Asian children’s verbal development: A comparison of the United States and Australia

Social Science Research 52 (2015) 389–407

Contents lists available at ScienceDirect

Social Science Research journal homepage: www.elsevier.com/locate/ssresearch

Asian children’s verbal development: A comparison of the United States and Australia Kate H. Choi a, , Amy Hsin b, Sara S. McLanahan c ⁄

a

University of Western Ontario, Canada Queens College, City University of New York, United States c Bendheim-Thoman Center for Research on Child Wellbeing, Office of Population Research, Princeton University, United States b

a r t i c l e

i n f o

Article history: Received 9 April 2014 Revised 5 February 2015 Accepted 18 February 2015 Available online 7 March 2015 Keywords: Asian model minority hypothesis Verbal development Cross-national research

a b s t r a c t Using longitudinal cohort studies from Australia and the United States, we assess the pervasiveness of the Asian academic advantage by documenting White-Asian differences in verbal development from early to middle childhood. In the United States, Asian children begin school with higher verbal scores than Whites, but their advantage erodes over time. The initial verbal advantage of Asian American children is partly due to their parent’s socioeconomic advantage and would have been larger had it not been for their mother’s English deficiency. In Australia, Asian children have lower verbal scores than Whites at age 4, but their scores grow a faster rate and converge towards those of Whites by age 8. The initial verbal disadvantage of Asian Australian children is partly due to their mother’s English deficiency and would have been larger had it not been for their Asian parent’s educational advantage. Asian Australian children’s verbal scores grow at a faster pace, in part, because of their parent’s educational advantage. Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction Asian Americans have been heralded as ‘‘model minorities’’ because of their educational and economic success (Sakamoto et al., 2009; Goyette and Xie, 1999). This perception is largely rooted in the academic success of Asians Americans in the Science, Technology, Engineering, and Math (STEM) fields. Specifically, prior studies have consistently shown that Asian American students score higher on standardized tests of mathematical abilities, enroll in four year universities in higher rates, are overrepresented in STEM fields; and have higher rates of college completion (Eaton and Dembo, 1997; Kao and Thompson, 2003; Xie and Goyette, 2003). The Asian academic advantage is generally attributed to the socioeconomic status of Asian parents and to Asian immigrant culture, which assigns greater symbolic and instrumental value to education (Kao, 1995; Sakamoto et al., 2009; Sue and Okasaki, 1990). Interestingly, the verbal performance of Asian Americans follows a different pattern. Specifically, studies show that Asian American children start out with a verbal advantage over White children at the time of school entry, but this advantage declines during the first few years of elementary school (Fryer and Levitt, 2004, 2006; Goyette and Xie, 1999; Han, 2008; Stiefel et al., 2003). By the time they reach high school, Asian American children are doing less well than Whites from similar socio-economic backgrounds (Glick and White, 2003). With few exceptions, this trajectory of verbal development is observed consistently across different Asian subgroups in the United States, although there is considerable heterogeneity

⁄

Corresponding author at: Department of Sociology, 5403 Social Science Centre, University of Western Ontario, London, Ontario N6A5C2, Canada. E-mail address: [email protected] (K.H. Choi).

http://dx.doi.org/10.1016/j.ssresearch.2015.02.010 0049-089X/Ó 2015 Elsevier Inc. All rights reserved.

390

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

in modes of incorporation and consequent socioeconomic conditions within the Asian American population (Han, 2008). The absence of a clear Asian academic advantage in verbal performance suggests that Asian children – even those who are socioeconomically advantaged and living in a country with immigrant admission policies conducive to higher levels of English proficiency- may be at a relative disadvantage in terms of their verbal development because they are more likely than Whites to grow up in families where English is not the primary language. Despite great interest in the academic performance of Asian Americans, including their verbal development, the question of why Asian American children lose their initial advantage in verbal performance has remained largely unexamined. In addition, most studies of Asian children’s academic performance have been conducted in the United States, and thus, we do not know if the pattern of verbal development observed in the United States extends to other countries as well. To address this gap in the literature, we use growth curve models to compare the trajectories of verbal development of children born to native-born White and Asian immigrant mothers and to ascertain the role that parent’s socioeconomic status and English proficiency play in explaining their trajectories. Specifically, we ask (1) whether differences in parents’ socioeconomic status can account for the advantage of Asian American children at the time they enter kindergarten, (2) whether differences in parents’ English proficiency can account for the relative decline of Asian children during elementary school, and (3) whether the patterns observed in the United States extends to Asian children born in Australia. All analyses are conducted separately for the various Asian subgroups (i.e., East, Southeast, and South Asians) in recognition of the vast heterogeneity in socioeconomic conditions and academic outcomes of children who belong to the various Asian subgroups. Australia is the selected country for our cross-national comparison. As former British colonies, the United States and Australia share the same cultural and historical roots, including the use of English as their official language. Moreover, Asian immigrants in Australia and the United States are generally high skilled immigrants who originate from similar regions within Asia, although their socioeconomic standing and English proficiency may differ due to variations in the immigrant admission criteria of the two countries.

2. Background Scholars have recently begun to study the academic performance of Asian American children during early childhood as a way to gain insights into when the Asian academic advantage emerges and how it changes over time (Fryer and Levitt, 2004, 2006; Han, 2008; Wang, 2008). This body of work finds that Asian children have a verbal advantage over their White counterparts at school entry, but this advantage fades over time (Fryer and Levitt, 2004, 2006; Han, 2008; Sun, 2011; Wang, 2008). Using data from the Early Childhood Longitudinal Study – Kindergarten Cohort (ECLS-K), Fryer and Levitt (2004, 2006) find that Asian American children score considerably higher than White children in reading assessments at the time they enter kindergarten. The gap, however, has become smaller by first grade, and has disappeared entirely by third grade (Fryer and Levitt, 2004, 2006). Han (2008) disaggregates the population of Asians into regional subgroups and finds that the pattern of White-Asian difference in verbal development is similar across Asian subgroups despite the considerable heterogeneity in modes of incorporation and consequent socioeconomic differentials across these groups. In fact, the only difference is observed in the size of the initial disadvantage as well as the pace of growth.

2.1. Do differences in parents’ socioeconomic status explain the advantage of Asian American children at school entry? The Asian-American academic advantage is generally attributed to the socioeconomic status of Asian parents (Kao, 1995; Sakamoto et al., 2009; Sue and Okasaki, 1990). According to this explanation, Asian parents are better educated and have higher income than parents in other race/ethnic groups because Asian immigrants are typically recruited into the United States as high-skilled laborers (Xie and Goyette, 2004; Sakamoto et al., 2009). Asian children perform better in school than children in other groups because their parents’ more favorable socioeconomic background gives them greater access to educational resources and shields them from the environmental toxins and maternal stressors arising due to economic hardship (Duncan and Magnuson, 2005; Kao and Thompson, 2003; Sakamoto et al., 2009). Empirical work, however, provides mixed accounts about the extent to which differences in parents’ socioeconomic status account for the verbal advantage of Asian American children at school entry. A study conducted by Sun (2011) using the ECLS-B shows that differences in parent’s socioeconomic status (e.g., parent’s education, family income, maternal employment) explain a portion (i.e., 15 percent) of the gap in verbal performance between Whites and East Asians at age 4, but East Asian children continue to have an educational advantage over Whites even after accounting for differences in socioeconomic status. In contrast, studies conducted by Fryer and Levitt (2004, 2006) find that socioeconomic differences (e.g., SES composite scores, mother’s WIC receipt) account for very little of the gap in verbal scores between Whites and all Asians at the time of entry into kindergarten. The contrasting accounts may be due to the fact that Sun’s study only includes East Asians, while Fryer and Levitt’s study includes all Asians. Han (2008) shows that while parent’s socioeconomic characteristics explain a considerable portion of the differences in verbal performance between White and East Asian/ Indian children, but they account for little of the differences in verbal performance between White and Southeast Asian children. In recognition of the heterogeneity in the impact of parent’s socioeconomic characteristics on verbal differences between Whites and Asian subgroups at the time of school entry, our study disaggregates Asians into various subgroups,

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

391

namely East, South, and Southeast Asians when examining the impact of parent’s socioeconomic status on White-Asian differences in verbal development. 2.2. Do differences in English proficiency explain the decline in the Asian academic advantage? Deficits suffered by children of Asian immigrants due to their parent’s limited English-language proficiency may explain why the Asian American advantage in verbal performance at school entry declines over time. Specifically, parents with limited English proficiency may adversely affect their children’s verbal development by engaging in fewer verbal interactions, providing limited exposure to vocabulary words, and relying on grammatically simple sentence structures when communicating with their children (Hart and Risley, 1995; Farkas and Beron, 2004). Immigrant parent’s limited English proficiency may also influence children’s academic performance, including verbal performance, indirectly by preventing their parents from effectively communicating with teachers and engaging in school activities (Glick and White, 2003; Turney and Kao, 2009). Finally, having an unfavorable language background may have a more pronounced effect on children’s verbal development at older ages as the English content of course materials becomes more complicated. Most prior work documents White-Asian differences in verbal performance at single points in time and seldom examines within person changes over time (Fryer and Levitt, 2004, 2006; Sun, 2011; an exception Han, 20081). Thus, it is difficult to disentangle the impact of the various determinants on White-Asian differences in initial verbal scores from their impact on changes in verbal performance. Furthermore, the main focus of research on the academic performance of Asian Americans has been to decipher the factors giving rise to Asian academic advantage. The question of why Asian children lose their initial advantage in verbal performance has remained largely unexamined. In this paper, we use growth curve models which allow us to distinguish between the effects of parental socio economic resources and English proficiency on White-Asian differences in children’s initial test scores as well as on changes in their test scores over time. 2.3. Do the patterns observed for Asian American children extend to Asian children outside of the United States? Although Asian immigrants dominate the global share of international immigrants, surprisingly little is known about how children of Asian ancestry fare outside of the United States. Thus, an important contribution of our study is to determine whether the patterns observed for Asian American children exist countries other than the United States. Australia provides a good platform for conducting a cross-national comparison of White-Asian differences in verbal development for several reasons. First, the skill level and ethnic composition of Asian immigrants in the two countries are somewhat similar. Asian immigrants typically represent a positively selected group of economic immigrants in both countries, entering the United States under the high skilled visa category (Sakamoto et al., 2009) and entering Australia only if they meet or surpass the minimum admission criteria under a point system designed to attract skilled workers (Walsh, 2008). Nonetheless, the two countries also have a sizable portion of Southeast Asian immigrants who migrated as refugees in the aftermath of the Vietnam War (Castles and Miller, 2009). The ethnic composition of Asian immigrants in the two countries is also somewhat similar, with East Asians (predominantly Chinese), South Asians (predominantly Indian), and Southeast Asians (predominantly Vietnamese) being the three largest Asian subgroups in both Australia and the United States (Jasso and Rosenzweig, 2008). Second, as former British colonies, the United States and Australia share the same cultural and historical roots, including the use of English as their official language. Third, comparable data is available in the two countries: the design of the Longitudinal Study of Australia Children – Kindergarten Cohort (LSAC-K) was closely modeled after that of the American Early Childhood Longitudinal Study – Kindergarten Cohort (ECLS-K) (Coley et al., 2013; Soloff et al., 2005). Most relevant for our comparison, both studies sampled children at the time they enrolled in kindergarten and both studies assessed children’s verbal development at three or more points in time (5.7, 7.3, 9.1 in the ECLS-K; and 4.2, 6.3, 8.2 in the LSAC-K). Despite their many similarities, Australia and the United States employ somewhat distinct immigrant admission criteria that may have important implications for the verbal development of Asian children in the two countries. Perhaps the most notable difference in immigrant admission policy across the two countries is the differential emphasis that Australia and the United States place on English proficiency: the US immigration system does not have English proficiency requirements, whereas Australia places a heavy emphasis on English proficiency (Jasso and Rosenzweig, 2008). Specifically, under the point system, Australia awards up to 20 points for English competence, which is also the maximum number of points awarded for educational attainment.2 The greater emphasis on language proficiency as a criterion for immigrant admission into Australia 1 Han (2008) uses growth curve models to document White-Asian differences in verbal development. Her full model adds several covariates capturing family background, including language use at home. Because language use is added along with several other variables capturing family background, we cannot isolate the impact of language use at home on White-Asian differences in verbal trajectories. Furthermore, because her primary goal was to evaluate the effects of school characteristics on children’s academic trajectories, Han (2008) does not show or explicitly discuss the impact of family background, including language use at home, on White-Asian differences in verbal development. In fact, she does not show the coefficients for family background in her tables (Han, 2008: Model 3 in Table 2). 2 65 points in the minimum number of points needed to enter Australia as a skilled immigrant. See: https://www.immi.gov.au/skilled/general-skilledmigration/pdf/points-test.pdf.

392

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

means that Asian immigrant parents in Australia will typically have greater English proficiency than their counterparts in the United States. The two countries also differ markedly in their requirements for employment visas, which may have implications for the socioeconomic wellbeing of Asian immigrant families. With some exceptions, employment visas in the United States are employer-sponsored; whereas, employment is not a pre-requisite for immigrant admission in Australia3 (Jasso and Rosenzweig, 2008). Although there is evidence that Asian immigrants in both countries are underemployed, the stronger coupling between legal and employment statuses in the United States suggests that economic immigrants of Asian origin in the United States are less likely to be underemployed than those in Australia (Zeng and Xie, 2004; Green et al., 2007). Thus, the magnitude of economic disparities between Whites and Asians with comparable levels of education is expected to be smaller in the United States than it is in Australia. This difference may affect children’s academic achievement, including their verbal performance, by determining the amount of educational resources available to the child at home. Also relevant for our study is the fact that the educational policies in the two countries also differ somewhat with respect to their emphasis on immigrant student’s English language proficiency. US educational policies seek to simultaneously promote student’s English proficiency and cultural and linguistic diversity (Calderon et al., 2011). For example, public schools, particularly those located in areas with large immigrant populations, offer ESL classes and English language support for immigrant students who are not proficient in English (Calderon et al., 2011; Han, 2008). At the same time, other programs also exist to foster linguistic diversity in the form of bilingual schools that allow non-native students to receive instruction in their native language (Calderon et al., 2011). In contrast, Australian schools place a near-exclusive emphasis on English acquisition to the point that researchers in Australia express concern over the maintenance and development of immigrant languages (Clyne, 2005; Hammond, 2001; Iredale and Fox, 1997; Rubino, 2010). For example, with few exceptions, the language of instruction in Australia is English only. Perhaps as a result of the different educational policies, immigrant children in Australia tend to abandon their native languages and develop a working proficiency of oral English relatively quickly following their arrival into Australia (Rubino, 2010). Given these differences in the educational policies, we may expect schools to play a larger role in reducing potential English language disadvantages among Asian students in Australia than among Asian students in the United States. Cross-national differences in immigrant admission policies can engender White-Asian differences in socioeconomic characteristics and English proficiency, which have implications for group differences in academic performance, including in verbal assessments. Furthermore, cross-national differences in educational policy regarding the English proficiency and literacy skills of immigrant children can influence Asian children’s verbal development. Thus, our study examines the extent to which White-Asian differences in socioeconomic characteristics and English language proficiency generate group differences in children’s verbal development between school entry and early elementary school years in each country and compares the pattern of group variation across countries. Past studies documenting the achievement of Asian children in Australia are sparse, and the few exceptions that exist do not distinguish between math and verbal scores, focusing on total exam scores. One of these few exceptions is CobbClark and Nguyen’s work (2010), which shows that Australian youth from non-English-speaking households have higher ENTER scores (i.e., university entrance rankings computed from their performance in several academic subjects) than their Australian-born peers and immigrant youth from English-speaking households. Although this study does not specifically compare the academic performance of Asian and White Australians, we can make inferences about the White-Asian gap, as most non-English speaking immigrants to Australia are East Asian immigrants and most Australian born youths are White. Taken together, these studies suggest that Asian adolescents in Australia have an educational advantage over their White counterparts. It remains to be seen whether this advantage is also observed in early elementary years and whether this advantage is observed when we focus solely on verbal performance.

3. Material and methods 3.1. Data Our study uses data from (1) the American Early Childhood Longitudinal Study – Kindergarten Cohort (ECLS-K) and (2) the Longitudinal Study of Australian Children – Kindergarten Cohort (LSAC-K). The ECLS-K4 is a nationally representative study that follows 21,409 children who enrolled in kindergarten programs in the United States in the fall of kindergarten in 1998. This survey oversamples Asian Pacific Islanders and asks the main caregiver (usually the mother) to provide detailed reports about the family’s socioeconomic status. Follow-up interviews were collected in spring of kindergarten (1999), the fall and spring of 1st grade (1999–2000), the spring of 3rd grade (2002), the spring of 5th grade (2004), and the spring of 8th grade (2007). Assessments in reading and math are collected in all waves. Average age of assessment in each wave is 5.7, 6.2, 6.7, 7.3, 9.1, 11.3, and 14.3, respectively. We limit our analysis to information collected in the fall of 1998, spring of first grade, and spring of third grade because (1) we wished to capture the developmental 3 Australian immigrants are awarded 5 points if they show proof of a job offer in an industry experiencing shortages of labor. See: https://www.immi.gov.au/ skilled/general-skilled-migration/pdf/points-test.pdf. 4 http://nces.ed.gov/ecls/.

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

393

trajectories of verbal scores during early and middle childhood and (2) we want to use time points that were roughly comparable with the Australia data. The LSAC-K is a nationally representative study that follows approximately 5000 children between 4 and 5 years of age living in Australia in 2004 (Soloff et al., 2005).5 This survey asks the main caregiver (usually the mother) to provide detailed reports about the family’s socioeconomic status. Follow-up interviews are conducted in 2006, and 2008, which roughly corresponds to first and third grades in Australia. Peabody Picture and Vocabulary Tests (PPVT) are collected at each wave, but math assessments are collected only at follow-up interviews. The average age of respondents was 4.2, 6.3, and 8.2 years in the three waves, respectively. These data sources are well-suited for the present analyses because they have verbal test scores in three or more waves of data. They also include detailed information on parent’s socioeconomic status (i.e., parent’s education, employment and household income) and mother’s language proficiency. Finally, the two datasets employ a similar sampling design, allowing us to document cross-national differences in the pattern of White-Asian variation in children’s verbal performance (Coley et al., 2013). 3.2. Analytic sample Our analytical sample is comprised of native-born children of Asian immigrant and native-born White mothers in each country. We exclude children born to US-born Asian mothers because (1) ECLS-K did not collect information about grandparent’s region of birth and we cannot disaggregate Asian mothers into subgroups; (2) Australia restricted the migration of non-Whites until the early 1980s and thus it is highly unlikely that LSAC-K includes children of Australian-born Asian mothers (Walsh, 2008)6; and (3) LSAC-K does not ask mothers to identify their race/ethnicity.7 We apply three additional restrictions to this subsample. First, we exclude children who were not living with their biological mothers on the basis of prior work showing that step parents make fewer monetary and non-monetary investments on their step children than biological mothers (Case and Paxson, 2001). Second, in recognition of the heterogeneity in academic performance and linguistic proficiency across Asian subgroups, we limit our sample to Asian children whose mothers were born in East, South, and Southeast Asia.8 Finally, we restrict our sample to children whose mothers provided valid information regarding their country of birth.9 These steps yield a sample size of 6841 children in the United States (6398 US-born Whites; 99 East Asian, 271 Southeast Asian, and 73 South Asian children) and 3202 children in Australia (2947 Australian-born Whites, 69 East Asian, 108 Southeast Asian, and 78 South Asian children). For simplicity of presentation, we use the term ‘‘Asian children’’ to refer to children with Asian immigrant mothers and apply an analogous strategy when we discuss ‘‘White children’’. 3.3. Measures 3.3.1. Outcome variables Children’s verbal ability is measured using the reading item response theory (IRT) scores from the ECLS-K and the PPVT scores from the LSAC-K. PPVT and reading IRT are both tests of verbal dexterity, and thus, are highly correlated. However, they measure somewhat different dimensions of verbal performance, with PPVT scores mainly capturing dexterity in English vocabulary and reading IRT scores capturing dexterity with letter recognition, vocabulary, reading comprehension, and critical stance (Rock et al., 2002). Additionally, the composition of the reading IRT composite scores changes between early and middle childhood, with IRT scores placing a heavier emphasis than PPVT on letter recognition at kindergarten entry and on reading comprehension and critical stance at later ages (Rock et al., 2002). To account for these differences, we standardize the verbal scores into Z-scores at each wave and conduct a cross-national comparison of patterns of White-Asian differences in verbal development. A similar approach was used by Washbrook et al. (2012). The ECLS-K did not administer reading assessments to children who did not make a predetermined cutoff point of verbal scores (Han, 2008). Partly because of this reason, verbal scores are missing for 2 percent of children in the fall of kindergarten, 3 percent of children in the spring of first grade, and 6 percent of students in the spring of third grade. Multiple imputation techniques with STATA’s mi commands are used to handle missing verbal scores.10,11 Missing verbal scores are 5 95.5% of these children were enrolled in kindergarten. We obtain virtually the same result before and after we restrict our sample to children who were enrolled in kindergarten. 6 LSAC-K children were born in 1998–1999. Mothers of LSAC-K children were born between 1950 and 1984. In Australia, immigration of non-Whites was restricted until the early 1980s; therefore, it is highly unlikely that LSAC-K includes children of Australian-born Asian mothers. 7 Preliminary analyses for the US also examined the academic trajectories of children with US-born Asian mothers. We found that the trajectories of children born to US-born Asian mothers fall consistently in between those of US-born White and Asian immigrant mothers. These results are available upon request. 8 These are the Asian regional subgroups with at least 65 respondents in both datasets. 9 Larger shares of children are held back in the United States (6%) than in Australia (4%). Rate of being held back a grade does not vary across the groups in Australia; however, they vary in the United States. As a supplementary analysis, we ran models which excluded children who were held back. Our results change little with the exclusion of children who were held back. Results from this supplementary analysis are available upon request. 10 Consistency checks were performed using a subsample of children who took the verbal assessments in all three waves of data. The pattern of White-Asian difference in verbal trajectories in Australia and the United States are similar in analyses performed using samples including and excluding imputed data. 11 We imputed missing verbal scores using only information about language spoken to child by both parents and substituted the minimum verbal score to missing cases. The distinct approaches yield similar results differing only with respect to the magnitude of White-Asian differences in initial scores and growth rates.

394

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

imputed at each wave using information about language spoken to child by both parents, mother’s education, and mother’s race/ethnicity. 3.3.2. Independent and control variables We use mothers’ self-reports of race and country of birth to categorize children into four groups: (1) Whites, (2) East Asian, (3) Southeast Asian, and (4) South Asian mothers.12 For simplicity, we use the term mother’s race to refer to this variable. In the United States, White mothers refer to US-born women who self-identify as non-Hispanic Whites. In Australia, White mothers refer to non-aborigine women born in Australia. Asian mothers are classified into three groups depending on their region of birth – East, Southeast, and South Asia. We measure parent’s socioeconomic status using each parent’s education, each parent’s employment status, and household income at wave 1. We selected these socioeconomic dimensions because they have been previously identified as key determinants of the economic wellbeing of immigrants and are the criteria to get a high-skilled visa in the two countries (Regets, 2001; Sakamoto et al., 2009).13 We consider both mother’s and father’s education and employment status due to variations in the gender selective nature of immigration flows across Asian subgroups and the consequent differences in the primary visa holder’s gender. Mother’s education classifies US children into three categories using reports about mother’s completed years of schooling (high school or less; some college; BA or higher) and Australian children into six categories using reports about their highest degree attained (610 years without certificates, 610 years with certificates, 11 to 12 without certificates, 11 to 12 with certificates, advanced diploma, and BA or more.14 Father’s education is constructed in an analogous fashion and includes a missing category.15 Mother’s employment status classifies respondents into three categories (employed, unemployed, and missing) using mother’s reports about their employment status. Father’s employment status is constructed in an analogous fashion. Household income at wave 1 uses mother’s reports about total household income to classify respondents in four categories (lowest, middle, second highest, highest). These categories roughly divide our entire sample into quartiles.16,17 Mother’s proficiency in English. We capture children’s language backgrounds using mother’s reports about how well they speak English at baseline and classify children into five categories: not well at all, not well, pretty well, very well, and missing.18,19 Our multivariate analyses also include several demographic controls, including child’s gender (female vs. male); child’s birth weight (less than 2500 g vs. 2500+ g); mother’s age at the focal child’s birth (<20; 20–24; 25–29; 30–34; 35+); family type (two parents plus siblings, two parents and no siblings, one parent with siblings, and one parent without siblings), and differentials between respondent’s age and average age at assessment (6 or more months younger, 3–5 months younger, younger by 3 months or more, 0–2 months older, 3–5 months, 6 months or older). 3.4. Methods We estimate two-level growth curve models to document White-Asian disparities in verbal development in the United States and Australia. This strategy allows us to estimate White-Asian differences in initial scores (i.e., intercepts) and rate of growth across the distinct waves (i.e., slopes) (Han, 2008). The level-1 equation in our growth curve models describe within-individual changes in test scores (i) over time (t) and can be represented as follows:

yit ¼ ai þ bi t þ eit

ð1Þ

Children’s verbal development ðyit Þ are characterized by a unique intercept (ai) and a time-dependent slope ðbi Þ. Time (t) is added as a series of dichotomous variables indicating the wave when assessments were conducted to capture the non-linearity in children’s verbal development. 12 We classify children according to their mother’s race instead of father’s race or parent’s joint race. This decision is made in consideration of children residing in single parent families, who usually reside with mothers and for whom information about father’s race may be missing. Among children for whom information about father’s race/ethnicity is available, father’s race was typically the same as mother’s race because racial endogamy is the most predominant form of marital sorting pattern; and as such, the addition of father’s race does not alter our general results. These results are available upon request. 13 We ran models with and without preschool attendance. They yield virtually the same results. 14 In most Australian territories, grades 9 and 10 are classified as secondary high school; grades 11 and 12 are classified as senior secondary schools. Mandatory schooling typically ends with grade 10. 15 We ran sensitivity tests using measures of mother’s and father’s education, which distinguished Australian children into three groups (high school or less, advanced diploma, and BA or more). The use of these alternate measures does not change our results. 16 All datasets report categories of income. 17 ECLS-K imputed total household income for 5.9 percent of children whose parents did not report household income. We use this measure with the missing flag for imputed income. We ran models using a time-varying measure of income and obtained the same results. 18 The ECLS-K asked mothers to rate their skill across three dimensions of English proficiency – writing, speaking, and reading – at wave 1 (i.e., fall of kindergarten). Using principal component factor analysis, we constructed a scale using all three measures of English proficiency – writing, reading, and speaking. Supplementary analyses were conducted using this scale. Results from these supplementary analyses are virtually the same as those obtained using just the measure of spoken English. They are available upon request. 19 LSAC-K asked mothers to rate how well they spoken English in all waves. We conducted supplementary analyses where we included mother’s proficiency in spoken English as a time-varying covariate. They yield virtually the same results as those obtained using a time-fixed measure of mother’s spoken English.

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

395

The level-2 equation models variation in verbal scores between Asian and White children (i.e. across individual effects). Formally, the level-2 relationships can be represented as follows:

ai ¼ a0 þ a1 xi1 þ a2 xi2 þ    þ ak xik þ ui

ð2Þ

bi ¼ b0 þ b1 xi1 þ b2 xi2 þ    þ bk xik þ v i

ð3Þ

These equations indicate that the random intercepts ðai Þ and slopes ðbi Þ are a function of time-invariant covariates ðxik Þ and error terms ðuit ; v it Þ, respectively. For each country, we fitted four successive models to address the following questions: (1) how does the verbal development of Asian children differ from that of White children?; (2) do differences in parents’ socioeconomic status explain White-Asian differences in verbal development?; and (3) do differences in English-language proficiency explain WhiteAsian differences in verbal development?. The first two models address the first question: Model 1 includes mother’s race, wave, and the interaction between mother’s race and wave and Model 2 adds demographic controls to Model 1. Model 3 addresses the second question by adding parent’s socioeconomic status to Model 2.20 Model 4 addresses the third question by adding mother’s English-language proficiency to Model 3. Three analytical approaches deserve further mention. First, our study focuses on verbal performance (instead of math performance) during early and middle childhood because (1) verbal achievement is an important determinant of later academic achievement, including math performance (Vukovic and Lesaux, 2013); (2) immigrant children’s verbal performance helps assess the degree of linguistic and consequent socio-cultural assimilation experienced by immigrant groups (Bloemraad and de Graauw, 2013); and (3) the Asian advantage in math scores is already well-established21 (Fryer and Levitt, 2006; Han, 2008; Sakamoto et al., 2009). Our decision to focus on verbal performance is also driven by data availability: (1) math scores are only available in 2 waves of LSAC-K data, preventing us from observing trajectories of verbal development; (2) math scores are missing for a significant proportion of Australian children22; and (3) math scores are missing in a non-random fashion (e.g., missing rates are higher among Asian children than they are among Whites). Supplementary analyses of White-Asian differences in math performance are included in the appendix section of this paper. Second, our results cannot be used to assess cross-national differences in children’s absolute skill level because the LSAC-K and ECLS-K use different assessments to measure verbal ability. Instead, we can assess how the verbal skills of Asians compare with those of Whites living in the same country, and later, how White-Asian disparities compare across countries. A similar approach was used by Washbrook et al. (2012). Third, we focus on early to middle childhood because this is the time period when studies show a turnaround in the advantage of Asian American children and because the impact of family background on children’s academic performance will likely be larger during this time than at older ages when schools, teachers, and peers increasingly play a more important role in the lives of children (Shah, 2011). 4. Results 4.1. White-Asian differences in verbal development in the United States We begin our analyses by examining how the verbal development of Asian American children compares with that of Whites during early and middle childhood. Our results, presented in Fig. 1,23 show that Asian children have higher verbal scores in the fall of kindergarten than Whites. Their scores, however, grow at a slower rate than those of Whites, especially between the spring of first and third grade. Their verbal advantage relative to Whites has eroded considerably by the spring of third grade. In-depth comparison of the trajectories of verbal development across the various Asian subgroups reveals that the only subgroup differences observed is the size of the initial differences (intercept) and the pace of growth across waves (slope). Specifically, South and East Asian children have an initial advantage over Whites, but their scores also grow at slower rate compared to Whites. Both the size of the initial advantage as well as the slower pace of growth is largest among South Asians. Despite the slower pace of growth, South and East Asian children are able to retain their verbal advantage relative to Whites by third grade due to their considerable initial advantage. This finding, however, does not apply to Southeast Asians whose initial verbal scores are on par with those of Whites. Like other Asian groups, their scores grow at a slower pace than Whites; and thus, they have a verbal disadvantage relative to Whites by the spring of third grade. These results are consistent with those presented by Han (2008). 20 Our models account for clustering of time points within individuals. Yet, we are unable to account for clustering arising due to the sampling design because STATA’s multiple imputation (mi) command is incompatible with svy commands (STATACorp, 2013). Robustness checks, however, reveal that models without the svy command and the models with the svy command (without the mi command) yield virtually the same results, suggesting that the additional employment of svy commands will likely not have a sizable impact on our results. Using Mplus, we were able to conduct analysis using imputed scores and sampling weights. Our general results remain consistent across software. 21 http://www.oecd.org/pisa/46643496.pdf. 22 30 percent of respondents in the LSAC-K did not report math scores at age 6 and 20 percent of respondents in LSAC-K did not report their math assessment scores at age 8. This compares with less than 3 percent for verbal assessment scores at each wave of LSAC-K. 23 The predicted scores presented in Figs. 1 and 2 are computed from Model 1 in Tables 2 and 4. They are consistent with our descriptive findings in Table 1.

396

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

Standardized Verbal Scores

1.2

0.8

0.4

0.0 Fall of kindergarten (5.7)

Spring of first (7.3)

Spring of third (9.1)

-0.4

-0.8 White

East Asian

Southeast Asian

South Asian

Fig. 1. White-Asian differences in standardized verbal scores, United States.

The observed pattern is consistent with the view that Asian parents are more likely than non-Asian parents to teach their children letter recognition and basic vocabulary before they start formal schooling because (1) they place greater emphasis on education, especially at early ages and (2) they are better able to provide their children greater access to educational resources due to their socioeconomic advantage (Hsin and Xie, 2014; Parmir et al., 2008; Sakamoto et al., 2009; Yu and Ruan, 2012). Greater exposure to early English education results in Asian children’s initial verbal advantage; however, upon school entry, White children will acquire the skills Asian children learned ahead of time and ‘‘catch-up’’ to Asians.

4.2. White-Asian differences in parent’s socioeconomic status and English language use in the United States Next we examine how the socioeconomic profiles and English language proficiency of Asian immigrant parents in the United States differ from those of US-born White mothers. As shown in Table 1, East and South Asian parents are more socioeconomically advantaged than Whites in several dimensions. The Asian advantage is most pronounced with respect to parent’s education and household income. For instance, 45 percent of East Asian and 68 percent of South Asian mothers have a college degree, as compared to only a third of White mothers. Over a third of East Asian and half of South Asian children reside in households in the top quartile of the income distribution, as compared to about 28 percent of White children. Our results for Southeast Asian children reveal a more complicated pattern of socioeconomic differentials than is found for East and South Asian children. The education distribution of Southeast parents follows a bi-modal pattern, with higher shares of Southeast Asian parents present in the highest and lowest category of parent’s education than Whites. Southeast Asian children are more likely to reside in families with lower household income than their White peers. Almost 40 percent of Southeast Asian children reside in households in the bottom quartile of the income distribution, as compared to 26 percent of White children. These results are consistent with past reports about the socioeconomic characteristics of Asian Americans (Sakamoto et al., 2009). Unsurprising given the fact that many South Asian countries were former British colonies, South Asian mothers exhibit greater English proficiency relative to other Asian groups. When asked how well they speak English, 93 percent of South Asian mothers report that they speak English ‘‘well’’ or ‘‘very well’’, as compared with 81 percent of East Asian and 83 percent of Southeast Asian mothers.

4.3. Explaining White-Asian differences in verbal development in the United States Table 2 presents the results from four additive growth curve models predicting White-Asian differences in verbal development in the United States. The main effects of mother’s race (a) captures differences in verbal scores between the various Asian subgroups and Whites at the time of entry into kindergarten and the slope (b) captures differences in growth rates between Asian subgroups and Whites across waves. For example, based on Model 1, we know that South Asian children have initial scores that are 1.10 standard deviations above those of White children, but this initial advantage erodes by 0.45 standard deviations between the fall of kindergarten and spring of first grade and 0.76 standard deviations between the fall of kindergarten and spring of third grade. Thus, by the spring of third grade, the verbal scores of East Asian children are 0.29 standard deviations and the verbal scores of South Asian children are 0.34 standard deviations above those of White children. We begin by describing the results from Model 2 because the results from Model 1 are discussed in Section 4.1. Model 2 adds demographic controls to the existing model. Demographic controls have little effect on White-Asian differences in verbal trajectories. The only exception is observed in the fact that demographic controls explain 7 percent [100 ⁄ (0.67  0.62)/0.67  7] of the East Asian children’s advantage over Whites. Supplementary analysis suggests that this effect accrues in large part because East Asian children have fewer siblings than other groups and number of siblings is negatively associated with academic skill.

397

K.H. Choi et al. / Social Science Research 52 (2015) 389–407 Table 1 White-Asian differences in parent’s socioeconomic status and mother’s linguistic proficiency, United States. USB White

FB Asian

(N = 6398)

East (N = 99)

Southeast (N = 271)

South (N = 73)

Verbal scores Fall of kindergarten (age = 5.7) Spring of 1st grade (age = 7.3) Fall of kindergarten (age = 9.1)

0.05 0.04 0.01

0.77 0.61 0.32

0.01 0.07 0.37

0.99 0.47 0.34

Parent’s socioeconomic status Mother’s education HS graduate or less Some college BA or higher Total

32 35 33 100

27 27 45 100

38 23 38 100

18 14 68 100

Mother’s employment Unemployed Employed Missing Total

29 71 0 100

44 56 0 100

25 74 0 100

32 66 2 100

Father’s education HS graduate or less Some college BA or higher Missing Total

32 25 33 10 100

17 19 62 2 100

35 24 33 8 100

22 11 64 4 100

Father’s employment Unemployed Employed Missing Total

3 85 12 100

6 92 3 100

9 81 10 100

1 95 4 100

Household income Low Second Third Highest Total

26 20 25 28 100

21 17 26 36 100

39 20 16 25 100

16 17 17 50 100

Language background How well do mothers speak English Not well at all Not well Well Very well Missing Total

0 0 0 100 0 100

5 14 19 62 0 100

1 15 14 69 2 100

0 4 21 72 4 100

Notes: (1) Percentages are weighted. Number of observations is not weighted. (2) The totals may not add to 100 percent due to rounding. (3) Parent’s education and mother’s language proficiency are time-invariant covariates. Father’s and mother’s employment and household income are time-varying covariates. We report the average distribution of time-varying covariates across all waves.

Model 3 adds parent’s socioeconomic status to the existing model. Our results show that parent’s socioeconomic status explains some of the initial advantage in verbal scores that East and South Asians have over Whites. Specifically, controlling for parent’s socioeconomic status reduces East Asian children’s initial advantage in verbal scores by approximately 10 percent [100 ⁄ (0.62  0.57]/0.62  10] and South Asian children’s initial advantage by about 20 percent [100 ⁄ (1.08  0.87)/ 1.08  20]. Interestingly, parent’s socioeconomic status does not explain why the verbal scores of Asian children in all subgroups grow at a slower rate relative to Whites. Model 4 adds mother’s English language deficiency to the existing model.24 Asian children’s initial advantage in verbal scores widens once we include controls for mother’s English deficiency. For example, net of mother’s English deficiency, the verbal advantage of East Asian children at the time of entry into kindergarten increases by 14 percent [100 ⁄ (0.65  0.57)/

24 We conducted sensitivity analysis using growth curve models which further disaggregated Asian children into two groups: those whose mothers have limited English proficiency and those whose mothers report being proficient or near native speakers of English. The trajectories of verbal development of Asian American children born to mothers with limited English proficiency follow the same pattern as those of Asian American children born to mothers who are proficient in English. The only difference is that the initial advantage is greater and the rate of decline is steeper among Asian American children born to mothers with limited English proficiency is smaller than those of their peers born to mothers who are proficient in English.

398

Table 2 Growth curve models predicting standardized verbal scores by parent’s socioeconomic status and English language proficiency, United States. Model 1 Zero order

Model 2 Model 1 + Demographic controls Intercept

Slope (b)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

(a)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

Intercept

Slope (b) Spring 1st vs. Fall K

Spring 3rd vs. Fall K

0.67*** 0.08

0.06 0.13*

0.38*** 0.30***

0.62*** 0.05

0.04 0.14*

0.39*** 0.30***

0.57*** 0.04

0.03 0.14*

0.37*** 0.28***

0.45***

0.76***

1.08***

0.43***

0.77***

0.87***

0.42***

Mother’s edu (6HS) Some college BA or more

0.13*** 0.35***

0.01 0.02

Father’s edu (6HS) Some college BA or more

0.12*** 0.34***

1.10***

Intercept

Slope (b)

(a)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

0.65*** 0.10

0.10 0.11

0.44** 0.32**

0.78***

0.91***

0.46***

0.81***

0.08** 0.10**

0.13*** 0.35***

0.02 0.02

0.02 0.01

0.07* 0.02

0.12*** 0.33***

0.02 0.01

0.08* 0.03

0.08** 0.10**

Mother’s employment Employed

0.05*

0.02

0.02

0.06*

0.02

0.02

Father’s employment Employed

0.11

0.06

0.02

0.11

0.07

0.02

Family income (Lowest) Second low Third Highest

0.13*** 0.17*** 0.25***

0.05 0.00 0.01

0.06 0.01 0.03

0.12** 0.17*** 0.24***

0.05 0.00 0.01

0.06 0.02 0.03

0.25 0.22 0.33

0.26 0.37 0.40

English proficiency (Not well at all) Not well Well Very well Notes: All models also includes controls for child’s gender, child’s birth weight, mother’s marital status, and mother’s age at birth. * p < 0.05. ** p < 0.01. *** p < 0.001.

0.03 0.23 0.29*

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

Slope (b)

(a)

Model 4 Model 3 + English proficiency

(a)

Race (White) East Asian Southeast Asian South Asian

Intercept

Model 3 Model 2 + SES

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

399

0.57 = 14]. Stated differently, Asian children’s initial verbal advantage would have been larger if it weren’t for the fact that they grow up in disadvantageous language backgrounds. In an earlier section, we hypothesized Asian children’s verbal scores grew at a slower rate relative to those of Whites because English classes become more difficult over time and parents with an English deficiency may increasingly have difficulty helping their children meet the requirements of English classes. Yet, contrary to our expectations, White-Asian differences in growth rates increase after we control for mother’s English deficiency. For example, net of controls for mother’s English deficiency, East and Southeast Asian children’s verbal scores grow 1.2 times [.44/.37 = 1.19] and 1.1 times [.32/.28 = 1.14] slower than Whites between the fall of kindergarten and spring of third grade, respectively. We conducted several supplementary analyses to ascertain why controlling for mother’s English proficiency widens the White-Asian differences in growth rates. Table A1 reports the findings from these analyses. If mother’s English deficiency is a sign of limited acculturation, then Asian children born to mothers with limited English proficiency will be more likely to subscribe to cultural norms specific to their region of origin (i.e., Confucianism) which protect their verbal scores from deteriorating at an even faster rate. Specifically, it is a well-established fact that Asian parents are more likely than non-Asian parents to express higher expectations for their children’s educational attainment and emphasize hard work as the means to achieve educational success (Kao, 1995; Sakamoto et al., 2009). Influenced by these views, Asian students spend more time than their non-Asian peers doing homework or studying for tests, which may explain the better than expected verbal outcomes of Asian children (Hsin and Xie, 2014; Peng and Wright, 1994; Sakamoto et al., 2009). We tested this explanation by adding parent’s expectations for their children’s educational attainment and amount of time parents spend on educational activities to Model 4. Cultural determinants explain a considerable portion of the effect of mother’s English deficiency on White-Asian differences in initial verbal scores, but they fail to explain its effect on White-Asian differences in growth rates. Alternatively, the unexpected finding may arise because mothers with limited English proficiency are more likely than those with greater English proficiency to reside in ethnic enclaves and to send their children to public schools with large minority populations which offer ESL classes and service for non-English speaking students. These programs will provide Asian students from disadvantageous linguistic backgrounds the resources necessary to ensure that their verbal advantage does not deteriorate at an even faster pace. We tested this possibility by adding three school characteristics- whether or not they attended a public school, the number of full-time faculty who teach ESL, and percent of minority students in school- to Model 4. School characteristics partly explain why controlling for mother’s English proficiency widens White-Asian differences in growth rates. They account for half [(0.40 + 0.44)/(0.37 + 0.44)  0.50] of the increase in White-East Asian differences in growth rates and all of the increase in White-South Asian growth rates following the addition of mother’s English deficiency. Overall, the findings from the supplementary analyses suggest that initial verbal scores are determined largely by parent’s socioeconomic and cultural resources, but changes in verbal scores after school entry are partly determined by availability of resources in school. 4.4. White-Asian differences in verbal development in Australia Next, we assess whether the patterns of White-Asian differences in verbal development observed in the US extends to children born in Australia. In Fig. 2, we see a pattern that is opposite of that observed in the United States. AsianAustralian children have lower verbal scores than Whites at the time of school entry, but their scores grow at a faster pace than those of Whites, resulting in a narrowing of the White-Asian gap as children age. This pattern is consistent across all Asian subgroups in Australia. The Asian subgroups differ only with respect to the size of the initial disadvantage and rates of growth. At age 4, the verbal scores of East Asians are 0.56 standard deviations below those of Whites. This gap persists between ages 4 and 6, but narrows down considerably between ages 6 and 8 and the verbal scores of East Asians surpass those of Whites by age 8. South Asians have an initial verbal disadvantage relative to Whites which is similar in magnitude as that of East Asians. This gap narrows down considerably between ages 4 and 6. As a result, South Asian children’s verbal disadvantage relative to Whites is considerably smaller by age 8 than their initial verbal disadvantage. Southeast Asians start out with a considerably larger disadvantage relative to the other Asian groups. This gap diminishes steadily across all age segments, with the rates of growth being particularly steep between ages 4 and 6. Although their verbal disadvantage has diminished considerably by age 8, they continue to have a verbal disadvantage relative Whites and other Asian subgroups given their large initial verbal disadvantage. It is noteworthy that the pattern of variation in initial scores differs in Australia and the United States: Asian American students have an initial advantage; whereas, Asian-Australian students have an initial disadvantage. Due to the paucity of research on the educational attainment of Asian-Australian children, it is hard to pinpoint exactly what factor gives rise to these cross-national differences. Nonetheless, we speculate that this cross-national difference arises because of two reasons. First, due to the higher prevalence of underemployment in Australia, Asian immigrants in Australia do not have a clear socioeconomic advantage relative to Whites (Zeng and Xie, 2004; Green et al., 2007). As a result, Asian-Australian children may not have access to the educational resources necessary to overcome their disadvantageous language backgrounds. Second, the datasets rely on distinct instruments to assess children’s verbal skills: the IRT in ECLS-K and the PPVT in LSAC-K. For children in kindergarten, IRT is a test of letter recognition and basic vocabulary; whereas, the PPVT is a vocabulary test (Rock et al., 2002). Teaching letter recognition and basic vocabulary may be a much easier task for mothers with an

400

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

Standardized Verbal Scores

1.2

0.8

0.4

6(6.3)

8(8.2)

0.0 4 (4.3)

-0.4

-0.8 White

East Asian

Southeast Asian

South Asian

Notes: Predicted scores are computed using the coefficients from Model 1. Fig. 2. White-Asian differences in standardized verbal scores, Australia.

English deficiency than it is to teach complex vocabulary. It is, therefore, possible that Asian children in both countries may be better at letter recognition but have a more limited vocabulary than their White counterparts. Equally noteworthy is the fact that White-Asian differences in verbal scores narrows down as children age, regardless of the destination country and disadvantaged group. Groups with an initial verbal disadvantage ‘‘catch-up’’ to groups with an initial verbal advantage during their early elementary years. This cross-national similarity supports the view that schools are important equalizers which reduce inequalities in academic and cognitive skills. 4.5. White-Asian differences in parent’s socioeconomic status and English language use in Australia Table 3 describes how parent’s socioeconomic status and mother’s English proficiency differ according to mother’s race in Australia. The results for Australia show a more complicated pattern of socioeconomic differentials than was found in the United States. Like in the United States, East and South Asian children in Australia have parents with higher levels of education and fathers with higher employment rates than White children. For example, 40 percent of East Asian mothers and 45 percent of South Asian mothers have a college degree, as compared with a quarter of White mothers. Over 90 percent of East and South Asian fathers are employed, as compared to about 80 percent of White fathers. Contrary to the United States, however, East and South Asian children grow up in families with lower income than White children. For instance, approximately a third of the East and 37 percent of Southeast Asian children reside in households with the lowest quartile of income, as compared to a quarter of White children. This pattern likely arises due to weaker coupling between legal and employment statuses in Australia where employment is not a pre-requisite for immigrant admission and the consequent high rates of underemployment in Australia (Jasso and Rosenzweig, 2008). Socioeconomic disparities between Southeast Asians and Whites in Australia follow a pattern which is very similar with those observed in the United States. Higher shares of Southeast Asian parents belong in the lowest and highest category of parent’s education. About 30 percent of Southeast Asian mothers are in the lowest educational category (i.e., 10 or fewer years of schooling and no vocational training certificate) and 40 percent of Southeast Asian mothers have college degrees. This compares to 16 and 25 percent of White mothers. Southeast Asian families appear to also have a significant income disadvantage relative to other groups. Almost 40 percent of Southeast Asian families belong in the lowest category of income, which compares to 22 percent of South Asians, 25 percent of Whites, and 31 percent of East Asians. Like in the United States, South Asian mothers in Australia are more proficient in English than Asian mothers from other subgroups. When they are asked how well they speak English, less than 10 percent of South Asian mothers answer ‘‘not well at all’’ or ‘‘not well’’, as compared with 32 percent of East Asian and 24 percent of Southeast Asian mothers. 4.6. Explaining White-Asian differences in verbal development in Australia Table 4 presents the results from four additive growth curve models predicting White-Asian differences in verbal development in Australia. Recall that the main effects of mother’s race (a) measures White-Asian differences in initial verbal scores and the slope (b) captures group differences in verbal growth between a follow-up wave and the baseline (i.e., age 4). We begin with Model 2 because the results obtained from Model 1 were discussed in section 4.4. Model 2 adds demographic controls into the model. Demographic controls have little effect on White-Asian differences in verbal trajectories. Here too, the only exception is observed in the fact that group differences in demographic profiles suppresses some of the verbal disadvantage of East and Southeast Asian children relative to their White counterparts. For example, with demographic controls, differences in initial scores between Whites and East Asians increase by 10 percent

401

K.H. Choi et al. / Social Science Research 52 (2015) 389–407 Table 3 White-Asian differences in parent’s socioeconomic status and mother’s linguistic proficiency, Australia. AUS Born White

FB Asian

(N = 2947)

East (N = 69)

South-east (N = 108)

South (N = 78)

Standardized vocabulary scores Age 4 Age 6 Age 8

0.05 0.05 0.02

0.63 0.49 0.07

0.79 0.27 0.20

0.47 0.08 0.15

Mother’s education 610, no certificate 610, certificate 11–12, no certificate 11–12, certificate Advanced diploma BA or more Total

16 13 21 15 8 25 100

15 2 14 12 18 40 100

30 4 14 7 3 40 100

3 4 8 18 21 45 100

Mother’s employment % employed

62

47

41

48

Father’s education 610, no certificate 610, certificate 11–12, no certificate 11–12, certificate Advanced diploma BA or more Missing Total

9 20 11 16 7 22 14 100

10 1 16 17 9 42 5 100

16 8 12 11 5 34 14 100

2 2 13 9 13 59 2 100

Father’s employment Unemployed Employed Missing Total

4 83 13 100

2 96 3 100

8 81 10 100

4 94 1 100

Household income Lowest Middle Second highest Highest Missing Total

24 38 18 15 5 100

31 29 13 15 11 100

37 25 17 15 7 100

22 43 11 12 10 100

Mother’s spoken English Not well at all Not well Well Very well Total

0 0 0 100 100

6 26 35 34 100

5 19 40 36 100

0 9 27 64 100

Notes: (1) Percentages are weighted. Number of observations is not weighted. (2) The totals may not add to 100 percent due to rounding. (3) Parent’s education and mother’s language proficiency are time-invariant covariates. Father’s and mother’s employment and household income are time-varying covariates. We report the average distribution of time-varying covariates across all waves.

[100 ⁄ (0.64 + 0.58)/0.58 = 10]. Supplementary analysis suggests that this effect accrues in large part because East Asian children have fewer siblings than other groups and number of siblings is negatively associated with academic skills. Stated differently, East Asian children would have an even larger verbal disadvantage relative to Whites if it wasn’t for the fact that they do not have to share their family resources with several siblings. Model 3 adds parent’s socioeconomic status to the existing model. The initial verbal disadvantage of Asian children relative to Whites widens once we control for parent’s socioeconomic status. For instance, initial differences in verbal scores between Whites and South Asians increase by 22 percent [100 ⁄ (0.66 + 0.54)/0.54 = 22] once we introduce controls for parent’s socioeconomic status. Stated differently, Asian children would have an even larger verbal disadvantage relative to their White counterparts had it not been for their parent’s educational advantage. Socioeconomic controls explain some of the White-Asian differences in the pace of growth in verbal scores. For example, the inclusion of socioeconomic controls explains 18 percent [100 ⁄ (0.41  0.34)/0.41  18] of the White-South Asian differences between ages 4 and 6. Stated differently, Asian children are able to close the verbal gap with Whites in part because of their parents’ educational advantage. Recall that Asian parents in Australia have an educational advantage but an income disadvantage relative to Whites. Supplementary analysis shows that income disadvantage explains the initial verbal disadvantage of Asian children while parent’s educational advantage suppresses Asian Australian children’s verbal disadvantage. Asian-Australian parent’s educational advantage explains White-Asian differences in growth rates; whereas Asian-Australian parent’s income disadvantage

402

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

Table 4 Growth curve models predicting standardized verbal scores by parent’s socioeconomic status and English language proficiency, Australia. Model 1 Zero order

Race (White) East Asian Southeast Asian South Asian

Model 2 Model 1 + Demographic controls

Model 3 Model 2 + SES

Intercept

Slope (b)

Intercept

Slope (b)

Intercept

Slope (b)

(a)

Age 6 vs. 4

Age 8 vs. 4

(a)

Age 6 vs. 4

Age 8 vs. 4

(a)

Age 6 vs. 4

0.58⁄⁄⁄ 0.76⁄⁄⁄ 0.56⁄⁄⁄

0.07 0.49⁄⁄⁄ 0.42⁄⁄⁄

0.63⁄⁄⁄ 0.62⁄⁄⁄ 0.42⁄⁄

0.64⁄⁄⁄ 0.79⁄⁄⁄ 0.54⁄⁄⁄

0.09 0.50⁄⁄⁄ 0.41⁄⁄

0.64⁄⁄⁄ 0.62⁄⁄⁄ 0.42⁄⁄

0.69⁄⁄⁄ 0.79⁄⁄⁄ 0.66⁄⁄⁄

Mother’s edu (6HS) 610, certificate 11–12, no certi 11–12, cert Advanced diploma BA or more Father’s edu (6HS) 610, certificate 11–12, no certi 11–12, cert Advanced diploma BA or more Mother’s employment Employed Father’s employment Employed Family income (Lowest) Second low Third Highest English proficiency (Not well at all) Not well Well Very well

0.03 0.45⁄⁄⁄ 0.34⁄⁄

Model 4 Model 3 + English proficiency

Age 8 vs. 4 0.60⁄⁄⁄ 0.60⁄⁄⁄ 0.36⁄⁄

Intercept

Slope (b)

(a)

Age 6 vs. 4

0.37⁄⁄⁄ 0.54⁄⁄⁄ 0.52⁄⁄⁄

0.03 0.44⁄⁄⁄ 0.33⁄

Age 8 vs. 4 0.46⁄⁄ 0.49⁄⁄⁄ 0.30⁄

0.06 0.16⁄⁄ 0.26⁄⁄⁄ 0.28⁄⁄⁄

0.10 0.05 0.09 0.03

0.09 0.02 0.10 0.07

0.05 0.15⁄ 0.24⁄⁄⁄ 0.27⁄⁄⁄

0.10 0.06 0.08 0.04

0.09 0.03 0.09 0.06

0.42⁄⁄⁄

0.00

0.03

0.40⁄⁄⁄

0.02

0.02

0.06 0.04 0.02 0.01

0.05 0.11 0.19 0.12

0.11 0.12 0.07 0.17⁄

0.28⁄⁄⁄

0.20⁄⁄

0.11 0.11 0.07 0.17 0.20⁄⁄

0.15⁄

0.05 0.04 0.02 0.00 0.16⁄

0.05 0.11 0.19⁄ 0.12 0.28⁄⁄⁄

0.07

0.04

0.02

0.06

0.04

0.03

0.13

0.00

0.01

0.12

0.01

0.02

0.01 0.00 0.05

0.06 0.04 0.08

0.02 0.03 0.13

0.01 0.00 0.05

0.05 0.03 0.08

0.40 0.65 1.08⁄⁄

0.20 0.61 0.42

0.11 0.45 0.53

0.03 0.04 0.14⁄

Notes: All models also includes controls for child’s gender, child’s birth weight, mother’s marital status, and mother’s age at birth. * p < 0.05. ** p < 0.01. *** p < 0.001.

suppresses White-Asian differences in growth rates. The fact that parent’s socioeconomic status overall suppresses WhiteAsian differences in initial verbal scores and explains White-Asian differences in growth rates suggest that the effect of parent’s education on children’s verbal scores dominates the effect of household income. Model 4 adds mother’s English language proficiency into the existing model.25 Mother’s English deficiency explains some of the Asian-Australian children’s initial verbal disadvantage. Specifically, controlling for mother’s English proficiency reduces East Asian children’s initial disadvantage in verbal scores by 46 percent [100 ⁄ (0.69 + 0.37)]/0.69  46]; Southeast Asian children’s initial disadvantage by 32 percent [100 ⁄ (0.79 + 0.54)/0.79 = 32]; and South Asian children’s initial disadvantage by 21 percent [100 ⁄ (0.66 + 0.52)/0.66 = 21]. Stated differently, Asian-Australians have lower initial verbal scores, in part, because they grow up in disadvantageous linguistic backgrounds. Interestingly, Asian mother’s English deficiency ‘explains’ why the verbal scores of Asian children increase at a faster pace than those of White children, which is not what we would expect. For example, controlling for mother’s English deficiency reduces approximately 23 percent [100 ⁄ (0.60  0.46)/0.60 = 23] of the White-East Asian difference in growth rates between ages 4 and 8. Based on the insights obtain from our supplementary analysis for the United States, we argue that this unexpected finding may arise because Australia places a heavy emphasis on mastery of working English and offers intense ESL classes to children with limited English proficiency (Clyne, 2005; Iredale and Fox, 1997). The verbal scores of children from disadvantageous language backgrounds may improve at a faster rate due to the additional training in English obtained in these courses. Because children born to mothers with English deficiency are more likely to enroll in and benefit from these 25 In Australia, the trajectories of verbal development of Asian children born to mothers have limited English proficiency follow the same pattern as those of Asian children born to mothers who are proficient or near native speakers of English. The only difference is that the initial disadvantage and the growth rates of Asian children born to mothers with limited English proficiency is larger than those of their peers born to mothers who are proficient in English.

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

403

courses than children born to mothers with greater English proficiency, mother’s English deficiency will explain the faster growth rate of Asian children. Equally possible is the fact that mother’s English deficiency may serve as a proxy for subscription to Asian cultural values, which prevent Asian children from languishing in a state of persistent verbal disadvantage (Kao, 1995; Sakamoto et al., 2009).26 Future studies of Asian Australian children’s verbal development should explore these explanations further.

5. Supplementary analysis Two additional supplementary analyses deserve particular mention. The results from the analysis in this section can be found in the appendix section. 5.1. White-Asian differences in verbal scores beyond middle childhood in the US We assessed whether pattern of White-Asian differences in verbal trajectories continue to hold beyond middle childhood by also considering the verbal scores in 5th and 8th grade. Our general results stay the same even after we expand our analysis to 8th grade. A noteworthy discovery is that Asian children’s verbal scores rebound between 5th and 8th grade; thus, South and East Asian children’s verbal advantage is larger in the 8th grade than it is in the 5th grade and Southeast Asian children’s verbal disadvantage is smaller in the 8th grade than it is in the 5th grade. Table A2 presents the results. 5.2. White-Asian differences in math scores in the United States The Asian academic advantage is usually tested by documenting White-Asian differences in math trajectories. Therefore, we conducted supplementary analyses where we documented differences in math trajectories between Whites and Asian subgroups to establish how the pattern of White-Asian differences in verbal performance compares with the corresponding pattern of differences in math performance. The results are presented in Table A3. As mentioned earlier, the present paper focuses on verbal achievement because (1) Asian advantage in math scores are well-established; (2) immigrant children’s verbal performance helps assess the degree of linguistic and socio-cultural assimilation experienced by the immigrant group, and (3) math scores are collected only at two waves and have high rates of missing data. In the United States, East and South Asian children have an advantage in math performance at the time of school entry and are able to maintain this advantage over time. The math performance of Southeast Asians is on par with that of Whites at the time of school entry and beyond. Like in the United States, Asian children in Australia score higher in math assessments than Whites at ages 6 and 8, with the advantage being particularly more pronounced among East and South Asians.

6. Conclusion The goal of this paper is to shed light on the Asian-American academic advantage by examining whether it extends to verbal skills and whether it extends to countries outside the United States. To this end, we use growth curve models to compare trajectories of verbal development of children born to Asian immigrant mothers with those of native born white children and to ascertain the role that parent’s socioeconomic status and English proficiency play in explaining their trajectories. We also examine whether the patterns observed in the US extend to Asian children born in Australia. Our analyses yield several noteworthy findings. First, we find that Asian American children have a verbal advantage over Whites at the time they enter kindergarten, but this advantage declines over time. East and South Asian children start out with a large verbal advantage over Whites, but this gap narrows considerably by third grade. The initial scores of Southeast Asian children are on par with those of Whites, but their scores grow at a slower pace and they are at a verbal disadvantage relative to Whites by the third grade. These findings are consistent with those by Han (2008). We attribute these findings to the fact that Asian parents are more likely than nonAsian parents to teach their children letter recognition and basic vocabulary before school enrollment, giving rise to the Asian children’s initial verbal advantage. However, upon school entry, White students learn the material that their Asian peers learned ahead of time and ‘‘catch-up’’ to their Asian American peers. Second, the initial verbal advantage of East and South Asian children in the United States is partly due to the higher socioeconomic status of Asian parents. Parent’s socioeconomic status explains about a tenth of East Asian children’s and a fifth of South Asian children’s verbal advantage at the time of entry into kindergarten. Parent’s socioeconomic status, however, exerts little influence on White-Asian differences in growth rates in the United States. 26 We conducted sensitivity analysis using growth curve models which further disaggregated Asian children into two groups: those whose mothers have limited English proficiency and those whose mothers report being proficient or near native speakers of English. The trajectories of verbal development of Asian American children born to mothers with limited English proficiency follow the same pattern as those of Asian American children born to mothers who are proficient in English. The only difference is that the initial advantage is greater and the rate of decline is steeper among Asian American children born to mothers with limited English proficiency is smaller than those of their peers born to mothers who are proficient in English.

404

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

Third, Asian children in the United States would have an even larger initial verbal advantage relative to their White counterparts had it not been for their mother’s limited English proficiency. This finding suggests that disadvantageous language backgrounds suppress children’s initial verbal scores. Fourth, contrary to our expectations, Asian American children’s verbal scores grow at an even slower rate when we control for mother’s English proficiency. Supplementary analyses suggest that this unexpected pattern partly arises because children of mothers with limited English proficiency are more likely than those with mothers with English proficiency to attend public schools with large minority population which offer ESL programs. ESL programs may offer students from disadvantaged language backgrounds the resources necessary to protect their verbal advantage from eroding at an even faster rate. Fifth, the pattern of White-Asian differences in verbal development observed in the United States does not extend to Australia. In fact, the opposite pattern is observed in Australia. Asian children in Australia score lower in verbal assessment tests at the time of school entry, but their scores grow at a faster pace relative to Whites. Sixth, Asian Australian children’s initial verbal disadvantage increase once we control for parents’ socioeconomic status. Stated differently, Asian children in Australia would have an even larger initial verbal disadvantage relative to their White counterparts had it not been for their parent’s educational advantage. Parent’s educational advantage also partially accounts for the faster rate of growth in Asian children’s verbal scores. Seventh, our results show that mother’s English deficiency explains a considerable portion of Asian-Australian children’s initial verbal disadvantage. Stated differently, like in the United States, disadvantageous language backgrounds suppress children’s initial verbal scores. Yet, contrary to expectations, we also find that mother’s English deficiency explains a portion of the faster pace of growth Asian Australian children’s verbal scores. Drawing from the results of our supplementary analyses for the United States, we speculate that one of the reasons why Asian-Australian children are able to catch-up to their White peers once they enter school is because Australian schools offer students from disadvantage language backgrounds the resources necessary to avoid languishing in a state of persistent verbal disadvantage. However, we acknowledge that we do not have the data necessary to empirically test this claim and recommend that future studies make efforts to identify the institutional factors (e.g., educational policy, school environment) which facilitate English acquisition for children of immigrant families, especially those who settled outside of the United States. Such efforts can be accomplished with cross-national data or state/county level data comparing the school environment and educational policies aimed at providing effective instruction for English language learners. Finally, despite cross-national variation in White-Asian differences in initial scores and speed of growth in verbal scores, the story that we observe across these two settings is one of convergence. That is, in both countries, groups with an initial disadvantage- Whites in the United States and Asian-Australians- are able to catch-up to groups with an initial advantage. This observation leads us to believe that schools are fulfilling their role as the great equalizer, impeding the reproduction of social inequality by providing disadvantaged children resources unavailable at home (Mare, 1995; Brand and Xie, 2010). We recommend that this hypothesis be explored further once comparable data on school contexts becomes available. A few limitations of our analyses should be kept in mind when interpreting our findings. First, as mentioned earlier, the assessments in the two datasets are not fully comparable. Although studies have not compared how the use of distinct instruments affects our predictions about White-Asian differences in verbal development, studies have shown that PPVT test produces a larger black-white gap in verbal development than do composite scores, including reading IRT scores (Rouse et al., 2005). If this pattern holds for White-Asian differences in verbal development, then the use of PPVT tests may overstate the magnitude of White-Asian differences in verbal development in Australia. To minimize this problem, we focus on WhiteAsian variations within countries and later compare the pattern of variation across countries. Second, the composition of reading IRT scores change over time, with heavier emphasis being placed on reading comprehension and critical stance over time. Although IRT scores are highly correlated across waves, it is possible that Asian children may fare worse in assessments of reading comprehension and critical stance (which require a comprehensive understanding of the English language) than in assessments of letter recognition or vocabulary acquisition (which require memorization). Although our findings are consistent with other studies of White-Asian differences in verbal performance, it is possible that the differential weights assigned to the various dimensions of verbal ability in the IRT tests may overstate the degree of convergence in verbal performance between Whites and Asians. Third, the assessments were collected at different ages (i.e., 5.7, 7.3, and 9.1 for the US; and 4.2, 6.3, and 8.3 for Australia), which may partly explain differences in growth patterns. Fourth, our study focuses on a specific aspect of academic achievement – ‘‘verbal achievement’’ – because this is the only achievement outcome available prior to school entry in Australia and the United States and available at 3 or more time points in the two datasets. We recognize that other dimensions of academic performance, such as math skills, teacher evaluations and grades, and socio-behavioral skills, also influence children’s overall academic performance. We also acknowledge that White-Asian disparities in the trajectories of these dimensions may differ from those of verbal test scores and that parent’s socioeconomic status and mother’s English deficiency may have a distinct impact on these other outcomes. We recommend that future work examine White-Asian disparities in the trajectories of these academic dimensions and answer these questions. Finally, a key finding in our study is that the verbal scores of Asian children typically converge with those of White between school entry and middle childhood, regardless of which racial group has the advantage. This finding suggests that schools may play important roles in determining White-Asian differences in verbal development. However, due to the absence of comparable information about school environments in the two datasets, we are unable to assess the role of schools in generating White-Asian differences in

405

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

verbal development. We recommend that future work examine the extent to which contextual factors, such as school environment and neighborhood characteristics, account for White-Asian differences in verbal trajectories. In conclusion, our findings offer a more nuanced picture about the academic performance of Asian children. With respect to verbal skill, Asian children do not appear to have a clear advantage over Whites during early and middle childhood. Rather, the educational performance of Asian children is an age- and context-specific phenomenon that is shaped by parent’s socioeconomic status and English proficiency. Acknowledgments An earlier version of this paper was presented in the 2012 Annual Meeting of the Population Association of America. We wish to thank Krista Perreira and Margot Jackson for their helpful comments. This paper uses unit record data from the Longitudinal Study of Australian Children (LSAC) Survey. The LSAC Project was initiated and is funded by the Australian Government Department of Families, Housing, Community Services and Indigenous Affairs (FaHCSIA) and is managed by the Australian Institute of Family Studies (AIFS). The findings and views reported in this paper, however, are those of the authors and should not be attributed to FaHCSIA or AIFS. Appendix A See Fig. A1 and Tables A1–A3. 0.80

Standardized Math Scores

0.68 0.60 0.48

0.46 0.40 0.31 0.22 0.20 0.10 0.00 Age 6

Age 8

-0.07

-0.11

-0.20 White

East Asia

Southeast Asia

South Asia

Notes: Scores predicted from Model 1. Fig. A1. White-Asian differences in standardized math scores, United States.

Table A1 Growth curve models predicting standardized verbal scores: Possible explanations for effects of mother’s English proficiency. Intercept

Slope (b)

(a)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

A. Model 3: Dem controls + SES Race (White) East Asian Southeast Asian South Asian

0.57⁄⁄⁄ 0.04 0.87⁄⁄⁄

0.03 0.14⁄ 0.42⁄⁄⁄

0.37⁄⁄⁄ 0.28⁄⁄⁄ 0.78⁄⁄⁄

B. Model 4: Dem controls + SES + Proficiency Race (White) East Asian Southeast Asian South Asian

0.65⁄⁄⁄ 0.10 0.91⁄⁄⁄

0.10 0.11 0.46⁄⁄⁄

0.44⁄⁄⁄ 0.32⁄⁄⁄ 0.81⁄⁄⁄

C. Model 4A1: Model 4 + Parental expectations + time Race (White) East Asian 0.61⁄⁄⁄ Southeast Asian 0.06 South Asian 0.86⁄⁄⁄

0.09 0.10 0.45⁄⁄⁄

0.43⁄⁄⁄ 0.32⁄⁄⁄ 0.80⁄⁄⁄

D. Model 4A2: Model 4 + School characteristics Race (White) East Asian 0.64⁄⁄⁄ Southeast Asian 0.08 South Asian 0.91⁄⁄⁄

0.09 0.14⁄ 0.45⁄⁄⁄

0.40⁄⁄⁄ 0.24⁄⁄ 0.78⁄⁄⁄

406

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

Table A2 Growth curve models predicting standardized verbal scores, US children from kindergarten to 8th grade. Intercept

Slope (b)

(a)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

Spring 5th vs. Fall K

Spring 8th vs. Fall K

0.66⁄⁄⁄ 0.07 1.08⁄⁄⁄

0.03 0.13⁄ 0.44⁄⁄⁄

0.33⁄⁄ 0.29⁄⁄⁄ 0.76⁄⁄⁄

0.33⁄⁄ 0.30⁄⁄⁄ 0.81⁄⁄⁄

0.22 0.20⁄ 0.63⁄⁄⁄

Model 2: Model 1 + Dem controls Mother’s race (White) East Asian 0.60⁄⁄⁄ Southeast Asian 0.06 South Asian 1.06⁄⁄⁄

0.02 0.14⁄ 0.43⁄⁄⁄

0.35⁄⁄⁄ 0.29⁄⁄⁄ 0.76⁄⁄⁄

0.35⁄⁄ 0.31⁄⁄⁄ 0.82⁄⁄⁄

0.25 0.22⁄⁄ 0.65⁄⁄⁄

Model 3: Model 2 + SES Mother’s race (White) East Asian Southeast Asian South Asian

0.55⁄⁄⁄ 0.03 0.85⁄⁄⁄

0.01 0.14⁄ 0.42⁄⁄⁄

0.34⁄⁄ 0.27⁄⁄⁄ 0.78⁄⁄⁄

0.33⁄⁄ 0.29⁄⁄⁄ 0.84⁄⁄⁄

0.23 0.20⁄ 0.67⁄⁄⁄

Model 4: Model 3 + Proficiency Mother’s race (White) East Asian 0.64⁄⁄⁄ Southeast Asian 0.09 South Asian 0.89⁄⁄⁄

0.08 0.10 0.45⁄⁄⁄

0.42⁄⁄⁄ 0.32⁄⁄⁄ 0.81⁄⁄⁄

0.45⁄⁄ 0.37⁄⁄⁄ 0.90⁄⁄⁄

0.36⁄ 0.28⁄⁄ 0.73⁄⁄⁄

Model 1: Zero order Mother’s race (White) East Asian Southeast Asian South Asian

Table A3 Growth curve models predicting standardized math scores, US children from kindergarten to 3rd grade. Model 1 Zero order

Race (White) East Asian Southeast Asian South Asian

Model 2 Model 1 + Demographic controls

Model 3 Model 2 + SES

Model 4 Model 3 + English proficiency

Intercept

Slope (b)

Intercept

Slope (b)

Intercept

Slope (b)

Intercept

Slope (b)

(a)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

( a)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

(a)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

(a)

Spring 1st vs. Fall K

Spring 3rd vs. Fall K

0.54⁄⁄⁄ 0.22⁄⁄ 0.50⁄⁄⁄

0.25⁄⁄ 0.07 0.18

0.03 0.04 0.04

0.48⁄⁄⁄ 0.17⁄ 0.49⁄⁄⁄

0.23⁄⁄ 0.07 0.16

0.03 0.03 0.02

0.45⁄⁄⁄ 0.07 0.28⁄

0.21⁄ 0.07 0.12

0.01 0.04 0.03

0.44⁄⁄⁄ 0.05 0.29⁄

0.27⁄⁄ 0.10 0.16

0.05 0.02 0.02

Mother’s edu (6HS) Some college BA or more

0.16⁄⁄⁄ 0.36⁄⁄⁄

0.01 0.04

0.04 0.04

0.16⁄⁄⁄ 0.36⁄⁄⁄

0.00 0.04

0.05 0.05

Father’s edu (6HS) Some college BA or more

0.14⁄⁄⁄ 0.35⁄⁄⁄

0.01 0.03

0.03 0.02

0.14⁄⁄⁄ 0.35⁄⁄⁄

0.01 0.03

0.03 0.02

Mother’s employment Employed

0.03

0.02

0.02

0.03

0.02

0.02

Father’s employment Employed

0.10

0.04

0.03

0.10

0.05

0.03

Family income (Lowest) Second low Third Highest

0.14⁄⁄⁄ 0.19⁄⁄⁄ 0.29⁄⁄⁄

0.02 0.01 0.01

0.04 0.01 0.03

0.14⁄⁄⁄ 0.19⁄⁄⁄ 0.29⁄⁄⁄

0.02 0.00 0.01

0.04 0.02 0.04

0.02 0.11 0.06

0.07 0.09 0.14

English proficiency (Not well at all) Not well Well Very well

0.37 0.25 0.26

References Bloemraad, I., de Graauw, E., 2013. Patchwork policies: immigrant integration in the United States. Can. Issues, 67–72. Brand, Jennie E., Xie, Yu, 2010. Who benefits most from college? Evidence for negative selection in heterogeneous economic returns to higher education. Am. Sociol. Rev. 75 (2), 273–302. Calderon, M., Slavin, R., Sanchez, M., 2011. Effective instruction for English learners. Future Child. 21 (1), 103–126.

K.H. Choi et al. / Social Science Research 52 (2015) 389–407

407

Case, A., Paxson, C., 2001. Mothers and others: who invests in children’s health? J. Health Econ. 20 (3), 301–328. Castles, S., Miller, M.J., 2009. The Age of Migration: International Population Movements in the Modern World, fourth ed. Palgrave MacMillan, Basingstoke. Clyne, M., 2005. Australia’s Language Potential. University of South Wales Press, Sydney. Cobb-Clark, D., Nguyen, Ha Trong, 2010. Immigration Background and Intergenerational Correlation in Education. IZA Working Papers. Coley, R., McPherrani, C., Sims, J., Votruba-Drzal, E., 2013. Early education and care experiences and cognitive skills development: a comparative perspective between Australian and American children. Family Matters 93, 36–49. Duncan, Greg, Magnuson, K., 2005. Can family socioeconomic resources account for racial and ethnic test score gaps? Future Child. 15 (1), 35–54. Eaton, Martin, Dembo, Myron, 1997. Differences in the motivational beliefs of Asian American and non-Asian students. J. Educ. Psychol. 89 (3), 433–440. Farkas, G., Beron, K., 2004. The detailed age trajectory of oral vocabulary knowledge: differences by class and race. Soc. Sci. Res. 33, 464–497. Fryer, Roland G., Levitt, Steven D., 2004. Understanding the Black-White test score gap in the first two years of school. Rev. Econ. Stat. 86 (2), 447–464. Fryer, Roland G., Levitt, Steven D., 2006. The Black-White test score gap through third grade. Am. Law Econ. Rev. 8 (2), 249–281. Glick, J., White, M., 2003. The academic trajectories of immigrant’s youths: analysis within and across cohorts. Demography 40 (4), 759–783. Goyette, Kimberly, Xie, Yu, 1999. Educational expectations of Asian American youths: determinants and ethnic differences. Sociol. Educ. 72 (1), 22–36. Green, C., Kler, P., Leeves, G., 2007. Immigrant overeducation: evidence from recent arrivals to Australia. Econ. Educ. Rev. 26 (4), 420–432. Hammond, J., 2001. Literacies in school education in Australia: disjunctions between policy and research. Language Educ. 15 (2–3), 162–177. Han, Wen-Jui, 2008. The academic trajectories of children of immigrants and their school environments. Dev. Psychol. 44 (6), 1572–1590. Hart, B., Risley, T., 1995. Meaningful Differences in Everyday Experience of Young American Children. Brookes, Baltimore. Hsin, A., Xie, Y., 2014. Explaining Asian Americans’ academic advantage over whites. Proc. Natl. Acad. Sci. 111 (23), 8416–8423. Iredale, R., Fox, C., 1997. The impact of immigration on school education in New South Wales, Australia. Int. Migr. Rev. 31 (3), 655–669. Jasso, G.M., Rosenzweig, 2008. Selection Criteria and the Skill Composition of Immigrants: A Comparative Analysis of Australian and US Employment Immigration. IZA Discussion Papers, No. 3564. . Kao, Grace., 1995. Asian Americans as model minorities? A look at their academic performance. Am. J. Educ. 103 (2), 121–159. Kao, Grace, Thompson, Jennifer S., 2003. Racial and ethnic stratification in educational achievement and attainment. Annu. Rev. Sociol. 29, 417–442. Mare, R.D., 1995. Changes in educational attainment and school enrollment. In: Farley, R. (Ed.), State of the Union: America in the 1990s. Economic Trends, vol. 5. Russell Sage Foundation, New York, pp. 155–213. Parmir, P., Harkness, S., Super, C., 2008. Teacher or playmate: Asian immigrant and Euro-American parents’ participation in their young children’s daily activities. Social Behav. Pers. 36 (2), 163–176. Peng, S., Wright, D., 1994. Explanation of academic achievement of Asian American students. J. Educ. Res. 87, 346–353. Regets, Mark, 2001. Research and Policy Issues on High Skill Migration. Innovative People: Mobility of Skilled Personnel in National Innovation Systems. OECD Publishing. Rock, D., Pollack, J., Germino Hasuken, E., 2002. Early Childhood Longitudinal Study-Kindergarten Class of 1998–99 (ECLS–K), Psychometric Report for Kindergarten through First Grade, NCES 2002–05. U.S. Department of Education, National Center for Education Statistics, Washington, DC. Rouse, C., Brooks-Gunn, J., McLanahan, S., 2005. Introducing the issue. Future Child. 15 (1), 5–13. Rubino, A., 2010. Multilingualism in Australia: Reflections on Current and Future Research Trends. Sakamoto, Arthur, Goyette, Kimberly A., Kim, ChangHwan, 2009. Socioeconomic attainments of Asian Americans. Annu. Rev. Sociol. 35 (1), 255–276. Shah, P., 2011. Asian Americans’ Achievement Advantage: When And Why Does It Emerge? (Doctoral Dissertation). . Soloff, Carol, Lawrence, David, Johnstone, Robert, 2005. Sample Design. LSAC Technical Paper 1: pp. 12–30. StataCorp, 2013. Stata 13 Base Reference Manual. Stata Press, College Station, TX. Stiefel, L., Schwartz, A., Chellman, C., 2003. Test Score Gaps in New York State Schools: What Do Fourth and Eighth Grade Results Show. Condition Report, Education Finance Research Consortium, New York State Education Department. Sue, S., Okasaki, S., 1990. Asian-American educational achievements: a phenomenon in search of an explanation. Am. Psychol. 45 (8), 913–920. Sun, Yongmin, 2011. Cognitive advantages of East Asian American children: when do such advantages emerge and what explains them? Sociol. Persp. 54 (3), 377–402. Turney, K., Kao, G., 2009. Pre-kindergarten child care and social and behavioral outcomes among children of immigrants. Early Childhood Res. Q. 24, 432– 444. Vukovic, R.K., Lesaux, N.K., 2013. The language of mathematics: investigating the ways language counts for children’s mathematical development. J. Exp. Child Psychol. 115 (2), 227–244. Walsh, James., 2008. Navigating globalization: immigration policy in Canada and Australia, 1945–2007. Sociol. Forum 23 (4), 786–813. Wang, Aubrey H., 2008. A pre-kindergarten achievement gap? Scope and implications. US–China Educ. Rev. 5 (9), 23–31. Washbrook, E., Waldfogel, J., Bradbury, B., Corak, M., Ghanghro, A., 2012. The development of young children of immigrants in Australia, Canada, the United Kingdom, and the United States. Child Dev. 83 (5), 1591–1607. Xie, Yu, Goyette, Kimberly, 2003. Social mobility and the educational choices of Asian Americans. Soc. Sci. Res. 32 (3), 467–498. Xie, Yu, Goyette, Kimberly, 2004. A Demographic Portrait of Asian Americans. Russell Sage Foundation, New York and Washington, DC. Yu, Z., Ruan, J., 2012. Early childhood English education in China. In: Ruan, J., Leung, C. (Eds.), Perspectives on Teaching and Learning English Literacy in China. Springer, New York. Zeng, Z., Xie, Y., 2004. Asian-Americans’ earnings disadvantage reexamined: the role of place of education. Am. J. Sociol. 109 (5), 1075–1108.