Analysis of students' self-efficacy, interest, and effort beliefs in general chemistry†
Brent
Ferrell
and
Jack
Barbera
*
Department of Chemistry and Biochemistry, University of Northern Colorado, Greeley, CO 80639, USA. E-mail: jack.barbera@unco.edu
First published on 19th February 2015
Abstract
Research in academic motivation has highlighted a number of salient constructs that are predictive of positive learning strategies and academic success. Most of this research has centered on college-level social sciences or secondary school student populations. The main purpose of this study was to adapt existing measures of personal interest and effort beliefs to a college chemistry context. In addition, a chemistry-specific measure of self-efficacy was evaluated in a modified form. This set of scales was initially administered at two time points in a first-semester general chemistry course to a sample of undergraduates (n1 = 373, n2 = 294). Confirmatory factor analyses (CFA) were conducted to determine whether the scales were functional in a chemistry context. Following revision of the scales, all CFA models demonstrated acceptable fit to the data. Cross-validation of the revised scales was performed using two different populations (n = 432, n = 728), with both studies producing similar model fits. Furthermore, our data shows that chemistry majors reported higher self-efficacy and interest than non-science majors. Cronbach's alpha estimates ranged from 0.75 to 0.92 for the revised scales across all studies. This set of scales could provide useful tools for assessing general chemistry students' motivation and the motivational impacts of various teaching practices.
Introduction
Introductory chemistry is a course required by several disciplines. At most institutions, enrollment overwhelmingly consists of students outside of the discipline of chemistry. Often, many students struggle through chemistry and are unsuccessful due to the complexity and abstract nature of the content (Nakhleh, 1992). The combination of content difficulty and the fact that most students are fulfilling a credit requirement for their non-chemistry majors generates an interesting classroom environment for the introductory-level chemistry course. Many factors can influence whether a student is successful in chemistry. There are some obvious characteristics of students including: inherent aptitude and prior experience in chemistry, which can be predictive of success in chemistry (Tai et al., 2005). However, research has shown that cognitive factors such as these are not sufficient to predict achievement, but must be augmented by adaptive motivational processes (Dweck, 1986; McCoach and Siegle, 2003; Zusho et al., 2003).The importance of motivation for learning and achievement in any classroom setting is indisputable (Dweck, 1986; Schunk, 1991; Ames, 1992; Hidi and Harackiewicz, 2000; Singh et al., 2002; Zusho et al., 2003). Student motivation has been described as a theoretical construct that can explain “the degree to which students invest attention and effort in various pursuits” (Brophy, 2010). Motivation and ability are the two major components for academic success among students (Hidi and Harackiewicz, 2000). Despite this, students' motivation can easily be ignored due to its complexity, or oversimplified as an unchanging facet of one's character. Motivation in the classroom is very complex, and can fluctuate in different situations (Nicholls et al., 1989; Pintrich, 2003) and among different subjects (Guay et al., 2008). Students' motivation toward school tends to become highly differentiated throughout grade school, as individuals encounter various situations and experiences that shape their interests and conceptions of ability. As such, it is important for researchers to study motivation in specific contexts (Pintrich, 2003).
Although much research has been conducted to understand the motivational and affective factors that influence performance and student engagement in the college classroom, only a small fraction of this has been directed towards science classrooms. In 2012, the National Research Council called for a collective effort on the part of science education experts across many disciplines to put together a report that would highlight current research areas in education, as well as areas that are lacking across these disciplines. As pointed out in this Discipline-Based Education Research (DBER) report, students' dispositions and motivations to learn science and engineering are largely understudied and are of “central importance” (National Research Council, 2012). Nevertheless, many researchers have explored specific motivational constructs and processes in the context of college chemistry both prior to and since this report (Dalgety et al., 2003; Zusho et al., 2003; Bauer, 2005; Dalgety and Coll, 2006; Taasoobshirazi and Glynn, 2009; Uzuntiryaki and Aydin, 2009; Xu et al., 2013; Villafane et al., 2014).
Our aim in this study is to expand upon the current body of research directed at understanding motivational processes among students in college chemistry. In particular, we were interested in modifying existing measures of motivational processes for use in college chemistry. We chose to examine three distinct constructs that have been linked to motivation in students: self-efficacy, interest, and effort beliefs (Zimmerman, 2000; Hidi and Renninger, 2006; Jones et al., 2012). We chose these three variables both for their salience within motivation literature and their influence on student performance and retention. Self-efficacy has been found to be positively correlated to achievement outcomes in many studies, as well as the adaptive motivational processes, effort and persistence (Pintrich and De Groot, 1990; Multon et al., 1991; Pajares and Miller, 1994; Zusho et al., 2003). Individual interest is less consistently correlated with performance, but is more strongly linked to positive learning strategies, such as mastery goals and attention (Hidi and Renninger, 2006). Furthermore, discipline-specific individual interest has been positively correlated with choice of major and number of classes taken within that discipline (Harackiewicz et al., 2008). Positive effort beliefs are very highly correlated with an incremental theory of intelligence, which states that competence is not fixed, but malleable (Blackwell et al., 2007; Dweck, 2012; Jones et al., 2012). Effort is absolutely vital for success in any college classroom. However, preceding the action of effort itself must be a positive, adaptive belief about the potential impacts of the action. Hence, effort beliefs are indicators about how a student perceives the impact of their effort on learning and performance, and students who believe their competence can be changed through effort tend to be more motivated and perform at higher levels than those who do not (Dweck, 2000).
The emphasis in academic motivational research has shifted from behavioral aspects such as drive and reinforcement to beliefs, goals, and expectations over the last 40 years (Wigfield and Eccles, 2002). These modern theories of motivation point to the critical role that expectancies and beliefs play in adaptive learning patterns among students (Eccles and Wigfield, 2002; Zeldin et al., 2008). By researching these beliefs among students in college chemistry, we can have a more clear understanding of the basis for motivational processes that exist in our classrooms. However, to do this we first need measures of the various motivational aspects that have been tested within the target population.
Self-efficacy
Self-efficacy is rooted in social cognitive theory and is defined as the self-appraisal of one's capacity to execute a specific task (Bandura, 1977). Efficacy beliefs or expectations are self-referent and guide an individual toward certain actions and away from others (Pajares, 1996). Self-efficacy must be distinguished from two theoretically similar constructs, outcome expectations and self-concept. Outcome expectations are beliefs that certain behaviors will lead to certain outcomes. Both self-efficacy and outcome expectations influence motivation, but self-efficacy is thought to play a larger role in predicting achievement (Zimmerman, 2000). While both self-efficacy and outcome expectations are task-specific, self-concept is much more broadly defined, and has more to do with one's beliefs about their self-worth and competence within a domain (Pajares and Miller, 1994). Moreover, self-concept has been defined in many different ways, is tied to affective as well as cognitive judgments, and is inherently norm-referenced (Hansford and Hattie, 1982; Bong and Clark, 1999). In contrast, the prevailing definition of self-efficacy has remained virtually unchanged since its inception (Bandura, 1977). Also, self-efficacy judgments are less likely to be influenced by social comparisons and affective swings due to the task-specific, objective nature of the construct (Bong and Clark, 1999). For these reasons, and the notable presence of self-efficacy in the literature, we chose to measure self-efficacy instead of other related constructs.Bandura (1986) argued that efficacy expectations are drawn from four sources of information: performance accomplishments, vicarious experiences, verbal persuasion, and physiological states. The most dominant of these sources are performance accomplishments, because they are founded upon personal mastery experiences (Bandura, 1977). However, some suggest the relative salience of sources for self-efficacy beliefs may be different for males and females (Zeldin et al., 2008). Nevertheless, a student in chemistry is more likely to have high efficacy expectations for a particular task if he or she has already successfully completed that task. When asked to explain their self-efficacy in college chemistry, students noted their prior success in chemistry as a common theme (Dalgety and Coll, 2006). From a quantitative approach, Lopez and Lent (1992) found that prior experience in math explained the most variance in self-efficacy scores when considering math self-concept, interest in math, and perceived value of mathematics. Equally important to considering the sources of self-efficacy, is the influence that self-efficacy has on the student.
Self-efficacy is hypothesized to have far-reaching implications in academics by influencing students' effort, perseverance, and emotional reactions to specific tasks in school (Lent et al., 1984; Lopez and Lent, 1992; Pajares and Kranzler, 1995; Pajares, 1996). Lent et al. (1984) investigated the relationship of students' self-efficacy beliefs to persistence in technical and science majors. They found that students who reported higher self-efficacy scores for completing their educational requirements were more likely to persist in their major. This supports Hackett and Betz's (1981) hypothesis that self-efficacy is linked to persistence in career goals.
Other studies across several disciplines suggest that self-efficacy is related to academic performance, problem solving, college entrance, and college major choice (Betz and Hackett, 1983; Lent and Hackett, 1987; Lopez and Lent, 1992; Pajares and Kranzler, 1995; Andrew, 1998; Britner and Pajares, 2001; Schunk and Pajares, 2002; Zusho et al., 2003; Parker et al., 2014). Zimmerman et al. (1992) found that students' academic self-efficacy scores significantly predicted (β = 0.39) their final grade in a high school social studies class. In a college setting, it was found that students' self-efficacy beliefs were significantly correlated to their grade-point average (GPA) and accounted for more variation in GPA than ACT scores (Gore, 2006).
Generalization can also occur with efficacy expectations, such that a student who has experienced mastery with one task could report high efficacy expectations for a similar task (Bandura et al., 1969). However, the degree of generalization is limited to the domain of functioning (Zimmerman, 2000). For example, one cannot assume that just because a student has high self-efficacy in biology, he or she will also have high self-efficacy in chemistry. Thus, it is important, when measuring and describing students' self-efficacy, to ensure that it is domain-specific.
Chemistry has been referred to as the “central science” and it is believed that mastery of chemistry concepts is influential for success in later science courses (Tai et al., 2005). Although chemistry course requirements are not as widespread as those of math, there are many professions unrelated to chemistry, particularly in health care, that require a background in chemistry (Brown et al., 2012). In fact, students pursuing degrees other than chemistry make up the bulk of the enrollment in introductory chemistry at larger universities, and it is these students who are least likely to exhibit high self-efficacy in chemistry (Uzuntiryaki and Aydin, 2009). Therefore, since self-efficacy could have a substantial influence on students' achievement and retention in college chemistry, it is an important factor to consider in chemical education research (Zusho et al., 2003).
Several researchers have investigated chemistry-specific self-efficacy in university-level classes (Dalgety et al., 2003; Zusho et al., 2003; Taasoobshirazi and Glynn, 2009; Uzuntiryaki and Aydin, 2009; Villafane et al., 2014). Taasoobshirazi and Glynn (2009) conducted a study where undergraduate introductory chemistry students were asked to solve a series of problems along with judging their self-efficacy in chemistry. It was found that students who reported high self-efficacy were more likely to use forward-working strategies and obtain the problem solution than students who reported low self-efficacy. In a related study, Zusho et al. (2003) investigated the role of several motivational processes on achievement in chemistry, as well as the correlations between them. They found that self-efficacy was the best predictor of final course grade, when measuring SAT-math, task value, and rehearsal strategies. Both of these studies support Bandura's (1986) notion that self-efficacy is positively related to achievement through persistence and effort. Villafane et al. (2014) explored student self-efficacy trajectories across a semester of preparatory chemistry. Their results suggest that individual characteristics (race/ethnicity and gender) could influence the degree to which students show an increase in self-efficacy across the semester. Although these studies have addressed important gaps in the literature, more research is needed that investigates the interplay between different motivational variables.
Interest
Interest, in academic settings, can be described as a psychological state where the student is engaging or has a predisposition to engage with content over time (Hidi and Renninger, 2006). Hidi and Baird (1986) argued that interest is more than “arousal,” but must be considered as a process. As a process, interest is said to endure and persist through time. There are two types of interest outlined in the literature: situational interest and individual interest (Hidi, 1990). Situational interest refers to an interest that is triggered spontaneously through interaction with the environment (Harackiewicz et al., 2008). This type of interest can be considered to deliver a sense of enjoyment and curiosity, but may or may not persist within the person (Renninger, 2000). Individual interest refers to a relatively stable interest that has developed over time and is associated with an enduring predisposition for the person to reengage with specific topics, subject areas, or activities (Hidi, 1990; Schiefele, 1991; Hidi and Renninger, 2006). Individual interest can be further broken down into two components: feeling-related and value-related (Schiefele, 1999). Feeling-related interest (e.g., Chemistry is fascinating to me) is tied to stimulation and enjoyment. Value-related interest (e.g., the material we are learning in chemistry is important for me to know) is associated with importance and personal significance and has been positively correlated with performance in academic contexts (Hulleman et al., 2008).Hidi and Renninger (2006) postulated that interest is a developmental process, which occurs in four phases: triggered situational interest, maintained situational interest, emerging individual interest, and well-developed individual interest. Triggered situational interest results from temporary cognitive or affective changes in the individual (Hidi and Baird, 1986). Maintained situational interest occurs after a triggered event, where the individual's interest is “held” and involves focused attention that endures for a period of time. If maintained situational interest persists long enough, it becomes emerging individual interest, characterized by stored value, positive feelings, and consistent reengagement with the material or activity. At some point, emerging interest can become well-developed interest following substantial reengagement with the material. Individuals with well-developed interest seek out answers to questions, and are likely to be resourceful when answers are not easily found. Also, it is possible for these individuals to expend effort, but feel as though the task is “effortless” (Renninger and Hidi, 2002). Although the development of interest is not the same for all individuals, Hidi and Renninger (2006) argued that there is no evidence suggesting well-developed interest can spawn without the individuals first being exposed to the area and experiencing triggered interest.
In the same vein as efficacy beliefs, interest is content-specific and represents a personal significance between the individual and the object of his or her interest (Schiefele, 1991; Renninger and Hidi, 2002). Science, as a subject matter in schools, is somewhat broad and diffuse in elementary and middle school years, but becomes differentiated and more focused during high school and college years. In light of this, several studies on interest in science and/or students' perceptions of the value of science at lower grade levels have focused on “science” as a whole (Anderman and Young, 1994; Singh et al., 2002; Tuan et al., 2005). However, studies on interest in science dealing with high school and college student populations tend to center on specific disciplines, such as chemistry, physics, and biology (Dalgety et al., 2003; Uitto et al., 2006; Gungor et al., 2007; Barbera et al., 2008). These studies underscore the importance of developing measures to target individual interest as a domain-specific construct. Hence, to understand more about interest as a component of motivation in the college chemistry classroom, the focus of the interest measure must be specifically directed towards the discipline of chemistry.
Effort beliefs
Effort can be considered as part of the attributional theory of motivation, and from this theoretical standpoint, is intimately tied to conceptions of ability (Weiner, 1985). Weiner (1985) pointed out that ability and effort are the most salient causal ascriptions to achievement. In short, students believe that those who have high ability and display high effort will be more successful than students who have low ability and display low effort. The notion that effort and persistence has a positive effect on the academic outcome of a student has been supported by empirical studies (Stipek and Gralinski, 1996; Elliot, 1999). For example, Elliot (1999) found that self-reported persistence and effort were positive predictors of academic performance. Effort was found to be a mediator between adaptive mastery goals and academic performance. Another study revealed that students who expressed positive beliefs toward the value of effort do not necessarily show increased performance, but do tend to focus more on mastery and the development of their abilities (Stipek and Gralinski, 1996). Effort is certainly a key component in the academic success of students. It must be considered when making judgments about academic performance, due to how it mediates the link between motivational constructs and academic outcomes (Elliot, 1999; Goodman et al., 2011).Students' beliefs about effort are a precedent to effortful actions and are highly correlated with their conceptions about intelligence and ability (Blackwell et al., 2007; Jones et al., 2012). These conceptions are referred to as implicit theories of intelligence. Implicit theories are “beliefs about the nature of human attributes”; two implicit theories of intelligence have been described in the literature: incremental and entity (Dweck, 2012). An incremental theory of intelligence is characterized by the view that intelligence is malleable and can change over time with the expenditure of effort. Conversely, individuals who hold an entity view of intelligence see it as fixed and unchanging, independent of effort (Dweck and Leggett, 1988). Students holding an incremental view of intelligence are more likely to see effort as enhancing ability and apply more effort to overcome obstacles. On the contrary, those endorsing an entity view of intelligence are less likely to put forth effort in the face of failure, less likely to be interested in a subject, and exhibit achievement gaps when compared to incremental theorists (Dweck and Sorich, 1999; Hong et al., 2004; Dweck, 2012). Effort beliefs and implicit theories of intelligence are fundamentally different constructs, but deeply related (Blackwell et al., 2007; Jones et al., 2012). Between these two, the vast majority of research has centered on implicit theories, leaving a gap with respect to effort beliefs. No studies were found on effort beliefs in a college setting, but a few studies have investigated effort beliefs in secondary school (Blackwell et al., 2007; Jones et al., 2012).
It is important that students feel they can improve upon their abilities with persistence and effort. This could be a challenge to get across to college students, as it is likely that their beliefs about intelligence and effort have already developed by the time they reach the college classroom. However, if we understand more about effort beliefs and how they can change among college-age students, new instructional strategies can be implemented so that more students will endorse positive effort beliefs in their classes. This is particularly important in math and science, because most students have experienced both prior to coming to college, and have likely developed beliefs about their abilities in those domains.
Purpose and rationale for current study
There is a need in the chemistry education community to understand some of the motivational and affective components of students enrolled in chemistry courses. What is true of one college subject is not necessarily true of another, and given the lack of motivational research in college-level sciences, it is important that this need be addressed.In order to effectively assess a large classroom of students on their motivational characteristics and dispositions, instructors and researchers must rely upon easy to administer self-report scales or instruments, consisting of items targeted at measuring a specific latent trait or group of traits. As with any scale or instrument, either in the physical or social sciences, the quality of data that can be produced from it depends largely on the quality of the data generated with the target population (Barbera and VandenPlas, 2011; Heredia and Lewis, 2012). Thus, steps must be taken to ensure that a scale or instrument will produce valid and reliable results when used with the target population.
While general and science-specific motivation instruments exist, such as the Motivated Strategies and Learning Questionnaire (MSLQ), and the Science Motivation Questionnaire (SMQ), the current availability of individual scales that measure motivational variables in college chemistry is limited (Pintrich et al., 1993; Glynn et al., 2009). In 2003, Bauer and colleagues published the Chemistry Self-Concept Inventory (CSCI), a 40-item survey that was adapted to measure students' self-concept in five domains: math, chemistry, academic, academic enjoyment, and creativity. There are two instruments (Colorado Learning Attitudes about Science Survey (CLASS) and Chemistry Attitude and Experiences Questionnaire (CAEQ)), which measure interest in chemistry (Dalgety et al., 2003; Barbera et al., 2008). The MSLQ and SMQ also have general academic interest scales, which could be adapted for a chemistry context. However, none of these instruments were designed based on the prevailing theoretical underpinnings of interest theory from educational psychology (Schiefele, 1991; Renninger, 2000). The CAEQ, being the largest instrument for motivation and affect in college chemistry, also has a self-efficacy scale. The CAEQ together with the College Chemistry Self-Efficacy Scale (CCSS) represent the only two instruments designed specifically to measure self-efficacy in a college chemistry setting. Both of these were designed using Bandura's (1986) theory of self-efficacy, which is widely accepted among researchers across many disciplines. The MSLQ and SMQ also have self-efficacy scales. However, the items in both scales are very general, relating more to the course as a whole than to tasks within the course. As mentioned above, self-efficacy is conceptualized as confidence at the task-level. Thus, scales purported to measure self-efficacy in a particular academic domain should be written with items targeting specific tasks encountered within that academic domain. We carefully examined the self-efficacy items from those that had been used in chemistry and related disciplines for wording that was most appropriate for an introductory chemistry class in a college setting. The CCSS contained items that were relevant, task-specific, and readable for college students in introductory chemistry. No chemistry-specific measures for effort beliefs, or implicit theories of intelligence were found.
The goal of the present study was to identify, utilize, and evaluate existing measures of motivational constructs (initial interest, maintained interest, self-efficacy, and effort beliefs) in a college chemistry setting. As doing this required modification of items and scales, this manuscript represents evidence of validity and reliability for these measures prior to their use in future studies of instructional styles. Items from existing scales must be modified to be discipline-specific, as the constructs under investigation are operationalized around a meaningful connection between the subject area and the individual (Bandura, 1986; Schiefele, 1991). Modifying items, scales, and instruments for subject-specific language and meaning is common practice in science and math education (Barbera et al., 2008; Taasoobshirazi and Glynn, 2009; Jones et al., 2012). However, any modifications to an item, scale, or instrument must be followed up by an investigation for validity and reliability evidence (Barbera and VandenPlas, 2011).
Items were taken from three independently published scales, each measuring a separate motivational construct (self-efficacy, interest, or effort beliefs) in academic settings. The items from the effort beliefs and interest scales were adapted to fit a college chemistry context. The self-efficacy items were originally developed for college chemistry and did not need modification. All scales were subjected to confirmatory factor analysis to determine if the structure of each scale matched that proposed by the individual scale developers. In addition, indicators of global and component model fit were used to assess whether any of the scales should be modified. Student interviews were also conducted and used as validity evidence for potential modifications of items and scales. Interviews with the target population are a vital part of evaluating an item, scale, or instrument for response process validity (Arjoon et al., 2013; Wren and Barbera, 2013). The data collection and subsequent analyses were guided by the following research questions:
(1) What modifications are needed to produce brief, chemistry-specific scales of self-efficacy, interest, and effort beliefs?
(2) What evidence supports the functioning of each of the modified scales?
(3) To what extent do students' interest, effort beliefs, and self-efficacy change across a semester of college chemistry?
The current study will contribute to the growing body of literature centered on motivational and affective processes among students in college chemistry courses. In addition, brief chemistry-specific scales for measuring three salient motivational beliefs will be made available to educators and researchers interested in gauging the motivational climate of their chemistry classrooms. These scales will be used in a follow-up study as variables in a path analysis to investigate a set of a priori motivation models that will include measures of academic performance in general chemistry. These scales provide important tools for educators who plan to implement new teaching strategies, and are interested in more than just the performance outcomes that result from those strategies. The follow-up study will provide detailed connections among the scales and student performance.
Methods
Participants
In accordance with Institutional Review Board (IRB) policy, students in both the quantitative and qualitative studies were informed that their participation had no impact on their course grade and that they would be volunteering for a research study regarding their academic motivation. Standard university policies for confidentiality and data handling were utilized throughout the study. Participation in the studies was voluntary and no incentives were given to students for participating.
Data collection
Scales
Qualitative
Quantitative data analysis
Global fit of each one-factor model was analyzed based on several indices including: Satorra–Bentler (SB) scaled chi-square (Satorra and Bentler, 1994), root mean squared error of approximation (RMSEA); (Steiger, 1990), non-normed fit index (TLI); (Tucker and Lewis, 1973), standardized root mean squared residual (SRMR), and comparative fit index (CFI); (Bentler, 1990). The SB scaled chi-square is a test for exact model fit, where the population covariances are fully reproduced by the hypothesized model. A non-significant result is desired and indicates that there are not significant discrepancies between the population covariances and those predicted by the model (Kline, 2011). However, the chi-square test is sensitive to sample size and will often produce a significant result for very small deviations in model fit. Thus, other descriptive test statistics are used to assess the fit of the model (Schermelleh-Engel et al., 2003).
There are two types of fit indices commonly used in the literature to assess model fit: absolute and incremental. Absolute fit indices (i.e., chi-square, RMSEA, and SRMR) are estimates of how well an a priori model fits the data. Incremental, or comparative fit indices (i.e., TLI and CFI) reflect improvement of model fit compared to a baseline model (Kline, 2011). The RMSEA can range from 0 to infinity and is a measure of the approximate model fit in the population (Steiger, 1990). Because exact fit of the model in the population is impractical, the RMSEA is a measure of “close fit,” and in general, values <0.05 are considered good and those <0.08 are considered reasonable (Browne and Cudeck, 1992). The SRMR value ranges from 0 to 1 and is a “badness of fit” measure based on the standardized fitted residuals. By standardizing the residuals, the scale of the variables is taken into account (Schermelleh-Engel et al., 2003). Hu and Bentler (1995) suggested that an SRMR value of <0.05 is indicative of good fit and <0.10 is acceptable. The TLI and CFI both take into account the chi-square values of the proposed model and the null baseline model (Brown, 2006). The TLI and CFI values are normed and range from 0 to 1, with values ≥0.95 indicating good fit (Hu and Bentler, 1999). Only when several fit indices (both incremental and absolute) are considered together can the quality model fit be assessed with reasonable propriety (Brown, 2006). Based on what is commonly accepted in the literature, we used the following cut-off values as an evaluation of acceptable model fit beyond the chi-square test statistic: RMSEA ≤ 0.05, SRMR ≤ 0.10, TLI and CFI ≥ 0.95 (Hu and Bentler, 1999; Hooper et al., 2008).
Component model fit was evaluated based on statistical significance (p < 0.05) and reasonable parameter estimates. In addition, modification indices were considered when significant. The modification index (MI) is represented as a one degree of freedom chi-square statistic that estimates the difference between two nested models. Modification indices are parameter-specific and reflect the approximate decrease in the model chi-square statistic when the fixed parameter is allowed to be freely estimated (Brown, 2006). With regard to CFAs, MI values are most commonly associated with correlated residuals between two indicator variables or between an indicator and two factors in the model. Thus, the higher the MI, the more likely it is that a particular indicator either is redundant or belongs with another factor. Modification indices are part of the evidence used to assess whether an indicator (item) should be dropped, aggregated with another indicator, or relocated to a different factor (latent trait). When the MI value exceeds the critical χ2 value (α = 0.05) using the degrees of freedom (df) from the model, then the appropriate modification should be considered (Hancock, 1999).
The one-factor CFAs were conducted on data collected at each time point. At time 1, all complete data sets were included for those students who consented to the study (n = 373). At time 2, only matched data sets were used in the analysis (n = 294). Although more than 294 students participated at the end of the semester, consent forms were only issued at the beginning of the semester, precluding the use of data from students who might have been absent at the first data collection or added the class late.
Qualitative data analysis
All interviews were transcribed and coded for significant statements and emergent themes, based on each item and its corresponding scale (Creswell, 2013). The strategy for coding centered on readability and the degree of consistency among participants' interpretations of the items. If students repeatedly report dissimilar interpretations of an item, then there can be no consensus on what the score of that item really means. This is problematic, as it negatively affects the validity of the scale and the inferences that can be drawn from the scores.Results
Interview results
A total of nine interviews (2 males and 7 females) were conducted in the fall of 2013 and five additional interviews (4 males and 1 female) in the spring of 2014. We felt that this number of interviews was sufficient to reach consensus on the items, especially given that we were not designing them from scratch. The students who participated in the fall 2013 interviews were asked to comment on the initial interest, self-efficacy, and effort beliefs scales. The interviews were conducted during the middle of the semester; therefore, students had not yet been given the maintained interest scale. For this reason, and to solidify the results from the self-efficacy scale, the interviews in the spring of 2014 covered the maintained interest and self-efficacy items. The effort beliefs scale and initial interest scales were not included in the spring interviews because we observed consistent responses from participants regarding the meaning and interpretability of items during the fall interviews.In-depth student interviews were conducted using the full versions of the scales. The results from the interviews were used in conjunction with quantitative results from the CFAs as support to flag any items that should be considered for modification or removal.
“Oh, it was a little confusing, I guess, yea. I would have said – I kind of connected it with the other ones and just said strongly agree. But I guess – but now I would say I disagree because…Oh, wait, let me re-phrase that. Yeah, it's a little confusing, the wording. I strongly agree with that too”.
This student started off with an affirmative response, then he switched his answer to a negative response, before returning to his original choice. The confusion over the wording of this item was consistent throughout the interviews.
In addition to readability, we were interested in how students attributed meaning to the items. Toward this end, participants were asked to explain their reasoning behind the answer choice they made. Most participants gave plausible reasons for the answers they chose, and provided rational explanations. However, there were two items from the self-efficacy scale (Items 3 and 8) for which the agreement of the meaning differed among several students. Item 3 (How well can you describe the structure of an atom?) was problematic because there were differing opinions about what it meant to “describe the structure of an atom.” Some students reported describing the structure of an atom simply meant knowing “the positions of things and charges”. Others reported that “interactions” and valence shell theory were part of the description. We observed, in several cases, that this item could be interpreted to varying degrees of depth and understanding. Item 8 (How well can you solve chemistry problems?) was also problematic for a similar reason. Participants regarded this item as “broad”, “vague”, and “depending on the problem”. Clearly, there are many types of problems students encounter in first-semester general chemistry. Diffuse tasks such as these can lead to problems when a student tries to self-appraise their ability to complete the task (Bandura, 1986). Additionally, item 8 is somewhat redundant in that every item that precedes it represents some type of chemistry problem.
We found that participants during the fall and spring interviews used dissimilar language when describing their reasons for answers to feeling-related items versus value-related items. Examples of feeling-related items from the initial interest scale are: I am fascinated by chemistry and I chose to take general chemistry because I'm really interested in the topic. Participants cited reasons for choosing their answers by using words and phrases such as: “I'm naturally gifted”, “I connect with the material”, “It excites me”, “Interested”, “Fascinated”. These words are evoked from “feelings of involvement, stimulation, and enjoyment” toward the topic of chemistry, and is exactly the type of interest that is characterized by feeling-related items (Schiefele, 1999). On the other hand, participants explaining their answer choices for value-related items used entirely different language. Examples of value-related items from the initial interest scale are: I think what we will study in general chemistry will be important for me to know and I think the field of chemistry is an important discipline. Participants commented on their answer choices by using phrases such as: “I'm going to be building off this”, “it will obviously be important in my field”, “this class will…help with future chemistry classes”, “chemistry is…everything around us”. Schiefele (1999) describes value-related interest as being directed toward something that is personally significant and important to the individual. The statements made by the participants during the interviews regarding value-related items were indicative of a personal significance and importance, as opposed to feelings of excitement or enjoyment. Based on the overwhelming difference we found in how participants described their interest using feeling-related items versus value-related items, the scale was tested as both a 1 and 2-factor model.
Data screening and descriptive statistics
Prior to analysis, all data sets were screened for careless responses from students (i.e., students selecting all of one response option). Only one case was found that exhibited this pattern. Missing item-level data were also screened for patterns. The only consistently missed item was the last item on the list. Of 37 total cases that had missing data, 17 of them failed to respond to the last item. As list-wise deletion was implemented, all cases with missing data were removed from each data set prior to analysis.Mean, standard deviation, skew and kurtosis were evaluated for each item on each scale (see Appendix, ESI†). Most items had skew and kurtosis values that were within acceptable ranges to be considered normal (−1 to +1) (Huck, 2012). However, some items were outside of this range with negative skew values down to ∼−1.5 and kurtosis values up to 3.5. Due to these deviations from normality, the robust maximum likelihood (RML) estimator was used in all CFA runs. The RML estimator utilizes the Satorra-Bentler scaled chi-square statistic, and is robust with respect to non-normal data (Chou et al., 1991).
| Scale | # of items | χ 2 value | dfa | p-value | RMSEA | TLI | CFI | SRMR | α |
|---|---|---|---|---|---|---|---|---|---|
| χ 2 – likelihood ratio test, RMSEA – root mean squared error of approximation, TLI – Tucker–Lewis index, CFI – comparative fit index, SRMR – standardized root mean squared residual.a Degrees of freedom (df) are based on RML estimator.b Cronbach's alpha.c 2-factor model.d Initial interest (feeling).e Initial interest (value). | |||||||||
| Initial interest | 7 | 158.81 | 14 | <0.001 | 0.17 | 0.93 | 0.95 | 0.09 | 0.88 |
| Initial interest (revised)c | 4,d 3e | 23.84 | 13 | 0.033 | 0.05 | 0.99 | 1.00 | 0.03 | 0.90,d 0.79e |
| Effort beliefs | 9 | 148.25 | 27 | <0.001 | 0.11 | 0.92 | 0.93 | 0.10 | 0.75 |
| Effort beliefs (revised) | 6 | 16.27 | 9 | 0.26 | 0.05 | 0.99 | 0.99 | 0.05 | 0.77 |
| Self-efficacy | 8 | 96.49 | 20 | <0.001 | 0.10 | 0.97 | 0.98 | 0.06 | 0.89 |
| Self-efficacy (revised) | 6 | 43.32 | 9 | <0.001 | 0.10 | 0.97 | 0.98 | 0.05 | 0.85 |
| Scale | # of items | χ 2 value | dfa | p-value | RMSEA | TLI | CFI | SRMR | α |
|---|---|---|---|---|---|---|---|---|---|
| χ 2 – likelihood ratio test, RMSEA – root mean squared error of approximation, TLI – Tucker–Lewis index, CFI – comparative fit index, SRMR – standardized root mean squared residual.a Degrees of freedom (df) are based on RML estimator.b Cronbach's alpha.c 2-factor model.d Maintained interest (feeling).e Maintained interest (value). | |||||||||
| Maintained interest | 8 | 222.34 | 20 | <0.001 | 0.19 | 0.92 | 0.94 | 0.12 | 0.91 |
| Maintained interest (revised)c | 4,d 4e | 22.49 | 19 | 0.26 | 0.03 | 1.00 | 1.00 | 0.04 | 0.92,d 0.87e |
| Effort beliefs (revised) | 6 | 15.49 | 9 | 0.08 | 0.05 | 0.99 | 0.99 | 0.03 | 0.82 |
| Self-efficacy (revised) | 6 | 36.03 | 9 | <0.001 | 0.10 | 0.97 | 0.98 | 0.04 | 0.87 |
| Fit index | Revised scales (n = 373) | Cross-validation 1 (n = 432) | Cross-validation 2 (n = 728) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| II | MI | EB | SE | II | MIc | EB | SE | II | EB | SE | |
| χ 2 – likelihood ratio test, RMSEA – root mean squared error of approximation, TLI – Tucker–Lewis index, CFI – comparative fit index, SRMR – standardized root mean squared residual.a Degrees of freedom (df) are based on RML estimator.b Cronbach's alpha.c Post-semester n = 175.d Feeling sub-scale.e Value sub-scale. | |||||||||||
| χ 2 | 23.84 | 22.49 | 16.27 | 31.31 | 49.02 | 18.31 | 60.01 | 62.74 | 80.44 | 48.61 | 66.78 |
| dfa | 13 | 19 | 9 | 9 | 13 | 19 | 9 | 9 | 13 | 9 | 9 |
| p-value | 0.03 | 0.26 | 0.26 | 0.00 | 0.00 | 0.50 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| RMSEA | 0.05 | 0.03 | 0.05 | 0.08 | 0.08 | 0.00 | 0.11 | 0.12 | 0.08 | 0.08 | 0.09 |
| TLI | 0.99 | 1.00 | 0.99 | 0.98 | 0.99 | 1.00 | 0.96 | 0.96 | 0.98 | 0.97 | 0.96 |
| CFI | 1.00 | 1.00 | 0.99 | 0.99 | 0.99 | 1.00 | 0.98 | 0.98 | 0.99 | 0.98 | 0.98 |
| SRMR | 0.03 | 0.04 | 0.05 | 0.04 | 0.05 | 0.04 | 0.08 | 0.04 | 0.04 | 0.06 | 0.05 |
| α | 0.90,d 0.79e | 0.92,d 0.87e | 0.77 | 0.85 | 0.91,d 0.90e | 0.94,d 0.86e | 0.84 | 0.88 | 0.90,d 0.84e | 0.75 | 0.82 |
Fit indices and alpha values from the cross-validation samples confirm the validity and reliability of the revised scales. All scales had consistently high Cronbach's alpha values across all samples, indicating similar reliability for each administration. With only a few exceptions, the fit indices were within the range of acceptable values. In the cross-validation 1 sample, the effort beliefs scale produced SRMR (0.08) and RMSEA (0.11) values above the standard cut-offs (≤0.06 and ≤0.08). However, Hu and Bentler report that acceptable SRMR values can be as high as 0.09 with CFI and TLI values >0.95 (Hu and Bentler, 1999). Therefore, with only the RMSEA value being out of range, the revised scale model has acceptable fit to the data from this sample. The self-efficacy scale produced inflated RMSEA values (0.12 and 0.09) in both cross-validation samples; however, the other indices were well within the acceptable ranges. Therefore, the revised self-efficacy scale is deemed to have acceptable fit to both populations. As both additional data sets were larger than the original data set, it is expected that the SB-scaled chi-squared values would be significant as the chi-square test is highly dependent on sample size (Brown, 2006). The only exception to this was the maintained interest scale, as this post-semester data set only contained 175 students. Taken together, the consistent fit indices and alpha values across all samples provide supporting evidence for the validity and reliability of the revised scales across three different samples from two different institutions.
| Scale | Mean (SD) time 1 | Mean (SD) time 2 |
|---|---|---|
| a Note: interest scales were different from time 1 to time 2 and cannot be directly compared. | ||
| Initial interest (feeling) | 3.46 (0.84) | — |
| Initial interest (value) | 4.06 (0.67) | — |
| Maintained interest (feeling) | — | 3.23 (0.90) |
| Maintained interest (value) | — | 3.61 (0.83) |
| Study | Scale | Mean (SD) time 1 | Mean (SD) time 2 | Mean differencea (effect sizeb) | p-value |
|---|---|---|---|---|---|
| a Based on paired samples t-test. b Effect size represented by Cohen's d – small (0.20), medium (0.50), large (0.80) (Cohen, 1992). | |||||
| Initial (n = 294) | Effort beliefs | 3.95 (0.57) | 3.77 (0.68) | −0.18 (0.29) | <0.001 |
| Self-efficacy | 3.29 (0.60) | 3.60 (0.65) | 0.31 (0.50) | <0.001 | |
| Cross-validation (n = 175) | Effort beliefs | 4.03 (0.57) | 3.91 (0.74) | −0.13 (0.18) | 0.022 |
| Self-efficacy | 2.87 (0.73) | 3.69 (0.64) | 0.71 (1.19) | <0.001 | |
Our results suggest that students' chemistry self-efficacy increased across the semester for both the initial sample and cross-validation sub-sample (see Table 5). Self-efficacy scores were based on a mean composite score of the revised scale (6 items). The mean difference in scores for the initial study (M1 = 3.29, M2 = 3.60) for all students across the semester was significant, t(293) = 8.23, p < 0.001). The effect size (d = 0.50) for this comparison was medium (Cohen, 1992). Data from the cross-validation study sub-sample showed a similar trend (M1 = 2.87, M2 = 3.69), with the difference also being significant, t(174) = 14.83, p < 0.001). The effect size (d = 1.19) for this test was large (Cohen, 1992). On average, students from both samples felt more confident in their abilities to solve chemistry problems at the end of the semester than at the beginning of the semester.
While our results suggest an increase in self-efficacy for our overall sample, this trend might not hold for all students in the sample. For example, Villafane et al. (2014) reported differing trajectories in chemistry self-efficacy based upon ethnic group. Similar trends were observed by Zusho et al. (2003) with regard to performance in chemistry. They reported that the self-efficacy of students who were “low achievers” in chemistry dropped sharply across a semester, and those who were “average achievers” dropped slightly. In contrast, students who were “high achievers” reported higher self-efficacy at the end of the semester than at the start. Collectively, these two studies demonstrate that students' self-efficacy trends across a semester depend on several factors, some of which may change during the semester. Hence, instructors should be aware and expectant of such differing trends in self-efficacy among their students, especially when evaluating the effectiveness of a novel approach to instruction.
| Scale | Chemistry majors n = 57 mean score (SD) | Non-science majors n = 136 mean score (SD) | Mean differencea (effect sizeb) | p-value |
|---|---|---|---|---|
| a Based on Tukey's post hoc tests. b Effect size represented by Cohen's d – small (0.20), medium (0.50), large (0.80) (Cohen, 1992). | ||||
| Intitial interest (feeling) | 4.14 (0.84) | 3.19 (0.74) | 0.95 (1.20) | <0.001 |
| Initial interest (value) | 4.38 (0.77) | 3.91 (0.62) | 0.47 (0.67) | <0.001 |
| Maintained interest (feeling) | 3.73 (0.94) | 2.93 (0.81) | 0.80 (0.91) | <0.001 |
| Maintained interest (value) | 3.95 (0.91) | 3.38 (0.79) | 0.57 (0.67) | <0.001 |
| Self-efficacy (time 1) | 3.49 (0.62) | 3.17 (0.59) | 0.32 (0.53) | 0.005 |
The overall models for the two components of initial interest (feeling- and value-related) were significant for the initial sample, F(3,293) = 20.87, p < 0.001, and F(3,293) = 7.12, p < 0.001, respectively. The assumption of homogeneity of variances was not violated for any of the tests performed, as indicated by non-significant results using Levene's test (p > 0.05). Post hoc analyses with Tukey tests revealed that chemistry majors reported higher feeling- (M1 = 4.14, SD1 = 0.84) and value-related interest (M2 = 4.38, SD2 = 0.77) than non-science majors (M1 = 3.19, SD1 = 0.74, M2 = 3.91, SD2 = 0.62) (see Table 6). The same trend was evident in the post-semester data with the two components of maintained interest being significant for the initial sample, F(3,293) = 12.93, p < 0.001, and F(3,293) = 7.72, p < 0.001, respectively. Post hoc analyses revealed that chemistry majors' reported higher maintained feeling- (M1 = 3.73, SD1 = 0.94) and value-related interest (M2 = 3.95, SD2 = 0.91) than non-science majors (M1 = 2.93, SD1 = 0.81, M2 = 3.38, SD2 = 0.79). All differences reported were significant at p < 0.01. The corresponding effect sizes (Cohen's d) are considered medium to large (Cohen, 1992).
A sub-sample used in the cross-validation study was also evaluated for differences in self-efficacy and interest by major. Participants did not differ in self-efficacy, at either pre or post-semester. Participants did, however, report different levels of interest (feeling and value) based on major at the start of the semester, F(3,174) = 9.30, p < 0.001, and F(3,174) = 4.59, p = 0.004, respectively. The same was true for maintained interest (feeling and value) at the end of the semester, F(3,174) = 6.53, p < 0.001, and F(3,293) = 4.68, p = 0.004, respectively. Based on our prior results reported above, we hypothesized that chemistry majors would report higher levels of interest than non-science majors. Hence, we performed planned contrasts to test this hypothesis. Results indicate that chemistry majors showed significantly more initial feeling-related interest, t(171) = 4.56, p < 0.001, and value-related interest, t(171) = 3.66, p < 0.001, than non-science majors. The same was true for maintained-feeling, t(171) = 3.67, p < 0.001, and maintained-value interests, t(171) = 2.93, p = 0.004 (see Appendix, ESI†).
These data support the findings from the study by Barbera et al. (2008) concerning personal interest among chemistry majors versus other majors. In addition, these results expand upon the findings reported by Uzuntiryaki and Aydin (2009) whereby “[chemistry] majors scored higher than non-majors [on self-efficacy for cognitive skills]; however, they did not appear significant.” Most importantly, these data demonstrate the ability of the modified items and revised scales to discriminate between populations of students who would be expected to score differently on self-efficacy and interest in chemistry.
Pre and post-semester comparisons of mean self-efficacy scores were analyzed by major (Table 7). These results indicate that all students with a declared major (n = 287) reported higher self-efficacy at the end of the semester than at the beginning of the semester, regardless of their major. All differences were significant at p < 0.05. Chemistry majors reported improved self-efficacy at the end of the semester (M2 = 3.70) compared to the start of the semester (M1 = 3.18), t(56) = 5.39, p < 0.001. The same was true for other science majors (M1 = 2.95, M2 = 3.62), t(93) = 10.03, p < 0.001; and non-science majors (M1 = 2.78, M2 = 3.45), t(135) = 12.49, p < 0.001. The effect sizes for the differences in self-efficacy from pre to post-semester were medium to large among all three groups of majors, ranging from d = 0.71 for chemistry majors to d = 1.0 for other science and non-science majors (Cohen, 1992). These results expand upon those reported in Table 5 by suggesting that students from all majors reported improved self-efficacy in chemistry after a semester of instruction.
| Major | Mean (SD) time 1 | Mean (SD) time 2 | Mean differencea (effect sizeb) | p-value |
|---|---|---|---|---|
| a Based on paired samples t-test. b Effect size represented by Cohen's d – small (0.20), medium (0.50), large (0.80) (Cohen, 1992). | ||||
| Chemistry | 3.17 (0.69) | 3.70 (0.79) | 0.53 (0.71) | <0.001 |
| Other science | 2.95 (0.66) | 3.62 (0.61) | 0.67 (1.00) | <0.001 |
| Non-science | 2.78 (0.65) | 3.44 (0.60) | 0.66 (1.00) | <0.001 |
Discussion
Self-efficacy, interest, and effort beliefs are salient factors associated with academic motivation and are supported by a strong foundation of research (Weiner, 1985; Bandura, 1986; Renninger, 2000; Blackwell et al., 2007). A chemistry-specific set of scales designed to measure interest, self-efficacy, and effort beliefs were administered to a sample of first-semester general chemistry students. The major goal of this study was to establish evidence of validity and reliability for scores from the scales such that they could be used in future studies regarding the impact of various teaching practices on these motivational factors and their relation to course performance. Demonstrating validity and reliability for data generated by a scale or instrument is paramount following any alterations to items or use in a new setting. Absence of such evidence renders the interpretations of scale scores invalid and can lead to misinformed decision-making (Arjoon et al., 2013). In this study, we have presented evidence to support the internal consistency as well as the response process and structural validity of the modified items and revised scales.Validity evidence based on response processes pertains to the agreement between the construct being measured and the actual processes respondents engage in when they generate an answer (AERA, APA, NCME, 1999). In the case of the interest scales, participants routinely described the meaning of the value-related items differently from the feeling-related items. These results, in conjunction with previous findings by Linnenbrink-Garcia et al. (2010), led us to split the larger interest scales into two smaller subscales, which resulted in improved fit of the CFA models. Our revisions to the self-efficacy and effort beliefs scales were guided more by the way students interpreted the items. When participants assign varied interpretations to a particular item, the meaning of that item is obscured and integrity of the score and construct associated with that item is compromised. Items that were found to illicit ambiguous or frequently incongruent responses from participants were flagged as problematic. Those items that also demonstrated lack of fit or redundancy in the CFA models were removed from the corresponding scale. Overall, our interview results for the items retained in each scale suggest that students consistently and adequately understood the meaning of the items.
To demonstrate functionality based on the internal structure of the individual scales, 1-factor CFAs were conducted to examine the degree to which the data fit each hypothesized model. Confirmatory factor analysis allows the researcher to test whether a proposed grouping of items, and the scores associated with them, appropriately describe a latent variable (Brown, 2006). Our preliminary psychometric evaluation of each scale revealed that the model fits were less than satisfactory. Revisions to each scale were informed by our qualitative studies with students from the target population as well as from the modification indices of the initial 1-factor CFA studies. Following revision of the effort beliefs and self-efficacy scales, and splitting of the interest scale into two factors, the model fit for each scale improved dramatically (see Tables 1 and 2). Nearly all fit indices improved to values considered acceptable for the CFAs conducted on data from both time points. The only exceptions to this were the RMSEA values for self-efficacy, and the chi-square values for initial interest and self-efficacy. As stated previously, the model chi-square test is a “badness of fit” index, where a significant result is not desired. Chi-square tests are sensitive to sample size and often, negligible deviations in fit produce significant results with large samples (Brown, 2006). The sample sizes (n = 373, n = 294) in our models would be considered large based on a recommended subject to indicator ratio of 1![[thin space (1/6-em)]](https://www.rsc.org/images/entities/char_2009.gif)
:10 (Bentler and Chou, 1987). Thus, it is not surprising that all of our preliminary models and two of our revised models failed to produce non-significant chi-square values. However, consistent with most studies involving factor analysis, we used approximate fit indices as alternative indicators of adequate model fit. The model describing the self-efficacy scale was the only one that remained problematic after revision, due to the significant chi-square value (p < 0.05) and high RMSEA value (0.10). Like the model chi-square test, the RMSEA value is considered a “badness of fit” index, where lower values are desired (Kline, 2011). Kline (2011) points out that values ≥0.10 might signal a “serious problem” with the model. Consequently, we urge readers to interpret our results for this model with caution, but also to consider that RMSEA values tend to be inflated in models with a small number of indicators (Breivik and Olsson, 2001).
As the revised scales were derived from qualitative and quantitative results with a single population, additional evidence for the structural validity of each revised scale was established using two additional samples. These cross-validation studies were comprised of data collected from students in the same course, at the same institution, during the year following the initial study. The second sample was obtained from students in the same course at a different institution. Supporting evidence for the structural validity of the revised scales was provided by obtaining similar fit indices with both student samples (Kline, 2011). Across all three populations the fit indices remained consistent. With the exception of the RMSEA values for the effort beliefs scale from sample 1 and the self-efficacy scale from both samples, all other indices were within recommended ranges. As the cross-validation samples sizes were larger, all of the SB-scaled chi-squared statistics were significant, however, this is not seen as a threat to the structural validity as this test is highly dependent on sample size (Brown, 2006). The maintained interest scale did have a non-significant chi-square value, however, data for this one scale was derived from a much smaller (n = 175) post-semester population.
To further support the functionality of the interest and self-efficacy scales in a college chemistry setting, we evaluated the extent to which chemistry majors differed from other majors. Based on prior studies, we operated under the assumption that declared chemistry majors would have higher self-efficacy and interest toward chemistry (Barbera et al., 2008; Uzuntiryaki and Aydin, 2009). Our initial data show that chemistry majors began the semester with higher self-efficacy and interest than non-science majors. However, the gap in self-efficacy scores closed by the end of the semester, indicating no significant difference based on major. For the cross-validation sub-sample, the difference in interest scores between chemistry majors and non-science majors was retained, but there was no significant difference in self-efficacy. It is possible that the lower sample size (n = 175), and hence, lower number of chemistry majors (n = 23) compared to the initial sample could be a cause of this discrepancy. Results from the maintained interest scale suggest that chemistry majors continued to have higher levels of interest than non-science majors through the end of the semester. This is certainly plausible and expected, given that enduring interest in a particular subject area has been shown to predict major choice and number of courses taken in that subject area (Harackiewicz et al., 2000, 2008). Taken together, these results suggest that the self-efficacy and interest scales can discriminate between groups for whom it is plausible to expect differences in confidence and interest toward chemistry, providing further validity evidence of the scales (Standard 1.14, AERA, APA, NCME, 1999).
The internal consistency of scales is an estimate of reliability that relates to how well items within a scale describe the same construct (Henson, 2001). Cronbach's alpha is the most commonly reported value of internal consistency, and a cutoff value of 0.70 is often used to indicate moderate internal consistency among items used in classroom rating scales (Murphy and Davidshofer, 2005). Following item modifications and scale revisions, all of our scales had reliability estimates ≥0.77. These results suggest that the items belonging to each scale are consistent with other items in the same scale in describing the specific construct. Consistently high alpha values were also obtained in the cross-validation studies, further supporting the internal consistency of the revised scales with the target population.
Limitations
While the results from our psychometric evaluation of these revised scales suggest that they function well among general chemistry students, we acknowledge several limitations to the study. First, our sample size from time 1 to time 2 in the initial study dropped by 79 participants (21%). While the time 2 sample size (n = 294) remained large enough for factor analysis, the missing participants could represent an important subset of the population (e.g., students who dropped the course). However, as with any study involving multiple collections of data from a single sample, there is always a risk of attrition. While our cross-validation studies provide positive support for the generalizability of the revised scales, we encourage researchers from other institutions to use and further evaluate this set of scales, so that educators can have a more complete understanding of the psychometric properties and generalizability of these scales. Finally, we acknowledge that the meaning attributed to items in the self-efficacy scale may be different among students in the same population. We excluded two items from the scale (items 3 and 8) due to ambiguity and a lack of consensus on the meaning among interview participants. However, these were items that had the most frequently incongruent responses. Participants in our sample did not necessarily assign the exact same meaning to the remaining items. For instance, item 4 reads, How well can you describe the properties of elements by using the periodic table? A student with a strong background in chemistry might, for example, interpret “properties of elements” as electronegativity, ionization energy, and bonding tendencies of elements. On the other hand, a student with a weak background in chemistry may view “properties of elements” as simply the number of protons, neutrons, and electrons in a given element. Thus, students with a strong chemistry background may, in fact, underestimate their ability because they have a deeper understanding of the theories and facets associated with certain tasks in chemistry. Students with less understanding of a chemistry task may inflate their appraisal of their ability due to an oversimplification of the task. In addition, students who are from non-English language backgrounds may interpret self-efficacy items differently than native English speakers (Lee and Fradd, 1998). Self-efficacy items involving tasks that are not completely objective and defined will always leave room for loose interpretations. However, items that are too narrowly focused and specific will lose generalizability and require the instrument to be long and arduous in order to cover the set of topics in a given course. On the other hand, Pajares (1996) cautions against using measures too general by stating that, “omnibus tests…transform self-efficacy into a generalized personality trait rather than a context-specific judgment.” The balance between specificity and generality with efficacy beliefs is a difficult aspect to fully resolve, a sentiment shared by other researchers as well (Tschannen-Moran and Hoy, 2001). We feel that our qualitative data, while limited, offers some useful insight into items from the self-efficacy scale, which would be most imprecisely interpreted by students in our target population. Although some of the items retained in the scale could be also interpreted in several ways, at varying degrees of understanding; our interview data demonstrates that most students had a consistent grasp of what the task meant. Researchers concerned with the interpretability of this and other self-efficacy scales in chemistry could extend upon these findings by conducting interviews with different groups of students. Students with diverse chemistry backgrounds and experiences, as well as those whose first language is not English may provide perspectives on items that are not obvious to the developer and users of the scale. This depth of information should be regarded as absolutely vital for the validity of any inferences drawn from scores from the scale.Implications and future research
Our interest when designing this study stemmed from our desire to study student motivation in chemistry. Due to the limited work in this area, our first step was to evaluate a set of modified items and revised scales, so that various aspects of motivation could be measured for a chemistry-specific population. Time constraints can often prevent instructors from administering lengthy scales or instruments, therefore, we tried to compile scales that would provide a balance between useful data and classroom administration time. Furthermore, student participation and completion rates tend to be lower with longer, more time-consuming instruments (Lichtenstein et al., 2008; Heredia and Lewis, 2012). Additionally, it is crucial that each scale is actually measuring what it's developers have purported it to be measuring. Therefore, it is incumbent upon researchers to thoroughly examine relevant psychometric evidence of such scales prior to, or as part of, their use. With the present work, we show that each of our revised scales to measure self-efficacy, interest, and effort beliefs demonstrate acceptable psychometric properties for use in a general chemistry setting. Additionally, our revised scales, which measure three well-defined latent traits, are comprised of a small number of items. Our revised scales consist of 7 initial interest, 8 maintained interest, 6 effort beliefs, and 6 self-efficacy items. When used together in future studies, this equates to 19 (with initial interest) or 20 (with maintained interest) total items to address three latent traits. By comparison, the CAEQ (measuring three distinct latent traits), and the CSCI (measuring five types of self-concept) are comprised of 69 and 40 items, respectively (Dalgety et al., 2003; Bauer, 2005).Our next step is to now utilize these scales to study the impact of teaching practice on students' self-efficacy, interest, and effort beliefs. Many studies are suggestive of the powerful influence of self-efficacy on performance (Zimmerman et al., 1992; Pajares, 1996; Zusho et al., 2003). To further corroborate these findings, and expand upon them in the college chemistry setting, studies involving measures of self-efficacy and performance together are needed. Even less explored are the relationships among interest, effort beliefs, and performance. There is a particular lack of research involving these latent traits in college level sciences. In future studies, we plan to investigate how different practices affect these motivational factors as well as how these factors affect each other and ultimately students' performance. Therefore, follow-up studies will use a priori path analysis models to evaluate the correlation between scales and their mediation of course performance (Xu et al., 2013).
In addition to our ongoing studies, we offer a few avenues for extension of the current study. A large-scale study could further utilize the power of structural equation modeling through invariance (or measurement equivalence) analysis. Invariance analysis allows the researcher to test whether a proposed model is equivalent across different groups of participants. One form of testing for invariance is to use a multi-group CFA, whereby the researcher is able to test for the equivalence of the measurement and structural solution (Brown, 2006). Put simply, invariance analysis can inform the researcher as to whether the same trait is being measured across different groups (race, gender, major), which is an important consideration for the validity of an instrument (Hutchinson et al., 2008).
Instructors who are interested in gauging the motivational atmosphere of their classes might find our chemistry-specific scales useful. Students' beliefs about motivation and effort precede and govern their actions in the course. At the start of a semester, an instructor might want to have knowledge of his or her students' interest and confidence toward chemistry for the purpose of tailoring certain aspects of the course to their group of students. Due to the brevity of the scales, data could be collected at multiple time points throughout the semester with minimal time commitment. This could be especially useful if an innovative instructional strategy were to be implemented. The instructor could evaluate the impact of their instructional strategy on dimensions beyond course performance measures. This would be informative as performance measures alone tell instructors nothing about a student's motivational or affective disposition toward the course, which are vital components of a student's academic success (Zusho et al., 2003). We encourage educators who employ novel instructional strategies to consider measuring the motivational and affective processes of their students, in order to add to the current understanding of the impacts of these strategies. As stated in the 2012 DBER report, “the interplay between faculty behavior [i.e., teaching strategies] and student affect merits further exploration.” We feel that the present work aids instructors in this exploration by providing measurement tools adapted for college chemistry, founded on prevailing theories from educational psychology, and subjected to the rigor of a thorough psychometric evaluation.
References
- AERA, APA, NCME (1999), Standards for educational and psychological testing, Washington, DC: American Educational Research Association.
- Ames C., (1992), Classrooms – Goals, Structures, and Student Motivation, J. Educ. Psychol., 84(3), 261–271.
- Anderman E. M. and Young A. J., (1994), Motivation and strategy use in science: individual differences and classroom effects, J. Res. Sci. Teach., 31(8), 811–831.
- Andrew S., (1998), Self-efficacy as a predictor of academic performance in science, J. Adv. Nurs., 27(3), 596–603.
- Arjoon J. A., Xu X. and Lewis J. E., (2013), Understanding the State of the Art for Measurement in Chemistry Education Research: Examining the Psychometric Evidence, J. Chem. Educ., 90(5), 536–545.
- Bandura A., (1977), Self-efficacy: toward a unifying theory of behavioral change, Psychol. Rev., 84(2), 191–215.
- Bandura A., (1986), Social foundations of thought and action: a social cognitive theory, Englewood Cliffs, NJ: Prentice Hall.
- Bandura A., Blanchard E. B. and Ritter B., (1969), Relative efficacy of desensitization and modeling approaches for inducing behavioral, affective, and attitudinal changes, J. Pers. Soc. Psychol., 13(3), 173.
- Barbera J. and VandenPlas J. R. (2011), All Assessment Materials Are Not Created Equal: The Myths about Instrument Development, Validity, and Reliability, Investigating Classroom Myths through Research on Teaching and Learning, American Chemical Society, vol. 1074, pp. 177–193.
- Barbera J., Adams W. K., Wieman C. E. and Perkins K. K., (2008), Modifying and validating the Colorado Learning Attitudes about Science Survey for use in chemistry, J. Chem. Educ., 85(10), 1435–1439.
- Bauer C. F., (2005), Beyond “student attitudes”: chemistry self-concept inventory for assessment of the affective component of student learning, J. Chem. Educ., 82(12), 1864–1870.
- Bentler P. M., (1990), Comparative fit indexes in structural models, Psychol. Bull., 107(2), 238–246.
- Bentler P. M. and Chou C.-P., (1987), Practical issues in structural modeling, Sociol. Methods Res., 16(1), 78–117.
- Betz N. E. and Hackett G., (1983), The Relationship of mathematics self-efficacy expectations to the selection of science-based college majors, J. Vocat. Behav., 23(3), 329–345.
- Blackwell L. (2002), Psychological mediators of student achievement during the transition to junior high school: the role of implicit theories, Unpublished Doctoral Dissertation, Columbia University.
- Blackwell L. S., Trzesniewski K. H. and Dweck C. S., (2007), Implicit theories of intelligence predict achievement across an adolescent transition: a longitudinal study and an intervention, Child Dev., 78(1), 246–263.
- Bong M. and Clark R. E., (1999), Comparison between self-concept and self-efficacy in academic motivation research, Educ. Psychol., 34(3), 139–153.
- Breivik E. and Olsson U. H., (2001), Adding variables to improve fit: the effect of model size on fit assessment in LISREL, Struct. Equ Modeling, 169–194.
- Britner S. L. and Pajares F., (2001), Self-efficacy beliefs, motivation, race, and gender in middle school science, J. Women Minor. Sci. Eng., 7(4), 15.
- Brophy J. E., (2010), Motivating students to learn, New York: Routledge.
- Brown T. A., (2006), Confirmatory factor analysis for applied research, New York: Guilford Press.
- Brown C. E., Henry M. L., Barbera J. and Hyslop R. M., (2012), A Bridge between Two Cultures: Uncovering the Chemistry Concepts Relevant to the Nursing Clinical Practice, J. Chem. Educ., 89(9), 1114–1121.
- Browne M. W. and Cudeck R., (1992), Alternative ways of assessing model fit, Sociol. Methods Res., 21(2), 230–258.
- Chou C.-P., Bentler P. M. and Satorra A., (1991), Scaled test statistics and robust standard errors for non-normal data in covariance structure analysis: a Monte Carlo study, Br. J. Math. Stat. Psychol., 44(2), 347–357.
- Cohen J., (1992), Statistical power analysis, Curr. Dir. Psychol. Sci., 1(3), 98–101.
- Creswell J. W., (2013), Qualitative inquiry and research design: choosing among five approaches, 3rd edn, Thousand Oaks, CA: Sage.
- Dalgety J. and Coll R. K., (2006), Exploring first-year science students' chemistry self-efficacy, Int. J. Sci. Math. Educ., 4(1), 97–116.
- Dalgety J. and Salter D., (2002), The development of the chemistry attitudes and experiences questionnaire (CAEQ), Chem. Educ. Res. Pract., 3(1), 19–32.
- Dalgety J., Coll R. K. and Jones A., (2003), Development of Chemistry Attitudes and Experiences Questionnaire (CAEQ), J. Res. Sci. Teach., 40(7), 649–668.
- Dawes J., (2008), Do data characteristics change according to the number of scale points used? An experiment using 5 point, 7 point and 10 point scales, Int. J. Market Res., 51(1), 61–77.
- Dweck C. S., (1986), Motivational Processes Affecting Learning, Am. Psychol., 41(10), 1040–1048.
- Dweck C. S., (2000), Self-theories: their role in motivation, personality, and development, Psychology Press.
- Dweck C. S. (2012), Implicit Theories, in Lange P. V., Kruglanski A. and Higgins T. (ed.), Handbook of Theories of Social Psychology, London: Sage, vol. 2, pp. 43–61.
- Dweck C. S. and Leggett E. L., (1988), A social cognitive approach to motivation and personality, Psychol. Rev., 95(2), 256–273.
- Dweck C. S. and Sorich L., (1999), Mastery-oriented thinking, Coping, 232–251.
- Eccles J. S. and Wigfield A., (2002), Motivational beliefs, values, and goals, Ann. Rev. Psychol., 53(1), 109–132.
- Elliot A. J., (1999), Approach and avoidance motivation and achievement goals, Educ. Psychol., 34(3), 169–189.
- Finney S. J. and DiStefano C., (2006), Non-normal and categorical data in structural equation modeling, Struct. Equ. Modeling, 269–314.
- Glynn S. M., Taasoobshirazi G. and Brickman P., (2009), Science Motivation Questionnaire: Construct Validation With Nonscience Majors, J. Res. Sci. Teach., 46(2), 127–146, DOI: 10.1002/tea.20267.
- Goodman S., Jaffer T., Keresztesi M., Mamdani F., Mokgatle D., Musariri M., Pires J. and Schlechter A., (2011), An investigation of the relationship between students' motivation and academic performance as mediated by effort, South African Journal of Psychology, 41(3), 373–385.
- Gore P. A., (2006), Academic self-efficacy as a predictor of college outcomes: two incremental validity studies, J. Career Assessment, 14(1), 92–115.
- Guay F., Ratelle C. F. and Chanal J., (2008), Optimal learning in optimal contexts: the role of self-determination in education, Can. Psychol., 49(3), 233–240.
- Gungor A., Eryılmaz A. and Fakıoglu T., (2007), The relationship of freshmen's physics achievement and their related affective characteristics, J. Res. Sci. Teach., 44(8), 1036–1056.
- Hackett G. and Betz N. E., (1981), A self-efficacy approach to the career development of women, J. Vocat. Behav., 18(3), 326–339.
- Hancock G. R., (1999), A sequential Scheffé-type respecification procedure for controlling type I error in exploratory structural equation model modification, Structural Equation Modeling: A Multidisciplinary Journal, 6(2), 158–168.
- Hansford B. C. and Hattie J. A., (1982), The relationship between self and achievement/performance measures, Rev. Educ. Res., 52(1), 123–142.
- Harackiewicz J. M., Barron K. E., Tauer J. M., Carter S. M. and Elliot A. J., (2000), Short-term and long-term consequences of achievement goals: predicting interest and performance over time, J. Educ. Psychol., 92(2), 316–330.
- Harackiewicz J. M., Durik A. M., Barron K. E., Linnenbrink-Garcia L. and Tauer J. M., (2008), The role of achievement goals in the development of interest: reciprocal relations between achievement goals, interest, and performance, J. Educ. Psychol., 100(1), 105–122.
- Henson R. K., (2001), Understanding internal consistency reliability estimates: a conceptual primer on coefficient alpha, Meas. Eval. Counsel. Dev., 34, 177–189.
- Heredia K. and Lewis J. E., (2012), A Psychometric Evaluation of the Colorado Learning Attitudes about Science Survey for Use in Chemistry, J. Chem. Educ., 89(4), 436–441.
- Hidi S., (1990), Interest and its contribution as a mental resource for learning, Rev. Educ. Res., 60(4), 549–571.
- Hidi S. and Baird W., (1986), Interestingness – A neglected variable in discourse processing, Cognitive Sci., 10(2), 179–194.
- Hidi S. and Harackiewicz J. M., (2000), Motivating the academically unmotivated: a critical issue for the 21st century, Rev. Educ. Res., 70(2), 151–179.
- Hidi S. and Renninger K. A., (2006), The four-phase model of interest development, Educ. Psychol., 41(2), 111–127.
- Hong Y. Y., Coleman J., Chan G., Wong R. Y., Chiu C. Y., Hansen I. G., Lee S. I., Tong Y. Y. and Fu H. Y., (2004), Predicting intergroup bias: the interactive effects of implicit theory and social identity, Pers. Soc. Psychol. B., 30(8), 1035–1047.
- Hooper D., Coughlan J. and Mullen M., (2008), Structural equation modelling: guidelines for determining model fit, Electron. J. Bus. Res. Methods, 6(1), 53–60.
- Hu L. T. and Bentler P. M. (1995), Evaluating model fit, in Hoyle R. H. (ed.), Structural equation modeling: concepts, issues, and applications, London, Sage, pp. 76–99.
- Hu L. T. and Bentler P. M., (1999), Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives, Struct. Equ. Modeling, 6(1), 1–55.
- Huck S., (2012), Reading Statistics and Research, 6th edn, Boston, MA: Pearson.
- Hulleman C. S., Durik A. M., Schweigert S. B. and Harackiewicz J. M., (2008), Task values, achievement goals, and interest: an integrative analysis, J. Educ. Psychol., 100(2), 398–416.
- Hutchinson S. R., Raymond K. J. and Black K. R., (2008), Factorial invariance of a campus climate measure across race, gender, and student classification, J. Divers. High. Educ., 1(4), 235–250.
- Jones B. D., Wilkins J. L. M., Long M. H. and Wang F. H., (2012), Testing a motivational model of achievement: how students' mathematical beliefs and interests are related to their achievement, Eur. J. Psychol. Educ., 27(1), 1–20.
- Jöreskog K. G. and Sörbom D., (2006), LISREL 8.80 for Windows, Computer Software, Lincolnwood, IL: Scientific Software International, Inc.
- Kline R. B., (2011), Principles and practice of structural equation modeling, 3rd edn, New York: Guilford Press.
- Knafl K., Deatrick J., Gallo A., Holcombe G., Bakitas M., Dixon J. and Grey M., (2007), The analysis and interpretation of cognitive interviews for instrument development, Res. Nurs. Health, 30(2), 224–234.
- Lee O. and Fradd S. H., (1998), Science for all, including students from non-English-language backgrounds, Educ. Researcher, 27(4), 12–21.
- Lent R. W. and Hackett G., (1987), Career self-efficacy: empirical status and future directions, J. Vocat. Behav., 30(3), 347–382.
- Lent R. W., Brown S. D. and Larkin K. C., (1984), Relation of self-efficacy expectations to academic achievement and persistence, J. Couns. Psychol., 31(3), 356.
- Lent R. W., Brown S. D. and Hackett G., (1994), Toward a unifying social cognitive theory of career and academic interest, choice, and performance, J. Vocat. Behav., 45(1), 79–122.
- Lewis S. E., Shaw J. L., Heitz J. O. and Webster G. H., (2009), Attitude Counts: Self-Concept and Success in General Chemistry, J. Chem. Educ., 86(6), 744.
- Lichtenstein M. J., Owen S. V., Blalock C. L., Liu Y., Ramirez K. A., Pruski L. A., Marshall C. E. and Toepperwein M. A., (2008), Psychometric reevaluation of the scientific attitude inventory-revised (SAI-II), J. Res. Sci. Teach., 45(5), 600–616.
- Linnenbrink-Garcia L., Durik A. M., Conley A. M., Barron K. E., Tauer J. M., Karabenick S. A. and Harackiewicz J. M., (2010), Measuring Situational Interest in Academic Domains, Educ. Psychol. Meas., 70(4), 647–671.
- Lopez F. G. and Lent R. W., (1992), Sources of mathematics self-efficacy in high school students, Career Dev. Quart., 41(1), 3–12.
- McCoach D. B. and Siegle D., (2003), Factors that differentiate underachieving gifted students from high-achieving gifted students, Gift. Child Quart., 47(2), 144–154.
- Multon K. D., Brown S. D. and Lent R. W., (1991), Relation of self-efficacy beliefs to academic outcomes: a meta-analytic investigation, J. Couns. Psychol., 38(1), 30.
- Murphy K. R. and Davidshofer C. O., (2005), Psychological testing: principles and applications, New Jersey: Prentice Hall.
- Nakhleh M. B., (1992), Why some students don't learn chemistry: chemical misconceptions, J. Chem. Educ., 69(3), 191–196.
- National Research Council, (2012), Discipline-Based Education Research: Understanding and Improving Learning in Undergraduate Science and Engineering, in Singer S. R., Nielsen N. R. and Schweingruber H. A. (ed.), Committee on the Status, Contributions, and Future Direction of Discipline-Based Education Research. Board on Science Education, Division of Behavioral and Social Sciences and Education, Washington, D.C.
- Nicholls J. G., Cheung P. C., Lauer J. and Patashnick M., (1989), Individual differences in academic motivation: perceived ability, goals, beliefs, and values, Learn. Individ. Differ., 1(1), 63–84.
- Pajares F., (1996), Self-efficacy beliefs in academic settings, Rev. Educ. Res., 66(4), 543–578.
- Pajares F. and Kranzler J., (1995), Self-efficacy beliefs and general mental ability in mathematical problem-solving, Contemp. Educ. Psychol., 20(4), 426–443.
- Pajares F. and Miller M. D., (1994), Role of self-efficacy and self-concept beliefs in mathematical problem solving: a path analysis, J. Educ. Psychol., 86(2), 193–203.
- Parker P. D., Marsh H. W., Ciarrochi J., Marshall S. and Abduljabbar A. S., (2014), Juxtaposing math self-efficacy and self-concept as predictors of long-term achievement outcomes, Educ. Psychol., 34(1), 29–48.
- Pintrich P. R., (2003), A motivational science perspective on the role of student motivation in learning and teaching contexts, J. Educ. Psychol., 95(4), 667–686.
- Pintrich P. R. and De Groot E. V., (1990), Motivational and self-regulated learning components of classroom academic performance, J. Educ. Psychol., 82(1), 33–40.
- Pintrich P. R., Smith D. A. F., García T. and McKeachie W. J., (1993), Reliability and predictive validity of the Motivated Strategies for Learning Questionnaire (MSLQ), Educ. Psychol. Meas., 53(3), 801–813.
- Raker J. R., Emenike M. E. and Holme T. A., (2013), Using Structural Equation Modeling To Understand Chemistry Faculty Familiarity of Assessment Terminology: Results from a National Survey, J. Chem. Educ., 90(8), 981–987.
- Renninger K. (2000), Individual interest and its implications for understanding intrinsic motivation, in Sansone C. and Harackiewicz J. (ed.), Intrinsic and extrinsic motivation: the search for optimal motivation and performance, San Diego, CA: Academic Press, pp. 373–404.
- Renninger K. and Hidi S. (2002), Student interest and achievement: developmental issues raised by a case study, in Wigfield A. and Eccles J. (ed.), Development of achievement motivation, San Diego, CA: Academic Press, pp. 175–191.
- Satorra A. and Bentler P. M. (1994), Corrections to test statistics and standard errors in covariance structure analysis, in von Eye A. and Clogg C. C. (ed.), Latent variables analysis: applications for developmental research, Thousand Oaks, CA: Sage Publications, Inc, pp. 399–419.
- Schermelleh-Engel K., Moosbrugger H. and Müller H., (2003), Evaluating the fit of structural equation models: tests of significance and descriptive goodness-of-fit measures, Methods Psychol. Res. Online, 8(2), 23–74.
- Schiefele U., (1991), Interest, learning, and motivation, Educ. Psychol., 26(3–4), 299–323.
- Schiefele U., (1999), Interest and learning from text, Sci. Stud. Read., 3(3), 257–279.
- Schunk D. H., (1991), Self-efficacy and academic motivation, Educ. Psychol., 26(3–4), 207–231.
- Schunk D. and Pajares F. (2002), The development of academic self-efficacy, in Wigfield A. and Eccles J. (ed.), Development of achievement motivation, San Diego, CA: Academic Press, pp. 16–29.
- Singh K., Granville M. and Dika S., (2002), Mathematics and science achievement: effects of motivation, interest, and academic engagement, J. Educ. Res., 95(6), 323–332.
- Sorich L. and Dweck C. S. (1997), Reliability data for new scales measuring students' beliefs about effort and responses to failure, Unpublished Raw Data, Columbia University.
- Steiger J. H., (1990), Structural model evaluation and modification: an interval estimation approach, Multivariate Behavioral Research, 25(2), 173–180.
- Stipek D. and Gralinski J. H., (1996), Children's beliefs about intelligence and school performance, J. Educ. Psychol., 88(3), 397–407.
- Taasoobshirazi G. and Glynn S. M., (2009), College Students Solving Chemistry Problems: A Theoretical Model of Expertise, J. Res. Sci. Teach., 46(10), 1070–1089.
- Tai R. H., Sadler P. M. and Loehr J. F., (2005), Factors influencing success in introductory college chemistry, J. Res. Sci. Teach., 42(9), 987–1012.
- Tschannen-Moran M. and Hoy A. W., (2001), Teacher efficacy: capturing an elusive construct, Teach. Teach. Educ., 17(7), 783–805.
- Tuan H. L., Chin C. C. and Shieh S. H., (2005), The development of a questionnaire to measure students' motivation towards science learning, Int. J. Sci. Educ., 27(6), 639–654.
- Tucker L. R. and Lewis C., (1973), A reliability coefficient for maximum likelihood factor analysis, Psychometrika, 38(1), 1–10.
- Uitto A., Juuti K., Lavonen J. and Meisalo V., (2006), Students' interest in biology and their out-of-school experiences, J. Biol. Educ., 40(3), 124–129.
- Uzuntiryaki E. and Aydin Y. C., (2009), Development and Validation of Chemistry Self-Efficacy Scale for College Students, Res. Sci. Educ., 39(4), 539–551.
- Villafane S. M., Garcia C. A. and Lewis J. E., (2014), Exploring diverse students' trends in chemistry self-efficacy throughout a semester of college-level preparatory chemistry, Chem. Educ. Res. Pract., 15(2), 114–127.
- Weiner B., (1985), An attributional theory of achievement-motivation and emotion, Psychol. Rev., 92(4), 548–573.
- Wigfield A. and Eccles J. S. (2002), Introduction, in Wigfield A. and Eccles J. S. (ed.), Development of achievement motivation, San Diego, CA: Academic Press, pp. 1–10.
- Wren D. and Barbera J., (2013), Gathering Evidence for Validity during the Design, Development, and Qualitative Evaluation of Thermochemistry Concept Inventory Items, J. Chem. Educ., 90(12), 1590–1601.
- Xu X. and Lewis J. E., (2011), Refinement of a Chemistry Attitude Measure for College Students, J. Chem. Educ., 88(5), 561–568.
- Xu X., Villafane S. M. and Lewis J. E., (2013), College students' attitudes toward chemistry, conceptual knowledge and achievement: structural equation model analysis, Chem. Educ. Res. Pract., 14, 188–200.
- Zeldin A. L., Britner S. L. and Pajares F., (2008), A comparative study of the self-efficacy beliefs of successful men and women in mathematics, science, and technology careers, J. Res. Sci. Teach., 45(9), 1036–1058.
- Zimmerman B. J., (2000), Self-efficacy: an essential motive to learn, Contemporary Educational Psychology, 25(1), 82–91.
- Zimmerman B. J., Bandura A. and Martinez-Pons M., (1992), Self-motivation for academic attainment: the role of self-efficacy beliefs and personal goal setting, Am. Educ. Res. J., 29(3), 663–676.
- Zusho A., Pintrich P. R. and Coppola B., (2003), Skill and will: the role of motivation and cognition in the learning of college chemistry, Int. J. Sci. Educ., 25(9), 1081–1094.
Footnote |
| † Electronic supplementary information (ESI) available. See DOI: 10.1039/c4rp00152d |
| This journal is © The Royal Society of Chemistry 2015 |