Developing an Innovative Sustainable Science Education Ecosystem: Lessons from Negative Impacts of Inequitable and Non-Inclusive Learning Environments

Abstract

Societal stereotypes and biases about who belongs in science courses and who can excel in them can impact short- and long-term outcomes of students from marginalized groups, e.g., women, including their grades and beliefs about science as well as retention in science disciplines. Therefore, developing a sustainable science education ecosystem requires fostering equitable and inclusive learning environments in which students from all demographic groups have comparable outcomes. Here we describe a study with more than five hundred students in an introductory physics course at a large research university in the US that investigated female and male students’ perceptions of the inclusiveness of the learning environment (including their sense of belonging, perceived recognition by others such as instructors, and perceived effectiveness of peer interaction) and how it predicted their physics course grades, self-efficacy, interest, and identity at the end of the course. We find gender differences in perceptions of the inclusiveness of the learning environment disadvantaging female students and that these perceptions played a major role in explaining student outcomes. These inequitable trends in the context of physics can be useful for contemplating how to develop an innovative sustainable science ecosystem using hands-on and minds-on science and create an equitable and inclusive learning environment to help all students excel in science.

1. Introduction

There is an urgent need for workforce growth in science, technology, engineering, and mathematics (STEM) fields in many countries particularly because the global economy has become increasingly dependent on rapid innovations in the STEM fields. However, individuals from certain demographic groups, e.g., women and ethnic/racial minority (ERM) students, have largely been left out from contributing to these exciting fields and continue to be severely underrepresented in STEM majors [1,2] and careers [3]. For example, in a study examining the most prestigious international research awards, a gender gap was found in who received the awards [4]. This lack of diversity not only disadvantages these individuals from contributing to the STEM enterprise, but it also negatively impacts sustainability through the productivity and innovation in the STEM workplace [5,6]. Prior research suggests that gender diversity in the workforce is associated with increased revenue for organizations [6] and an organization that values the talents of diverse individuals with different backgrounds can create fertile grounds for disruptive innovation and better harness the unmet needs in under-leveraged markets [5]. Thus, there has been a focus on investigating the experiences and participation of traditionally underrepresented students, e.g., women and ERM students in many STEM fields [7,8,9,10,11,12,13], to improve equity and inclusion in these fields and increase the participation of these underrepresented groups in the STEM workforce, which is critical for innovation and sustainability of the STEM enterprise.

However, increasing the enrollment of traditionally underrepresented people in STEM fields is not sustainable unless the environment is equitable and inclusive and there is an established supportive culture so that all individuals regardless of their demographic group feel that they belong and can freely contribute their ideas. For example, some strategies to create an equitable and inclusive environment and increase innovation include ensuring that everyone is respected, valued, and heard, making it safe to propose novel ideas without any fear of judgment, and giving actionable feedback [5]. Creating this type of sustainable STEM ecosystem in which everyone can succeed begins with creating an innovative science education ecosystem with research-based active learning pedagogies that centers equity and inclusion [14,15,16,17]. In particular, it is important to keep in mind that prior research shows that if the active learning pedagogy is not implemented using teaching strategies that are equitable and inclusive, men have been shown to not only dominate in asking and responding to questions in class but they also dominate while working in groups, which can lower women’s self-efficacy [16]. In addition, in the lab context, men and women have been shown to fall into gender roles when splitting up the work which may disadvantage women [13]. Therefore, to make a sustainable science education ecosystem, the learning environment in STEM classes must be curated and implemented with equity and inclusion as a foundational and central tenet.

To develop an innovative sustainable science education ecosystem, our conceptualization of equity in STEM learning includes three pillars: equitable access and opportunity to learn, equitable and inclusive learning environment, and equitable outcomes. Thus, by equity in STEM learning, we mean that not only should all students have equitable opportunities and access to resources, but they should also have an equitable and inclusive learning environment with appropriate support and mentoring so that the learning outcomes are equitable. For there to be equitable learning outcomes, students from all demographic groups (e.g., regardless of their gender) who have the pre-requisites to enroll in STEM courses should have comparable learning outcomes. This conceptualization of equitable outcome is consistent with Rodrigues et al.’s equity of parity model [18]. An equitable and inclusive learning environment should be student-centered in which students are provided appropriate support and have an equal sense of belonging regardless of their prior preparation. It would also ensure that students from all demographic groups enjoy the hands-on and minds-on learning and embrace challenges as opportunities to grow their knowledge instead of being threatened by them. Equitable learning outcomes for STEM majors include the ability of STEM courses to empower all students and make them passionate about pursuing further learning and careers in related areas. We note that these three pillars are strongly entangled with each other. For example, if the learning environment is not equitable and inclusive, the learning outcomes are unlikely to be equitable.

Our conceptualization of equity in STEM learning is mindful of the pervasive societal stereotypes and biases about physics as well as the lack of role models that can have a detrimental psychological impact on women who are severely underrepresented. In general, when students struggle to solve challenging problems, they often respond in one of two ways. Some question whether they have what is needed to excel in STEM. Others enjoy the struggle because it means that they are tackling new concepts and learning. The negative reaction is a manifestation of a fixed mindset (believing that intelligence is immutable and struggling is a sign of a lack of intelligence), whereas the positive reaction is the sign of a growth mindset (believing in the brain’s capabilities can grow with deliberate effort). In an inequitable and non-inclusive learning environment, due to societal stereotypes and lack of role models, marginalized students are more likely than others to fall prey to the fixed mindset trap and view their struggle with challenging problems in a negative light.

We note that for a sustainable STEM ecosystem, STEM courses should not only have learning outcomes based upon STEM-related knowledge and skills we want students to learn, but also those that focus on whether all students (and especially those from marginalized groups) have a high self-efficacy, interest, positive perception of recognition from others such as instructors, sense of belonging, peer interaction, and identity as people who can excel in physics. Drawing analogy with sports, we note that to help players excel in any game, such as tennis, coaches must ensure both good defense and offense. Likewise, helping students learn requires that instructors equip all students with both defensive (strong motivational beliefs so students believe they can excel) and offensive strategies (strong problem solving and meta-cognitive skills). Instructors can strengthen students’ defenses by creating equitable and inclusive learning environments in which all students have high beliefs. Only if the learning environment is equitable and inclusive so that all students have strong defenses about learning can they effectively engage with the offense by tackling challenging problems and developing problem solving, reasoning, and meta-cognitive skills. In the absence of equitable and inclusive classrooms, students without strong defenses are unlikely to risk struggling with challenging problems and engage fully with hands-on and minds-on activities.

Here we discuss a research study involving physics classrooms at a large research university in the US to understand women’s and men’s perceptions of the inclusiveness of the learning environment, and how it predicts student outcomes (including their performance and beliefs) in introductory physics courses. Lessons learned from our investigation about equity and inclusion in physics classrooms are invaluable for all STEM disciplines that suffer from similar issues with students from marginalized demographic groups, particularly because societal stereotypes and biases are some of the worst in physics, a field whose history is often told through the stories of brilliant men [19]. These stereotypes, biases, and lack of role models can contribute to lower student beliefs (such as self-efficacy, interest, and identity) and performance outcomes (e.g., grades) for women in physics [20,21,22] unless explicit efforts are made to make the learning environment equitable and inclusive. We emphasize that in addition to performance outcomes, students’ STEM beliefs are important to investigate since students’ beliefs in different STEM domains can influence their continuation in related courses, majors, and careers [23,24,25,26]. Student beliefs such as identity, self-efficacy, and interest in a particular STEM field are important for their career interests [27,28,29], learning [30], and continuation in STEM fields [31,32,33,34]. Furthermore, in physics, prior research shows that gender gaps can persist in many of these beliefs as well as in the performance outcomes at the end of the course with certain student populations [20,21,22]. Thus, in order to create a sustainable STEM education ecosystem, equity and inclusion must be central to making sure that innovative hands-on and minds-on activities benefit all students and that all students (and particularly those from marginalized groups) can be supported equitably.

Therefore, investigating these beliefs and performance outcomes for students from different demographic groups, e.g., women and men, can provide important information on how students are persisting in these STEM courses and how educators can create an innovative sustainable ecosystem that centers on an equitable and inclusive learning environment in which students from all demographic groups have comparable outcomes. Prior work has mainly investigated students’ performance outcomes as well as self-efficacy, interest, and identity and connections between these factors in physics courses in which women are underrepresented [35,36]. However, societal stereotypes and biases may impact female students’ beliefs and performance even in STEM courses in which they are not numerical minorities if the learning environment is not equitable and inclusive. Therefore, research is necessary to examine how mechanisms for structuring courses in instructors’ control can influence women’s and men’s course outcomes in contexts not frequently studied in the past, e.g., introductory physics courses for students on the bioscience track in which women are not numerical minorities. The findings of this research can provide guidelines for developing an innovative and sustainable hands-on and minds-on science education ecosystem that fosters equitable and inclusive learning environments. Since in the research presented here we investigate students’ motivational beliefs and their perception of the inclusiveness of the learning environment, we start with a brief background on each.

Self-efficacy in a particular discipline is a student’s belief in their ability to solve a particular problem or goal [37,38]. Self-efficacy has been shown to impact students’ engagement, learning, and persistence in science courses [28,30,33,37,39,40,41,42,43,44,45,46,47,48,49,50,51]. When tackling difficult problems, students with high self-efficacy tend to view the problems as challenges that can be overcome, whereas people with low self-efficacy tend to view them as personal threats to be avoided [37]. However, in introductory physics courses in which women are underrepresented, studies have found a gender gap in self-efficacy favoring men that sometimes widens by the end of the course even in interactive engagement courses [39,52]. Similarly, interest in a particular discipline may affect students’ perseverance, persistence, and achievement in a course [49,50,53,54,55,56]. One study showed that changing the curriculum to stimulate the interest of the female students helped improve all of the students’ understanding at the end of the year [57]. Within the expectancy-value theory, interest and competency beliefs (closely related to self-efficacy) are connected constructs that predict students’ academic outcomes and career expectations [58]. Additionally, perceived recognition has been shown to play an important role in women’s motivation [59]. However, studies have shown that female students do not feel recognized appropriately even before they enter college [20,60,61]. In a study on students’ perception of support, teacher support was more strongly linked to the motivation and engagement of girls than boys [59].

A student’s identity in STEM disciplines is important to study since it plays a key role in students’ participation in classes as well as career decisions [29,60,61,62,63,64]. For example, “physics identity” has been studied in physics classes and has been shown to be connected with a student’s self-efficacy, interest, and perceived recognition [35,36,62]. The science identity framework draws on the work of Gee, who defined an individual’s identity as being recognized as a certain kind of person in a given context and emphasized that identity can change over time [62]. This has been adapted in the science context to address the identities of both students and scientists [61]. Carlone and Johnson’s science identity framework includes three interrelated dimensions: competence (“I think I can”), performance (“I am able to do”), and recognition (“I am recognized by others”) [61]. Hazari et al. modified the framework by adapting it to physics specifically rather than science more generally [60]. “Competence” and “performance” were defined as students’ beliefs in their ability to understand the subject and students’ belief in their ability to perform physics tasks. Additionally, recognition was framed as recognition by others as being a good physics student. Lastly, a fourth dimension, interest, was added to the framework since students have highly varying levels of interest in physics [65,66]. In later studies by Hazari for introductory students, performance and competence are combined into one variable [36]. In a slightly reframed version of Hazari’s physics identity framework by Kalender et al. [35], performance/competence was framed as self-efficacy (closely related to competency belief). Additionally, recognition was framed explicitly as “perceived recognition” by students for clarity.

Students’ identity in STEM disciplines plays an important role in students’ participation in classes as well as career decisions [29,60,61,62,63,64]. However, prior studies have shown that it can be more difficult for women to form a physics identity than men [20,21,22,67]. This could be due to gender stereotypes and biases about who can excel in physics courses. In general, the image of a physicist is portrayed as male, which can make women feel less welcome and accepted in the physics community. In addition, the innate abilities of genius and brilliance are often seen as important factors necessary to succeed in physics [68]. However, genius is often associated with boys [69], and girls from a young age tend to shy away from fields associated with innate brilliance or genius. Archer et al. [20] investigated the impact of physics-related cultural attributions on girls’ or women’s decisions to pursue physics and reported that science-keen girls or young women who name physics as their favorite subject slowly lose their interest due to alienation, discrimination, and gender-biased beliefs about physics. All of the stereotypes and biases can influence women’s perception of their ability to do physics before they enter the classroom. In addition, faculty members’ unconscious gendered beliefs regarding the students’ ability can be one source of the threat and alienation that women in STEM experience [26]. One study showed that science faculty members in biological and physical sciences exhibit biases against female students by rating men significantly more competent when the curriculum vitae are identical except for the name of the student being a male or female name [26]. These highly problematic stereotypes and biases are founded in the historical marginalization of certain groups, e.g., women in physics, and continue to manifest today in many ways, including gendered beliefs and barriers to women excelling in physics when there is no explicit focus on making the learning environment equitable and inclusive.

In addition to motivational factors that predict physics identity, other motivational factors that contribute to the student perception of the inclusiveness of the learning environment can also influence how women perform in their STEM classes and beyond [70]. For example, students’ interaction with peers has been shown to enhance understanding and engagement in courses. In addition, students’ sense of belonging in physics has been shown to correlate with retention and self-efficacy [59,71], so it is important to understand how it predicts both the performance and motivational outcomes of women and men at the end of the course.

Inequitable outcomes in students’ beliefs and performance may be a result of inequitable access to resources, inadequate support, and inequitable learning environments. Thus, it is important to investigate student perception of the inclusiveness of the learning environment in STEM courses to foster an inclusive education ecosystem in which all students regardless of their demographic group affiliations can succeed. The study reported here used structural equation modeling (SEM) to investigate factors in physics 1 that are part of the inclusiveness of the learning environment and which can influence student motivational and academic outcomes. We analyzed factors that instructors have control over as part of the inclusiveness of the learning environment in their courses, specifically students’ interaction with their peers, students’ sense of belonging in the course, and their perceived recognition by others (including instructors and teaching assistants or TAs).

While many studies have investigated gender differences in beliefs in introductory physics courses, most have not considered the inclusiveness of the students’ learning environment. The inclusiveness of the learning environment includes experiences students have in the classroom as well as interactions outside of the classroom, such as students’ experiences during office hours or via email correspondences with the instructor or TA and students studying or doing homework together. We control for students’ high school factors that include their high school grade point average (GPA) and standardized math scores (SAT math score) as well as their self-efficacy and interest at the beginning of physics 1 since these are students’ beliefs about physics when they enter the course based on prior experiences. The perception of the inclusiveness of the learning environment consists of the student’s perception of the effectiveness of peer interaction, their sense of belonging, and perceived recognition (from instructors, TAs, friends, and family). We discuss an investigation of the students’ outcomes pertaining to their physics performance (as measured by the end of the semester grade) as well as their physics self-efficacy, interest, and identity at the end of physics 1 to answer the following research questions. Our final statistical model that shows the path analysis is shown in Figure 1.

• RQ1 Are there gendered differences in students’ beliefs at the end of the physics 1 course and do they change from the beginning to the end of the course?
• RQ2 Do academic measures (e.g., high school grade point average and standardized math scores) predict students’ motivational beliefs and performance at the end of the physics 1 course?
• RQ3 How do the students’ perceptions of the inclusiveness of the learning environment predict their motivational and performance outcomes at the end of the physics 1 course?

2. Materials and Methods

2.1. Participants

In this study, a motivational survey covering the components in our model was administered to students at the beginning and end of one semester of a traditionally taught introductory algebra-based physics 1 course at a large research university in the US. The course is typically taken by students on the bioscience track in their junior or senior year of undergraduate studies. The course had many sections and had three primarily traditionally taught lecture classes and one recitation per week. There was a similar grading policy across sections such that in addition to some points for homework, students’ grades were mainly based on 2–3 midterm exams and 1 final exam. In general, there were little to no evidence-based active learning strategies implemented in the courses and there was no intentional effort to make the course equitable and inclusive. We analyzed the data for 501 students who completed the survey. The university provided demographic information such as gender using an honest broker process by which the research team received the information without knowledge of the identities of the participants. From the university data, the participants were 35% male and 65% female students. We recognize that gender is not a binary construct. However, the data provided by the university only included binary options of male or female (less than 1% of the students did not provide this information and thus were not included in the analysis).

2.2. Motivational Survey Instrument Validity

The validated motivational survey instrument used in the study measured students’ physics identity, self-efficacy, and interest as well as their perception of the inclusiveness of the learning environment as measured by their sense of belonging, perceived recognition, and interaction with their peers. Thus, the survey questions asked about different aspects of the students’ beliefs at two points in time (beginning and end of the course) and student perception of the inclusiveness of the learning environment at the end of the course. Students were asked to answer all of the survey questions with regard to the physics course they were enrolled in. The physics identity questions evaluated whether the students saw themselves as a physics person, i.e., a person who can excel in physics [27]. The physics self-efficacy questions measured students’ confidence in their ability to answer and understand physics problems [27,72,73,74,75]. The interest in physics questions measured students’ enthusiasm and curiosity to learn physics and ideas related to physics [73]. The sense of belonging questions evaluated whether students felt like they belonged in the introductory physics classroom [59,76]. The perceived recognition questions measured the extent the student thought that other people see them as a physics person [27]. Lastly, the effectiveness of peer interaction questions measured whether students thought that working with their peers was beneficial to their confidence to do physics [77,78]. The questions in the study were designed on a Likert scale of 1 (low endorsement) to 4 (high endorsement) except for the sense of belonging questions which were designed on a scale of 1 to 5 [79]. A lower score was indicative of a negative endorsement of the survey construct while a higher score was related to a positive belief in the construct.

The survey items were adapted from previous research [74,76,78,80] and revalidated in our own context using one-on-one student interviews, Exploratory Factor Analysis (EFA), Confirmatory Factor Analysis (CFA) [81], Pearson correlations, and Cronbach alpha. The survey questions for each construct and factor loadings for each question are given in

Table 1. Survey questions for each of the motivational constructs along with factor loadings (lambdas, $\lambda )$ from the Confirmatory Factor Analysis (CFA) results for all students (n = 501). The rating scale for most of the self-efficacy and interest questions was NO!, no, yes, YES! while the rating scale for the physics identity, peer interaction, and perceived recognition questions was strongly disagree, disagree, agree, strongly agree. The rating scale for the physics belonging questions was not at all true, a little true, somewhat true, mostly true, and completely true. Superscript ^† means the rating scale was never, once a month, once a week, every day. Superscript ^†† means that the rating scale was very boring, boring, interesting, very interesting. All p-values (of the significance test of each item loading) are p < 0.001.

Construct and Item	$\mathit{\lambda }$
Physics Identity
I see myself as a physics person	1.00
Physics Self-Efficacy
I am able to help my classmates with physics in the laboratory or recitation	0.60
I understand concepts I have studied in physics	0.74
If I study, I will do well on a physics test	0.76
If I encounter a setback in a physics exam, I can overcome it	0.72
Physics Interest
I wonder about how physics works †	0.57
In general, I find physics interesting ††	0.75
I want to know everything I can about physics	0.74
I am curious about recent discoveries in physics	0.64
Physics Perceived Recognition
My family sees me as a physics person	0.89
My friends see me as a physics person	0.92
My physics instructor and/or TA sees me as a physics person	0.71
Physics Belonging
I feel like I belong in this class	0.81
I feel like an outsider in this class	0.74
I feel comfortable in this class	0.82
I feel like I can be myself in this class	0.62
Sometimes I worry that I do not belong in this class	0.69
Physics Peer Interaction
My experiences and interactions with other students in this class…
Made me feel more relaxed about learning physics	0.75
Increased my confidence in my ability to do physics	0.93
Increased my confidence that I can succeed in physics	0.96
Increased my confidence in my ability to handle difficult physics problems	0.90

.

Additionally,

Table 2. Pearson correlations are given between all of the predictors and outcomes. All p-values are < 0.001.

Pearson Correlation Coefficient
Observed Variable	1	2	3	4	5	6	7	8	9	10
1. SAT Math	--	--	--	--	--	--	--	--	--	--
2. H.S. GPA	n.s.	--	--	--	--	--	--	--	--	--
3. Pre Self-Efficacy	n.s.	n.s.	--	--	--	--	--	--	--	--
4. Pre Interest	n.s.	n.s.	0.65 *	--	--	--	--	--	--	--
5. Perceived Recognition	0.15 **	n.s.	0.30	0.33	--	--	--	--	--	--
6. Peer Interaction	n.s.	n.s.	0.21 **	0.25	0.46	--	--	--	--	--
7. Belonging	0.26	n.s.	0.36	0.25	0.59	0.62	--	--	--	--
8. Post Self-Efficacy	0.30	n.s.	0.41	0.29	0.62	0.62	0.81	--	--	--
9. Post Interest	n.s.	0.13 *	0.19 *	0.66	0.61	0.47	0.51	0.60	--	--
10. Grade in physics 1	0.48	n.s.	0.13	n.s.	0.37	0.23	0.44	0.49	0.20	--
11. Physics Identity	0.14 **	n.s.	0.31	0.38	0.79	0.43	0.58	0.60	0.62	0.29

* = p < 0.05, ** = p < 0.01, no * = p < 0.001, n.s. = non significant (p > 0.05).

shows the Pearson’s correlation r values which signify the strength of the relationship between constructs. The table shows that the correlations vary in strength, but none of the correlations in our survey are so high that the constructs cannot be separately examined. The only high inter-correlations were between pre interest and post interest (0.86). Past work has shown that there is a high correlation in interest throughout the introductory courses in which no explicit hands-on and minds-on efforts are made to make students more interested [81]. All correlations in Table 2 are low enough that they can be considered separate constructs.

2.3. Analysis

Initially, we compared female and male students’ mean scores for the predictors and outcomes in our model (see Figure 1) for statistical significance using t-tests and for the effect size using Cohen’s d [82]. Cohen’s d is defined as $d=\left({\mu }_m-{\mu }_f\right)/{\sigma }_{pooled}$, where ${\mu }_m$ is the average score of male students, ${\mu }_f$ is the average score of female students and ${\sigma }_{pooled}$ is the pooled standard deviation (or weighted standard deviation for men and women) for all students. ${\sigma }_{pooled}$ is defined as ${\sigma }_{pooled}=\left({\sigma }_m{}^2+{\sigma }_f{}^2\right)/2$, where ${\sigma }_m$ is the standard deviation for men and ${\sigma }_f$ is the standard deviation for women.

To quantify the statistical significance and relative strength of our framework’s path links, we used Structural Equation Modeling (SEM) as a statistical tool by using R (lavaan package) with a maximum likelihood estimation method [80]. We report the model fit for SEM by using the Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residuals (SRMR). Commonly used thresholds for the goodness of fit are as follows: CFI and TLI > 0.90, and RMSEA and SRMR < 0.08 [83]. Additional details about SEM can be found in the following papers [70,84].

3. Results

3.1. Gendered Differences in Predictors and Outcomes

Statistically significant differences were found in the majority of the predictors and outcomes in favor of male students, except for the high school grade point average which favored women (see

Table 3. Mean predictor and outcome values as well as effect sizes (Cohen’s d) by gender (n = 292 for women and n = 144 for men for the pre values and n = 327 for women and n = 174 for the post values and high school factors since we only required students to have the post values). Statistical significance (p-values) is given by superscript a for p = 0.002, superscript b for p = 0.171, and no superscript for p < 0.001.

Predictors and Outcomes (Score Range)	Mean		Cohen’s d
	Male	Female
High School Grade Point Average (0–5)	4.02	4.17	−0.32 a
Standardized Math Score (200–800)	680	670	0.16 b
Pre Self-Efficacy (1–4)	3.07	2.83	0.58
Pre Interest (1–4)	2.74	2.45	0.58
Post Self-Efficacy (1–4)	2.92	2.55	0.66
Post Interest (1–4)	2.66	2.27	0.69
Physics 1 Grade (0–4)	3.17	2.89	0.36
Perceived Recognition (1–4)	2.22	1.92	0.43
Peer Interaction (1–4)	2.90	2.58	0.43
Belonging (1–5)	3.58	3.01	0.60
Physics Identity (1–4)	2.04	1.64	0.59

). Additionally, there was no statistically significant difference in the standardized math scores between men and women. The societal stereotypes, biases, and previous experiences women have before they come into the class could be one reason women have lower scores on these beliefs than men even at the beginning of the course (pre). Both women’s and men’s self-efficacy and interest drops from the pre test to the post test. Moreover, student perceptions of the inclusiveness of the learning environment at the end of the course are lower for women than men (see perceived recognition, perceived effectiveness of peer interaction, and sense of belonging in Table 3). Since students’ beliefs have important implications for their course engagement and learning outcomes, it is critical for instructors to create a sustainable educational ecosystem in which hands-on and minds-on activities center equity and inclusion to ensure that all students regardless of their demographic group have similar outcomes (both in terms of their performance and their beliefs). In Appendix A, we provide the percentage of men and women who selected each response to the questions. This provides a sense of how students answered each question.

3.2. SEM Path Model

We used SEM to investigate the relationships between the constructs and to unpack each construct’s contribution to explaining the self-efficacy, interest, physics identity, and grades of women and men at the end of physics 1. We initially tested moderation analysis of variables using multi-group SEM between female and male students to see if any of the relationships differed across gender. There were no group differences at the level of weak or strong measurement invariance and regression coefficients, so we proceeded to mediation analysis. We used mediation analysis to understand the extent that gendered differences in students’ outcomes at the end of physics 1 (self-efficacy, interest, grade, and physics identity) were mediated by differences in student’s initial self-efficacy, interest, pre-college academic measures (high school grade point average and standardized math scores), and the perception of the inclusiveness of the learning environment in the physics class.

In our model, the inclusiveness of the learning environment (students’ sense of belonging, perception of the effectiveness of peer interaction, and perceived recognition) mediated the outcomes. The result of the SEM is presented in Figure 2. The model fit indices indicate a good fit to the data: CFI = 0.924 (>0.90), TLI = 0.914 (>0.90), RMSEA = 0.051 (<0.08), and SRMR = 0.080 (<0.08). Self-efficacy, interest, and perceived recognition had a direct effect on physics identity, and there were no direct effects from gender, grade in physics 1, or any of the pre college and pre test factors. Students’ perceived recognition had the largest correlation (β = 0.59) while self-efficacy (β = 0.13) and interest (β = 0.18) had similar, but smaller direct effects on physics identity.

Additionally, self-efficacy at the end of physics 1 is largely correlated with student’s sense of belonging (β = 0.44) with additional effects from students’ pre self-efficacy (β = 0.26), peer interaction (β = 0.26), standardized math scores (β = 0.14), and perceived recognition (β = 0.12). Similarly, interest at the end of physics 1 had the largest correlation with pre interest (β = 0.71) with smaller effects from peer interaction (β = 0.21), perceived recognition (β = 0.11), and high school grade point average (β = 0.12). Lastly, student’s grade in physics 1 was correlated with their standardized math score (β = 0.40), student’s sense of belonging (β = 0.25), and perceived recognition (β = 0.15).

Interestingly, we found that gender only directly correlated with students pre self-efficacy (β = 0.32) and pre interest (β = 0.33) as well as smaller correlations with peer interaction (β = 0.14) and belonging (β = 0.13). The gendered differences in the pre-test values may be partially caused by past experiences and societal stereotypes about physics women have before they enter the physics classroom. However, women have additional negative experiences in the classroom from the perceptions of the inclusiveness of the learning environment factors, despite women making up the majority of students in the class.

4. Discussion and Implications

In accordance with our framework, all students, especially those from marginalized demographic groups such as women, must be given adequate support to excel in their coursework as well as to develop high STEM-related motivational beliefs in order to foster a sustainable science education ecosystem and a sustainable STEM workforce. If the STEM learning and work environments do not provide equitable opportunities for individuals from all demographic groups to contribute their talents, quality and innovation will be compromised. In other words, in order to create a sustainable STEM ecosystem, equity and inclusion must be centered so that all students can be supported in order for them to realize their potentials. However, using an example from an introductory physics 1 course for students on the bioscience track at a large university in the US in which women are not underrepresented, we find evidence of an inequitable and non-inclusive learning environment that can be a major impediment in fostering a sustainable STEM ecosystem. In particular, we find that women have lower grades and motivational beliefs than men at the end of physics 1. This trend is comparable to the calculus-based physics courses [35,85] in which women are underrepresented, showing that these inequities cannot simply be explained by the numerical representation of women and have their roots in stereotypes and biases related to who belongs in these STEM disciplines and who can excel in them as well as the dominant culture in these disciplines that perpetuates the inequities.

In response to RQ1, we find that women had lower motivational beliefs and perceptions of the inclusiveness of the learning environment in physics 1. The course learning environment was not equitable and inclusive enough to eliminate these gender differences. Our finding shows that the percentage of women in a physics course is not enough to create a sustainable educational ecosystem; instead, the learning environment must be equitable and inclusive in order for women (and other marginalized students) to excel. Since students’ performance and beliefs in physics courses can impact their future career choices, it is important to create a sustainable educational ecosystem and make physics learning environments equitable and inclusive so that the gaps between marginalized and dominant groups can be eliminated.

In response to RQ2, our SEM model indicates that the high school factors (high school grade point average and standardized math scores) predicted students’ interest, self-efficacy, perceived recognition, and grade at the end of the course. However, the perception of the inclusiveness of the learning environment factors had larger correlations with students’ beliefs at the end of the course.

In response to RQ3, the inclusiveness of the learning environment factors was important for explaining students’ physics self-efficacy, interest, grade, and identity at the end of the course. Perceived recognition predicted all of the motivational and academic outcomes at the end of the course while the effectiveness of peer interaction predicted students’ self-efficacy and interest and sense of belonging predicted students’ grades and self-efficacy. Although interest at the end of the course was primarily predicted by interest at the beginning of the course, instructors have the potential to improve students’ interest in physics if they explicitly focus on it as a goal. One possible way to influence students’ interest is to engage them with problems that relate to their prospective majors and occupations or that are of interest to them in general.

A limitation of our quantitative study focusing on descriptive and inferential quantitative analysis is that we can only gain insight into the relative values of motivational beliefs of men and women at the beginning and end of the course and how the relation between gender and physics identity is mediated by physics perceived recognition, self-efficacy and interest, but we cannot establish causal effects. Therefore, future studies should include qualitative data, such as interviews or focus groups, that can be used to gain more insight into how to improve the learning environment for all students. Another limitation of the study is that the sample size in the pre test was different than in the post test.

We note that the physics course in this study was a traditionally taught lecture-based course in which student grades heavily depended on two or three midterm exams and a final exam. The courses consisted of 3 h of lecture per week taught by the instructor and 1 h of recitation per week taught by a TA. It is important to recognize that even in this traditionally taught lecture-based course, students often received feedback from their instructors in multiple ways, including receiving praise for asking or answering a question in class (which often advantages male students since they dominate these situations) and their interactions with students during office hours or over email. In addition, students interact with their TAs in recitation classes by asking questions about the homework or class material at the start of recitation, when completing group work during recitation, and during the TA’s office hours.

Instructors have the ability to positively transform the inclusiveness of the learning environment (which would lead to an increased sense of belonging, effectiveness of peer interaction, and perceived recognition) and make their classes more equitable and inclusive in order to foster a sustainable STEM ecosystem. These beliefs can influence each other as well. For example, if an instructor can improve the students’ perception of the effectiveness of peer interaction, by allowing students to work in groups during class and ensuring that all students feel safe, valued, and respected participating in the discussions without the fear of being wrong, they could influence the students’ sense of belonging as well. In other words, if instructors can provide support for one of the factors they can most readily control (e.g., effectiveness of peer interaction) and make their classes more equitable and inclusive, they are likely to improve student outcomes in the process. In the hands-on lab, it would be beneficial to have students contribute equally to each task as opposed to splitting the group work so that all students have the opportunity to engage in all the types of work that make up science (as opposed to women becoming secretaries and managers and men doing the tinkering with lab equipment) [86]. Moreover, brief social-psychological classroom interventions have also been shown to eliminate or reduce the gender gap in performance [87,88,89,90]. Creating an inclusive learning environment and inculcating a growth mindset, i.e., intelligence is not immutable and one can excel in physics by working hard and working smart, can go a long way in helping all students engage effectively and benefit from research-validated tools and approaches [90,91]. However, if the instructor does not make a concerted effort to inoculate students against stereotype threats [92](i.e., fear of confirming a negative stereotype about one’s group), the inequitable and non-inclusive learning environment is more likely to hurt women and the goal of creating a sustainable STEM ecosystem will be compromised.

Other researchers have pointed to structural changes to implement at the institution and classroom level to make the learning environment more equitable and inclusive and make the science education ecosystem sustainable [57,93,94,95]. Structural changes at the institution level require centering disadvantaged students in the design of curricula and pedagogies. Instructors can make their courses student-centered, e.g., by adopting pedagogy that focuses on societal implications of physics [57] in addition to providing mentoring/support for students who are marginalized [94]. These students must be provided appropriate mentoring, guidance, scaffolding, and support in college so that the structural hurdles they encounter in STEM fields can be dismantled and they are not put at a disadvantage relative to their privileged peers [95].

In summary, instructors and teaching assistants need to provide an inclusive learning environment that emphasizes recognizing their students, allows for positive peer interactions, and provides a space where all students can feel that they belong. From our analysis presented here, it is clear that student perceptions of the inclusiveness of the learning environment factors play a central role in predicting not only students’ grades but also their self-efficacy, interest, and identity at the end of the course. We emphasize again that it is important to note that student perception of the inclusiveness of the learning environment is not shaped only by what happens in the classroom. Student interactions with each other while they do their homework, students’ experiences during an instructor’s or TA’s office hours, interactions between students and the instructor over email, and other circumstances all contribute to the students’ perceptions of the inclusiveness of the learning environment.

We hope that this research conducted in traditionally taught physics classes in the US can serve as an example of how the current science education ecosystem is not sustainable because students from marginalized groups, e.g., women, are continuing to have concerns about the inclusiveness of the learning environment and this perception predicts their performance as well as beliefs at the end of the course. In accordance with our framework, it is important to make intentional efforts to create an innovative equitable and inclusive learning environment to help create a sustainable science education ecosystem so that all students can benefit from the hands-on and minds-on learning regardless of their demographics.