The Mediating Role of Home-School Dissonance Linking Maternal Discrimination to Latin American-origin Adolescent Academic Performance

Abstract

Latin American-origin parents play an important role in supporting the formal education of their youth, but cultural, linguistic and systemic barriers make parent involvement difficult. The aim of the present study was to examine how Latina mothers’ experiences with discrimination were associated with short-term changes in their adolescent children’s academic performance, directly and indirectly through mothers’ appraisal of home-school dissonance. Data were drawn from an ongoing longitudinal study of 547 mother-youth dyads in suburban Atlanta. Results showed a significant indirect relationship between mothers’ experiences of discrimination and declines in adolescents’ grade point average by way of increased home-school dissonance. Thus, Latina mothers’ discriminatory experiences may have spillover effects on adolescent academic achievement through increasing proximal barriers to parent involvement.

Although overall population growth among U.S. Latin American-origin population has slowed over the last decade, this group accounted for half (52%) of all population growth during this period and grew faster in the South than any other region (Noe-Bustamante, Lopez, & Krogstad, 2020). While births spurred most of this growth, immigration contributed 27.5% to the increase between 2010–2015 (Flores, 2017). These demographic changes have important implications for the context of schooling for Latin American-origin youth (Bauman, 2017). In many immigrant Latino families, parents are unlikely to speak English or be familiar with the U.S. education system. These linguistic and cultural barriers can elevate perceived dissonance between home and school, limiting parents’ involvement in their youth’s schooling. Experiences of discrimination may exacerbate this perceived dissonance even further, an important consideration given the increasingly anti-immigrant climate of recent years (Lopez, Gonzalez-Barrera, & Krogstad, 2018). Understanding how discrimination and perceived home-school dissonance influence Latin American-origin adolescents’ academic performance can help elucidate our understanding of the structural factors salient to parents’ formal and informal investments in their adolescents’ schooling.

The academic success of Latin American-origin students has long lagged behind that of White and Black students, and this disparity persists despite steady improvements in this population’s education outcomes over the past 20 years (de Brey et al., 2019). Understanding whether and how parents’ discriminatory experiences have spillover effects on Latin American-origin adolescents’ academic success is both timely and significant. In the present study, we draw on ecological and parent involvement frameworks (Bronfenbrenner, 1986; Hornby & Lafaele, 2011) to examine whether Latina mothers’ discrimination, a distal societal factor, may compromise adolescents’ academic performance by way of maternal perceptions of home-school dissonance, a proximal barrier to parent involvement in their child’s school. We utilize three time points of contemporary data for mother-youth dyads to test a prospective mediation model in which mothers’ perceived discrimination is related to short-term changes in adolescent grade point average (GPA) by way of home-school dissonance. It is especially valuable to examine these processes among adolescents. Although direct forms of parent school involvement (e.g., volunteering at school) decline as children progress through the adolescent years, direct parental school involvement has been shown to be more strongly related to student achievement for students in secondary, as compared to elementary, school (Kim & Hill, 2015). Importantly, our sample is drawn from Latino families living in suburban Atlanta, an important new immigrant destination where exponential growth in foreign-born residents occurred over the last three decades (Singer, 2015). When compared to more established immigrant enclaves, new immigrant areas tend to be characterized by fewer Spanish-language supports, and less Latino cultural inclusion in communities and schools (Perreira, Potochnick, & Brietzke, 2020). Thus, there is an especially strong need for research examining Latino-specific barriers to youth’s school performance in new immigrant areas.

Home-School Dissonance, Parental Involvement, and Latin American-origin Adolescent School Performance

By many accounts, the U.S. education system is firmly rooted in Eurocentrism. Schools are typically administered and staffed by White educators (de Brey et al., 2019;), values and norms tend to prioritize individualism over collectivism, and curricula and extra-curricula often center White culture and history. These characteristics may be particularly evident in new immigrant destinations such as suburban Atlanta, where only 5% of school teachers are of Latin American-origin (GOSA, 2020) and where the State Board of Education has recently opposed teaching of critical race theory (Office of the Governor, 2021). School-based strategies often aim to promote parents’ involvement in their youth’s schooling, a well-established correlate of better educational outcomes among children and youth, including for Latino families living in the U.S. (Benner, Boyle, & Sadler, 2016). Though research has traditionally classified parent involvement into school or home-based strategies, such as communicating with school staff or monitoring homework (Epstein, 1987), parents also engage in a number of informal strategies (i.e., academic socialization), such as emphasizing the importance of formal education or discussing future opportunities (Ceballo, Maurizi, Suarez et al., 2014). Standards of parent involvement set forth by national parent-teacher organizations and based on empirical research (Epstein et al., 2009) emphasize six home and school domains: communication, parenting, student learning, volunteering, school decision making and advocacy, and collaborating with community. These myriad forms of parental involvement are facilitated by parental beliefs and efficacy regarding parents having a role in supporting their child’s formal education; parent perceptions of schools and teachers welcoming parental school involvement; and fewer conflicting demands between the needs of home and school (Hoover-Dempsey & Sandler, 1997). This kind of involvement may be especially challenging for marginalized populations who are not familiar or comfortable with school-based expectations (Marks & García Coll, 2009). Although traditional Latino values such as educación (i.e., socializing children according to moral and behavioral standards) and familismo (i.e., valuing family closeness, interdependence, and duty) promote academic achievement in Latin American-origin adolescents, these values are often not consonant with those espoused by U.S. schools. For example, the value of educación, while seemingly aligned with educational goals, has been shown to clash with the school setting (Arellanes, Viramontez Anguaino, & Lohman, 2019). Collectivist values like familismo are also incongruent as teachers, regardless of ethnicity, tend to endorse more individualistic values (Aganza et al., 2015; Greenfield & Quiroz, 2013). Moreover, teachers perceive Latino values to be misaligned with academic goals (Aganza et al., 2015). Theoretically, these misalignments shape the experiences of non-white students and their families in the education setting (Ogbu & Simons, 1998).

Differences in parent and school-held cultural values, beliefs and norms —manifestations of home-school dissonance—may undermine parent involvement behaviors because of related misunderstandings, misperceptions, and mistreatment of Latin American-origin parents by the schools that serve their children (Arellanes, et al., 2019; Nelson & Guerra, 2014). For Latin American-origin parents, home-school dissonance makes it more difficult to communicate with teachers, trust schools, and feel welcome (Hill & Torres, 2010), especially when their children reach middle and high school. Compared to primary schools, middle and high schools are larger and more diverse, and parents may find it harder to navigate this new environment even as it is more likely for their children to encounter racial bias (Hill, 2015). Additionally, secondary schools are more complex and bureaucratic, requiring parents to communicate with multiple teachers who are unlikely to speak their home language or be familiar with their background culture. When faced with discrimination, Latin American-origin parents tend to feel less efficacious and find it more difficult to remain involved (Hill, Witherspoon, & Bartz, 2018).

Whether experienced by youth or parents, home-school dissonance appears to undermine adolescent academic performance. For example, in one study of secondary school students, researchers found that GPA declined when Latin American-origin adolescents attributed lack of parent involvement to an unwelcoming school climate (Alexander, Cox, Behnke, et al., 2017). In contrast, Latin American-origin adolescents have been found to perform better academically when they perceive parent ease of cross-cultural interactions with their school (Chun & Devall, 2019) and when mothers report a welcoming and positive school climate (Lin, Pomerantz, Zheng, & Robins, 2020). Some scholars posit that parents’ perception of a positive school climate may support parent school involvement during adolescence when youth are more likely to become disengaged from school and at the same time, discourage parent involvement out of a desire for more autonomy (Kim & Hill, 2015; Lin et al., 2020). When parents feel connected to their child’s school, they may be more likely to seek school resources to promote their children’s academic success, particularly during the transition to middle and high school when academic curricula become increasingly complex and demanding (Bhargava, Bámaca-Colbert, Witherspoon et al., 2017; Lin et al., 2020).

Latin American-origin Parent Discrimination and Adolescent Academic Performance

Ethnic discrimination refers to everyday incidences of unfair treatment or negative interpersonal experiences based on one’s ethnicity (Brondolo et al., 2005; Contrada et al., 2001). There is ample evidence of the deleterious effects of adolescent perceived discrimination on their adjustment across psychosocial, behavioral and academic domains (see Benner et al., 2018 for review), and a growing literature suggests that parental experiences with discrimination are also associated with adverse psychosocial outcomes among youth (e.g., Espinoza, Gonzales, & Fuligni, 2016). It remains unclear whether parent experiences with discrimination impact the academic performance of youth, but a relevant study on acculturative stress suggests so. Benner and Kim (2009) found that parental stress around being a perpetual foreigner explained the pathway linking perceived discrimination among Chinese American parents and declines in the educational attitudes of their adolescents. Similarly, in a two-wave study among Mexican American adolescents, mothers’ workplace discrimination was indirectly related to youth’s decreased educational attitudes by way of mothers’ psychological distress (Wheeler, Updegraff, & Crouter, 2015). The present study builds on past findings by examining the link between parent experiences of discrimination and adolescent academic performance.

Current Study

It is important to understand how U.S. societal factors such as discrimination may foster Latina mothers’ perceived dissonance between home and school and, in turn, affect adolescent academic performance. Focusing on maternal experiences of discrimination and perceptions of home-school dissonance is important. Not only do mothers play a more central role in the ethnic and cultural socialization of their adolescent children than fathers (Knight, Carlo, Mahrer et al., 2016) but school involvement of mothers has been shown to matter more for adolescent academic achievement than the school involvement of fathers (Kim & Hill, 2015). The present study used prospective data from Latin American-origin mother-adolescent dyads living in suburban Atlanta, an important emerging immigrant area, to address two aims. First, we sought to identify how mothers’ experiences of discrimination were associated with short-term changes in their adolescents’ GPA. Second, we investigated whether mothers’ perception of home-school dissonance mediated associations between maternal experiences of discrimination and changes in adolescents’ GPA. We hypothesized that time 1 (T1) maternal discrimination would be related to T1 to T3 decreases in GPA directly and indirectly via mothers’ T1 to T2 increased appraisal of home-school dissonance. To account for the fact that parents of higher socioeconomic status and with greater English-language abilities tend to be more involved in their child’s school and to have children who experience better academic performance (Altschul, 2012; Bhargava et al., 2017; Stacer & Perrucci, 2013), our models controlled for maternal education and tested for structural pathway differences by mothers’ primary language. We adjusted for adolescent age and sex given research suggesting that academic success declines with age and is poorer for adolescent males compared to females (Fite, Rubens, & Cooley, 2014; Piña-Watson et al., 2018; Toyokawa & Toyokawa, 2019). We also controlled for adolescent generational status since second and third-/later- generation immigrant youth may have lower academic success compared to first generation youth (May & Whiterspoon, 2019). Additionally, we controlled for household structure given research showing that Latin American-origin adolescents living in single or stepparent homes may have lower odds of school success compared to those in two parent households (LeFevre & Shaw, 2012). Finally, since school ethnic composition may have important implications for parents’ perception of home-school dissonance and involvement (Hill & Tyson, 2010) and students’ academic success (Morales-Chicas & Graham, 2017), we accounted for whether or not the adolescent attended a school with a low, moderate, or high percent of Latin American-origin students.

Our study advances current scholarship in a number of ways. First, our use of three time-points of data limits biased estimates resulting from cross-sectional and half-longitudinal mediation models (Cole & Maxwell, 2003; Maxwell & Cole, 2007). Second, although there is evidence for the negative impact of adolescent perceived home-school dissonance on academic outcomes (e.g., Tyler, Burris, & Coleman, 2018), we draw on Latin American-origin parent involvement literature (Hill & Torres, 2010) to address whether parent perceived home-school dissonance similarly predicts academic outcomes. Finally, by including a sample of families living in a new immigrant area, this research advances knowledge about educational outcomes in communities and schools unaccustomed to the cultural and language orientations of Latino immigrant families.

Methods

Sample and Procedure

Data for this study derive from Caminos al Bienestar, an ongoing longitudinal study of 547 parent-adolescent dyads living in suburban Atlanta. Using 2017–2018 enrollment lists from 14 middle schools in a large socioeconomically diverse school district, students identified as “Hispanic” were randomly sampled within sex and grade strata and evenly across schools with “low” (< 13%), “moderate” (18% – 25%), and “high” (>40%) Latin American-origin student concentration to ensure heterogeneity in family and school characteristics. Adolescent exclusion criteria included not self-identifying as “Hispanic” (or a related term), having an individualized education plan (IEP), not being able to read and write in Spanish or English, or having a sibling already recruited into the study. The response rate among eligible adolescents whose parents were contacted and provided permission was 65.2% (547/839), and the response rate among eligible adolescents who were contacted was 95.3% (547/574). An intentionally missing study design was implemented as a cost-effective way to collect mother data while ensuring data quality (Little & Rhemtulla, 2013). For 50% of adolescents, a mother or mother-like figure was randomly selected for inclusion and was eligible to participate if they self-identified as Latin American-origin and spoke Spanish or English. The response rate among eligible mothers was 83.9% (271/386). Baseline data collection for 77.1% of adolescents (the “main” sample) occurred in schools from February to May 2018, with absent students completing the survey online using a url link provided to them at school. T1 data collection unexpectedly stopped after May 2018 when the school district requested an end to in-school data collection. Therefore, the study transitioned to online data collection, and the remaining 22.9% of adolescents (the “lagged” sample) completed their baseline surveys online from September 2018 – January 2019. Further details on sampling and recruiting procedures are described elsewhere (Roche et al., 2020).

At T1, the majority were in a 6^th – 8^th grade middle school (6^th: 26%; 7^th: 33.1%; 8^th: 31.6%). Due to the lagged sample beginning their T1 survey later than the main sample, a small percentage of students were in 9^th grade at T1 (9%). The sample was diverse with respect to socioeconomic status (as assessed by maternal education) and national origin. Among adolescents, 10.4% were first-generation immigrants (mother and adolescent born outside the U.S.), 65.3% were second-generation immigrants (adolescent born outside the U.S. but having at least one parent who was foreign-born), and 24.3% were third or later generation immigrants (both youth and parents born in the U.S.). Just over half of mothers were born in Mexico (52.3%), with approximately 12% from a country in Central America (11.7%) or in South America (12.4%), and 8.8% from another country. Additional sample characteristics are shown in Table 1. About half of student participants were male (44.6%), and on average, students were 12.78 years old at T1. The majority of youth lived in two parent households (67.3%) and were second-generation immigrants (69.3%), and the majority of mothers spoke Spanish as a primary language (69.8%). There was also diversity in terms of mothers’ educational attainment. A quarter (25.2%) had completed high school, and slightly over one-third (34.4%) had at least some college experience, with 17.4% having completed college.

Table 1.

Baseline Sample Characteristics (N = 547)

Demographic Variable	n (%) or M (SD)	Range
Adolescent gender
Female	303 (55.4%)
Male	244 (44.6%)
Adolescent generation status
First	57 (10.4%)
Second	379 (69.3%)
Third	111 (20.3%)
Mother’s primary language
English	165 (30.2%)
Spanish	382 (69.8%)
Household structure
Two parents	368 (67.3%)
Stepparent	74 (13.5%)
One or no parent	105 (19.2%)
School Latino Concentration
Low	153 (28%)
Moderate	186 (34%)
High	208 (38%)
Adolescent age in years	12.78 (1.03)	11–16
Mother’s education	2.99 (1.61)	1–6

Data was collected at three time points, spaced approximately six months apart, from 2018 to 2019. At each time point, adolescent surveys were self-administered on a computer, tablet, or mobile phone using the Qualtrics XM Research Core Survey Software (2018) program, and mother surveys were administered over the telephone by trained bilingual, bicultural staff. Parents provided oral or written permission and adolescents provided written assent. For mother surveys, mothers gave oral consent to participate. Investigators obtained a Certificate of Confidentiality from the National Institutes of Health and institutional review board approval. All study materials were provided in Spanish and English using the double-translation and double back-translation method combined with a review team approach (Knight, Roosa, & Umaña-Taylor, 2009).

Measures

Ethnic discrimination.

The Brief Perceived Ethnic Discrimination Questionnaire-Community Version (Brondolo et al., 2005) was used to assess ethnic discrimination. Mothers responded to 15 items indicating their experiences with being marginalized, treated unfairly or threatened because of being Latina (e.g., “others made you feel like an outsider because of dress, speech or other characteristics related to being Latina”). Response categories ranged from 1 = almost never or never to 5 = almost always or always, and scores ranged from 1.00 to 3.90 (T1: M = 1.46, SD = .44; T2: M = 1.38, SD = .44; T3: M = 1.23, SD = .29). Items were translated into Spanish by the study team. The scale demonstrated high reliability in English (T1: α = 0.85; T2: α = 0.88; T3: α = 0.84) and Spanish (T1: α = 0.84; T2: α = 0.94; T3: α = 0.97).

Home-school dissonance.

The six-item home-school dissonance subscale from Patterns of Adaptive Learning Scales (PALS; Midgley et al., 1998) was modified by the research team to assess mothers’, rather than students’, perceptions of incongruent home-school values. For example, the item “I don’t like to have my parents come to school because their ideas are very different from my teachers’ ideas” was modified to “my ideas about what is important for my child at school is different from teachers’ ideas.” Response options included 1 = not true, 2 = somewhat true, and 3 = true, and mean scores ranged from 1 to 3 across all time points (T1: M = 1.62, SD = .45; T2: M = 1.58, SD = .42; T3: M = 1.52, SD = .43), with higher scores indicating greater home-school dissonance. Items were translated into Spanish by the study team and demonstrated high reliability in both English (α = 0.77, T1; α = .77, T2; α = 0.84, T3) and Spanish (α = 0.80, T1; α = 0.88, T2; α = 0.91, T3).

GPA.

At each time point, adolescents self-reported their previous semester grades in math, science, history, and English (0 = F; 1 = D; 2 = C; 3 = B; 4 = A). Letter grades were averaged to indicate GPA. The average GPA across all time points approximately corresponded to a B average (T1: M = 2.89, SD = .97; T2: M = 3.17, SD = .72; T3: M = 3.09, SD = .79).

Covariates.

T1 adolescent demographics included age in years, self-reported sex (0 = female, 1 = male), and generational status, indicated by dummy coded variables for first, second, and third-/later-generation, the omitted reference group. Adolescent report for household structure was dummy coded to create variables for single parent, stepparent, and two parent homes, the omitted reference group (a small number of youth, n = 6, who reported not living with any parent were included in with the single-parent category). Dummy coded variables for school Latino concentration indicated high, moderate, or low Latino concentration in the school, with low omitted as the referent group. Mother self-report of educational attainment was treated as a continuous variable (1 = 8^th grade or less, 2 = some high school, 3 = completed high school, 4 = some college, 5 = completed college, and 6 = graduate or profession school after college). Mothers’ primary language was assessed by the language in which mothers completed their baseline survey (0 = English, 1 = Spanish).

Analytic Strategy

Missing data.

Intentionally missing mother data (50%) was deemed to be missing at random (MAR), because there were no significant differences in demographics among parents who participated in the survey compared to mothers who did not consent to participate or were unreachable for consent. Missing data due to item non-response ranged .4 to 1% across T1–T3. Attrition from T1 to T3 was 25.2% for youth and 12.1% for mothers, with the majority of youth attrition (22.1%) occurring at T2 when data collection transitioned from in-school to online. Due to the shift in data collection procedures requested by the school district, GPA was missing at T1 for the main sample. Youth report of T2–T3 GPA along with other study variables were used to inform the imputation of T1 GPA, which was shown to be valid based on extensive post-imputation checks. Overall, missing data was treated as MAR, and 100 multiply imputed datasets were produced and averaged to create the multiply imputed grand mean dataset (van Buuren & Groothuis-Oudshoorn, 2011). To preserve sufficient statistics (means, variances, covariances), imputation was conducted at a parcel level as opposed to individual items.

Measurement models.

For our measurement models, we estimated loadings for each parcel-level indicator on their respective latent variable in our measurement models using confirmatory factor analysis (CFA). One ethnic discrimination item was dropped due to poor fit (“others threatened to damage your property because you are Latina”). Measurement model fit was deemed acceptable if the comparative fit index (CFI) was greater than .90 and if the root mean square error of approximation (RMSEA) and the standardized root mean residual (SRMR) were less than .08 (Browne & Cudeck, 2002; Little, 2013). Measurement models were tested for invariance across time and by adolescent sex and maternal primary language. Invariance of loadings and intercepts was determined when change in CFI was less than 0.01, and the RMSEA remained within the 95% confidence interval of the preceding model when adding increasing constraints (Putnick & Bornstein, 2016).

Structural models.

We ran a series of structural equation models (SEM) using Mplus 8.2 (Muthén & Muthén, 1998–2017) to examine whether mothers’ experiences with discrimination were related to adolescent changes in GPA directly and indirectly via mothers’ appraisal of home-school dissonance. We rescaled latent constructs on a metric using phantom variables, which converts variances into standard deviations and covariances into correlations (Little, Slegers, & Card, 2006; Rindskopf, 1984). As a preliminary step, we first assessed stability and homogeneity of the variance-covariance matrix across time and group (adolescent sex and parent primary language). T1–T3 GPA scores were initially regressed on all demographic covariates and non-significant covariates were pruned in the final model. Due to underestimated standard errors resulting from analyses using the grand mean imputed data, we determined statistical significance for direct pathways using the more stringent p-value of less than .005. To test for significance of indirect effects, we used bootstrapping with 5,000 resamples to estimate bias-corrected 95% confidence intervals (Hayes, 2009). Multigroup models using chi-square difference testing for direct effects and the Wald χ² test for the indirect effect were run to examine possible differences in structural paths by mothers’ primary language. Finally, we conducted a series of alternate models to confirm the validity of the mediation model.

Results

Measurement Models and Bivariate Correlations

The CFA results indicated acceptable fit for the overall measurement model (χ² = 818.722, df = 262, p < .00001; RMSEA = .062 90% CI [.058–.067]; SRMR = .039; CFI = .928). As shown in Table 2, standardized factor loadings for the two parceled constructs, maternal discrimination and home-school dissonance, remained consistently strong at each time point. Changes in model fit statistics indicated measurement invariance across both sex and maternal primary language (results available from authors). Bivariate correlations shown in Table 3 indicated that maternal ethnic discrimination was associated with significantly higher levels of home-school dissonance. In addition, discrimination and home-school dissonance demonstrated modest, inverse correlations with adolescent GPA at some time points.

Table 2.

Unstandardized (Standard Error) and Standardized Factor Loadings for Final Measurement Model

Indicator Parcels		Time 1	Time 2	Time 3
	b (SE)	β (SE)	β (SE)	β (SE)
Ethnic discrimination
Parcel 1: felt like outsider + hinted lazy + others think can’t do things + others hurt you + or threatened to hurt	0.95 (.01)	0.89 (.01)	0.90 (.01)	0.90 (.01)
Parcel 2: felt like outsider because of language + ignored + hinted unclean + not trusted + property damaged	1.06 (.02)	0.85 (.01)	0.90 (.01)	0.84 (.01)
Parcel 3: people nice to your face but say bad things + hinted dishonest, untrustworthy + treated unfairly by child’s school + treated unfairly by police, security	0.99 (.02)	0.82 (.02)	0.83 (.02)	0.80 (.02)
Home-school cultural value gap
Parcel 1: ideas about what is important for child different than teachers’ + home life different from school life	0.98 (.03)	0.66 (.03)	0.71 (.02)	0.73 (.02)
Parcel 2: uncomfortable going to child’s school because not like other parents + concerned ideas about what child should learn different than teacher’s	1.09 (.03)	0.83 (.02)	0.86 (.02)	0.81 (.02)
Parcel 3: uncomfortable talking with non-Latino parents at child’s school because culture different than theirs + uncomfortable talking with other Latino parents at child’s school because culture different than theirs	0.93 (.03)	0.67 (.03)	0.67 (.03)	0.65 (.03)

Note. Fit statistics for model included: χ² = 818.722, df = 262, p < .00001; RMSEA = .062 [90% CI .058–.067]; CFI = .928; SRMR = .039. All parameter estimates significant at p < .001

Table 3.

Correlations and Descriptive Statistics Among Study Variables (N = 547)

Variable	1	2	3	4	5	6	7	8	9
1. Ethnic discrimination (T1)	-
2. Ethnic discrimination (T2)	.70***	-
3. Ethnic discrimination (T3)	.44***	.63***	-
4. Home-school dissonance (T1)	.31***	.34***	.21***	-
5. Home-school dissonance (T2)	.34***	.43**	.33***	.59***	-
6. Home-school dissonance (T3)	.26***	.40***	.42***	.50***	.69***	-
7. GPA (T1)	−.07	−.11**	−.10*	−.16***	−.15**	−.16***	-
8. GPA (T2)	−.04	−.11	−.05	−.05	−.07	−.26***	.34***	-
9. GPA (T3)	−.05	−.21***	−.12	−.15***	−.13**	−.25***	.42***	.68***	-
M	1.46	1.38	1.23	1.62	1.58	1.52	2.89	3.17	3.09
SD	0.44	0.44	0.29	0.45	0.42	0.43	0.97	0.72	0.79

Note: Values shown are based on correlations of manifest variables. Imputed parcels for each construct (home school dissonance and discrimination) were averaged together to create manifest variables.

^**p < .005

^***p <.001

Structural Models

In our preliminary analyses, we found no differences in latent means or the variance-covariance matrix by parent survey language or adolescent sex. Therefore, we collapsed groups and conducted the final structural models with the entire sample. There was evidence that the latent means differed across time, suggesting that average discrimination and home-school dissonance did not remain stable across the three time points. The results of the direct effects model shown in Figure 1 indicated that mothers’ discrimination at T1 was associated with T1 to T2 increases in home-school dissonance (β = 0.30, SE = 0.06, p < .001), such that as mothers experienced more ethnic discrimination in the community, they perceived greater incongruence in home-school values approximately six months later. Further, as mothers endorsed greater home-school dissonance at T2, adolescents experienced short-term declines in T2 to T3 GPA (β = −0.14, SE = 0.04, p < .001). With regard to demographic variables (results available from authors), males had lower GPAs than females at T1 and T2 (β = −0.28, SE = 0.06, p < .001; β = −0.25, SE = 0.06, p < .001, respectively), and adolescents who lived in a stepparent household at T1 had a lower GPA at T3 than did those living with biological/adoptive parents (β = −0.21, SE = 0.07, p < .005). Further, mothers’ educational attainment was related to higher GPA at T2 (β = .06, SE = 0.02, p < .005), such that mothers with higher levels of education had adolescent children with higher GPAs. As an indicator of effect size, the amount of variance explained by each outcome is also provided in Figure 1. Half (49.1%) of the variance of T3 GPA was accounted for by T1 discrimination, T2 home-school dissonance, autoregressive paths, and covariates, indicating a large effect size (Cohen, 1988). As shown in Table 4, the association between greater maternal discrimination and declines in GPA over time was mediated by home-school dissonance (β = −0.04, SE = 0.02, 95% CI = −.078, −.007). Additionally, results showed no direct relationship between discriminatory experiences and changes in school performance, suggesting that any relationship between mothers’ experiences with discrimination and adolescent GPA is fully mediated by home-school dissonance. Although it was not possible to calculate a ratio (P_M) as an indicator of magnitude for the indirect effect due to the direct and indirect effects having opposite signs and the total effect being smaller than the indirect effect (Wen and Fan, 2015), the indirect effect was robust given that it emerged over and above the autoregressive paths and the direct effect of discrimination on GPA. Finally, there were no significant differences in direct or indirect effects by mothers’ primary language. The final structural model indicated acceptable model fit (χ² (346) = 1057.530, p <.00001, RMSEA = .061 (90% CI: .057, .066); SRMR = .060; CFI = .908).

Figure 1.

Final direct effects structural model of maternal ethnic discrimination, home-school dissonance, and adolescent grade point average (N = 547). Standardized coefficients are shown based on the use of phantom variables. Dashed lines are not significant. Model controls for mothers’ educational attainment, adolescent gender, household structure, adolescent age and generation status, and school Latino concentration. χ² (346) = 1057.530, p ≤ .00001; RMSEA = .061 (90% CI: .057, .066); CFI = .908; SRMR = .060. *** p < .001

Table 4.

Standardized Path Estimates from Mediation Model (N = 547)^{a, b}

	β (SE)	95% CIc
Effects from ethnic discrimination to GPA
Total effect	.01 (.05)	−.089, .107
Direct Effect
T1 Ethnic discrimination → T3 GPA	.05 (.06)	−.040, .111
Indirect Effect
T1 Ethnic discrimination → T2 HSD → T3 GPA	−.04 (.02)	−.058, −.010

Notes.

^aStandardized coefficients are shown for latent constructs based on the use of phantom variables.

^bModel controls for mothers’ educational attainment, household structure, school Latino concentration, and adolescent gender, generation status, and age.

^cBootstrapping = 5,000.

HSD = Home-school dissonance; GPA = Grade point average; T1 = Time 1; T2 = Time 2; T3 = Time 3. χ² (346) = 1057.530, p ≤ .00001; RMSEA = .061 (90% CI: 0.057, 0.066); SRMR = .060; CFI = .908.

Alternate Models

Results from alternative models confirmed the validity of structural model findings. First, findings for direct and indirect effects run for the 50% (n = 271) of mothers enrolled in the study were essentially the same as those shown using imputed data for the sample (n = 547) that included the intentionally missing 50% of mothers. Second, in models with all second-order autoregressive pathways, findings remained unchanged from the direct and indirect effects shown in Figure 1 and Table 4. These additional autoregressive pathways did not emerge as statistically significant. Third, results from an alternate model excluding T1 GPA, which was imputed for the main sample, were essentially the same as those shown in the model with all three time points of GPA. Finally, results from a model including non-statistically significant paths for demographic covariates were essentially the same as those shown in models with all paths for covariates. Alternative models fit the data well based on criteria established for the structural model.

Discussion

Latin American-origin parents face significant barriers to supporting their children’s formal education (Hill et al., 2018). Although educators and policy makers emphasize parent-school partnerships as an important mechanism for promoting academic success in youth, such relationships are often formed in a context of inequity and exclusion (Auerbach, 2007; Susmita, 2007). For Latin American-origin immigrant populations, the social context is shaped by xenophobic rhetoric and practices, particularly in recent years as immigration has become increasingly criminalized (Brown, Jones, & Becker, 2018). Drawing on ecological and parent involvement frameworks (Bronfenbrenner, 1986; Hornby & Lafaele, 2011), this study considered how Latina mothers’ experiences of discrimination may impact adolescents’ academic performance.

Although GPA, which was roughly 3.0 (i.e., “B”) on average, was generally stable across three time points, maternal reports of experiencing discrimination were associated with significant declines in students’ grades. Although it is known that Latin American-origin parents are confronted with discrimination in the school setting (Hill & Torres, 2010; Hill et al., 2018), our research provides empirical support for the impact of these parental experiences on adolescent academic achievement. Specifically, experiences with discrimination were related to increased appraisals of home-school dissonance, which, in turn, were associated with short-term declines in GPA. Similar to prior research linking parents’ discriminatory experiences to adolescents’ poor academic outcomes by way of parents’ increased stress (Benner & Kim, 2009) and depression (Wheeler et al., 2015), our findings highlight the spillover effect of parents’ discriminatory experiences on student performance. Our research builds on current scholarship by identifying home-school dissonance as an additional mechanism by which parents’ discrimination can indirectly compromise Latin American-origin youth’s educational outcomes.

Theoretically, discrimination would be expected to increase home-school dissonance by limiting the ways in which parents feel comfortable navigating the culture of U.S. schools. It is also possible that perceived discrimination led to stress and/or depression among mothers, impacting their parenting generally and parent involvement specifically (Holloway & Varner, 2021; Nomaguchi & House, 2013). Prior research suggests that in the context of discrimination, Latin American-origin parents are less likely to be proactive in school involvement compared to African American parents (Hill et al., 2018). When Latin American-origin parents perceive discrimination, they may be less likely to have the will or the support to communicate with school staff, be present in the school building, or convey positive feelings about schooling to their youth. Parents’ negative feelings around the school context may be compounded during middle and high school years; transitions to more complex school systems coupled with natural developmental changes in parent-adolescent relationships, such as increased parent-child conflict and youth demands for autonomy, make it difficult for parents to remain involved and connected to school (Hill 2015; Lin et al., 2020). Yet, while youth may seek greater independence from their parents and increasingly discourage some forms of school involvement, they still rely on their parents as primary sources of information for educational planning and aspirations and want them to remain academically involved (Hill, 2015; Hill et al., 2018). If parents feel unwelcome at and disconnected to their child’s school, particularly after experiencing ethnic discrimination, they may be less likely to seek school resources to support their children’s academic success during a time when they are at greater risk for academic disengagement and school dropout (Wang & Eccles, 2012; Wang & Fredricks, 2014). In this way, parents’ experiences of discrimination may compromise adolescent academic achievement by limiting the ways parents feel comfortable seeking resources and information from schools to support their children.

Contrary to expectations, there were no sex differences in pathways from discrimination and home-school dissonance to adolescent GPA. Prior literature has shown that the effect of adolescents’ experiences with discrimination on academic performance is stronger for Latino males than Latina females (e.g., Benner et al., 2018), possibly due to especially negative societal attitudes toward Latino males (Cooper & Sánchez, 2016). We speculate that when parents are the target of discrimination and school marginalization, males and females are similarly harmed by virtue of parents’ limited capacity to remain involved in their child’s education.

Study results did not support a direct relation between parents’ experiences with discrimination and adolescent GPA, which was unexpected given prior research (Espinoza et al., 2016; Ford et al., 2013; Holloway & Varner, 2021). In a two-year longitudinal study among Mexican Americans, Espinoza and colleagues (2016) found that parents’ experiences of discrimination were directly related to youth’s increased internalizing symptoms and lowered self-esteem one year later. It may be that discrimination is more directly consequential for youth’s emotional adjustment but primarily operate through indirect pathways with regard to educational outcomes. Observing or learning about another’s experience with discrimination (i.e. vicarious discrimination) can increase one’s fears around personally being discriminated against and contribute to psychological distress (Truong, Museus, & McGuire, 2016). Thus, while witnessing or becoming aware of discrimination targeted at parents can increase adolescents’ sense of vulnerability and directly compromise their mental health, it may be too distal an extrafamilial stressor to directly influence their performance in schools. However, given the dearth of research on the topic, additional studies are needed to assess whether direct associations exist between parents’ discriminatory experiences and adolescent school success.

Although the coefficient for the indirect effect was of a small magnitude, this finding was robust and theoretically meaningful. Specifically, indirect effects shown here provide important information about the process by which the predictor – maternal discrimination – may influence adolescent school performance above and beyond the autoregressive effect of GPA and the direct effect of discrimination. The use of longitudinal data to establish temporal ordering, the inclusion of theoretically informed covariates, and the large magnitude of main effects further strengthens the validity of findings. Finally, it is valuable to note that the current study focused on testing a theoretical model, rather than on determining the magnitude of impact that parents’ discriminatory experiences may have on adolescent school success. Future studies spanning a longer developmental period and accounting for parents’ chronic experiences of discrimination and home-school dissonance might elucidate the extent to which home-school dissonance acts as a mechanism linking parents’ discriminatory experiences to adolescents’ school performance.

Importantly, this study’s findings have implications for implementing strategies across schools with Latino families to attenuate the detrimental impact of incongruent home and school values on adolescent academic performance. Given the salient role of cultural and linguistic barriers to school involvement for Latino families (Anderson, Cox, Giano et al., 2020; Hill & Torres, 2010), scholars note that there is a need to identify intervention models focused on reducing home-school dissonance (Chun & Dickson, 2011), specifically by working with schools to challenge cultural biases and assumptions (Hill & Torres, 2010). Ethnographic research suggests that bilingual parent liaisons may play a critical role in helping schools recognize and promote Latino values (López, Anguiano, Galinod et al., 2017), which could facilitate feelings of cultural acceptance among families and leverage the protective effects of traditional Latino values emphasized in the home setting. Further, some evidence suggests that increasing ethnic consonance between teachers and parents may increase parent-school involvement (Calzada et al., 2014; Shah, 2009) and ultimately narrow home-school cultural gaps.

Limitations and Conclusion

Despite the novel findings of this study, results should be interpreted in light of a few limitations. First, although home-school dissonance is believed to stem from cultural differences, the measure does not clearly ascertain whether parents’ experiences with home-school dissonance is rooted in culture (Brown-Wright et al., 2013). Additional research is needed to investigate the cultural context of home-school dissonance specific to Latino families, such as the mismatch of collectivism values learned at home and individualism ideals promoted in schools. Second, adolescents’ own experiences of discrimination were not accounted for in the study, an important consideration given that adolescent experiences of discrimination may directly affect academic outcomes (Benner et al., 2018). Third, the present study did not include a direct and comprehensive measure of parent involvement, an important direction for future research. Fourth, while a little over half of study participants were of Mexican-origin, the Latino population is a heterogenous pan-ethnic group, and the relation between discrimination and home-school dissonance may vary considerably by race, language use, and/or country of origin. Although little is known about ethnic differences in parent involvement for Latin American-origin adolescents, research among young children suggest that parenting style and parent cultural socialization may have differential effects on academic achievement for Dominican- and Mexican-origin children (Kim et al., 2018). Due to small cell sizes for country of origin and unavailable data regarding race and documentation status, we could not determine whether pathways differed by subgroup and caution that our findings may not generalize to specific Latin American-origin groups. Fifth, while results from an alternate model did not change when T1 GPA was omitted, the imputation of GPA for three-fourths of the sample at T1 is a limitation. Finally, some evidence suggests that teachers’ emphasis of Eurocentric values and ideas around education may contribute to biased grading for Latin American-origin students experiencing high levels of cultural discontinuity across home and school environments (Taggart, 2017). Future studies should consider incorporating other indicators of school achievement to understand fully the influence of home-school dissonance on academic outcomes.

Notwithstanding the limitations, the findings from the current study advance scholarship on incongruent home-school values and its influence on adolescents’ academic performance using longitudinal data with an immigrant Latino population. In particular, in a departure from previous studies examining youth perspectives of home-school dissonance (e.g., Tyler et al., 2018), this research focuses on parents’ perspective, which to date has been limited in the literature. Additionally, findings provide initial empirical support for the mediating role of home-school dissonance in explaining the link between parents’ discriminatory experiences and youth’s school performance. Considering the heightened levels of discrimination experienced by Latin American-origin adults (Almeida, Biello, Pedraza et al., 2016), schools can adopt programs and policies to reduce gaps in home-school values, thereby bolstering Latin American-origin parent involvement and supporting adolescent academic achievement.