Before the lecture begins: unpacking how affective measures impact performance in general chemistry 1
Vanessa P.
Bustamante
and
Regina F.
Frey
*
Department of Chemistry, University of Utah, Salt Lake City, UT 84112, USA. E-mail: gina.frey@utah.edu
First published on 23rd October 2025
Abstract
General chemistry is often the first course taken by students interested in STEM and health; therefore, it is considered critical for their retention in these fields. Success in these courses is shaped by multiple factors, including economic disparities, academic preparation, and both affective and cognitive dimensions. While affective measures have been increasingly studied in relation to student performance and retention, few studies have explored the interrelationships between affective variables and how they collectively influence performance or retention. Using structural equation modeling (SEM) path analysis, this study examined the connections between four early-semester affective variables (sense of belonging, academic mindset entity (fixed mindset), science resources, and imposter syndrome) and their impacts on exam performance in a large first-semester general chemistry course (N = 354). Additionally, mediation analyses were conducted to further identify indirect effects among these variables. Key findings revealed that exam performance was directly predicted by students’ early semester academic mindset entity and experiences of imposter syndrome, and indirectly affected by science resources. Science resources also directly affected their early-semester academic mindset entity, while indirectly affecting imposter feelings through academic mindset entity. These findings suggest the importance of recognizing the varied science backgrounds and experiences students bring into the classroom. Actively designing pedagogical strategies to support them may improve not only cognitive outcomes but also affective experiences such academic mindset and imposter syndrome. Supporting STEM success requires addressing not only the cognitive domain, but also the beliefs, emotions, and prior experiences students bring with them into the classroom.
Introduction
Introductory STEM courses, such as General Chemistry, often serve as early requirements for students pursuing careers in STEM or healthcare. However, success in these courses is shaped by a range of factors, including economic disparities (Aikens and Barbarin, 2008; Brown et al., 2016; Doerschuk et al., 2016), academic preparation (Tai et al., 2005; Micari et al., 2016), as well as affective and cognitive factors (Cromley et al., 2016; Han et al., 2017; Hilts et al., 2018; Edwards et al., 2021; Van Sickle and Frey, 2025). While research regarding cognitive factors has shown that mathematical ability (Lewis and Lewis, 2007; Korpershoek et al., 2015; Willis et al., 2022), prior conceptual knowledge (Seery, 2009; Xu et al., 2013), and scientific reasoning ability (Thompson et al., 2018) influence student performance, there is still uncertainty in research on how affective constructs directly or indirectly affect student performance and how these affective constructs relate to each other. Hence, student success in introductory STEM courses, such as General Chemistry 1, continues to be a concern for educators. Further research is needed in examining the roles and interrelationships of affective constructs in shaping student outcomes in introductory STEM courses, such as General Chemistry, and improving student success.To develop and implement well-informed, effective interventions or teaching strategies to enhance student retention, we need to better understand what factors influence student course-level success. Previous research studies have looked at what factors influence the lack of persistence within STEM fields and have found that affective constructs such as social belonging, academic mindset, self-concept, and imposter syndrome affect all students (Choi, 2005; Fink et al., 2018, 2020; Edwards et al., 2021; Moreno et al., 2021; Demirdöğen and Lewis, 2023). The current study adds to this effort by offering new insights into how various affective factors–such as academic mindset entity (fixed mindset), imposter syndrome, sense of belonging, and childhood science resources–interrelate and influence student performance in a General Chemistry 1 course.
Background
Affective constructs such as mindset, belonging, and impostorism, to name a few, have been individually studied in educational settings. These studies have provided foundational insights, enabling researchers and instructors to better understand these “unseen” measures that often influence student outcomes. In this study, we add to the literature by approaching these constructs as a collective. Herein, we discuss each construct in further detail.Academic mindset
Mindset theory, also known as implicit theories of intelligence, posits that individuals have one of two mindsets of intelligence: an entity/fixed mindset or an incremental/growth mindset (Dweck, 2008). In an academic setting, students with a fixed mindset believe that intelligence cannot be changed, seeing academic failures as a reflection of their intelligence. Conversely, students with a growth mindset believe intelligence is malleable and that they can learn from their mistakes or failures (Dweck, 2008). The belief in the capacity to change intelligence has significant consequences for achievement (Dweck, 1999; Robins and Pals, 2002; Blackwell et al., 2007). Importantly, however, believing that intelligence can be developed does not guarantee that students feel confident in their ability to improve. De Castella and Byrne found that students’ (adolescent-youth) perception of their own intelligence, rather than intelligence in general, was a better predictor of their motivation, engagement, and performance in school (De Castella and Byrne, 2015).Research has shown that college students with a growth mindset have been associated with greater achievement than those with fixed mindsets (Blackwell et al., 2007), resulting in persistence, resilience, and success in learning (Dweck and Molden, 2000; Dweck, 2008; Yeager and Dweck, 2012). Additionally, in a study of 643 Australian high school students aged 15 to 19, De Castella and Byrne found that entity (fixed) mindset beliefs were predictive of lower endorsement of achievement goals, greater helplessness attributions and poorer self-reported academic grades. Entity mindset beliefs were also predictive of academic self-handicapping, truancy and disengagement (De Castella and Byrne, 2015). As a result, growth mindset interventions were designed to counteract a fixed mindset by teaching students about the malleability of their intelligence and increasing their openness to take on challenges, promoting academic achievement and persistence. Successful mindset interventions have been shown to boost student grades and increase persistence in STEM (Yeager and Walton, 2011; Yeager et al., 2019; Fink et al., 2020, 2023).
While much research focuses on students, mindsets also significantly influence instructors (Canning et al., 2019; Muenks et al., 2020; LaCosse et al., 2021; Kroeper et al., 2022). Canning and O’Leary's work emphasizes the importance of instructors' mindset in creating a supportive classroom (Canning et al., 2019; O’Leary et al., 2020). A Structural Equation Modeling (SEM) path analysis conducted by Kattoum and colleagues found that when students perceived their instructor as endorsing a fixed mindset, they experienced a greater sense of academic misfit, which in turn led to lower chemistry grades. These findings underscore the important role that mindset plays in shaping academic experiences and outcomes for students, stressing the need for continued efforts to understand what affects students’ mindsets.
Imposter syndrome
The terms imposter syndrome (IS) and imposter phenomenon (IP) are used interchangeably in the literature. The concept was first introduced in 1978 by Clance and Imes, who described it as the persistent feeling of inadequacy and fear of being exposed as a fraud, despite evidence of competence, and is an experience that can affect anyone (Clance and Imes, 1978). According to Clance and Imes, individuals experiencing these feelings may attempt to compensate by working excessively hard, leading to strong performance and approval from authority figures. Over time, the definition has broadened to include experiences of discomfort when succeeding, attributing successes to factors other than ability, and denial of competence (Clance, 1985; Harvey and Katz, 1985). Moreover, Clance noted that these individuals would “attempt to avoid these feelings and prevent the discovery of their impostorism by working extra hard, which pays off in excellent performances and approval from authorities.” (Clance and Imes, 1978). In 2000, Leary and colleagues hypothesized a modification of Clance and Ime's definition to include two types of imposters: true imposters who believe that others perceive them too positively, and strategic imposters who claim they are not as good as others think (Leary et al., 2000). While their results did not support this distinction, evidence of the strategic nature of impostorism was observed. In other words, Leary and colleagues suggest that people may experience true feelings of impostorism, but the characteristics attributed to “‘so-called’ impostors are partly interpersonal, self-presentational behaviors designed to minimize the implications of poor performance” (Leary et al., 2000).Imposter syndrome can affect anyone. In a longitudinal study of college STEM courses, Canning and colleagues found that perceived classroom competition was associated with greater daily in-class imposter feelings among all students and was especially significant among first-generation students (Canning et al., 2020). Additionally, Canning et al. found that these imposter feelings predicted students’ end-of-term course engagement, attendance, dropout intentions, and course grades (Canning et al., 2020). However, research has seen a disproportionate effect on women (Clance and Imes, 1978; McGregor et al., 2008; Tao and Gloria, 2019; Vaughn et al., 2020). Altogether, these findings capture the complex nature of impostor syndrome, raising essential questions about the influence of additional affective factors on this phenomenon. For example, previous research studies on BIPOC, women, and first-generation university students have shown that imposter syndrome impedes a sense of belonging (Pulliam and Gonzalez, 2018; Stone et al., 2018; Trefts, 2019; Collins et al., 2020; Vaughn et al., 2020). However, the majority of these studies focused on STEM professionals, such as faculty (Clance and Imes, 1978; Chakraverty, 2022), graduate students (Hutchins, 2015), and medical students (Gottlieb et al., 2020). As such, there is a need for additional research on imposter syndrome in undergraduate STEM majors, specifically in introductory STEM courses such as general chemistry.
Social belonging
Psychologist Abraham Maslow described the need to belong as a basic human necessity, placing it above the desire for knowledge and understanding in his hierarchy of needs (Maslow, 1962). Earlier models of belonging in higher education involved the integration of how student persistence stems from both social and academic integration into the campus environment, which would in turn predict student engagement and involvement (Tinto, 1975, 1993; Naylor et al., 2018). In an educational setting, student belonging has been shown to be a key ingredient for persistence and achievement. Furthermore, belonging research has found that students who felt a lower sense of belonging in their science courses (i.e., chemistry, physics, and biology) were less likely to persist in STEM (Hazari et al., 2017; Rainey et al., 2018; Fink et al., 2020; Edwards et al., 2023; Hansen et al., 2023).Course-level social belonging comprises two factors: sense of belonging and belonging uncertainty (Edwards et al., 2021). Sense of belonging refers to a student's perception of being accepted, valued, included, and encouraged by others (teachers and peers): “Do I belong in this course?” (Edwards et al., 2021; Hansen et al., 2023). In contrast, belonging uncertainty refers to a feeling of insecurity in relation to being part of a particular background: “Do people like me belong in this course?” where “people like me” may be in reference to being part of a certain group (e.g., gender, race/ethnic group, first vs. continuous generation students, socio-economic status) or to their perceived academic ability. An early study in a college computer science course found that women were more likely to have a higher belonging uncertainty and a lower sense of belonging (Höhne and Zander, 2019a, b). Additionally, there is a relationship between a student's sense of belonging and their performance in STEM courses (Gopalan and Brady, 2020; Chen et al., 2021; Edwards et al., 2021, 2022; Gehringer et al., 2022; Jackson et al., 2023; Brown et al., 2024; Fong et al., 2025), showing the importance of the belonging construct. Although student belonging is a significant measure in its own right, it is also essential to understand the associations belonging might have with other affective measures, as well as student achievement.
Childhood science currency
In a general context, capital refers to assets that can be used to generate income. From a psychological standpoint, Pierre Bourdieu conceptualized capital (namely economic, social, cultural, and symbolic) as legitimate, valuable, and exchangeable resources within a society that can provide social advantage to individuals or groups, particularly within specific fields (e.g., education) (Bourdieu, 1977, 1984, 1986). Expanding on Bourdieu's foundation, Archer and colleagues (Archer et al., 2014) extended the concept to the STEM education field by introducing science capital, which they define as:“Conceptual device for collating various types of economic, social and cultural capital that specifically relate to science—notably those which have the potential to generate, use, or exchange value for individuals or groups to support and enhance their attainment, engagement and/or participation in science.”
In this way, science capital captures the augmentation of Bourdieu's conceptualization of capital through a scientific lens. Archer's science capital can be conceptualized through the following “currencies”: (1) science-related cultural capital, which includes scientific literacy, science-related preferences/dispositions, and symbolic knowledge about the transferability of science in the labor market, (2) science-related behaviors and practices, which include consumption of science-related media and participation in out-of-school science learning contexts, and (3) science-related social capital, which includes knowing someone who works a science job, parental science qualifications, talking to others about science, future science affinity, and science identity (Archer et al., 2014).
Studies have demonstrated that families play a crucial role in shaping students’ engagement, aspirations, and achievement/persistence in science (Dabney et al., 2013). For example, the interplay between capital and social culture within families can shape children's values, attitudes, expectations, and behaviors in ways that promote academic advancement (Sandefur et al., 2006; Archer et al., 2012; Gaddis, 2013). Additionally, families may leverage their resources to access additional opportunities such as private schooling or additional enrichment activities such as music lessons, after-school tutoring, and additional language programs for their children, aiming to enhance and set them apart within the educational landscape (Vincent and Ball, 2007; Hodges et al., 2017). Furthermore, research has shown that families with greater science-related resources actively foster their children's interest and aspirations in science by integrating science into everyday life, such as providing science kits, watching science programs together, discussing scientific topics, and visiting science museums. (Archer et al., 2012; Sha et al., 2016; Chakraverty et al., 2018; Wenner, 2024). To date, there have not been studies in higher education on the effect of “Childhood Science Currency” or their individual factors (parental influence and childhood science resources) on student performance in STEM courses.
Affective constructs and their effect on course exam performance
Previous studies have shown that students’ early semester sense of belonging (SB) can affect exam grades in general chemistry (Fink et al., 2020; Edwards et al., 2021) (GC1) and introductory physics courses (Edwards et al., 2022). Therefore, this relationship was explored in our current analysis. Additional studies explored how our other variables of interest affected academic outcomes. In a longitudinal study, 818 students across several STEM courses were surveyed to assess the relationship between perceived classroom competition and impostor syndrome (Canning et al., 2020). Students were asked to report their perceptions of classroom competition in their STEM classes in relation to these impostor feelings. Their results showed that impostor feelings not only affected students' course grades but also their end-of-term course engagement, attendance, and dropout intentions. Limeri et al. found that mindset affects STEM students’ course grade, intent to persist, and sense of belonging (Limeri et al., 2023). However, other studies have found no significant relationship between students’ mindset and their course performance. For example, Cavanagh et al. (2018) examined an anatomy and physiology course for science majors and found that while both student trust in the instructor and students’ views of their own intelligence were associated with commitment to and engagement in active learning, only trust in the instructor was significantly related to final grades; i.e., students’ intelligence mindsets were not predictive of course performance (Cavanagh et al., 2018). Similarly, Kattoum and Baille's study of 625 STEM students found that the context of mindset interventions influenced growth beliefs more strongly among traditional-aged students (18–22 years) than among non-traditional students (over 22) (Kattoum and Baillie, 2025). However, previous studies suggest that mindset interventions can influence academic outcomes. For example, Fink et al. (2018) found that a chemistry-specific, growth-mindset intervention improved GC1 performance among first-year college students from backgrounds underrepresented in STEM (Fink et al., 2018). Separately, Fink et al. (2023) found that a growth mindset intervention significantly improved exam performance in a sample of 155 nontraditional students in introductory psychology at a 2-year community college (Fink et al., 2023). Similarly, a study conducted in a K-12 setting found that a fixed mindset in 10th grade predicted lower gains in academic achievement by 12th grade (Hwang et al., 2019). Therefore, in this study, we used SEM path analysis to explore the relationship between students’ affective measures (social belonging, imposter syndrome, and academic mindset entity) collected at the beginning of the semester related to performance on Exam 1 (given midway through the semester) and the Final Exam (given at the end of the semester). We included both exams to observe the influence of these variables at two different time points in the semester.Hypothesized relationships between affective constructs
Previous works have hypothesized relationships between the latent variables discussed in this study. Kenneally et al. found a correlation between fixed-mindedness (entity) and imposter syndrome in pharmacy students at the University of Kentucky (Kenneally et al., 2023). Separately, Noskeau et al. found that adults across the United Kingdom, Ireland, and the United States (N = 201) who have a fixed mindset tend to experience more imposter phenomena at work and that this relationship was predominantly explained by their fear of failure (Noskeau et al., 2021). Based on these findings, we hypothesized the directionality of the relationship between AME and IS; i.e., students’ academic mindset will predict their imposter syndrome measure. Additionally, past works hypothesized a relationship between students’ sense of belonging and feelings of impostorism. Data gathered in a longitudinal study (first six months – end of the academic year) from 58 first-year college students found that a greater sense of social and academic belonging was correlated with lower imposter syndrome during the first six months of college (Dao et al., 2024). This observed correlation led us to explore the directionality between social belonging and imposter syndrome in this study. Separately, in a study conducted on undergraduate STEM students who were interviewed about their social capital (i.e., assets within a social network comprising information, influence, social credentials, and reinforcement), the researchers found that gaining social capital and experiencing counterspaces (i.e., settings that challenge deficit-based narratives and foster positive self-concept among marginalized individuals (Case and Hunter, 2012)) can contribute to undergraduate STEM students developing a sense of belonging (Gray et al., 2025). Similar to social capital, science capital is a conceptual device for collating various types of economic, social, and cultural capital that specifically relates to science (Archer et al., 2014). While there have been studies regarding science capital (Turnbull et al., 2020) and how it may affect students’ self-concept, there is still a gap in understanding how exposure to science-related material as a child and/or adolescent affects future college students in STEM with regards to the affective domains (social belonging, imposter syndrome, and academic mindset).Research questions (RQs)
Three of these affective constructs (belonging, mindset, and impostorism) have been shown to affect student performance individually. However, students often experience multiple affective influences simultaneously, and many may enter a course already exhibiting varying levels of these measures. As such, this study looks at the relationship between these early-semester affective measures and their effects on performance. To our knowledge, these inter-relationships and their effects on performance have not been studied, especially in General Chemistry 1 (GC1). Using structural equation modeling (SEM) path analysis, we tested path models to address the following research questions:1. To what extent do these four measures; sense of belonging, academic mindset entity, imposter syndrome, and childhood science resources, affect student exam performance in GC1?
2. What are the relationships between academic mindset entity, imposter syndrome, sense of belonging, and childhood science resources?
Methods
Study setting
This study took place at a large (∼35![[thin space (1/6-em)]](https://www.rsc.org/images/entities/char_2009.gif)
000 undergraduate students), public, research-intensive university in the Mountain-West region of the United States during the Fall 2023 semester. Student surveys, including four affective measures (academic mindset, imposter syndrome, childhood science currency, and social belonging), were collected in General Chemistry I (GC1). GC1 is the first course in a two-course general chemistry sequence required for all chemistry majors, pre-health majors, and most physics, engineering, and biology majors. This course consisted of two sections (593 students total; 250–340 per section) with a separate laboratory course. The course combined asynchronous online content with in-person components, which included lectures (three times a week), discussion sessions (once weekly on Thursdays), and optional study techniques sessions (once weekly on Tuesdays). Coursework included weekly online graded homework and timed quizzes via Canvas, along with three in-person exams (two midterms and one non-cumulative final). Additionally, attendance was taken for points at the lecture and discussion sessions.
As a team, two instructors managed the sections as one course using the same assessments, absolute grading scale, grading procedure, and course policies. Weekly Tuesday study-technique sessions were led by Learning Assistants (LAs) (Van Sickle et al. In preparation (2025)). In these sessions, students learned a new study technique, discussed it, practiced the technique, and then worked on problems related to the class material using this technique. While these sessions were purely optional, instructors highly recommended that students attend them. The study techniques that were taught in the Fall 2023 semester have been shown to improve students’ performance in GC1 (Van Sickle and Frey, 2025; Van Sickle et al. In preparation (2025)). Weekly Thursday discussion sessions were led by graduate Teaching Assistants (TAs) and assisted by undergraduate TAs and LAs. LAs at this institution are paid undergraduate students who are formally trained through a semester-long pedagogy course and work with small groups of students to facilitate collaborative active learning during the course based on the nationally recognized LA model (Barrasso and Spilios, 2021). Students were assigned a mandatory discussion section. In these sessions, students worked on instructor-selected questions in small groups and individually completed a practice quiz, followed by a group discussion of this practice quiz. Instructors, TAs, and LAs held weekly office hours, and all students could attend any office hour. All course content was posted on a common Canvas page; therefore, there were no differences in access to materials or instructors across sections.
Course grades were based on “homework” assignments (which included weekly graded homework, quizzes, a Pre-knowledge Assessment, and two surveys; 30% of grade), an introductory quiz (2% of grade), attendance in lecture and discussion (25.3% of grade), two midterm exams (27.7% of grade), and one non-cumulative final exam (15% of grade).
Participants
Participants were recruited from GC1 during the Fall 2023 semester. Of the 593 students enrolled in the course, 398 consented to participate in the study and completed all required surveys. Outliers were identified using Mahalanobis’ distance. Participants with a Mahalanobis’ distance corresponding to p < 0.001 were considered outliers and were excluded from analyses (n = 38; please see SI for list of outliers). Additionally, it was found that six consenting students who completed all required surveys had missing exam data; as such, these students were also removed from the study, resulting in a final sample size of N = 354. Of those 354, 61% were first-year students and 25% were second-year students. All recruitment and study procedures were approved by the university's Institutional Review Board (IRB_00145042). Students received course credit for completing the surveys, independent of their consent to allow their data to be used for research; no compensation was given for research participation.Demographics
Students' self-reported demographic information (race, gender, first-generation status, etc.) was obtained from the Early Semester (ES) survey, focusing on three variables: gender, first-generation status, and race or ethnicity. In response to the question, “To which gender identity do you most identify?” students selected from options: “Woman” (n = 144), “Man” (n = 210), “Non-binary” (n = 0), “Prefer to self-describe” (n = 0), and “Prefer not to answer” (n = 0). To determine first-generation status, students were asked, “Did any of your parents/legal guardians obtain a college degree?” with responses including “Yes” (n = 288), “No” (n = 62), or “I do not know/prefer not to answer” (n = 4). In response to the question, “I identify my race/ethnicity as: [select all that apply],” students indicated their identity with options: “Asian” (n = 31), “Black/African American” (n = 7), “Hispanic/Latinx” (n = 30), “Native American/Alaskan Native” (n = 7), “Pacific Islander” (n = 1), “White” (n = 238), “Prefer to self-describe” (n = 2), “Prefer not to answer” (n = 2), with 36 students selecting more than one race. We recognize the limitations of the term “Asian,” as students under this category may come from diverse cultural and ethnic backgrounds, leading to substantially different experiences in STEM (McGee et al., 2017; Chen and Buell, 2018). However, demographic variables were not included in the statistical analyses in the current study; instead, all students were analyzed as a single group.Academic preparation
Chemistry preparation was measured by initial performance on a multiple-choice, chemistry pre-knowledge assessment (average 15.18 ± 5.36 out of 25 total possible points), which was developed by experienced GC1 instructors at the university. This assessment has been used at this institution for over 10 years, and has shown, in prior GC1 studies, to be correlated with exam performance (Edwards et al., 2021). The Pre-knowledge Assessment (referred to as ‘pre-assessment’ or ‘PA’ henceforward) was administered online through Canvas, with a time limit of 50 minutes (On average, students finished within 22 minutes). The preassessment used in this study can be obtained by contacting the corresponding author.Exam grades
The first midterm of the semester, represented by the variable Exam 1 (%), was given after approximately 15 lectures with the students. Exam 3 was a non-cumulative final based on course material across approximately 18 lectures, was given during the university's final-exam testing period, and is represented by the Final Exam (%) variable. Students in all Fall 2023 GC1 sections took the same exams during the same testing periods, so exam scores were expressed in the analyses as raw percentages. Exams 1 and 3 were of interest as we wanted to see the effect of the affective variables on performance at two different time points, the midterm which was 40% of the way through the semester and the final exam which was at the end of semester.Data collection
Data for this study were collected using the online survey platform Qualtrics (Qualtrics, 2020). All students who completed the survey received credit for the assignment, regardless of whether they provided consent for their data to be used in the study. Surveys were administered twice during the semester as part of assignments in each section, with credit awarded for thoughtful completion. The first survey was distributed during the first two weeks of class to capture baseline measures of students’ academic mindset, imposter syndrome, childhood science currency, and social belonging—prior to their experiencing the course or receiving feedback on their performance. The second survey was administered during the final week of class, just before the final exam testing period. At the end of the semester, grades for assignments, participation, and exams were collected from instructors for consenting students and matched with their corresponding survey responses.Measures
As described in the Data Collection sub-section, data were collected at two time points: early semester (pr_) and late semester (po_). For the purposes of this study, only the early semester data were analyzed via path analysis to examine how students’ incoming affective variables, alongside their initial chemistry knowledge, influenced exam performance in GC1. Therefore, the data provided in the main text and SI is from the early-semester survey (pr_). For the following measures, Academic Mindset and Imposter Syndrome, a CFA is sufficient to confirm the internal structure of these measures because we used adapted surveys from literature. However, we included EFAs in the SI to provide additional reference for comparisons in other studies using the same survey.In this study, we first confirmed the internal structure of the two-factor model using confirmatory factor analysis (CFA). To incorporate the use of a total score from each individual factor, such as Entity, into the SEM path analyses, we conducted separate unidimensional CFAs for each factor. For the adapted survey questions, EFA, and CFA results of the two-factor structure (incremental and entity), as well as the unidimensional entity CFA, please see the SI.
Due to this specific social belonging instrument having been used at the same institution and course (Edwards et al., 2021, 2023), we confirmed the internal structure of the two-factor model using confirmatory factor analysis (CFA). To incorporate individual factors, such as Sense of Belonging, into the SEM path analyses, we conducted separate unidimensional CFAs for each factor. The unidimensional CFA for the ‘Belonging Uncertainty’ factor did not support adequate internal structure for these two items as a single factor, whereas Sense of Belonging' demonstrated a suitable fit for a unidimensional model. Therefore, the SEM path analyses used only ‘Sense of Belonging.’ Please see the SI for the survey and CFA results for the original two-factor structure (Sense of Belonging and Belonging Uncertainty), and the CFA results for the unidimensional Sense of Belonging factor.
Data analysis
All statistical analyses were completed with the statistical program RStudio (R version 4.4.2) with the use of the lavaan (0.6 series) (Rosseel, 2012), psych (version 2.4.6.26) (Revelle, 2017), dplyr (version 1.1.4) (Wickham et al., 2023), manymome (Cheung and Cheung, 2024), polycor (version 0.8–1) (Fox and Dusa, 2022), fungible (version 2.4.4) (Waller, 2016), semTools (version 0.5–7) (Jorgensen et al., 2016), MBESS (version 4.9.41) (Kelley, 2007), and FactorAssumptions (version 2.0.1) (Storopoli, 2022) packages.Exploratory factor analysis (EFA)
EFA is a multivariate technique for uncovering the underlying structure of relationships among a set of observed variables (items) by identifying their common underlying dimensions (latent variables) (Beaujean, 2014; Hair et al., 2019). This method does not rely on prior assumptions about the factor structure; factors (latent variables) are extracted from a dataset without pre-specifying the number of factors or the pattern of factor loadings between the observed and latent factors. To support the internal consistency of the data from the measures in our sample population, we conducted comprehensive data screening and factor analyses. Data representing social belonging, imposter syndrome, and academic mindset were collected at the beginning and end of the semester, while data for Childhood Science Currency was only collected at the beginning of the semester. For factor analysis, the full early-semester sample (N = 398) was randomly split: one half was used for exploratory factor analysis (EFA), and the other for confirmatory factor analysis (CFA).Confirmatory factor analysis (CFA)
CFA is a theory-driven statistical method used to test hypotheses about the relationships between observed variables (items) and their underlying latent constructs (factors). This method involves specifying a predefined measurement model, estimating the covariance matrix implied by this model, and comparing it to the observed covariance matrix to evaluate how well the model fits the data. Model fit is assessed using standard indices as suggested by Hu and Bentler (1999). This includes; the Comparative Fit Index (CFI), which should be close (i.e., ≥0.95), the Tucker-Lewis Index (TLI), which should be ≥0.95 (Watkins, 2021), the root mean square error of approximation (RMSEA), which should be less than 0.06 (Browne and Cudeck, 1993), and the Standardized Root Mean Square Residual (SRMR), which should be less than 0.08 (Hu and Bentler, 1999). Model fit thresholds were assessed using descriptors, proposed by Yuan and Marcoulides, which assign a range of adjectives to specific values of RMSEA (0.01 = “excellent,” 0.05 = “close,” 0.08 = “fair,” and 0.10 = “poor”) and CFI (0.99 = “excellent,” 0.95 = “close,” 0.92 = “fair,” and 0.90 = “poor”) (Yuan et al., 2016). Please note that these cutoff values are intended as flexible guidelines and should not be considered alone for fit decisions (Goretzko et al., 2024).Structural equation modeling (SEM) path analysis
Structural equation modeling (SEM) is a versatile and powerful statistical technique that enables researchers to investigate complex relationships among both observed and latent variables (Chen and Yung, 2024). SEM allows for the evaluation of direct and indirect effects, making it particularly useful for testing theoretical models that involve mediating or moderating relationships (Hancock and Mueller, 2013). Typically, SEM analysis is conducted in three stages (Byrne, 2013): (1) a theoretical model is hypothesized based on prior research or theory, (2) the overall fit of the data to the model is assessed using goodness-of-fit indices, and (3) specific parameters within the model are evaluated to determine their significance and contribution to the overall structure (Hancock and Mueller, 2013; Kline, 2016).Path Analysis, on the other hand, is a specialized form of multiple regression analysis that uses path diagrams to visually represent hypothesized causal relationships among measured variables. The primary goal of path analysis is to estimate the magnitude and significance of these hypothesized relationships. While path analysis provides valuable insights into how variables influence one another, it is important to note that it cannot establish causality or identify a “correct” model (Streiner, 2005). Instead, it assesses whether the observed data are consistent with the proposed model.
When combined, SEM and path analysis create a more robust analytical framework that extends beyond the limitations of each technique individually. SEM enhances path analysis by incorporating latent variables—unobservable constructs inferred from measured indicators—which allows researchers to account for measurement error and test more sophisticated models (Hoyle, 2012). Additionally, SEM supports parameter constraints, mediation effects, multi-group comparisons, and hierarchical modeling, enabling researchers to explore complex theoretical frameworks with greater precision and flexibility. Together, SEM path analysis provides a comprehensive approach for analyzing both simple and intricate relationships in multivariate data (Chavance et al., 2010). Path model fits assess how well the hypothesized model aligns with the observed data. Commonly used indices include the Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI), which measure relative fit compared to a baseline model; the root mean square error of approximation (RMSEA), which evaluates approximate fit based on residuals; and the standardized root mean square residual (SRMR), which assesses the average discrepancy between observed and predicted correlations (Kline, 2010).
Bootstrapping
Bootstrapping is a procedure in which multiple new samples are generated from the original dataset by randomly selecting observations, allowing for the same observation to be chosen more than once (resampling with replacement). This process is repeated many times (e.g., 5000 iterations) to produce robust confidence intervals and provide a more accurate assessment of effect significance within the SEM framework (Bollen and Stine, 1992; Sharma and Kim, 2013; Kattoum et al., 2024). This approach is especially useful in mediation analysis, where it is often used to calculate confidence intervals and standard errors for both direct and indirect effects. Given the complexity of the relationships under study, we employed structural equation modeling (SEM) with bootstrapping to analyze both direct and indirect effects via mediation analysis, which was used in this study in a post hoc manner (Tables 1–3).Mediation analysis
Mediation analysis is a statistical method used to examine the pathway by which an independent variable (X) influences a dependent variable (Y), by assessing whether this relationship occurs indirectly through an intervening variable known as a mediator (M). These pathways are found by tracing the ways one can get from X to Y, ensuring to never trace in a direction opposite to the direction an arrow points (Hayes, 2018). A pathway leading from X to Y, passing through M represents an indirect effect. However, a path leading from X to Y without passing through M represents a direct effect (Chen and Yung, 2024). These direct effects are estimated from the model coefficients, while indirect effects are tested by examining the product of the relevant regression coefficients. Instead of reporting only a z- or p-value, researchers recommend presenting statistical significance along with a confidence interval (CI), as this provides meaningful information about the magnitude and precision of the estimated effect (Hancock et al., 2018). Therefore, mediation analysis results should include four key outputs: the estimate (β), the standard error (S.E.) – which provides the amount of variability in the estimate, the lower level and upper-level confidence intervals (LLCI and ULCI, respectively). Last, authors should be cautious when phrasing the results from mediation analysis. Hancock et al. warn that language such as “causes’, “impacts on”, or “influences”, should be reserved for longitudinal datasets (Hancock et al., 2018). However, others in the field generally allow the use of causal language in experimental designs (i.e., X causes Z, which in turn, subsequently causes Y) (MacKinnon and Luecken, 2008; Hayes, 2009; Kenny and Judd, 2014). Mediation analysis results are presented in Tables 1–3.Results
EFAs
In this study, there were four constructs of interest: academic mindset (incremental and entity), social belonging (sense of belonging and belonging uncertainty), imposter syndrome, and childhood science currency (parental influence and science resources). The full dataset for early semester responses (N = 398) was randomly split in half: one subgroup (G1, n = 199) was used for the exploratory factor analysis (EFA), while the other (G2, n = 199) was reserved for CFA. For each construct, we created a training (G1) and testing (G2) dataset. Data normality and variability were assessed using descriptive statistics, including histograms, means, standard deviations, skewness, and kurtosis (please see SI for detailed results). Mardias test confirmed violations of multivariate normality across all measures. As a result, polychoric correlations were used for the EFAs, while ensuring that a substantial number of correlations exceeded 0.30, indicating sufficient intercorrelation among items to justify factor analysis (Watkins, 2018). To identify the appropriate number of factors to retain for rotation, two post-estimations were conducted: minimum average partial (MAP) (Velicer, 1976) and scree plots (please see SI for detailed results). EFAs were conducted using Minimum Residual (MinRes) extraction with an oblique, Promax rotation. Kaiser-Myer Olkin (KMO) and Barlett's tests indicated sampling adequacy (please see SI for test results for each respective measure). Due to non-normality, EFAs were performed using Polychoric correlations with MinRes extraction and Promax oblique rotation, referencing standard cutoffs for factor loading and model fit. Outliers and multicollinearity were checked using Mahalanobis D2 and VIFs, respectively. These analyses ensured that our model assumptions were met and that the latent structures were appropriate for the sample. Full procedural details are available in the SI.Childhood Science Currency is a measure based on Turnbull's work with 693 science students at the University of Auckland in New Zealand (Turnbull et al., 2020). Originally, the instrument comprised 20-items: Self-concept (3-items), Parents (4-items), Peers (4-items), and Resources (5-items) (Turnbull et al., 2020). However, due to the in-applicability of question five in the Resources construct: “Growing up, did you go to a lunchtime or after-school science club”, this question was omitted. To avoid over-extraction, five, four, three and two factor pre-models were examined sequentially. The five, four, and three pre-models were found to be inadequate using residuals and modification indices, as well as examining item loadings, uniqueness and communalities. The final factor structure for childhood science currency (SciRes) involved two factors: Science Resources (4-items) and Parental Influence (4-items)–which we renamed from Turnbull's prior work. An EFA was conducted on this factor structure using Minimum Residual extraction with an oblique rotation. Kaiser-Myer Olkin (KMO) and Barlett's tests indicated sampling adequacy (please see SI for test results).
Multidimensional CFAs
Multidimensional CFAs were conducted using MLM estimators for academic mindset (AM), and social belonging (SB), while MLR was used for science currency (SciCur). Due to the multivariate non-normality distribution of our data, a Satorra-Bentler scaled chi-squared test for model goodness of fit was selected, which is robust to non-normality (Satorra and Bentler, 1994). For AM, CFA results demonstrated “close” model fit (Yuan et al., 2016), as indicated by the fit indices (CFI = 0.994, TLI = 0.991, RMSEA = 0.062, and SRMR = 0.023) all meeting recommended thresholds (Kline, 2016). For social belonging, CFA results demonstrated “close” model fit, as indicated by fit indices (CFI = 0.996, TLI = 0.992, RMSEA = 0.033, and SRMR = 0.019). For SciCur, CFA results demonstrated “close” model fit, as indicated by fit indices (CFI = 0.980, TLI = 0.970, RMSEA = 0.053, and SRMR = 0.046). CFAs were used to evaluate the measurement model and to provide reliability evidence based on the data, as they require strong a priori hypotheses about the relationships between latent constructs (factors) and observed item responses, allowing explicit testing of the measurement structure (Flora, 2020). To determine the reliability of the data representing each multidimensional construct, McDonald's omega coefficients were calculated separately for each factor, as omega assumes unidimensionality of the items contributing to a given factor (Komperda et al., 2018). Since academic mindset, social belonging, and science currency models include distinct factors representing different constructs, estimating omega for each factor individually ensures that reliability reflects the proportion of variance attributable to that specific construct. Internal consistency reliability of the data representing each latent factor was assessed using McDonald's omega (ω). Commonly cited cutoff values for omega are 0.70 (Streiner, 2005; Lance et al., 2006; Watkins, 2017); however, others simply put that higher values are preferrable (Bandalos, 2018; Komperda et al., 2018; Lewis, 2022). To quantify the precision of these estimates, 95% bias-corrected and accelerated bootstrap confidence intervals (CIs) were computed using the MBESS package in R (Kelley, 2007). Confidence intervals were generated based on 1000 bootstrap resamples with bias-corrected and accelerated (BCa) adjustment to provide robust interval estimates around the omega coefficients. For academic mindset, Entity ω = 0.959 (95% C.I. [0.932, 0.972]), and Incremental ω = 0.948 (95% C.I. [0.926, 0.958]). For social belonging, sense of belonging ω = 0.841 (95% C.I. [0.790, 0.872]), and belonging uncertainty ω = 0.707 (95% C.I. [0.582, 0.768]). For science currency, parents ω = 0.794 (95% C.I. [0.731, 0.837]), and science resources ω = 0.812 (95% C.I. [0.751, 0.861]). For the comprehensive CFA results, please see SI.Unidimensional CFAs
To support the use of a total score for the specific factors for our path analysis, we conducted separate unidimensional CFAs for each construct under study. This approach allowed us to confirm that the items for each factor were adequately represented by a single latent variable, supporting the use of subsequent modeling steps. However, imposter syndrome (IS) is typically conceptualized as a single-factor construct with four indicators; thus, conducting a unidimensional CFA for IS aligns with established theory for this scale. As described in the ‘EFAs’ subsection in Methods, the testing dataset was used for CFAs (G2, n = 199).For IS, CFA results showed a non-significant Chi-square test result for the user model, χ2 = 2.648, p = 0.266. Global fit indices demonstrated a “close” model fit, as indicated by fit indices (CFI = 0.998, TLI = 0.995, RMSEA = 0.048, and SRMR = 0.017). Standardized factor loadings for all items were statistically significant and above recommended thresholds, with loadings ranging from 0.720 to 0.958. Model-based composite reliability (McDonald's omega) for the data reflecting imposter syndrome was 0.84, indicating high internal consistency reliability in participants’ responses.
For academic mindset entity (AME, four items), unidimensional CFA results showed a non-significant Chi-square test result for the user model, χ2 = 2.628, p = 0.269. Global fit indices demonstrated an “close” model fit, as indicated by fit indices (CFI = 0.998, TLI = 0.993, RMSEA = 0.073, and SRMR = 0.012). Standardized factor loadings for all items were statistically significant and above recommended thresholds, with loadings ranging from 0.848 to 0.968. Model-based composite reliability (McDonald's omega) for the data from the academic mindset entity measure was 0.96, indicating high internal consistency reliability in the observed responses.
For sense of belonging (SB, four items), unidimensional CFA results showed a non-significant Chi-square test result for the user model, χ2 = 1.538, p = 0.464, df = 2. Global fit indices demonstrated a “perfect” model fit, as indicated by fit indices (CFI = 1.00, TLI = 1.00, RMSEA = 0.00, and SRMR = 0.014). While “perfect” model fits are always something to be cautious about, we have additional support for this model by checking modification indices for this factor, as well as checking to ensure no residuals were above 0.05 (Kim et al., 2016) (please see SI for modification indices and residual results of the sense of belonging factor). Additionally, standardized factor loadings for all items were statistically significant and above recommended thresholds, with loadings ranging from 0.577 to 0.847. Model-based composite reliability (McDonald's omega) for the data from the sense of belonging measure was 0.84, indicating high internal consistency reliability in participants’ responses. Again, the reasoning for our additional tests for the sense of belonging construct was in response to a “perfect” model fit indicated by a CFI and TLI of 1 and an RMSEA of 0. While these are the traditional fit indices reported to ensure a proper fitting model, several researchers recommend a holistic approach for confirming internal structure (Maydeu-Olivares and Shi, 2017; Shi et al., 2019; Goretzko et al., 2024).
For science resources (SciRes, four items), unidimensional CFA results showed a non-significant Chi-square test result for the user model, χ2 = 2.883, p = 0.237. Global fit indices demonstrated a “close” model fit, as indicated by fit indices (CFI = 0.995, TLI = 0.986, RMSEA = 0.055, and SRMR = 0.022). Standardized factor loadings for all items were statistically significant and above recommended thresholds, with loadings ranging from 0.697 to 0.819. Model-based composite reliability (McDonald's omega) for the data from the science resources measure was 0.81, indicating high internal consistency reliability in participants’ responses.
SEM path analysis
SEM path analysis was used to test a set of path models, which describe a network of relationships simultaneously. All models were tested using the entire data set with outliers removed (N = 354). Each path model included the four early semester, affective measures of interest. All the proposed models were grounded in prior studies (please see the Background section). As mentioned in Background portion, we hypothesized the directionality for AME → IS and IS → SB. However, we tested IS → AME and SB → IS, and our hypothesized directionality provided a more suitable model fit based on lower AIC and BIC values (please see SI for these results). Results below provide the parameter estimate where β is the unstandardized path coefficient and the reported p-value, which has been subject to False Discovery Rate (FDR) correction (Benjamini and Hochberg, 1995; Narum, 2006). Herein, we discuss the two models with the best fits (please see the Methods section for fit guidelines). The model that included the path from pr_SB → Exam 1 (Fig. 1A) demonstrated “close” fit, with CFI = 0.988, TLI = 0.986, RMSEA = 0.030, and SRMR = 0.045. Similarly, the model that included the path from pr_SB → Final Exam (Fig. 1B) showed “close” fit indices: CFI = 0.990, TLI = 0.988, RMSEA = 0.027, and SRMR = 0.045. These cutoff thresholds are supported by Yuan et al. (Yuan et al., 2016). Please see the SI for all tested models, including full model fits and parameter reports. | ||
| Fig. 1 Path model illustrating the relationships between latent constructs and observed exam performance variables. Latent variables (e.g., sense of belonging) are represented as bubbles, and observed variables (e.g., exam scores) as boxes. Panels 1a and 1b display models predicting Exam 1 and Final Exam performance, respectively. Arrow colors specify relationship direction: green for positive and red for negative associations. Asterisks signify significance level, where * denotes p-value <0.05, ** denotes p-value <0.01, *** denotes p-value <0.001. | ||
Although these statistically significant direct paths discussed in the models above provide valuable insights, we also identified several post hoc mediated pathways that warrant further investigation (as seen on Table 1). Specifically, our analyses focused on the following mediated/indirect relationships (shown in Table 1): (a) pr_SciRes → pr_AME → Exam 1 or Final Exam, (b) pr_IS → PA → Exam1 or Final Exam, and (c) pr_AME → pr_IS → Exam 1 or Final Exam. In mediated model (a), only the indirect effect was statistically significant. This indicates that the impact of childhood science resources (pr_SciRes) on exam performance is fully mediated by academic mindset entity (pr_AME), with no significant direct effect from pr_SciRes to exam outcomes. For mediated model (b), all effects–direct, indirect, and total, were statistically significant. Suggesting that imposter syndrome (pr_IS) has an effect on exam performance both directly and through the pathway of performance on the preassessment (PA). Slightly different results were observed in mediated model (c) for Exam 1: only the direct and total effects were significant, while the indirect effects were not. This suggests that the independent variable (pr_AME) influences the dependent variable (Exam 1) mainly through a direct pathway rather than via the proposed mediator (pr_IS). In contrast, all pathways were significant in model (c) for the final exam, indicating a more complex mediated relationship whereby early-semester entity mindset, mediated by early-semester imposter syndrome, plays a role in predicting performance at the end of the semester.
| Pathwaya | Estimateb (β) [LLCIc, ULCId], S.E.e |
|---|---|
| a Bolded pathways represent significant effects. b Bootstrapped unstandardized estimate. c LLCI = bootstrapped lower-level 95% confidence interval. d ULCI = bootstrapped upper-level 95% confidence interval. e Bootstrapped standard error. | |
| Mediated model (a) | |
| Direct | |
| pr_SciRes → Exam1 | −0.021 [−0.046, 0.004], 0.049 |
| Indirect | |
| pr_SciRes → pr_AME → Exam1 | 0.007 [0.001, 0.017], 0.016 |
| Total effects | −0.014 [−0.040, 0.011], 0.050 |
| Direct | |
| pr_SciRes → Final Exam | −0.013 [−0.046, 0.017], 0.062 |
| Indirect | |
| pr_SciRes → pr_AME → Final Exam | 0.008 [0.001, 0.018], 0.017 |
| Total effects | −0.005 [−0.037, 0.026], 0.062 |
| Mediated model (b) | |
| Direct | |
| pr_IS → Exam1 | −0.023 [−0.045, −0.002], 0.042 |
| Indirect | |
| pr_IS → PA → Exam1 | −0.009 [−0.015, −0.003], 0.011 |
| Total effects | −0.032 [−0.055, −0.010], 0.044 |
| Direct | |
| pr_IS → Final Exam | −0.026 [−0.048, −0.004], 0.043 |
| Indirect | |
| pr_IS → PA → Final Exam | −0.010 [−0.017, −0.004], 0.013 |
| Total effects | −0.035 [−0.057, −0.013], 0.043 |
| Mediated model (c) | |
| Direct | |
| pr_AME → Exam1 | −0.044 [−0.072, −0.015], 0.056 |
| Indirect | |
| pr_AME → pr_IS→ Exam1 | −0.013 [−0.027, 0.001], 0.025 |
| Total effects | −0.057 [−0.080, −0.032], 0.047 |
| Direct | |
| pr_AME → Final Exam | −0.051 [−0.084, −0.014], 0.069 |
| Indirect | |
| pr_AME → pr_IS → Final Exam | −0.014 [−0.027, −0.002], 0.024 |
| Total effects | −0.065 [−0.095, −0.032], 0.062 |
Again, while the statistically significant direct effects discussed in the models above offer important insights, there are mediated/indirect pathways that merit additional exploration. Specifically, we examined the following paths (shown in Table 2): pr_AME → pr_IS → pr_SB. In the model above, only the indirect and total effects were significant, while the direct path from pr_AME to pr_SB, was insignificant (see Table 2). This result can be interpreted as evidence that the relationship between students’ academic mindset and their sense of belonging is fully mediated by imposter syndrome. In other words, students’ perceptions of themselves as imposters appear linked to the pathway connecting mindset and sense of belonging. Indeed, the absence of significant direct effects highlights just how nuanced these connections truly are.
| Pathwaya | Affective variable 1 | Mediator | Affective variable 2 | Estimateb (β) [LLCIc, ULCId] S.E.e |
|---|---|---|---|---|
| a Bolded pathways represent significant effects. b Bootstrapped unstandardized estimate. c LLCI = bootstrapped lower-level 95% confidence interval. d ULCI = bootstrapped upper-level 95% confidence interval. e Bootstrapped standard error. | ||||
| Direct | pr_AME | — | pr_SB | −0.060 [−0.193, 0.070] 0.063 |
| Indirect | pr_AME | pr_IS | pr_SB | −0.146[−0.231, −0.080] 0.039 |
| Total | — | — | — | −0.205 [−0.341, −0.087] 0.065 |
To further explore the relationship between Science Resources (SciRes) and Imposter Syndrome (IS), we tested a post hoc mediation model to examine whether SciRes influences imposter syndrome indirectly through Academic Mindset Entity (AME) as a mediator (as seen on Table 3). This analysis revealed statistically significant confidence intervals for both the indirect effect (SciRes → AME → Imposter Syndrome) and the total effect (SciRes → Imposter Syndrome) in the mediated pathway (Table 3). In other words, students with greater childhood science resources are more likely to develop a particular academic mindset (such as a fixed or growth mindset), and this mindset, in turn, affects their feelings of imposter syndrome. Additionally, to further explore the relationship between SciRes and sense of belonging (SB), we tested a post hoc mediation model to examine whether SciRes influences SB indirectly through imposter syndrome (IS) as a mediator. This analysis revealed non-significant results for all pathways (direct or indirect), and for total effects. These results can be seen in Table 3.
| Pathwaya | Affective variable 1 | Mediator | Affective variable 2 | Estimateb (β) [LLCIc, ULCId] S.E.e |
|---|---|---|---|---|
| a Bolded pathways represent significant effects. b Bootstrapped unstandardized estimate. c LLCI = bootstrapped lower-level 95% confidence interval. d ULCI = bootstrapped upper-level 95% confidence interval. e Bootstrapped standard error. | ||||
| Direct | pr_SciRes | — | pr_IS | 0.037 [−0.142, 0.219] 0.354 |
| Indirect | pr_SciRes | pr_AME | pr_IS | −0.088 [−0.180, −0.18] 0.159 |
| Total | — | — | — | −0.051 [−0.252, 0.144] 0.388 |
| Direct | pr_SciRes | — | pr_SB | 0.123 [−0.014, 0.266] 0.274 |
| Indirect | pr_SciRes | pr_IS | pr_SB | −0.011 [−0.062, 0.044] 0.104 |
| Total | — | — | — | 0.112 [−0.038, 0.277] 0.309 |
Discussion and implications
This study explored the interrelationships between four affective constructs–academic mindset entity, imposter syndrome, sense of belonging, and childhood science resources–and their associations with student performance. Student performance was measured using a chemistry pre-knowledge assessment (PA), first midterm exam scores (%), and final exam scores (%). Data were collected from a first-semester General Chemistry I (GC1) course at a Mountain West R1 university consisting of predominantly White and non-first-generation students.Finding 1: interplay of affective measures on performance: students’ first and final exam performances were directly predicted by their early semester academic mindset and experiences of imposter syndrome and indirectly affected by childhood science currency
Our findings indicate that, when considering all four affective measures together, early-semester academic mindset entity and early-semester imposter syndrome were the two that directly impacted exam performance. That is, students who entered the course with a more fixed mindset or higher levels of imposter syndrome—prior to experiencing GC1's environment, assessments, or course structure—tended to perform lower on both the first midterm and the final exam of the semester. Building on these findings, our mediation analyses provided further insight into how academic mindset and imposter syndrome influence student achievement. Specifically, we found that the effect of students’ childhood science resources on exam performance is fully mediated by their academic mindset (SciRes → pr_AME → Exam1/Final Exam).Our results also show that students’ academic mindset influences exam performance both directly and indirectly through imposter syndrome (pr_AME → pr_IS → Exam 1/Final Exam). This finding suggests an intricate interplay between students’ beliefs about their own intelligence and their self-perceptions of inadequacy or fear of being exposed as a fraud, despite evidence of competence. This finding aligns with Kumar and Jagacinski (2006), who surveyed 135 college students about the imposter phenomenon (Clance and Imes, 1978) and other measures relevant to achievement goal theory (Kumar and Jagacinski, 2006). Their results showed that among women, imposter fears were significantly associated with endorsement of an entity viewpoint (fixed mindset). Additionally, they reported that for both men and women, imposter fears were positively related to test anxiety and negatively related to confidence in one's own intelligence. Similarly, Canning et al. (2020) noted that students’ early-semester entity mindset (regardless of gender) predicts students’ early-semester imposter syndrome (Canning et al., 2020). We also found this relationship and further observed that higher early-semester imposter syndrome levels predicted lower exam performance in GC1.
Given these connections, based on Kumar and Jagacinski's findings, it is plausible that students with higher imposter syndrome experience more test anxiety than those with lower IS (Kumar and Jagacinski, 2006), which may contribute to lower exam performance. If mindset contributes to impostorism, then conscious efforts to promote a growth mindset culture in classrooms should be made. For example, instructors can support students by normalizing struggle as part of the learning process and emphasizing that abilities grow with effort and support, thus creating a more welcoming environment (Dweck, 2006; Yeager and Dweck, 2012; Canning et al., 2019). Beyond promoting a growth mindset, building student-instructor trust is another crucial component to creating a supportive course environment and may be even more influential in improving student engagement (Wang et al., 2021). Additionally, an instructor's mindset influences not only student performance (Kattoum et al., 2024), but also student motivation (Heyder and Pegels, 2025), and overall classroom environment (Lytle and Shin, 2023). Together, these findings highlight the importance of instructor practices that support both cognitive and affective aspects of student learning.
Our study also offers new insights into the role of course-level sense of belonging. While previous research has primarily examined social belonging, both sense of belonging and belonging uncertainty, as a singular affective factor linked to academic outcomes, findings have been mixed. As recent review by Fong et al. (2024), which analyzed 79 studies involving college students, found that most studies reported small but significant effects of social belonging on performance or persistence (Fong et al., 2024). However, larger effects were generally observed for marginalized students, though a few studies found no significant association. In STEM courses, several studies have shown a relationship between social belonging and performance or persistence, some even accounted for prior knowledge (Höhne and Zander, 2019a, 2019b; Fink et al., 2020; Edwards et al., 2021, 2022; Krause-Levy et al., 2021; Fong et al., 2024; Hewitt et al., 2024). However, our results suggest that when multiple affective variables are considered simultaneously, stronger affective predictors such as academic mindset entity and imposter syndrome, may overshadow sense of belonging's effect on performance. These results highlight the complex relationship among affective variables and their collective influence on student performance. Instructors may want to consider the influence of multiple affective factors when designing and implementing their courses. Our finding aligns with Fong et al.'s (2024) call for further research is needed to better clarify the concept of belonging and its relationship to student achievement and persistence.
We want to build on this recommendation by emphasizing the need to explore how belonging interacts with other affective variables such as mindset and imposter syndrome. For example, our mediated model provided greater explanatory power than direct models; imposter syndrome mediated the effect of early-semester fixed mindset on early-semester sense of belonging (pr_AME → pr_IS → pr_SB). In other words, although fixed mindset did not directly predict sense of belonging, the indirect and total effects indicate that imposter syndrome offers a more compelling explanatory pathway through which students’ mindsets relate to their sense of belonging in the course. These findings extend previous research. For example, Shin and Lytle (2024) found, in a study of 2808 undergraduate students across two United States universities, that higher levels of impostorism were associated with lower academic sense of belonging (Shin and Lytle, 2024) and greater reported difficulty seeking help compared to those with lower levels. Our findings support Shin and Lytle's results, while further illuminating the important role academic mindset plays in shaping students’ affective experiences in STEM learning environments.
Finding 2: students’ childhood science resources directly affects their early semester academic mindset entity and indirectly affects their feelings of impostorism
Based on prior research, we hypothesized that childhood science resources may influence students’ affective constructs (belonging, imposter syndrome, and mindset) as they enter GC1. While we were uncertain about the exact relationships, our results suggest that student's childhood science resources – the access to science resources (e.g., science kits, science tv programs, and books about science) – directly affected their fixed mindset, as they entered GC1.Interestingly, childhood science resources did not have a direct effect on students’ early-semester imposter syndrome or sense of belonging. To better understand this relationship, we conducted a mediation analysis to explore whether (a) academic mindset served as an indirect pathway between childhood science resources and imposter syndrome and (b) whether imposter syndrome served as an indirect pathway between childhood science resources and sense of belonging. While the results from mediated model (b) yielded insignificant pathways, the results from mediated model (a) revealed that although science resources was not directly associated with imposter syndrome, students’ access to science resources did shape their beliefs about learning and intelligence and these mindsets in turn linked the extent to which students felt like imposters in the classroom.
These findings have several instructor implications. First, it is important for instructors to recognize that students arrive in the classroom with varied science backgrounds and experiences that can shape how connected and confident they feel in class from the start. Instructors can help bridge these differences by not assuming all students have similar science exposure and incorporate multiple pathways to expose their students to science, such as highlighting a variety of science careers, connecting course content to popular media, or using STEM examples with real-world contexts. It may also be helpful for students to see scientists of all backgrounds (gender, age, ethnicity, geography, etc.) in their course content, which can reinforce the idea that science is broad and assessable to all (Schinske et al., 2016; Yonas et al., 2020; Metzger et al., 2023; Ovid et al., 2024; Rivera et al., 2024).
The connection between childhood science resources and mindset also suggests the importance of promoting a growth mindset culture in the classroom. Research has shown that encouraging effort, fostering confidence, and supporting future learning play a critical role in shaping the development of a growth mindset in children (Haimovitz and Dweck, 2017). Instructors can support this by modeling a growth-oriented perspective themselves and promoting a growth mindset culture, thus helping students’ build their self-efficacy and confidence–which has been shown to predict academic achievement, adjustment, and health in first-year university students (White et al., 2020). When instructors communicate a growth mindset, students are more likely to feel motivated and engaged. In contrast, students in courses led by instructors who hold a fixed mindset, such as believing intelligence is unchangeable or suggesting that all students are capable of succeeding in introductory STEM courses, may experience lower motivation (Muenks et al., 2020).
Last, these findings also have implications for addressing imposter syndrome. Since childhood science resources indirectly affects imposter syndrome through academic mindset, thoughtful strategies that support student confidence in their abilities may help reduce their feelings of impostorism and promote greater belonging. The classroom environment plays a key role in setting the tone for student–student interaction. For example, first-generation students who perceived higher classroom competition were linked to more feelings of impostorism (Canning et al., 2020). Therefore, instructors can structure their courses to emphasize and foster collaboration and peer support (Adams et al., 2005). Since our findings show that imposter syndrome affects student's sense of belonging, instructors can help ease students’ concerns by emphasizing that growth and improvement are possible through effective study habits and getting support from the teaching team (instructor, LA, and TAs) (Dunlosky et al., 2013). Additionally, the use of active learning strategies have been shown to improve student performance (Freeman et al., 2014), enhance motivation and self-efficacy (Cicuto and Torres, 2016), and foster a supportive community among classmates and with the instructional team (Allsop et al., 2020).
Limitations
While our study identifies statistically significant links between affective factors and exam performance in GC1, the data were obtained from a single research-intensive university setting. Additionally, the course studied was a large introductory course with multiple sections containing 220–350 students. Also, although the institution enrolls students with a range of academic backgrounds, it remains a predominantly White campus, and most students in GC1 are not first-generation college students. These characteristics–research-intensive institution, large classes, predominantly White campus, traditional students–may limit the extent to which our findings can be generalized to other institutional or demographic contexts.Another potential limitation is that the unidimensional CFA for the sense of belonging indicated a “perfect” model fit. While additional statistics supported the internal structure of this construct, it is important to note that sense of belonging only contributed significantly in paths where it acted as an endogenous variable (i.e., arrows coming into the variable, but none going out). Additionally, it did not emerge as a statistically significant predictor of either the first or final exams in GC1. Given these findings, we believe that the role of sense of belonging in the models does not raise concerns for the overall interpretation of the results.
Conclusion
In this study, we have added to the growing body of research examining how student's affective experiences shape their performance in introductory STEM courses. Specifically, we explored how four early-semester affective measures (academic mindset, sense of belonging, imposter syndrome, and childhood science resources) interacted with each other and affected student exam performance in a large GC1 course. Our findings show that academic mindset (entity) and imposter syndrome are particularly influential, both directly predicting student performance and serving as mediators to help explain the effect of childhood science experiences and sense of belonging.Importantly, our results show that early-semester fixed mindsets and high levels of impostorism, taken before students engage with the structure and assessments of a course, can negatively impact student exam performance. Moreover, the indirect pathways of childhood science resources to imposter syndrome via academic mindset and of mindset to sense of belonging via imposter syndrome highlight the need to treat these affective constructs as interconnected dimensions of student experience in the classroom and not as isolated effects.
Taken together, these findings have several implications for instructors. Efforts to cultivate a growth mindset culture, normalize academic struggle, foster trust between students and instructors, and promote a supportive and collaborative classroom may help buffer students from the negative effects of impostorism and fixed mindset beliefs. Furthermore, recognizing the varied science backgrounds and experiences students bring into the classroom and actively designing pedagogical strategies to support them may improve not only cognitive outcomes but also affective experiences such as confidence and belonging. Improving student outcomes in STEM requires not only addressing the cognitive domain, but also paying attention to the beliefs, emotions, and prior experiences students bring with them into the classroom.
Conflicts of interest
There are no conflicts to declare.Data availability
The data are not publicly available as approval for this study did not include permission for sharing data publicly.Supplementary information (SI): EFAs and CFAs for the measures (academic mindset entity, imposter syndrome, and childhood science currency); CFA for social belonging measure; SEM path model details; mediation analyses. See DOI: https://doi.org/10.1039/d5rp00295h.
Acknowledgements
We would like to thank the instructors of the general chemistry class for allowing us to conduct the study with the students enrolled in their classes. We would also like to thank the students who willingly participated in the study.References
- Adams J. D., Black G. C., Clemmons J. R. and Stephan P. E., (2005), Scientific teams and institutional collaborations: evidence from US universities, 1981–1999, Res. Policy, 34, 259–285.
- Aikens N. L. and Barbarin O., (2008), Socioeconomic differences in reading trajectories: the contribution of family, neighborhood, and school contexts, J. Educ. Psychol., 100, 235.
- Allsop J., Young S. J., Nelson E. J., Piatt J. and Knapp D., (2020), Examining the Benefits Associated with Implementing an Active Learning Classroom among Undergraduate Students, Int. J. Teach. Learn. High. Educ., 32, 418–426.
- Archer L., DeWitt J., Osborne J., Dillon J., Willis B. and Wong B., (2012), Science aspirations, capital, and family habitus: how families shape children's engagement and identification with science, Am. Educ. Res. J., 49, 881–908.
- Archer L., DeWitt J. and Willis B., (2014), Adolescent boys' science aspirations: masculinity, capital, and power, J. Res. Sci. Teach., 51, 1–30.
- Bandalos D. L., (2018), Measurement theory and applications for the social sciences, New York: Guilford Publications.
- Barrasso A. P. and Spilios K. E., (2021), A scoping review of literature assessing the impact of the learning assistant model, Int. J. STEM Educ., 8, 1–18.
- Beaujean A. A., (2014), Latent variable modeling using R: A step-by-step guide, New York: Routledge.
- Benjamini Y. and Hochberg Y., (1995), Controlling the false discovery rate: a practical and powerful approach to multiple testing, J. R. Stat. Soc., Ser. B, 57, 289–300.
- Blackwell L. S., Trzesniewski K. H. and Dweck C. S., (2007), Implicit theories of intelligence predict achievement across an adolescent transition: a longitudinal study and an intervention, Child Dev., 78, 246–263.
- Bollen K. A. and Stine R. A., (1992), Bootstrapping goodness-of-fit measures in structural equation models, Soc. Methods Res., 21, 205–229.
- Bourdieu P., (1977), Ecole des Hautes Etudes en Sciences Sociales.
- Bourdieu P., (1984), A social critique of the judgement of taste, Cambridge, MA.
- Bourdieu P., (1986), in Richardson J. (ed.), Handbook of Theory and Research for the Sociology of Education, Westport, Connecticut: Greenwood, ch. 1, pp. 241–258.
- Brown C. S., Usher E. L., Coleman C. and Han J., (2024), Perceptions of discrimination predict retention of college students of color: connections with school belonging and ethnic identity, J. Coll. Stud. Retent.: Res., Theory & Pract., 26, 240–258.
- Brown M. G., Wohn D. Y. and Ellison N., (2016), Without a map: college access and the online practices of youth from low-income communities, Comput. Educ., 92, 104–116.
- Browne M. W. and Cudeck R., (1993), Alternative ways of assessing model fit, Soc. Methods Res., 21, 230–258.
- Byrne B. M., (2013), Structural equation modeling with EQS: Basic concepts, applications, and programming, California: Routledge.
- Canning E. A., LaCosse J., Kroeper K. M. and Murphy M. C., (2020), Feeling like an imposter: the effect of perceived classroom competition on the daily psychological experiences of first-generation college students, Soc. Psychol. Person. Sci., 11, 647–657.
- Canning E. A., Muenks K., Green D. J. and Murphy M. C., (2019), STEM faculty who believe ability is fixed have larger racial achievement gaps and inspire less student motivation in their classes, Sci. Adv., 5, eaau4734.
- Case A. D. and Hunter C. D., (2012), Counterspaces: a unit of analysis for understanding the role of settings in marginalized individuals’ adaptive responses to oppression, Am. J. Commun. Psychol., 50, 257–270.
- Cavanagh A. J., Chen X., Bathgate M., Frederick J., Hanauer D. I. and Graham M. J., (2018), Trust, growth mindset, and student commitment to active learning in a college science course, CBE—Life Sci. Educ., 17, ar10.
- Chakraverty D., (2022), Faculty experiences of the impostor phenomenon in STEM fields, CBE—Life Sci. Educ., 21, ar84.
- Chakraverty D., Newcomer S. N., Puzio K. and Tai R. H., (2018), It runs in the family: the role of family and extended social networks in developing early science interest, Bull. Sci., Technol. Soc., 38, 27–38.
- Chavance M., Escolano S., Romon M., Basdevant A., de Lauzon-Guillain B. and Charles M. A., (2010), Latent variables and structural equation models for longitudinal relationships: an illustration in nutritional epidemiology, BMC Med. Res. Methodol., 10, 1–10.
- Chen S., Binning K. R., Manke K. J., Brady S. T., McGreevy E. M., Betancur L., Limeri L. B. and Kaufmann N., (2021), Am I a science person? A strong science identity bolsters minority students’ sense of belonging and performance in college, Person. Soc. Psychol. Bull., 47, 593–606.
- Chen G. A. and Buell J. Y., (2018), Of models and myths: Asian (Americans) in STEM and the neoliberal racial project, Race Ethn. Educ., 21, 607–625.
- Chen D. G. and Yung Y. F., (2024), Structural Equation Modeling using R/SAS: a Step-by-Step approach with Real Data Analysis, Boca Raton, Florida: CRC Press, pp. 1–408.
- Cheung S. F. and Cheung S.-H., (2024), manymome: an R package for computing the indirect effects, conditional effects, and conditional indirect effects, standardized or unstandardized, and their bootstrap confidence intervals, in many (though not all) models, Behav. Res. Methods, 56, 4862–4882.
- Choi N., (2005), Self-efficacy and self-concept as predictors of college students' academic performance, Psychol. Schools, 42, 197–205.
- Cicuto C. A. T. and Torres B. B., (2016), Implementing an active learning environment to influence students’ motivation in biochemistry, J. Chem. Educ., 93, 1020–1026.
- Clance P. R., (1985), Clance Impostor Phenomenon Scale (CIPS), in The impostor phenomenon: When success makes you feel like a fake, Toronto: Bantam Books, pp. 20–22.
- Clance P. R. and Imes S. A., (1978), The imposter phenomenon in high achieving women: dynamics and therapeutic intervention, Psychotherapy: Theory, Res. Pract., 15, 241.
- Collins K. H., Price E. F., Hanson L. and Neaves D., (2020), Consequences of stereotype threat and imposter syndrome: the personal journey from stem-practitioner to stem-educator for four women of color, Taboo: J. Culture Educ., 19, 10.
- Cromley J. G., Perez T. and Kaplan A., (2016), Undergraduate STEM achievement and retention: cognitive, motivational, and institutional factors and solutions, Policy Insights Behav. Brain Sci., 3, 4–11.
- Dabney K. P., Chakraverty D. and Tai R. H., (2013), The association of family influence and initial interest in science, Sci. Educ., 97, 395–409.
- Dao A., Pegg S., Okland S., Green H. and Kujawa A., (2024), Longitudinal associations between sense of belonging, imposter syndrome, and first-year college students’ mental health, J. Am. College Health, 1–10.
- De Castella K. and Byrne D., (2015), My intelligence may be more malleable than yours: the revised implicit theories of intelligence (self-theory) scale is a better predictor of achievement, motivation, and student disengagement, Eur. J. Psychol. Educ., 30, 245–267.
- Demirdöğen B. L. and Lewis S. E., (2023), Investigating How Chemistry Students’ Reported Challenges Inform the Relationship between Mindset and Academic Performance, J. Chem. Educ., 100, 3252–3260.
- DeWitt J., Archer L., Osborne J., Dillon J., Willis B. and Wong B., (2011), High aspirations but low progression: the science aspirations–careers paradox amongst minority ethnic students, Int. J. Sci. Math. Educ., 9, 243–271.
- Doerschuk P., Bahrim C., Daniel J., Kruger J., Mann J. and Martin C., (2016), Closing the gaps and filling the STEM pipeline: a multidisciplinary approach, J. Sci. Educ. Technol., 25, 682–695.
- Dunlosky J., Rawson K. A., Marsh E. J., Nathan M. J. and Willingham D. T., (2013), Improving students’ learning with effective learning techniques: promising directions from cognitive and educational psychology, Psychol. Sci. Public Interest, 14, 4–58.
- Dweck C. S., (1999), Self-theories: Their role in motivation, personality, and development, Philadelphia: Psychology Press.
- Dweck C. S., (2006), Mindset: The new psychology of success, New York: Random house.
- Dweck C. S., (2008), Can personality be changed? The role of beliefs in personality and change, Curr. Direct. Psychol. Sci., 17, 391–394.
- Dweck C. and Molden D. C., (2000), Self theories, Handbook of competence and motivation, pp. 122–140.
- Edwards J. D., Barthelemy R. S. and Frey R. F., (2021), Relationship between course-level social belonging (sense of belonging and belonging uncertainty) and academic performance in General Chemistry 1, J. Chem. Educ., 99, 71–82.
- Edwards J. D., Laguerre L., Barthelemy R. S., De Grandi C. and Frey R. F., (2022), Relating students’ social belonging and course performance across multiple assessment types in a calculus-based introductory physics 1 course, Phys. Rev. Phys. Educ. Res., 18, 020150.
- Edwards J. D., Torres H. L. and Frey R. F., (2023), The effect of social belonging on persistence to General Chemistry 2, J. Chem. Educ., 100, 4190–4199.
- Fink A., Cahill M. J., McDaniel M. A., Hoffman A. and Frey R. F., (2018), Improving general chemistry performance through a growth mindset intervention: selective effects on underrepresented minorities, Chem. Educ. Res. Pract., 19, 783–806.
- Fink A., Frey R. F. and Solomon E. D., (2020), Belonging in general chemistry predicts first-year undergraduates’ performance and attrition, Chem. Educ. Res. Pract., 21, 1042–1062.
- Fink A., Mattson P. S., Cahill M. J., Frey R. F. and McDaniel M. A., (2023), Growth mindset intervention improves performance of highly nontraditional introductory psychology students at a community college, Teach. Psychol., 50, 393–406.
- Flora D. B., (2020), Your coefficient alpha is probably wrong, but which coefficient omega is right? A tutorial on using R to obtain better reliability estimates, Adv. Methods Pract. Psychol. Sci., 3, 484–501.
- Fong C. J., Adelugba S. F., Garza M., Pinto G. L., Gonzales C., Zarei P. and Rozek C. S., (2024), A scoping review of the associations between sense of belonging and academic outcomes in postsecondary education, Educ. Psychol. Rev., 36, 138.
- Fong C. J., Fathi Z., Adelugba S. F., García A. J., Garza M. and Pinto G. L., (2025), Belongingness moderates student mindsets’ influence among first-generation, community college students of color, Soc. Psychol. Educ., 28, 1–17.
- Fox J. and Dusa A., (2022), Polychoric and Polyserial Correlations (version 0.8-1).
- Freeman S., Eddy S. L., McDonough M., Smith M. K., Okoroafor N., Jordt H. and Wenderoth M. P., (2014), Active learning increases student performance in science, engineering, and mathematics, Proc. Natl. Acad. Sci. U. S. A., 111, 8410–8415.
- Gaddis S. M., (2013), The influence of habitus in the relationship between cultural capital and academic achievement, Soc. Sci. Res., 42, 1–13.
- Gehringer T. A., Folberg A. M. and Ryan C. S., (2022), The relationships of belonging and task socialization to GPA and intentions to re-enroll as a function of race/ethnicity and first-generation college student status, J. Diversity High. Educ., 15, 744.
- Gopalan M. and Brady S. T., (2020), College students’ sense of belonging: a national perspective, Educ. Res., 49, 134–137.
- Goretzko D., Siemund K. and Sterner P., (2024), Evaluating model fit of measurement models in confirmatory factor analysis, Educ. Psychol. Measure., 84, 123–144.
- Gottlieb M., Chung A., Battaglioli N., Sebok-Syer S. S. and Kalantari A., (2020), Impostor syndrome among physicians and physicians in training: a scoping review, Med. Educ., 54, 116–124.
- Gray M. J., Yacobian J. N. and Shortlidge E. E., (2025), Elements of Social Capital and Counterspace Processes Contribute to Undergraduate STEM Student Development of a Sense of Belonging, Sci. Educ., 109, 911–927.
- Haimovitz K. and Dweck C. S., (2017), The origins of children's growth and fixed mindsets: new research and a new proposal, Child Dev., 88, 1849–1859.
- Hair J. F., Black W. C., Babin B. J. and Anderson R. E., (2019), Multivariate data analysis, 102–410.
- Han C.-W., Farruggia S. P. and Moss T. P., (2017), Effects of academic mindsets on college students' achievement and retention, J. College Student Dev., 58, 1119–1134.
- Hancock G. R. and Mueller R. O., (2013), Structural equation modeling: A second course, Charlotte, North Carolina: Iap.
- Hancock G. R., Stapleton L. M. and Mueller R. O., (2018), The Reviewer's Guide to Quantitative Methods in the Social Sciences, New York: Routledge.
- Hansen M. J., Palakal M. J. and White L. J., (2023), The Importance of STEM Sense of Belonging and Academic Hope in Enhancing Persistence for Low-Income, Underrepresented STEM Students, J. STEM Educ. Res., 1–26.
- Harvey J. C. and Katz C., (1985), If I'm so successful, why do I feel like a fake? The impostor phenomenon, New York: St. Martin's Press.
- Hayes A. F., (2009), Beyond Baron and Kenny: statistical mediation analysis in the new millennium, Commun. Monogr., 76, 408–420.
- Hayes A. F., (2018), Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach (Methodology in the Social Sciences), New York, NY: Guilford Press.
- Hazari Z., Brewe E., Goertzen R. M. and Hodapp T., (2017), The importance of high school physics teachers for female students’ physics identity and persistence, Phys. Teach., 55, 96–99.
- Hewitt A., Fall R., Puvanendran S., Mwavita M. and Mutambuki J. M., (2024), Comparison of the Effects of Metacognition Instruction, Voices of Role Models, and the Synergy of the Interventions on Students’ Achievement Scores and Perceived Belonging in a General Chemistry Course, J. Chem. Educ., 102, 183–193.
- Heyder A. and Pegels H., (2025), Detrimental effects of instructors’ fixed mindsets on students’ anticipated motivation and emotions in secondary and higher education, Soc. Psychol. Educ., 28, 32.
- Hilts A., Part R. and Bernacki M. L., (2018), The roles of social influences on student competence, relatedness, achievement, and retention in STEM, Sci. Educ., 102, 744–770.
- Hodges J., McIntosh J. and Gentry M., (2017), The effect of an out-of-school enrichment program on the academic achievement of high-potential students from low-income families, J. Adv. Acad., 28, 204–224.
- Höhne E. and Zander L., (2019a), Belonging uncertainty as predictor of dropout intentions among first-semester students of the computer sciences, Z. Erziehungswissenschaft, 22, 1099–1119.
- Höhne E. and Zander L., (2019b), Sources of male and female students’ belonging uncertainty in the computer sciences, Front. Psychol., 10, 1740.
- Hoyle R. H., (2012), Path analysis and structural equation modeling with latent variables.
- Hu L. T. and Bentler P. M., (1999), Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives, Struct. Equ. Model.: Multidiscip. J., 6, 1–55.
- Hutchins H. M., (2015), Outing the imposter: a study exploring imposter phenomenon among higher education faculty, New Horiz. Adult Educ. Human Res. Dev., 27, 3–12.
- Hwang N., Reyes M. and Eccles J. S., (2019), Who holds a fixed mindset and whom does it harm in mathematics? Youth Soc., 51, 247–267.
- Jackson Z. A., Harvey I. S. and Sherman L. D., (2023), The impact of discrimination beyond sense of belonging: predicting college students’ confidence in their ability to graduate, J. College Stud. Ret.: Res., Theory Pract., 24, 973–987.
- Jorgensen T. D., Pornprasertmanit S., Schoemann A. M., Rosseel Y., Miller P., Quick C., Garnier-Villarreal M., Selig J., Boulton A. and Preacher K., (2016), Package ‘semtools’, Website: https://cran.r-project.org/web/packages/semTools/semTools.pdf.
- Kattoum R. N. and Baillie M. T., (2025), A more positive mindset context is associated with better student outcomes in STEM, particularly for traditional-age students, Int. J. STEM Educ., 12, 15.
- Kattoum R. N., Montenegro-Montenegro E. and Baillie M. T., (2024), Students Who Perceive Instructors to Have a Fixed Mindset Report a Greater Sense of Academic Misfit That Leads to Lower Chemistry Grades, J. Chem. Educ., 101, 3613–3623.
- Kelley K., (2007), Methods for the behavioral, educational, and social sciences: an R package, Behav. Res. Methods, 39, 979–984.
- Kenneally A. M., Schadler A. and Cain J., (2023), Fixed mindset and imposter phenomenon: determining the correlation and prevalence among pharmacy students, Curr. Pharm. Teach. Learn., 15, 62–68.
- Kenny D. A. and Judd C. M., (2014), Power anomalies in testing mediation, Psychol. Sci., 25, 334–339.
- Kim H., Ku B., Kim J. Y., Park Y.-J. and Park Y.-B., (2016), Confirmatory and exploratory factor analysis for validating the phlegm pattern questionnaire for healthy subjects, J. Evidence-Based Complementary Altern. Med., 2016, 2696019.
- Kline R. B., (2010), Principles and practice of structural equation modeling, 3rd edn, New York: Guilford Press.
- Kline R. B., (2016), Principles and practice of structural equation modeling, 4th edn, New York: The Guilford Press.
- Komperda R., Pentecost T. C. and Barbera J., (2018), Moving beyond alpha: a primer on alternative sources of single-administration reliability evidence for quantitative chemistry education research, J. Chem. Educ., 95, 1477–1491.
- Korpershoek H., Kuyper H. and van der Werf G., (2015), The relation between students’math and reading ability and their mathematics, physics, and chemistry examination grades in secondary education, Int. J. Sci. Math. Educ., 13, 1013–1037.
- Krause-Levy S., Griswold W. G., Porter L. and Alvarado C., (2021), The relationship between sense of belonging and student outcomes in CS1 and beyond.
- Kroeper K. M., Muenks K., Canning E. A. and Murphy M. C., (2022), An exploratory study of the behaviors that communicate perceived instructor mindset beliefs in college STEM classrooms, Teach. Teach. Educ., 114, 103717.
- Kumar S. and Jagacinski C. M., (2006), Imposters have goals too: the imposter phenomenon and its relationship to achievement goal theory, Person. Indiv. Diff., 40, 147–157.
- LaCosse J., Murphy M. C., Garcia J. A. and Zirkel S., (2021), The role of STEM professors’ mindset beliefs on students’ anticipated psychological experiences and course interest, J. Educ. Psychol., 113, 949.
- Lance C. E., Butts M. M. and Michels L. C., (2006), The sources of four commonly reported cutoff criteria: What did they really say? Org. Res. Methods, 9, 202–220.
- Leary M. R., Patton K. M., Orlando A. E. and Wagoner Funk W., (2000), The impostor phenomenon: self-perceptions, reflected appraisals, and interpersonal strategies, J. Person., 68, 725–756.
- Lewis S. E., (2022), Considerations on validity for studies using quantitative data in chemistry education research and practice, Chem. Educ. Res. Pract., 23, 764–767.
- Lewis S. E. and Lewis J. E., (2007), Predicting at-risk students in general chemistry: comparing formal thought to a general achievement measure, Chem. Educ. Res. Pract., 8, 32–51.
- Limeri L. B., Carter N. T., Lyra F., Martin J., Mastronardo H., Patel J. and Dolan E. L., (2023), Undergraduate Lay Theories of Abilities: mindset, universality, and brilliance beliefs uniquely predict undergraduate educational outcomes, CBE—Life Sci. Educ., 22, ar40.
- Lytle A. and Shin J. E. L., (2023), Self and professors’ incremental beliefs as predictors of STEM engagement among undergraduate students, Int. J. Sci. Math. Educ., 21, 1013–1029.
- MacKinnon D. P. and Luecken L. J., (2008), How and for whom? Mediation and moderation in health psychology, Health Psychol., 27, S99.
- Maslow A. H., (1962), Some basic propositions of a growth and self-actualization psychology, Perceiving, behaving, becoming: A new focus for education, pp. 34–49.
- Maydeu-Olivares A. and Shi D., (2017), Effect sizes of model misfit in structural equation models, Methodology, 13, 23–30.
- McGee E. O., Thakore B. K. and LaBlance S. S., (2017), The burden of being “model”:rRacialized experiences of Asian STEM college students, J. Divers. High. Educ., 10, 253.
- McGregor L. N., Gee D. E. and Posey K. E., (2008), I feel like a fraud and it depresses me: the relation between the imposter phenomenon and depression, Soc. Behav. Person.: Int. J., 36, 43–48.
- Metzger K. J., Dingel M. and Brown E., (2023), “No matter what your story is, there is a place for you in science”: students’ Ability to Relate to Scientists Positively Shifts after Scientist Spotlight Assignments, Especially for First-Generation Students and Women, CBE—Life Sci. Educ., 22, ar12.
- Micari M., Van Winkle Z. and Pazos P., (2016), Among friends: the role of academic-preparedness diversity in individual performance within a small-group STEM learning environment, Int. J. Sci. Educ., 38, 1904–1922.
- Moreno C., Pham D. and Ye L., (2021), Chemistry self-efficacy in lower-division chemistry courses: changes after a semester of instruction and gaps still remain between student groups, Chem. Educ. Res. Pract., 22, 772–785.
- Muenks K., Canning E. A., LaCosse J., Green D. J., Zirkel S., Garcia J. A. and Murphy M. C., (2020), Does my professor think my ability can change? Students’ perceptions of their STEM professors’ mindset beliefs predict their psychological vulnerability, engagement, and performance in class, J. Exp. Psychol.: General, 149, 2119.
- Narum S. R., (2006), Beyond Bonferroni: less conservative analyses for conservation genetics, Conser. Gen., 7, 783–787.
- Naylor R., Baik C. and Arkoudis S., (2018), Identifying attrition risk based on the first year experience, High. Educ. Res. Dev., 37, 328–342.
- Noskeau R., Santos A. and Wang W., (2021), Connecting the dots between mindset and impostor phenomenon, via fear of failure and goal orientation, in working adults, Front. Psychol., 12, 588438.
- O’Leary E. S., Shapiro C., Toma S., Sayson H. W., Levis-Fitzgerald M., Johnson T. and Sork V. L., (2020), Creating inclusive classrooms by engaging STEM faculty in culturally responsive teaching workshops, Int. J. STEM Educ., 7, 1–15.
- Ovid D., Rose Acosta-Parra A., Alemayehu A., Francisco Gomez J., Tran D. and Tripp B., (2024), “All of us are capable, and all of us can be scientists.” The impact of Scientist Spotlight assignments with undergraduates in physiology courses, Adv. Physiol. Educ., 48, 770–783.
- Pulliam N. and Gonzalez C. E., (2018), Success or fraud? Exploring the impacts of the impostor phenomenon among high achieving racial/ethnic minority and first-generation college students, J. Access, Retent., Inclusion High. Educ., 1, 4.
- Qualtrics, (2020), Provo, UT, (https://www.qualtrics.com).
- Rainey K., Dancy M., Mickelson R., Stearns E. and Moller S., (2018), Race and gender differences in how sense of belonging influences decisions to major in STEM, Int. J. STEM Educ., 5, 1–14.
- Revelle W. R., (2017), psych: procedures for personality and psychological research.
- Rivera A. T., Chong S., Kim J. and Owens M. T., (2024), Low-stakes Scientist Spotlight assignment demonstrates high value and multiple effects for introductory biology students, CBE—Life Sci. Educ., 23, ar47.
- Robins R. W. and Pals J. L., (2002), Implicit self-theories in the academic domain: implications for goal orientation, attributions, affect, and self-esteem change, Self Identity, 1, 313–336.
- Rosseel Y., (2012), lavaan: an R Package for Structural Equation Modeling, J. Stat. Softw., 48, 1–36.
- Sandefur G. D., Meier A. M. and Campbell M. E., (2006), Family resources, social capital, and college attendance, Soc. Sci. Res., 35, 525–553.
- Satorra A. and Bentler P. M., (1994), Latent Variables Analysis: Applications for Developmental Research, 171, p. 399.
- Schinske J. N., Perkins H., Snyder A. and Wyer M., (2016), Scientist spotlight homework assignments shift students’ stereotypes of scientists and enhance science identity in a diverse introductory science class, CBE—Life Sci. Educ., 15, ar47.
- Seery M. K., (2009), The role of prior knowledge and student aptitude in undergraduate performance in chemistry: a correlation-prediction study, Chem. Educ. Res. Pract., 10, 227–232.
- Sha L., Schunn C., Bathgate M. and Ben-Eliyahu A., (2016), Families support their children's success in science learning by influencing interest and self-efficacy, J. Res. Sci. Teach., 53, 450–472.
- Sharma P. N. and Kim K. H., (2013), A comparison of PLS and ML bootstrapping techniques in SEM: A Monte Carlo study.
- Shi D., Lee T. and Maydeu-Olivares A., (2019), Understanding the model size effect on SEM fit indices, Educ. Psychol. Measure., 79, 310–334.
- Shin J. E. and Lytle A., (2024), The roles of impostorism and academic help-seeking in undergraduate students’ sense of belonging and college completion intention, Soc. Psychol. Educ., 27, 2589–2602.
- Stone S., Saucer C., Bailey M., Garba R., Hurst A., Jackson S. M., Krueger N. and Cokley K., (2018), Learning while Black: a culturally informed model of the impostor phenomenon for Black graduate students, J. Black Psychol., 44, 491–531.
- Storopoli J., (2022), Factor Assumptions: set of assumptions for factor and principal component analysis.
- Streiner D. L., (2005), Finding our way: an introduction to path analysis, Canadian J. Psychiatry, 50, 115–122.
- Tai R. H., Sadler P. M. and Loehr J. F., (2005), Factors influencing success in introductory college chemistry, J. Res. Sci. Teach., 42, 987–1012.
- Tao K. W. and Gloria A. M., (2019), Should I stay or should I go? The role of impostorism in STEM persistence, Psychol. Women Quart., 43, 151–164.
- Thompson E. D., Bowling B. V. and Markle R. E., (2018), Predicting student success in a major's introductory biology course via logistic regression analysis of scientific reasoning ability and mathematics scores, Res. Sci. Educ., 48, 151–163.
- Tinto V., (1975), Dropout from higher education: a theoretical synthesis of recent research, Rev. Educ. Res., 45, 89–125.
- Tinto V., (1993), Leaving college: Rethinking the causes and cures of student attrition, Chicago, Illinois: University of Chicago press.
- Trefts S., (2019), The imposter phenomenon in female, first-generation STEM majors, California Lutheran University.
- Turnbull S. M., Meissel K., Locke K. and O'Neale D. R., (2020), The impact of science capital on self-concept in science: a study of university students in New Zealand Front. Educ., 5, 27–43.
- Van Sickle L. L., Bustamante V. P. and Frey R. F., (2025), Effect of weekly Study Technique Training Sessions on students' performance and study behaviors in a General Chemistry I course, Chem. Educ. Res. Pract., in preparation.
- Van Sickle L. L. and Frey R. F., (2025), Student's study behaviors as a predictor of performance in general chemistry I, Chem. Educ. Res. Pract., 26, 88–111.
- Vaughn A. R., Taasoobshirazi G. and Johnson M. L., (2020), Impostor phenomenon and motivation: women in higher education, Stud. High. Educ., 45, 780–795.
- Velicer W. F., (1976), Determining the number of components from the matrix of partial correlations, Psychometrika, 41, 321–327.
- Vincent C. and Ball S. J., (2007), Making up'the middle-class child: families, activities and class dispositions, Sociology, 41, 1061–1077.
- Waller N. G., (2016), Fungible correlation matrices: a method for generating nonsingular, singular, and improper correlation matrices for Monte Carlo research, Multivariate Behav. Res., 51, 554–568.
- Walton G. M. and Cohen G. L., (2007), A question of belonging: race, social fit, and achievement, J. Person. Soc. Psychol., 92, 82.
- Wang C., Cavanagh A. J., Bauer M., Reeves P. M., Gill J. C., Chen X., Hanauer D. I. and Graham M. J., (2021), A framework of college student buy-in to evidence-based teaching practices in STEM: the roles of trust and growth mindset, CBE—Life Sci. Educ., 20, ar54.
- Watkins M. W., (2017), The reliability of multidimensional neuropsychological measures: from alpha to omega, Clin. Neuropsychol., 31, 1113–1126.
- Watkins M. W., (2018), Exploratory factor analysis: a guide to best practice, J. Black Psychol., 44, 219–246.
- Watkins M. W., (2021), A step-by-step guide to exploratory factor analysis with SPSS, New York, New York: Routledge.
- Wenner J. A., (2024), What do families have to say about it? family supports for children's interest, engagement, and aspirations in science, Int. J. Sci. Educ., 1–26.
- White K. N., Vincent-Layton K. and Villarreal B., (2020), Equitable and inclusive practices designed to reduce equity gaps in undergraduate chemistry courses, J. Chem. Educ., 98, 330–339.
- Wickham H., François R., Henry L. and Müller K. D. V., (2023), dplyr: A Grammar of Data Manipulation.
- Willis W. K., Williamson V. M., Chuu E. and Dabney A. R., (2022), The relationship between a student's success in first-semester general chemistry and their mathematics fluency, profile, and performance on common questions, J. Sci. Educ. Technol., 31, 1–15.
- Xu X., Villafane S. M. and Lewis J. E., (2013), College students’ attitudes toward chemistry, conceptual knowledge and achievement: structural equation model analysis, Chem. Educ. Res. Pract., 14, 188–200.
- Yeager D. S. and Dweck C. S., (2012), Mindsets that promote resilience: when students believe that personal characteristics can be developed, Educ. Psychol., 47, 302–314.
- Yeager D. S., Hanselman P., Walton G. M., Murray J. S., Crosnoe R., Muller C., Tipton E., Schneider B., Hulleman C. S. and Hinojosa C. P., (2019), A national experiment reveals where a growth mindset improves achievement, Nature, 573, 364–369.
- Yeager D. S. and Walton G. M., (2011), Social-psychological interventions in education: They’re not magic, Rev. Educ. Res., 81, 267–301.
- Yonas A., Sleeth M. and Cotner S., (2020), In a “Scientist Spotlight” intervention, diverse student identities matter, J. Microbiol. Biol. Educ., 21, 1–12 DOI:10.1128/jmbe.v21i1.2013.
- Yuan K.-H., Chan W., Marcoulides G. A. and Bentler P. M., (2016), Assessing structural equation models by equivalence testing with adjusted fit indexes, Struct. Equ. Model.: Multidiscip. J., 23, 319–330.
| This journal is © The Royal Society of Chemistry 2026 |