Critical Consciousness and Vocational Development Among Latina/o High School Youth

Critical Consciousness Measurement

In spite of interest in the construct of critical consciousness that spans several decades, only very recently have independent efforts to develop specific measures of adolescent critical consciousness been introduced (Diemer, 2014; Diemer, Rapa, Park, & Perry, 2014). The benefits of a valid and reliable measure include the opportunity to further investigate correlates of critical consciousness, to assess the effects of interventions designed to enhance critical consciousness, to connect different dimensions of critical consciousness (e.g., agency vs. behavior) with different outcomes, and to increase comparability across studies. In order to be most useful, such a measure should be easy to administer, brief, sensitive to change, and domain-specific such that it captures the relevant dimensions of critical consciousness for the target population. Encouraged by the growing body of evidence that critical consciousness may benefit youth career development, and consistent with a call for strength-based orientations in research addressing Latina/Latino youth (Acevedo-Polakovich et al., 2013), we set out to develop a measure of critical consciousness in the context of a pilot intervention project with Latina/Latino youth. The purpose of the present study is to describe the development and initial validity testing of this measure using exploratory (Study 1) and confirmatory (Study 2) factor analyses as well as testing associations between critical consciousness dimensions and relevant vocational development variables.

We began in 2009 with a set of 10 items that we developed to assess critical consciousness among participants in a pilot after school intervention we developed and conducted for Spanish speaking Latino/Latina high school students (McWhirter et al., 2010). The high school at which the intervention was conducted was designated by the Oregon Department of Education as “Needing Improvement” in meeting state education standards (Oregon Department of Education, 2009). The school served a predominantly White low- and middle-income student population, with a growing number of Latina/Latino students who were also recent immigrants, some of whom were undocumented. The high school drop-out rate in Oregon was then, and continues to be, quite high, particularly among Latina/Latino students and English Language Learners (Oregon Department of Education, 2014). Our aim was to promote the educational persistence of Spanish-speaking Latina/Latino students through provision of academic and vocational support and through enhancing participants’ advocacy skills and critical consciousness.

Critical consciousness items were developed after an extensive review of the literature on Latinas/Latinos in the education system, barriers they encounter, and the available literature on critical consciousness. Items were intended to reflect the components of critical consciousness development described by Diemer and Hsieh (2008). Thus, items focused on participants’ awareness of racism and discrimination in general and in their own lives, the importance of and motivation for helping their community and ending racism, and participation in discussions with family and in community or school groups seeking to end discrimination or promote equality. Our focus was on racism and discrimination, as we expected this to be the most salient aspect of inequity experienced by our Latina/Latino adolescent participants. A team of four doctoral students, of whom two were bilingual and bicultural Latinas and two were Spanish fluent White females, reviewed the items for face validity and comprehension. The small sample size of our program participants (n = 19) precluded evaluation of the psychometric properties of this pilot measure. Prior to initiating Study 1, we made a variety of item modifications and additions. For example, we added more items and we altered wording of other items to reduce item similarities and improve item clarity. This resulted in the set of 17 items administered in Study 1.

Study 1

In this first study, our aims were to explore the factor structure of our 17-item measure of adolescent critical consciousness (MACC) in a sample of Latina/Latino high school youth and to test whether higher critical consciousness would be associated with plans to obtain postsecondary education and higher vocational outcome expectations (VOE). Prior findings of Diemer (2009), Diemer and Blustein (2006), and Diemer et al. (2010) showed positive associations between urban youth’s consciousness of and motivation to reduce structural inequities and their educational attainment, as well as commitment to their future career, work role salience, and VOE. Thus, our hypotheses were that critical consciousness (1) would distinguish between Latina/Latino students planning to attend college and those who did not plan to attend college and (2) would be positively associated with VOE. The lack of validated measures of adolescent critical consciousness has limited research in this area, and extant work has not tested or identified evidence of gender differences in adolescent critical consciousness of racial inequality. Thus, our third hypothesis was that we would not find gender differences in critical consciousness.

Method

Results

Preliminary analyses examined variable distributions and potential biases associated with missing data. Skewness and kurtosis values were acceptable (Curran, West, & Finch, 1996). Rates of missing data ranged from 8% to 12%. Most participants (74%) completed all (17) items. Those that completed all items were compared to those that did not (26%) on study demographic characteristics. No significant differences were found with the exception of gender, χ²(1,475) = 5.57, p = .018; females were more likely to complete all items. To maximize statistical power and reduce bias, a single imputed data set was created using the IVEware program (Raghunathan, Solenberger, & Van Hoewyk, 2002). Although it is not possible to know for sure that data are missing at random (MAR), the MAR assumption can be made more tenable as the imputation model is made more general. The inclusion of additional predictors in the imputation model can reduce bias and make the MAR assumption more plausible (Allison, 2009; He, Zaslavsky, Landrum, Harrington, & Catalano, 2010; Rubin, 1996). Therefore, the imputation model included auxiliary variables (e.g., age and gender) in addition to the VOE and MACC items.

Exploratory Factor Analyses (EFAs)

The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy (.85, p < .001) and Barlett’s test of sphericity (p < .001) indicated that the sample was suitable for factor analysis (Hair, Anderson, Tatham, & Black, 1995). An EFA of the items was conducted, using principal axis factoring with an oblimin rotation, given that factors were expected to correlate. To determine the optimal number of factors, we utilized the criteria of Eigenvalues > 1, examination of the scree plot, interpretability of the factors, and parsimony. Communalities ranged from .07 to .71. An initial five-factor solution explained 52.6% of the variance, but yielded a factor with a single item. The scree plot suggested two-, three-, or four-factor models were plausible. We compared pattern matrices of two-, three-, or four--factor models. In all three models, there were 4 items with pattern coefficients <.32. The four-factor model also contained a single-item factor, and the four- and three-factor models each had 2 cross-loading items. The two-factor model had no cross-loading items, thus the two-factor model was deemed most parsimonious and retained. Next, items were removed one at a time, starting with the lowest pattern coefficient, until all items loaded at .4 or higher on one factor. This resulted in 8- and 2-item factors, respectively, with pattern coefficients ranging from .47 to .71 (see Table 1). Factor 1, with 8 items, was labeled Critical Agency (α = .80). Factor 2 was labeled Critical Behavior. With only 2 items, the Spearman–Brown prophesy formula was used to calculate internal consistency (Eisinga, Te Grotenhuis, & Pelzer, 2013), yielding a coefficient of .61.

OPEN IN VIEWER

Table 1.

Final Pattern Coefficients for Exploratory Factor Analysis of 17 Critical Consciousness Items.

Item	Critical Agency Factor	Critical Behavior Factor
17. In the future, I will participate in activities and groups against racism and discrimination	.67
16. It is important to me to contribute to my community	.66
13. More effort is needed to end racism and discrimination	.63
15. In the future, I will participate in activities and groups to promote equality and justice	.62
10. It is important to fight against social and economic inequality	.62
11. I can make a difference in my community	.61
5. There are ways that I can contribute to my community	.52
7. I am motivated to end racism and discrimination	.48
12. I am involved in activities or groups that promote equality and justice		.84
4. I am involved in activities or groups against racism and discrimination		.48
Mean	3.30	2.63
SD	.49	.79
Internal consistency	.80	.61

Note. Item range 1–4. Study 1 N = 558.

Discussion of Study 1

The Study 1 EFA resulted in a 10-item, two-factor structure of Critical Consciousness. This structure differs from Watts et al.’s (2011) suggestion of awareness, agency, and behavior as three distinct components of critical consciousness, in that we did not find a distinct “awareness of inequity” factor. Two items we conceptualized as indicators of awareness were included in the Critical Agency factor. As expected, Critical Agency was highest for those planning on attending 4-year postsecondary institutions. Critical Behavior distinguished between those with 4-year and 2-year college plans. Both Critical Agency and Critical Behavior were significantly and positively correlated with VOE, though the magnitude of the relationship was quite low for Critical Behavior. The Latina girls had higher Critical Agency scores than the Latino boys. Perhaps this is a function of gender role socialization for contributing to the welfare of others or that Latina girls’ experience of sexism enhances their agency and motivation for addressing inequity relative to that of boys. The data do not allow exploration of these possibilities.

The Critical Behavior factor included only 2 items and had a relatively low internal consistency. A factor with fewer than 3 items is generally weak and unstable (Little, Lindenberger, & Nesselroade, 1999; Velicer & Fava, 1998), thus we developed an additional Critical Behavior item. In light of state and national attention to issues of immigration and higher education access to undocumented students, the new item focused on participation in demonstrations or signing petitions regarding justice issues.

Study 2

Results

Preliminary analyses examined variable distributions and potential biases associated with missing data. Skewness and kurtosis values were acceptable (Curran et al., 1996). Rates of missing data ranged from 2% to 4%. Most participants (91%) completed all 11 items. Those who completed all items were compared to those who did not (9%) on study demographic characteristics. No significant differences were found with the exception of grades, χ²(5,849) = 12.10, p = .033; lower grades were associated with missing responses. Following the same logic and procedures described in Study 1, one imputed data set was created.

The hypothesized two-factor (Critical Agency and Critical Behavior) structure produced by the Study 1 EFA, plus the additional item hypothesized to load on the Critical Behavior factor, was tested for stability with the 680 participants that completed the English version of the assessment. Model specification included scaling of the latent variables by fixing one observed measure per factor to 1.0 and allowing the factors to covary; no cross-loadings of items or correlated item residuals were permitted. The resulting overidentified model with 34 degrees of freedom was fit to the data with the Mplus software (version 7.11, Muthén & Muthén, 1998–2012) and estimated with the maximum likelihood function.

Evaluation of the CFA model included assessment of overall goodness of fit, localized areas of misfit in the solution, and interpretability, size, and statistical significance of the model’s parameters. Model fit was based on recommended cut-off values by Hu and Bentler (1999); confirmatory fit index (CFI) > .95, root mean square error of approximation (RMSEA) < .06, and standardized root mean residual (SRMR) < .08. Standardized residuals and modification indices were used to help identify areas of model misfit. Given the large sample size, scrutiny of modification indices relied on the expected parameter change (EPC) values rather than statistical significance. Model respecification was undertaken to improve parsimony and interpretability of the model. We compared one-factor and two-factor structures. Lastly, the model structure was tested on data from the 190 participants who completed the measure in Spanish, to evaluate how well the factor structure was maintained with the Spanish translation.

The 11-item two-factor model showed marginal fit to the data, and the single-factor model provided a poor fit (see Table 2). Examination of the modification indices for the two-factor model indicated that Item 8 had a cross-loading with the behavior factor and had the largest standardized EPC and the largest set of standardized residuals across all items. Therefore, Item 8 was dropped and the model was reestimated. The fit indices improved to an acceptable level and the decrease in the sample size adjusted Bayesian Information Criterion (14349 − 13219 = 1130) was substantial. Additionally, the parameter estimates make statistical (e.g., no Heywood cases) and substantive sense (e.g., moderate to large correlation between the Critical Behavior and Critical Agency factors).

OPEN IN VIEWER

Table 2.

Summary of Model Fit.

Model	χ2(df), p value	CFI	RMSEA	SRMR
English: One-factor 11 items	517.05 (44), <.001	.845	.126	.078
English: Two-factor 11 items	245.86 (43), <.001	.934	.083	.041
English: Two-factor 10 items	128.01 (34), <.001	.962	.064	.032
Spanish: Two-factor 10 items	61.05 (34), .003	.978	.065	.039

Note. CFI = confirmatory fit index; RMSEA = root mean square error of approximation; SRMS = standardized root mean residual. Study 2 English subsample: n = 680; Spanish subsample: n = 190.

Table 3 contains the standardized regression weights, standard errors, and factor correlations associated with the final two-factor CFA model. All parameter estimates are statistically significant at p < .001, positive, and large in magnitude, indicating each variable is related to the latent constructs with which it is associated.

OPEN IN VIEWER

Table 3.

Final Confirmatory Factor Analysis Results for English and Spanish Measure.

		English n = 680		Spanish n = 190
		Estimate	SE	Estimate	SE
	Critical Agency
2.	There are ways that I can contribute to my community	.62	.026	.82	.026
3.	I am motivated to try to end racism and discrimination	.69	.023	.84	.024
4.	It is important to fight against social and economic inequality	.68	.024	.86	.021
5.	I can make a difference in my community	.71	.022	.89	.017
7.	More effort is needed to end racism and discrimination	.64	.025	.82	.026
9.	It is important to me to contribute to my community	.79	.018	.90	.016
10.	In the future, I will participate in activities or groups that struggle against racism and discrimination	.80	.018	.75	.034
	Critical Behavior
1.	I am involved in activities or groups against racism and discrimination	.68	.032	.39	.075
6.	I am involved in activities or groups that promote equality and justice	.74	.032	.76	.068
11.	I have participated in demonstrations or signed petitions about justice issuesa	.59	.034	.69	.068

Note: Two-factor Critical Consciousness model with standardized parameter estimates; factor correlations .43 and .50 for English and Spanish measures, respectively. All estimates are statistically significant at p < .001. Cronbach’s α for Critical Agency = .89, Critical Behavior = .69.

^aItem added for Study 2.

The final model was then imposed on the sample that completed the survey in Spanish. The fit of the model was comparable to the English version, and like the English version, all parameter estimates were statistically significant and positive. The magnitude of the relationships between the English and Spanish versions differ somewhat in that the Spanish version showed stronger correlations between factors and items, with the exception of Item 1 (see Table 3).

Follow-Up Analyses

Next, we conducted a series of analyses to test the construct validity of the measure. First, we conducted a one-way analysis of variance (ANOVA) to determine whether levels of critical consciousness differed as a function of measure language. There were no language differences for Critical Agency, F(1, 868) = .054, p = .82, or Critical Behavior, F(1, 868) = 1.93, p = .17. Based on this finding, in conjunction with our finding of the same factor structure of Critical Agency and Critical Behavior in the Spanish language data, we combined the English and Spanish data for all subsequent analyses.

We conducted a MANOVA with Critical Agency and Critical Behavior as the DVs and gender (male and female), and language (English and Spanish) as the independent factors. Box’s M was significant at p < .001, thus Pillai’s Trace was used as the multivariate effect indicator. There was no interaction effect, F(2, 865) = 1.45, p = .24, η² = .003, nor were there main effects for gender, F(2, 865) = 3.39, p = .03, or language, F(2, 865) = 1.58, p = .21. Next, we conducted a multivariate analysis of covariance (MANCOVA) to test whether Critical Agency and Critical Behavior differed as a function of postsecondary education plans (no school plan, 2-year degree, or 4-year degree), controlling for the effect of grades. Box’s M was not significant. There was a main effect for postsecondary plans (Wilks’s Λ = .966, F(4, 1678) 7.39, p < .001, η² = .017, and the covariate of grades had a significant effect on the combined DV (Wilks’s Λ = .986, F(2, 838) = 6.02, p = .003, η² = .014). Univariate ANOVA results indicate that only the DV of Critical Agency was significantly associated with postsecondary school plans, F(2, 843) = 11.72, p < .001, η² = .027, and the covariate of grades, F(1, 843) = 12.02, p = .001, η² = .014. Comparison of adjusted means indicate significant differences in Critical Agency for each level of postsecondary schooling plans, adjusting for grades, with the lowest Critical Agency for those not planning to pursue postsecondary education immediately after high school, and the highest for those planning to enroll in 4-year colleges. Critical Behavior did not differ as a function of postsecondary plans after adjusting for grades (Table 4).

OPEN IN VIEWER

Table 4.

Adjusted Means and Standard Deviations of Critical Agency and Critical Behavior by Postsecondary Plans, Adjusted for Grades.

	Postsecondary Plan	Mean	SD	n
Critical Agencya	No school	3.04	.70	106
	2-Year college	3.28	.58	379
	4-Year college	3.38	.56	358
	Total	3.29	.60	843
Critical Behavior	No school	2.59	.68	106
	2-Year college	2.54	.69	379
	4-Year college	2.67	.68	358
	Total	2.60	.69	843

Note. Range 1–4. Participants with postsecondary plans that did not include those listed here (n = 27) were excluded from this analysis.

^aSignificant differences at each level of postsecondary plans for Critical Agency after adjusting for grades.

Next, we classified participants as high, medium, or low Critical Agency and Critical Behavior on the basis of scores in the upper, middle, and lower third of means for each subscale. We compared participants high in both Critical Agency and Behavior (high CC; n = 127) to participants low in both Agency and Behavior (low CC; n = 118) with respect to grades (Mostly As and Bs vs. Mostly Cs and Ds) and conference attendance (first time vs. second, third, or fourth time). Box’s M was not significant at p < .001, thus Wilks’s Λ was used as the multivariate effect indicator. There was a significant multivariate effect, Wilks’s Λ = .81, F(2, 242) = 14.10, p < .001, η² = .104. Examination of between-subjects effects and cell means indicated that, as expected, high CC participants had higher grades, F(1, 245) = 13.96, p < .001, η² = .054, and had attended the conference more often, F(1, 245) = 13.7, p < .001, η² = .053, than low CC participants.

We next examined differences in engagement between the high CC and low CC groups. School engagement, extracurricular engagement, helping engagement, and engagement in Spanish served as the four DVs. Box’s M was not significant at p < .001, thus Wilks’s Λ was used as the multivariate effect indicator. There was a significant multivariate effect, Wilks’s Λ = .81, F(4, 240) = 13.88, p < .001, η² = .19. Examination of between-subjects effects and cell means indicated that high CC participants had higher School Engagement, F(1, 245) = 20.67, p < .001, η² = .078, Extracurricular Engagement, F(1, 245) = 43.62, p < .001, η² = .133, Spanish Engagement, F(1, 245) = 9.15, p = .003, η² = .036, and Helping Engagement, F(1, 245) = 37.36, p < .001, η² = .133, than low CC participants. Finally, we conducted Pearson product moment correlations and found significant but very small negative correlations between Critical Agency and past thoughts of dropping out (r = −.13, p < .001) and present thoughts of dropping out (r = −.15, p < .001). Critical Behavior was not correlated with either dropout item.

General Discussion

This study contributes to the growing literature base on the potential value of critical consciousness for Latina/Latino youth educational persistence and vocational development. We present initial validity data for a new measure of critical consciousness, and evidence of a relationship between the two dimensions of critical consciousness (Critical Agency and Critical Behavior) and specific educational and vocational outcomes. This measure was specifically developed for and tested with Latina/Latino high school students. The combined results of both studies support a two-factor model. Critical Agency combines commitment to and efficacy for taking action against racism and discrimination. It is akin to Watts and Guessous’s (2006) sense of agency and commitment to make change. Critical Behavior reflects the action component of critical consciousness and can be described as actions to promote justice and end racism.

Our two-factor structure differs from the awareness, efficacy, and action components of critical consciousness proposed by Watts, Diemer, and Voight (2011). Rather than a distinct “awareness of inequity” factor, Critical Agency includes 2 awareness items. It may be that for this group of youth, agency to address racism and discrimination only manifests in conjunction with recognition of racism and inequity. It may also be that we simply did not develop appropriate items for capturing critical awareness.

Critical Agency was strongly and favorably associated with postsecondary plans to attend 4-year colleges in both Studies 1 and 2. In Study 2, Critical Agency was significantly different for each postsecondary pathway, and this association remained even after adjusting for the influence of grades. These results are consistent with our hypotheses, and with prior findings that indicators of critical consciousness were associated with urban minority adolescents’ work role salience, career commitment, and positive vocational expectations (Diemer & Blustein, 2006; Diemer et al., 2010). Contrary to our expectations, although Critical Behavior scores differed between those with 2-year and 4-year college plans in Study 1, it was not significantly associated with postsecondary plans in Study 2 after controlling for grades. Further, in Study 1, Critical Behavior scores did not differ between those with no postsecondary education plans and those with 4-year college plans. As such, Study 2 differences in postsecondary plans in the high versus low CC groups were likely due to the effects of Critical Agency alone.

Perhaps participation in demonstrations and in groups promoting equity and reducing racism among high school students is more subject to practical constraints (e.g., transportation and home responsibilities) and thus less connected to outcomes such as postsecondary plans. In prior research, perceived support (from peers, family, schools, and community) for challenging racism, social injustice, and sexism among urban ethnic minority adolescents was associated with higher critical awareness of inequity, but not with higher levels of action to address inequity (Diemer, Kauffman, Koenig, Trahan, & Hsieh, 2006). These authors suggested that different school contexts might operate to support or constrain sociopolitical action. Future research might explore potential home and school constraints on behavioral manifestations of critical consciousness among Latina/Latino adolescents, as well as differential associations between these 2 subscales and educational and vocational development outcomes.

Critical Agency and Critical Behavior did not vary by gender or by the language in which the survey was completed in Study 2. Attention to gender differences should continue in future studies of Latina/Latino adolescent critical consciousness, particularly, given increasing gender differences in educational expectations, college enrollment, and the differential effects of social capitol on expectations (Wells, Seifert, Padgett, Park, & Umbach, 2011).

Our comparison of the highest and lowest Critical Agency/Behavior groups showed higher critical consciousness was associated with attending the conference more than once. The conference is intended to enhance leadership, instill pride, and highlight access to resources for college and careers. It is possible that conference attendance builds critical consciousness; it is also possible that students higher in critical consciousness seek to attend the conference more often. The present data do not allow for exploring these possibilities. We also found that higher critical consciousness was associated with positive school outcomes such as grades, school engagement, and extracurricular engagement. Our findings contrast with those of Fine (1991) and Conchas (2001), who suggested that critical consciousness of structural inequities may reduce motivation and decrease engagement with school and work. Our participants with a stronger commitment to addressing inequity and higher participation in justice efforts also appear to be more invested in doing well in school and participate more in extracurricular school and community activities. These key types of engagement reduce the likelihood of dropping out of high school (Davalos, Chavez, & Guardiola, 1999; Ream & Rumberger, 2008), and as such may be more promising correlates of critical consciousness than our items assessing the past and present thoughts about dropping out. Fine (1991) and Conchas (2001) focused on critical awareness of inequity rather than critical agency to address inequity, which may explain the differences in our findings. It is possible that a lack of variability in responses to our 2 dropout items artificially suppressed the relationship between critical consciousness and dropout in this sample.

Students high in both critical consciousness subscales had higher engagement in speaking Spanish (speaking and understanding Spanish and translating for family members), and higher engagement in helping behaviors such as caring for younger children and helping others at school, than their low critical consciousness counterparts. The documented challenges of serving as family translators and responsibilities at home have been identified as barriers for Latina/Latino youth (Martinez, McClure, & Eddy, 2009) and yet also are activities that connect young people to their families and communities, provide important roles to play in others’ well-being, and have been linked with academic benefits (Buriel, Perez, De Ment, Chavez, & Moran, 1998; Villanueva & Buriel, 2010; Weisskirch, 2005). It may be that engagement in caring for their families and communities increases awareness of and motivation to challenge the larger ecology of inequity, while they carry out responsibilities and make contributions that enhance their agency. Awareness of parental sacrifices (e.g., immigration and working multiple jobs) for their children’s future opportunities has been described by Latina high school girls as a strong motivator for persisting in school (McWhirter et al., 2013).

Summary and Conclusion

Prior research has identified critical consciousness as a potentially powerful and transformative resource for Latina/Latino youth as they navigate structural barriers and pursue their educational and vocational pathways. The present study introduces a measure of critical consciousness designed for and validated with Latina/Latino high school youth and provides additional support for associations between critical consciousness and important indicators of Latina/Latino youth’s educational persistence and career development. A reliable and valid measure of critical consciousness for Latina/Latino youth is an important tool for continued research on this potential asset to Latina/Latino educational attainment and vocational development.