A Single-Item Measure for Assessing STEM Identity

McDonald, Melissa M.; Zeigler-Hill, Virgil; Vrabel, Jennifer K.; Escobar, Martha

doi:10.3389/feduc.2019.00078

ORIGINAL RESEARCH article

Front. Educ., 26 July 2019

Sec. Assessment, Testing and Applied Measurement

Volume 4 - 2019 | https://doi.org/10.3389/feduc.2019.00078

A Single-Item Measure for Assessing STEM Identity

$\nMelissa M. McDonald$ Melissa M. McDonald^*

Virgil Zeigler-Hill

Jennifer K. Vrabel

Martha Escobar

Department of Psychology, Oakland University, Rochester, MI, United States

Science identity based frameworks have proven fruitful in predicting persistence in careers in Science, Technology, Engineering, and Mathematics (STEM). However, much of the research in this area is qualitative or relies on measures of science identity that have not been validated. Here, we propose and provide initial evidence for the validity and reliability of the single-item STEM Professional Identity Overlap measure (STEM-PIO-1) that aims to assess the broader construct of STEM identity via students' perceived overlap between the image they have of themselves and the image they have of STEM professionals. Across three studies, we provide evidence of convergent, discriminant, and criterion validity—the measure displays moderate positive associations with adapted measures of STEM identity, explains unique variance relative to related but distinct measures of STEM identity, and is positively associated with STEM attitudes, STEM self-efficacy, mastery goal orientation, and agentic behavior toward one's graduation goals. The measure differentiates between STEM and non-STEM majors, and is associated with self-reported persistence in one's STEM major. The single-item measure displays moderate test-retest reliability and an expanded four-item version yields good internal consistency. Although continued validation is needed, the simplicity of the STEM-PIO-1 may prove valuable in its ability to promote consistency in measurement of the STEM identity construct across time, age groups, and disciplines.

Introduction

In recent years, there has been a growing interest in professional identity as a factor that may improve our understanding of the difficulties involved in motivating and retaining students in science, technology, engineering, and mathematics (STEM) fields. Despite widespread knowledge that STEM jobs tend to have higher salaries than other jobs (U.S. Department of Education: Institute of Education Sciences, 2014), and that there is a shortage of qualified individuals to fill STEM positions (Morrison et al., 2011), student interest in pursuing college majors in STEM remains relatively low (Chen and Soldner, 2013). Furthermore, approximately half of the students who initially pursue STEM majors eventually switch to a non-STEM major (Strenta et al., 1994; Seymour and Hewitt, 1997; Daempfle, 2003). The most common explanation for this trend is that students who decide to switch from a STEM major to a non-STEM major do so because they perceive that they lack the requisite competence or skill to be successful. However, even high-ability students fail to persist in STEM (e.g., Webb et al., 2002). This has prompted researchers to examine psychological factors that guide the decisions of students about the costs and benefits of allocating effort toward persistence in their major.

One factor that appears to contribute to persistence in STEM fields is the extent to which the individual possesses a science identity (Carlone and Johnson, 2007; Herrera et al., 2012; Hazari et al., 2013; Perez et al., 2014). Science identity can be broadly defined as the aspect of the self that relates to science (Carlone and Johnson, 2007). Overall, science identity among students is quite low, with ~30% of all students indicating that they view themselves as a “science person,” a number that is even lower for women and individuals from groups underrepresented in STEM fields (Hazari et al., 2013). Consequently, interventions that foster STEM identity may be particularly effective in promoting greater persistence in STEM majors. However, before this ideal can be realized, researchers must come to agree on what science identity is, and how it can be measured. The current research seeks to contribute to this conversation by developing and validating a simple and versatile measure of STEM identity.

Research on science identity has predominantly relied on qualitative research, which was necessary to define and provide a rich understanding of the construct (Sfard and Prusak, 2005; Carlone and Johnson, 2007; Herrera et al., 2012). However, qualitative approaches are of limited utility to assess the effectiveness of large-scale interventions aimed at increasing science identity. Efforts to quantify the construct have resulted in a number of author-generated and adapted measures of science identity, most of which have not been validated (e.g., Young et al., 2013; Starr, 2018). As research continues to grow in this area, it is important to develop measures that are validated and which can promote consistency of measurement across research samples. Here, we propose and validate a broad and versatile single-item measure of identification with STEM, the STEM Professional Identity Overlap measure (i.e., the STEM-PIO-1). Owing to its simplicity, it is our hope that the measure may provide researchers with an efficient means of assessing STEM identity in large-scale intervention studies, longitudinal work, and with a diverse set of populations that range in age, STEM field, and demographic variables.

What Is Science Identity?

Identity has been defined as a “core sense of self” (Jones and McEwen, 2000). However, this “core” is not unitary, as it is comprised of personal identities as well as many intersecting social identities that serve to foster a sense of belonging to one or more groups (Tajfel and Turner, 1986; Jones and McEwen, 2000). In the development of Social Identity Theory, Tajfel (1981) described social identity as “that part of an individual's self-concept which derives from his [or her] knowledge of his [or her] membership in a social group (or groups) together with the value and emotional significance attached to that membership” (p. 255). Developments in self-categorization theory (Turner and Reynolds, 2012) highlighted the important role of context in determining which elements of our identity are activated, such that some contexts elicit more of our personal identity, and others more of our social identity. Moreover, which social identity is activated also varies with context, such that a woman in a STEM classroom that is filled with men is likely to have her social identity as a woman activated, whereas when she is with her family, her identity as a daughter or sister may be more likely to be activated. Owing to this, measures of social identity typically reflect specific social identities, rather than a generalized social identity. This specificity has the advantage that context relevant identities (e.g., student, scientist) predict context-specific outcomes (e.g., academic performance and persistence) better than broader social identities (Eccles and Barber, 1999; Bonous-Hammarth, 2000; Osborne and Walker, 2006).

The notion of social identity proposed by Tajfel and Turner (1986) explicitly focuses on the importance and value of social identity to a person's self-concept. This approach to social identity can be broadly construed as identity centrality (Leach et al., 2008). However, other research has suggested that social identity is multi-faceted, and includes a facet of typicality (referred to as self-stereotyping by Leach et al., 2008; Starr, 2018). This perspective recognizes that two elements are essential for developing a social identity: (1) seeing oneself as a member of the group, and (2) feeling that the members of the group accept you as a member (Kim et al., 2018). Although the first may be captured by the importance of a social group to one's self-concept, the latter may be better captured by the self-perception that you are a prototypical member of the group—that you possess the skills, knowledge, beliefs, practices, and principles of a member of that particular group (Ibarra, 1999; Nadelson et al., 2015).

Measures of science and STEM identity vary in whether they emphasize identity centrality or typicality. For example, Young et al. (2013) developed a multi-item scale assessing the importance of science to one's self-concept, including the item: “Being a STEM student is an important reflection of who I currently am.” Other measures have instead emphasized typicality. For example, Hazari et al. (2013) asked participants whether they see themselves as a biology/chemistry/physics person, which appears to reflect typicality more so than importance (i.e., participants must consider whether they match the characteristics of a “biology person”). Nadelson et al. (2015) argued that, for professional identities, it is important to ask students proxy questions for determining the extent to which their attributes, skills, knowledge, etc. match those of a highly identified professional, because otherwise students tend to assume a higher level of identification development than is warranted based on their actual characteristics. Starr (2018) suggested that a focus on typicality may be especially important when people possess stereotypes about a group that do not match their own self-concept – such as an African American female student who possesses the stereotype that scientists are White men.

Consistent with Starr's (2018) rationale for assessing STEM typicality, Carlone and Johnson (2007) made the strong case that students from under-represented groups in science fields may be especially likely to struggle with feeling that they are recognized by others as belonging to a field of science professionals. In their qualitative analysis of science identity, Carlone and Johnson proposed that, in order for individuals to have a strong science identity, they must feel (1) confident that they have (or are capable of acquiring) the knowledge necessary to understand science concepts (i.e., competence), (2) confident in their ability to showcase their science skills in public settings (i.e., performance), and (3) that others—particularly those within the scientific community—recognize their competence and performance (i.e., recognition). Similarly, Herrera et al. (2012) suggested that competence, performance, and recognition interact with an individual's multiple social and cultural identities such that an individual's STEM identity is influenced by the perceived recognition from others, as well as how that individual integrates STEM into his or her sociocultural context.

The process of learning in science is more than just accumulating knowledge because it also involves socializing students into the culture of science. It is, therefore, important to investigate how students come to connect with the culture of science or, alternatively, become alienated from that culture. Carlone and Johnson (2007) argued that, even when students feel competent in their STEM knowledge and possess the requisite STEM skills, they may lack recognition by their peers, family, and—perhaps most importantly—their STEM teachers and professors. Lewis (2003) argued that, “science career attainment is a social process, and the desire of an aspirant is only one factor in this process. An aspiring scientist relies on the judgment and invitation of practicing scientists throughout every phase of the educational and career process” (p. 371). Consequently, those individuals who are perceived as not fitting the traditional “mold” of a STEM professional may highly value science as part of their self-concept, but receive less recognition of their skills and competencies, and ultimately perceive that they do not belong in the field owing to a lack of perceived overlap with the prototypical image of a scientist.

Why Is Science Identity Important?

Individuals who highly identify with science are more likely to make decisions that validate that identity (Eccles, 2009; Herrera et al., 2012; Perez et al., 2014). These students may be better able to maintain their motivation to persist in STEM fields because their efforts are directed toward the pursuit of a science career that aligns closely with their science identity (Oyserman, 2015). Consequently, decisions that divert an individual away from a science career trajectory carry the additional cost of severing part of a highly valued identity. Along these lines, Perez et al. (2014) found that individuals who arrived at the decision to major in a STEM field through a deep exploration of the field, believing that it matched their interests, skills, and prior experiences, were more likely to have positive beliefs about their competence in STEM and the value of a STEM major, than did students who chose their major based on other factors (e.g., parental pressure). Additionally, they perceived the costs associated with a STEM major (e.g., time spent studying) as more worthwhile. Importantly, these variables subsequently predicted students' intention to persist in their STEM major.

Similarly, Kuchynka et al. (2017) found that women's STEM identity provided a buffer against the experience of sexism, such that women with low STEM identity (but not high STEM identity) reported reduced STEM major intentions, STEM self-efficacy, and STEM GPA. Arguing from the perspective of Motivational Intensity Theory (Brehm and Self, 1989; Richter et al., 2016), the authors suggested that STEM identity fosters investment in one's STEM work, which begets motivation to perform well in that domain, and fosters the exertion of effort even when faced with hardship. In contrast, those with lower STEM identity are likely to invest less, be less motivated to perform well, and be more likely to desist in the face of hardship. In other words, STEM identity may play a crucial role in rallying effort during critical moments of hardship when students might otherwise consider changing their major.

Measuring Science Identity

Given the recent interest in promoting the development of science identity, it is important to identify a method for measuring the construct. Most of the research concerning science identity has focused on qualitative measures that allow for a rich characterization of the factors that promote identification with science (e.g., Sfard and Prusak, 2005; Carlone and Johnson, 2007; Herrera et al., 2012). Although extremely informative, qualitative approaches to the measurement of science identity do not provide a way to quantify this identity and are extremely difficult to implement in larger samples. Quantitative approaches to the measurement of science identity have relied on a variety of different assessment strategies. As noted above, Hazari et al. (2013) adapted an item from previous research asking individuals to report the extent to which they see themselves as a science person. In their research, they created separate items for each of three scientific disciplines (i.e., “Do you see yourself as a biology/chemistry/physics person?”). Although face valid, this item may be problematic in that it assumes that one either is, or is not, a science person. That is, it suggests that science identity is an all-or-nothing construct, and that being a scientist is not something that can be learned. This is problematic because identity is dynamic and changes with life experiences (Sfard and Prusak, 2005; Ceglie, 2011).

Measures assessing the importance of science to one's self-concept (e.g., Young et al., 2013) generally focus on the impact that being a science student has on learning material at the present time, not how the student compares himself or herself to other science professionals. This may be an important point given that what limits the development of STEM identity for some students is the perception that their self-concept does not match the attributes of a typical STEM professional. Sfard and Prusak (2005) suggested that identity can be measured as stories about an individual told by an author (the individual or someone else) to a recipient (the individual or someone else). This perspective suggests that identity is shaped by the context, and that it can represent the current state of the individual (the actual identity) or a potential state of the individual (the designated identity). Thus, educational choices can be viewed as a way to close the gap between a student's actual and designated identities; measuring this gap may be useful for predicting persistence, as it connotes the likelihood that effort toward one's goal will be well-spent or wasted. Thus, although existing measures of science and STEM identity have face validity, there is little consistency with regard to how science identity is measured, making it difficult to compare findings across studies.

The STEM Professional Identity Overlap Measure

In creating our measure of STEM identity, we prioritized an emphasis on typicality, that is, an assessment of the extent to which students feel that who they are as a person is compatible with being a STEM professional. This was done with an eye toward future research that will seek to understand why under-represented groups in STEM do not persist in their majors. The qualitative literature strongly suggests that persistence may be related to recognition from valued mentors, and lack of such recognition may lead to a perceived lack of “fit” with the profession (Carlone and Johnson, 2007). Moreover, a measure that can assess the gap between one's current self and his or her idealized future professional identity was seen as beneficial because gaps that are too large may be highly predictive of the perception of diminishing returns on investments in one's major.

The development of our measure was based on Aron's et al. (1992) measure of interpersonal closeness, in which participants are presented with a set of circles with varying levels of overlap, and are asked to identify one of seven pairs of overlapping circles that best represents the degree to which they feel they have included another person in their own self-concept. This interpersonal measure has been adapted by previous researchers to study other facets of social identity. For example, Tropp and Wright (2001) adapted the measure to assess women's gender identity. Their measure was framed more in terms of identity centrality, asking participants to select the pair of circles “that best represents your own level of identification with your group” (or in other studies, their sense of connection with the group, and their relationship to the group). More recently, Ahlqvist et al. (2013) used Aron et al.'s overlapping circles to measure the stability of female college students' perceived compatibility between their gender and STEM identities.

Our adaptation of this method asks participants to: “Select the picture that best describes the current overlap of the image you have of yourself and your image of what a STEM professional is” and uses a similar 7-point pictorial response scale to that of Aron et al. (1992) (see Figure 1). This framing permits us to assess perceived overlap with a desired future identity. There is a degree of subjectivity built into the concept of a STEM professional. This is intentional because students' concept of what a STEM professional is may vary considerably based on their experiences. For example, although the stereotype of a scientist may be a White male, some students may have had the opportunity to be mentored by professors or research supervisors who defy that stereotype, thereby changing their image of what a STEM professional is. This shift is important, particularly for students from underrepresented groups (URGs), as it may increase their perceived overlap with STEM professionals, and foster their persistence to achieve that desired end-goal. Moreover, it is likely to be the case that students' self-perceived overlap with their image of a STEM professional will be more predictive of their STEM outcomes, as compared to their self-rated overlap with a set of objective standards for defining a STEM professional (which they may or may not recognize as being important in their pursuit of a STEM career). The reference to a STEM professional also means that the measure is not specific to a particular science, which allows respondents to imagine any STEM professional that they are aspiring to become, across a variety of disciplines.

FIGURE 1

Figure 1. STEM Professional Identity Overlap Measure. Participants were instructed to select the picture that best describes the current overlap of the image they have of themselves and their image of what a STEM professional is.

Although there are clearly drawbacks to using a single-item measure, there are also many advantages to assessing STEM identity in this way. For example, it allows for repeated assessments that are not taxing for the individual being assessed, thereby providing a “snapshot” of the individual's identity at a particular moment, and enabling researchers and teachers to track how students' STEM identity evolves over time, and what experiences foster a stronger identification with STEM. Importantly, the measure does not pre-suppose that one's identity as a STEM professional is fixed, thereby making it possible to observe change in one's identity. The measure is also simple enough to be used with a wide array of age groups (i.e., from children to adults).

The Current Research

The aim of the current research was to provide initial evidence for the validity and reliability of the STEM-PIO-1 as a measure of STEM identity. Across three studies, we examined the convergent validity of the measure with an existing measure of STEM identity, the criterion validity by examining its associations with concurrent outcomes for which we expect STEM identity to be able to predict as well as its ability to predict criterion group membership as a STEM major, and its discriminant validity by examining whether it predicts unique variance in relevant outcomes over and above that predicted by centrality measures of STEM identity. We also examined the test-retest reliability of the measure over a 6-month time span, and the inter-item reliability of an expanded four-item version of the measure.

We selected constructs to test the criterion validity of the STEM-PIO-1 based on the perspective of Motivational Intensity Theory (Brehm and Self, 1989; Richter et al., 2016), which states that effort is directed toward a goal, even as it increases in difficulty, as long as the importance of success is sufficiently high. For students who highly identify with STEM, becoming a STEM professional is an idealized state, thereby justifying a high degree of effort. As a consequence, we expected that students more strongly identified with STEM (relative to students who weakly identify with STEM) would have more positive attitudes toward STEM, feel more efficacious in their pursuit of STEM learning, be more likely to engage in agentic behaviors to facilitate their end-goal, and be more likely to persist toward their STEM career goals and therefore report a reduced likelihood of changing their major. We also explored whether STEM identity would be associated with mastery goal orientation (vs. performance goal orientation). Mastery goals reflect an intrinsic desire to develop skills and expertise and are associated with an incremental implicit theory about intelligence, that is, that intelligence is malleable and capable of growth. In contrast, performance goals are simple shows of competence, that are associated with an entity implicit theory about intelligence, that is, that intelligence is a fixed trait that cannot be increased (Dweck, 1996; Elliot and Church, 1997). Given the difficulty of coursework in STEM, and the strong likelihood that students will often struggle to perform at a high level of competence, it is expected that students with performance goals will find other majors more appealing as a means of displaying competence, whereas those with a mastery orientation will be more willing to invest effort in learning, and continue to identify as a STEM professional, even in the face of hardship and reduced performance.

In Study 2, we additionally tested the reliability of the STEM-PIO-1. This was done in two ways. First, we examined the test-retest reliability of the measure across a 6-month time span for a small subset of students. Second, we expanded the measure with three additional items that assessed students' perceived overlap with those aspects of science identity outlined as important by Carlone and Johnson (i.e., competence, performance, and recognition). This multi-item measure allowed us to assess the internal reliability of the measure, and determine whether our measure was capturing the breadth of these important identity elements. Finally, in Study 3, we examined the association between our measure and measures of STEM identity that focus on identity centrality, in order to determine whether something unique was captured by our measure of typicality.

Study 1

To provide an initial validation of the single-item STEM Professional Identity Overlap (STEM-PIO-1) measure, we administered it to a sample of college and university students, across a variety of STEM and non-STEM majors. We included an existing measure of STEM identity to provide evidence of convergent validity, as well as measures of relevant criteria that should be associated with STEM identity, including attitudes toward STEM, self-efficacy in STEM, agency toward STEM goals, goal orientation, and desire to persist in the STEM major. We also examined the extent to which scores on the STEM-PIO-1 varied as a function of demographic characteristics.

Method

Participants and Procedure

Participants were college and university students in various institutions across the state of Alabama (see Table 1). Participants were invited to participate in the study via an email solicitation. Those electing to participate provided written informed consent. Participation required completing an online survey assessing a variety of demographic, academic, and psychological variables. Participation was incentivized via inclusion of participants into a raffle to win one of several gift cards. This study and those that follow were carried out with the institutional review board approval of participating institutions and were conducted in accordance with the 1964 Helsinki declaration and its later amendments.

TABLE 1

Table 1. Descriptive statistics for scholar and control samples in studies 1–3.

The sample included students in STEM fields and non-STEM fields. Among the STEM students, there was a sub-sample of students who had recently been accepted into a STEM scholarship program (MAKERS: Making to Advance Knowledge, Excellence, and Recognition in STEM). The MAKERS program is funded by a collaborative S-STEM Grant from the National Science Foundation (NSF), and provides scholarships, academic support, and social support to students majoring in STEM fields (biological, physical, mathematical, geological, and computer and information sciences; engineering; and associated technology areas). However, these students had not yet begun their participation in the MAKERS program, so they were combined with the rest of the sample (descriptive statistics comparing the two samples are provided in Table 1). In total, 67 students from the MAKERS program participated in the survey, but five were excluded due to substantial missing data (n = 62). The sample of students who were not part of the MAKERS program consisted of 499 participants; however, 67 were excluded due to substantial missing data, and 18 were excluded for being a univariate outlier (i.e., more than three standard deviations above or below the mean) on at least one measure (n = 414). The final sample consisted of 476 participants.

Measures

STEM professional identity overlap (STEM-PIO-1)

Students' identification as a STEM professional was assessed with a one-item pictorial scale derived from the measure of interpersonal closeness developed by Aron et al. (1992). Participants were given the following instructions, “STEM professionals are individuals whose professional activities relate to the STEM fields (Science, Technology, Engineering, or Mathematics). Select the picture that best describes the current overlap of the image you have of yourself and your image of what a STEM professional is.” Participants selected among a set of seven overlapping circles varying in the degree of overlap: 1 = no overlap and 7 = near complete overlap (see Figure 1).

STEM attitudes and identity

Existing measures of science identity and attitudes toward science were adapted to assess STEM identity and attitudes toward STEM (Young et al., 2013). Responses were provided on a scale ranging from 1 (strongly disagree) to 5 (strongly agree). STEM identity was assessed with five items, including “Being a STEM student is an important reflection of who I currently am.” Attitudes toward STEM were assessed with four items, such as, “In general, I find working on assignments in my STEM courses very interesting.” For students who were not STEM majors, instructions asked them to reflect on the STEM courses they had recently taken in high school or college. Participants were also asked to indicate their current satisfaction with their STEM major, ranging from “I think I have found my true calling” to various levels of dissatisfaction (i.e., “I am not sure this is what I want to do for a living;” “I have not enjoyed most of the courses I have taken so far;” “I am considering changing majors”). Responses expressing dissatisfaction were collapsed into a single category and compared to those indicating that their major represents their true calling.

STEM self-efficacy and agency

A subset of six items were adapted from a measure of self-efficacy in biology (Baldwin et al., 1999) to assess self-efficacy in STEM. For example, “I am confident I have the ability to learn the material taught in my STEM courses.” Responses were scored on a scale ranging from 1 (strongly disagree) to 5 (strongly agree). Non-STEM majors were asked to reflect on STEM courses they had recently taken in high school or college. To assess agentic behaviors, participants were asked “How confident are you that you have knowledge of all requirements needed to graduate in your major(s)?” with response options ranging from 1 (not sure at all) to 4 (completely sure).

Performance and mastery goal orientation

Performance and mastery goal orientations were assessed using a measure adapted by Shell and Husman (2008). Instructions told participants that, “Students differ in what they want to get out of the courses they take. Use the scale given to rate how important achieving each of the following is for you.” Responses were recorded on a scale ranging from 1 (very unimportant) to 5 (very important). Performance goal orientation was captured by eight items, for example, “Remembering enough from the class to impress other people.” Mastery goal orientation was assessed with five items, including, “Learning new knowledge or skills in the class just for the sake of learning them.”

Results

We first sought to examine the construct validity of the STEM-PIO-1 by examining its associations with measures assessing the same or highly related constructs (Table 2 presents descriptive data for measured variables). There were moderate associations with Young's et al. (2013) adapted measure of STEM identity [r₍₄₇₆₎ = 0.42, p < 0.001], a measure of attitudes toward STEM [r₍₄₇₆₎ = 0.39, p < 0.001], and a measure of self-efficacy in STEM [r₍₄₇₆₎ = 0.37, p < 0.001]. We then examined whether STEM-PIO-1 would predict goal orientation among STEM majors, with the expectation that students who highly identified with STEM professionals would exhibit greater mastery goal orientation, but not necessarily performance goal orientation. This prediction was partially supported—STEM-PIO-1 was positively associated with an orientation toward mastery goals [r₍₄₇₅₎ = 0.14, p = 0.002], and weakly positively associated with performance goal orientation [r₍₄₇₅₎ = 0.09, p = 0.048].

TABLE 2

Table 2. Descriptive statistics for measured variables in study 1.

To further illustrate the construct validity of the measure, we examined whether STEM-PIO-1 would detect differences in identification with STEM between students majoring in a STEM field vs. those in non-STEM fields. Students were categorized into non-STEM (n = 162), soft-STEM (n = 43), hard-STEM (n = 196), and health (n = 63) (the designations of soft and hard STEM were consistent with those proposed by Biglan, 1973). An ANOVA revealed significant differences in STEM-PIO-1 scores across majors, F_{(3, 460)} = 10.06, MSE = 2.81, p < 0.001. Bonferroni post-hoc analysis of mean differences (see Table 3) indicated that non-STEM majors reported significantly lower STEM-PIO-1 (M = 3.75, SD = 1.71) than hard-STEM majors (M = 4.70, SD = 1.68; d = 0.56). Thus, the STEM-PIO-1 was sensitive to differences in STEM identification among individuals who are, and are not, currently pursuing a STEM major. We also examined whether STEM-PIO-1 in STEM majors would predict their degree of confidence in knowing the requirements to graduate for their major. STEM-PIO-1 was positively associated with this index of agency [r₍₃₁₄₎ = 0.23, p < 0.001] which is consistent with previous research showing a connection between identity and agency (Herrera et al., 2012).

TABLE 3

Table 3. Differences in STEM-PIO by major category.

The relationship between STEM identity and satisfaction with one's STEM major was examined by comparing the STEM-PIO-1 between those students endorsing the statement, “I think I have found my true calling” vs. those expressing various levels of dissatisfaction. Among STEM majors, STEM-PIO-1 scores were greater for students who reported a positive view of their STEM major (M = 4.68, SD = 1.61), relative to those who expressed any level of dissatisfaction with their STEM major [M = 3.94, SD = 1.80; t₍₃₁₂₎ = 3.50, p = 0.001; d = 0.44].

We then examined whether scores on the STEM-PIO-1 varied as a function of demographic characteristics. Scores for the STEM-PIO-1 were not correlated with participant age [r₍₄₇₆₎ = −0.01, p = 0.760], parental income [r₍₃₆₇₎ = 0.05, p = 0.352], mother's education [r₍₄₅₈₎ = 0.04, p = 0.391], or father's education [r₍₄₃₀₎ = −0.004, p = 0.926]. Scores for the STEM-PIO-1 also did not differ between men (M = 4.35, SD = 1.64) and women (M = 4.21, SD = 1.76), t₍₄₇₄₎ = −0.72, p = 0.475; d = 0.08, even when restricting the analysis to hard-STEM majors. STEM-PIO-1 also did not vary as a function of whether the individual is part of a racial or ethnic group that is typically underrepresented in STEM (M = 4.30, SD = 1.75) or a group that is not typically underrepresented (M = 4.19, SD = 1.69), t₍₄₆₆₎ = 0.64, p = 0.523; d = 0.06, regardless of major.

Study 2

The results from Study 1 provide preliminary evidence for the validity of the STEM-PIO-1. It displayed moderate association with an adapted self-report measure of STEM identity (Young et al., 2013), and was positively associated with agency and self-efficacy within the STEM major, as well as an intrinsic motivation to pursue academic goals. However, as a single-item measure, it is difficult to ascertain its reliability, as well as the extent to which a single-item measure can capture meaningful aspects of one's STEM identity. To address these issues, a second study was conducted in order to test the reliability of a multi-item measure of STEM professional identity overlap. In particular, the STEM-PIO-1 measure from Study 1 was adapted to measure the specific components of STEM identity presented by Carlone and Johnson (2007): competence, performance, and recognition. These new items were included alongside the original measure to examine how strongly they converge, and to assess the reliability of the total set of items. We also examined the test-retest reliability of the original STEM-PIO-1 with a small subset of our sample that overlapped with Study 1. Finally, we sought to replicate the findings from Study 1, using both the single-item measure, and the new composite measure.