Achievement Goal Theory at the Crossroads: Old Controversies, Current Challenges, and New Directions

Abstract Achievement goal theory has been one of the most prominent theories of motivation in educational research for more than 25 years. It has undergone considerable revision during that span, most notably with the distinction between approach and avoidance goals, debate concerning the critical features of performance goals, and the emergence of a multiple goal perspective that emphasizes the positive potential of performance-approach goals alongside mastery goals. This multiple goal perspective has met several criticisms from theorists taking the traditional perspective that emphasizes mastery goals over performance goals. We review these criticisms and the ongoing debate in light of the relevant research. We then spotlight two areas for future research, with the aim of advancing theory development and bridging these perspectives. Notes 1Our review of the history of achievement goal theory is necessarily brief. Readers interested in a more thorough review are directed to CitationElliot (2005); Senko, Durik, and CitationHarackiewicz (2008); and Urdan (1997a). 2The few studies showing directrelationships between mastery goals and achievement are far outnumbered by studies showing no direct link. Of course it is possible for mastery goals to improve achievement indirectly by promoting behaviors that do boost performance. Indeed, a small number of other studies found that mastery goals were associated with persistence (J. K. Ford, Smith, Weissbein, Gully, & Salas, 1998; CitationSimons, Dewitte, & Lens, 2004) or deep studying strategies (CitationGreene, Miller, Crowson, Duke, & Akey, 2004; CitationLee, Sheldon, & Turban, 2003; CitationWolters, 1998) and that these behaviors were in turn associated with achievement. However, several others studies found that persistence and deep learning strategies failed to predict achievement (Al-Emadi, 2001; CitationBandalos, Finney, & Geske, 2003; CitationElliot, McGregor, & Gable, 1999; CitationHarackiewicz, Barron, Tauer, & Elliot, 2002; CitationStipek & Gralinki, 1996). 3In this vein, Vansteenkiste, Mouratidis, and Lens (2010) advised exploration of performance goal effects for people pursuing the goal for autonomous reasons versus controlled reasons, and CitationMolden and Dweck (2006) advised exploration of performance goal effects for people with fixed versus malleable views of intelligence. 4Our discussion here has been limited to performance-approach goals. Performance-avoidance goals have almost always been defined in non-normative ways, with an emphasis either on trying to avoid appearing incompetent or on fears about doing poorly (see CitationHulleman et al., 2010). Only recently have theorists begun to define these goals with an emphasis solely on avoiding doing worse than others (CitationElliot & Murayama, 2008). It is too early, therefore, to test whether normative-avoidance versus appearance-avoidance goals yield different effects. 5The size of the normative goal relationship with achievement generally ranges from .10 to .25, a small to medium effect size by conventional standards for correlations (CitationRosenthal, Rosnow, & Rubin, 2000). It is also consistent with a meta-analysis showing that, aside from students’ prior achievement, motivational factors such as goals and self-efficacy remain the strongest predictors of school achievement, above socioeconomic status, learning strategies, and other variables (CitationRobbins et al., 2004). 6By contrast, a negative interaction effect, such that mastery goals are less effective when pursued alongside performance goals, would reveal support for the mastery goal perspective. This pattern has been seldom found in the literature (CitationMidgley et al., 2001). 7Timing is essential with this statistical approach. Several studies (e.g., CitationHarackiewicz et al., 2000; CitationHarackiewicz, Durik, Barron, Linnenbrink, & Tauer, 2008; CitationSenko & Harackiewicz, 2005b; CitationWolters et al., 1996) measured goals at the beginning of the semester (Time 1) and achievement (or perceived competence) at multiple separate points later in the semester, for example, a midterm exam (Time 2) and then final grade (Time 3). It is tempting to treat the Time 2 achievement measure as an ability covariate when testing the goal effects on Time 3 achievement. Linnenbrink-Garcia et al. (2008) took this approach and concluded from their review of seven studies that the relationship between performance goals and grades often disappears when controlling for ability. We believe this approach is misguided, however, because the Time 2 achievement measure is not a valid indicator of baseline ability; statistically controlling it removes not only ability but also any effects that achievement goals have on ongoing achievement. The best baseline measures are those taken prior to the assessment of achievement goals. 8Because college entrance exams (e.g., the SAT) and high school performance (rank or grade point average) rate among the most robust and reliable predictors of college achievement (CitationRobbins et al., 2004), researchers use them as baseline ability measures. Nonetheless, one's prior performance in a similar class(es) or on a similar task might provide an even better index of baseline ability. Fortunately, a few of the studies listed in Table 1 did use measures of pretask performance or cumulative baseline grade point average, and they showed the same effects as those using college entrance exams or high school achievement. So, too, do the studies using baseline competence perceptions, which are likely to be strongly influenced by one's performance history on similar tasks (see Table 2). The consistency across these different baseline measures is impressive and, we believe, allows reasonable confidence in the use of SAT or high school performance as baseline ability measures. Nonetheless, use of various other baseline ability measures would be welcome in future tests of the ability confound hypothesis. 9Though Nicholls (1984) never used the performance-approach and performance-avoidance terminology, his hypothesis clearly captured the basic goal-switching process described here. 10 CitationVan Yperen and Renkema (2008) also tested the effects of competence feedback on goal pursuit. They found that normatively based performance goals were more likely to be chosen after receiving positive instead of negative feedback in two studies. The effects on performance-avoidance goals were less consistent: Positive feedback produced high performance-avoidance goal pursuit in their Study 1 but low pursuit in their Study 2. However, their studies lacked a baseline goal measure and thus were unable to examine change in goal pursuit. 11Murdock and colleagues’ studies showed that evaluative classroom climates, but not students’ personal appearance-based performance goals, predicted tolerance of cheating. 12Although we disagree with Brophy's call to “move on” from performance goals, our focus on mechanism as a new direction is partially inspired by his earlier attention to this topic, in particular at a special session he organized during the 2005 meeting of the American Educational Researcher Association. Prominent theorists on the panel were invited to discuss possible reasons for why performance goals, but not mastery goals, are often unrelated to achievement. 13Three of these studies (Barker et al., 2001; CitationBereby-Meyer & Kaplan, 2005; CitationGraham & Golan, 1991, Study 1) also provided 14 total comparisons between appearance goals and a no-goal control group on the deep learning tasks. Each study observed one case of a disadvantage of appearance goals, yet the remaining 11 deep learning tests showed no difference between these two conditions. However, we recommend some caution in generalizing these findings to the broader debate about normative goal benefits, as each experiment used appearance-based performance goal manipulations that focused on social presentation concerns (i.e., wanting to look smart) rather than normative-based performance goal manipulations.