Examining Goal Orientation 1 Running head: GOAL ORIENTATIONS AMONG COLLEGE MAJORS Examining Goal Orientation Similarities and Differences Among College Majors: An HLM Analysis

This study employed hierarchical linear modeling (HLM) to explore differences among college majors on five goals orientations. Specifically, measures of mastery-approach, performance-approach, mastery-avoidance, performance-avoidance, and work-avoidance goals were collected from a sample of 1889 college students representing 26 different majors. Statistically significant differences existed among majors in mastery-approach, performance-approach, and work-avoidance, although measures of practical significance indicated that the differences among majors were small. Moreover, once gender and SAT scores were controlled for, differences among majors in mastery-approach and workavoidance disappeared. Only significant differences in performance-approach among majors remained when controlling for other variables. These results provide initial information for better understanding differences in goal pursuit that may be a function of contextual influences (i.e., the academic environment created by one’s major) or characteristics of students who self-select into certain majors. Examining Goal Orientation 3 Examining Achievement Goal Similarities and Differences Among College Majors: An HLM Analysis The purpose of the present study was to examine similarities and differences in the types of goal orientation being pursued by college students majoring in a variety of different disciplines (e.g., Nursing, Computer Science, Psychology, English). Specifically, we investigated the pursuit of five different goal orientations currently being discussed in the literature: mastery-approach, mastery-avoidance, performance-approach, performance-avoidance, and work-avoidance goals. Within the framework of goal orientation, researchers have often focused on examining differences in goal pursuit between groups. For instance, some studies examined differences between the goal orientations of males and females (e.g., Elliot & McGregor, 2001; Finney & Davis, 2003; Patrick, Ryan, & Pintrich, 2002; Roeser, Midgley, & Urdan, 1996), while other studies examined differences in goal orientations among ethnic groups (e.g., Kaplan & Maehr, 1999; Pastor & RiCharde, 2003). The findings from such studies can help identify demographic groups that may be adopting less advantageous goal orientations. In addition, contextual effects of the classroom and school environment have been examined to determine their influence on goal orientation. Specifically, studies have considered the environmental factors of classrooms in different subject domains using students in elementary schools (e.g., Midgley, 2002), middle schools (e.g., Anderman & Anderman, 1999; Midgley, 2002; Roeser, et al., 1996), and high schools (e.g., Midgley, 2002). Similarly, Barron, Harackiewicz, and colleagues (e.g., Barron & Harackiewicz, in press; Harackiewicz, Barron, Carter, Lehto, & Elliot, 1997) have explored differences in Examining Goal Orientation 4 achievement goals of college students across contexts within the same subject area (e.g., introductory vs. advanced psychology courses). When group differences are identified, researchers are faced with the challenge of trying to determine why such differences exist. For instance, if in the present study, differences are found in the types of goals adopted by students in different majors, we are forced to ask to what extent are the differences due to characteristics of the person and/or characteristics of the environment? An explanation using a characteristic of the person to explain the difference might suggest that students are inclined to seek out environments which promote the same goals that they endorse. In this sense, goal orientation may be a cause for which major a student selects. In contrast, an explanation using a characteristic of the environment to explain the difference would be that students do not select their major based on their goal orientation, rather, their goal orientation is shaped by whatever goal orientation is predominant and adaptive in their new environment (major). In this sense, the major (i.e., the climate promoted by faculty and other students in that major) may be the cause for a student adopting a particular goal orientation. The first possible explanation of differences between majors implies that person variable influences on one’s goal orientation should be considered. A search of the Internet results in hundreds of websites devoted to helping students identify which college major is best suited for them. These websites suggest that students consider their interests, abilities, and values. One website (Hansen, n.d.) uses an instrument in which students must identify the values that they feel are most important in a future career. For example, individuals must determine if they desire a career that involves mentally challenging tasks, professional development, on-going learning and growth, recognition Examining Goal Orientation 5 for quality of work in a public way, or obtainment of social status. Within these example values we can see signs of qualities that are reflected within the framework of achievement goals. Dweck (1986, p. 1040) describes persons adopting learning (or mastery) goals as those who “seek to increase their competence, to understand or master something new” and persons adopting performance goals as those who “seek to gain favorable judgments of their competence or avoid negative judgments of their competence.” If one’s goal orientation drives the selection of their college major, students that are mastery-oriented would choose majors that lead to careers characterized by challenging tasks and on-going learning and growth, whereas students that are performance-oriented would select majors that lead to careers promising social status or recognition of their work in a public way. There are other types of personal characteristics that may help predict differences among majors in goals orientation. Given that certain majors are traditionally considered to be more “female-oriented” or “male-oriented”, research on gender differences in goal orientation should be considered. For example, previous research has established that college-aged females more strongly endorse mastery-approach goals than males (e.g., Bouffard, Boisvert, Vezeau, & Larouche, 1999; Elliot & McGregor, 2001; Finney & Davis, 2003), and adolescent males usually report higher levels of performance-approach goals than females (e.g., Middleton & Midgley, 1997; Midgley & Urdan, 1995; Patrick et al., 1999; Roeser et al., 1996; Ryan, Hicks, & Midgley, 1997). Therefore, it would be expected that if a major was predominantly populated by females, the overall major may appear to be higher in mastery-approach, not due to a characteristic of the major, but rather a characteristic of its members. Examining Goal Orientation 6 In addition to gender, the influence of prior ability must also be considered. For example, when students apply to college, admittance into their major of choice may require a given level of ability as demonstrated by high school grades or SAT scores. Therefore, students may be restricted by their choice in major due to ability. Elliot and McGregor (2001) found that SAT scores positively predicted performance-approach goals and negatively predicted performance–avoidance goals in college students. Therefore, those majors that have higher requirements for admittance would be expected to have students with higher levels of performance-approach and lower levels of performance-avoidance goals. The second possible explanation for differences between majors implies that contextual or environmental influences on one’s goal orientation should also be considered. Research on classroom context has indicated that endorsement of personal goals is often reflective of the perceived contextual goals (e.g., Anderman & Anderman, 1999; Anderman, Maher, & Midgley, 1999; Church, Elliot, & Gable, 2001). For example, students entering a middle school that was perceived to be performance-oriented were more likely to report an increase in their personal performance goals than students entering a middle school that was perceived to be more mastery-oriented (Anderman et al., 1999). Therefore, the personal adoption of goals may depend on what goal orientation the student perceives as being advocated by their particular major. Purpose of the Current Study Given that there are reasons to expect differences between majors in goal pursuit, the primary purpose of this study was to determine if significant differences between majors existed in the pursuit of five goal orientations. If differences were found, it was of Examining Goal Orientation 7 interest to explore whether these differences were attributable to person-level characteristics (specifically, gender or ability) or a contextual-level characteristic (specifically, classification of majors within colleges). The achieve this purpose, the following research questions were addressed: 1) Are there significant differences among majors in goal orientation? 2) To what extent are students within a given major similar in goal orientation? 3) If significant differences among majors do exist, where do the various majors fall along the scale of each goal in relation to other majors (e.g., which majors are highest on endorsing a particular goal and which are lowest)? 4) If significant differences are found between majors, do they still exist after controlling for certain person-level or context-level characteristics? An ideal technique for analyzing the differences among majors on these five goals is hierarchical linear modeling (HLM). The technique of HLM has been used in studies examining achievement goals with elementary school students (e.g., Anderman et al., 2001; Urdan, Midgley, & Anderman, 1998), middle school students (e.g., Anderman et al., 2001), high school students (e.g., Kaplan, Gheen, & Midgley, 2002), and college students (e.g., Church et al., 2001). Using HLM over traditional methods such as ANOVA is advantageous for several reasons (Raudenbush & Bryk, 2002). First, HLM can account for the natural hierarchical structures that can be found in many datasets (e.g., students nested within majors, majors nested within universities) and allows for a simultaneous test of all group differences with a large number of groups (e.g., in this study we have 26). Second, HLM allows for use of both person level (level-1) and group level (level-2) predictor variables. Third, the empirical Bayes estimates of group means Examining Goal Orientation 8 (described below) are considered to be more precise than raw score means (Hox, 2002). Beyond the scope of this study, HLM is also a useful technique when analyzing longitudinal data (see Raudenbush & Bryk, 2002 for a discussion). For those unfamiliar with this technique, an explanation of HLM (and how it was used in the current study) is detailed below. Method Sample The data for this study were collected in the spring of 2002 during a universitywide assessment day at a mid-sized university. Students were required to take a series of tests beginning with the goal orientation measure used in this study and followed by other measures of student learning and development. The sample of 2112 college students consisted primarily of Caucasian sophomores, with 60 percent of the sample consisting of females. Because students were eliminated if there were less than 25 respondents in their respective major, the final sample consisted of 1889 respondents in 26 majors, including a group of students who had not yet selected a major (Undeclared). Measures Over the past 5 to 10 years, the achievement goal framework has grown from a 2factor conception (Dweck, 1986; Nicholls, 1984), to a 3-factor (Elliot 1999; Elliot & Church, 1996; Middleton & Midgley, 1997), and to a 4-factor model of achievement goals (Elliot & McGregor, 2001). This most recent representation includes the goals of mastery-approach, mastery-avoidance, performance-approach, and performanceavoidance. Many of the previous studies that examined the relationship between context and achievement goal pursuits have done so by using either the 2-factor (e.g., Archer & Examining Goal Orientation 9 Scevak, 1998; Roeser et al., 1996) or 3-factor (Church et al., 2001; Kaplan et al., 2002) framework. The current study utilized a modified version of Elliot and McGregor’s (2001) Achievement Goal Questionnaire (AGQ), which represents the 4-factor structure. In addition, a measure of work-avoidance was included, which has been cited in the literature as being related to several achievement goals (e.g., Archer, 1994; Barron & Harackiewicz, in press; Harackiewicz et al., 1997). In fact, recent research has argued that work-avoidance should be included as a goal orientation (Pieper, 2003). Therefore, this study explored differences among college majors using five goal orientations. Students were administered the 12-item AGQ (Elliot & McGregor, 2001) modified for use in a more general context (Finney, Pieper, & Barron, 2004). Specifically, students were asked to consider each statement in reference to their courses for the current semester. The AGQ uses 4 subscales of 3 items each to measure masteryapproach, mastery-avoidance, performance-approach, and performance-avoidance goals. Included in the questionnaire were 4 questions measuring students’ work avoidance tendencies that were adapted from Barron and Harackiewicz (in press). Students were asked to rate the extent to which each statement was true of them on a scale of 1 (Not at all true of me) to 7 (Very true of me). Students’ gender, SAT scores, and majors were obtained from university records for this study. Specifically, their SAT total score was computed by taking the student’s highest reported SAT math and verbal scores. Student major was coded as the major on record for the student at the beginning of semester during which the assessment day took place. For those students who were double majors (less than 1% of the sample), their first listed major was used for these analyses. Examining Goal Orientation 10 Data Analysis HLM was used to answer each of our research questions in the current study. The major focus of estimating models in HLM is to partition the variance of a goal orientation measure into: 1) a part attributable to student-to-student variation within a major (withinmajor variance), and 2) a part attributable to major-to-major variation (between-major variance). In this study, we considered two levels of data: the student-level (level-1) and the major-level (level-2). An example of the student-level model, ij j j i r MAP + = 0 β , (1) demonstrates how the mastery-approach score of student i in major j is modeled as a function of the mean (intercept) for major j (β0j), and the difference between student i's level of mastery-approach and the major mean (level-1 error, rij). These difference values are often referred to as offsets or errors and are typically assumed to be distributed as rij ~N(0,σ). The variation of mastery-approach scores from student to student within a major is captured in σ, the level-1 error variance. If this value is significant, significant differences in mastery-approach exist among students within the same major. At the major-level (level-2) of this model, the model is represented as: oj j u + = 00 0 γ β . (2) The major intercept, β0j, is modeled as a function of the grand mean, γ00 and the difference between the grand mean and the mean for major j (level-2 error, u0j). These error terms, u0js, are typically assumed to be distributed as u0j ~N(0,τ00). The variation of mastery-approach scores among majors is captured in τ00, the level-2 error variance. If τ00 is significantly different from zero, significant differences in mastery-approach between majors exist. The student-level and major-level models can be written in a single Examining Goal Orientation 11 equation, which highlights how the mastery-approach score for a single student i is modeled as a function of the grand mean (γ00), a major effect ( oj u ), and a student effect