The importance of motivation as a predictor of school achievement

https://doi.org/10.1016/j.lindif.2008.05.004Get rights and content

Abstract

The present study examined to which extent different motivational concepts contribute to the prediction of school achievement among adolescent students independently from intelligence. A sample of 342 11th and 12th graders (age M = 16.94; SD = .71) was investigated. Students gave self-reports on domain-specific values, ability self-perceptions, goals, and achievement motives. Hierarchical regression and relative weights analyses were performed with grades in math and German as dependent variables and intelligence as well as motivational measures as independent variables. Beyond intelligence, different motivational constructs incrementally contributed to the prediction of school achievement. Domain-specific ability self-perceptions and values showed the highest increments whereas achievement motives and goal orientations explained less additional variance. Even when prior achievement was controlled, some motivational concepts still proved to contribute to the prediction of subsequent performance. In the light of these findings, we discuss the importance of motivation in educational contexts.

Introduction

It is well-established that general school achievement is highly related to general intelligence (e.g., Kuncel, Hezlett, & Ones, 2004). Given that general intelligence explains only about 25% of the variance in scholastic achievement (Kuncel et al., 2004), it is worthwhile to look for other concepts that might add to the explained variance. Motivation is one of the constructs thought to cover a share of school performance variance not explained by intelligence. Despite the importance that is often ascribed to motivation in school contexts, only a few studies have so far investigated the incremental validity of motivation above and beyond general intelligence. There is even less research on how motivational constructs compare with intelligence and prior performance as another important predictor of subsequent school performance. Furthermore, there are only a few studies investigating how much unique criterion-related variance can be attributed to certain motivational constructs, intelligence, and prior school performance, respectively, considering criterion-related validity shared by the different predictors. Studies of this kind allow estimating the relative importance of different predictors, providing information beyond incremental validity analyses (cf. LeBreton, Hargis, Griepentrog, Oswald, & Ployhart, 2007). The current study aims at two goals. First, we investigate the incremental validity of different motivational constructs above and beyond intelligence. Second, we calculate the relative importance of these motivational constructs, intelligence, and prior school performance when all these constructs predict subsequent school achievement together.

Contemporary achievement motivation literature discusses a great variety of concepts all subsumed under the term motivation (cf. Murphy & Alexander, 2000). For the purpose of our investigation we shall focus on three of the most prominent approaches: need achievement theory, expectancy-value theory, and goal theories. All of these theories have been extensively investigated in school settings (Covington, 2000, Eccles and Wigfield, 2002). We have chosen to neglect theories that are either conceptually very close to the investigated motivational constructs [e.g., self-efficacy (Bandura, 1977) which is very similar to ability self-perceptions] or have not been as extensively investigated with regard to academic achievement as the concepts listed above (cf. Murphy & Alexander, 2000). In the following, we take a closer look at the above mentioned three motivational theories, the measurement of their central constructs, and their relations to school achievement. One aspect important for our hypotheses is the theorized generality or specificity of the motivational concepts. Thus the following sections explain why we operationalize the variables either domain-specifically or -generally.

Murray (1938) considered need for achievement as one of the basic human needs. In his conception needs, such as need for achievement, are “a more or less consistent trait of personality” (Murray, 1938, p. 61). As need for achievement is part of a person's personality, it is thought to trigger behavior across different situations. Consequently, need for achievement is theorized to be domain-general and, thus, assessed without referring to a certain domain or situation. Need for achievement assessed in the vein of Murray's personology covers thoughts and behaviors associated with success, accomplishment, and overcoming obstacles. Jackson (1967) used Murray's personology as the basis for his Personality Research Form (PRF) and one scale assesses need for achievement. The PRF is one of the personality questionnaires most frequently used world wide and a gold standard in personality research.

Murray's concept of need for achievement was extended by McClelland and others who founded the classical achievement motive research (cf. McClelland et al., 1953). According to McClelland and colleagues need for achievement is the result of an emotional conflict between the hope to approach success and the desire to avoid failure. Hope for success, on the one hand, is associated with positive emotions and the belief to succeed. Fear of failure, on the other hand, is related to negative emotions and the fear that the achievement situation is out of one's depth. The balance of these two motives is thought to determine the direction, intensity, and quality of achievement-related behavior. Need for achievement according to McClelland's is measured by describing affective experiences or associations like fear or joy in achievement situations. Like Murray's need for achievement conception, hope for success and fear of failure are also thought to possess omnibus importance in achievement-related behaviors and are thus operationalized in a domain-general manner, i.e., referring to general achievement situations such as problem solving. McClelland's and Murray's conceptions differ with regard to at least two points. First, McClelland's conception of need for achievement acknowledges that achievement situations are not only characterized by the need or the hope to achieve (as in Murray's conception) but also by the possibility to fail. Second, in contrast to Murray's conception of need for achievement McClelland strongly underlines the affective components of achievement situations. Therefore, need for achievement according to McClelland is assessed with different instruments. A frequently used instrument in achievement motivation research is the Achievement Motives Scale by Gjesme and Nygard (1970).

Due to the fact that school largely consists of achievement situations, need for achievement, both in the sense of McClelland and Murray, is triggered quite often. Since accomplishments in school can only be achieved by learning or fulfilling the demands of school, students high on need for achievement or hope for success are supposed to work hard to achieve. Correlations of achievement motives assessed via self-report, both in the sense of McClelland and Murray, and academic success are mostly of weak to medium magnitude in high school and college student samples (e.g., Paunonen and Ashton, 2001, Spangler, 1992).

In the expectancy-value model of Eccles and her colleagues (Eccles et al., 1983, Wigfield and Eccles, 2000) achievement-related behavior is explained by expectancies for future success and the values ascribed to a task. Unlike need for achievement, both expectancies and values are conceptualized in a domain-specific manner, i.e., focusing on specific tasks or subjects. This is in line with empirical research demonstrating that ability self-concepts and values have been shown to be domain-specific from the early school years on (e.g., Gottfried, 1985, Wigfield et al., 1997).

According to the Eccles model, expectations of future success are largely determined by ability self-perceptions (synonyms are ability beliefs or ability self-concepts). The construct is measured by asking students how good they think they are at certain tasks or subjects. Academic ability self-concepts and academic achievement are usually moderately to highly correlated within domains (e.g., Guay et al., 2003, Marsh et al., 1983, Skaalvik and Valas, 1999). The importance of ability self-perceptions is further emphasized by the body of research on causal ordering of ability self-concepts and school achievement (e.g., Guay et al., 2003, Marsh and Yeung, 1997). It was shown that not only prior achievement influences ability self-perception but that prior ability beliefs influence subsequent achievement.

With reference to values, the model of Eccles and her colleagues focuses on intrinsic, importance, and utility values as the major reasons for achievement behavior (Eccles & Wigfield, 1995). Task values are most commonly measured by means of asking an individual to rate how important, useful, or interesting a task is to them (cf. Jacobs, Lanza, Osgood, Eccles, & Wigfield, 2002). Interest is the best investigated of the three values mentioned. Domain-specific correlations between interest and values, respectively, and school achievement are weak to moderate (e.g., Gottfried, 1985, Gottfried, 1990, Lloyd and Barenblatt, 1984, Steinmayr and Spinath, 2007). This is in line with predictions derived from expectancy-value theory (Eccles et al., 1983, Wigfield and Eccles, 2000) according to which values should be better predictors of choices in achievement contexts whereas ability self-concepts should be more strongly correlated with actual performance.

In the early literature on achievement goals, goals are separated into learning and performance goals (e.g., Dweck, 1986, Nicholls, 1984). Within this dichotomous framework, learning goals focus on the development of competence while performance goals center the demonstration of competence. Later on, Elliot and others further distinguished between performance-approach (striving to demonstrate competence) and performance-avoidance goals (striving not to demonstrate incompetence) (Elliot & Church, 1997). Furthermore, some researchers include one component in their considerations which represents the opposite of high achievement motivation, namely the tendency of work-avoidance (e.g., Nicholls, 1984, Harackiewicz et al., 1997). Work-avoidance refers to the goal to invest as little effort as possible.

Goals can be considered as situation-specific states or as cross-situationally consistent and stable-over-time traits. For example, Nicholls considered goal orientations as depending on different notions of success (Nicholls, Patashnick, & Nolen, 1985) and Dweck reasoned that goal orientations rely on implicit theories about intelligence (Dweck & Leggett, 1988). Both theoretical conceptions imply that domain-general goal orientations influence achievement-related behavior in different situations, domains, or at different tasks in the same way. Thus, goal orientations are often measured in a domain-general way, although situationally specific measures might add valuable information for specific situations. Experimental studies showed that, on average, an orientation towards learning goals leads to a better performance than an orientation towards performance goals (cf. the meta-analysis by Utman, 1997). Furthermore, learning goals are typically associated with higher achievement in real-life settings (e.g., Greene and Miller, 1996, Meece and Holt, 1993). Performance-approach goals are also positively related to achievement (e.g., Elliot and McGregor, 1999, Lopez, 1999, Urdan, 2004), whereas performance-avoidance goals are mostly negatively correlated with performance (Elliot and Church, 1997, Elliot et al., 1999, Zusho et al., 2005). Work-avoidance is consistently shown to be negatively correlated with achievement (e.g., Dupeyrat and Marine, 2005, Spinath et al., 2002).

Summarizing the depicted motivational concepts and their relation to performance, the following observations are of special importance. First, all depicted motivational constructs are related to school performance. Second, domain-specifically assessed motivational constructs are typically stronger related to school achievement than domain-generally assessed constructs. The former are moderately to highly associated with school performance whereas the latter are weakly to moderately correlated with academic achievement. Third, even though the utility of motivation for the prediction of achievement-related criteria has frequently been demonstrated, astonishingly few studies have investigated the importance of motivation above and beyond intelligence. This is in so far of significance as some authors question the importance of motivation as a determinant of scholastic achievement (e.g., Gagné & St. Père, 2002). Other authors regard intelligence as the predominant determinant of scholastic achievement (e.g., Kuncel et al., 2004), leaving no or little room for variance explanation by other constructs such as motivation. A common way to prove the utility of a construct is the proof of its incremental validity (cf. LeBreton et al., 2007). It is therefore surprising to note, especially in school contexts, that only few studies have examined whether motivation can predict academic achievement independently of intelligence and which motivational constructs show the highest increment in this regard.

Intelligence is the most valid psychological predictor of academic achievement (e.g., Kuncel et al., 2004). The average correlation of intelligence with school achievement is r = .50 (cf. Gustafsson & Undheim, 1996). The strength of the association depends, among other factors, on the school performance criterion considered. When using standardized achievement tests as achievement criterion, correlations are very high with coefficients ranging from .61 to .90 (e.g., Deary et al., 2007, Frey and Detterman, 2004, Rindermann, 2006). This high correlation can partly be attributed to methodological variance shared by school achievement and intelligence tests. In contrast to standardized scholastic achievement tests, grades given by teachers are composites made up of different types of information. Apart from just being a measure of knowledge they are also a reflection of students' social behavior in the classroom, their motivation, and other aspects. Therefore, intelligence test scores correlate less high with grades. Concerning single grades, the highest correlation is usually the one between intelligence and mathematics (about r = .40; cf. Amthauer, Brocke, Liepmann, & Beauducel, 2001). Compared to general intelligence, domain-specific intelligence is usually more highly correlated with school performance in the corresponding domains (e.g., Amthauer et al., 2001), although this is not always the case (e.g., Spinath, Spinath, Harlaar, & Plomin, 2006).

For some motivational constructs it has been shown that, above and beyond intelligence, motivation explains variance in academic achievement (e.g., Gose et al., 1980, Schicke and Fagan, 1994, Spinath et al., 2006). Addressing academic self-concept, Schicke and Fagan (1994) as well as Gose et al. (1980) demonstrated additional variance explanation independently of intelligence in samples of elementary and middle-school students. Investigating intrinsic motivation, Gottfried (1990) as well as Lloyd and Barenblatt (1984) found an incremental validity beyond intelligence for elementary and high school students, respectively. In contrast to these findings, in a sample of female high school students, a measure of intrinsic motivation did not predict scholastic achievement over intelligence (Gagné & St. Père, 2002). The reasons for these conflicting results are not quite clear. Possible explanations might be a restricted range of intrinsic motivation in the sample from an all-girl high school, psychometric problems of the measurement instrument (the authors mention a lack of construct validity) or the use of a domain-unspecific measure of intrinsic motivation in the study by Gagné and St. Père (2002).

Taken together, most of the reported studies are in favor of an incremental validity of motivation concerning the prediction of academic success. The same is true for a recent study by Spinath et al. (2006). The authors simultaneously investigated intrinsic motivation and academic self-concept. When intelligence and both motivational constructs were entered in a regression analysis with domain-specific academic achievement as the dependent variable, only ability self-concept showed incremental validity predicting performance in English and math. Entered separately, both motivational constructs showed an increment.

The meta-analysis by Robbins et al. (2004) gives further evidence on the incremental power of different motivational constructs. Although the authors did not include any measures of intelligence as predictors, it should be noted that the included cognitive measure, the SAT score, is highly correlated with intelligence (e.g., Frey & Detterman, 2004). Thus, the meta-analysis gives valuable evidence on the importance of motivational measures above a cognitive measure. Achievement motivation, academic goals, and academic self-efficacy were investigated as motivational concepts. All of the motivational concepts were positively related to academic success, with academic self-efficacy showing the highest correlation and academic goals the lowest. The corrected correlation between academic self-efficacy and college GPA was even higher than the corrected correlation between the cognitive predictors and the criterion.

Both studies, Robbins et al. (2004) and Spinath et al. (2006), provide valuable evidence on which motivational construct yields the highest increment over intelligence or a related cognitive measure. Nevertheless, these studies also have some limitations. Even though Robbins et al. (2004) used the ACT they did not include a genuine measure of g when comparing the incremental validity of the investigated motivational constructs. Moreover, academic goals were not further differentiated in learning, performance-approach, performance-avoidance, and work-avoidance goals. Considering the description of academic goals presented in Robbins et al., academic goals are conceptionally close to learning goals. The question whether the other above mentioned goal orientations incrementally contribute to academic achievement remains unsettled. Other important motivational concepts, such as values, were also not considered in the meta-analysis. Furthermore, the meta-analysis only concentrated on studies in secondary education. Spinath et al. (2006) addressed the relative importance of ability self-concept and intrinsic motivation. Goal orientations and achievement motives were not investigated. Furthermore, Spinath et al. (2006) examined elementary-school children aged nine. These results need to be replicated in different age groups.

Moreover, none of the aforementioned studies controlled for prior school performance. If it was shown that, after controlling for prior achievement, motivation still predicted subsequent achievement over and beyond intelligence, this would be especially strong evidence for the importance of motivation (Marsh, Byrne, & Yeung, 1999). Marsh, Trautwein, Lüdtke, Köller, and Baumert (2005) conducted a longitudinal study with two measurement occasions considering grades in math, a mathematical scholastic aptitude test, domain-specific intrinsic motivation, and ability self-concept. When testing all constructs simultaneously in a structural equation model, ability self-concept and performance in the math test still proved to be incrementally valid when controlling for prior grades in math whereas intrinsic motivation did not. These results should be replicated using a measure of intelligence and considering domains beside math.

The present study was designed to simultaneously explore the relative importance of some of the most prominent motivational achievement constructs (achievement motives, goal orientations, ability self-concept, and values) in comparison to intelligence when predicting scholastic achievement. Motivational constructs differ according to their theoretical backgrounds in whether they are domain-general or domain-specific. Domain-specific motivational concepts as well as domain-general ones were tested against domain-specific intelligence measures and against each other when predicting domain-specific school performance (grades in math and German). Furthermore, domain-general motivational concepts were tested against general intelligence when predicting general school performance (GPA). Moreover, we tested the incremental validity of intelligence and different motivational constructs when controlling for prior school performance. Based on the findings reported above, the following hypotheses were derived:

  • (1)

    Achievement motives, goal orientations, ability self-concepts, and values incrementally contribute to the prediction of school performance in math, German, and general school performance beyond intelligence.

  • (2)

    When predicting domain-specific school performance, domain-specifically operationalized motivational concepts show a higher increment than domain-general ones with ability self-concepts showing the highest increment.

  • (3)

    When predicting general school performance, domain-general motivational constructs show a higher increment than when predicting domain-specific school performance.

  • (4)

    When controlling for prior school performance, motivational constructs and intelligence will still explain additional variance in math, German, and in general school performance.

Section snippets

Sample and procedure

The sample was recruited from a German school preparing children for university (Gymnasium).1 The school was located in a mid-sized town and its pupils can be considered as the typical population of this type of school in Germany (i.e., the majority being Caucasian from medium to high socio-economic status homes). In three consecutive cohorts, 342 11th and 12th grade students were tested,

Descriptives and intercorrelations

Means (M), standard deviations (SD), and internal consistencies (α) of all measures are presented in Table 1. Internal consistencies for the cognitive ability test and the self-report measures are mostly good and at least satisfactory.

With regard to our hypotheses, it is important to check for indications of range restrictions in the central variables because such restrictions would influence their predictive power. The mean intelligence score of the sample was about four standard points higher

Discussion

The results of the present study emphasize the general importance of motivation in school contexts. Furthermore, they give evidence on which motivational construct to choose when predicting scholastic performance. In the following, we discuss the implications of the results along the hypotheses and look at some limitations of our study.

References (67)

  • AjzenI. et al.

    Attitude–behavior relations: A theoretical analysis and review of empirical research

    Psychological Bulletin

    (1977)
  • AmthauerR. et al.

    Intelligenz-Struktur-Test 2000 R [Intelligence-Structure-Test 2000 R]

    (2001)
  • BanduraA.

    Self-efficacy: Toward a unifying theory of behavioral change

    Psychological Review

    (1977)
  • BühnerM. et al.

    Is the I-S-T 2000 R Rasch scaleable?

    Diagnostica

    (2006)
  • CovingtonM.V.

    Goal theory, motivation, and school achievement: An integrative review

    Annual Review of Psychology

    (2000)
  • DweckC.S.

    Motivational processes affecting learning

    American Psychologist

    (1986)
  • DweckC.S. et al.

    A social-cognitive approach to motivation and personality

    Psychological Review

    (1988)
  • EcclesJ.S. et al.

    Expectancies, values, and academic behaviors

  • EcclesJ.S. et al.

    In the mind of the actor: The structure of adolescents' achievement task values and expectancy-related beliefs

    Personality and Social Psychology Bulletin

    (1995)
  • EcclesJ.S. et al.

    Motivational beliefs, values, and goals

    Annual Review of Psychology

    (2002)
  • ElliotA.J. et al.

    A hierarchical model of approach and avoidance achievement motivation

    Journal of Personality and Social Psychology

    (1997)
  • ElliotA.J. et al.

    Test anxiety and the hierarchical model of approach and avoidance achievement motivation

    Journal of Personality and Social Psychology

    (1999)
  • ElliotA.J. et al.

    Achievement goals, study strategies, and exam performance: A mediational analysis

    Journal of Educational Psychology

    (1999)
  • FreyM.C. et al.

    Scholastic assessment or g? The relationship between the scholastic assessment test and general cognitive ability

    Psychological Science

    (2004)
  • GagnéF. et al.

    When IQ is controlled, does motivation still predict achievement?

    Intelligence

    (2002)
  • Gjesme, T., and Nygard, R. (1970). Achievement-related motives: Theoretical considerations and construction of a...
  • GoseA. et al.

    The relative potential of self-concept and intelligence as predictors of achievement

    Journal of Psychology

    (1980)
  • GöttertR. et al.

    AMS — Achievement Motives Scale von Gjesme und Nygard — Deutsche Fassung [AMS — German version]

  • GottfriedA.E.

    Academic intrinsic motivation in elementary and junior high school students

    Journal of Educational Psychology

    (1985)
  • GottfriedA.E.

    Academic intrinsic motivation in young elementary school children

    Journal of Educational Psychology

    (1990)
  • GuayF. et al.

    Academic self-concept and academic achievement: Developmental perspectives on their causal ordering

    Journal of Educational Psychology

    (2003)
  • GustafssonJ.E. et al.

    Individual differences in cognitive functions

  • HarackiewiczJ.M. et al.

    Predictors and consequences of achievement goals in the college classroom: Maintaining interest and making the grade

    Journal of Personality and Social Psychology

    (1997)
  • Cited by (0)

    View full text