Academic achievement Official math and science grades fromhigh school (Time 1 achievement) and from the first completed se-mester of college (Time 2 achievement) were obtained directly fromthe Ministry of Education of Quebec Grades from each subject werethen compiled into a percent average to create one variable

414 Statistical analysesIn this study we used path analysis with Amos 7 to test the fit

between the data and the hypothesized reciprocal effects model usingthe maximum likelihood estimation procedure This analysis waschosen given the recommendation by Kline (2011) who suggeststhe use of a path analysis with observed variables in studies wherethere are only two indicators per factor To evaluate model fit wefollowed the same guidelines that were outlined in Study 2

42 Results

421 Missing dataThe descriptive analyses indicated that 372 of the initial sample

at T1 completed the second assessment wave This attrition re-

sulted from absence on the day of the T2 assessment as well thestudents changing programs or colleges Comparisons for all mainvariables were performed to examine whether students who com-pleted both waves of assessment were equivalent to those whoprovided data at T1 only We performed a MANOVA to test the maineffect of participant group (1 wave vs 2 waves) on the observed vari-ables at T1 Using Wilksrsquo lambda results revealed a significantdifference between the two groups (Λ = 80 F[6 1057] = 4437p lt 001) Specifically students who completed both assessmentshad a higher high school science achievement than students whoonly participated at T1 (F[1 1063] = 24245 p lt 001 R2 = 18) In ad-dition the students who completed both assessments had higheridentified regulation (F[1 1063] = 2420 p lt 001 R2 = 02 ) loweramotivation (F[1 1063] = 1866 p lt 001 R2 = 02) but higher exter-nal regulation (F[1 1063] = 1929 p lt 001 R2 = 02) As mentionedin Study 2 to account for these differences we followed the guide-lines by Buhi et al (2008) and Schlomer et al (2010) and used fullinformation maximum likelihood (FIML) to estimate missingobservations

422 Preliminary analyses4221 Descriptive statistics Means standard deviations and inter-correlations between observed variables across both assessmenttimes are presented in Table 5

Overall the most popular reasons students gave for studyingscience were identified and external regulation They reported mod-erate levels of intrinsic motivation and introjected regulation andrelatively low levels of amotivation Similar to Study 1 correla-tions seemed to follow the pattern expected based on our meta-analysis of the current literature Moreover correlations show thatT1 academic achievement was strongly positively associated withlater T2 achievement Longitudinal stability coefficients between T1and T2 were relatively strong ranging from r = 43 for amotivationto r = 68 for intrinsic motivation Finally the correlations among mo-tivational subscales were examined to see whether they reflectedthe simplex pattern in line with the autonomy continuum postu-lated by SDT We found that motivation types at both assessmentwaves were related to each other in the simplex pattern predictedby SDT with adjacent motivations (eg intrinsic motivation and iden-tified regulation) correlated more strongly than distal ones (egintrinsic motivation and external regulation)

423 Stability and reciprocal effects path modelThe results of the reciprocal effects path model are presented

in Fig 2Since all the parameters in this model were estimated (Model

4) thus leaving no degrees of freedom the model fit could not betested Several paths were significant Results show stability across

Table 5Study 3 descriptive statistics and intercorrelations matrix for all variables

1 2 3 4 5 6 7 8 9 10 11 M SD

1 T1 ACH 7292 20952 T2 ACH 39 6862 14183 T1 IM 02 24 373 574 T1 ID 07 15 33 431 535 T1 INTROJ minus02 minus11 26 21 330 676 T1 EXT 14 02 08 48 32 406 627 T1 AMOT minus06 minus22 minus34 minus51 minus06 minus28 156 628 T2 IM 02 25 68 21 18 minus07 minus28 366 809 T2 ID minus05 18 22 47 20 21 minus40 37 408 4610 T2 INTROJ minus07 minus09 23 11 67 31 minus08 26 17 318 5911 T2 EXT minus09 minus16 02 35 34 64 minus08 minus01 26 47 337 6012 T2 AMOT 01 minus23 minus28 minus33 05 minus12 43 minus40 minus58 minus03 minus05 194 52

p lt 01

time of both academic achievement and academic motivation typessince most direct path coefficients within one construct betweenthe two waves were moderately strong and positive ranging from30 to 45 The path coefficient connecting amotivation at bothassessment waves was significant but lower (β = 21) The resultsalso showed the following positive significant paths between dif-ferent motivation types the path connecting T1 intrinsic motivationto T2 amotivation (β = minus11) the path connecting T1 identified reg-ulation to T2 introjected regulation (β = minus12) the path connectingT1 introjected regulation to T2 external regulation (β = minus11) and toT2 amotivation (β = 10) the path connecting T1 external regula-tion to T2 intrinsic motivation (β = minus12) and to T2 introjectedregulation (β = 13) and the path connecting T1 amotivation to T2identified regulation (β = minus18)

Regarding the relationship between motivation types and aca-demic achievement the model showed that T1 intrinsic motivationsignificantly predicted increases in T2 academic achievement (β = 17)At the same time T1 introjected regulation and T1 amotivation bothsignificantly predicted decreases in T2 academic achievement(β = minus11 and β = minus09 respectively) Finally the model showed thatT1 academic achievement predicted decreases in T2 external reg-ulation (β = minus10) In order to test whether the path going from T1intrinsic motivation to T2 achievement was stronger than the pathgoing from T1 amotivation to T2 achievement we tested anothermodel in which these parameters were constrained to equalityWe then compared the fit of this model to the final unconstrainedone (Model 4) The chi-square difference test was significantΔχ2(1) = 2595 p lt 001 However the fit of the constrained model

was significantly worse (χ2(20) = 2595 p = 15 CFI = 98 TLI = 93RMSEA = 029 90 CI [020 037]) meaning that a model where therelation of T1 intrinsic motivation to T2 achievement is almost twiceas strong as the one between T1 amotivation and T2 achievementfit the data in a significantly better way Overall the final model(Model 4) explained a significant proportion of variance in all sixoutcomes at T2 including achievement (R2 = 15) intrinsic motiva-tion (R2 = 25) identified regulation (R2 = 15) introjected regulation(R2 = 24) external regulation (R2 = 24) and amotivation (R2 = 11)

43 Brief discussion

Overall our results replicate the findings from Study 2 showingthat studentsrsquo intrinsic motivation positively predicts their aca-demic achievement one year later above and beyond the effect ofbaseline academic achievement extending them to a college sciencestudent population Importantly the grades were not self-reportedas in Study 2 instead they were obtained directly from the collegeadministration Moreover results of this study also showed thatintrojected regulation and amotivation were significantly negative-ly associated with science achievement over time The negativerelation for amotivation is not surprising given the results of ourmeta-analysis which indicated that amotivation was the moststrongly related with school achievement However it is interest-ing to note that the positive relation of intrinsic motivation withschool achievement was twice as strong as the negative associa-tion of amotivation with the same outcome

The negative relation of introjected regulation with achieve-ment suggests that college students who choose to study sciencefor internally coercive reasons such as wanting to prove to othersthat they are intelligent may experience more negative outcomesthan high school students who have the same motivation given thatthey operate in an environment where there is more freedom ofchoice and where they have to rely on their own motivational devices(Ratelle et al 2007) This result supports the cross-sectional find-ings of Assor et al (2009) who showed that students who regulatedin an introjected way were more likely to experience lower achieve-ment than those who were more autonomously motivated Anotherinteresting result was that studentsrsquo science achievement in highschool negatively predicted their external regulation at T2 showinganother reciprocal effect between achievement and motivation

A potential problem with the present study was that the measureof academic motivation contained only two items for each type ofregulation Moreover the internal reliability coefficients were lowerwhich is typical for subscales that contain only two items (Eisingaet al 2013) It would have been better to include more items toassess motivation in this study as well as to employ the same 7-pointLikert scale as in the other studies Nevertheless the descriptive andpredictive results closely matched those obtained in Study 2

5 Study 4

Study 4 aimed to replicate the relation of different academic mo-tivation types to academic motivation found in Studies 2 and 3 andto extend these results cross-culturally In this study we exploredthe relation between these variables in a sample of Swedish highschool students who had chosen to study natural sciences BecauseSDT postulates that autonomy is a basic psychological need that isrelevant to all humans it becomes important to examine the rela-tion of different types of motivation to academic achievement acrosscultures While there is a growing literature on cross-cultural ap-plications of SDT in education (Assor et al 2009 Chirkov amp Ryan2001 Levesque Zuehlke Stanek amp Ryan 2004) to our knowl-edge no studies have examined the relation of the different typesof motivation to academic achievement separately using a longi-tudinal design and controlling for prior achievement

Fig 2 Cross-lagged panel modelling of academic achievement and types of moti-vation in Study 3 Notes Correlations between variables at Time 1 and betweendisturbances were estimated Only significant paths are presented but all were allowedto covary

The relevance of attempting to replicate these relations in Swedenbecomes clearer once we consider how it differs from Canada alongHofstedersquos (1983) dimensions of national cultures which are re-flected in the education system Although both countries scoresimilarly on individualism and power distance they differ signifi-cantly in their emphasis on working toward goals because ofcompetition versus cooperation or interest Hofstede (1983) distin-guishes a cultural dimension (called masculinityfemininity) whichfocuses on what motivates peoplemdashwanting to be the best (mascu-line) versus liking what you do (feminine) Compared to CanadaSweden scores extremely low on the masculinity dimension meaningthat Swedish culture is mostly based on values of quality of life andcooperation This is reflected in the Swedish education system wherestudents do not receive official grades until they reach grade 6 andwhere teachers tend to encourage cooperation and create equal op-portunities by focusing on group projects and refraining from usingexternal rewards and competition between students (Ministry ofEducation and Science 2001) This key difference between the cul-tural and educational systems of Sweden and Canada (wherecompetition rewards and grades are much more present) makes itinteresting to examine the relations between academic motivationand achievement in order to establish whether autonomous typesof motivation have a similar role in an environment that seems tobe supportive of autonomy (Sweden) compared with an environ-ment that focuses mostly on competence (Canada) Given theuniversality perspective of SDT we hypothesized that results of theprevious studies would replicate in a different cultural context iethat intrinsic motivation would be more strongly related to higherlevels of academic achievement than the other motivation types

51 Method

511 ParticipantsStudents attending a science program in metropolitan high

schools in Sweden completed a questionnaire in September and Aprilof their final school year In Sweden high school begins in the stu-dentsrsquo tenth year when they are 15 to 16 years old and lasts forthree years Students select a high school orientation in the springof their ninth school year for example social science business orscience They must complete high school in order to move on to uni-versity The sample at the first measurement time consisted of 440participants (226 males 214 females) Out of this first sample 288participants completed the follow-up questionnaire This follow-up sample was composed of 143 males and 145 females Officialgrade transcripts were received regarding 247 of the students Theage of the participants ranged from 18 to 19 years old

512 ProcedureThe students completed a questionnaire assessing academic

motivation once a year for two years on a voluntary basis with theauthorization of the schoolrsquos principal and teachers (Time 1 and Time2) The questionnaire was in Swedish and was distributed duringclass time by one researcher who informed them that the ques-tionnaire concerned adolescentsrsquo attitudes toward school in aneducational setting The students were also informed that their par-ticipation was voluntary and that they were allowed to skip itemsof the questionnaire The researcher was present to answer any ques-tions that the students may have had during the completion of thequestionnaire The questionnaire took approximately twenty minutesto complete Three months after the students had completed thesecond questionnaire official grades in science courses were ob-tained directly from the school administration (Time 2) Moreoverstudentsrsquo science grades for their second year of high school wereobtained from the school administration and represented the base-line grade assessment (Time 1)

513 Measures5131 Academic motivation As in Studies 2 and 3 we assessed ac-ademic motivation by adapting items from the English version ofthe Academic Motivation Scale (AMS Vallerand et al 1992 1993)described above The scale was composed of 20 items from the fivesubscales proposed by Self-Determination Theory (Deci amp Ryan2002) Students rated their agreement with each reason for goingto high school on a 4-point Likert type scale ranging from 1 (totallydisagree) to 4 (totally agree) In this study the Cronbach alphasobtained were as follows intrinsic motivation (91 at Time 1 90at Time 2) identified regulation (80 at Time 1 and Time 2)introjected regulation (75 at Time 1 74 at Time 2) external reg-ulation (75 at Time 1 78 at Time 2) and amotivation (91 at Time1 90 at Time 2) All scales were translated from English to SwedishThe translation was independently carried out by two Swedish-speakers The Swedish items were then back translated into Englishand discrepancies were arbitrated by two consultants English-speaking college Science teachers from Canada and solutions werereached by consensus

5132 Academic achievement Grade transcripts were received fromeach high schoolrsquos administrative office The Swedish grading systemuses letter grades for each course These letter grades were con-verted to a numerical scale fail = 1 pass = 2 pass with distinction = 3and pass with great distinction = 4 Science achievement at Time1 was calculated as the mean of grades in the two Mathematicscourses that students take in their second year in high school anda course in basic Physics that they take during their first two yearsin high school The students received their grades in these threecourses during their second year which would thus represent scienceachievement at Time 1 Achievement at Time 2 was calculated asthe mean of three compulsory courses an advanced course in math-ematics an advanced course in physics and a course in chemistryof their final year of high school

514 Statistical analysesThe same analytical approach outlined in Study 2 was used in

this study

52 Results

521 Preliminary analyses5211 Missing data On average the proportion of missing dataacross all variables included in the model at both measurement timeswas 27 Missing data resulted from usual factors in longitudinalresearch with adolescents such as absence on the day of assess-ment or refusal to complete the questionnaire Sample attrition fromT1 to T2 was 34 with these students joining non-participating class-rooms in special education classes graduating or changing schoolsafter the first year of the study Comparisons for all main variableswere performed to examine whether students who completed bothwaves of assessment were equivalent to those who provided dataat T1 only A MANOVA was performed to test the main effect of par-ticipant group (1 wave vs 2 waves) on the 23 indicators of latentconstructs at T1 Using Wilksrsquo lambda results showed no signifi-cant difference between the two groups (Λ = 90 F[23 217] = 108p = 37) Nevertheless to avoid decreases of power that typically resultfrom missing data we followed the guidelines by Buhi et al (2008)and Schlomer et al (2010) and used full information maximum like-lihood (FIML) to estimate missing observations

Overall the most popular reasons students reported for study-ing science were identified regulation and intrinsic motivation

Students reported moderate levels of external regulation andintrojected regulation and relatively low levels of amotivation Onceagain the correlations seemed to follow the pattern expectedbased on our meta-analysis of the current literature Finally the cor-relations among motivational subscales were examined to testwhether they reflected the pattern in line with the autonomy con-tinuum postulated by SDT We found that motivation types mostlyrelated to each other in a continuum-like way with adjacent mo-tivations (eg intrinsic motivation and identified regulation)correlated more strongly than distal ones (eg intrinsic motivationand external regulation) However some correlations did not rep-licate this simplex pattern (a) at both assessment waves thecorrelations of identified regulation with introjected (T1 r = 29 andT2 r = 32) were lower than with external regulation (T1 r = 47 andT2 r = 37) and (b) the correlations between identified and externalregulations was higher than the correlation between introjected andexternal regulations This issue will be addressed in the discussion

Table 7First we tested the adequacy of the measurement model at T1

and T2 (Model 5) According to the guidelines described in Study1 this model yielded adequate fit indices which provides good ev-idence for the factorial validity of scores ie the fact that indicatorsrelate to their respective factor in the ways proposed by the mea-surement model Factor loadings were all acceptable across bothmeasurement waves After establishing the adequacy of the mea-surement model we tested the invariance of factor loadings acrossmeasurement times (Model 6) The fit indices for Model 6 were ad-equate and the chi-square difference test showed no significantdifference with Model 5 indicating that the meaning of the con-structs did not change over time Therefore the factor loadings werefixed to equality for subsequent analyses

523 Stability and reciprocal effects modelAccording to the guidelines presented above the reciprocal

effects model (Model 7) yielded a good fit to the data Moreoveras depicted in Fig 3 the results show stability across timeof both academic achievement and academic motivation types sinceall direct path coefficients within one construct between thetwo waves were strong and positive ranging from 43 forachievement to 86 for intrinsic motivation The results also showeda positive significant path between T1 intrinsic motivationand T2 amotivation (β = minus18) No other significant pathsbetween different motivation types across measurement wavesemerged

Regarding the relationship between motivation types andacademic achievement results showed that T1 intrinsic motiva-tion significantly positively predicted T2 science achievementMoreover T1 external regulation was positively associated withT2 science achievement We also found that T1 amotivation wasnegatively related to T2 science achievement However absolutevalues of the coefficients show that this relation was not as strongas the relation of T1 intrinsic motivation to T2 science achieve-ment (β = 41 for intrinsic motivation and β = minus22 for amotivation)Finally T1 achievement negatively predicted T2 introjectedregulation

In order to test whether the path going from T1 intrinsic moti-vation to T2 achievement was stronger than the path going fromT1 amotivation to T2 achievement we tested another model in whichthese parameters were constrained to equality We then com-pared the fit of this model to the final unconstrained one (Model4) The chi-square difference test was significant Δχ2(1) = 3048p lt 001 However the fit of the constrained model was slightly lower(see Table 7) meaning that a model where the relation of T1 in-trinsic motivation to T2 achievement is almost twice as strong asthe one between T1 amotivation and T2 achievement fits the datain a significantly better way Overall the model explained a large

1 2 3 4 5 6 7 8 9 10 11 M SD

1 T1 ACH 309 552 T2 ACH 40 305 693 T1 IM 04 45 278 824 T1 ID 05 26 63 291 705 T1 INTROJ minus00 03 22 29 221 716 T1 EXT 02 10 19 47 42 251 717 T1 AMOT minus10 minus46 minus65 minus56 minus07 minus15 155 738 T2 IM 04 43 83 52 23 14 minus53 284 809 T2 ID 03 23 54 72 26 37 minus44 67 294 6610 T2 INTROJ minus11 minus03 11 18 66 27 minus02 26 32 230 6611 T2 EXT minus02 09 00 29 31 67 minus03 12 37 47 261 7112 T2 AMOT minus13 minus47 minus53 minus37 minus03 minus03 70 minus60 minus51 minus03 minus03 158 72

p lt 01

SEM modelsModel 7 (full model disturbances and uniquenesses correlated) 169656 918 92 91 040 [037 043]Model 8 (constrained model) 1727034 919 92 91 041 [038 044] 3048b M7

Note CFI = comparative fit index TLI = TuckerndashLewis Index RMSEA = root mean square error of approximation CI = confidence interval CM = comparison modela Non significantb Significant

proportion of variance in all six outcomes at T2 including achieve-ment (R2 = 44) intrinsic motivation (R2 = 66) identified regulation(R2 = 52) introjected regulation (R2 = 51) external regulation(R2 = 47) and amotivation (R2 = 44)

53 Brief discussion

Overall we replicated results found in Studies 2 and 3 showingthat intrinsic motivation was the most strongly positively

Fig 3 Cross-lagged panel modelling of academic achievement and types of motivation in Study 4 Notes Correlations between each motivation itemrsquos error term at eachassessment between factors at Time 1 and between disturbances were estimated Equality constraints were imposed on same itemsrsquo loading across time Only significantpaths are presented but all were allowed to covary

associated with achievement over a one-year period in a popula-tion of Swedish high school science students This indicates thatscience students who feel that they have freely chosen the aca-demic orientation that they are interested in and stimulated by willhave the needed engagement and drive to succeed over time Theresults also showed that amotivation was related to lower scienceachievement over one year which is not surprising and confirmsprevious research (Guay et al 2008) If a science student lacks mo-tivation it can be very difficult to achieve well in science courseswhich are known to be demanding Moreover it is interesting tonote that the positive relation of intrinsic motivation with schoolachievement was nearly twice as strong as the negative associa-tion of amotivation with the same outcome

Our results also showed that T1 external regulation was posi-tively associated with achievement at T2 However this relation wasweaker than the relation of intrinsic motivation to achievement Thisresult does not support the predictions of SDT (Deci amp Ryan 2000)One possible explanation for it is that science students who are aboutto graduate are thinking about university applications and futurejobs and that this external focus may push them to do better inschool during their final year In other words extrinsic motivationmay have a positive effect at this stage of their education No re-lationships were obtained between the other types of motivation(identified introjected) Finally T1 achievement was found to neg-atively predict T2 introjected regulation suggesting that the highera studentrsquos achievement the less she will be motivated by wantingto prove her worth or by wanting to avoid feeling guilty

6 General discussion

The objective of the present investigation was to examine therelations of different types of motivation to overall academic achieve-ment in order to adequately test which types play the mostimportant role We first performed a meta-analysis of the litera-ture showing that in general intrinsic motivation and identifiedregulation (ie both types of autonomous motivation) have the stron-gest positive relations to academic achievement whereas amotivationhas the strongest negative relation Introjection and external reg-ulation had weaker and more inconsistent relations to academicachievement However it also uncovered that most studies have usedcross-sectional designs or have not controlled for baseline levelsof achievement These methodological issues were addressed withthree controlled longitudinal studies of high school and college stu-dents in Canada and in Sweden The results across all three studiesshowed that intrinsic motivation was the only motivation type tobe consistently positively associated with academic achievement overa one-year period controlling for baseline achievement and for itsreciprocal relation to subsequent academic motivation types Thisreplicates results reported in one of the few controlled longitudi-nal studies in the field (eg Baker 2003) and supports Self-Determination Theory which highlights intrinsic motivation as theprototype of autonomy (Deci amp Ryan 2000)

Amotivation was significantly associated with lower schoolachievement in our final two studies The fact that students whofeel like they do not know why they are in school have difficultymaintaining their grades is not surprising Amotivation entails feel-ings of alienation and incompetence both of which will naturallylead to problems in academic self-regulation (Legault Green-Demersamp Pelletier 2006)

Interestingly our findings also provide some evidence that therelations between academic motivation types and achievement maybe reciprocal Indeed we showed that prior achievement can pos-itively predict subsequent intrinsic motivation (Study 2) and thatit can negatively predict external regulation (Study 3) and introjectedregulation (Study 4) This is not surprising in light of SDT whichposits that the satisfaction of competence along with the other basic

needs of autonomy and relatedness leads to the development ofmore autonomous forms of motivation (Deci amp Ryan 2000) In ourstudies achievement may have influenced studentsrsquo perceptions oftheir academic competence at T1 which then may have led to moreintrinsic motivation and less introjected and external regulation atT2 These results also corroborate findings by Garon-Carrier et al(2014) and Goldberg amp Cornell (1998) who showed that priorachievement was associated with later intrinsic motivation over timeHowever our results were not consistent across studies and thepattern of relations found between motivation and achievement pro-vides more evidence to suggest that the direction of the relationgoes from motivation to achievement

61 The importance of intrinsic motivation foracademic achievement

Interestingly several researchers argue that intrinsic motiva-tion is not necessarily the most important for young learners Forexample Eccles and Wigfield (2002) have argued that extrinsic mo-tivation in the form of attainment or utility value (eg pursuingan activity because it is important for the pursuit of future goalsor to please onersquos parents) is crucial in determining a young pe-rsonrsquos achievement in school Moreover researchers working withSDT would also expect identified regulation another form of au-tonomous motivation to be very important for the regulation ofschool-related activities that are not necessarily interesting for chil-dren such as homework (Koestner amp Losier 2002) However ourstudies provide consistent evidence that intrinsic motivation seemsto be the most important ldquomotivational ingredientrdquo in the recipefor academic achievement

Our result regarding intrinsic motivation is also surprising giventhat this type of motivation has been repeatedly shown to de-crease as children move into higher grades (Lepper et al 2005 Otiset al 2005 Ratelle et al 2005 Wigfield amp Eccles 2002) Like thosebefore us we also found evidence that intrinsic motivation de-creased over timemdashas reflected in lower mean scores at Time 2 thanTime 1 Moreover we found that intrinsic motivation was less com-monly endorsed by high school and college students than identifiedregulation or even introjected regulation Nonetheless it seems thatintrinsic motivation was the type of motivation that played the mostrobust role in predicting school achievement

This finding provides strong support for the prediction of SDTthat intrinsic motivation is positively associated with school achieve-ment (Deci et al 1991 Niemiec amp Ryan 2009 Pintrich 2003)because it reflects a sense of volition and personal interest ratherthan external pressure Thus a student who goes to school becausehe enjoys learning new things and is stimulated by his accomplish-ments will be more likely to work harder to receive better gradesand to want to stay in school This finding is consistent across allthree studies showing that intrinsic motivation had a positive re-lation to achievement in high school and that it could predict positivechanges in this outcome over time as well as during an importantschool transition and in different cultures Our findings are also con-sistent with other research on the benefits of intrinsic motivationfor conceptual learning (Benware amp Deci 1984) creativity (Amabile1983 Lepper et al 1973) flow (see Ryan amp Deci 2000) vitality (egMoller Deci amp Ryan 2006) and psychological well-being (see Mi-quelon et al 2005) obtained across various spheres of life such asphysical education (Ntoumanis 2001) management (Gagneacute amp Deci2005 Taylor amp Adalsteinsdottir 2003) parenting (Grolnick ampApostoleris 2002) and weight loss and health (Koestner Otis PowersPelletier amp Gagnon 2008 Williams et al 2006)

Previous studies appear to have missed the unique importanceof intrinsic motivation in predicting school achievement over timeThis is largely due to the fact that most studies measuring academ-ic motivation according to SDT have been cross-sectional and only

a few prospective studies have controlled for previous achieve-ment (Baker 2003 Black amp Deci 2000 Burton et al 2006) and forreciprocal relations between motivation types and achievement(Guay et al 2010) Because it is known that students who are highin intrinsic motivation and identified regulation also tend to havehigher achievement (Niemiec amp Ryan 2009) it is thus difficult tointerpret the findings of studies that have failed to control for oneof the most important predictors of future school achievement Inaddition cross-sectional studies do not provide any informationabout the potential predictive power of motivational variables overtime Our findings extend current knowledge by using a prospec-tive design and controlling for previous school achievement toexamine the specific role of academic motivation in the predic-tion of achievement over time

Amotivation was found to be a significant predictor of achieve-ment in two of our studies This supports past research suggestingthat amotivation is highly detrimental for school achievement(Lavigne et al 2007 Otis et al 2005) However the relation ob-tained between amotivation and changes in achievement wasconsiderably weaker than in previous cross-sectional studiesand those that did not control for a baseline measure of achieve-ment We believe it is noteworthy that amotivation whichreflects strong feelings of alienation and incompetence was notas strongly related to poor school achievement as was having alack of intrinsic motivation Our results are similar to those ob-tained in a study conducted by Otis et al (2005) examining differentmotivation types separately and showing that students who wereintrinsically motivated at the end of junior high school were theleast vulnerable to the long-term negative effects of the seniorhigh school transition whereas amotivated students were the mostvulnerable

62 Intrinsic motivation across different cultures

It is interesting to note that the significantly positive path betweenintrinsic motivation and achievement was replicated across differ-ent school settings and across two cultures ie Canada and Swedenwhich differ in their emphasis on what motivates peoplemdashwantingto be the best versus liking what one does A difference betweenthe educational systems of Sweden (where cooperation supportfor autonomy and lack of external constraints are the rule) andCanada (where competition rewardsgrades are much more present)made it interesting to examine the relations between academicmotivation and achievement in order to test the universality hy-pothesis of SDT ie that autonomous forms of motivation will bepositively related to achievement even across different culturesReplicating the results of several studies conducted across IsraeliBelgian German and Korean samples in education (Assor et al2009 Jang Reeve Ryan amp Kim 2009 Levesque et al 2004)our longitudinal findings provide additional support for thegeneralizability of motivational processes across cultures (Deci ampRyan 2000)

However it must be noted that the relation between T1 intrin-sic motivation and T2 achievement was much stronger in theSwedish sample than in the other Canadian samples One expla-nation for this could be based on the way achievement is assessedacross these two educational systems Because the Swedish systemis more autonomy-supportive and less focused on rigid external con-straints than the Canadian system achievement is assessedaccordingly using more open-ended performance criteria such aswritten essays demonstrating reasoning processes or collabora-tive problem solving (Nusche Halaacutesz Looney Santiago ampShewbridge 2011) Since intrinsic motivation has been shown topromote conceptual learning (Benware amp Deci 1984) and creativ-ity (Amabile 1983 Lepper et al 1973) which are best measuredwith open-ended performance tests it is likely that the intrinsic mo-

tivation that Swedish students develop within this environment ismore strongly linked with achievement when it is measured in sucha congruent way

63 Relations between different academic motivationtypes over time

It is interesting that our studies not only indicate that differentacademic motivation types can influence achievement over timebut that they can also influence each other Specifically in Study1 high school studentsrsquo intrinsic motivation positively predicted theirlevels of identified regulation one year later More interestingly wealso found that for Swedish and Canadian science students (Studies3 and 4) intrinsic motivation was associated with less amotivationone year later which suggests that intrinsic motivation can also serveto prevent academic disengagement at a time that can define therest of their educational path By contrast there was some evi-dence that introjected regulation was associated with greateramotivation over time

64 Practical implications

Our studies provide consistent support for the beneficial role ofengaging in school activities because of interest and enjoymentThese findings can serve to provide practical guidelines forteachersrsquo professional development as well as for the design of in-terventions to promote school achievement by focusing on ignitingstudentsrsquo interests for different subjects or fields and by maintain-ing this intrinsic motivation through the support of studentsrsquo basicpsychological needs of autonomy competence and relatedness byteachers (Deci amp Ryan 2000 Deci et al 1991 Reeve Jang CarrellJeon amp Barch 2004 Vallerand et al 1997) The findings also implythat parents would do well to focus on encouraging their adoles-cents to pursue studies that are interesting and exciting to them

65 Limitations and future research

The central limitation of our prospective studies was that werelied mostly on self-report measures of motivation It would havebeen useful to collect measures from other sources such as parentsand peers Another limitation of the present investigation con-cerns the cross-cultural application of SDT Even though Swedendiffers from Canada in terms of masculinity and school environ-ments they are both Western cultures and have many othersimilarities It would be interesting for future research to examinewhether intrinsic motivation predicts further achievement inmore radically different cultures (eg countries in Asia or Africa)given that some cross-cultural studies based on different theoret-ical frameworks have shown that some external types of motivationssuch as social goals (Cheng amp Lam 2013) and performance goals(King McInerney amp Watkins 2012) seem to be similarly predic-tive of achievement

Finally our studies adopted a variable-oriented approach Wefound that prior intrinsic motivation is consistently positively as-sociated with subsequent achievement in school However thisapproach did not allow us to identify whether a student who en-dorsed autonomous motivations for going to school also endorsedcontrolled motivations Because these two types of motivation arecorrelated in most studies (Ryan amp Deci 2002) it would be impor-tant for future studies to examine whether reporting both types ofmotivations simultaneously is beneficial for students Research in SDThas recently started using such a person-centered approach in orderto evaluate which motivational profiles are most beneficial for stu-dentsrsquo academic adjustment (Poulin Duchesne amp Ratelle 2010Ratelle et al 2007)

7 Conclusion

Many studies have shown that autonomous motivation is pos-itively related to academic achievement (Niemiec amp Ryan 2009)However these relations have rarely been investigated in prospec-tive studies of high school students that controlled for baselineachievement The objective of the present investigation was to usea controlled prospective design to examine whether these motiva-tion types could predict changes in academic achievement over timeduring the transition from high school to college and in differentcultures We showed that intrinsic motivation is consistently themost beneficial form of motivation for studentsrsquo achievement Ourfindings highlight the importance of encouraging students to pursuesubjects that they are passionate about They also appear to givecredence to Socratesrsquo idea that education is about ldquothe kindling ofa flame not the filling of a vesselrdquo (Socrates 470 BC)

References marked with an asterisk indicate studies included in themeta-analysis

References

Amabile T M (1983) The social psychology of creativity A componentialconceptualization Journal of Personality and Social Psychology 45 357ndash376doi10103700223514452357

Archambault I Janosz M Morizot J amp Pagani L (2009) Adolescent behavioralaffective and cognitive engagement in school Relationship to dropout TheJournal of School Health 79 408ndash415 doi101111j1746-1561200900428x

Assor A Vansteenkiste M amp Kaplan A (2009) Identified versus introjectedapproach and introjected avoidance motivations in school and in sports Thelimited benefits of self worth strivings Journal of Educational Psychology 101482ndash497 doi101037a0014236

Baker S R (2003) A prospective longitudinal investigation of social problem-solvingappraisals on adjustment to university stress health and academic motivationand performance Personality and Individual Differences 35 569ndash591 doi101016S01918869(02)00220-9

Benware C A amp Deci E L (1984) Quality of learning with an active versus passivemotivational set American Educational Research Journal 21 755ndash765 doi10310200028312021004755

Black A E amp Deci E L (2000) The effects of instructorsrsquo autonomy support andstudentsrsquo autonomous motivation on learning organic chemistry A self-determination theory perspective Science Education 84 740ndash756 doi1010021098-237X(200011)846 lt 740AID-SCE4gt30CO2-3

Boicheacute J C S Sarrazin P G Grouzet F M E Pelletier L G amp Chanal J P (2008)Studentsrsquo motivational profiles and achievement outcomes in physical educationA self-determination perspective Journal of Educational Psychology 100 688ndash701doi1010370022-06631003688

Buhi E R Goodson P amp Neilands T B (2008) Out of sight not out of mindStrategies for handling missing data American Journal of Health Behavior 32(1)83ndash92

Burton K D Lydon J E DrsquoAlessandro D U amp Koestner R (2006) The differentialeffects of intrinsic and identified motivation on well-being and performanceProspective experimental and implicit approaches to self-determination theoryJournal of Personality and Social Psychology 91 750ndash762 doi1010370022-3514914750

Chatzisarantis N L D Hagger M S Biddle S J H Smith B amp Wang J C K(2003) A meta-analysis of perceived locus of causality in exercise sport andphysical education contexts Journal of Sport amp Exercise Psychology 25(3) 284ndash306

Cheng R W amp Lam S (2013) The interaction between social goals and self-construalon achievement motivation Contemporary Educational Psychology 38(2) 136ndash148doi101016jcedpsych201301001

Chirkov V amp Ryan R M (2001) Parent and teacher autonomy-support in Russianand US adolescents Common effects on well-being and academic motivationJournal of Cross-Cultural Psychology 32(5) 618ndash635

Cohen J (1992) A power primer Psychological Bulletin 112(1) 155ndash159 doi1010370033-29091121155

Cokley KO Bernard N Cunningham D amp Motoike J (2001) A psychometricinvestigation of the Academic Motivation Scale using a United Statessample Measurement and Evaluation in Counseling and Development 34(2)109ndash119

DrsquoAilly H (2003) Childrenrsquos autonomy and perceived control in learning A modelof motivation and achievement in Taiwan Journal of Educational Psychology 9584ndash96 doi1010370022-066395184

Deci E L amp Ryan R M (1985) Intrinsic motivation and self-determination in humanbehavior New York Plenum

Deci E L amp Ryan R M (2000) The ldquowhatrdquo and ldquowhyrdquo of goal pursuits Human needsand the self-determination of behavior Psychological Inquiry 11 227ndash268doi101207S15327965PLI1104_01

Deci E L Koestner R amp Ryan R M (1999) A meta-analytic review of experimentsexamining the effects of extrinsic rewards on intrinsic motivation PsychologicalBulletin 125 627ndash668 doi1010370033-29091256627

Deci E L amp Ryan R M (2002) Overview of Self-Determination Theory Anorganismic dialectical perspective In E L Deci amp R M Ryan (Eds) Handbookof Self-Determination Research (pp 3ndash33) Rochester NY University of RochesterPress

Deci EL Vallerand RJ Pelletier LG amp Ryan RM (1991) Motivation and educationThe self-determination perspective Educational Psychologist 26 325ndash346doi1010800046152019919653137

DeCharms R (1968) Personal causation The internal affective determinants of behaviorNew York Academic Press

Dornbusch S M Ritter P L Leiderman P H Roberts D F amp Fraleigh M J (1987)The relation of parenting style to adolescent school performance ChildDevelopment 58 1244 doi1023071130618

Duchesne S Ratelle C F Larose S amp Guay F (2007) Adjustment trajectories incollege science programs Perceptions of qualities of parentsrsquo and college teachersrsquorelationships Journal of Counseling Psychology 54 62ndash71 doi1010370022-016754162

Eccles J S amp Wigfield A (2002) Motivational beliefs values and goal Annual Reviewof Psychology 53 109ndash132 doi101146annurevpsych53100901135153

Eisinga R Grotenhuis M amp Pelzer B (2013) The reliability of a two-item scalePearson Cronbach or Spearman-Brown International Journal of Public Health58(4) 637ndash642 doi101007s00038-012-0416-3

Elliot A J amp Moller A C (2003) Performance-approach goals Good or bad formsof regulation International Journal of Educational Research 39(4ndash5) 339ndash356doi101016jijer200406003

Field A P amp Gillett R (2010) How to do a meta-analysis British Journal ofMathematical and Statistical Psychology 63 665ndash694 doi101348000711010X502733

Fortier MS Vallerand RJ amp Guay F (1995) Academic motivation and schoolperformance Toward a structural model Contemporary Educational Psychology20 257ndash274 doi101006ceps19951017

Gagneacute M amp Deci E L (2005) Self-determination theory and work motivation Journalof Organizational Behavior 26 331ndash362 doi101002job322

Garon-Carrier G Boivin M Forget-Dubois N Guay F Kovas Y Lemelin J-P et al(2014) Une eacutetude longitudinale de lrsquoassociation entre la motivation intrinsegravequeet le rendement en matheacutematiques [A longitudinal study of the associationbetween intrinsic motivation and mathematics achievement] Paper presentedat the Association francophone pour le savoir (ACFAS) Montreal QuebecCanada

Goldberg M D amp Cornell D G (1998) The influence of intrinsic motivationand self concept on academic achievement in second- and third-gradestudents Journal for the Education of the Gifted 21 179ndash205 doi101177016235329802100204

Grolnick W S amp Apostoleris N H (2002) What makes parents controlling In EL Deci amp R M Ryan (Eds) Handbook of self-determination research (pp 161ndash203)Rochester NY University of Rochester Press

Grolnick W S Ryan R M amp Deci E L (1991) Inner resources for schoolachievement Motivational mediators of childrenrsquos perceptions of their parentsJournal of Educational Psychology 83 508ndash517 doi1010370022-0663834508

Guay F Ratelle C F amp Chanal J (2008) Optimal learning in optimal contexts Therole of self-determination in education Canadian Psychology 49 233ndash240doi101037a0012758

Guay F Ratelle C F Roy A amp Litalien D (2010) Academic self-concept autonomousacademic motivation and academic achievement Mediating and additiveeffects Learning and Individual Differences 20(6) 644ndash653 doi101016jlindif201008001

Hardre P L amp Reeve J (2003) A motivational model of rural studentsrsquo intentionsto persist in versus drop out of high school Journal of Educational Psychology95 347ndash356 doi1010370022-0663952347

Harter S (1981) A new self-report scale of intrinsic versus extrinsic orientation inthe classroom Motivational and informational components DevelopmentalPsychology 17 300ndash312 doi1010370012-1649173300

Hedges L V amp Olkin I (1985) Statistical methods for meta-analysis New YorkAcademic Press

Hennan M R Dornbusch S M Herron M C amp Herting J R (1997) The influenceof family regulation connection and psychological autonomy on six measuresof adolescent functioning Journal of Adolescent Research 12 34ndash67 doi1011770743554897121004

Hofstede G (1983) National cultures in four dimensions A research-based theoryof cultural differences among nations International Studies of Management ampOrganization 46ndash74

Jang H Reeve J Ryan R M amp Kim A (2009) Can self-determination theory explainwhat underlies the productive satisfying learning experiences of collectivisticallyoriented Korean students Journal of Educational Psychology 101 644ndash661doi101037a0014241

Johnson BT (1993) DSTAT 110 Software for the meta-analytic review of literaturesHillsdale NJ Erlbaum

King R B McInerney D M amp Watkins D A (2012) Competitiveness is not thatbad at least in the East Testing the hierarchical model of achievementmotivation in the Asian setting International Journal of Intercultural Relations36(3) 446ndash457 doi101016jijintrel201110003

Kline R B (2011) Principles and practice of structural equation modeling New YorkGuilford Press

Koestner R amp Losier G F (2002) Distinguishing three ways of being highlymotivated A closer look at introjection identification and intrinsic motivationIn E L Deci amp R M Ryan (Eds) The handbook of self-determination theory research(pp 101ndash121) Rochester NY University of Rochester Press

Koestner R Otis N Powers T A Pelletier L amp Gagnon H (2008) Autonomousmotivation controlled motivation and goal progress Journal of Personality 761201ndash1230 doi101111j1467-6494200800519x

Lavigne G L Vallerand R J amp Miquelon P (2007) A motivational model ofpersistence in science education A self-determination theory approach EuropeanJournal of Psychology of Education 22(3) 351ndash369

Legault L Green-Demers I amp Pelletier L (2006) Why do high school students lackmotivation in the classroom Toward an understanding of academic amotivationand the role of social support Journal of Educational Psychology 98 567ndash582doi1010370022-0663983567

Lekes N Joussemet M Koestner R Taylor G Hope N H amp Gingras I (2011)Transmitting intrinsic value priorities from mothers to adolescents Themoderating role of a supportive family environment Child Development Research2011 1ndash9 doi1011552011167146

Lepper M R Corpus J H amp Iyengar S S (2005) Intrinsic and extrinsic motivationalorientations in the classroom Age differences and academic correlates Journalof Educational Psychology 97 184ndash196 doi1010370022-0663972184

Lepper M R Greene D amp Nisbett R E (1973) Undermining childrenrsquosintrinsic interest with extrinsic reward A test of the ldquooverjustificationrdquo hypothesisJournal of Personality and Social Psychology 28 129ndash137 doi101037h0035519

Levesque C Zuehlke A N Stanek L R amp Ryan R M (2004) Autonomy andcompetence in German and American university students A comparative studybased on self-determination theory Journal of Educational Psychology 96 68ndash84doi1010370022-066396168

Marsh H W amp Hau K T (1996) Assessing goodness of fit Is patrimony alwaysdesirable Journal of Experimental Education 64(4) 364

Marsh H W Hau K T amp Grayson D (2005) Goodness of fit in structural equationmodels In R P McDonald A Maydeu-Olivares amp J J McArdle (Eds) Contemporarypsychometrics A festschrift for Roderick P McDonald Mahwah NJ LawrenceErlbaum Associates

Ministegravere de lrsquoEacuteducation du Loisir et du Sport (MELS) (2013) Indices dedeacutefavorisation 2012ndash2013 [Socioeconomic index 2012ndash2013] Retrieved fromhttpwwwmelsgouvqccafileadminsite_webdocumentsPSGstatistiques_info_decisionnelleIndices_par_CS2013ppdf

Ministry of Education and Science (2001) The development of education Nationalreport of Sweden lthttpwwwibeunescoorgInternationalICEnatrapSwedenpdfgt

Miquelon P Vallerand R J Grouzet F M E amp Cardinal G (2005) Perfectionismacademic motivation and psychological adjustment An integrative model PersSoc Psychol Bull 31 913ndash924 doi1011770146167204272298

Moller AC Deci EL amp Ryan RM (2006) Choice and ego-depletion The moderatingrole of autonomy Personality and Social Psychology Bulletin 32 1024ndash1036doi1011770146167206288008

Morin A J S Madore I Morizot J Boudrias J-S amp Tremblay M (2009) Multipletargets of workplace affective commitment Factor structure and measurementinvariance of the Workplace Affective Commitment MultidimensionalQuestionnaire In A M Columbus (Ed) Advances in psychology research (Vol 59pp 45ndash75) Hauppauge NY Nova Science Publishers Inc

Niemiec C P amp Ryan R M (2009) Autonomy competence and relatedness in theclassroom Applying self-determination theory to educational practice Theoryand Research in Education 7 133ndash144 doi1011771477878509104318

Noels K A Clement R amp Pelletier L G (1999) Perceptions of teachersrsquocommunicative style and studentsrsquo intrinsic and extrinsic motivation The ModernLanguage Journal 83 23ndash34 doi1011110026-790200003

Ntoumanis N (2001) A self-determination approach to the understanding ofmotivation in physical education The British Journal of Educational Psychology71 225ndash242 doi101348000709901158497

Ntoumanis N (2002) Motivational clusters in a sample of British physical educationclasses Psychology of Sport and Exercise 3(3) 177ndash194 doi101016S1469-0292(01)00020-6

Ntoumanis N Barkoukis V amp Thoslashgersen-Ntoumani C (2009) Developmentaltrajectories of motivation in physical education Course demographic differencesand antecedents Journal of Educational Psychology 101 717ndash728 doi101037a0014696

Nusche D Halaacutesz G Looney J Santiago P amp Shewbridge C (2011) OECD reviewsof evaluation and assessment in education Sweden lthttpwwwoecdorgeduschool47169533pdfgt

Otis N Grouzet F M E amp Pelletier L G (2005) Latent motivational change in anacademic setting A 3-year longitudinal study Journal of Educational Psychology97 170ndash183 doi1010370022-0663972170

Petersen I Louw J amp Dumont K (2009) Adjustment to university and academicperformance among disadvantaged students in South Africa EducationalPsychology An International Journal of Experimental Educational Psychology 2999ndash115 doi10108001443410802521066

Peugh J L amp Enders C K (2004) Missing data in educational research A reviewof reporting practices and suggestions for improvement Review of EducationalResearch 74 525ndash556

Pintrich P R (2003) A motivational science perspective on the role of studentmotivation in learning and teaching contexts Journal of Educational Psychology95 667ndash686 doi1010370022-0663954667

Poulin R Duchesne S amp Ratelle CF (2010) Profils de buts drsquoapprentissage etcaracteacuteristiques personnelles des eacutelegraveves au deacutebut du secondaire [Learning goalprofiles and personal characteristics of junior high school students] CanadianJournal of Behavioural Science Revue canadienne des Sciences du comportement42 44ndash54 doi101037a0016544

Ratelle C F Vallerand R J Seneacutecal C amp Provencher P (2005) The relationshipbetween school-leisure conflict and educational and mental health indexes Amotivational analysis Journal of Applied Social Psychology 35 1800ndash1823doidoi101111j1559-18162005tb02196x

Ratelle C Guay F Vallerand RJ Larose S amp Seneacutecal C (2007) Autonomouscontrolled and amotivated types of academic motivation A person-orientedanalysis Journal of Educational Psychology 99 734ndash746 doi1010370022-0663994734

Reeve J Jang H Carrell D Jeon S amp Barch J (2004) Enhancing studentsrsquoengagement by increasing teachersrsquo autonomy support Motivation and Emotion28 147ndash169 doi101023BMOEM0000032312954996f

Retelsdorf J Koumlller O amp Moumlller J (2014) Reading achievement and readingself-conceptmdashTesting the reciprocal effects model Learning and Instruction 2921ndash30 doi101016jlearninstruc201307004

Rosenthal R amp DiMatteo M R (2001) Meta-analysis Recent developments inquantitative methods for literature reviews Annual Review of Psychology 5259ndash82 doi101146annurevpsych52159

Ryan R M amp Connell J P (1989) Perceived locus of causality and internalizationExamining reasons for acting in two domains Journal of Personality and SocialPsychology 57 749ndash761 doi1010370022-3514575749

Ryan R M amp Deci E L (2000) Self-determination theory and the facilitation ofintrinsic motivation social development and well-being The AmericanPsychologist 55 68ndash78 doi1010370003-066X55168

Schlomer G L Bauman S amp Card N A (2010) Best practices for missing datamanagement in counseling psychology Journal of Counseling Psychology 57(1)1ndash10 doi101037a0018082

Sitzmann T Kraiger K Stewart D amp Wisher R (2006) The comparativeeffectiveness of web-based and classroom instruction A meta-analysis PersonnelPsychology 59 623ndash664 doi101111j1744-6570200600049x

Soenens B amp Vansteenkiste M (2005) Antecedents and outcomes of self-determination in 3 life domains The role of parentsrsquo and teachersrsquo autonomysupport Journal of Youth and Adolescence 34 589ndash604 doi101007s10964-005-8948-y

Taylor G amp Adalsteinsdottir E B (2003) Empowerment job characteristics and intrinsicmotivation An examination of the role of managerial autonomy-support(Unpublished Masterrsquos thesis) London School of Economics and Political ScienceLondon UK

Vallerand R J Blais M R Briegravere N M amp Pelletier L G (1989) Construction etvalidation de lrsquoeacutechelle de motivation en eacuteducation (EME) [Construction andvalidation of the Academic Motivation Scale (AMS)] Canadian Journal ofBehavioural Science Revue Canadienne Des Sciences Du Comportement 21 323ndash349doi101037h0079855

Vallerand R J Fortier M S amp Guay F (1997) Self-determination and persistencein a real life setting Toward a motivational model of high school dropout Journalof Personality and Social Psychology 72 1161ndash1176 doi1010370022-35147251161

Vallerand R J Pelletier L G Blais M R Briegravere N M Seacuteneacutecal C amp Valliegraveres E F(1992) The academic motivation scale A measure of intrinsic extrinsic andamotivation in education Educational and Psychological Measurement 521003ndash1017 doi1011770013164492052004025

Vallerand R J Pelletier L G Blais M R Briegravere N M Seacuteneacutecal C amp Valliegraveres EF (1993) On the assessment of intrinsic extrinsic and amotivation in educationEvidence on the concurrent and construct validity of the Academic MotivationScale Educational and Psychological Measurement 53 159ndash172 doi1011770013164493053001018

Vansteenkiste M Niemiec CP amp Soenens B (2010) The development of the fivemini theories of self-determination theory An historical overview emergingtrends and future directions In T Urdan amp S Karabenick (Eds) (Vol 16 pp105ndash165 ) Advances in motivation and achievement Bradford UK EmeraldGroup Publishing Limited doi101108S0749-7423(2010)000016A007

Walls T A amp Little T D (2005) Relations among personal agency motivation andschool adjustment in early adolescence Journal of Educational Psychology 9723ndash31 doi1010370022-066397123

Wigfield A amp Eccles J S (2000) Expectancy-value theory of achievement motivationContemporary Educational Psychology 25 68ndash81 doi101006ceps19991015

Wigfield A amp Eccles J S (2002) Development of achievement motivation San DiegoCalifornia Academic Press

Williams GC McGregor HA Sharp D Levesque C Kouides RW amp Ryan RMet al (2006) Testing a self-determination theory intervention for motivatingtobacco cessation Supporting autonomy and competence in a clinical trial HealthPsychology 25 91ndash101 doi1010370278-613325191

Wrzesniewski A Schwartz B Cong X Kane M Omar A amp Kolditz T (2014)Multiple types of motives donrsquot multiply the motivation of West Point cadetsProceedings of the National Academy of Sciences 111(30) 10990ndash10995doi101073pnas1405298111

A self-determination theory approach to predicting school achievement over time the unique role of intrinsic motivation

Introduction

Self-determination theory in education

Academic motivation and educational achievement

Overview of studies

Study 1

Method

Selection of studies

Dependent measure and computation of effect sizes

Moderator analyses

Results

Effect sizes for individual types of motivation

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Perceived academic achievement

Statistical analyses

Results

Preliminary analyses

Missing data

Descriptive statistics

Measurement model and factor loadings invariance

Stability and reciprocal effects model

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Academic achievement

Results

Missing data

Stability and reciprocal effects path model

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

The importance of intrinsic motivation for academic achievement

Intrinsic motivation across different cultures

Relations between different academic motivation types over time

Practical implications

Limitations and future research

Conclusion

References

that is extrinsically motivated is not performed out of interest butfor the consequence it is thought to be instrumentally linked to(Wrzesniewski et al 2014) Extrinsic motivation is thought to beimportant for socially prescribed activities such as doing home-work because they are often not inherently interesting Unlike manyconceptualizations of motivation (eg Harter 1981) SDT does notview extrinsic motivation as one-dimensional and opposed to in-trinsic motivation Instead it specifies different types of extrinsicmotivation which vary in the extent to which they are autono-mous (Deci amp Ryan 1985)

These types from the least to the most autonomous consist ofexternal introjected and identified regulation External regulationrefers to behaviours that are initiated by an external contingencyfor example being offered a reward to do onersquos homeworkIntrojected regulation refers to internalizing a regulation withoutfully accepting it into onersquos sense of self It involves feelings of in-ternal coercion and pressure and refers to attempts to avoid feelingunworthy guilty or ashamed or to prove onersquos worth (AssorVansteenkiste amp Kaplan 2009) An example of introjected regula-tion would be a student who studies long hours to prove to herselfthat she is worthy Identified regulation takes place when the valueof an instrumental behaviour has come to be identified with onersquossense of self This type of regulation is considered to be more au-tonomous than the other types of extrinsic motivation because itis initiated from a sense of personal meaning and volition (Deci ampRyan 2000 Koestner amp Losier 2002) A student who does extra ex-ercises at the end of a history chapter because she believes it willhelp her fully understand the subject matter is regulated byidentification1 SDT also considers amotivation the absence of mo-tivation that happens when an individual does not experienceintentionality or a sense of personal causation These different formsof motivation have been proposed to lie along a continuum of rel-ative autonomy starting with the form that exhibits the lowest levelto the one that represents the highest level of autonomy (Deci ampRyan 1985) SDT (Ryan amp Connell 1989) also mentions that adja-cent motivations on the continuum (eg intrinsic motivation andidentified regulation) should relate more strongly to each other thandistal ones (eg intrinsic motivation and external regulation)However evidence for the continuum is inconsistent While somefindings corroborate this pattern others deviate from it in variousways (eg intrinsic motivation being more strongly related tointrojected than to identified regulation) (for examples see BoicheacuteSarrazin Grouzet Pelletier amp Chanal 2008 Ntoumanis 2002 OtisGrouzet amp Pelletier 2005 Ratelle Guay Vallerand Larose amp Seneacutecal2007)

12 Academic motivation and educational achievement

Although many studies have examined the relation of academ-ic motivation to school achievement from the SDT perspective (egChatzisarantis Hagger Biddle Smith amp Wang 2003 Deci VallerandPelletier amp Ryan 1991) the majority have been cross-sectional andhave yielded inconsistent results (Cokley Bernard Cunningham ampMotoike 2001 DrsquoAilly 2003 Fortier Vallerand amp Guay 1995

Grolnick Ryan amp Deci 1991 Hardre amp Reeve 2003 Noels Clementamp Pelletier 1999 Otis et al 2005 Petersen Louw amp Dumont 2009Soenens amp Vansteenkiste 2005 Walls amp Little 2005) A careful ex-amination of past research shows that only a few studies haveadopted a prospective design while also controlling for previousachievement (Baker 2003 Black amp Deci 2000 Burton LydonDrsquoAlessandro amp Koestner 2006 Study 2b) The details of these con-trolled prospective studies merit review Burton et al (2006)conducted a 6-week prospective study of university students toexamine the relations of intrinsic motivation and identified regu-lation to final exam performance in a single psychology courseResults showed that controlling for previous grades identified reg-ulation significantly positively predicted final examination gradeswhereas intrinsic motivation was unrelated to the final grades Blackand Deci (2000) examined the relation of relative autonomy in asample of college chemistry students over a one-semester periodThey found that relative autonomy did not significantly predict finalcourse grade after controlling for previous ability and grade pointaverage (GPA) Results for specific types of motivation were not re-ported Finally Baker (2003) examined the relations of academicmotivation types to total GPA in a sample of university students andcontrolled for academic achievement as measured by entry quali-fications upon entering university Her results showed that intrinsicmotivation assessed during the second semester of the first yearof university was the only type of motivation to significantly predictoverall academic performance measured one year later control-ling for entry qualifications

An example of another longitudinal study that has controlled forbaseline achievement is one by (Guay Ratelle Roy amp Litalien 2010)Using a cross-lagged model to examine the reciprocal relations ofacademic motivation and achievement in a population of high schoolstudents they found that autonomous motivation as defined by arelative autonomy score was positively associated with academicachievement over the course of one year even after controlling forbaseline achievement However they did not estimate the contri-bution of each type of motivation to later achievement making itdifficult to understand which type of motivation was driving thisrelation

Given the inconsistent results of past cross-sectional studies andthe paucity of longitudinal studies that have controlled for base-line levels of achievement a more systematic review of the researchis needed in order to fully understand the effect of each differentmotivation type on school achievement Moreover as Ratelle et al(2007) have suggested more longitudinal studies are necessary toprovide some information about the causal mechanisms betweenmotivation and achievement Finally no studies have examined theselongitudinal relations in samples of high school students Since failureto achieve is a prevalent problem in high school and leads to unde-sirable consequences such as dropout it is imperative to conductcarefully controlled studies in such a pre-university population

13 Overview of studies

In an attempt to overcome the limitations of past research weconducted a meta-analysis and a series of three empirical studiesto systematically examine the contribution of the different moti-vation types to school achievement The meta-analysis (Study 1)reviewed cross-sectional and longitudinal studies that have as-sessed the relation of motivation types to school achievementaccording to SDT using the Academic Motivation Scale (AMS) de-signed by Vallerand et al (1992) This is the most widely used scaleof school motivation from the SDT framework We also designedthree controlled longitudinal studies that used the AMS to measurefive different types of academic motivation and to examinetheir relation to school achievement over time To ensure that ourfindings were robust and generalizable we varied the school context

1 Integrated regulation the most autonomous form of extrinsic motivation occurswhen the value of the instrumental behaviour has come to be in harmony with othervarious aspects of a personrsquos values and identity to form a coherent sense of self Astudent who does not like math but understands the importance and benefits oftaking a statistics class and does so because he wants to eventually become a psy-chologist displays integrated regulation It must be noted that integrated regulationrequires much effort self-awareness and reflection (Vansteenkiste Niemiec ampSoenens 2010) Moreover this type of motivation has not typically been includedin measures of academic motivation because some early studies showed that stu-dents could not differentiate it from identified regulation on self-report scales (RobertJ Vallerand et al 1992)

2 Study 1

21 Method

22 Results

23 Brief discussion

k N ofstudies

d 95 CI

From To

3 Study 2

31 Method

32 Results

1 2 3 4 5 6 7 8 9 10 M SD

p lt 01

33 Brief discussion

4 Study 3

41 Method

42 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

2 Study 1

21 Method

22 Results

23 Brief discussion

k N ofstudies

d 95 CI

From To

3 Study 2

31 Method

32 Results

1 2 3 4 5 6 7 8 9 10 M SD

p lt 01

33 Brief discussion

4 Study 3

41 Method

42 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

22 Results

23 Brief discussion

k N ofstudies

d 95 CI

From To

3 Study 2

31 Method

32 Results

1 2 3 4 5 6 7 8 9 10 M SD

p lt 01

33 Brief discussion

4 Study 3

41 Method

42 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

3 Study 2

31 Method

32 Results

1 2 3 4 5 6 7 8 9 10 M SD

p lt 01

33 Brief discussion

4 Study 3

41 Method

42 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

32 Results

1 2 3 4 5 6 7 8 9 10 M SD

p lt 01

33 Brief discussion

4 Study 3

41 Method

42 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

33 Brief discussion

4 Study 3

41 Method

42 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

42 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

42 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

43 Brief discussion

5 Study 4

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

51 Method

this study

52 Results

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

1 2 3 4 5 6 7 8 9 10 11 M SD

p lt 01

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

53 Brief discussion

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

7 Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

Introduction

Overview of studies

Study 1

Method

Moderator analyses

Results

Moderator analyses

Brief discussion

Study 2

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 3

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

Study 4

Method

Participants

Procedure

Measures

Academic motivation

Results

Missing data

Brief discussion

General discussion

Conclusion

References

Documents

A self-determination theory approach to predicting school ... et al...Title A self-determination theory approach to predicting school achievement over time: the unique role of intrinsic