There has been a push to create more research opportunities for science, technology, engineering, and mathematics (STEM) undergraduates (American Association for the Advancement of Science, 2011), with the intention of producing more STEM graduates (President’s Council of Advisors on Science and Technology, 2012)—and for good reason (Lopatto, 2010). Students who participate in research not only demonstrate gains in their understanding of the scientific method but also report greater feelings of belonging to the scientific community, a heightened sense of science self-efficacy, and a stronger science identity (Chemer et al., 2011; Estrada et al., 2011; Hunter et al., 2007; Lopatto, 2004, 2007; Mraz-Craig et al., 2018; Russell et al., 2007; Seymour et al., 2004). They also express a greater interest in pursuing a career in science and graduate with STEM degrees at higher rates (Lopatto, 2010; Russell et al., 2007).

Institutions use a wide range of models to involve undergraduates in research (Corwin Auchincloss et al., 2014). Virtually all models use elements of active learning (Freeman et al., 2014) and try to promote authentic science (Dolan et al., 2008). The most common course-based research model is the traditional laboratory course, in which the instructor defines the topic of inquiry and the methods. The most common research model outside of courses is the research internship, in which students work directly with a more senior researcher, typically guided by a top-down approach to hypothesis generation. Although research participation is beneficial, access to these experiences is not equitable (Bangera & Brownell, 2014). Unfortunately, barriers to participating in undergraduate research disproportionately impact students from historically excluded backgrounds, who may have the most to gain from such experiences (Estrada et al., 2011; Hernandez et al., 2013; Villarejo et al., 2008). One method for equalizing access to undergraduate research is course-based undergraduate research experiences (CUREs; Bangera & Brownell, 2014; Corwin, Graham, & Dolan, 2015; Corwin Auchincloss et al., 2014).

CUREs

Key features of CUREs include use of scientific practices, collaboration, discovery, broadly relevant or important work, and iteration (Corwin Auchincloss et al., 2014; Corwin, Graham, & Dolan, 2015). Published studies on CUREs point to student gains comparable to those achieved from research internships (Corwin, Graham, & Dolan, 2015) in four general outcome categories: cognitive, psychosocial, affective, and behavioral (Dolan, 2016). From an institutional perspective, CUREs represent an economical and pragmatic way to expose large numbers of students to research experiences, particularly introductory students (Bangera & Brownell, 2014; Harrison et al., 2011).

CURE limitations

In response to institutional data suggesting we needed to better support first-year STEM majors, we were excited by the potential benefits a CURE could offer (Corwin, Graham, & Dolan, 2015). However, we could not find an appropriate model to fit our needs, so we developed our own CURE, the KEYSTONE (KEYs to Success Through year ONE) Science Bootcamp. In doing so, we aimed to address the following limitations of existing CURE models:

1. A majority of CUREs are not guided by theoretical frameworks (Dolan, 2016; Krim et al., 2019). A guiding theoretical framework can shape CURE design, implementation, and assessment (Dolan, 2016), allowing for greater unification of CUREs with broader curricular efforts.
2. A majority of CUREs are discipline specific rather than interdisciplinary (Dolan, 2016). Discipline-specific CUREs are often geared toward students in a single major, typically within the context of a majors’ course (Drew & Triplett, 2008; Shaffer et al., 2014) but do not build a broader STEM community.
3. A majority of CUREs do not collect valid and reliable assessment measures (Dolan, 2016; Linn et al., 2015; Shortlidge & Brownell, 2016), making it difficult to compare CURE models objectively.
4. A majority of CUREs utilize assessment designs that are vulnerable to threats to internal validity, such as volunteer bias, small sample sizes, and lack of comparison groups (Brownell et al., 2013; Flannelly et al., 2018).

Science Bootcamp: An interdisciplinary, authentic CURE for first-year STEM majors

The KEYSTONE Science Bootcamp is an interdisciplinary, authentic CURE for first-year STEM students and is one piece of a larger STEM retention initiative on our campus—the KEYSTONE Program (Guenther et al., 2019; Majka et al., 2020, 2021). In developing our CURE, we wished to overcome the limitations discussed in the previous section, starting with a guiding theoretical framework.

Community of practice (CoP) framework

Learning communities are a high-impact practice that promote positive STEM outcomes (Carrino & Gerace, 2016; Dagley et al., 2015). They aim to add an informal, social component to the STEM student experience (Fenichel & Schweingruber, 2010) and facilitate the development of psychosocial learning factors through social interaction (Carrino & Gerace, 2016). A number of scholars have suggested that CURE design would benefit from a community of practice (CoP) framework (Dolan, 2016; Krim et al., 2019), yet, to our knowledge, no CUREs have drawn on this theorizing.

A CoP is a group of people who regularly interact through engagement in a shared passion (Smith et al., 2017). CoPs emerge when three elements are united: a domain (the sense of common identity and area of knowledge), community (the group of people who come together to learn), and practice (the specific knowledge and skills of a discipline; Snyder & Wenger, 2010; Wenger, 2005). CoPs function as both a social learning system and social network (Barab et al., 2002). Together, the informal community and learning activities engender the collective knowledge of a group, with new members privy to that knowledge (Crawford, 2014; Snyder & Wenger, 2010; Wenger, 1998, 2005, 2010).

Designing the CURE

We deliberately designed our Science Bootcamp CURE to establish an interdisciplinary CoP for first-year STEM students. The Science Bootcamp CoP elements include the domain (STEM major identity), community (STEM learning community), and practice (scientific method practiced in the CURE). The primary goal of the CURE is to expose students to the practice of doing science without the added stress of discipline-specific content. This shift of focus from content to process matches a pedagogical shift seen in many STEM disciplines (American Association for the Advancement of Science, 2011). To this end, students work in small, interdisciplinary groups to carry out novel research projects, providing opportunities to engage in use of scientific practices, collaboration, discovery, broadly relevant or important work, and iteration—the key design features of a CURE (Corwin, Runyon et al., 2015; Corwin Auchincloss et al., 2014). They are given total autonomy in project topic selection (with only financial and pragmatic limits) and are encouraged to enjoy the ups and downs of doing “real research.”

In establishing a STEM CoP, we attend to students’ inherent needs for relatedness, competence, and autonomy (for a similar approach, see Findley-Van Nostrand & Pollenz, 2017; Kuchynka et al., 2019). According to self-determination theory, when these needs are supported, individuals are more intrinsically motivated and better able to self-regulate, and they experience greater well-being (Ryan & Deci, 2000). Relatedly, engaging in situated social learning with peers in a CoP should enhance students’ self-perceptions of competence, which can impact self-regulation and goal pursuit (Bandura, 1997).

In short, by establishing a STEM CoP, we predicted that STEM majors who participated in the Science Bootcamp would experience psychosocial gains (e.g., belonging, self-efficacy, science identity), which would be associated with their intention to stay in their STEM major (a proxy for STEM persistence; e.g., Chemer et al., 2011; Estrada et al., 2011; Hausmann et al., 2009; see Figure 1). Prior to testing our hypothesis, we present data that (i) validate our CURE design (Corwin, Runyon et al., 2015) and (ii) demonstrate that we successfully established a CoP in the Science Bootcamp. We also explored the relation between the Science Bootcamp CURE design features and psychosocial gains as an additional contribution to the CURE literature. We sought to answer the following research questions:

1. Does the Science Bootcamp include key CURE design features?
2. Does the Science Bootcamp establish a CoP for undergraduate STEM majors?
3. Does participation in the Science Bootcamp promote psychosocial gains?
4. Do design features of the Science Bootcamp CURE relate to psychosocial outcomes?

Methods

Participants and design

Data were collected from Science Bootcamp cohorts pre-CURE and post-CURE in 2018 (n = 19), 2019 (n = 21), and 2020 (n = 22). Presenting assessment data for a single sample with only pre- and posttest data obviously has limitations—namely, an inability to identify a volunteer or selection bias, as well as threats to internal validity (Brownell et al., 2013; Flannelly et al., 2018). Therefore, in 2020, we also collected data pre-CURE and post-CURE for a comparison group recruited from the population of traditional first-year STEM major students (n = 26). Thus, for 2020, we utilized a mixed factorial pretest/posttest design. See Table 1 for student demographics.

Science Bootcamp course logistics

The Science Bootcamp CURE learning outcomes center on scientific practices (e.g., experimental design), but the intervention objectives are to use key CURE design features (e.g., opportunities to engage in collaboration, discovery/relevance, iteration) to establish a STEM CoP and promote psychosocial gains (see the online appendix for the syllabus and detailed course notes). The course is a 4–credit hour elective course that counts toward general graduation requirements but not any specific major. It is open to first-year STEM-oriented students and has no prerequisites. To date, the course has been offered during the 4-week January term (48 contact hours, class size = 20–25). The course is team taught (most often by a biologist and psychologist), with instructors acting as both traditional instructors and research mentors. We believe the course functions best when taught by instructors who represent a range of disciplines and are outwardly enthusiastic, have a history of conducting research with undergraduates, and genuinely believe in and promote the idea that most students have high scientific aptitude (Rattan et al., 2018).

In Week 1, the instructors set the stage for students to subsequently develop their study idea (e.g., they model how to read empirical articles, discuss identifying a “gap” in the literature, etc.). In Week 2, students define their hypothesis and develop their study design (oral proposal, written proposal); in Week 3, they conduct their study and begin writing individual manuscripts; and in Week 4, they interpret and share findings (manuscript and poster symposium). The faculty members support students at each step by serving as research mentors. The course includes significant instruction in scientific writing, with all students completing a research paper and receiving constructive feedback on drafts from instructors and peers.

A unique, hallmark feature of our CURE is that students have wide latitude over their project topic. This represents a bottom-up model that contrasts with a traditional research experience for undergraduates, giving students greater project ownership. Student projects are typically restricted only by the available time frame and lab resources. Over the years, students have completed a range of projects (see Figure 2) that exemplify the interdisciplinary approach and, at times, have led to surprising student groupings and topics. We see this as a strength of the course and a testament to students being free to explore topics and recognize early that modern scientific research is not restricted to individual disciplines.

Measures

The university’s Institutional Review Board approved all data collection.

Lab Course Assessment Survey

The Lab Course Assessment Survey (LCAS) assesses lab course design on three dimensions: (i) collaboration, (ii) discovery and relevance, and (iii) iteration (Corwin, Runyon et al., 2015). These three subscales correspond to key features of CUREs (Corwin Auchincloss et al., 2014), and the scale has successfully differentiated biology CUREs and traditional biology lab courses (Corwin, Runyon et al., 2015), with CURE students scoring higher on the discovery/relevance and iteration subscales than non-CURE students. For individual items, response format descriptions, and scale reliability, see Table 2.

STEM CoP Index

To assess the establishment of a STEM CoP, we created a course assessment with seven items corresponding to key components of a STEM CoP (i.e., domain, community, practice). All items were measured on a scale ranging from 1 (not at all) to 7 (very much). The index used the average of the seven items (see validation procedures in the next paragraph; Cronbach’s α ≥ 0.79 for every sample reported). For responses for individual items, see Table 3.

To justify the creation of an index, we used a three-part validation procedure. To start, using data from STEM students (N = 101) with data at the pre-CURE time point (2018–20 CURE cohorts, 2020 non-CURE comparison group), we conducted two analyses. First, we conducted a principle component analysis (PCA) to determine if there were one or multiple underlying components represented in the assessment items. A one-component solution was suggested by the scree plot, eigenvalues (Kaiser, 1960), and proportion of variance explained by one component (56.43%). Second, we conducted a Cronbach’s alpha for the final scale to evaluate internal consistency (reliability), which yielded a respectable alpha of 0.87 (Lance et al., 2006). Third, using combined data from CURE (n = 22) and non-CURE (n = 26) students in 2020, we conducted a set of bivariate correlations to assess convergent/criterion validity. The STEM CoP Index was positively associated with a range of psychosocial outcomes (e.g., belonging to school, r = 0.38; science self-efficacy, r = 0.61; science identity, r = 0.77), as well as students’ intention to stay in their STEM major (r = 0.66)—a proxy for STEM persistence, all p < 0.05. Taken together, the PCA, Cronbach’s alpha, and bivariate correlations provide justification for assessing a STEM CoP using the STEM CoP Index.

Psychosocial gains

Psychosocial outcomes were assessed using valid/reliable scales that measured belonging to school (Layous et al., 2017), science self-efficacy, and science identity (items adapted from Findley-Van Nostrand & Pollenz, 2017). All items were measured on a 7-point scale, with higher scores representing higher values of the psychosocial outcome. Scales were constructed using means (all Cronbach’s alphas ≥ 0.76 for pre- and post-CURE in 2020 cohort).

Intention to stay in STEM

As a proxy for STEM persistence, two items assessed students’ intention to stay in their STEM major. Students were asked to rate the two items on a scale ranging from 1 (strongly disagree) to 7 (strongly agree), with responses averaged together prior to analyses; a higher score reflected a greater intention to stay in STEM (items adapted from Findley-Van Nostrand & Pollenz, 2017; 2020 pre-CURE α = 0.78, 2020 post-CURE α = 0.86).

Qualitative feedback

A series of open-ended questions asked students what they had personally gained from the course, as well the strengths and weaknesses of the course.

Results

Does the Science Bootcamp interdisciplinary CURE include key CURE design features?

In 2020, we formally assessed the CURE using the LCAS. As seen in Table 2, a majority of students agreed or strongly agreed that the course encouraged them to work closely with their classmates (collaboration subscale) and expected them to generate new knowledge relevant to a broader community (discovery/relevance subscale). We think it is notable that 100% of students agreed that the course encouraged them to “formulate [their] own research question or hypothesis to guide an investigation,” which speaks to students’ autonomy in selecting project topics—a feature relatively unique to the Science Bootcamp CURE.

Although a majority of students agreed or strongly agreed with most items on the iteration subscale, their responses were more variable in this area. This was not surprising given the time frame of the course—students carry out a novel project from start to finish, so collecting additional data, for example, is beyond the scope of the course. Although our scale is modified from the original LCAS, results look comparable to the CURE student data in the LCAS validation paper (Corwin, Runyon et al., 2015). Taken together, these data verify that the Science Bootcamp interdisciplinary CURE includes key CURE course design features.

Does the Science Bootcamp establish a CoP for undergraduate STEM majors?

STEM Community of Practice Index data from pre- to post-CURE (2018–20 cohorts)

To evaluate whether the Science Bootcamp establishes a CoP, we conducted several analyses. First, combining data across the 2018 (n = 19), 2019 (n = 21), and 2020 (n = 22) Science Bootcamp cohorts to maximize sample size (Brownell et al., 2013), we conducted a series of paired-samples t-tests to compare all students’ (N = 62) pre-CURE and post-CURE scores for each of the individual STEM CoP Index items. As can be seen in Table 3, Science Bootcamp students reported gains on each of the individual items (all p < 0.05). For example, they reported feeling they understood the scientific method better from pre-CURE (M = 5.37, SD = 1.09) to post-CURE (M = 6.45, SD = 0.71), t(61) = 8.54, p < 0.001, d = 1.08. Second, we conducted a series of paired-samples t-tests to compare each Science Bootcamp cohort’s pre-CURE to post-CURE scores for the entire STEM CoP Index. As expected, students in all three Science Bootcamp cohorts reported higher scores on the STEM CoP Index from pre-CURE to post-CURE (all p < 0.001; see Figure 4, left three clusters of bars).

Pre-post STEM CoP Index data from 2020 cohort relative to comparison group

We conducted additional analyses in 2020 to address potential threats to internal validity (Brownell et al., 2013). Using data from the 2020 Science Bootcamp cohort and a comparison group, we conducted a 2 (CURE student status: CURE student vs. non-CURE student) x 2 (Time: pre-CURE vs. post-CURE) mixed-model analysis of variance (ANOVA) on STEM CoP Index scores. Both the main effect of CURE student status and the main effect of time were significant, F(1, 46) = 7.47, p = 0.009, η2p = 0.14 and F(1, 46) = 12.37, p = 0.001, η2p = 0.21, respectively. Most critically, the two-way interaction was also significant, F(1, 46) = 17.56, p < 0.001, η2p = 0.28, suggesting that CURE and non-CURE students showed different patterns of data from pre- to posttest (see Figure 4, right two clusters of bars).

To probe the two-way interaction, we conducted pairwise follow-up tests. As already reported, CURE students showed an increase in STEM CoP Index scores from pre-CURE to post-CURE, p < 0.001, but non-CURE students showed no gain across time points, p = 0.62. At the pre-CURE time point, CURE and non-CURE students did not differ in their STEM CoP Index scores, p = 0.25—which rules out a volunteer/selection bias. Moreover, at the post-CURE time point, CURE students reported greater STEM CoP Index scores than non-CURE students. Taken together, these data confirm that the Science Bootcamp successfully establishes a STEM CoP.

Does participation in the Science Bootcamp lead to psychosocial gains?

To test our hypothesis that the Science Bootcamp would promote psychosocial gains, we again used data from the 2020 CURE cohort and comparison group and conducted three separate 2 (CURE student status: CURE student vs. non-CURE student) x 2 (Time: pre-CURE vs. post-CURE) mixed-model ANOVAs with the three psychosocial outcomes: belonging to school, science self-efficacy, and science identity. The results for belonging to school and science self-efficacy parallel the results described on the STEM CoP Index—for both belonging to school and science self-efficacy, the two-way interactions were significant, F(1, 46) = 5.23, p = 0.027, η2p = 0.10 and F(1, 46) = 12.79, p = 0.001, η2p = 0.22), with pairwise comparisons revealing gains for the CURE students but not the comparison group (see Figure 5).

For science identity, only the main effect of time was marginally significant, F(1, 46) = 2.84, p = 0.10, η2p = 0.06), indicating that all STEM students showed a slight increase in their science identity over the January term. Taken together, these results provide initial support for our hypothesis that the Science Bootcamp promotes psychosocial gains. Students who participated in the bootcamp experienced an increase in their sense of belonging to school and science self-efficacy.

What outcome variables predict intention to stay in STEM?

We also conducted a series of bivariate correlations to investigate which outcome measures predicted students’ intention to stay in their STEM major. As can be seen in Table 4, all the primary outcome measures—the STEM CoP Index, belonging to school, science self-efficacy, and science identity—positively predicted students’ intention to stay in their STEM majors.

Do design features of the Science Bootcamp CURE relate to psychosocial gains?

Finally, we conducted bivariate correlations to explore the relation between the LCAS subscales and gains on the three psychosocial outcomes (belonging to school, science self-efficacy, science identity). To calculate gains, we subtracted pre-CURE scores from post-CURE scores, such that higher scores represent an increase on the construct from pre-CURE to post-CURE. The collaboration subscale was marginally related to belonging to school gains (r = 0.41, p = 0.067); the discovery/relevance subscale was positively associated with science self-efficacy gains (r = 0.49, p = 0.026); and the iteration subscale was positively associated with science self-efficacy gains (r = 0.44, p = 0.049) and science identity gains (r = 0.55, p = 0.009).

Discussion

We have described the details of the KEYSTONE Program Science Bootcamp, an interdisciplinary CURE course for first-year STEM students that successfully establishes a STEM CoP while implementing key features of CURE course design (Corwin Auchincloss et al., 2014; Wenger, 2005). As hypothesized, students who participated in the Science Bootcamp experienced psychosocial gains (increased belonging to school, science self-efficacy), which were positively associated with their intention to remain in their STEM major. Students’ comments about the course echo these findings (see Figure 3).

By drawing from a CoP framework (Smith et al., 2017), the KEYSTONE Science Bootcamp CURE is among the first CUREs to implement a course design driven by a clear theoretical framework (Dolan, 2016; Krim et al., 2019). It is also among the first to provide an early interdisciplinary research experience decoupled from traditional course content, an approach consistent with a broader emphasis on “process” across many STEM disciplines (American Association for the Advancement of Science, 2011; Rajapaksha & Hirsch, 2017). Modelling modern scientific research, the Science Bootcamp provides a unique opportunity for students to work in teams with students from other STEM majors and to have total autonomy over topic selection, a feature rare in most CURE models (Drew & Triplett, 2008; Shaffer et al., 2014) but that can positively impact retention (Hanauer & Dolan, 2014). Through the Science Bootcamp experience, a STEM CoP emerges, as STEM students engage in academic and social learning with supportive STEM peers and faculty. Although many institutions have adopted learning communities as a means to improving STEM student outcomes (Brown et al., 2018; Johnson et al., 2020; Solanski et al., 2019), the Science Bootcamp CURE provides a direct and inclusive pathway for new STEM students to enter such a community.

As we continue to refine and assess the Science Bootcamp CURE, we recognize the need to use a larger sample size, develop a more psychometrically rigorous STEM CoP measure, and conduct more formal mediation analyses. Although data from our quasi-experimental design argue against a volunteer bias, a randomized experiment would better address concerns related to internal validity (Brownell et al., 2013; Flannelly et al., 2018), which would assist in the important task of identifying which features of CURE design impact specific psychosocial outcomes. Our data showed that each CURE feature (collaboration, discovery/relevance, iteration) was associated with at least one psychosocial outcome, but greater efforts are needed in this area (Corwin, Graham, & Dolan, 2015; Corwin et al., 2018; Corwin, Runyon et al., 2015). Finally, guided by self-determination theory (Ryan & Deci, 2000), we look forward to assessing a broader range of psychosocial outcomes, as well as probing more deeply into why effects emerge for particular psychosocial outcomes.

Our novel KEYSTONE Science Bootcamp CURE is affordable, scalable, and modifiable; allows for flexible staffing; and can complement a variety of STEM curricular efforts. We look forward to assessing the long-term outcomes of Science Bootcamp, including whether it better prepares students for later research experiences. Investigating the impact of the bootcamp on retention and graduation rates, particularly among students from underrepresented minority backgrounds, is also important. Few CUREs are intentional about encouraging participation or tracking outcomes for underserved or underrepresented populations (Corwin et al., 2018; Krim et al., 2019). With such data in hand, we can offer more tailored supports within our community of practice. This would ensure inclusiveness and equity, allowing the KEYSTONE Science Bootcamp to be an intentional mechanism for further expanding access to STEM education and diversifying the pool of promising STEM graduates entering STEM fields.

Acknowledgments

This program was supported by National Science Foundation Grant, DUE-1160956, part of the STEP Program.

Elizabeth A. Majka (liz.majka@elmhurst.edu) is an associate professor and Thomas P. Sawyer is a professor, both in the Department of Psychology; Kyle F. Bennett is an associate professor and Merrilee F. Guenther is a professor, both in the Department of Biology; and Jon L. Johnson is a professor emeritus in the Department of Mathematics, all at Elmhurst University in Elmhurst, Illinois.