Development and Validation of the College Social–Emotional Crossroads Inventory

Abstract

The purpose of this study was to develop and provide initial validity evidence for the College Social–Emotional Crossroads Inventory (C-SECI). A sample (N = 751) of undergraduate students was randomly split into two samples for exploratory and confirmatory factor analysis. Results of exploratory factor analysis indicated that three factors should be extracted from the data and that the items comprised three subscales: Campus Cultural Fit, Academic Capital, and School–Family Integration. A confirmatory factor analysis suggested a bifactor structure was the best representation of the C-SECI items. Furthermore, scores on the C-SECI subscales correlated in expected directions with measures of institutional classism, academic self-efficacy, academic progress, global stress, first-generation college student status, subjective social status, and family income. The C-SECI is a brief measure that can be used to capture tensions students may experience between their postsecondary institutions and families and communities of origin.

Keywords

scale development college adjustment higher education

Higher education is undergoing rapid change (Bastedo, Altbach, & Gumport, 2016). Political, economic, and social forces have affected how colleges and universities are funded, who they allow access, and how they frame their social purpose. With these macro-level changes, much attention has been given to how contextual factors affect students’ academic and career trajectories. Attention has been given to race, gender, and social class along with how social determinants affect career development in the vocational literature (Garriott, Faris, Frazier, Nisle, & Galluzzo, 2017; Lee et al., 2017).

While some gains have been made in advancing the understanding of how contextual factors inform academic and career development (Brown & Lent, 2017; Fouad & Santana, 2017), more work is needed to unpack how environmental factors shape career opportunities and success. Furthermore, although research shows that environmental and contextual factors are related to students’ social–emotional experiences while in college (Jehangir, Williams, & Jeske, 2012), these experiences are not always well captured by extant career theory. For example, social cognitive career theory (SCCT; Lent, Brown, & Hackett, 1994) integrates contextual factors into the explanation of career interests, choice, and performance. However, SCCT posits direct and indirect relations between contextual supports and barriers and goals through self-efficacy (Lent et al., 1994; Lent et al., 2003). Intermediary social–emotional connections between one’s environment and career-related variables are omitted in SCCT, despite having been documented in prior research and may be particularly important for historically marginalized groups (Gloria, Castellanos, & Orozco, 2005; Gloria & Ho, 2003). For example, first-generation college students often describe challenges reconciling the norms and expectations of their home environments with those of the academy (Davis, 2010; Jehangir et al., 2012). This may in part explain some findings suggesting that relations between environmental variables and academic and career outcomes using extant frameworks are relatively weak (Garriott, Hudyma, Keene, & Santiago, 2015; Lent & Brown, 2019).

Vocational scholars have called for research that addresses how systems of privilege and oppression function as social determinants of well-being (Blustein, 2011). More recent theoretical frameworks, such as psychology of working theory (PWT), were designed to address these concerns (Duffy, Blustein, Diemer, & Autin, 2016). However, PWT does not directly address the academic and career development of college students. The critical cultural wealth model (CCWM; Garriott, 2019) of academic and career development was designed to address these issues. The CCWM captures the role of structural forces in shaping academic and career trajectories via social–emotional tensions students may experience between their home environments and postsecondary institutions. The CCWM is grounded in critical theory (Garcia, López, & Vélez, 2017; Yosso, 2005) and includes five primary dimensions: structural and institutional conditions, social–emotional crossroads, cultural wealth, career self-authorship, and academic, career, and psychological outcomes. Propositions of the CCWM have received empirical support (Duffy, Kim, et al., 2019).

Theoretical Framework

Within the CCWM, social–emotional crossroads refer to the interactions between community, culture, family, and institution often encountered by students in higher education. We identified three overarching crossroads themes within empirical studies and personal narratives of historically marginalized college students based on a systematic review of the literature: campus cultural fit, normative capital, and school–family integration.

Campus cultural fit refers to the degree a student feels they are welcomed, valued, and belongs at their institution (Garriott, 2019). This definition was adopted given the conceptual and statistical overlap between instruments designed to measure constructs such as sense of belonging, cultural congruity, and perceptions of the university environment used in prior literature (Gloria & Kurpius, 1996; Wells & Horn, 2015). This construct was also defined broadly as previous measures have been designed to capture specific dimensions of identity, such as ethnicity (Gloria & Kurpius, 1996). Therefore, the goal in this study was to design an instrument that could be used broadly to capture feelings of inclusion across various dimensions of social inequality.

Normative capital is defined as a student’s familiarity and access to forms of capital typically prioritized by postsecondary institutions (Garriott, 2019). Sometimes referred to as the “hidden curriculum,” normative culture in higher education is frequently omnipresent while remaining invisible or unspoken. The forms of capital privileged within university settings are grounded in White upper middle-class norms and tend to advantage individualism, extroversion, and competition (Cabrera, 2014; Stephens, Fryberg, Markus, Johnson, & Covarrubias, 2012). Examples of normative capital needed to navigate higher education institutions include knowing how to get support with academic, financial, and emotional concerns as well as personal comfort with behavioral expectations (e.g., approaching professors during office hours; Davis, 2010). Although the function of social and cultural capital has been examined qualitatively in prior research (Sandoval-Lucero, Maes, & Klingsmith, 2014), no instruments are available for researchers who wish to capture this construct using quantitative designs.

School–family integration is the degree to which students feel they are able to bridge school and family during their college experience (Garriott, 2019). The prioritization of individualism in university settings may disadvantage students with strong familial bonds (Winkle-Wagner, 2009). Specifically, an institution that prioritizes individualism may disrupt both emotional (e.g., ability to talk with family about one’s college experience) and logistical (e.g., ability to care for family members at home) familial connections. Although family roles and responsibilities have been examined extensively in terms of their relation to work, no measures assess the extent to which college students feel their family and school lives are integrated during college. Furthermore, although existing measures capture familial support during college and for career decisions, none of these measures assess perceived compatibility between school and family. This is a notable gap in the literature, given the dominant culture of higher education—designed to reward White, economically advantaged students—can itself be a barrier (Stephens et al., 2012).

Within the CCWM, structural and institutional conditions are hypothesized to predict social–emotional crossroads. Specifically, the extent to which structural and institutional conditions are oppressive or liberating will be associated with perceptions of campus cultural fit, normative capital, and school–family integration (Garriott, 2019). Social–emotional crossroads are also hypothesized to relate to academic and career outcomes. In particular, when a student experiences high levels of campus cultural fit, normative capital, and school–family integration, they will be more likely to experience high levels of academic persistence, performance, and satisfaction as well as direction in their career (Garriott, 2019).

The purpose of this study was to develop a measure of social–emotional crossroads consistent with the CCWM (Garriott, 2019). We hypothesized (Hypothesis 1) that scores on a measure of institutional classism would negatively correlate with factors for the scale given that negative structural and institutional factors are hypothesized to be associated with lower campus cultural fit, normative capital, and school–family integration in the CCWM (Garriott, 2019). Furthermore, we hypothesized (Hypothesis 2) that academic progress and academic self-efficacy would positively correlate with factors for the scale given the CCWM proposes that social–emotional crossroads will be associated with academic adjustment (Garriott, 2019). We hypothesized that scores on the College Social–Emotional Crossroads Inventory (C-SECI) would be weakly correlated with global stress (Hypothesis 3) because this variable is an indicator of general feelings of distress rather than college-specific stressors. We also hypothesized (Hypothesis 4) that scores on the C-SECI would be positively correlated with first-generation student status, subjective social status, and total family income based on literature suggesting that students with greater social class privilege report higher levels of academic and life satisfaction (Allan, Garriott, & Keene, 2016).

Method

Item Development and Researcher Positionality

The researchers’ institutional review board approved all study procedures. All authors participated in developing items for the C-SECI. The research team included one assistant professor, three doctoral students, and five masters students in a counseling psychology program. The research team members occupied both dominant (e.g., White, straight, continuing-generation college students, men) and marginalized (e.g., students of color, gay, first-generation college students, women) seats of power, privilege, and oppression. This allowed for the integration of experiential and institutional knowledge in the development of the measure. Specifically, research team members were able to draw from their experiences working with, and experiences living as, college students with intersecting positions of advantage and marginalization. Collectively, the research team expressed a desire to challenge preexisting narratives of historically marginalized college students as deficient or damaged. The team’s values centered on highlighting ways in which institutional norms and practices may fail students.

Throughout the item development process, the team also discussed potential biases and assumptions that might be grounded in our privileged statuses (e.g., deficit thinking). Members of the team reviewed literature and first-person narratives on the experiences of first-generation, low-income students and also drew from personal experiences to inform development of items for the scale. This resulted in 58 items spread across the three domains labeled campus cultural fit (15 items), financial and cultural capital (25 items), and school–family integration (18 items). Items were written with positive and negative valence to reduce the impact of common method variance on scale items (Seng Kam & Meyer, 2015).

Five undergraduate students and one director affiliated with a program for first-generation college students at the authors’ academic institution reviewed the 58 items. Two members of the research team facilitated a focus group with the students, and the first author met individually with the director to receive verbal feedback on the items. The students and director were provided with the item categories and definitions and asked if any additional themes or item content should be added. They were also asked about the items’ readability and relevance to their experience. After documenting feedback, the research team made wording changes to several items to improve clarity and added 1 item for a final pool of 59 items.

Participants and Procedure

A sample of college students (N = 790) was recruited from two public 4-year universities in the United States. The majority (n = 704) of students in the sample attended a predominantly White institution located in the Midwest. Carnegie classifications for this institution include doctoral university with very high research activity, high undergraduate enrollment, full-time, more selective, large, and primarily residential. A second group (n = 86) of students attended a Hispanic- and minority-serving public university in the Northeast. Carnegie classifications for this institution include doctoral university with high research activity, high undergraduate enrollment, full-time, selective, large, and primarily nonresidential.

A survey link was e-mailed to all eligible students at the institution in the Midwest and to an introductory psychology research pool at the institution in the Northeast. The first page of the survey contained information regarding informed consent. Consent to participate in the research was required before an individual could complete additional questions on the survey. The survey took approximately 10–15 minutes to complete, and participants were allowed to enter their name into a raffle for a gift certificate to an online shopping center upon completing the survey. A total of 1,507 students opened the survey. Within this group, 790 students consented and completed items for an approximate response rate of 52%. This response rate exceeded what is typical for online surveys with incentives administered in higher education settings (Sax, Gilmartin, & Bryant, 2003). The 790 usable surveys were then checked for excessive missing data, which were defined as any case missing 20% or more of the values for any measure in the survey (Peng, Harwell, Liou, & Ehman, 2006). This resulted in the deletion of 37 additional cases. Two participants identified as graduate students and were also removed from the data set. This larger sample (N = 751) was then randomly split into two smaller samples for the purposes of conducting exploratory factory analysis (EFA; n = 378) and confirmatory factor analysis (CFA; n = 373). A series of χ² tests showed that the EFA and CFA samples were not significantly different with regard to race and ethnicity, gender, institutional affiliation, first-generation college student status, income, or year in school (all p values > .05).

The age of the total sample ranged from 18 to 55 years with a mean of 20.29 years. Participants identified as women (n = 391, 52%), men (n = 265, 35.3%), gender-fluid (n = 2, 0.3%), or another option not provided (n = 6, 0.7%). Eighty-seven (11.5%) participants did not identify their gender. Participants identified as White (n = 488, 65%), Asian or Asian American (n = 120, 16%), Latinx or Hispanic (n = 73, 9.7%), Black or African American (n = 23, 3.1%), Multiracial (n = 35, 4.7%), or another option not provided (n = 1, 0.3%). This participant identified as “Arab.” Eleven (1.5%) participants did not identify their race or ethnicity. The majority of the participants (n = 666, 88.7%) attended the institution in the Midwest. Participants reported that they would (n = 241, 32.1%) or would not (n = 504, 67.1%) be the first person in their family to earn a bachelor’s degree. Six (0.8%) participants did not report this information. The year of study of participants was first year (n = 216, 28.8%), second year (n = 165, 22%), third year (n = 176, 23.4%), fourth year (n = 164, 21.8%), and fifth year and beyond (n = 20, 2.7%). Ten participants (1.3%) did not provide their year in school. Participants reported their family’s total annual income as less than $10,000 (n = 22, 2.9%), $10,000–14,999 (n = 6, 0.8%), $15,000–24,999 (n = 17, 2.3%), $25,000–34,999 (n = 39, 5.2%), $35,000–49,999 (n = 38, 5.1%), $50,000–74,999 (n = 90, 12.0%), $75,000–99,000 (n = 60, 8.0%), $100,000–149,999 (n = 127, 16.9%), $150,000–199,999 (n = 53, 7.1%), and $200,000 or more (n = 127, 16.9%). Seventy-seven (10.3%) participants reported that they did not know their annual family income, and 95 (12.6%) participants did not provide this information.

Measures

Academic progress

Academic progress was measured with 7 items used in prior research (Lent et al., 2005). Participants rated their perceived progress toward a list of academic goals on a Likert-type scale ranging from 1 (no progress at all) to 5 (excellent progress). A sample item is “Completing academic requirements of your major satisfactorily.” Scores are averaged with high scores indicative of perceptions of strong progress toward one’s academic goals. Evidence of criterion validity for scores on the measure has been demonstrated through a positive correlation with academic persistence (Lent et al., 2005). Coefficient αs for scores on the scale have ranged from .84 to .90 (Lent et al., 2005). Coefficient α for scale scores in this study was .93.

Academic self-efficacy

Academic self-efficacy was measured with the Course subscale of the College Self-Efficacy Inventory (Solberg, O’Brien, Villareal, Kennel, & Davis, 1993). Seven items were rated on a Likert-type scale ranging from 0 (no confidence at all) to 10 (very confident). A sample item is “Do well on your exams.” Scores are averaged with high scores indicating high academic self-efficacy. Scores on the Course Self-Efficacy subscale have been shown to correlate in expected directions with measures of psychological adjustment and stress (Solberg et al., 1993). Coefficient α for scores on the course Self-Efficacy subscale was .88 in an initial validation study (Solberg et al., 1993). Coefficient α for scale scores in this study was .86.

C-SECI

Participants rated the 59 items developed for the C-SECI on a Likert-type scale ranging from 1 (not at all like me) to 7 (very much like me).

Institutional classism

Institutional classism was measured with the Institutional Classism subscale of the Classism Experiences Questionnaire—Academe (Langhout, Rosselli, & Feinstein, 2007). Participants rated 5 items on a Likert-type scale ranging from 1 (never) to 5 (many times). A sample item is “You could not afford a class (e.g., music, science, film) because you could not afford the fees for the class (for materials, travel, etc.).” Scores are averaged with high scores indicative of more frequent experiences with institutional classism. Scores on the Institutional Classism subscale have been shown to correlate in expected directions with school persistence and academic adjustment (Langhout et al., 2007). Coefficient α for scores on the Institutional Classism Scale was .83 in an initial validation study (Langhout et al., 2007). Coefficient α for scale scores in this study was .81.

Stress

Perceived stress was measured with the Perceived Stress Scale (PSS; Cohen, Kamarck, & Mermelstein, 1983). Participants rated 10 items on a Likert-type scale ranging from 1 (never) to 5 (very often) and were asked to reference the past month when completing the measure. A sample item is “How often have you been able to control irritations in your life?” Scores are averaged and several items are reverse-scored such that high scores indicate higher levels of perceived stress. Scores on the PSS have been shown to correlate in expected directions with measures of health, academic progress, and coping (Cohen et al., 1983; Garriott & Nisle, 2018). Coefficient α for scores on the PSS has ranged from .84 to .88 in prior research (Cohen et al., 1983; Garriott & Nisle, 2018). Coefficient α for scores in this study was .87.

Subjective social status

The MacArthur Scale of Subjective Social Status was used to measure subjective social status (Adler, Epel, Castellazzo, & Ickovics, 2000). Participants were shown a picture of a ladder and given the following instructions: “Think of this ladder as representing where people stand in our society. At the top of the ladder are the people who are the best off, those who have the most money, most education, and best jobs. At the bottom are the people who are the worst off, those who have the least money, least education, and worst jobs or no job.” Participants then rated their standing on the ladder using a scale ranging from 1 (bottom rung) to 10 (top rung).

Results

Sample 1

Preliminary analyses

Missing data within the full data set were assessed prior to conducting the main data analyses. There were 14 cases with a total of 28 (0.05%) missing values of 47,285 possible values in the data set. Percentage of missing data ranged from 0.3% to 0.5% on the C-SECI items. Results of Little’s missing completely at random (MCAR) test were statistically significant, χ² = 1,037.55, p < .001, indicating data were not MCAR. Therefore, a dummy variable was created for cases with missing values (0 = not missing, 1 = missing), and this variable was regressed on all other variables in the data set to determine the pattern of missingness (Schlomer, Bauman, & Card, 2010). There was no significant association between the dummy variable and any of the items for the C-SECI. However, there was a significant correlation between institution and the dummy variable (r = .32, p < .001), suggesting attending the Northeastern university was associated with missing values in the data set. Items were examined individually for this subgroup, and no reliable pattern of missing values emerged. Therefore, the data were determined to be missing at random, and expectation–maximization was used to replace missing values in the data set (Schlomer et al., 2010).

Primary analyses

Parallel analysis with principle axis factoring was conducted to gain an initial estimate of the number of factors to extract from the items. A total of 1,000 random data sets were requested, and factors from eigenvalues corresponding to the original data set were compared to those randomly generated from the parallel analysis. Factors corresponding to eigenvalues produced from the actual data set that were greater than those produced by the parallel analysis at the 95% confidence interval (CI) were retained (Hayton, Allen, & Scarpello, 2004). The parallel analysis suggested the presence of three factors. Therefore, a three-factor solution was extracted and examined in subsequent analyses.

Principle axis factoring with direct oblimin rotation was next used to explore a three-factor solution for the items. Scree plots, interpretability of the factor solution, and factor loadings for the items were examined to finalize the number of factors to be extracted. All three of these criteria strongly suggested the presence of three factors. Therefore, a three-factor solution was specified and used to retain items for the scale.

Item retention was based on several criteria: (a) factor loadings > .32, (b) cross-loadings < .15, (c) absolute factor loadings > .32 on only one factor, (d) communalities > .40, and (e) the inclusion of at least 3 items on a given subscale (Worthington & Whittaker, 2006). Using these criteria resulted in an interpretable three-factor scale comprised of 14 items that explained 56% of the variance in scores on the items. The first factor (5 items; α = .84) was named Campus Cultural Fit and included items reflecting students’ perceptions that they are welcomed, valued, and belong at their university. Higher scores on this subscale indicate high levels of campus cultural fit. The second factor (3 items; α = .81) was named Academic Capital and included items reflecting students’ perceptions that they are familiar with institutional academic resources and norms. Higher scores on this subscale indicate high levels of academic capital. The third and final factor (6 items; α = .85) was named School–Family Integration and contained items reflecting students’ perceptions that their school and family lives are compatible. Items were reversed-scored on this subscale such that high scores indicate higher perceptions of compatibility between school and family. Table 1 includes factor loadings for the 14 C-SECI items. The three factors were significantly correlated, but these correlations did not exceed levels suggestive of construct redundancy (r > .80; see Table 2).

Table 1.

Item Factor Loadings for the College Social–Emotional Crossroads Inventory.

Variable	1	2	3
Factor 1: Campus cultural fit
1. I feel connected to a community on campus	.88
2. I feel I belong to a social group on campus	.85
3. I attend campus events	.65
4. I feel welcomed as a student by this college	.64
5. I feel very comfortable on campus	.58
Factor 2: Academic capital
6. I know where to get help with my academics		.86
7. I am comfortable approaching professors with questions about the class material or assignments		.76
8. If I was experiencing an academic problem, I would know where to go		.69
Factor 3: School–family integration
9. I feel resentment from my family for being in college			.78
10. I feel like my family does not understand why I am attending college			.75
11. Gender roles in my family do not support me getting a higher education			.74
12. I feel like an outcast in my family because I am in college			.71
13. My family responsibilities have conflicted with my academic responsibilities			.69
14. My family obligations interfere with my ability to have a social life at college			.60

Table 2.

Means, Standard Deviations, and Correlations—Sample 1.

Variable	1	2	3
1. Campus cultural fit	__	.45***	.17**
2. Academic capital		__	.14*
3. School–family integration			__
M	4.62	5.02	1.60
SD	1.37	1.44	1.60

*p < .05. **p < .01. ***p < .001.

Sample 2

A CFA with a second sample of undergraduate college students was conducted to confirm the factor structure of the C-SECI retained from the first sample. Drawing from the CCWM (Garriott, 2019), four plausible factor structures were examined. The first model tested was a first-order model in which all C-SECI items were loaded onto a general factor. Retention of this model would suggest that all C-SECI items represent a unidimensional construct. The second model tested was a correlational model in which items loaded onto their respective factor and the factors were correlated with one another. Retention of this model would suggest separate, but related factors underlying the C-SECI items. The third model tested was a second-order model in which items were loaded onto their respective factors and the factors were loaded onto a general factor. Subfactors were not correlated in this model. Retention of the third model would suggest the presence of subfactors and a general factor for the items. The fourth and final model tested was a bifactor model in which items were loaded onto their respective factor and a general factor. In a bifactor model, all latent factors are orthogonal. Retention of this model would suggest the presence of subfactors as well as a general factor and that C-SECI items load on both.

Preliminary analyses

Data were first assessed for normality. Skewness statistics for the main study variables ranged from −1.05 to 2.45, and kurtosis statistics ranged from −.63 to 5.90. These values were in the acceptable range for structural equation modeling (SEM) analyses (Weston & Gore, 2006). However, Mardia’s coefficient was 158.82, suggesting the presence of multivariate nonnormality. Therefore, robust maximum-likelihood estimation was used to correct for nonnormality in subsequent analyses.

Primary analyses

The CFA models were examined using Mplus Version 8.0 (Muthén & Muthén, 1998–2017). The comparative fit index (CFI), root mean square error of approximation (RMSEA), and standardized root mean residual (SRMR) were examined to determine the adequacy of model-to-data fit. Recommended values for these fit indices have ranged from less (CFI ≥ .90, RMSEA ≤ .08, SRMR ≤ .10) to more conservative (CFI ≥ .95, RMSEA ≤ .06, SRMR ≤ .06; Kline, 2016). In addition to fit indices, residual correlations were examined for evidence of potential model misspecification. In general, a pattern of residual correlations >|.10| can be indicative of poor model fit (Kline, 2016). Model comparisons were made using the Aikake information criterion (AIC), given models were nonnested. Lower AIC values were taken as evidence of improved model-to-data fit.

The first-order model, with all items loaded on a single factor, demonstrated poor fit to the data, χ²(77) = 851.21, p < .001; CFI = .424; RMSEA = .164; 90% CI [.154, .174]; SRMR = .149, as all fit indices were out of the generally acceptable range. Fit indices for the correlational model, with items loaded onto correlated subfactors, suggested adequate fit to the data, χ²(74) = 170.48, p < .001; CFI = .928; RMSEA = .059; 90% CI [.047, .071]; SRMR = .073. All items significantly loaded onto their respective factors at values of .44 or above. Covariances between the subfactors were also significant and ranged from .24 to .81. Residual correlations for the model ranged from −.002 to .25, and there were 15 residual correlations at values >|.10|. The majority of these residual correlations were between individual items on the Campus Cultural Fit subscale. Comparison of AIC values for the first-order (AIC = 17,652.76) and correlational models (AIC = 16,614.28) favored the correlational model.

The second-order model, which included a general factor and subfactors, produced the same fit statistics and AIC value as the correlational model, χ²(74) = 170.48, p < .001; CFI = .928; RMSEA = .059; 90% CI [.047, .071]; SRMR = .073. All items significantly loaded onto their respective factors at values of .44 or above, and each subfactor significantly loaded onto the general factor (βs ranging from .36 for school–family integration to .73 for academic capital). The degrees of freedom and fit statistics for the correlational and second-order models were identical. The pattern of residual correlations for the second-order model was also the same as the correlational model, with residual correlation values >|.10| primarily among the Campus Cultural Fit subscale items.

The bifactor model, in which items loaded onto a general factor and their respective subfactors, exhibited close fit to the data, χ²(63) = 66.50, p = .357; CFI = .997; RMSEA = .012; 90% CI [.000, .034]; SRMR = .030 (see Figure 1). All items significantly loaded onto their respective factors with the exception of two of the campus cultural fit items. These two items did not significantly load onto their subfactor but did significantly load onto the general factor. Factor loadings for the bifactor model ranged from −.01 to .79. These results suggested the presence of a suppressor effect (Beckstead, 2012). That is, the general factor appeared to explain most of the variance in scores on these two campus cultural fit items, whereas the subscale appeared to add little additional meaningful variance. Residual correlations for this model were all <|.10|. Additionally, the AIC value (AIC = 16,494.25) was lower when compared to the correlational and second-order models, suggesting the bifactor model provided a superior fit to the data.

Figure 1.

Confirmatory factor analysis model for Sample 2. All coefficients are standardized. Error terms are not included to reduce visual clutter. ***p < .001.

Retaining the bifactor model suggested that it was next appropriate to calculate omega (ω), omega hierarchical (ωH), and explained common variance (ECV) statistics. These statistics were computed using the code provided by McNeish (2017) for calculating McDonald’s ω. ω is an expression of reliability that calculates the variance in total scores accounted for by subfactors versus error. ω for the bifactor model was .92, suggesting approximately 92% of the variance in total C-SECI scores was attributable to the subfactors and the remaining 8% of variance was attributable to error. ωs for each of the subscales were as follows: Campus Cultural Fit (ω = .87), Academic Capital (ω = .84), and School–Family Integration (ω = .85). ωH provides information on the amount of variance in scores attributable to the general factor when the subfactors are treated as error. ωH for the general factor was .74. When calculated in relation to the ω value for the general factor, relative ωs for the bifactor model indicated that 80% of the reliable variance in C-SECI scores was attributable to the general factor (.74/.92 = .80) and 19% (.92/.74 = .18/.92 = .19) was attributable to the subfactors. ωH values for the subscales were Campus Cultural Fit (ω_H = .59), Academic Capital (ω_H = .33), and School–Family Integration (ω_H = .14). The ECV value indicates how much of the common variance in scores is attributable to the general factor versus subfactors. The ECV value for the bifactor model was .61, indicating 61% of the common variance in scores was attributable to the general factor and the remaining 39% of the variance was spread among the three subfactors.

Table 3 contains means, standard deviations, and correlations between study variables. In support of Hypothesis 1, institutional classism was significantly negatively correlated with campus cultural fit, academic capital, and school–family integration. Hypothesis 2 was also supported. Academic progress and academic self-efficacy were positively and moderately correlated with campus cultural fit, academic capital, and school–family integration. In support of Hypotheses 3 and 4, scores on the C-SECI were weakly correlated with global stress and were moderately correlated with first-generation student status and subjective social status. Specifically, higher levels of perceived campus cultural fit, academic capital, and school–family integration were significantly positively correlated with higher subjective social status and being a continuing-generation college student.

Table 3.

Means, Standard Deviations, and Correlations—Sample 2.

Variable	1	2	3	4	5	6	7	8	9	10
1. Campus cultural fit	__	.46	.24	.25	.31	.36	−.04	.25	.32	.27
2. Academic capital		__	.26	.16	.56	.48	−.13	.15	.27	.18
3. School–family integration			__	−.36	.25	.17	−.05	.27	.29	.24
4. Institutional classism				__	−.15	−.22	.18	−.25	−.34	−.23
5. Academic self-efficacy					__	.66	−.12	.13	.26	.18
6. Academic progress						__	−.12	.28	.34	.28
7. Stress							__	−.13	−.17	−.07
8. First-generation status								__	.43	.46
9. Subjective social status									__	.48
10. Total family income										__
M	4.64	4.99	1.49	3.41	4.93	2.01	3.13	__	6.32	7.60
SD	1.33	1.50	0.90	1.24	1.76	0.83	0.44	__	1.86	2.60

Note. N = 373. Correlations ≥|.19| are statistically significant at the p > .001 level. Correlations between |.18| and |.13| are statistically significant at the p < .01 level. Correlations between |.08| and |.12| are statistically significant at the p < .05 level. Correlations below |.12| are not statistically significant at the p < .05 level.

Discussion

The purpose of this study was to develop and provide initial validity evidence for a measure of college social–emotional crossroads consistent with the CCWM (Garriott, 2019). Results supported the presence of general and subfactors for C-SECI scores. Items retained for the C-SECI provide researchers with the advantage of assessing education-related barriers broadly, rather than confining these to narrow ranges of identity (i.e., race and gender). Furthermore, the C-SECI has undergone systematic validation procedures and is theory-based, unlike other similar instruments used in educational and vocational research that rely on clusters of items taken from national surveys or haphazardly created by researchers. Importantly, the C-SECI captures facets of the college experience, such as family, not included in previous measures.

Researchers using the C-SECI may use several strategies to incorporate the measure into quantitative analyses. The bifactor model retained from the CFA suggests that the C-SECI can be represented by an observed general factor or a general factor and its three subfactors using latent variable SEM. If using the C-SECI in this manner, researchers would ideally model paths from the C-SECI general factor and subfactors to other variables simultaneously. This analytic strategy accounts for the shared and unique variance contributed by each dimension of the measure. Researchers may also wish to use the C-SECI subfactors outside of a latent SEM framework. In this case, researchers may use ipsative scoring in order to remove common variance associated with the general factor from its subfactors (Tracey, 2012). The subfactors of the C-SECI may then be used as observed variables in subsequent analyses.

It should be noted that a second-order model also provided an adequate fit to the data in this study. This model is theoretically consistent with the manner in which social–emotional crossroads are presented in the CCWM and is more parsimonious when compared to the bifactor model. Therefore, this model may also be appropriate for researchers depending on their goals and likely presents a more straightforward and less cumbersome option for scoring the measure in future research.

The C-SECI may be used in future research to test assumptions of the CCWM (Garriott, 2019). Specific research questions that may be examined include the following (a) Do structural and institutional conditions predict social–emotional experiences measured by the C-SECI? (b) Do scores on the C-SECI predict career self-authorship (work volition and career adaptability)? and (c) Do scores on the C-SECI mediate the relationship between structural and institutional conditions and career self-authorship? Results of these studies may point to potential areas and modes of intervention at colleges and universities.

The C-SECI may also be used as a program evaluation tool. Because university programs designed to support minoritized college students are increasingly focusing on students’ social–emotional experiences in addition to traditional outcomes (e.g., grade point average [GPA], retention), the C-SECI could provide valuable data on the relative impact of these programs. For example, in a qualitative study of the utility of multicultural learning communities for first-generation, low-income college students, researchers found that multicultural curriculum and critical pedagogy appeared to facilitate students’ interpersonal and cognitive development (Jehangir, Williams, & Pete, 2011). Program personnel and researchers may use the C-SECI to supplement similar evaluative designs with quantitative data.

On a broader scale, university administrators seeking to conduct campus climate surveys could use the C-SECI. Researchers have called for more frequent use of validated instruments in the design of campus climate surveys and the need to assess psychological dimensions of campus climate (Hurtado, Griffin, Arellano, & Cuellar, 2008). The C-SECI accomplishes both these aims and is a brief measure that may be attractive to institutions wishing to assess a large number of variables in a single survey.

Researchers may also choose to adapt the C-SECI to suit their specific aims. Although the items were administered generically in this study, researchers may wish to ask participants to complete items in response to specific social locations to capture particular forms of marginalization, as has been done with similar measures in prior research (Duffy, Gensmer, et al., 2019). Specifically, the Campus Cultural Fit subscale may be modified to capture how this construct is experienced based on particular social identities. For example, if examining links between institutional racism and campus cultural fit, researchers may ask students to reference their racial identity when completing items for the subscale. It is also important to highlight that the items comprising the C-SECI could be misinterpreted to suggest that students are unprepared or lack the cultural capital to succeed in higher education. This interpretation is consistent with a deficit-thinking framework in which students, rather than systems, are conceived of as problems to be solved (Patton & Museus, 2019). Instead, as evidenced by the moderate correlation between the C-SECI subscales and institutional classism, researchers should interpret scores on the C-SECI as symptomatic of unhealthy structural conditions.

Limitations and Future Directions

Although the C-SECI may provide a useful tool for researchers, this study has several limitations. First, data collection in this study was limited to two universities in the United States, with students at one of these institutions disproportionately represented in the samples. Thus, researchers using the C-SECI at institutions unlike that at which most of the data in this study were collected should interpret scores with caution. For example, it is possible that items on the C-SECI function differently for students attending community colleges or who are enrolled in online degree programs.

Additionally, the demographics of participants in this study were not representative of the college student population nationally. Specifically, the sample obtained for this study included a higher proportion of White and a lower proportion of Latinx students compared to national averages at colleges and universities (U.S. Census Bureau, 2018). A larger multi-institutional design may have increased the external validity of results from this study. Unfortunately, sample size restrictions did not allow for the test of invariance of the C-SECI by demographic or institutional characteristics. Therefore, little is known regarding how the C-SECI might perform when administered in more discrete groups of college students. More research is needed to determine if the factor structure of the C-SECI holds within various institutional contexts and across student demographic groups. The use of multiple-group SEM and item response theory models may be helpful in this regard.

Additionally, although results provide initial validity evidence for scores on the C-SECI, more research is needed to determine the degree of conceptual overlap between constructs measured by the C-SECI and other measures. For example, items on the Campus Cultural Fit subscale of the C-SECI share thematic content with items on the previously validated University Environment and Cultural Congruity Scales (Gloria & Kurpius, 1996) and items used to measure sense of belonging (e.g., Hoffman, Richmond, Morrow, & Salomone, 2002). Existing measures of family support could also be examined in relation to the School–Family Integration subscale. Tests of incremental validity in examining the associations between scores on these measures and relevant academic or career outcomes might be especially helpful in determining advantages of using the C-SECI in comparison to other measures.

Although items were developed to measure forms of academic, financial, and cultural capital, collectively described as normative capital in the CCWM (Garriott, 2019), only one facet of this construct emerged from the items in this study. More research is needed to develop measures that capture other forms of normative capital identified in the higher education literature, such as financial or peer capital. It is also possible that the items that emerged for this subscale in this study were reflective of the sample, which included large numbers of students with racial and economic privilege.

Responses to all instruments in this study were also captured at a single time point and in self-report format. It is possible that both these issues introduced some degree of error in participants’ responses to survey items. To gain a better understanding of students’ experiences of the three dimensions captured by the C-SECI, researchers may incorporate mixed-method research designs to aid in revealing particular aspects of the campus culture that are alienating, specific norms and expectations that are perceived as dominant, as well as common ruptures in school–family dynamics. The cross-sectional nature of the data also limited our ability to test a key proposition of the CCWM, that social–emotional crossroads mediate the relationship between structural and institutional conditions and student outcomes. Longitudinal research using the C-SECI will be necessary to test this hypothesis.

Despite these limitations, the C-SECI provides researchers and university personnel with a new tool for examining associations between structural and institutional conditions and student outcomes. The C-SECI may be used in future research and program evaluation efforts to examine the impact of institutional practices on college students’ social–emotional functioning.

Footnotes

Authors’ Note

A preliminary draft of this article was presented at the American Psychological Association Convention in Chicago, IL, August 2019. Ree Ae Jordan is currently a PhD student at the University of Wisconsin–Madison. Yeji Son is currently a PhD student at the University of Iowa.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Patton O. Garriott

References

Adler

N. E.

Epel

E. S.

Castellazzo

Ickovics

J. R.

(2000). Relationship of subjective and objective social status with psychological and physiological functioning: Preliminary data in healthy, White women. Health Psychology, 19, 586–592. doi:10.1037/0278-6133.19.6.586

Allan

B. A.

Garriott

P. O.

Keene

C. N.

(2016). Outcomes of social class and classism in first- and continuing-generation college students. Journal of Counseling Psychology, 63, 487–496. doi:10.1037/cou0000160

Bastedo

M. N.

Altbach

P. G.

Gumport

P. J.

(2016). American higher education in the 21st century: Social, political, and economic challenges (4th ed.). Johns Hopkins University Press.

Beckstead

J. W.

(2012). Isolating and examining sources of suppression and multicollinearity in multiple linear regression. Multivariate Behavioral Research, 47, 224–246. doi:10.1080/00273171.2012.658331

Blustein

D. L.

(2011). Vocational psychology at the fork in the road: Staying the course or taking the road less traveled. Journal of Career Assessment, 19, 316–322. doi:10.1177/1069072710395537

Brown

S. D.

Lent

R. W.

(2017). Social cognitive career theory in a diverse world: Closing thoughts. Journal of Career Assessment, 25, 173–180. doi:10.1177/1069072716660061

Cabrera

N. L.

(2014). Exposing whiteness in higher education: White male college students minimizing racism, claiming victimization, and recreating white supremacy. Race Ethnicity and Education, 17, 30–55. doi:10.1080/13613324.2012.725040

Cohen

Kamarck

Mermelstein

(1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396. doi:10.2307/2136404

Davis

(2010). The first generation student experience: Implications for campus practice, and strategies for improving persistence and success. Sterling, VA: Stylus.

10.

Duffy

R. D.

Blustein

D. L.

Diemer

M. A.

Autin

K. L.

(2016). The psychology of working theory. Journal of Counseling Psychology, 63, 127–148. doi:10.1037/cou0000140

11.

Duffy

R. D.

Gensmer

Allan

B. A.

Kim

H. J.

Douglass

R. P.

England

J. W.

Autin

K. L.

Blustein

D. L.

(2019). Developing, validating, and testing improved measures within the psychology of working theory. Journal of Vocational Behavior, 112, 199–215. doi:10.1016/j.jvb.2019.02.012

12.

Duffy

R. D.

Kim

H. J.

Gensmer

N. P.

Pendleton

L. H.

Boren

Garriott

P. O.

(2019). Testing a critical cultural wealth model of well-being among first-generation students. Journal of Counseling Psychology.

13.

Fouad

N. A.

Santana

M. C.

(2017). SCCT and underrepresented populations in STEM fields: Moving the needle. Journal of Career Assessment, 25, 24–39. doi:10.1177/1069072716658324

14.

Garcia

N. M.

López

Vélez

V. N.

(2017). QuantCrit: Rectifying quantitative methods through critical race theory. Race, Ethnicity, and Education, 21, 149–157. doi:10.1080/13613324.2017.1377675

15.

Garriott

P. O.

(2019). A critical cultural wealth model of first-generation and economically marginalized college students’ academic and career development. Journal of Career Development.

16.

Garriott

P. O.

Faris

Frazier

Nisle

Galluzzo

(2017). Multicultural and international research in four career development journals: An 11-year content analysis. The Career Development Quarterly, 65, 302–314. doi:10.1002/cdq.12109

17.

Garriott

P. O.

Hudyma

Keene

Santiago

(2015). Social cognitive predictors of first- and non-first-generation college students’ academic and life satisfaction. Journal of Counseling Psychology, 62, 253–263. doi:10.1037/cou0000066

18.

Garriott

P. O.

Nisle

(2018). Stress, coping, and perceived academic goal progress in first-generation college students: The role of institutional supports. Journal of Diversity in Higher Education, 11, 436–450. doi:10.1037/dhe0000068

19.

Gloria

A. M.

Castellanos

Orozco

(2005). Perceived educational barriers, cultural fit, coping responses, and psychological well-being of Latina undergraduates. Hispanic Journal of Behavioral Sciences, 27, 161–183. doi:10.1177/0739986305275097

20.

Gloria

A. M.

T. A.

(2003). Environmental, social, and psychological experiences of Asian American undergraduates: Examining issues of academic persistence. Journal of Counseling & Development, 81, 93–105.

21.

Gloria

A. M.

Kurpius

S. E. R.

(1996). The validation of the cultural congruity scale and the university environment scale with Chicano/a students. Hispanic Journal of Behavioral Sciences, 18, 533–549. doi:10.1177/07399863960184007

22.

Hayton

J. C.

Allen

D. G.

Scarpello

(2004). Factor retention decisions in exploratory factor analysis: A tutorial on parallel analysis. Organizational Research Methods, 7, 191–205. doi:10.1177/1094428104263675

23.

Hoffman

Richmond

Morrow

Salomone

(2002). Investigating “sense of belonging” in first-year college students. Journal of College Student Retention: Research, Theory & Practice, 4, 227–256. doi:10.2190/DRYC-CXQ9-JQ8V-HT4V

24.

Hurtado

Griffin

K. A.

Arellano

Cuellar

(2008). Assessing the value of climate assessments: Progress and future directions. Journal of Diversity in Higher Education, 1, 204–221. doi:10.1037/a0014009

25.

Jehangir

Williams

Jeske

(2012). The influence of multicultural learning communities on the intrapersonal development of first-generation college students. Journal of College Student Development, 53, 267–284. doi:10.1353/csd.2012.0035

26.

Jehangir

Williams

Pete

(2011). Multicultural learning communities: Vehicles for developing self-authorship in first-generation college students. Journal of the First-Year Experience & Students in Transition, 23, 53–73.

27.

Kline

R. B.

(2016). Principles and practice of structural equation modeling (4th ed.). Guilford Press.

28.

Langhout

R. D.

Rosselli

Feinstein

(2007). Assessing classism in academic settings. The Review of Higher Education, 30, 145–184. doi:10.1353/rhe.2006.0073

29.

Lee

B. H.

Zhu

Diaz

Fischer

Flores

L. Y.

Lin

… Atilano

(2017). Racial/ethnic minority vocational research trends: An 11-year update. The Career Development Quarterly, 65, 288–301. doi:10.1002/cdq.12108

30.

Lent

R. W.

Brown

S. D.

Hackett

(1994). Toward a unifying social cognitive theory of career and academic interest, choice, and performance. Journal of Vocational Behavior, 45, 79–122. doi:10.1006/jvbe.1994.1027

31.

Lent

R. W.

Brown

S. D.

Schmidt

Brenner

Lyons

Treistman

(2003). Relation of contextual supports and barriers to choice behavior in engineering majors: Test of alternative social cognitive models. Journal of Counseling Psychology, 50, 458–465. doi:10.1037/0022-0167.50.4.458

32.

Lent

R. W.

Brown

S. D.

(2019). Social cognitive career theory at 25: Empirical status of the interest, choice, and performance models. Journal of Vocational Behavior, 115. doi:10.1016/j.jvb.2019.06.004

33.

Lent

R. W.

Singley

Sheu

H.-B.

Gainor

K. A.

Brenner

B. R.

Treistman

Ades

(2005). Social cognitive predictors of domain and life satisfaction: Exploring the theoretical precursors of subjective well-being. Journal of Counseling Psychology, 52, 429–442. doi:10.1037/0022-0167.52.3.429

34.

McNeish

(2017). Thanks coefficient alpha, we’ll take it from here. Psychological Methods, 23, 412–433. doi:10.1037/met0000144

35.

Muthén

L. K.

Muthén

B. O.

(1998–2017). Mplus user’s guide (8th ed.). Muthén & Muthén.

36.

Patton

L. D.

Museus

S. D.

(2019, July 19). Identifying and disrupting deficit thinking [Blog post]. Retrieved from https://medium.com/national-center-for-institutional-diversity/identifying-and-disrupting-deficit-thinking-cbc6da326995

37.

Peng

C.-Y. J.

Harwell

Liou

S.-M.

Ehman

L. H.

(2006). Advances in missing data methods and implications for educational research. In Sawilowsky

(Ed.), Real data analysis (pp. 31–78). Information Age.

38.

Sandoval-Lucero

Maes

Klingsmith

(2014). African American and Latina(o) community college students’ social capital and student success. College Student Journal, 12, 522–533.

39.

Sax

L. J.

Gilmartin

S. K.

Bryant

A. N.

(2003). Assessing response rates and nonresponse bias in web and paper surveys. Research in Higher Education, 44, 409–432.

40.

Schlomer

G. L.

Bauman

Card

N. A.

(2010). Best practices for missing data management in counseling psychology. Journal of Counseling Psychology, 57, 1–10. doi:10.1037/a0018082

41.

Seng Kam

C. C.

Meyer

J. P.

(2015). Implications of item keying and item valence for the investigation of construct dimensionality. Multivariate Behavioral Research, 50, 457–469. doi:10.1080/00273171.2015.1022640

42.

Solberg

V. S.

O’Brien

K. M.

Villareal

Kennel

Davis

(1993). Self-efficacy and Hispanic college students: Validation of the college self-efficacy instrument. Hispanic Journal of Behavioral Sciences, 15, 80–95. doi:10.1177/07399863930151004

43.

Stephens

N. M.

Fryberg

S. A.

Markus

H. R.

Johnson

C. S.

Covarrubias

(2012). Unseen disadvantage: How American universities’ focus on independence undermines the academic performance of first-generation college students. Journal of Personality and Social Psychology, 102, 1178–1197. doi:10.1037/a0027143

44.

Tracey

T. J. G.

(2012). Problems with single interest scales: Implications of the general factor. Journal of Vocational Behavior, 81, 378–384. doi:10.1016/j.jvb.2012.10.001

45.

U. S. Census Bureau. (2018, December). Current population survey, school enrollment supplement, October 2007 and 2017. Retrieved from www.census.gov/programs-surveys/cps.html

46.

Wells

A. V.

Horn

(2015). The Asian American college experience at a diverse institution: Campus climate as a predictor of sense of belonging. Journal of Student Affairs Research and Practice, 52, 149–163. doi:10.1080/19496591.2015.1041867

47.

Weston

Gore

P. A.

Jr (2006). A brief guide to structural equation modeling. The Counseling Psychologist, 34, 719–751. doi:10.1177/0011000006286345

48.

Winkle-Wagner

(2009). The perpetual homelessness of college experiences: Tensions between home and campus for African American women. The Review of Higher Education, 33, 1–36. doi:10.1353/rhe.0.0116

49.

Worthington

R. L.

Whittaker

T. A.

(2006). Scale development research: A content analysis and recommendations for best practices. The Counseling Psychologist, 34, 806–838. doi:10.1177/0011000006288127

50.

Yosso

T. J.

(2005). Whose culture has capital? A critical race theory discussion of community cultural wealth. Race, Ethnicity, and Education, 8, 69–91. doi:10.1080/1361332052000341006