My question is: The article uses the term 'self-efficacy' several times. Briefly describe what you understand by self-efficacy in the context of the paper and whether you think it has been clearly explained.

 

1. Introduction

 

According to 2009 US Department of Education data, 44% of undergraduates initially declaring an engineering major in 2004 had switched out of the major within 4 years (National Center for Education Statistics 2009). Although this percentage compares somewhat favourably to attrition from other STEM disciplines, it presents a concern for engineering educators, and has been a prominent subject of discussion among educators and policymakers over some 20 years. Understanding the experiences that contribute to - or detract from - undergraduates' persis[1]tence in the engineering major is critical to improving retention rates. Factors that have been shown to predict persistence in the major for engineering undergraduates include a wide array of both academic and extra-academic factors: GPA and intrinsic motivation (French, Immekus, and Oakes 2005); self-efficacy, personal and academic support within the engineering environment, and progress towards academic goals (Lent et al. 2007); class size, class attendance, access to the instructor, and peer collaboration (Martínez-Caro and Campuzano-Bolarín 2011); and use of academic support services (Jackson, Gardner, and Sullivan 1993); among others. *Corresponding author.  © 2015 SEFI European Journal of Engineering Education 381 Existing research addresses both non-malleable factors, such as gender and high school grades (e.g. French, Immekus, and Oakes 2005), and malleable factors, such as confidence and interac[1]tions with instructors and peers (e.g. Litzler and Young 2012). Malleable factors are generally of more interest to educators and policymakers, since by definition they can be modified (although not necessarily easily) to help ensure more successful student outcomes. This study addresses two malleable factors which, taken together, can be conceptualised as an indicator of the students' perceived 'fit' between themselves and the program - the degree to which students feel connected to academic staff/faculty members, and the degree to which students feels well-aligned with their engineering peers - and their relationship to both self-efficacy for engineering and satisfaction with engineering coursework. 

 

2. Background

 

2.1. Student-instructor relationships 

A number of studies have shown that the better college students' relationships are with their teachers, the better are their overall college experiences and learning outcomes. Pascarella and Terenzini (1978, 1979), in early studies on college instructor-student interaction, found that informal contact with instructors had a significant relationship with persistence in college among first-year students, even after controlling for a number of factors known to impact student reten[1]tion. Other studies have addressed particular aspects of the undergraduate experience that are impacted by instructor contact. For instance, Kuh and Hu (2001) found that the quality of student-instructor interaction had an effect on students' effort in college, which in turn had an effect on their satisfaction and learning gains. Umbach and Wawrzynski (2005) likewise found that greater instructor-student interaction was related to a more engaged student body, and Anaya and Cole (2001) found that interaction with faculty members/academic staff predicted academic performance. Komarraju, Musulkin, and Bhattacharya (2010) found that students' academic self[1]concept was related to their relationships with instructors, including their sense of instructors' approachability, accessibility, and respect for students. Similarly, Eimers (2001) found that stu[1]dents who reported better relationships with instructors were more likely to feel they had made strides in math and science, as well as in problem-solving ability, general intellectual ability, and career development; Graunke and Woosley (2005) found satisfaction with instructor interaction to be related to academic success; and Lundberg and Schreiner (2004) found positive and fre[1]quent interactions with faculty to predict perceived learning gains. Kim and Sax (2009) found interaction with faculty to be related to a wide array of cognitive and non-cognitive outcomes. In the engineering realm in particular, frequent instructor-student interaction has been shown to have a positive influence on the confidence of engineering students as it relates to profes[1]sional and interpersonal skills (Bjorklund, Parente, and Sathianathan 2004; Chen, Lattuca, and Hamilton 2008). Based on previous research demonstrating the positive impact of student-instructor interaction, we would expect connectedness with instructors to predict student satisfaction with engineering. 

 

2.2. Peer relationships 

Astin has written that peers are 'the single most potent source of influence' (1993, 398) in col[1]lege students' lives, and in many respects this claim has been supported in the scholarly literature. 382 M. Micari and P. Pazos General engagement with peers has been studied within several broad areas of research, including academic and social integration, as well as belongingness. Academic integration has been defined as 'development of a strong affiliation with the college academic environment both in the class[1]room and outside of class' (Nora 1993, 235); and social integration as 'development of a strong affiliation with the college social environment both in the classroom and outside of class' (Nora 1993, 237). Integration into the college community generally, which encompasses engagement with peers, has been shown in numerous studies to predict student persistence in college (Berger and Milem 1999; Bers and Smith 1991; Pascarella, Smart, and Ethington 1986; Pascarella and Terenzini 1979), as well as academic achievement (McKenzie and Schweitzer 2001; Próspero and Vohra-Gupta 2007). Belongingness - which has been defined as 'interpersonal relatedness most dissimilar to loneliness and most closely associated with social support' (Hoffman et al. 2002, 229) and for which peer engagement is critical - is also an important predictor of stu[1]dent outcomes such as self-efficacy, intention to persist, and grades (Freeman, Anderman, and Jensen 2007; Good, Rattan, and Dweck 2012; Hausmann, Schofield, and Woods 2007; Ostrove and Long 2007; Walton and Cohen 2011). Additionally, several studies have more specifically found peer connections to predict persistence in college (Nora et al. 1996; Porter and Swing 2006). Alignment with peers - the sense that one's peers reflect one's own interests and values - is narrower than and conceptually distinct from integration and belongingness. Peer alignment in college has not been studied per se, but alignment with the overall environment - what is termed person-environment fit - has been studied widely, mainly in vocational contexts but also in university contexts (Feldman, Smart, and Ethington 1999). This line of research has found that when the individual's personality matches the prevailing orientation of the environment (e.g. academic department), the individual will have higher levels of satisfaction and longevity within that environment (Feldman, Smart, and Ethington 1999; Porter and Umbach 2006; Smart, Feldman, and Ethington 2000). Based on the findings that both peer engagement and environmental alignment predict positive student outcomes, we would expect alignment with peers to predict satisfaction with the academic experience of engineering students. 

 

2.3. Self-efficacy 

In the academic realm, self-efficacy refers to the beliefs a learner holds about his or her own ability to perform well in a particular learning area (Bandura 1997). Previous research has shown that high levels of self-efficacy positively affect academic motivation and achievement (Pajares 1996; Schunk 1995), as well as persistence in problem-solving (Bouffard-Bouchard, Parent, and Parivee 1991) and deep learning strategies that promote understanding (Liem, Lau, and Nie 2008). Self-efficacy has also been shown to be a precursor of interest in particular subject matter (Silvia 2003) and a predictor of cognitive engagement (Walker, Greene, and Mansell 2006) and of 'optimal experience', or a state of intense concentration and satisfaction (Csikszentmihalyi 1990), in learning activities. Sources of self-efficacy are theorised to include personal mastery experiences, vicarious mastery experiences, persuasion by others, and particular emotional states (Bandura 1997), but a wide range of particular factors have been studied as potential sources of self-efficacy. Students' interaction with instructors appears to be a predictor of self-efficacy (Vogt 2008), with low levels of interaction predicting lower levels of self-efficacy in students. Peers are another key source of input for academic self-efficacy beliefs, acting as models of what is possible for the individual (Schunk 1987), and prior research has suggested that the sense of belonging within the classroom, linked to interaction with both faculty and peers, is a predictor of academic self-efficacy (Freeman, Anderman, and Jensen 2007). European Journal of Engineering Education 383 

 

3. Hypotheses 

 

First, based on previously established connections between self-efficacy and overall academic experience, as well as between instructor connection and fit on the one hand and academic experience on the other, we expect that self-efficacy, instructor connectedness, and peer alignment will all predict satisfaction with the engineering academic experience. Further, based on the previously demonstrated relationships between self-efficacy on the one hand and peer interaction, instructor interaction, and belongingness on the other, we hypothesise that fitting in (through peer alignment and instructor connections) promotes satisfaction via self-efficacy; in other words, fitting in increases one's belief that one can do well, which in turn increases satisfaction. We propose the following hypotheses: 

1 Self-efficacy will predict satisfaction with the major. 

(2) Peer alignment will predict satisfaction with the major. 

(3) Instructor connectedness will predict satisfaction with the major. 

(4) Self-efficacy will mediate the relationship between peer alignment and satisfaction. 

(5) Self-efficacy will mediate the relationship between instructor connectedness and satisfaction. 

 

4. Methods 

 

This study took place in a Fluid Mechanics course in a highly selective university in the Midwest region of the United States. This course was chosen because it is a required course for students in two engineering majors (biomedical and mechanical engineering), and thus would contain a fairly representative sample of engineering students in those majors. Participants were recruited from this course over two consecutive years, during the spring 2012 and spring 2013 quarters. Participants took a survey consisting of 47 items, 16 of which were used in this study. Items used a 5-point Likert scale, with 'strongly disagree' (1) and 'strongly agree' (5) as anchors. Students took the survey in both online and paper formats. No incentives were offered for participating in the study, and the research team was not involved in teaching the classes included in the study. A total of 275 students were invited to participate, with a final sample of 135 students. Gender breakdown was 53% males and 47% females; 87% were sophomores (second-year university students) and 13% juniors (third-year students). In terms of ethnicity, 59% where White, 26% Asian American, 7% Hispanic American, 6% African American, and 2% other. All students completing the survey were included in the study. 

 

5. Variables 

 

Prior to establishing the final measures, we used exploratory factor analysis (a method of identifying underlying factors, or variable clusters, within a larger set of variables; Fabrigar 2011) to evaluate the underlying factor structure of the questionnaire items. The analysis was con[1]ducted using SPSS version 22. We used the maximum likelihood method with oblimin rotation (a particular approach to exploratory factor analysis) to determine the factor structure and the scree plot (a graphical display of the variance of each factor) to determine the optimal number of factors. This analysis yielded four factors, which can be considered the underlying dimensions that are measured by the survey questions. A 0.5 loading was used as the lower limit for inclusion in the scales. Table 1 (see Appendix) shows the loadings (a measure of the correlation between an individual item and a factor) and Chronbach's alpha value (a measure of internal 384 M. Micari and P. Pazos consistency) for each of the factors. Based on the loadings, the composition of each of the underlying dimensions was extracted leading to the following variables: satisfaction with engineering, instructor connectedness, peer alignment, and self-efficacy. 

 

5.1. Satisfaction with engineering 

Because persistence itself is hard to measure, requiring longitudinal studies and careful data tracking, satisfaction with the major is often used as a study outcome variable, and has been shown to predict persistence (Borden 1995; Litzler and Young 2012; Sanders and Burton 1996; Schreiner and Nelson 2013-14). In this study, we use satisfaction with the major as an outcome variable. The Satisfaction variable, adapted from the Academic Pathways of People Learning Engineering Survey (APPLES) (Sheppard et al. 2010), is composed of six items. See Table 1 (Appendix) for items and factor loadings. Sample items from this scale include 'I made the right decision when I decided to study engineering' and 'On the whole, I think engineering classes are fun'. The calculated reliability for this variable is α = 0.93. 

 

5.2. Instructor connectedness 

Instructor connectedness was defined as the extent to which students feel supported, connected to, and guided by faculty members/academic staff. This variable comprised two survey items, developed by the authors. A sample item for this scale is 'I have gotten to know one or more engineering faculty members fairly well'. See Table 1 (Appendix) for items and loadings. The reliability of this scale was α = 0.74. 

 

5.3. Peer alignment 

The Peer Alignment variable was defined as the extent to which students report a sense of fit with their engineering classmates in terms of common interests and relatedness. The peer alignment scale included three items and yielded a reliability coefficient of α = 0.85. A sample item for this scale is 'I can relate to the people around me in my classes'. It should be noted that we define peer alignment not as a general sense of comfort in classes, but rather as the sense that one shares interests, outlook, values, and so on with one's peers. The peer alignment items were adapted from the Longitudinal Assessment of Engineering Self-Efficacy survey developed by the Assessing Women in Engineering Project (AWE 2014). See Table 1 (Appendix) for items and loadings. 

 

5.4. Self-efficacy 

Self-efficacy items were adapted from the Patterns of Adaptive Learning Survey (PALS) (Midgley et al. 1996). This variable is composed of five items; sample items include 'I can do even the hardest work in engineering if I try' and 'Even if my engineering work is hard, I can learn it'. Alpha reliability for the self-efficacy scale is 0.93. See Table 1 (Appendix) for items and loadings. 

 

6. Analysis 

 

The first part of the analysis presents the descriptive statistics of the sample for all the intervening variables. The second part will include the hypotheses tests. European Journal of Engineering Education 385 

 

6.1. Descriptive statistics 

Means and standard deviations for the four key variables are listed in Table 2 (see Appendix). 

 

6.2. Hypothesis tests 

Because there has been previous research suggesting that social connections to instructors and peers may be especially important for women (Amelink and Meszaros 2011) - particularly in the engineering realm, where women may find it difficult to develop meaningful academic connections to faculty in a male-dominated department (Chesler and Chesler 2002) - we compared means by gender. An ANOVA (a test comparing group means) showed no significant differences by gender on any of the four key variables. As a result, we did not use gender as an additional variable in the hypothesis testing. Results are shown in Table 3 (see Appendix). To test hypotheses 1 through 3, we ran a linear regression analysis (an approach to modelling the relationship between predictor and dependent variables; Montgomery, Peck, and Vining 2012), with Instructor Connection, Peer Alignment, and Self-Efficacy as predictors and Satisfaction as the dependent variable, using SPSS 22. As predicted, Instructor Connection significantly predicted Satisfaction scores, Peer Alignment significantly predicted Satisfaction scores, and Self-Efficacy significantly predicted Satisfaction scores. See Appendix for details of the regression model. These results indicate that as Instructor Connection, Peer Alignment, and Self-Efficacy increase, student's satisfaction with engineering is also expected to increase. To test hypotheses 4 and 5, we ran regression analyses using the PROCESS procedure (which allows for analysis of mediating and moderating variables; Hayes 2013), model 4, in SPSS 22. This method is widely accepted as a means of better understanding the conditions or factors that affect the relationship between two variables. In this analysis, we want to determine whether: (1) Self-efficacy mediates the relationship between Instructor Connection and Satisfaction (hypothesis 4); (2) Self-efficacy mediates the relationship between Peer Alignment and Satisfaction (hypothe[1]sis 5). For hypothesis 4, we used Instructor Connection as the independent variable, Self-Efficacy as the mediator, and Satisfaction as the dependent variable. Both the direct effect and indirect effect of Instructor Connection on Satisfaction were significant, indicating that Self-Efficacy plays a mediating role between Instructor Connection and Satisfaction. We also found a sig[1]nificant mediation effect of self-efficacy, suggesting that Instructor Connection contributes to Satisfaction partially by enhancing student Self-Efficacy. That is, an increased level of connec[1]tion with the instructor has a positive effect on students' self-efficacy, which in turn contributes to increased levels of students' satisfaction. For hypothesis 5, we used Peer Alignment as the independent variable, Self-Efficacy as the mediator, and Satisfaction as the dependent variable. Both the direct effect and indirect effect of Peer Alignment on Satisfaction were significant, indicating that Self-Efficacy played a mediating role between Peer Alignment and Satisfaction. In other words, an increased sense of connection with student peers had positive effect on students' self-efficacy, which in turn contributed to increased levels of students' satisfaction. 

 

7. Discussion 

 

This study tested whether two malleable factors which together can be conceived of as 'fit' in the engineering program - connection to instructors and alignment with peers - significantly 386 M. Micari and P. Pazos contribute to student satisfaction with engineering, and what the relationship of those variables is to self-efficacy. The study found that the sense of connection to instructors and alignment with peers, as well as self-efficacy, predicted students' levels of satisfaction with the engineering major; and that self-efficacy mediated the relationships between peer alignment and satisfac[1]tion and between instructor connectedness and satisfaction. It appears, then, that both students' sense of connection to instructors and their sense of fitting into the classroom peer environment contribute to self-efficacy, which contributes to satisfaction with engineering school. Our findings highlight the importance of social influences for engineering students. The results contribute to the body of previous research on student connectedness in STEM fields which points to the critical importance of social interaction in student satisfaction and retention. In particular, previous research has demonstrated that students in STEM fields are often dissatisfied with the support they receive from instructors (Haag et al. 2007), and that students who leave engineering tend to feel isolated and lack positive, meaningful contact with both instructors and peers (Baillie 2000; Litzler and Young 2012). The importance of the instructor is often overlooked in discussions of STEM student well[1]being, which tend to focus on broader issues of climate, support services, school demographics, and so on. Calls do continue to emerge, however, for university STEM departments to system[1]atically promote teaching, and more broadly to promote a culture in which teaching is seen as a critical part of each instructor's role (Anderson et al. 2011; Brownell and Tanner 2012). We must recognise that good teaching involves not just enabling understanding, but also creating an environment that encourages students to engage and persist. Environments in which students feel connected will support feelings of inclusion, critical to retention, particularly among students underrepresented in engineering (Johnson 2012; Marra et al. 2009; Strayhorn et al. 2013). Given the lack of emphasis on the teaching relationship in many engineering schools, engineering instructors may not view their interactions with students as central to student success. There is an opportunity, then, to encourage what Gillespie (2002) describes as a 'connected' student-teacher relationship, one which supports 'coparticipation in the learning process' and exists 'in sharp contrast with the fearfulness and anxiety that often [characterize] ... nonconnected student-teacher relationships'. Indeed, there has been a recent increase in concern over college student anxiety and mental health generally (Galatzer-Levy, Burton, and Bonanno 2012), but the reaction to this often overlooks negative student interactions with faculty as a potential trigger - or positive ones as a potential source of support. Positive student-faculty relationships can promote a healthy classroom and campus climate, which in turn can reduce stu[1]dent stress levels (Cress 2008), and efforts to improve climate should include both faculty input and opportunities for faculty to reflect on and, where appropriate, improve their interactions with students. The finding that feeling in sync with one's peers is important for engineering student satis[1]faction corresponds with previous research demonstrating the important role of peers in student satisfaction and persistence in college (Astin 1984; Pascarella and Terenzini 1991), and promot[1]ing peer interaction makes intuitive sense. But students are not often given the opportunity to get to know their classroom peers at a level which would allow them to identify common traits, interests, or values; this may seem to many instructors to fall outside of the domain of the course. Group project work has become increasingly popular in engineering education, but such work is not necessarily structured in such a way that students can gain a meaningful sense of who their group members are. Such opportunities are important for all students, but appear to be especially important for underrepresented students in STEM fields (Ong et al. 2011), and can counter the sense of isolation that underrepresented students can easily feel. A feeling of congruence with peers might also help reduce the fierce competitiveness sometimes found in STEM departments which can discourage engineering student persistence (Amelink and Meszaros 2011; Bonous[1]Hammarth (2006). Again, this can be a particular concern in regard to underrepresented students, European Journal of Engineering Education 387 who, some research has shown, can be more negatively affected by competitiveness than their majority peers (Hurtado et al. 2007). 

 

8. Suggestions for practice

 

Both individual instructors and departments or schools can influence the degree to which students interact meaningfully with instructors. Following is a set of suggestions for encouraging greater and more meaningful instructor-student interaction: • Instructors should be aware of the important role of feedback in student learning and student motivation (Bjorklund, Parente, and Sathianathan 2004; Gibbs 2006). Feedback will be more meaningful when students are encouraged to engage with it in some way, for instance by redo[1]ing assignments based on feedback provided. Peer feedback can be used as a way to help both the feedback-receiver and the feedback-giver learn more deeply, and offers the added benefit of relieving some of the responsibility for providing feedback from the instructor. When peer feedback is used, students should receive training and guidelines on how to give and receive it most effectively. • Instructors can interact with students during class, even in large lectures. Active-learning opportunities for students to grapple with conceptual information and to think through and solve problems in class offer far more benefit from the classroom experience than lecture alone (De Graaff, Saunders-Smits, and Nieweg 2005; Smith et al. 2005). Peer discussion can be used in large and small classes to enable students to work out ideas and benefit from expo[1]sure to others' approaches, and offers the added benefit of drawing away attention from the individual, reducing the intimidation many students feel speaking in class. • Instructors can actively encourage students to visit their offices to discuss course material or questions students may have. Many students are intimidated by the idea of facing an instructor one-on-one with questions about a course, so that often the only students who visit instruc[1]tor offices are those who are already fairly confident with the material (Karabenick 2003; Karabenick and Knapp 1988). Apart from announcing more frequently in class that they are glad to meet with students who have questions, instructors can offer to have students visit their offices in pairs or groups, which takes the spotlight off of the individual and can help reduce anxiety; can offer periodic alternative office hours in more comfortable environments, such as a local café or dining hall; and can attend any existing student group study sessions to provide informal guidance on homework and exam review. • Instructors can mentor undergraduate research projects, not just as supervisors, but as true mentors, helping to bring students into the scholarly community. True mentoring involves helping students explore their interests, allowing them to work independently with a guiding hand, and providing constructive feedback and encouragement throughout the process (John[1]son 2007). Engaging undergraduates in research or other projects with instructors also makes an impact on undergraduate engagement and persistence (Hunter, Laursen, and Seymour 2007). Other ideas for increased interaction include instructors playing a role in residential colleges, or sharing a meal periodically with students; conversations that go beyond the class material to research, careers, and personal/professional development; and simply demonstrating respect for students and being approachable. Previous research (e.g. Micari and Pazos 2012) has shown that students who feel respected by and comfortable approaching instructors do better in their courses than students who feel otherwise. Instructors may be unaware that their behaviours send the opposite message (see Cotton and Wilson 2006), particularly for underrepresented 388 M. Micari and P. Pazos students (Allan and Madden 2006; Salter and Persaud 2003; Solórzano, Ceja, and Yosso 2000). While alignment with peers is difficult to influence directly, there are steps instructors and administrators can take to help students feel more in sync with their classmates. It has long been known that peers exert an important influence on college student outcomes (Astin 1984; Pas[1]carella and Terenzini 1991), and peer-to-peer interaction in academic settings - for instance, peer tutoring or peer assessment - has been shown repeatedly to bear cognitive and social benefits for undergraduates (Topping 2005), especially for underrepresented (Micari and Drane 2007) and first-generation college students (Pascarella et al. 2004). Giving undergraduates opportunities to interact meaningfully with one another in the classroom might also provide opportunities for them to learn more about one another and discover affinities. Indeed, enhancing the sense of peer alignment should not be seen as discouraging students from relating with different others. On the contrary, experiencing meaningful interactions with a wide range of peers - including those with different ethnic or cultural backgrounds, different socioeconomic backgrounds, and so on - pro[1]vides a number of academic and psychosocial benefits (Chang 1999; Denson and Chang 2009; Milem and Hakuta 2000). Out-of-class study groups - for instance, the supplemental instruction or peer workshop models (Drane, Micari, and Light 2014; Hensen and Shelley 2003) - can pro[1]vide less-structured academic engagement that may be even more conducive than the classroom to developing personal relationships. Engineering education has undergone significant change in recent decades (Crawley et al. 2007; Felder et al. 2000), with emphasis on career readiness, curriculum, and teaching and assess[1]ment methods. Instructor rapport with students and students' own sense of alignment with their peers have received less attention. Helping students feel they 'fit in' to their engineering environments, by creating opportunities for students to engage meaningfully with instructors and to get to know their fellow engineering students, can be a critical step in promoting student satisfaction, which in turn should promote persistence in the engineering major. 

 

9. Limitations of the study 

 

Although the sample in this study was restricted to two sections of a single engineering course, the nature of the course as a core requirement for two majors, mechanical and biomedical engineering, makes it a fairly representative sample of engineering students in those majors. The results may not be generalisable to all engineering disciplines, but can provide good insight about the influence of academic fit on students' academic experience.