Task Values 1 Running head: TASK VALUES Task Values, Achievement Goals, and Interest: An Integrative Analysis Chris S. Hulleman University of Wisconsin-Madison Amanda M. Durik Northern Illinois University Shaun A. Schweigert and Judith M. Harackiewicz University of Wisconsin-Madison Hulleman, C. S., Durik, A.M., Schweigert, S., & Harackiewicz, J. M. (2008). Task values, achievement goals, and interest: An Integrative analysis. Journal of Educational Psychology, 100, 398-416.
Task Values 2 Abstract The research presented in this paper integrates three theoretical perspectives in the field of motivation: expectancy-value, achievement goals, and interest. We examined the antecedents (initial interest, achievement goals) and consequences (interest, performance) of task value judgments in two learning contexts: a college classroom and a high school sports camp. The pattern of findings was consistent across both learning contexts. Initial interest and mastery goals predicted subsequent interest, and task values mediated these relationships. Performanceapproach goals and utility value predicted actual performance as indexed by final course grade (classroom) and coach ratings of performance (sports camp). Implications for theories of motivation are discussed.
Task Values 3 Task Values, Achievement Goals, and Interest: An Integrative Analysis The definition of optimal motivation is motivation that produces optimum intellectual development. Though optimum intellectual development is the ultimate goal, progress toward it is assessed in terms of motivation. (Nicholls, 1979, p. 1072) As Nicholls (1979) asserted, motivation and achievement are inherently connected. An understanding of the motivational dynamics at work in achievement settings will therefore allow us to better understand how to promote learning. Motivation is defined here as a motive (e.g., wish, intention, drive) to engage in a specific activity (Austin & Vancouver, 1996; Schiefele, 1999; Weiner, 1985), and can be conceptualized as a behavioral antecedent, a process experienced during task engagement, and as an outcome. Utilizing this multi-faceted conceptualization of motivation enables us to integrate three distinct, yet overlapping, theoretical perspectives: expectancy-value (Eccles, 2005), achievement goal (Dweck, 1986; Nicholls, 1984) and interest theory (Hidi & Renninger, 2006). These three perspectives allow us to understand motivational dynamics in a way that any one single perspective may not completely capture. Interest in activities has been considered to be one of the central components of motivation and motivated behavior (Dewey, 1913; Deci & Ryan, 1985; Schiefele, 1991). One way to develop interest in activities is to find meaning and value in those activities (Hidi & Renninger, 2006; Renninger & Hidi, 2002). Achievement goals and initial interest can predispose individuals to find value in educational activities (Hidi & Harackiewicz, 2000; Pintrich, 2003; Wigfield & Eccles, 2002). Thus, we propose a model of motivation wherein achievement goals and initial interest lead to the perception of task value, which then promotes the development of subsequent interest and learning. In the present paper we examine the
Task Values 4 antecedents and consequences of task value judgments in two learning contexts: a college classroom and a high school sports camp. Task Values Research from the expectancy-value perspective has examined the values that individuals perceive when engaging in tasks, and how these task values are related to subsequent achievement choices. Eccles and her colleagues (Eccles et al., 1983; see Eccles, 2005 for a review) have identified several types of task value that are important in predicting motivation and achievement, in addition to the well-documented effects of success expectancies (Bandura, 1997; Pintrich & Schunk, 2002). Two of these task values are utility and intrinsic value. Tasks with utility value are important because they are useful and relevant beyond the immediate situation, for other tasks or aspects of a person s life. Tasks with intrinsic value are important to the individual because they are enjoyable and fun. Individuals can discover and appreciate the value of activities through interaction and experience. Perceiving utility and/or intrinsic value in tasks has been associated with motivation and interest in activities. For example, both intrinsic and utility value have been found to predict motivational outcomes such as course enrollment decisions (Harackiewicz, Durik, Barron, Linnenbrink, & Tauer, 2007; Meece, Wigfield, & Eccles, 1990; Updegraff, Eccles, Barber, & O Brien, 1996; Wigfield, 1994a), self-reported effort in science classes (Cole, Bergin, & Whittaker, 2006; Mac Iver, Stipek, & Daniels, 1991), intentions to continue a school-based running program (Xiang, Chen, & Bruene, 2005), amount of free-time spent on sports (Eccles & Harold, 1991), and classroom interest (Durik, 2003). Whereas both intrinsic and utility task values have been linked to motivation, utility value may be uniquely related to achievement. For example, Simons, Dewitte, and Lens (2003) found that highlighting the usefulness of an activity by telling participants how it could help them
Task Values 5 achieve their future goals increased persistence and performance in a physical education class. Bong (2001a) found that the perceived usefulness of a course predicted self-efficacy in the course, which in turn predicted exam performance. Malka and Covington (2005) found that the relevance of schoolwork to student s future goals (i.e., perceived instrumentality) predicted classroom performance. These studies indicate that there is a relationship between perceiving utility in a task and subsequent performance. Given the role of intrinsic and utility value in predicting interest and performance, it is essential to understand how individuals come to perceive tasks as intrinsically valuable and useful. The expectancy-value model proposes that goals lead to task values, which are considered to be the more proximal predictors of achievement choices (Eccles et al., 1983; Wigfield & Eccles, 1992, 2002). A large body of research suggests that the achievement goals that students and athletes pursue serve to orient them to tasks (Duda, 1995), influence their cognitive processes and effort during task engagement (Linnenbrink & Pintrich, 2000), and help them find more or less value in what they do (Pintrich, 2003; Wigfield & Eccles, 1992). For example, a student whose goal is to learn and understand course material may be more likely to experience the intrinsic value of the material and see how the course is relevant to his/her life. This enjoyment and personal connection with the material may facilitate attention, cognitive processing, effort, and subsequent interest (Hidi, 1990; Hidi & Harackiewicz, 2000). Achievement Goals According to the achievement goal perspective (Ames, 1984; Dweck, 1986; Nicholls, 1984), goals are mental representations of individuals competence strivings during achievement activities (Maehr, 1989; Shah & Kruglanski, 2000). Although different labels have been used by researchers over the years, achievement goals have been divided into two general classes:
Task Values 6 mastery and performance (Ames & Archer, 1988). Although achievement goals have been subdivided into approach and avoidance components (Elliot, 1999; Elliot & McGregor, 2001; Pintrich, 2000a), we will focus on approach goals because we believe that approach motivation is the most relevant for the discovery of value and development of interest. Mastery-approach goals focus on developing knowledge and learning new skills, whereas performance-approach goals focus on doing well compared to other people. Research has demonstrated that mastery-approach goals positively predict a variety of motivational variables including classroom interest (Ames & Archer, 1988; Harackiewicz, Barron, Pintrich, Elliot, & Thrash, 2002; Lee, Sheldon, & Turban, 2003), as well as deep processing, effort and persistence (Elliot, McGregor, & Gable, 1999; Harackiewicz, Barron, Tauer, Carter, & Elliot, 2000; Midgley, Kaplan, & Middleton, 2001; Wolters, Yu, & Pintrich, 1996). On the other hand, performance-approach goals are positively linked to academic performance (Harackiewicz, Barron, Pintrich et al., 2002; but see Grant & Dweck, 2003). When individuals pursue performance goals, they focus on the outcome of task engagement (i.e., success/failure) and may not become deeply engaged in the activity. In contrast, individuals who pursue mastery goals focus on learning and improvement, and are able to experience and explore the activity because they are focused on the process rather than the product of task engagement (Dewey, 1913; Flum & Kaplan, 2005; Renninger & Hidi, 2002). This task focus may allow them to discover or experience both intrinsic and utility value, which may then motivate effort, persistence, and subsequent interest. From this perspective, achievement goals operate as a framework for the perception of task value, and perceived task value becomes the more proximal predictor of interest and performance. Experimental studies have documented the effects of mastery goals on task involvement and task interest (Senko &
Task Values 7 Harackiewicz, 2005; Zimmerman & Kitsantas, 1997; for a review see Rawsthorne & Elliot, 1999). Other research has demonstrated that mastery-approach goals are correlated with both utility and intrinsic value (Bong, 2001b; DeBacker & Nelson, 1999; Linnenbrink, 2005; Mac Iver et al., 1991; Shim & Ryan, 2005; Simons, Dewitte, & Lens, 2004; Wigfield & Eccles, 2002; Xiang, McBride, & Bruene, 2004). In summary, our hypothesized model suggests that mastery goals create a framework within which individuals focus on the task and perceive value in it. Perceiving both intrinsic and utility value can lead to the development of subsequent interest. Although mastery goals have not been consistently associated with achievement (Harackiewicz, Barron, Pintrich et al., 2002), mastery goal adoption may have indirect effects on performance through perceptions of utility value. Perceived utility value may emerge from deep engagement in an activity (facilitated by a mastery goal), and may motivate attention, effort, and persistence that result in higher levels of performance. The Development of Interest How does the perception of value influence the development of interest? Interest can be defined as a psychological state, e.g. being engaged, engrossed, or entirely taken up with some activity (Dewey, 1913, p. 17), and as a process that emerges over time (Hidi & Renninger, 2006). The psychological state of interest that arises through interactions with task features, such as stimulating pictures and humor, is known as situational interest (Hidi, 1990). Interest that is more enduring and trait-like, and that develops and deepens over time, is known is individual interest (Renninger, 2000). Perceiving value in activities has been hypothesized to be a key contributor in the progression from situational to individual interest, and to the deepening of existing individual interest (Hidi & Renninger, 2006). As momentary interest in the situation is
Task Values 8 activated, an individual may come to perceive value in the activity and desire to continue pursuing the activity over time, resulting in maintained situational interest. If situational interest is maintained and an individual continues to engage in the activity and perceive value, then individual interest may develop (emerging individual interest). Thus, perceived value may play a critical role in the beginning stages of interest development as well as the deepening of individual interest over time. Integration of Motivational Perspectives Although we have conceptualized value and interest as distinct constructs, many theorists consider value to be an integral component of interest (Dewey, 1913; Hidi & Harackiewicz, 2000; Hidi & Renninger, 2006; Krapp, 2002; Mitchell, 1993; Schiefele, 1991). We agree, but believe these constructs can be distinguished over time. In particular we are interested in how the perception of task value early in a learning experience influences the development of subsequent interest. For example, students might find a specific lecture or reading assignment in a psychology class to be enjoyable (intrinsic value) or relevant to their lives (utility value), and this perception of value could then lead to the development of subsequent interest in psychology. In terms of Hidi and Renninger s four-phase model of interest development (2006), perceived task values would correspond to both the triggering of situational interest and the early stages of maintained situational interest, whereas subsequent interest in psychology would correspond to the latter stages of maintained situational interest or the emergence of individual interest. This conceptualization of task values as situational interest is not inconsistent with Eccles perspective; rather, it is simply a more specific formulation that allows us to integrate these theoretical perspectives. In the Eccles et al. model of task values, intrinsic and utility task values could be considered as more general beliefs or as situation-specific reactions to the task. In this
Task Values 9 paper we consider task values to be situation-specific predictors of subsequent interest and performance; whereas interest refers to more general beliefs about the activity over time. This important distinction allows us to consider perceived task value and subsequent interest as separate outcomes, and to evaluate how the emergence of value during task engagement contributes to the development of subsequent interest. This model of interest development is supported by our previous theoretical (Hidi & Harackiewicz, 2000) and empirical work (Harackiewicz et al., 2007), and requires that we consider the initial level of interest that students and athletes bring into learning situations. Hidi and Renninger s (2006) four-phase model of interest development outlines how interest develops and deepens in a situation depending upon the value, positive affect, and knowledge experienced with an activity. Individuals who have different levels of experience and knowledge of an activity may enter a situation with different levels of initial interest. Thus, in order for us to assess the role of value in interest development, we will need to consider initial levels of interest in the activity (Hidi & Renninger, 2006; Maxwell & Cole, 2007). The integration of achievement goal, interest, and expectancy-value models of motivation complements the previously established notion that there can be multiple pathways to optimal motivation (Barron & Harackiewicz, 2001; Harackiewicz, Barron, Pintrich et al., 2002; Pintrich, 2000b, 2003). For example, achievement goal researchers have emphasized the importance of testing the independent and interactive effects of achievement goals on outcomes, and identified several patterns of multiple goal effects (c.f., Harackiewicz, Barron, Pintrich et al., 2002). In this paper we will also assess the indirect effects of achievement goals through task values. This analysis complements the multiple goals perspective in that it provides additional pathways through which achievement goals might promote interest and performance. Although not
Task Values 10 exhaustive, our theoretical integration incorporates three related perspectives in which we have conducted prior research, and as such is an initial step toward theoretical synthesis. For example, our framework is generally consistent with other motivational models, such as those that propose dual-regulation systems (Boekaerts, 2003; Boekaerts & Corno, 2005; Heckhausen, 2000; Krapp, 2002, 2005; Kuhl, 2001; Schiefele, 1991). These models suggest that behavior is regulated by both cognitive-rational and emotional-feeling subsystems, which parallels our explication of utility and intrinsic value as two components of situational interest. Current Research In two studies, we investigated the predictors and consequences of task value in classroom and sports contexts. We were primarily interested in: 1) the role of task value in interest development and performance, and 2) the role of achievement goals in creating a framework for the perception of task value. In Study 1, we examined college students perceptions of task values in an introductory psychology course, as well as their achievement goals, as predictors of their subsequent interest in the topic and academic performance. In Study 2, we examined high school athletes perceptions of task values at a summer football camp, as well as their achievement goals, as predictors of their subsequent interest and performance at camp. In both studies we controlled for initial interest in the activity. College classes and sports camps are similar in terms of their educational mission and definition of achievement. In both settings there is a focus on learning: Students learn lessons, do homework, and take tests, whereas athletes learn techniques, practice and do drill work, and play competitive games. For some individuals, sports may be more interesting than academics (and vice versa), but both can involve long hours of homework and repetitive drills. Success in both domains can require the ability to practice and work through difficult or boring activities. Given
Task Values 11 these similarities, theories of achievement motivation and interest should be applicable to both domains. There is also an important difference between sports and school that should not be overlooked, particularly in terms of interest development: Most athletes have chosen to participate in their sport, whereas students often have little or no choice in which classes they take, particularly when fulfilling academic requirements. Although the decision to take an introductory course may indicate a certain level of initial interest in the topic, most students are unlikely to have had extensive experience with the topic. The introductory course therefore provides the opportunity to examine the emergence and development of interest in its early phases (Hidi & Renninger, 2006). In contrast, athletes willing to attend a summer camp at their own expense typically have extensive experience with the sport and a well-developed interest in it. Thus, the high school summer sports camp provides the opportunity to examine the role of situational interest in deepening already well-developed interest. Study 1 The College Classroom We investigated the role of task values and achievement goals in predicting subsequent interest and performance in a college classroom, controlling for initial levels of interest. Based on previous findings (Harackiewicz, Barron, Pintrich, et al., 2002; Wigfield & Eccles, 2002), we hypothesized that mastery goals would predict the perception of value and the development of subsequent interest in the course, and that performance-approach goals would predict course grades. Including a measure of initial interest allowed us to clarify the relationship between mastery goals, task values, and the development of subsequent interest. In addition, we hoped to explore relationships between values and performance by examining utility and intrinsic values
Task Values 12 separately. Specifically, we hypothesized that perceiving utility value in the course might also predict performance. Method Overview This study took place during the course of a semester at a large, Midwestern university and consisted of three waves of data collection during the semester. Students initial interest in the course topic and achievement goals were assessed on the second day of class (Wave 1); their perceptions of course value (intrinsic and utility) were measured several weeks into the semester, but before the first exam (Wave 2); and their interest in the course was measured during the last week of class (Wave 3). We also obtained students final course grades from department records. Participants and Setting Participants were recruited from four sections of Introductory Psychology classes (approximately 350 students per section). Only students who were taking the course for graded credit were included in the sample. Classes were almost entirely lecture format. Students grades were determined by their performance on several multiple-choice exams given throughout the semester. Final grades were assigned based upon a normative curve recommended by the Psychology Department. Wave 1 data collection was part of a departmental-wide survey (1145 participants) for which students received extra credit in exchange for participation. Participants enrolled in our study at Time 2 (830 participants) when they completed consent forms and the accompanying survey during class time. Students were not compensated for their participation in the second and third waves. Some attrition did occur after enrollment in our study at Wave 2. First, 35 participants dropped the course before the end of the semester. Second, because data were
Task Values 13 collected during class time, students who were absent at Wave 3 (n = 132) were not included in the study. Thus, the final sample used in data analysis included 663 students (215 males and 448 females) and represents 79.9% of the students who enrolled in our study at Wave 2. Additionally, two participants who received a grade of incomplete were not included in the course grade analysis. Measures Initial measures. During the second meeting of the course, after students had become familiar with the course syllabus and materials, students completed a questionnaire that contained items assessing their initial interest in psychology, and their achievement goals for the semester. Initial interest in psychology was assessed with two items ( I think psychology is a very interesting subject, I don t think psychology is a very interesting subject (reversed) ; α =.78). Mastery-approach goals were assessed with two items ( My goal is to learn as much as possible about psychology, I want to develop an understanding of psychology ; α =.80). Performance-approach goals were assessed with two items ( I would like to do better than other students in this class, My goal is to get a better grade than other students in this class ; α =.75). The initial interest and achievement goal items were based on prior research (Harackiewicz, Barron, Tauer, & Elliot, 2002; Harackiewicz et al., 2006). Participants responded to all selfreport items on a scale from 1 strongly disagree to 7 strongly agree. Mid-semester measures. Four weeks into the course, and about one week prior to the first exam, students completed a questionnaire that contained items assessing their perceived values in the course and their interest in the course. Utility value was assessed with three items ( What I am learning in this class is relevant to my life, The topics in this class are important for my career, In general, material from this class is not useful to me (reversed) ; α =.72). Intrinsic
Task Values 14 value was assessed with three items ( Lectures in this class are entertaining, Lectures in this class drag on forever (reversed), I enjoy coming to lecture ; α =.82). Interest in the course was assessed with the same two items used for the initial measure of interest (α =.78). Subsequent interest. At the end of the semester, but prior to the final exam, students final interest in the course was assessed. In addition to the items used to measure initial interest, the measure of subsequent interest included two behavioral inclination items ( I would like to take more psychology courses, My experience in this course has made me want to take more psychology courses ; α =.87). Final grade. Students final course grades were obtained from departmental records. Each student could receive one of seven possible grades, based on the university s 4-point scale. The average grade in this study was 2.92 (SD =.84). Grades were distributed as follows: A = 21.5%, AB = 19.5%, B = 16%, BC = 17.4%, C = 20.7%, D = 4.7%, and F = 0.2%. Two students in our sample received an incomplete grade for the semester, but were retained in the interest analysis. The average grade for students in our sample was slightly higher than for all students in Introductory Psychology (M = 2.70, SD =.97, N = 1267), but the distribution of grades was similar. Results Preliminary Analysis Attrition. We conducted an analysis that compared the final sample (N = 663) to those individuals who enrolled in our study at Time 2 but who were not included in the final sample because they either dropped the course or failed to attend class on the day of data collection (N = 167). Comparisons were made on all the major variables in the study: initial interest, mastery-
Task Values 15 approach and performance-approach goals, intrinsic and utility value, subsequent interest, and final grade. Independent samples t-tests revealed two significant differences. Effect sizes (Cohen s d) were computed for these differences, revealing small to large effects. The final sample had higher scores on mastery-approach goals, t(825) = -2.57, p =.01, d =.24, and course grade, t(793) = -7.28, p <.01, d =.82. Descriptive and correlational analyses. Means, standard deviations, correlations, and scale reliabilities for the variables in this study are presented in Table 1. Although the structure and content and grading distributions of the four classes were comparable, we tested for instructor differences on all variables. There were significant instructor effects on intrinsic value, utility value, and subsequent interest, indicating that students reported higher levels of intrinsic value, utility value, and subsequent interest with some instructors compared with others. Therefore, we included a set of three dummy-code terms to test and control for mean-level differences between the instructors in all subsequent analyses (Cohen, Cohen, West, & Aiken, 2003). Missing data. It is possible that the attrition that occurred in this data set could be influencing the findings in a myriad of ways. A complete account of causality and interpretation issues due to missing data can be found elsewhere (Rubin, 1976). The number of people with available data is presented for each variable in Table 1. We created a dichotomous variable reflecting whether individuals had complete data across all study variables (coded 0) or had incomplete data on at least one variable (coded 1). This variable is included in Table 1. From the correlations between this variable and the other variables we can determine the extent to which incomplete data is related to other variables in the study. The bottom row of correlations shows
Task Values 16 students with a lower final grade, lower initial interest, and lower mastery goals were more likely to have incomplete data. These associations indicate that the incomplete data are not missing completely at random, because if they were, then the missing pattern would not be related to any variables in the dataset (Rubin, 1976). In addition, these associations enabled us to create a new data set that accounted for at least some of the variability in the missing data. One approach to missing or incomplete data is to create imputed data sets that model what the data may have looked like if none of the data were missing. We used Mplus version 3.01, using full information maximum likelihood, to impute a complete data set that includes values on each variable for each individual (Muthén & Muthén, 2004). The results of that analysis did not diverge from the analysis of only the originally complete data. Therefore, in this paper we will report only the analyses using the original data set that did not account for missing data. A complete report of the maximum likelihood analysis is available from the first author upon request. Scale construction. To test our hypothesis that the interest, utility value, and intrinsic value items would form separate factors, we conducted two types of factor analyses. First, in order to examine the factor loadings on our hypothesized three-factor model, we conducted an exploratory factor analysis of the items used to measure subsequent interest, utility value, and intrinsic value using a separate sample of undergraduates. As a part of a larger study on student attitudes towards learning, participants (N = 188) were recruited from four sections of an introductory statistics course, and completed a brief survey during class time. The items and the pattern matrix of factor loadings are presented in Table 2. Each item loaded onto its respective factor (all values >.678), and there was an absence of cross loadings onto other factors (all cross loadings <.300).
Task Values 17 Second, in order to test how well our model fit the data, we conducted a confirmatory factor analysis (CFA) using LISREL 8.72 (Jöreskog & Sörbom, 1998) on the current sample of introductory to psychology students (N = 663). Five nested models were tested. The three factor model which separated interest, intrinsic value, and utility value was tested against all other possible factor groupings: 1) a single factor that combined interest, intrinsic and utility value items, 2) two factors: one interest factor and one combined intrinsic/utility value factor, 3) two factors: one utility value factor and one combined interest/intrinsic value factor, and 4) two factors: one intrinsic value factor and one combined interest/utility value factor. For each model, we calculated multiple indices of fit: chi-square, comparative fit index (CFI), standardized root mean square residual (SRMR), and root-mean-square error of approximation (RMSEA). The results of the five CFA s are presented in Table 3. The hypothesized model with distinct interest, utility and intrinsic value factors provided a better fit than any of the other models. The worst fitting model was the one that combined all three factors into one combined factor. Regression Analyses We used hierarchical multiple regression to analyze these data in three stages. First, we investigated the antecedents of achievement goals. Second, we examined the direct effects of task values, achievement goals, and initial interest on subsequent interest and course grade. Finally, we tested the indirect effects of all variables on subsequent interest and course grade. For all analyses, all continuous variables were standardized, and all two- and three-way interaction terms were computed (Aiken & West, 1991). Significant interaction effects were interpreted in two ways. First, we computed predicted values for representative high and low groups (one standard deviation above and below the mean) from the regression equations using the unstandardized regression coefficients. Second,
Task Values 18 we calculated simple slope values for representative high and low groups from the regression equations and tested them according to Aiken and West (1991). Predictors of achievement goals. The first set of analyses examined the antecedents of achievement goals. The achievement goal model contained the three instructor dummy codes, initial interest, and an effects code for gender (male = -1, female = +1). This 5-term achievement goal model was used to predict both mastery- and performance-approach goals. Initial interest was the only significant predictor for both mastery-approach, t(662) = 19.69, p <.01, (β =.61) and performance-approach goals, t(662) = 2.64, p =.01, (β =.11). Individuals with higher levels of initial interest reported higher levels of mastery- and performance-approach goals. Direct effects models. The second set of analyses tested the direct effects of initial interest, mastery- and performance-approach goals, and all of their two- and three-way interactions on subsequent interest and final grade. We also included the gender effects code and the three instructor dummy codes. Preliminary testing indicated that none of the three-way interactions attained significance on any measure and they were dropped from all subsequent analyses. All two-way interactions were retained in the models. Therefore, the direct effects model contained ten terms: the seven main effect terms as well as three two-way interactions (mastery-approach X performance-approach, mastery-approach X initial interest, performanceapproach X initial interest). Direct effects on subsequent interest. There was a significant main effect of initial interest, t(652) = 9.86, p <.01, (β =.46), such that students who entered the course with higher levels of initial interest reported more interest in psychology at the end of the semester. The main effect of mastery-approach goals, t(652) = 3.60, p <.01, (β =.17), indicated that students with higher levels of mastery goals at the beginning of the course were more interested in psychology.
Task Values 19 These two main effects were qualified by the significant interaction between mastery-approach goals and initial interest, t(652) = 5.35, p <.01, (β =.25), which indicated that mastery-approach goals were a stronger predictor of final interest in psychology for those students with high levels of initial interest (β =.42), than for those with low levels of initial interest (β =.01). This interaction is graphed in the top panel of Figure 1. Direct effects on final grade. There was a significant main effect of performanceapproach goals, t(650) = 4.56, p <.01, (β =.18), such that students with higher levels of performance-approach goals obtained higher final course grades. The direct effects path model for final interest and final grade is presented in Figure 2. Although we tested for the effects of initial interest in all analyses, we did not include it in the path diagrams to enhance parsimony. Mediated (and indirect effects) models. The third set of analyses tested the role of perceived task values as mediators or indirect paths between initial interest and goals and the final outcome measures. First, we tested whether mastery goals and initial interest were significantly related to task values by using the direct effects regression model to predict intrinsic and utility value. Second, we tested whether the mediators were related to the outcomes, while controlling for the original predictors, by including task values in the direct effects model. Therefore, we added intrinsic and utility value to the direct effects model, as well as all possible two- and three-way interactions between values, initial interest, and goals. For each set of analyses, only significant interactions were retained in the model. Thus, unless otherwise noted, the mediation model contained 12 terms: 10 terms from the direct effects model with the addition of the two value terms. Finally, to test the significance of mediated and indirect effects, we followed procedures outlined by Kenny, Kashy, & Bolger (1998) in which the indirect path is
Task Values 20 tested by creating a product term and dividing by its standard error. This technique has been shown to be robust to Type I errors (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002). Direct effects on intrinsic value. The significant main effects of initial interest, t(652) = 3.13, p <.01, (β =.16), and mastery-approach goals, t(652) = 2.53, p =.01 (β =.13), indicated that students with higher levels of initial interest and mastery-approach goals were more likely to find intrinsic value in the course topic. These main effects were qualified by their marginally significant interaction, t(652) = 2.08, p =.038, (β =.11), which indicated that mastery-approach goals led to more intrinsic value for those students with high levels of initial interest (β =.24), but not for those with low levels of initial interest (β =.02). This interaction is presented in the middle panel of Figure 1. Direct effects on utility value. The significant main effects of initial interest, t(652) = 6.32, p <.01, (β =.30), and mastery-approach goals, t(652) = 6.68, p <.01, (β =.32), indicated that students with higher levels of initial interest and mastery-approach goals were more likely to find utility value in the course topic. These main effects were qualified by their significant interaction, t(652) = 5.86, p <.01, (β =.28), which indicated that initial interest led to more utility value for those students with high levels of initial interest (β =.60), but not for those with low levels of initial interest (β =.04). This interaction is presented in the lower panel of Figure 1. Mediated and indirect effects on subsequent interest. In the mediation model, the significant main effects of intrinsic value, t(650) = 3.84, p <.01, (β =.14), and utility value, t(650) = 6.33, p <.01, (β =.25), indicated that students who perceived each type of value in the course reported more interest in psychology at the end of the semester. The main effect of initial interest was also significant, t(650) = 8.03, p <.01, (β =.36), but reduced in size from the direct
Task Values 21 effects model (β =.46). The formal test of mediation revealed that the effect of interest was partially mediated by both intrinsic value, z = 2.43 (p =.02), and utility value, z = 4.47 (p <.01). In addition, the mastery-approach goal main effect was no longer significant (β =.07, p =.12) and reduced in size from the direct effects model (β =.17). The formal test of mediation revealed that the effect of mastery goals was partially mediated through both intrinsic, z = 2.11 (p =.03), and utility value, z = 4.49 (p <.01). The complete path model is presented in Figure 3. The interaction between initial interest and mastery goals remained significant in the mediated model, t(650) = 3.68, p <.01, (β =.17), but reduced in size from the direct effects model (β =.25). The formal test of mediation revealed that this interaction effect was partially mediated through both intrinsic value, z = 1.83 (p =.07), and utility value, z = 4.30 (p <.01). Mediated and indirect effects on final grade. In the mediated model, the significant main effect of performance-approach goals was unchanged from the direct effects model, t(648) = 4.44, p <.01, (β =.18). However, there was a significant main effect of utility value, t(648) = 3.94, p <.01, (β =.19), such that students who perceived higher levels of utility value in psychology received higher grades than students who perceived lower levels of utility value. Although initial interest and mastery-approach goals did not directly predict final grade, there were significant indirect paths from these variables to final grade through utility value, z = 3.34 (p <.01) and z = 3.39 (p <.01), respectively. The complete path model for final interest and final grade is presented in Figure 3. Structural Equation Model In order to test the overall fit of the model, Lisrel 8.72 was used to simultaneously test the model presented in Figure 3. The results indicated that the overall model provided a satisfactory fit to the data, χ 2 (80) = 309.44, p <.001, RMSEA =.07, CFI =.97, SRMR =.07.
Task Values 22 Discussion Study 1 The results of this study suggest that perceptions of task value play an important role in the development of subsequent interest, and are also associated with academic performance. Adopting mastery-approach goals was associated with the perception of utility and intrinsic value in the coursework, particularly for students already high in initial interest, and these task values were in turn associated with the development of subsequent interest. Perceptions of utility value were also positively associated with final grades. Taken together, these results highlight the important role of mastery goals in optimal motivation as they promote the perception of value. This study replicated prior work on achievement goal effects in the college classroom and demonstrated that mastery-approach goals predicted subsequent interest in the course whereas performance-approach goals predicted higher course grades (Harackiewicz, Barron, Pintrich, et al., 2002). In addition, this study extends prior research in several important ways. First, controlling for initial interest and achievement goals, utility and intrinsic task values were unique predictors of subsequent interest. In addition, the effects of both initial interest and masteryapproach goals on subsequent interest were mediated through the value students perceived in the course. Initial interest and mastery-approach goals predicted utility and intrinsic value in the course material, which predicted subsequent interest, supporting the four-phase model of interest development (Hidi & Renninger,2006). Second, perceiving utility value in the course was positively associated with higher course grades. This finding represented an indirect pathway through which initial interest and mastery-approach goals positively influenced academic performance. It is important to note that our results are not informative about the processes whereby task values were associated with
Task Values 23 subsequent interest and performance. Perhaps finding the material useful led students to make more meaningful connections between themselves and the material, which allowed them to learn the material more thoroughly, remember it better (Markus, 1977), and put more effort into the course (Mac Iver et al., 1991). Future research will need to explicate the mechanisms of the task value effects. Third, the effects of mastery-approach goals on subsequent interest remained significant even when controlling for initial interest. In fact, despite being highly correlated (r =.61, p <.01), initial interest and mastery-approach goals had unique, independent effects on subsequent interest. In addition, the interaction between initial interest and mastery-approach goals revealed that mastery-approach goals were strongly predictive of subsequent interest only for students who were high in initial interest. In contrast, mastery-approach goals were unrelated to subsequent interest for individuals who entered the class with low levels of initial interest. Thus, the combination of initial interest and mastery-approach goals was particularly powerful in predicting task values and subsequent interest. In Study 2, we sought to extend our analyses to a different achievement context, and we investigated task values and achievement goals at a sports camp. Participants were high school football players who participated in a summer football camp designed to prepare them for the upcoming season. In contrast to introductory psychology students in Study 1, the football players in Study 2 were more likely to have extensive experience playing football and thus entered the camp situation with more well-developed interest. This allowed us to investigate a proposition of Hidi & Renninger s model regarding the deepening of already well-developed individual interest. According to their model, well-developed individual interest is maintained and deepened through continued interaction with the activity: By re-engaging in the activity,
Task Values 24 individuals with well-developed interests can experience new-found situational interest through knowledge and skill acquisition, finding new meaning and value, and experiencing positive affect - and thereby deepen their interest in the activity. In Study 2 we investigated the first step in this process by examining the development of situational interest in a domain where individuals begin with a more well-developed individual interest. Study 2 Sports Camp This study investigated the role of task values and achievement goals in predicting subsequent interest and performance at a summer football camp for high school boys, controlling for initial interest. High school sports camps provide opportunities for athletes to engage in their sport outside of the competitive season and to develop and extend their existing skills. These camps tend to attract individuals with high levels of experience and interest in the sport. Indeed, not only did our athletes report being interested in football (M = 5.54, SD = 0.78), but they also had been playing the sport for several years (M = 4.26, SD = 1.93). Those campers who get the most out of camp are likely to leave camp with new skills as well as with a sense of enjoyment and satisfaction, having optimized their camp experience. Thus, our primary outcome measure in this study was camp satisfaction, reflecting situational interest in the camp experience. This study extends prior achievement goal research by examining the influence of both types of approach achievement goals in the sports context. Prior research in sports contexts has assessed mastery-approach goals, but performance goals have typically been assessed as a combination of approach and avoidance motivation (Elliot, 2005) or within the ego-orientation perspective (Duda, 1995). Based upon prior research in the classroom, we predicted that mastery-approach goals would predict satisfaction in a sports camp, whereas performanceapproach goals would predict performance at camp. We also hypothesized that mastery goals
Task Values 25 would predict both intrinsic and utility task values. Consistent with the results of Study 1, we hypothesized that intrinsic value would predict camp satisfaction and that utility value would predict both camp satisfaction and performance. In addition, because this research extends beyond the classroom, we can speculate about how achievement goal effects might differ in this particular sports context. Sporting contexts can differ from classrooms in that competition, rewards, and performance are inherent to the activity (Hidi, 2000; Kruglanski, 1975), and these components of the activity might enhance subsequent interest and intrinsic motivation (Hidi & Harackiewicz, 2000). Therefore, it is possible that performance-approach goals might positively predict camp satisfaction. However, it is also possible that mastery-approach goals might predict performance in this context. This is because football camp occurs outside of the regular competitive season and skill development could be more salient than normative performance at football camp, and more salient than in the classroom. Thus, this particular sporting context, where skill development is an inherent part of the purpose of the activity, could provide the opportunity for achievement goals to work in ways that are different from the college classroom and reveal multiple pathways for optimal motivation (Barron & Harackiewicz, 2001; Hidi & Harackiewicz, 2000; Pintrich, 2000b). Method Overview This study took place during a summer football camp for high school boys in the Midwestern United States. Participants initial interest in football and achievement goals were assessed on the first day of the camp (Wave 1), their task values for specific football drills were assessed during the middle of the camp (Wave 2), and their interest in camp was assessed at the
Task Values 26 conclusion of camp (Wave 3). Our dependent measure of performance consisted of coaches ratings of campers effort and performance at camp. Participants and Setting Participants were recruited from two summer football camps for high school boys (approximately 200 campers per camp). Participants were entering the 8 th to 12 th grade in the coming fall (Grade-level of participants: 8 th = 0.8%; 9 th = 14.3%, 10 th = 24.4%, 11 th = 26.9%, 12 th = 33.6%). The football camps were either three- or four-day over-night camps that focused on teaching players football-relevant skills. Each day of camp included two or three practice sessions that were focused on skill development, allowing only minimal opportunity to scrimmage or play competitive games. Camp drills were divided into two categories: technique drills and competitive drills. Technique drills were defined as structured practice drills in which players were working individually on form or footwork, and not competing against someone. Examples of technique drills are form running, footwork drills, running routes, or pushing a blocking sled. Competitive drills were defined as structured practice drills in which players were competing against others, but not actually scrimmaging as an entire team. Examples of competitive drills are 1-on-1 passing and blocking drills. Participants enrolled in our study when they completed the Wave 1 consent form and questionnaire on the first day of camp (237 participants). Players were not compensated for their participation in any of the surveys. Of the original 237 participants, 198 completed the Wave 2 measures, and 155 completed the Wave 3 measures. The participant attrition was primarily due to players needing to leave camp early due to schedule conflicts or participation in other summer sports events. Because the coach ratings were not dependent upon participant attrition, the final
Task Values 27 sample for coach ratings was higher (n = 198, 84% of the original sample) than that for camp interest (n = 155, 65% of the original sample). Measures Initial interest and achievement goals. Participants were mailed an initial questionnaire to complete and return when they checked into camp. This measure assessed participants interest in football as well as mastery- and performance-approach goals for football camp. We adapted the initial interest measures for use in the football context from the final interest measures used in Harackiewicz, Barron, Tauer et al. (2002), e.g. Football is my favorite sport to play. The specific items in this six-item scale had good reliability (α =.78) and can be found in the Appendix. Mastery-approach goal items focused on skill development at camp ( I want to develop my football skills at this camp, I want to learn as much as possible at this camp, I don t really care how much I learn at this camp (reversed) ; α =.72). Performance-approach goal items focused on normative comparisons at camp ( It is important for me to do well compared to others at this camp, My goal is to be one of the best football players at this camp, I would like to do better than other players at this camp, I don t care how I do compared to other players at this camp (reversed) ; α =.79). These items were adapted for the camp context based on Harackiewicz, Barron, Tauer et al., (2002) and Conroy, Elliot, & Hofer (2003). All self-report measures in this study used a seven-point Likert scale that ranged from 1 strongly disagree to 7 strongly agree. Utility and intrinsic value. Approximately half-way through the camp participants completed a ten-item questionnaire that contained the task value measures. These value items were based upon the model of values proposed by Eccles and Harold (1991) and assessed participants self-reported utility and intrinsic values of specific drills at camp. The six-item