The Relationships Among High School STEM Learning Experiences and Students’ Intent to Declare and Declaration of a STEM Major in College
by Martha Cecilia Bottia, Elizabeth Stearns, Roslyn Arlin Mickelson, Stephanie Moller & Ashley Dawn Parker - 2015
Background/Context: Schools are integral to augmenting and diversifying the science, technology, engineering, and mathematics (STEM) workforce. This is because K–12 schools can inspire and reinforce students’ interest in STEM, in addition to academically preparing them to pursue a STEM career. Previous literature emphasizes the importance of high-quality STEM academic preparation in high school and the role of informal and formal exposure to STEM as important influences on students’ chances of following a STEM career. Interestingly, although many students decide to major in STEM fields while they are in high school, the majority of the extant literature about why students choose STEM majors primarily focuses on students’ experiences during the college years.
Purpose/Objective/Research Question/Focus of Study: Through our research, we seek to investigate how learning experiences of inspiration/reinforcement/preparation toward STEM that students have during high school can help explain the stark differences in STEM involvement by gender and ethnicity. We first investigate the importance of high school inspirational/ reinforcing/ preparatory experiences for students’ intent to major in STEM while in high school. We then see how they relate to students’ actual choice of a STEM major. We do this focusing on gender and racial/ethnic differences in outcomes. Specifically, we analyze the impact of the timing of high school STEM courses (algebra, biology, and physics), the quantity of STEM-related classes, and the quality of these courses on students’ decision to pursue a college STEM major.
Research Design: This is an analysis of quantitative data gathered about members of North Carolina’s 2004 high school graduating class who also matriculated to one of the 16 campuses of the University of North Carolina system. Our research developed in two different stages. In the first stage, we utilize multilevel binomial models to examine students’ intent to declare a STEM major in their senior year of high school. In the second stage, we employ multilevel multinomial models to analyze chances of declaring a STEM major during the years 2005–2011, when students are in college.
Findings/Results: Findings suggest that STEM experiences of inspiration/reinforcement/preparation during high school interact with demographic variables to moderate students’ interest in STEM. Taking physics and intending to major in STEM during high school are the variables most closely associated with students’ choice of STEM as a major. In addition, taking physics is especially important for young women’s odds of declaration of STEM.
Conclusions/Recommendations: Findings suggest several policy recommendations: Provide a variety of high school learning STEM experiences that will link and augment students’ interest in STEM; change the way physics is presented to female students; utilize curricula and pedagogy that focus on ways that physics is personally relevant may increase the number of young women who take the course in high school; increase the quality of the STEM-related academic preparation of students; particular attention should be given to underrepresented subgroups of students; increase the offering of math and science-focused program at schools; and increase the availability of more STEM-related co- and extracurricular experiences available to youth.
Worldwide, the importance of mastering science, technology, engineering, and mathematics (STEM) constantly increases for individuals and their societies. Consequently, the need to cultivate a cutting-edge STEM workforceas well as to prepare a population that is scientifically, mathematically, and technologically literatehas become even greater. Whereas other developed nations appear to be making rapid advances in preparing their youth in math, engineering, and science and technology, U.S. childrens interest in and academic preparation for STEM careers have not kept pace with projected societal needs. The STEM labor force in the United States remains primarily male and White, and the absence of widespread participation of adults from diverse ethnic, gender, and social class backgrounds in the STEM workforce preparation exacerbates the current and projected problems. As underrepresented groups in the STEM workforce become an increasingly larger proportion of the U.S. population, the strength of the STEM workforce may further decline unless action is taken to broaden participation of all sectors of society (National Science Foundation [NSF], 2005, p. 3).
According to the Presidents Council of Advisors on Science and Technology (2010), there is a need for a two-step strategy to transform K12 education so it provides a more diverse and abundant supply of people who are able to join, and are interested in joining, the STEM workforce. First, students must be inspired so that they are motivated to study STEM subjects in school and subsequently excited about the prospect of pursuing careers in STEM fields. Second, it is important to prepare students so they have a strong foundation in STEM subjects and thus are able to use this knowledge in their personal and professional lives.
Primary and secondary schools are integral to augmenting, diversifying, and equalizing the STEM workforce because these schools can inspire students to pursue STEM or reinforce the interests of previously inspired students. They can also prepare students to pursue STEM majors once they enroll in college. Students who plan early and strategically for STEM learning and who have access to high-level and rigorous secondary school coursework are more likely to be well prepared for and successful in the STEM fields (ACT, 2006). Moreover, early exposure to science and math can awaken students interests in the fields early on (OECD, 2011) and can aid youth in making the appropriate sequences of curricular choices that enable them to continue in a STEM area.
Previous literature has emphasized the importance of high-quality STEM academic preparation in high school (Griffith, 2010; Hoepner, 2010; Tyson, Lee, Borman, & Hanson, 2007). Research also highlights the role of informal and formal exposure to STEM as important determinants of students chances of following a STEM career (Griffith, 2010; Kokkelenberg & Sinha, 2010). High school experiences that help awaken, reinforce, or increase students interests in STEM have been a less common focus of research, although their role certainly has been recognized by a number of programs and policies currently implemented to increase childrens curiosity in STEM.1 When people are interested in something, they become more attentive and alert; in fact, playful engagement with science during childhood and youth influences interest in science during adolescence (Krapp, Hidi, & Renninger, 1992).
Although many students decide to major in STEM fields while they are in high school (Maltese & Tai, 2011), the majority of the extant literature about why students choose STEM majors primarily focuses on students experiences during the college years. The few studies that incorporate important precollege experiences and characteristics in their analysis of students choice of STEM majors include Crisp, Nora, and Taggart (2009), Engberg and Wolniak (2013), and Wang (2013). Nevertheless, Crisp et al. (2009) exclusively focused on a sample of undergraduate students who attended Hispanic predominant institutions; consequently, their findings are only generalizable to this sample of students. Engberg and Wolniak (2013) and Wang (2013) addressed how individual and institutional factors affect students likelihood of majoring in a STEM field in college. Although these studies offer insights on some important precollege variables, they do not emphasize the importance of specific high school STEM-related learning experiences on students decision to major in STEM, nor do they discuss how these learning experiences are differentially associated with students of different races/ethnicities and gender. In general, there is a lack of research on contextual and learning factors affecting student decisions and choices as they relate to STEM in postsecondary education.
We build directly on Wangs (2013) study, which also focuses on the longitudinal impact of precollege settings. Unlike Wang, we focus on a set of high school experiences that could serve as potential inspiration/reinforcement/preparation experiences in a period that is crucial for students decisions of which college major to declare. By potential inspirers, we mean activities that capture the curiosity and imagination of students and provide them with access to exciting individual STEM experiences inside and outside of schools. In addition, these experiences can also help maintain excitement for STEM and could eventually lead students to choose one of these majors (thereby reinforcing their interests). By preparing opportunities, we mean experiences that help develop students capacities in the STEM subjects to a level of proficiency. It is important to note that experiences that could prepare some students might serve as inspirations or reinforcements for some others, and vice versa. Last, our study examines whether these inspiration/reinforcement/preparation experiences are similarly related to students odds of declaring/graduating with a STEM major, given differences in their gender and racial/ethnic backgrounds.
Specifically, for a sample of college-bound North Carolina students, our study examines the influence of high school exposure to basic STEM courses, high school exposure to STEM-related environments and activities, high school quantity of exposure to precollege STEM classes, and the quality of the STEM classes on matriculants likelihood of declaring a STEM major. We first investigate the importance of high school inspirational/ reinforcing/ preparatory experiences for students intent to major in STEM while in high school. We then move on to see how they relate to students actual choice of a STEM major. We do this keeping in mind gender and racial/ethnic differences. Our final objective is to identify the educational policy strategies that are most effective for increasing the number of students who pursue a STEM field, taking into account the gender and racial/ethnic diversity of the current pool of K12 students in the United States.
For decades, scholars and policy makers have referred to the STEM pipeline as the channel through which students flow from schools into the STEM workforce (Astin, 1982; Berryman, 1983). The Presidents Council of Advisors on Science and Technology (2010) stated that to build a strong STEM pipeline, it is not only critical to increase math and science proficiency among American students but also crucial to augment their interest in STEM fields. In fact, recent evidence reveals that significant numbers of even the most proficient students, including women and minority students, are drawn away from science and engineering toward other areas.
To better understand the role of students precollege school experiences in their decision-making process regarding pursuing a STEM field, we follow Wangs (2013) approach and draw from a theoretical model that integrates the social cognitive career theory (SCCT) based on Banduras (1986) social cognitive theory (SCT), and previous literature that identifies factors that have a strong association with college students academic outcomes and choices. SCCT considers career/academic choice as a dynamic enterprise (Lent, Brown, & Hackett, 1994). This model proposes that students intent to major in STEM is affected by a set of high school learning experiences that may be inspirational, reinforcing, or preparatory. All the preceding factors are subject to the influence of prior achievement in math. Students STEM intentions then affect their actual choice of STEM as a college major. Given that the major selection is not a static process, choosing a STEM major is also directly associated with college characteristics that might act as supports and barriers. SCCT holds that determination to produce a particular choice is the result of goals and interests; thus, we also examine the influence of students intent to pursue STEM fields upon postsecondary entry.
Learning experiences are crucial in the model of SCCT because these are contextual factors that comprise the opportunity structure within which choices are made and pursued. These factors may strengthen or constrain the associations between interests, goals, and actions. In our model, we focus on learning experiences of inspiration/reinforcement/preparation toward STEM that youth might have during high school and how these are related to their choice of major. In addition to the environmental support and barriers that constrain and enable students choices, we focus on these experiences of inspiration/reinforcement/preparation that help explain students interests and self-reference beliefs. To sum up, the model of SCCT states that personal inputs and contextual background determine students learning experiences, which then are associated with their self-efficacy and outcome expectations. These factors then influence students career interest and career intent and, ultimately, their career choice.
Because we were also interested in identifying racial/ethnic differences in the impact of these early experiences on students chances of pursuing a STEM degree, we also drew from the tracking literature. Sociologists of education uniformly have found that the assessments of academic ability and subsequent placement in different tracks or ability-grouped classes often parallel race and social class differences (Carbonaro, 2005; Lucas, 1999; Mickelson, 2001, 2007; Oakes, 2005). There could be a differential impact of the inspiration/reinforcement/ preparation experiences on students decisions to pursue a STEM field based on their race/ethnicity, in part because of the unequal STEM inspirational/reinforcing/preparation experiences students receive depending on their track and/or ability group. These groupings, in turn, reflect racial and social class stratification in public schools opportunities to learn. In fact, previous research suggests that racial disparities in STEM participation occur precisely because fewer Black and Latino/a students are prepared for STEM in high school because they are more likely to learn in lower level tracks (Tyson et al., 2007).
Last, we also expected to find different results for young women compared with young men. Previous studies have shown that during the high school years, the percentage of young men interested in a STEM career remains stable, whereas the percentage of young women with similar interests declines (Sadler, Sonnert, Hazari, & Tai, 2012). Even contemporary American students perceive explicit and implicit messages about appropriate gender roles regarding the STEM fieldsnamely, that STEM-related careers are associated with primarily male professions (Lee, 1998; Seymour & Hewitt, 1997). Studies have found that self-efficacys influence on intentions to enroll in math and math achievement was significantly stronger for boys than for girls, whereas interest was significantly more important in the prediction of math enrollment intentions for girls than for boys (Stevens, Wang, Olivarez, & Hamman, 2007). Other studies have acknowledged the importance of unequal prior achievement between boys and girls to explain gender gaps in STEM (Fryer & Levitt, 2010; Good, Aronson, & Harder, 2008; Penner & Paret, 2008). However, more recently, Riegle-Crumb, King, Grodsky, and Muller (2012) reported that the gender discrepancy in STEM majors is not largely explained by disparities in prior achievement. Through our research, we sought to investigate how learning experiences of inspiration/reinforcement/preparation toward STEM that students have during high school are associated with differences in STEM involvement by gender.
LEARNING EXPERIENCES OF INSPIRATION/REINFORCEMENT/PREPARATION
The learning experiences students have during their high school years are framed by the opportunity structure that, in part, conditions students career choice. School exposure to STEM courses and STEM-related activities can inspire/reinforce students interest in pursuing a STEM major early in their educational trajectory. Doing so likely helps them to make appropriate decisions to successfully follow a STEM course-taking pathway. In addition, the greater the opportunities students have to learn about science and math during high school, the higher their chances of committing to a STEM field as college students. Therefore, a broader academic preparation augments their capacity to pursue science, technology, mathematics, and engineering career pathways.
Previous research has linked STEM inspiring/reinforcing/preparing learning experiences with students interest, intent, and actual choice of major in STEM. For example, studies show that early exposure to STEM-related courses and higher quantity and quality of STEM-related courses are linked to higher STEM course taking in college and in students decision to pursue a STEM-related degree (Enberg & Woniak, 2013; Hoepner, 2010; Lee & Judy, 2011; NSF, 2005; Newton, Torres, & Rivero, 2011; Wang, 2013). In addition, past studies have also found that high school physics is the chief STEM pathway (Tyson et al., 2007). Furthermore, evidence also indicates that intent to major in STEM and early career aspirations are also closely associated with enrollment in STEM major fields during college (Newton et al., 2011; Sadler et al., 2012).
Numerous studies have also documented the efficacy of various programs that aim to renew student enthusiasm, interest, and knowledge in the sciences. In doing so, these programs reinforce students interest in STEM careers. Many reinforcing experiences occur outside the classroom. They include outreach programs, extracurricular activities, school clubs, and summer science enrichment programs (Atwater, Colson, & Simpson, 1999; Gibson & Chase 2002; Howe 2009; Knox, Moynihan, & Markowitz, 2003).
Past evidence has also shown that there is a differential impact of students STEM-related learning experiences during high school on their choice of major based on gender and/or race. Studies have reported that women and disadvantaged minority students math and science courses are often of lower quality than those taken by men and Whites. These disparities in opportunities to learn reflect the likely existence of structural inequalities in the educational system (Davenport et al., 1998; Laird, Alt, & Wu, 2009; Riegle-Crumb & Grodsky, 2010; Schneider, Swanson, & Riegle-Crumb, 1998). Research has also found that there is a stronger positive relationship between early exposure to STEM and exposure to higher quality STEM-related courses, and womens and minorities interest and persistence in college STEM (Griffith, 2010; NSF, 2005). Tyson and colleagues (2007) concluded that gender disparities in STEM occur because women are less likely to pursue STEM, but racial disparities occur because fewer Black and Hispanic students are prepared for STEM in high school.
Most research analyzing STEM majors focuses on the college years and on variables that help explain retention in STEM majors rather than aiming to explain why students decide to enroll in a STEM major (Bettinger & Long, 2005; Carrell, Page, & West, 2010; Hoffman & Oreopoulos, 2007; Newmark & Gardecki, 1998; Price, 2010; Qian, Zafar, & Xie, 2010; Rothstein, 1995; Robst, Keil, & Russo, 1998). To a lesser extent, research has tried to incorporate precollege characteristics and experiences to help understand students choice of major (Crisp et al., 2009; Wang, 2013). Our research has a longitudinal approach that acknowledges the dynamic nature of career choice by including learning experiences that students had during the high school years and relating them to students choice of major during the first college years.
Although Crisp and colleagues (2009) performed an analysis that included precollege variables to explain college choice of STEM major, these authors focused on student-level precollege characteristics rather than students lived experiences during their precollege years. Additionally, their sample only included students attending one Hispanic-serving institution. In contrast, our study utilizes a diverse sample of North Carolina college-bound students and centers attention on the STEM inspiring/reinforcing/preparing learning experiences of students during the middle school and high school years. Therefore, we are able to compare the impact of all these experiences for our diverse sample of students.
Our study builds directly from Wangs (2013) article, which also provides a longitudinal approach to explaining STEM choice of major. In her study, Wang focused on the direct and indirect influences of high school exposure to math and science, achievement and motivational attributes as related to math, and initial postsecondary experiences on entrance into STEM fields of study in college. Our study expands Wangs (2013) research by emphasizing the importance of the learning experiences students have during the high school years, which, according to SCCT theory, constrain the opportunity structures students have to make their career choice decision. Given the breadth and depth of our data, we are able to go beyond just looking at exposure to math and science, and math achievement to analyze the impact of the timing of high school STEM courses (algebra, biology, physics), the quantity of STEM-related classes, and the quality of these courses on students decision to pursue a college STEM field (see Figure 1). We advance the literature by doing this analysis by racial/ethnic categories and gender categories. The level of detail of our results regarding the relationship of each inspiring/reinforcing/preparing experience with students odds of declaring a STEM major by gender and racial/ethnic background offers numerous possibilities for policy implementation during the K12 years that could help reduce the STEM participation gender and racial gaps.
We also include many other school and individual experiences and characteristics, students own interests and self-reference beliefs, and environmental supports and barriers that ultimately enable or constrain students choices. An important note is that we employ a data set that targets the STEM context of secondary schools, including a variable that accounts for whether a high school has a math- and science-focused course of study.
Figure 1 presents a conceptual model of the relationships among secondary school experiences of inspiration/reinforcement/preparation and students odds of STEM declaration that we test in this study. Drawing from SCCT, we look at the persons demographic characteristics as inputs and at the background context to examine how these relate to the learning experiences (opportunity structure) students are exposed to during the high school years. We expect these learning experiences of inspiration/reinforcement/preparation to have an association with students ranking in classroom (academic self-efficacy) and with students math SAT scores. These variables are associated with students interest, which at the same time is linked to students intention to major. Last, the intent to major in STEM is expected to be closely related to the students actual chances of declaring a STEM major during college. By focusing on experiences of inspiration/reinforcement/preparation, we center our analysis on important life events, choices, or intermediate phenomena that are most likely to occur as a result of the initial conditions that critically affect the direction of childrens lives. Together, these dynamic forces shape their interest in pursuing a STEM field. We also include a measure for prior achievement at the end of high school that might have a direct and/or indirect effect on the final outcomes.
Figure 1. Conceptual model
Note: Adaptation of the model of personal, contextual, and experiential factors affecting career-related choice behavior (Lent, Brown, & Hackett, 1994).
As this model suggests, we will test if:
Hypothesis 1: The intent of majoring in a STEM field and actual STEM declaration are greater for students who have had a greater number and higher quality of STEM inspiring/reinforcing/preparing experiences while in high school than for those who have had fewer of these experiences.
1.1. The earlier students take algebra and biology classes in high school, the more likely they will intend to major in STEM and then declare a STEM major.
1.2. The chances of intending to major in STEM and later declaring a STEM major increase if the students attend a high school that offers a math and science program.
1.3. The greater the number of years of biology and mathematics a student has taken, the more likely they will intend to major in STEM and then declare a STEM major.
1.4. The chances of intending to major in STEM and later declaring a STEM major increase if the students took physics during high school.
1.5. The more advanced the STEM related classes the student took during high school, the more likely he or she will intend to major in STEM and then declare a STEM major.
1.6. The more extra STEM-related coursework/experiences a student has, the more likely he or she will intend to major in STEM and then declare a STEM major.
Hypothesis 2: The significant influence of these STEM inspiring/reinforcing/preparing experiences is stronger for disadvantaged minorities, who are generally placed on the lower tracks and receive fewer STEM learning opportunities and STEM learning experiences of lower quality as compared with White students.
Hypothesis 3: The significant influence of STEM inspiring/reinforcing experiences is stronger for young women, who receive less societal encouragement to enter STEM fields and whose interest in STEM tends to decline more rapidly than that of young men.
Hypothesis 4: The greater students intent to major in STEM, the greater the odds they will declare a STEM major in college. This holds for students of all races/ethnicities and gender.
Hypothesis 5: Mathematics achievement in high school and students intent to major in STEM mediate the influence that high school STEM inspiring/reinforcing/preparing experiences have on students odds of declaring a STEM major in college.
The next section of this article presents our data and research design. After presenting our findings, we discuss their limitations and implications for policy, and we suggest future lines of research.
DATA SET AND ANALYTIC SAMPLE
We used the NC Roots of STEM data set for our investigation. This data set contains longitudinal information on all North Carolina public high school graduates for the class of 2004. In addition to academic performance indicators from seventh grade through college graduation, the data set has information about the characteristics of the middle schools, high schools, and colleges that students attended throughout their educational careers in North Carolina. Data that follow students from Grades 712 were provided by the North Carolina Education Research Data Center at Duke University and from the National College Board. Data on college experiences were provided by the University of North Carolina General Administration. We focus on a racially, ethnically, and socioeconomically diverse sample of 18,0002 college-bound students who attended 510 middle school and 350 high schools in North Carolina and then later attended any of the 16 University of North Carolina campuses beginning in 2004 (as a result, our findings are only generalizable to this sample of college-bound students). Once we eliminated cases with random missing values for other variables, we ultimately arrived at a sample of 12,000 students.
VARIABLES IN THE STUDY
As dependent variables, we first use whether a student had intent to declare a STEM major in his or her high school senior year. The intent of STEM major is a dichotomous variable that comes from a question asked in the SAT questionnaire in 20032004, in which students indicated the major or area of study that most interested them, followed by their other choices (up to four). Students who reported on their SAT questionnaire that they intended to major in agriculture and natural sciences, biological sciences, computer and information science technology, engineering and engineering technician studies, mathematics and/or physical sciences were categorized as students intending to major in STEM.
Our second dependent variable was whether the student actually declared a STEM major in college. The declaration STEM variable is a multinomial dependent variables, where 0 indicates no declaration of a STEM major, 1 stands for nondeclaration of major (undeclared), 2 stands for declaration of a major in biology,3 and 3 indicates declaration of a major in any other STEM field other than biology (physical sciences, engineering, and mathematics). While 2/3 of the students who declared physical sciences/engineering/mathematics majors were men, the biology major did not have the issue of underrepresentation of women and students of color that other STEM disciplines had. To define a STEM major, we used the categorization used by the National Science Foundation ADVANCE program (http://www.nsf.gov/crssprgm/advance/index.jsp), in which majors such as engineering; physical sciences; earth, atmospheric, or ocean sciences; mathematical and computer sciences; and biological and agricultural sciences are considered within the STEM category. We excluded social sciences from our STEM category in this analysis, although in certain contexts, they are considered STEM majors (see Table 1 for descriptive statistics). For the analysis of declaration of STEM major, we chose a multinomial approach because it is most appropriate given the different demographic composition of students in biology and other STEM disciplines4 (Newton et al., 2011).
Table 1. Descriptive Statistics of Key Variables by Subsample of Students
The key independent variables include a set of exogenous indicators at the student level that are related to inspirational/reinforcing/preparing experiences that students have had during their K12 schooling that prior research suggests help awaken and/or reinforce their interest in STEM majors. The four sets of experiences follow:
1. Attending a school that offers a STEM-related program: To create this variable, the authors contacted each of the high schools in our database and asked the guidance counselor or faculty member in charge of curriculum the following question: Does your school currently have, or has it ever had, any programs or academies [or other organizational structures] that focus on science, math, engineering, or technology? If they responded positively, we asked for the start date of the STEM-related curriculum/program. If they offered a math- and science-focused program and the expressed dates coincided with the years when the students from our sample attended high school (20012004), these schools were flagged with a dummy variable indicating the availability of a STEM-related program. Thus, attending a school that offers a math- and science-focused program does not necessarily mean that the student was part of the program; rather, it connotes that there was strong possibility of access to STEM-related activities/courses.
2. Having taken algebra 1, algebra 2, physical sciences, and biology earlier than the mean of the sample (for this sample of students, the usual pattern is students taking algebra 1 in 9th grade, algebra 2 in 10th grade, physical sciences in 9th grade, and biology in 10th grade): If students took these math and/or science classes earlier than the typical student (sample mean), they were exposed earlier to STEM material. Such experiences likely offer more chances of stimulating interest in STEM and obtaining better STEM preparation (Newton et al., 2011).
3. Measures of academic preparation during high school: This variable consisted of number of years students took biology and math (Engberg & Wolniak, 2013; Lee & Judy, 2011; Wang, 2013), if they took physics (Tyson et al., 2007),5 and the quality of the STEM-related courses they took (Griffith, 2010; Hoepner, 2010). Quality was operationalized as the proportion of honors STEM-related classes the students took given the availability of honors STEM-related classes at their high school. These variables are related to the knowledge and skills that students acquire during high school that are necessary to pursue a STEM major.
4. Other STEM-related activities: Whether a student took coursework or had experience in any of several STEM-related activities, including spreadsheets, databases or statistical programs, Internet activity, computer graphics, and computer programming. The inclusion of this variable is consistent with the recommendation of the Presidents Council of Advisors on Science and Technology (2010, September) that students develop personal connections to the knowledge and excitement of STEM fields. These connections can happen outside the classroom through extracurricular activities and advanced courses (see Appendix A for more details).
The influence of the set of four inspirational/reinforcing/preparatory experiences may potentially be reflected in previous achievement scores. Thus, we also included previous mathematics achievement scores in our models to account for this fact. For the models that predict intent to major in STEM in 12th grade, we controlled for students mathematics end of grade (EOG) scores in eighth grade. For the models predicting odds of declaration of STEM as a major, we included the mathematics SAT scores, which students took in 12th grade (doing so controlled for the possibility of a reciprocal causal relationship with inspiration/reinforcement and academic preparation of students lived experiences during high school). In addition, we also included a measure for students estimate of high school class rank (academic self-efficacy), a factor that has been shown to predict STEM interest and intent to graduate with a STEM degree (Byars-Winston, Estrada, & Howard, 2008). Academic self-efficacy is an important component of the social cognitive career theory.
In addition to the core set of independent variables described earlier, we included control variables that reflected students school and family background. Family background is likely to enable or constrain students decision to pursue a STEM major. These variables were: (a) if the student transferred middle schools between seventh and eighth grade; (b) being a first-generation college student; and (c) receiving free/reduced lunch in eighth grade. We also incorporated into the model characteristics of students schools that could facilitate or limit their opportunities to pursue a STEM major. These characteristics were: (d) whether the school was in a rural, suburban, or urban area; (e) percent of students in an advanced college track in 10th grade; and (f) percent female math and science teachers in 10th grade. Last, we also included in the analysis (g) demographic variables such as race and gender of the student. For the models that predict odds of declaration of STEM as a major, we included measures indicating (h) which of the UNC college campuses the student attended (North Carolina State University is the excluded category) and (i) if the student received a Pell grant in his or her first year of college. Appendix B provides further details of these variables. The models analyzing intent of majoring in STEM do not include any college-level variables.
To address our research questions, we utilized multilevel models using the GLIMMIX procedure in SAS. In hierarchical models, a different variance structure is assumed to exist at each level of analysis, and regression coefficients and standard errors can be estimated at each level (in this case, at the student level and school level) without bias. Our research developed in two different stages. In the first stage of our analysis, we utilized multilevel binomial models to examine students intent to declare a STEM major in their senior year in high school. In the second stage of our analysis, we employed multilevel multinomial models to analyze chances of declaring a STEM major during the years 20052011, when they were in college. Multilevel multinomial regression analysis is appropriate for predicting students odds of declaring a STEM major because it allows one to compare more than two groups. We included dummies for categories of college campuses in the NC University system to control for college fixed effects (employing NC State University as a reference category because it is the UNC systems flagship STEM campus). The hierarchical multinomial models allowed us to examine the effect of school characteristics that impact college students decision to declare a STEM major, taking into consideration that certain groups of students attended the same high schools.
To run models with no interactions and with gender interactions, we used the entire sample of students. We also ran models of subgroups of female students and male students. To analyze racial differences, we used a sample with only White and African American students, and we also later performed analysis of subgroups of White students and African American students.6
Before we discuss our findings, there are some methodological limitations to keep in mind as we report and interpret our results. The limitations concern sample selectivity biases that limit the generalizations we are able to make from our findings. First, it is important to acknowledge that we can only generalize our results to these students in our study because of the nature of our sample. It was restricted to the pool of students who attended secondary public school in North Carolina and later pursued their undergraduate studies in the UNC system and declared a major on any of these campuses.7 Second, because of the economy of North Carolina compared with other regions of the country, the demand for jobs in STEM in this region could also help explain decisions to enroll in a STEM field. Third, our sample size is decreased by including the variable high school class rank, which is our proxy measure for self-efficacy. We decided that, theoretically, this was a key variable that needed to be included. Therefore, we opted to go with a smaller sample that still allows us to conduct our detailed analysis. Nevertheless, our samples before and after listwise deletion are not substantially different. For example, in both samples, the same percent of students declared non-STEM majors, declared biology as a major, were male, received free-reduced lunch, took physics, and attended, on average, schools with the same percentage of students in advanced/college track and with the same percentage of female math and science teachers.
Descriptive Statistics of Data
Descriptive statistics in Table 1 show that 38.2% of the sample expressed an intention to major in STEM while they were in high school. This intent significantly varied depending on the students gender. Only 24.5% of the female students reported intent to major in STEM, compared with 58.0% of the male students. Intent to major in STEM is 34.2% for African American students and 39.5% for White students. Regarding students declaration of major, our data show that women and White students more often declare a non-STEM field as a major (63.9% and 60.6% respectively) compared with male and African American students (50.5% and 55.6%, respectively). Biology is slightly more popular among women and White students, whereas declaring a major in physical sciences, engineering, and mathematics is far more popular among male students (20.4%) than for female students (only 6.6%). In fact, there is a substantial 13.8% gender gap in declaration of physical sciences, engineering, and mathematics as a major.
With respect to our key independent variables of learning experiences during high school associated with STEM, we found significant differences by gender in students participation in physics during high school. Whereas 34.9% of male students take physics during high school, only 20.4% of the female students enroll in this subject. The racial/ethnic differences in students early STEM experiences are also substantial. On average, African American students in our sample of NC college-bound students took a much smaller percent of honors STEM-related classes during high school; took algebra 1, algebra 2, physical sciences, and biology later; and took physics less frequently than their White counterparts.
Intent to Major in STEM
Table 2 shows predicted probabilities of intent to major in STEM. Model 1 displays results of regressions with no interaction terms. Models 2 and 3 include interaction terms by gender categories (keeping female students as the reference category) and by race/ethnicity categories (keeping White students as the reference category). Models 4, 5, 6, and 7 show results for specific subsamples of male, female, White, and African American students, and they include a control for math standardized test performance (EOG score).
Table 2. Unstandardized Coefficients From Binomial Multilevel Models - DV: Intent to Major in STEM
Continuation Table 2. Unstandardized Coefficients from Binomial Multilevel Models - DV: Intent to Major in STEM
Models 17 in Table 2 indicate the importance of taking physics during high school and having coursework and/or experience in other STEM-related activities during high school for all students intent to major in STEM. Both variables have a positive significant relationship with students intent to major in STEM. In addition, taking a larger proportion of honors STEM-related courses given the availability of honors courses at their schools, attending a school that offers a math- and science-focused program, and taking a greater number of years of biology and mathematics during high school are also related to higher intentions to major in STEM in 12th grade. Taking algebra 1, algebra 2, physical sciences, and biology early does not appear to have a significant association with students intentions of majoring in STEM.
In Model 2, interactions with gender show that taking higher proportions of the STEM courses in advanced tracks, taking physics during high school and taking a greater number of years of biology help explain the gap in intention to major in STEM between men and women. Two experiences significantly increases the chances that female students intend to major in STEM during their senior year, and another learning experience increases even more male students intention to major in STEM. Figures 2, 3, and 4 present a visual representation of predicted probabilities calculated at their mean values. Figure 2 shows that female students intention to major in STEM increases from .25 to .32 if they take 90% of their STEM-related classes in an advanced track rather than taking no STEM classes in an advanced track. Figure 3 shows that taking a greater number of biology courses also significantly increases female students intentions to major in STEM, raising the intent from .18 among those female students who take 0 years of biology in high school to .34 of those female students who take 2 years of biology. Although taking biology early appears to have a positive relationship with female students intent to major in STEM, this relationship does not appear to be a significant predictor of males intent to major in STEM. Finally, Figure 4 provides evidence that taking physics in high school increases male students intent to major in STEM even more. In this case, the intent to major in STEM of those male students who took physics during high school increased to .74 from .54 among those women who did take physics. As Figure 4 shows, taking physics in high school also significantly increases female students intent to major in STEM.
Figure 3. Predicted probability of intent to major in STEM
Figure 4. Predicted probability of intent to major in STEM if student took physics or not during high school
Model 3 presents results of models that include race interactions. Because of the small number of Asian, Latino, and American Indian students in our sample, this analysis could be performed only with the sample of White and African American students. These models provide evidence that the number of years of biology taken in high school is the only high school experience of inspiration/reinforcement/preparation to help explain the gap in intent to major in STEM between White and African American students. Additionally, the models of subgroup of White and African American students (Models 4 and 5) provide evidence that several factors, including the proportion of honors STEM-related classes taken, attending a school that offered a math and science focus, and number of years of biology taken during high school have significant associations with White students intent to major in STEM, whereas they do not have a meaningful relationship with African American students intentions to major in STEM. Again, taking physics and being involved in other STEM-related activities during high school is significantly related to greater intentions to major in STEM of both White and African American students.
Table 3. Unstandardized Coefficients from Multinomial Multilevel Models - DV: Declaration of a Major in STEM
Cont. Table 3. Unstandardized Coefficients from Multinomial Multilevel Models - DV: Declaration of a Major in STEM
Cont. Table 3(2).Unstandardized Coefficients from Multinomial Multilevel Models - DV: Declaration of a Major in STEM
Cont. Table 3(3).Unstandardized Coefficients from Multinomial Multilevel Models - DV: Declaration of a Major in STEM
Models 8 and 9 (Table 3) show that taking physics during high school and the number of years of biology taken have a direct effect on students odds of declaring a STEM major, given that they are significant in all models regardless of whether we control for previous math achievement scores and intent to major in STEM. An important note is that intent to major in STEM appears to have a strong and positive relationship with students odds of declaring STEM once they are in college. In addition, a number of variables are significant in models that do not control for the prior mathematics achievement and intent to major in STEM but are not significant in the models that do control for previous achievement and intent to major in STEM. In other words, taking algebra 1 early has an indirect but significant impact on students interest in physical sciences/engineering/mathematics, but it only remains significant in the models that do not control for previous math achievement. Taking a higher proportion of advanced STEM courses in high school and being involved in other STEM-related activities during high school have an indirect but significant impact on students interest in biology and PSEM majors that only remains a significant predictor of declaring the major in models that do not control for intent to major in STEM in 12th grade. These findings are evidence that part of the influence that inspiration/reinforcing/academic preparation experiences have on students odds of declaring a STEM major is mediated by students previous SAT math scores and students intent to major in STEM.
The results also indicate gender differences in paths to STEM major declaration. Model 11 (Table 3) shows that taking physics is a factor that helps account for the gender gap in STEM declaration in college. Specifically, taking physics has a more positive association with female students chances of declaring biology compared with male students odds.
We were also interested in investigating differences that may occur by students race/ethnicity. Results presented in Model 12 (Table 3) show that none of the learning experiences of STEM inspiration/reinforcement/preparation helps explain the gap in declaration of STEM majors. Nevertheless, the analysis by subgroups presented in Models 15 and 16 show that White and African American students interest in nonbiology STEM fields is positively and significantly linked to whether they took physics during high school. Additionally, the proportion of STEM-related honors courses taken during high school is only directly related to African American students odds of declaring a major in biology. Last, if White students take a greater number of years of biology during high school, their odds of declaring a biology major increase, whereas no significant association exists between number of years of biology and African American odds of biology declaration in college.
Appendix C shows other variables that have a significant positive relationship with students odds of declaration of STEM, including being male, being Asian, having had a higher percentage of female math and science teachers during high school, and high school class rank.
In general, our findings provide support for our hypotheses. We found that students STEM-related learning experiences during high school are associated with students intent in STEM and, consequently, their choice of a STEM discipline as a major. Our findings provide a great deal of support for Hypothesis 1. Almost all the learning experiences of inspiration/reinforcement/preparation included in our study proved to have an important association with students intent to major in STEM and/or with their odds of declaring a major in STEM. Our findings show that taking physics is the variable most powerfully related to increases in the intent to major in STEM (biology and PSEM) and to the odds of declaring a STEM major (biology and PSEM) (Hypothesis 1.4). Given that previous studies show that females take fewer physics classes because they seem to think that they understand a concept only if they can put it into a broader world view, unlike young men, it is important to make physics appear more relatable to young womens interests during the high school years (Stadler, Duit, & Benke, 2000). Taking a greater number of biology courses during high school is also very important for students who follow a biology pathway (Hypothesis 1.3). Attending a school with a math- and science-focused program is positively related with students intent to major in STEM but not directly with their odds of declaring a PSEM major (Hypothesis 1.2). Variables such as proportion of honors STEM-related courses, having other STEM-related activities during high school, and number of years of math taken during high school increase students intent to major but are not related directly with students odds of STEM declaration (but they are related indirectly through intent to major in STEM ) (partial support for Hypotheses 1.5, 1.6, and 1.3). Our results do not show that taking algebra and/or biology classes earlier has a significant relationship with students intent to major in STEM or with students actual odds of STEM declaration, and therefore do not provide support for Hypothesis 1.1.
We found mixed support for Hypothesis 2. Although we found that taking a higher proportion of honors STEM courses in high school is indirectly and directly related with students odds of declaring STEM majors, we did not find that any other of the inspiration/reinforcement/preparation STEM learning experiences are particularly beneficial for African Americans. Results support Hypothesis 3 in that there is a significantly stronger influence of inspiring/reinforcing/preparing experiences on female students odds of declaring STEM as compared with the influence for males.
Our models also show support for Hypothesis 4 by showing that the greater the students intent to major in STEM, the greater the odds that he or she will declare a STEM major in college (similar to Wangs 2013 findings). And, as expected, this relationship holds for all races/ethnicities and for both genders. This is consistent with the SCCT precept that determination to produce a particular choice (for example, choosing a STEM major) is the result of goals and interests (intent to major in STEM). It is also consistent with findings of previous research.
In conclusion, this study indicates that students who have more STEM inspiration/reinforcement/preparation learning experiences during their high school years are more likely to intend to major in STEM and to actually declare STEM as a major.
We draw a number of policy implications from the studys findings. First and foremost, there is a continued need to address STEM preparation as a pipeline issue. Preparation issues involve key stakeholders across all levels of education. We recognize that an effective strategy to push more students toward STEM pathways is to provide a variety of high schools inspiring/reinforcing/preparing STEM experiences that will link with and augment students interest in STEM and increase their odds of declaring a STEM major. Doing so seems to be particularly important for women and African Americans. Our study shows that physics is strongly associated with young womens chances of declaring a STEM major. In addition, previous evidence (Murphy & Whitelegg, 2006) suggests that young women need to feel a personal relevance in the physics curriculum to motivate them to take physics in high school. Given that taking physics will substantially increase female students odds of declaring a STEM major, a reasonable strategy would be to change the way physics is presented to female students. Curricula and pedagogy that focus on ways that physics is personally relevant may increase the number of young women who take the course in high school.
Additionally, our results suggest that if all students have the opportunity to take more honors STEM-related classes, the likelihood that students will pursue a STEM degree will probably increase. For this to be possible, the STEM-related academic preparation of students should be of higher quality, and the number of honors class offerings and access to them should be expanded at all high schools. Previous research shows that minorities have less access to these advanced STEM courses. Therefore, particular attention should be given to underrepresented subgroups of students to increase their chances to follow STEM pathways. Doing so will require a serious evaluation and redesign of tracking practices to eliminate their correlation with students racial and SES backgrounds, including offering honors courses for all students.
Our study also highlights the importance of offering a math- and science-focused programs at schools and increasing the availability of more STEM-related experiences available to youth. Both factors appear to be significantly associated with increases in students interest in STEM. In addition, we find evidence that participating in other STEM-related activities during high school is positively related with higher chances of declaring a nonbiology STEM major. Increasing opportunities for students to participate in other STEM-related activities or courses could also help augment the number of students declaring a STEM major.
If improving STEM education is a national priority, it is certainly necessary to cultivate interest in and knowledge of math and science in order to spark a sense of wonder and excitement in the next generation (Obama, 2009). This articles findings offer evidence that STEM-related inspirational/reinforcing/preparatory experiences have an important and positive association with students chances of pursuing a future in STEM. Our recommendations suggest feasible approaches for providing inspiring and reinforcing experiences and for preparing more of the nations youth to choose a STEM pathway for their future.
1. For example, IQUEST, Massachusetts STEM Pipeline Fund Programs Using Promising Practices, DIGITS; Engineering is Elementary; Got Math?: STEM Summer Camps; Bringing Up Girls in Science (BUGS).
2. Sample sizes are rounded to the nearest 10th.
3. In this study, agricultural and earth sciences are grouped together with biology. Together, these fields only account for approximately 100 students out of the almost 15,000 cases we analyzed.
4. Although women comprise only 1/3 of the majors in the physical science and engineering fields, the biology major has an overrepresentation of women (65%) among the analyzed sample of students in the North Carolina University system. The sample is roughly reflective of the population of the University of North Carolina systems college seniors.
5. Only 27% of the students take physics during high school (either early in high school or as seniors).
6. Because of sample size limitations, we were not able to analyze subgroups of Asian, Latino/a, and American Indian students.
7. The sample of students we employ in this study is representative of North Carolinas in-state four-year college-going population: first, because most of the students in North Carolina who go to college stay in a college in-state, and second, because the UNC system includes very competitive public universities, such as UNC Chapel Hill and NC State University. Based on SAT survey data provided to the Roots Project by the College Board, the sample of young men and women included in our study had, on average, higher math and reading SAT scores than the North Carolina students who did not attend colleges in the UNC system but planned to attend a four-year college when they took the SAT.
ACT. (2006). Developing the STEM education pipeline. Iowa City, IA: Author.
Astin, A. W. (1982). Minorities in American higher education. San Francisco, CA: Jossey-Bass.
Atwater, M. M., Colson, J., & Simpson, R. D. (1999). Influences of a university summer residential program on high school students commitment to the sciences and higher education. Journal of Women and Minorities in Science and Engineering, 5(2), 155173.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall.
Berryman, S .E. (1983). Who will do science? Trends, and their cause in minority and female representation among holders of advanced degrees in science and mathematics. New York, NY: Rockefeller Foundation.
Bettinger, E., & B. Long. (2005). Do faculty serve as role models? The impact of instructor gender on women students. American Economic Review Papers and Proceedings, 95(2), 152157.
Byars-Winston, A., Estrada, Y., & Howard, C. (2008). Increasing STEM retention for underrepresented students: Factors that matter. Madison, WI: Center for Education and Work.
Carbonaro, W. (2005). Tracking, student effort, and academic achievement. Sociology of Education, 78(1), 2749.
Carrell, S. E., Page, M. E., & West, J. E. (2010). Sex and science: How professor gender perpetuates the gender gap. Quarterly Journal of Economics, 125(3), 11011144.
Crisp, G., Nora, A., & Taggart, A. (2009). Student characteristics, pre-college, college, and environmental factors as predictors of majoring in and earning a STEM degree: An analysis of students attending a Hispanic serving institution. American Educational Research Journal, 46(4), 924942.
Davenport, E. C., Jr., Davison, M. L., Kuang, H., Ding, S., Kim, S., & Kwak, N. (1998). High school mathematics course taking by gender and ethnicity. American Educational Research Journal, 35(3), 497514.
Engberg, M., & Wolniak, G. (2013). College student pathways to the STEM disciplines. Teachers College Record, 15(1), 127.
Fryer, R. G., & Levitt, S. D. (2010). An empirical analysis of the gender gap in mathematics. American Economic Journal: Applied Economics, 2(2), 210240.
Gibson, H., & Chase, C. (2002). Longitudinal impact of an inquiry-based science program on middle school students attitudes toward science. Science Education, 86(5), 693705.
Good, C., Aronson, J., & Harder, J. A. (2008). Problems in the pipeline: Stereotype threat and womens achievement in high-level math courses. Journal of Applied Developmental Psychology, 29(1), 1728.
Griffith, A. L. (2010). Persistence of women and minorities in stem field majors: Is it the school that matters? Economics of Education Review, 29(6), 911922.
Hoepner, C. C. (2010). Advanced placement math and science courses: Influential factors and predictors for success in college stem majors (Doctoral dissertation). Retrieved from ProQuest Digital Dissertations. (AAT 3437520)
Hoffmann, F., & Oreopoulos, P. (2007). A professor like me: The influence of instructor gender on college achievement (NBER Working Paper No. 13182). Cambridge, MA: National Bureau of Economic Research.
Howe, A. (2009, November). Tracking students knowledge of and interest in science, technology, engineering, and mathematics careers. Paper presented at the Evaluation 2009 Conference, Orlando, FL.
Knox, K. L., Moynihan, J. A., & Markowitz, D. G. (2003). Evaluation of short-term impact of a high school summer science program on students perceived knowledge and skills. Journal of Science Education and Technology, 12(4), 471478.
Kokkelenberg, E. C., & Sinha, E. (2010). Who succeeds in stem studies? An analysis of Binghamton University undergraduate students. Economics of Education Review, 29(6), 935946.
Krapp, A., Hidi, S., &Renninger, K. A. (1992). Interest, learning, and development. In K. A. Renninger, S. Hidi, & A. Krapp (Eds.), The role of interest in learning and development (pp. 325). Hillsdale, NJ: Erlbaum.
Laird, J., Alt, M., & Wu, J. (2009). Stem coursetaking among high school graduates, 1990-2005. MPR research brief, MPR Inc.
Lee, J. D. (1998). Which kids can become scientists? Effects of gender, self-concepts, and perceptions of scientists. Social Psychology Quarterly, 61(3), 199219.
Lee, J., & Judy, J. (2011). Choosing a stem path: Course-sequencing in high school and postsecondary outcomes. Paper presented at the Society for Research on Education Effectiveness, Washington, DC. Abstract retrieved from https://www.sree.org/conferences/2011f/program/downloads/abstracts/315.pdf
Lent, R. W., Brown, S. D., & Hackett, G. (1994). Toward a unified social cognitive theory of career/academic interest, choice, and performance [Monograph]. Journal of Vocational Behavior, 45, 79122.
Lucas, S. R. (1999). Tracking inequality: Stratification and mobility in American high schools. New York, NY: Teachers College Press.
Maltese, A. V., & Tai, R. H. (2011). Pipeline persistence: Examining the association of educational experiences with earned degrees in STEM among U.S. students. Science Education, 5, 877907.
Mickelson, R. A. (2001). Subverting Swann: First- and second- generation segregation in Charlotte, North Carolina. American Educational Research Journal, 38(2), 215252.
Mickelson, R. A. (2007, September/October). The social science evidence on the effects of diversity in K-12 schools. Poverty and Race, 1214.
Murphy, P., & Whitelegg, E. (2006). Girls in the physics classroom: A review of the research into the participation of girls in physics. Institute of Physics Report. Retrieved from https://www.iop.org/education/teacher/support/girls_physics/review/file_41599.pdf
National Science Foundation. (2005). Pathways to STEM careers: Preparing the STEM workforce of the 21st century. Arlington, VA: Author.
Newmark, D., & Gardecki, R. (1998). Women helping women? Journal of Human Resources, 33(1), 220246.
Newton, X. A., Torres, D., & Rivero, R. (2011). Making the connection: Timing of algebra and future college STEM participation. Journal of Women and Minorities in Science and Engineering, 17(2), 111128.
Oakes, J. (2005). Keeping track: How schools structure inequality. New Haven, CT: Yale University Press.
Obama, B. (2009, April). National Academy of Sciences speech. Presented at the National Academy of Sciences annual meeting, Washington, DC.
OECD. (2011). Education at a glance 2011: OECD indicators. Retrieved from http://dx.doi.org/10.1787/eag-2011-en
Penner, A. M., & Paret, M. (2008). Gender differences in mathematics achievement: Exploring the early years and the extremes. Social Science Research, 37(1), 239253.
Presidents Council of Advisors on Science and Technology. (2010, September). Prepare and inspire: K-12 science, technology, engineering, and math (STEM) education for Americas future. Washington, DC: Author.
Price, J. (2010). The effect of instructor race and gender on student persistence in STEM fields. Economics of Education Review, 29(6), 901910.
Qian, Y., Zafar, B., & Xie, H. (2010). Do female faculty influence female students choice of college major, and why? Working paper, Northwestern University.
Riegle-Crumb, C., & Grodsky, E. (2010). Racial-ethnic differences at the intersection of math course-taking and achievement. Sociology of Education, 83(3), 248270.
Riegle-Crumb, C., King, B., Grodsky, E., & Muller, C. (2012). The more things change, the more they stay the same? Prior achievement fails to explain gender inequality in entry to STEM college majors over time. American Education Research Journal, 20(10), 126.
Robst, J., Keil, J., & Russo, D. (1998). The effect of gender composition of faculty on student retention. Economics of Education Review, 17(4), 429439.
Rothstein, D. S. (1995). Do female faculty influence female students educational and labor market attainments? Industrial and Labor Relations Review, 48, 515530.
Sadler, P., Sonnert, G., Hazari, Z., & Tai, R. (2012). Stability and volatility of STEM career interest in high school: A gender study. Educational Science, 96(3), 411427.
Schneider, B., Swanson, C. B., & Riegle-Crumb, C. (1998). Opportunities for learning: Course sequences and positional advantages. Social Psychology of Education, 2, 2553.
Seymour, E., & Hewitt, N. (1997). Talking about leaving: Why undergraduates leave the sciences. Boulder, CO: Westview Press.
Stadler, H., Duit, R., & Benke, G. (2000) Do boys and girls understand physics differently? Physics Education, 35, 417422.
Stevens, T., Wang, K., Olivarez, A., & Hamman, D. (2007). Use of self-perspectives and their sources to predict the mathematics enrollment intentions of girls and boys. Sex Roles, 56(56), 351363.
Tyson, W., Lee, R., Borman, K. M., & Hanson, M. A. (2007). Science, technology, engineering, and mathematics (STEM) pathways: High school science and math coursework and postsecondary degree attainment. Journal of Education for Students Placed at Risk, 12(3), 243270.
Wang, X. (2013). Why students choose STEM majors: Motivation, high school learning, and postsecondary context of support. American Educational Research Journal, 50(5), 10811121.
Multinomial Multilevel Model - DV: Declare a STEM Major (Other variables)
Continuation Appendix C. Multinomial Multilevel Model - DV: Declare a STEM Major (Other variables)