Sage Journals: Discover world-class research

Abstract

The importance of family background in determining a student’s academic achievement has long been acknowledged by researchers. Nonetheless, the effect of schooling on this relationship has also been widely investigated. Some studies have shown that family background plays a far stronger role while the effect of schooling is minimal; others have found that schooling has much to do in either reinforcing or mitigating the effect of family background. Through analyzing the PISA (Programme for International Student Assessment) 2018 dataset for Taiwan, the present study explores the moderating effect of schooling on the relationship between family background and academic achievement. To understand the effect of schooling on the relationship between family background and academic achievement, the present study employs two-level hierarchical linear modeling to take into account the fact that in the PISA 2018 dataset, students are nested within classes, and classes are nested within schools. The results indicate that after taking into account school-level factors and their interaction with family background, the effect of family background on academic achievement diminishes significantly. Among the school-level factors examined in this study, the SES composition of the student body of a school appears to be the most significant predictor of academic achievement. Residence-based admissions and perceived teacher interest in students seem to play a role as well.

Keywords

family background schooling academic achievement PISA 2018 Taiwan

Introduction

Conflict sociologists perceive credentialism as an effective mechanism in modern societies to differentiate individuals by controlling and overseeing entry into key occupational positions (Tholen, 2020). They define credentialism as “a closure mechanism” devised to regulate access to crucial roles within the division of labor (Tholen, 2020, p. 286). Essential in this process are academic/professional qualifications and credentials, which ensure that individuals who hold cultural capital have the greatest chance to pass on the advantages of professional status to their children¹ (Ayling, 2021; Jæger, 2022). In other words, individuals who are familiar with the knowledge and skills required by important qualifications and credentials from childhood are more likely to acquire them (Ayling, 2021; Jæger, 2022). Hence, although an important goal of schooling is to promote social equity, it does not remove the special advantage held by individuals from certain socioeconomic (SES) backgrounds² (Bernardi & Gil-Hernández, 2021; Jæger, 2022; Mcdossi, 2023).

In this context, the significance of cultural capital stands out. In Bourdieu’s view, cultural capital can be transformed into economic capital through institutionalized academic qualifications (Ayling, 2021; Jæger, 2022). This is evident in the fact that a holder of a bachelor’s degree typically receives a higher monthly salary than someone who has merely finished secondary education. It is also noteworthy that the academic achievement³ of children from distinct SES backgrounds tends to differ significantly, mainly due to the unequal distribution of cultural capital across these SES backgrounds (Ayling, 2021; Jæger, 2022). As Bourdieu reveals, the child’s academic results from educational activities rely on the cultural capital already invested by parents (Jæger, 2022). This investment heavily depends on economic capital; given that the accumulation of cultural capital is time-consuming, the family must be able to free the child from financial obligations for a significant duration to acquire and accumulate cultural capital (Fiske, 2023). This not only implies that the child’s schooling is extended, but also the usable time for cultural capital accumulation is increased (Fiske, 2023). In essence, what is commonly perceived as innate ability or talent is, to a large extent, nurtured within the domestic sphere; in other words, academic aptitudes are closely associated with home-based education (Jæger, 2022).

As a result, it is not surprising to find that children from higher SES backgrounds tend to perform better academically in empirical studies (e.g., Harju-Luukkainen et al., 2020; Muench et al., 2023; Vázquez-Cano et al., 2020). Furthermore, academic skills such as literacy and numeracy are significantly correlated with labor market outcomes in terms of earnings (Glewwe et al., 2022; Ozawa et al., 2022). In brief, family background serves as an important predictor of academic achievement, and academic achievement in turn plays a significant role in determining an individual’s occupational status and income.

However, the role of educational institutions should not be disregarded. They have a critical function in distributing individuals into different positions in the labor market (Jelonek, 2021; Traini, 2023). It is widely agreed that an individual’s placement in the labor market is largely attributed to the combined effects of family background and educational institutions (Jelonek, 2021). As a matter of fact, a considerable portion of the effect of family background on labor market outcomes is transmitted through educational attainment, often manifesting as academic credentials (Jelonek, 2021). Blau and Duncan’s classical status attainment model also indicates a link between family background and educational attainment, which subsequently contributes to variation in occupational status (Zhou, 2023). Additionally, empirical evidence demonstrates that schools can either reinforce or mitigate the effect of family background on academic achievement (e.g., Ersan & Rodriguez, 2020; Giancola & Salmieri, 2020). Therefore, this study focuses on the moderating effect of schooling on the relationship between family background and academic achievement and uses Taiwan as an illustrative case. The data utilized is the PISA 2018 dataset for Taiwan, which includes 7,243 15-years-old students. The present study aims to answer the following research questions:

What is the effect of family background on academic achievement?

How does schooling moderate the relationship between family background and academic achievement?

By answering these questions, the present study may provide empirical evidence showing the importance of schooling in addressing SES inequality. Furthermore, as the school is a site where government policy can more easily intervene, this study may offer some policy implications that can be applied across societies.

Family Background and Academic Achievement

Lareau (2000) sheds light on how affluent parents foster a positive home environment to enhance their children’s academic skills, equipping them better for school, in her famous book Home Advantage. They employ their economic and cultural assets to enhance educational experiences via tutoring and private lessons (Lareau, 2000). Additionally, due to their familiarity with the educational system, they actively engage with their children’s schooling, scrutinize the quality of education, and even influence academic programs (Lareau, 2000). Such opportunities are often inaccessible to children from lower-SES families. This emphasizes the early advantages of children from higher SES backgrounds.

J. Coleman (1968) similarly contends that the extra educational resources invested by wealthier families, such as tutoring and supplementary schools, can widen the achievement gap. Likewise, Teachman (1987) suggests parents utilize both material and immaterial resources to foster an environment nurturing academic skills and motivation. Drawing from the National Longitudinal Study of the High School Class of 1972 in the U.S., Teachman illustrates how domestic educational resources significantly impact educational attainment.

Sirin’s (2005) meta-analysis further reveals a moderate to strong correlation between SES and academic achievement. However, factors such as how SES is assessed, school level, location, and minority status moderate this relationship. SES is a commonly used indicator of family background and usually represents a person’s or a family’s position in a hierarchy based on their capability to access or control valuable resources such as wealth, power, and social status (Sirin, 2005). Parental education, occupation, and income are common SES indicators, while home resources are used to a lesser extent (Sirin, 2005). As Sirin notes, family SES directly affects academic achievement through home resources and indirectly through social capital crucial for school success (e.g., parent-school collaborations). Additionally, family SES can influence the type of schools attended, closely linked to education quality (Sirin, 2005).

There are also empirical studies that have explored the effect of family background on academic achievement. For example, in a cross-national comparative study, Jerrim (2013) reported that children from disadvantaged family backgrounds in England, the U.S., and Australia were much less likely to have the required academic skills for attending an elite university, and even if they were qualified, it was too expensive for them to attend an elite university. In a comparison of 11 European countries, Mazzonna (2014) showed that family SES during childhood was a strong predictor of old-age economic, cognitive, and health status. Furthermore, education was the most significant factor that explained both within- and between-country heterogeneity (Mazzonna, 2014). Põder et al. (2016) unveiled that in Estonia and Russia, family background significantly explained student test scores and its effect was stronger in urban areas than in rural areas. Vázquez-Cano et al. (2020) discovered that in Canada, Finland, and Singapore, parental education, parental occupation, and parents’ educational role were the most important predictors of children’s reading ability.

In Croatia, Puzić et al. (2016) found that students from higher SES backgrounds scored higher on the science literacy test in PISA 2009. Moreover, possession of cultural capital had a positive effect on test scores (Puzić et al., 2016). Q. Chen et al. (2018) examined the effect of family SES on children’s reading ability in China and uncovered that it had a positive effect on reading ability, but this relationship was mediated by parent-child relationships and moderated by learning motivation. Harju-Luukkainen et al. (2018) compared the mathematical achievement of immigrant students in three different Finnish cities—Helsinki, Espoo, and Vantaa—and the rest of Finland. Harju-Luukkainen et al. showed that among first-generation students, those in Espoo had the highest mean score and SES; this finding may not be surprising since Espoo has attracted a number of immigrant experts in the electrical and information technology industries. On the other hand, among second-generation students, those in the southern parts of Helsinki performed the best and had the highest SES. Furthermore, Harju-Luukkainen et al. (2020) demonstrated that basic mathematical skills students bring to school in the beginning of primary education were the most powerful predictor of mathematical achievement at fourth grade in Finland. Finally, Muench et al. (2023) showed a growing influence of SES on test scores in Sweden and Finland.

In brief, we can see a constant positive relationship between family background and academic achievement across countries. Yet, the effect of family background can be transmitted indirectly through schooling as well. It is also possible that schools reinforce or reduce the effect of family background. The role of schooling is discussed in the next section.

Effect of Schooling

In Equality of Educational Opportunity, J. S. Coleman et al. (1966) demonstrate that there are greater within-school than between-school differences in academic achievement. Second, school factors have little to do with these differences, except the SES makeup of students (J. S. Coleman et al., 1966). Third, family background is a far stronger contributor to achievement differences than any school factor. In short, the findings in the report suggest that school resources do not matter much in determining a student’s academic achievement, especially for white students. J. Coleman (1968) and J. S. Coleman et al. (1966) also specify a number of school inputs that can impact academic achievement, even though their effects are not as strong as family background. From the weakest to the strongest, they are school facilities and curriculum, teacher quality (including educational qualification, teaching experience, verbal skills, etc.), and family backgrounds of fellow students (J. Coleman, 1968; J. S. Coleman et al., 1966).

In contrast to J. S. Coleman et al. (1966), Heyneman and Loxley (1983) contend that the conclusion that the effect of school quality is weaker than the effect of family background on academic achievement is mainly based on research in economically developed countries in Europe and North America as well as Japan. After examining data from economically less-developed countries in Africa, Asia, Latin America, and the Middle East, Heyneman and Loxley point out that primary-school children in these countries have learned much less than their counterparts in wealthy countries within similar amounts of time. Further, there is a positive relationship between a nation’s income level and the effect of family background on academic achievement (Heyneman & Loxley, 1983). Hence, school effects are much greater in low-income countries (Heyneman & Loxley, 1983). There is also empirical evidence from developing countries that schooling has the potential to mitigate the negative effect of disadvantaged family backgrounds (Abbasi et al., 2023; Tan, 2022).

Baker et al. (2002) further complicate the picture by demonstrating that school effects are not as strong as they were in the 1970s in economically less-developed countries. Instead, the relationship between family background and academic achievement and the portion of academic achievement explained by family background is similar across countries. A major reason for the decreasing school effects in economically less-developed countries is the expansion of formal schooling that has resulted in at least a minimum standardization of curriculum, instruction, and other resources in the past decades (Baker et al., 2002). Moreover, schooling has become a common part of life for most children and families and a key to upward social mobility around the world (Baker et al., 2002). Consequently, academic achievement has been taken more seriously by families and students (Baker et al., 2002). As the quality of schools has reached a threshold, which has reduced differences in school resources both across and within countries, family background may play a stronger role now (Baker et al., 2002).

Although past research has tried to determine the respective proportions of the family’s and the school’s contribution to SES disparity in academic achievement, Alexander (2016) argues that we should see the family and the school as co-contributors to these differences. Furthermore, the strong effect of school SES composition, according to Alexander, in fact reflects the influence of family background. That is, family SES largely determines the community where the child resides and the school where he/she attends (Alexander, 2016). However, Alexander indicates that it is also a school influence since when the characteristics of family, neighborhood, and school overlap, especially under conditions of significant neighborhood and school segregation, children from low-income families face a triple disadvantage. Therefore, a socioeconomically integrated school may have the potential to narrow the SES achievement gap; yet, this must be done in an authentic way, namely, lower-SES and minoritized students are not simply placed in low-level remedial programs and segregated from their wealthy peers (Alexander, 2016). For Alexander, schools can serve as either agents of social reproduction or agents of social mobility. Moreover, instead of reinforcing the effect of family background, schools actually do more to lift up lower-SES students academically (Alexander, 2016).

Downey and Condron (2016) hold a similar view and point out that schools can both reproduce (or exacerbate) and compensate for inequality across different SES backgrounds. As a matter of fact, empirical evidence demonstrates that schools actually reduce the SES achievement gap that is already formed before formal schooling (Downey & Condron, 2016). Downey and Condron go further on by stating that if we meticulously isolate school effects, there is unexpectedly scant evidence that high-SES schools are more effective at promoting math and reading learning than low-SES schools.

Carter (2016) adds that we must pay attention to both “material” and “sociocultural” factors (e.g., teachers’ attitudes toward students, relationships between different student groups, students’ sense of belonging, etc.) in the school context to better understand the impact of schooling on academic achievement. Echoing Alexander (2016), Carter indicates that school SES integration does not necessarily benefit minoritized students. In fact, tracking within schools deprives these students of access to high-quality academic programs and opportunities for building interracial friendships, which can be foundations of valuable social capital (Carter, 2016). In addition, tracking within schools affects students’ attitudes toward their peers, classes and teachers as well as teachers’ perceptions of their students. Thus, in addition to material factors, the school’ sociocultural context must be taken into consideration (Carter, 2016).

Empirically, Waldinger’s (2007) cross-national comparison indicated that ability tracking at the country-level did not reinforce the effect of family background on student test scores. In their cross-national comparison of 15 European-Union countries, Martins and Veiga (2010) found SES-related inequality in math achievement in all the countries. Yet, in some countries, this inequality was explained mostly by family background while in others, school SES composition played a more important role (Martins & Veiga, 2010). Sousa et al. (2012) compared 10 OECD countries and showed that school variables such as curriculum, institutional factors, resources, and school/classroom environment had significant effects on math and science test scores. Raitano and Vona (2013) investigated peer effects in OECD countries. An important finding was that peer heterogeneity had a negative relationship with student test scores in the comprehensive educational systems, but exerted a positive effect in the early-tracking ones (Raitano & Vona, 2013). In addition, the effect of peer heterogeneity was stronger for low-ability students in both types of educational systems (Raitano & Vona, 2013). Using the PISA 2012 database, Hansen and Strietholt (2018) uncovered that although schools did widen the achievement gap between low- and high-SES students, the effect size was small.

Põder et al. (2016) revealed that in Estonia and Russia, merit-based school admissions policies intensified the effect of family background. Based on a comparison of the PISA 2015 math scores of students in Indonesia, Japan, and Algeria, Kartianom and Ndayizeye (2017) pointed out that both family and school SES could significantly predict student test scores in Indonesia and Japan, but not in Algeria. Põder et al. (2017) compared the effect of family background on student test scores in Sweden, Finland, and Estonia and found that family background was a strong determinant of test scores. Nevertheless, the effect of family background was moderated by school choice policy: while matching students and schools based on residency reduced the effect of family background, achievement-based admissions reinforced it (Põder et al., 2017).

Fertig and Schmidt (2002) unveiled that school physical conditions and teacher shortages had strong effects on U.S. students’ reading ability. Fertig (2003) showed that in the German context, higher student-teacher ratios and teacher shortages had negative effects on academic achievement. In another study, Willms (2006) uncovered that in Switzerland, ability tracking considerably impacted students’ reading performance. Alacacı and Erbaş (2010) further revealed that in Turkey, school characteristics explained more variance in student math performance than family background. Moreover, selectivity in admissions, time for math learning, and geographical regions were significant explanatory variables (Alacacı & Erbaş, 2010). On the other hand, Ersan and Rodriguez (2020) confirmed that while SES was still a strong predictor of math achievement, school SES composition also positively influenced school environment and academic achievement in Turkey. In Hong Kong, Sun et al. (2012) demonstrated that school size, school SES composition, and instruction time for science per week had positive effects on student science scores.

In the Australian context, Mahuteau and Mavromaras (2014) examined the effects of private and public schools on student test scores and uncovered that after controlling for school- and student-level factors, there was no significant difference between private and public schools, but Catholic school students performed better than students from both types of schools. In the U.S., Morgan and Jung (2016) found that school expenditures and facilities had very weak correlations with test scores, high school graduation rates, and college attendance. Chesters (2019) further showed that the average test score of low-SES students in high-SES schools was higher than that of their counterparts in low-SES schools. Furthermore, low-SES students were less likely to pursue higher education than their high-SES peers (Chesters, 2019). Net of individual SES and test scores, student in low-SES schools were less likely to go to college (Chesters, 2019). The mediating role of schools in the relationship between family background and educational attainment was clear in the study (Chesters, 2019).

In Italy, Giambona and Porcu (2018) demonstrated that there was an inverse U-shaped relationship between school size and student test scores in villages and large cities, but it was not true in towns. Furthermore, this relationship was true only when there were two or more other competing schools in the same area (Giambona & Porcu, 2018). In addition, Graziosi et al. (2018) employed a pseudo-panel approach to combine the TIMSS 2007 and PISA 2012 datasets and reported that low-SES and vocational school students performed worse. Moreover, students in northern Italy outperformed students in the south (Graziosi et al., 2018). Giancola and Salmieri (2020) revealed that family background, school choice, and school SES were strong explanatory factors for student learning outcomes in Italian upper secondary schools. Moreover, there were regional differences: school choice had a stronger effect in the south and southern islands while the effect of school SES was stronger in the central region and the northeast (Giancola & Salmieri, 2020).

As shown above, the role of schooling may differ based on the national context. Nonetheless, there seems to be more evidence that schools do have significant effects on academic achievement. It is also likely that discrepancies between research findings can be at least partly attributed to how school effects are operationalized. As Carter (2016 suggests, we should not merely look into the material aspect of schooling; rather, both material and sociocultural factors must be taken into account to paint a more holistic picture. Based on the literature, factors need to be considered in the present study are: (1) geographic and demographic characteristics such as school location, school size, school SES composition, and student-teacher ratio; (2) quality factors such as educational resources, teacher quality, and instruction time per week; and (3) sociocultural factors such as admissions policy, within-school tracking, and relationships between students and teachers.

Taiwanese Educational System in Brief

In the Taiwanese educational system, students entering high school are divided into either general or vocational tracks. While general high school students typically focus on academic courses, vocational high school students receive vocational training (Han et al., 2024). In terms of prestige and quality, public general high schools are positioned at the hierarchy’s summit, whereas private vocational high schools are placed at the bottom (Chang & Lin, 2015; S. C. Hsieh, 2020; Peng et al., 2011). Similar to many other societies, empirical evidence in Taiwan indicates a connection between family background and school type attended. Specifically, higher family SES heightens the chances of enrollment in public general high schools, while lower SES associates positively with attending private vocational high schools (K. J. Chen et al., 2017; S. C. Hsieh, 2020; Huang & Hwang, 2014; Lee & Hwang, 2010).

In the Taiwanese educational landscape, not only secondary but also tertiary education is stratified; public universities are not only ranked higher academically but are also more cost-effective compared to private ones (Chang & Lin, 2015; S. C. Hsieh, 2020; Peng et al., 2011). Graduating from a public university typically results in more prestigious occupational status (Hwang & Lin, 2016). Given the prominence of college entrance exams in determining university attendance in Taiwan, and considering that public high schools tend to achieve better results in these exams, school type attended profoundly influences a student’s educational path (Chang & Lin, 2015). Generally speaking, higher-SES students are more likely to attend public universities (Chang & Lin, 2015; Lin, 2020). Hence, it is evident that students from lower SES backgrounds face a twofold disadvantage as they often attend lower-ranked private high schools and colleges with higher tuition fees due to their lower academic achievement (Chang & Lin, 2015; S. C. Hsieh, 2020; Lin, 2020). Subsequently, educational attainment significantly shapes labor market outcomes. In short, students from more privileged backgrounds are more likely to attend more privileged public universities, which in turn provides greater opportunities for securing higher positions in the occupational hierarchy (Chang & Lin, 2015; Chung & Chen, 2011).

Methods

Data Source

The current study draws its data from the results of the Programme for International Student Assessment (PISA) 2018, a triennial international survey targeting 15-year-old students that has been conducted by the Organisation for Economic Cooperation and Development (OECD) in OECD member countries and some partner countries/regions since 2000 (OECD, 2009, 2019). What PISA assesses is students’ competencies in reading, math, and science in both school and non-school contexts (OECD, 2009, 2019). The PISA database contains not only standardized test scores, but also background information on the student, the family, and the school, which enables researchers to carry out in-depth analyses of relationships between academic achievement and various factors (OECD, 2009, 2019). The total number of participating students in the Taiwanese sample is 7,243, of which 3,624 are female and 3,619 are male; two are in eleventh grade, 4,899 are in tenth grade, 2,335 are in ninth grade, and seven are in eighth grade; 6,177 are non-immigrant students, 988 are immigrant students, and 78 are students whose immigration background cannot be identified. It should be noted that in the original dataset, students who have only one immigrant parent is classified as “native.” To better reflect the Taiwanese context in which they are also classified as children of immigrants, they are recoded as “immigrant children” in the present study.

PISA uses a two-stage sampling design, that is, schools are sampled first and then in each participating school, students are sampled (OECD, 2009, 2019). In PISA 2018, there are 192 participating schools in Taiwan. Each student and each school are given a particular weight because: (1) “students and schools in a particular country did not necessarily have the same probability of selection”; (2) “differential participation rates according to certain types of school or student characteristics required various non-response adjustments”; and (3) “some explicit strata were over-sampled for national reporting purposes” (OECD, 2009, p. 36). Hence, statistical analyses using the PISA database must be weighted throughout the entire process to obtain unbiased estimates (OECD, 2009).

Furthermore, a student’s competencies in reading, math, and science are expressed as plausible values (PVs) and there are 10 PVs for each domain. PVs are generated based on the Rasch Model (OECD, 2009). Briefly speaking, PISA uses a balanced incomplete block design, in which participating students are randomly assigned to a group and then answer one of the available test forms in the group (ETS, 2016; OECD, 2009). There are four 30-minute clusters in each test form and each cluster is a combination of a number of test items (ETS, 2016; OECD, 2009). The Rasch Model estimates a student’s ability based on the number of correct answers and test item difficulties and the student’s ability is expressed as the probability that the student answers an item correctly (OECD, 2009). Therefore, a student does not need to answer the entire item battery as long as there are link items that connect different clusters, which makes it possible to estimate how the student is likely to perform on the items he/she does not answer (OECD, 2009). In general, PVs are random draws from a distribution of possible values of a student’s abilities (OECD, 2009). The OECD average score and standard deviation are designed to be 500 and 100 respectively for all the three domains (OECD, 2009). A major advantage of PVs is that they provide unbiased estimates of student performance (OECD, 2009). OECD (2009) warns that averaging PVs at the student level should never be performed and any analysis involving PVs must follow the following steps:

(1) The required statistic and its respective standard error have to be computed for each plausible value. 81 estimates (one student final weight and 80 replicates for each PV) are necessary to get the final estimate and its standard error. Therefore, an analysis involving 10 plausible values requires 810 estimates. For instance, if a mean needs to be estimated, 810 means will be calculated.

(2) The final mean estimate is equal to the average of the 10 mean estimates.

(3) The final sampling variance is equal to the average of the 10 sampling variances.

(4) The imputation variance, also denoted measurement error variance, is computed.

(5) The sampling variance and the imputation variance are combined to obtain the final error variance.

(6) The standard error is equal to the square root of the error variance. (pp. 118–119).

Variables

In the present study, the dependent variables are student test scores in reading, math, and science in PISA 2018. Based on the main findings of the literature, this study utilizes a number of independent variables, which can be classified into two categories: student-level (level 1) and school-level (level 2) variables. Table 1 presents the details and descriptive statistics of the independent variables. There are four student-level variables: ESCS, gender, grade, and immigration background. The school-level variables can be further divided into three subgroups: (1) school geographic and demographic characteristics, (2) school quality factors, and (3) school sociocultural factors.

Table 1.

Details and Descriptive Statistics of Independent Variables.

Continuous variables
Label	Description	Mean	SD
ESCS	Index of economic, social and cultural status	−0.32 (0.02)	0.91 (0.01)
SCHSIZE	School size	2,194.61 (74.44)	1,624.15 (82.58)
SCHESCS	School ESCS composition	−0.32 (0.02)	0.43 (0.02)
SCHHET	School ESCS heterogeneity	0.80 (0.01)	0.11 (0.01)
LMINSTU	Percentage of linguistic minority students	12.84 (2.11)	17.78 (2.38)
NEEDSTU	Percentage of students with special needs	6.99 (0.82)	8.93 (1.33)
LSESSTU	Percentage of students from socioeconomically disadvantaged homes	18.98 (1.32)	17. 35 (1.56)
PROATCE	Proportion of fully certified teachers	0.92 (0.01)	0.12 (0.01)
STRATIO	Student-teacher ratio	14.97 (0.27)	5.81 (0.47)
CLSIZE	Class size	35.49 (0.33)	7.04 (0.23)
EDUSHORT	Educational material shortage	−0.17 (0.06)	0.76 (0.04)
STAFFSHORT	Educational staff shortage	0.07 (0.06)	0.74 (0.04)
LMINS	Instruction time per week (reading)	253.60 (2.19)	94.99 (6.36)
MMINS	Instruction time per week (math)	236.36 (2.35)	89.77 (5.31)
SMINS	Instruction time per week (science)	187.92 (2.96)	121.3 (4.46)
TEACHINT	Perceived teacher interest in students	0.11 (0.02)	0.90 (0.01)
STUBEHA	Student behavior hindering learning	0.04 (0.08)	1.13 (0.07)
TEACHBEHA	Teacher behavior hindering learning	0.33 (0.09)	1.12 (0.07)
Categorical variables
Label	Description	Number	Cumulative %
GENDER	Gender
GENDER	1 = female2 = male	3,6243,619	50.04100
GRADE	Grade
GRADE	891011	72,3354,8992	0.1032.3399.97100
IMMIG	Immigration background
IMMIG	1 = native2 = immigrantMissing	6,17798878	85.2898.92100
SCHLOC	School location
	1 = village, hamlet or rural area (fewer than 3,000 people)	208	2.87
	2 = small town (3,000 to about 15,000 people)	493	9.68
	3 = town (15,000 to about 100,000 people)	2,288	41.27
	4 = city (100,000 to about 1,000,000 people)	1,991	68.76
	5 = large city (with over 1,000,000 people)	2,183	98.90
	Missing	80	100
COMSCH	Competitor schools in the same area
COMSCH	1 = no other2 = one other3 = two or more otherMissing	2025146,370157	2.799.8997.83100
SCHLTYPE	School type
SCHLTYPE	1 = public2 = privateMissing	4,3632,758122	60.2498.32100
PROTYPE	Study program type
PROTYPE	1 = general2 = vocational	4,5112,732	62.28100
MERITADM	Merit-based admissions
MERITADM	1 = never2 = sometimes3 = alwaysMissing	1,4572,4633,126197	20.1254.1297.28100
RESIDADM	Residence-based admissions
RESIDADM	1 = never2 = sometimes3 = alwaysMissing	2,9212,3331,751238	40.3372.5496.71100
TRACKING1	Students are grouped by ability into different classes
TRACKING1	1 = never2 = for some subjects3 = for all subjectsMissing	1511,7544,664674	2.0826.3090.69100
TRACKING2	Students are grouped by ability within their classes
TRACKING2	1 = never2 = for some subjects3 = for all subjectsMissing	1952,8803,798370	2.6942.4594.89100

Note. Standard errors are in parentheses.

It is worth mentioning that the current study uses the PISA index of economic, social and cultural status (ESCS), which is provided in the original dataset, to measure family background. ESCS is a composite measure that combines three variables: (1) parents’ highest educational attainment (PARED), which is measured using estimated years of schooling converted from parents’ ISCED (International Standard Classification of Education) levels; (2) parents’ highest occupational status (HISEI), which is measured using the international socio-economic index of occupational status (ISEI); and (3) home possessions (HOMEPOS), which is an index of specific household items such as cars, computers, art works, and books available at home (OECD, 2019). In the computation of ESCS, all the three variables are standardized across all participating countries/regions and assigned equal weight (OECD, 2019). The OECD average score and standard deviation for ESCS are 0 and 1, respectively (OECD, 2019). There are two variables that are derived from ESCS in this study: (1) a school’s ESCS composition (SCHESCS) is the average ESCS score of its student body; and (2) a school’s ESCS heterogeneity is measured using the standard deviation for the ESCS scores of its student body.

Models

To explore the effect of schooling on the relationship between family background and academic achievement, this study employs two-level hierarchical linear modeling to better reflect the fact that in the PISA 2018 dataset, students are nested within classes, and classes are nested within schools, which means it is unlikely that observations are independent of one another (OECD, 2009).

The modeling procedure in the present study includes three steps. First, a null model (Model 0) taking into account only student test scores is produced. This model allows us to decompose the variance of test scores into the within-school level and the between-school level (OECD, 2009), which can be expressed as the following set of equations:

Y_{ij} = β_{0 j} + ε_{ij}

(1)

β_{0 j} = γ_{00} + U_{0 j}

where $Y_{ij}$ represents the test score of student $i$ in school $j$ , $β_{0 j}$ denotes the intercept for school $j$ , $ε_{ij}$ is the student-level residual, $γ_{00}$ indicates the overall intercept, and $U_{0 j}$ is the school-level residual. Both $ε_{ij}$ and $U_{0 j}$ are assumed to be normally distributed with a mean of zero while $ε_{ij}$ has a variance $σ^{2}$ , which is equal to the within-school variance, and $U_{0 j}$ has a variance $τ_{0}^{2}$ , which is equal to the between-school variance. Thus, the intraclass correlation ( $ρ$ ), which tells us the proportion of total variance in the outcome variable that can be attributed to the between-school variance, can be determined using the following equation:

ρ = \frac{τ_{0}^{2}}{τ_{0}^{2} + σ^{2}}

(2)

The second step involves the development of a student-level model without school-level variables (Model 1), which allows us to investigate the effect of ESCS on test scores with three control variables: gender, grade, and immigration status. The general form of this model can be written as follows:

Y_{ij} = β_{0 j} + β_{1 j} X_{ij} + ε_{ij}

β_{0 j} = γ_{00} + U_{0 j}

(3)

β_{1 j} = γ_{10}

where $X_{ij}$ denotes the student-level variable ESCS (including gender, grade, and immigration status as the control variables), and $β_{1 j}$ represents the combination of the within-school coefficients of the student-level variables. In this model, the effects of the student-level variables are treated as fixed and constant across schools.

At the third step, school-level variables are added into the model. This full model (Model 2) is used to determine the moderating effects of the school-level variables—geographic and demographic characteristics, quality factors, and sociocultural factors—on the relationship between ESCS and test scores. The general form of the model can be written as follows:

Y_{ij} = β_{0 j} + β_{1 j} X_{ij} + ε_{ij}

β_{0 j} = γ_{00} + γ_{01} W_{j} + U_{0 j}

(4)

β_{1 j} = γ_{10} + γ_{11} W_{j} + U_{1 j}

where $W_{j}$ is the pool of the above-mentioned school-level variables, $γ_{01}$ and $γ_{11}$ denote the effects of the school-level variables on test scores and ESCS, respectively. In this model, the effects of the student-level variables are treated as fixed and constant across schools. Statistical analyses in the current study are performed in the R environment (Version 4.2.1, R Core Team, 2022) with the package “BIFIEsurvey” (Version 3.4-15, BIFIE et al., 2022). The results are presented in Tables 2 to 4 and only the variables of interest and those with statistically significant coefficients (p < .005) are included in the tables. It should be noted that as the sample size increases, the probability of committing a Type I error (claiming that there is a significant effect when there is none) also grows significantly. Therefore, with a large sample size of 7,243 students, the present study sets the significance level at 0.005 rather than the conventional 0.05 to reduce the false positive rate, as suggested by Benjamin et al. (2018).

Table 2.

Estimated Effects of Student-Level and School-Level Factors on Reading Scores.

	Model 0		Model 1		Model 2
Label	$\hat{β}$	s.e.	$\hat{β}$	s.e.	$\hat{β}$	s.e.
Intercept	500.2471***	3.1277	499.4339***	2.5114	497.4063***	2.4003
Student-level factors
ESCS			36.7967***	2.2771	16.2551***	2.4382
School geographic and demographic characteristics
SCHLTYPE					−32.5519***	5.8143
PROTYPE					−15.8037***	3.4430
SCHSIZE					0.0052***	0.0014
SCHESCS					52.7609***	7.0671
School quality factors
LMINS					0.0859**	0.0303
School sociocultural factors
TEACHINT					7.5866***	2.0343
Between-school variance	2,646.8439***	359.873	1,420.6073***	213.5085	73.3006	46.8426
Within-school variance	9,530.2705***	304.662	9,448.7284***	309.0859	9,112.2111***	339.2081
$R_{level 2}^{2}$			0.1447***	0.0324	0.9739***	0.0159
$R_{level 1}^{2}$			0.1012***	0.0126	0.0786***	0.0148
$R_{total}^{2}$			0.1071***	0.0139	0.2750***	0.0203
ICC ( $\hat{ρ}$ )	0.2174***	0.0239	0.1307***	0.0176	0.0080	0.0053
Number of students	7,243		7,243		7,243
Number of schools	192		192		192

***

$p < 0.001$ . ** $p < 0.005$ .

Table 3.

Estimated Effects of Student-Level and School-Level Factors on Math Scores.

	Model 0		Model 1		Model 2
Label	$\hat{β}$	s.e.	$\hat{β}$	s.e.	$\hat{β}$	s.e.
Intercept	528.5298***	3.3076	527.6713***	2.7531	526.4274***	2.5651
Student-level factors
ESCS			38.2457***	2.4548	16.8417***	2.2494
School geographic and demographic characteristics
SCHLTYPE					−37.0963***	6.1159
PROTYPE					−17.2500***	3.4723
SCHESCS					52.9821***	7.8956
School quality factors
CLSIZE					0.9027***	0.2601
MMINS					0.0911**	0.0307
School sociocultural factors
RESIDADM					−7.7775**	2.8207
Between-school variance	2,795.6099***	399.3799	1,489.6303***	217.0976	215.7642	87.4926
Within-school variance	9,434.6509***	440.6737	9,407.7224***	449.9601	9,049.9906***	458.1033
$R_{level 2}^{2}$			0.1326***	0.0262	0.9314***	0.0268
$R_{level 1}^{2}$			0.1003***	0.0133	0.0660***	0.0141
$R_{total}^{2}$			0.1048***	0.0138	0.2774***	0.0205
ICC ( $\hat{ρ}$ )	0.2287***	0.0262	0.1368***	0.0181	0.0233	0.0094
Number of students	7,243		7,243		7,243
Number of schools	192		192		192

***

$p < 0.001$ . ** $p < 0.005$ .

Table 4.

Estimated Effects of Student-Level and School-Level Factors on Science Scores.

	Model 0		Model 1		Model 2
Label	$\hat{β}$	s.e.	$\hat{β}$	s.e.	$\hat{β}$	s.e.
Intercept	513.3298***	3.4258	512.5194***	2.9190	510.2153***	2.7185
Student-level factors
ESCS			35.3185***	2.6072	13.7978***	2.7581
School geographic and demographic characteristics
SCHLOC					−0.8009***	0.2114
SCHLTYPE					−37.2428***	5.7954
PROTYPE					−15.5987***	3.6128
SCHESCS					55.6872***	8.1410
School quality factors
SMINS					0.0961***	0.0228
School sociocultural factors
TEACHINT					6.7317**	1.9148
Between-school variance	2,726.2024***	446.796	1,518.6181***	254.0782	52.8910	50.4562
Within-school variance	9,022.7373***	280.7317	9,066.8079***	301.5811	8,782.5691***	310.0441
$R_{level 2}^{2}$			0.1153***	0.0292	0.9816***	0.0169
$R_{level 1}^{2}$			0.0896***	0.0139	0.0689***	0.0154
$R_{total}^{2}$			0.0933***	0.0148	0.2808***	0.0210
ICC ( $\hat{ρ}$ )	0.2319***	0.0302	0.1434***	0.0211	0.0059	0.0055
Number of students	7,243		7,243		7,243
Number of schools	192		192		192

***

$p < 0.001$ . ** $p < 0.005$ .

Results

Table 2 displays the estimated effects of student-level and school-level factors on reading scores. The null model (Model 0) shows that significant between-school and within-school differences do exist. The between-school variance is 2,646.8439 and the within-school variance is 9,530.2705. The intraclass correlation is thus .2174, which means that about 21.74% of the total variance in reading scores can be explained by school-level factors. In Model 1, we can see that after controlling for gender, grade, and immigration background, an increase of one point on the ESCS index can increase 36.7967 points on the reading scale at the student level. In addition, after taking into account student-level factors, the between-school variance decreases to 1,420.6073 while the within-school variance decreases to 9,448.7284. That is, about 46.33% of the between-school variance and 0.86% of the within-school variance can be explained by student-level factors. Moreover, the intraclass correlation becomes 0.1307.

In the full model (Model 2), it appears that the effect of ESCS shrinks significantly to 16.2551 while still being statistically significant. At the school level, school type, study program type, school size, school ESCS composition, instruction time per week, and perceived teacher interest in students are shown to be effective predictors of reading scores at the school level. Specifically, on average, private schools perform 32.5519 points lower than public schools, whereas vocational programs perform 15.8037 points lower than general programs. In addition, an increase of one point in the average school ESCS results in a 52.7609-point increase on the reading scale. While the effects of school size and instruction time per week are significant, their coefficient are very small, that is, 0.0052 and 0.0859, respectively, and thus may have no practical significance. Perceived teacher interest in students has an effect of 7.5866. Finally, the between-school variance decreases to 73.3006 and becomes insignificant while the within-school variance decreases to 9,112.2111. This indicates that about 97.23% of the between-school variance and 4.39% of the within-school variance are explained by school-level factors. Further, the intraclass correlation decreases to .0080 and becomes insignificant. Overall, the full model explains about 97.39% of the between-school variance ( $R_{level 2}^{2}$ = 0.9739) and 7.86% of the within-school variance ( $R_{level 1}^{2}$ = 0. 0786).

As can be seen in Tables 3 and 4, there are similar trends for math and science scores. However, it should be noted that perceived teacher interest in students appears to have no significant effect on math scores while residence-based admissions have a significant effect of −7.7775 on math scores.

Discussion

The present study explored the relationship between family background and academic achievement and the moderating effect of schooling on this relationship. Through analyzing the PISA 2018 dataset for Taiwan, family background in terms of SES has been found to exert a significant positive effect on academic achievement, that is, students from higher SES backgrounds tend to have higher academic achievement. This finding is consistent with previous studies undertaken in a variety of societies. It is shown in the literature that students from higher SES backgrounds tend to have access to more educational resources at home and to attend more resourceful schools, which positively contributes to their academic achievement (Alexander, 2016; Lareau, 2000). Moreover, middle- and upper-class parents tend to have interpersonal ties with school personnel and thus are able to make schooling more optimal for their children (Lareau, 2000). Nonetheless, after taken into account school-level factors and their interactions with SES, the effect of family background on academic achievement shrinks significantly. This seems to imply that schools have an important role to play in either widening the SES gap or leveling the playing field (Alexander, 2016; Downey & Condron, 2016).

Among school-level factors, in line with previous studies, it appears that school SES composition plays the most prominent role. That is, school SES composition has a strong positive effect on academic achievement. A plausible explanation is that when students from higher SES backgrounds, who enjoy a “home advantage” since childhood as Lareau (2000) argues, concentrate in a certain school, it might not be surprising to find that the school exhibits better academic performance. It is possible that the effect of family background is significantly mediated by school SES composition, which reduces the effect of ESCS, which is employed to measure SES, in Model 2. However, the specific way school SES composition works needs further investigation, which may have the potential to narrow the SES achievement gap.

Another significant finding is that private and vocational schools perform worse than public and general schools for all the three domains. This reflects the stratified nature of the Taiwanese educational system: generally speaking, in terms of quality, prestige, and student performance, public and general schools are ranked higher than private and vocational schools (Chang & Lin, 2015; S. C. Hsieh, 2020) and test scores are one of the most apparent indicators of this fact. Furthermore, residence-based admissions appear to have a negative effect for math scores. This may indicate that students with lower academic achievement in math tend to attend schools in their own communities. Yet, it requires further research to determine whether this finding denotes that significant SES segregation in housing exists in Taiwan. In addition, Perceived teacher interest in students has been found to be a significant predictor of both reading and science scores, which seems to highlight the fact that when students feel teachers care about them, they are more likely to be engaged in learning and achieve higher academically (Valenzuela, 1999). It is noteworthy that there is no significant interaction between ESCS and school-level factors, which seems to indicate that schooling has no significant moderating effect on the relationship between SES and academic achievement.

Conclusions

Implications

As Alexander (2016) and Downey and Condron (2016) contend, schools can either reproduce or compensate for the SES achievement gap. Given that the findings reveal that school-level factors, particularly school SES composition, plays an important role in Taiwan, it is necessary to think about how schools can better serve lower-SES and minoritized students. A critical step may be increasing school SES integration as J. S. Coleman et al. (1966) indicate that attending a higher-SES school helps lower-SES and minoritized students cultivate high educational aspirations. A plausible policy intervention is voucher programs. In fact, efforts have been made to provide more equal educational opportunities through voucher programs in countries such as the United States (Chingos & Peterson, 2012; Howell et al., 2002; Wolf et al., 2013), Colombia (Angrist et al., 2002, 2006), and Chile (C. T. Hsieh & Urquiola, 2006; Murnane et al., 2017). A typical practice is that students from lower SES backgrounds are financially supported to attend better-performing schools attended by more affluent peers. In general, these programs have been found to have a positive impact on educational outcomes including test scores, high school graduation rates, and college enrollment rates (Angrist et al., 2002, 2006; Barnard et al., 2003; Chingos & Peterson, 2012; Howell et al., 2002; Murnane et al., 2017; Wolf et al., 2013). On the other hand, the government may also make an effort to ensure that the majority of schools have certain percentages of higher- and lower-SES students by providing incentives for both groups of students to attend the same school.

Nonetheless, Alexander (2016) and Carter (2016) suggest that school SES integration must be done in an authentic way, namely, lower-SES and minoritized students are not segregated from their peers within the school. Voucher programs do not guarantee that students from lower SES backgrounds are fully integrated within schools. Therefore, how to provide incentives for schools to improve SES integration after voucher programs have successfully motivated more lower-SES and minoritized students to attend higher-SES schools is another issue needs to be address. Given that students from different SES backgrounds may have academic skills at various levels, teaching such a heterogeneous class might be very challenging for teachers. In addition to facilitating school SES integration, how to best support teachers is another important issue policy makers have to consider carefully. A relevant finding is that improving teacher quality, especially the attitudinal aspect, may be necessary as perceived teacher interest in students is shown to have a significant positive effect on reading and science scores. Teacher education programs and relevant teacher workshops may be where the government can enhance and maintain teachers’ interest in students, which may help boost student academic performance.

Moreover, given the significant gaps between public and private high schools as well as between general and vocational programs, it might be necessary for the government to intervene before students reach high school. If it is true that lower-SES students are more likely to attend private and vocational high schools, it might be necessary to provide them with tutoring programs that can compensate for their insufficient cultural capital at home when they are still in elementary and middle school, which may enhance their chances of enrolling in public and general high schools. On the other hand, the government may need to invest more effort into improving the quality of vocational and private schools such as providing more funding and decrease the student-teacher ratio to let their students receive quality education and thus have more opportunities for upward mobility.

Caveats

As Sirin (2005) contends, researchers should strive to incorporate multiple components of SES in their operationalization because relying on a single component tends to overestimate the effect of SES. According, this study employed ESCS that measures multiple dimensions of SES. The use of ESCS in measuring family background is also supported by empirical evidence. For instance, using the PISA 2003 and 2006 databases, Schulz (2005) found that the internal consistency and stability of ESCS is reasonable across countries and years. Further, this composite index explains more variance in student performance in the majority of the participating countries than using parental occupational status, parental educational attainment, and home possessions separately. Nonoyama-Tarumi’s (2008) analysis of the PISA 2000 database shows that a multidimensional SES index combining the effects of parental educational attainment, parental occupational status, and home cultural resources has a stronger explanatory power and a better model fit across both industrialized and developing countries.

On the other hand, although Marks (2011) demonstrates that a composite measure of parental occupational status and educational attainment can better explain student performance than single indicator measures, he does not recommend the inclusion of indicators of home material and cultural resources as they may jeopardize “the distinction between indicators of a concept and mediators of its effects” (p. 244). In addition, Rutkowski and Rutkowski (2013) warn that the ESCS index needs to be used with caution as the home possessions index, one of its three main components, suffers from inconsistent reliability across participating countries, weak model-to-data consistency in a number of countries, and poor cultural comparability. Hence, findings in the present study must be interpreted cautiously.

Concluding Remarks

In sum, the present study aimed to examine (1) the effect of family background on academic achievement and (2) how schooling moderates the relationship between family background and academic achievement. In terms of the first purpose, it is clear that after taking into account school-level factors, family SES still had a significant effect on academic achievement even though its effect size shrank considerably. Regarding the second purpose, given that there was no significant interaction between family background and school-level factors, the moderating effect of schooling cannot be proved in the present study. Nevertheless, school type, study program type, school ESCS composition, and perceived teacher interest in students appeared to be significant predictors of academic achievement. While school type, study program type, school ESCS composition may reflect the facts that the Taiwanese education system is stratified and that SES backgrounds can be reproduced through schooling, this finding also illuminates potential targets for education reform if we are to provide more equal educational opportunities for all students. Moreover, the significant effect of perceived teacher interest in students highlights the importance of the attitudinal aspect of teacher quality, which teacher education programs and in-service training should focus more on.

Footnotes

Acknowledgements

Not applicable

Author Contributions

Not applicable

Declaration of Conflicting Interests

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

Funding

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

Ethics Approval and Consent to Participate

Not applicable.

Consent for Publication

Not applicable.

ORCID iD

Tzung-Ruei Tsou

Data Availability Statement

The datasets generated and/or analyzed during the current study are available in the PISA repository,

Notes

References

Abbasi

B. N.

Yang

Sohail

(2023). Uncovering basic educational skills and intelligence among women and children in rural Pakistan. SN Social Sciences, 3(7), 116.

Alacacı

Erbaş

A. K.

(2010). Unpacking the inequality among Turkish schools: Findings from PISA 2006. International Journal of Educational Development, 30(2), 182–192.

Alexander

(2016). Is it family or school? Getting the question right. RSF The Russell Sage Foundation Journal of the Social Sciences, 2(5), 18–33.

Angrist

Bettinger

Bloom

King

Kremer

(2002). Vouchers for private schooling in Colombia: Evidence from a randomized natural experiment. American Economic Review, 92(5), 1535–1558.

Angrist

Bettinger

Kremer

(2006). Long-term educational consequences of secondary school vouchers: Evidence from administrative records in Colombia. American Economic Review, 96(3), 847–862.

Ayling

(2021). International education and the pursuit of ‘Western’ capitals: Middle-class Nigerian fathers’ strategies of class reproduction. British Journal of Sociology of Education, 42(4), 460–474.

Baker

D. P.

Goesling

LeTendre

G. K.

(2002). Socioeconomic status, school quality, and national economic development: A cross-national analysis of the “Heyneman-Loxley Effect” on mathematics and science achievement. Comparative Education Review, 46(3), 291–312.

Barnard

Frangakis

C. E.

Hill

J. L.

Rubin

D. B.

(2003). Principal stratification approach to broken randomized experiments: A case study of school choice vouchers in New York City. Journal of the American Statistical Association, 98(462), 299–323.

Benjamin

D. J.

Berger

J. O.

Johannesson

Nosek

B. A.

Wagenmakers

E. J.

Berk

Johnson

V. E.

(2018). Redefine statistical significance. Nature Human Behaviour, 2(1), 4–10.

10.

Bernardi

Gil-Hernández

C. J.

(2021). The social-origins gap in labour market outcomes: Compensatory and boosting advantages using a micro-class approach. European Sociological Review, 37(1), 32–48.

11.

BIFIE, Robitzsch

Oberwimmer

(2022). BIFIEsurvey: Tools for survey statistics in educational assessment (Version 3.4-15). BIFIE. https://CRAN.R-project.org/package=BIFIEsurvey

12.

Carter

P. L.

(2016). Educational equality is a multifaceted issue: Why we must understand the school’s sociocultural context for student achievement. RSF The Russell Sage Foundation Journal of the Social Sciences, 2(5), 142–163.

13.

Chang

Y. C.

Lin

T. H.

(2015). How does the expansion of higher education reproduce class inequality? The case of Taiwan. Taiwan Journal of Sociology of Education, 15(2), 85–129.

14.

Chen

K. J.

Lyau

N. M.

Chen

F. C.

(2017). A study of the influence of family education resources and academic performance on educational tracking for economically disadvantaged high school students. Educational Policy Forum, 20(4), 33–62.

15.

Chen

Kong

Gao

(2018). Effects of socioeconomic status, parent-child relationship, and learning motivation on reading ability. Frontiers in Psychology, 9, 1297.

16.

Chesters

(2019). Alleviating or exacerbating disadvantage: Does school attended mediate the association between family background and educational attainment? Journal of Education Policy, 34(3), 331–350.

17.

Chingos

M. M.

Peterson

P. E.

(2012). The effects of school vouchers on college enrollment: Experimental evidence from New York City. The Brookings Institution.

18.

Chung

Y. C.

Chen

Y. L.

(2011). Education and social mobility in Taiwan. Journal of Social Sciences and Philosophy, 23(1), 61–91.

19.

Coleman

(1968). The concept of equality of educational opportunity. Harvard Educational Review, 38(1), 7–22.

20.

Coleman

J. S.

Campbell

E. Q.

Hobson

C. J.

McPartland

Mood

A. M.

Weinfeld

F. D.

York

R. L.

(1966). Equality of educational opportunity. U.S. Department of Health, Education, and Welfare, Office of Education.

21.

Downey

D. B.

Condron

D. J.

(2016). Fifty years since the Coleman Report: Rethinking the relationship between schools and inequality. Sociology of Education, 89(3), 207–220.

22.

Ersan

Rodriguez

M. C.

(2020). Socioeconomic status and beyond: A multilevel analysis of TIMSS mathematics achievement given student and school context in Turkey. Large-scale Assessments in Education, 8(1), 15.

23.

ETS. (2016). PISA 2018 integrated design. PISA, OECD Publishing.

24.

Fertig

(2003). Who’s to blame? The determinants of German students’ achievement in the PISA 2000 study (IZA discussion paper, No. 739). Institute for the Study of Labor (IZA).

25.

Fertig

Schmidt

C. M.

(2002). The role of background factors for reading literacy: Straight national scores in the PISA 2000 study (IZA discussion paper, No. 545). Institute for the Study of Labor (IZA).

26.

Fiske

(2023). Transforming lost time into migration capital: Hazara refugee social and cultural capital development in Indonesia. Journal of Refugee Studies, 36(4), 917–936.

27.

Giambona

Porcu

(2018). School size and students’ achievement. Empirical evidences from PISA survey data. Socio-Economic Planning Sciences, 64, 66–77.

28.

Giancola

Salmieri

(2020). Family background, school-track and macro-area: The complex chains of education inequalities in Italy (DiSSE Working Papers No. 04/2020).

29.

Glewwe

Song

Zou

(2022). Labor market outcomes, cognitive skills, and noncognitive skills in rural China. Journal of Economic Behavior & Organization, 193, 294–311.

30.

Graziosi

Agasisti

De Witte

Pauli

(2018). Evaluation of students’ performance at lower secondary education. An empirical analysis using TIMSS and PISA data [Paper presentation]. Paper presented at the 49th Scientific Meeting of the Italian Statistical Society, Palermo, Italy.

31.

Han

P. L.

Chiou

H. J.

Y. F.

(2024). A multilevel mediation study of students’ reading competence on teachers’ reading activities volume, classroom management, teachers’ reading autonomy supports, and student reading performance: The effects on educational tracking. Contemporary Educational Research Quarterly, 32(1), 3–40.

32.

Hansen

K. Y.

Strietholt

(2018). Does schooling actually perpetuate educational inequality in mathematics performance? A validity question on the measures of opportunity to learn in PISA. ZDM, 50(4), 643–658.

33.

Harju-Luukkainen

Tarnanen

Nissinen

Vettenranta

(2018). Economic, social and cultural status (ESCS) andmath performance of immigrant students’ in the Finnish metropolitan area in PISA 2012. In Garvis

Ødegaard

E. E.

(Eds.), Nordic dialogues on children and families (pp. 108–125). Routledge.

34.

Harju-Luukkainen

Vettenranta

Wang

Garvis

(2020). Family related variables effect on later educational outcome: A further geospatial analysis on TIMSS 2015 Finland. Large-scale Assessments in Education, 8(1), 1–13.

35.

Heyneman

S. P.

Loxley

W. A.

(1983). The effect of primary-school quality on academic achievement across twenty-nine high- and Low-Income countries. American Journal of Sociology, 88(6), 1162–1194.

36.

Howell

W. G.

Wolf

P. J.

Campbell

D. E.

Peterson

P. E.

(2002). School vouchers and academic performance: Results from three randomized field trials. Journal of Policy Analysis and Management, 21(2), 191–217.

37.

Hsieh

C. T.

Urquiola

(2006). The effects of generalized school choice on achievement and stratification: Evidence from Chile’s voucher program. Journal of Public Economics, 90(8-9), 1477–1503.

38.

Hsieh

S. C.

(2020). The construction of social justice indicators for technical and vocational education policy [Unpublished doctoral dissertation, University of Taipei]. National Taiwan Normal University.

39.

Huang

M. F.

Hwang

Y. J.

(2014). Correlation among students’ family background, academic performance in junior high school, and senior high school tracking in Taiwan. Journal of Educational Practice and Research, 27(2), 67–98.

40.

Hwang

Y. J.

Lin

H. M.

(2016). Individual education, accessed and mobilized social capital, and status attainment. Bulletin of Educational Research, 62(2), 97–130.

41.

Jæger

M. M.

(2022). Cultural capital and educational inequality: An assessment of the state of the art. In Gërxhani

de Graaf

N. D.

Raub

(Eds.), Handbook of sociological science: Contributions to rigorous sociology (pp. 121–134). Edward Elgar Publishing.

42.

Jelonek

(2021). Universities and the labour market: Graduate transitions from education to employment. Routledge.

43.

Jerrim

(2013). Family background and access to ‘high status’ universities. Sutton Trust.

44.

Kartianom

Ndayizeye

(2017). What’s wrong with the Asian and African students’ mathematics learning achievement? The multilevel PISA 2015 data analysis for Indonesia, Japan, and Algeria. Jurnal Riset Pendidikan Matematika, 4(2), 200–210.

45.

Lareau

(Ed.). (2000). Home advantage: Social class and parental intervention in elementary education (2nd ed.). Rowman & Littlefield Publishers.

46.

Lee

P. H.

Hwang

Y. J.

(2010). Aboriginals and Hans, family background and their relationship with the Basic Competence Test, and educational tracking: A study in Taitung. Journal of Research in Science Education, 56(1), 193–226.

47.

Lin

Y. H.

(2020). The influence of expanding universities on the investment of higher education human resources. Taiwan Educational Review Monthly, 9(2), 119–134.

48.

Mahuteau

Mavromaras

K. G.

(2014). Student scores in public and private schools: Evidence from PISA 2009 (IZA Discussion Paper, No. 8471). Institute for the Study of Labor (IZA).

49.

Marks

G. N.

(2011). Issues in the conceptualisation and measurement of socioeconomic background: Do different measures generate different conclusions? Social Indicators Research, 104(2), 225–251.

50.

Martins

Veiga

(2010). Do inequalities in parents’ education play an important role in PISA students’ mathematics achievement test score disparities? Economics of Education Review, 29(6), 1016–1033.

51.

Mazzonna

(2014). The long-lasting effects of family background: A European cross-country comparison. Economics of Education Review, 40, 25–42.

52.

Mcdossi

(2023). Inequality reproduction, higher education, and the double major choice in college. Higher Education, 85(1), 157–186.

53.

Morgan

S. L.

Jung

S. B.

(2016). Still no effect of resources, even in the new gilded age? RSF The Russell Sage Foundation Journal of the Social Sciences, 2(5), 83–116.

54.

Muench

Wieczorek

Dressler

(2023). Equity lost: Sweden and Finland in the struggle for PISA scores. European Educational Research Journal, 22(3), 413–432.

55.

Murnane

R. J.

Waldman

M. R.

Willett

J. B.

Bos

M. S.

Vegas

(2017). The consequences of educational voucher reform in Chile (NBER Working Paper No. 23550). National Bureau of Economic Research.

56.

Nonoyama-Tarumi

(2008). Cross-national estimates of the effects of family background on student achievement: A sensitivity analysis. International Review of Education, 54(1), 57–82.

57.

OECD. (2009). PISA data analysis manual: SPSS^® (2nd ed.). PISA, OECD Publishing.

58.

OECD. (2019). PISA 2018 results (volume III): What school life means for students’ lives. PISA, OECD Publishing.

59.

Ozawa

Laing

S. K.

Higgins

C. R.

Yemeke

T. T.

Park

C. C.

Carlson

Y. E.

Guterman

L. B.

Omer

S. B.

(2022). Educational and economic returns to cognitive ability in low- and middle-income countries: A systematic review. World Development, 149, 105668.

60.

Peng

L. H.

Hsung

Chi

C. S.

(2011). How cohort, gender, and gendered majors affect the acquirement of first job’s SES in the context of Taiwan’s higher education expansion. Taiwan Journal of Sociology of Education, 11(2), 47–85.

61.

Põder

Lauri

Ivaniushina

Alexandrov

(2016). Family background and school choice in cities of Russia and Estonia: Selective agenda of the soviet past and present. Studies of Transition States and Societies, 8(3), 5–28.

62.

Põder

Lauri

Veski

(2017). Does school admission by zoning affect educational inequality? A study of family background effect in Estonia, Finland, and Sweden. Scandinavian Journal of Educational Research, 61(6), 668–688.

63.

Puzić

Gregurović

Košutić

(2016). Cultural capital – A shift in perspective: An analysis of PISA 2009 data for Croatia. British Journal of Sociology of Education, 37(7), 1056–1076.

64.

Raitano

Vona

(2013). Peer heterogeneity, school tracking and students’ performances: Evidence from PISA 2006. Applied Economics, 45(32), 4516–4532.

65.

R Core Team. (2022). R: A language and environment for statistical computing (Version 4.2.1). R Foundation for Statistical Computing. https://www.R-project.org

66.

Rutkowski

(2013). Measuring socioeconomic background in PISA: One size might not fit all. Research in Comparative and International Education, 8(3), 259–278.

67.

Schulz

(2005). Measuring the socio-economic background of students and its effect on achievement in PISA 2000 and PISA 2003 [Paper presentation]. Paper presented at the Annual Meeting of the American Educational Research Association, San Francisco.

68.

Sirin

S. R.

(2005). Socioeconomic status and academic achievement: A meta-analytic review of research. Review of Educational Research, 75(3), 417–453.

69.

Sousa

Park

E. J.

Armor

D. J.

(2012). Comparing effects of family and school factors on cross-national academic achievement using the 2009 and 2006 PISA surveys. Journal of Comparative Policy Analysis Research and Practice, 14(5), 449–468.

70.

Sun

Bradley

K. D.

Akers

(2012). A multilevel modelling approach to investigating factors impacting science achievement for secondary school students: PISA Hong Kong sample. International Journal of Science Education, 34(14), 2107–2125.

71.

Tan

(2022). School socioeconomic desegregation and student academic performance: Evidence from a longitudinal study on middle school students in China. Social Psychology of Education, 25(5), 1135–1155.

72.

Teachman

J. D.

(1987). Family background, educational resources, and educational attainment. American Sociological Review, 52(4), 548–557.

73.

Tholen

(2020). Degree power: Educational credentialism within three skilled occupations. British Journal of Sociology of Education, 41(3), 283–298.

74.

Traini

(2023). Educational gradient in occupational attainment: Does the stratification of education systems really matter? Comparative Education Review, 67(1), 123–146.

75.

Valenzuela

(1999). Subtractive schooling: U.S. - Mexican youth and the politics of caring. State University of New York Press.

76.

Vázquez-Cano

De la Calle-Cabrera

A. M.

Hervás-Gómez

López-Meneses

(2020). Socio-family context and its influence on students’ PISA reading performance scores: Evidence from three countries in three continents. Educational Sciences Theory & Practice, 20(2), 50–62.

77.

Waldinger

(2007). Does ability tracking exacerbate the role of family background for students’ test scores?’. Unpublished manuscript London School of Economics.

78.

Willms

J. D.

(2006). Variation in socioeconomic gradients among cantons in French- and Italian-speaking Switzerland: Findings from the OECD PISA. Educational Research and Evaluation, 12(2), 129–154.

79.

Wolf

P. J.

Kisida

Gutmann

Puma

Eissa

Rizzo

(2013). School vouchers and student outcomes: Experimental evidence from Washington, DC. Journal of Policy Analysis and Management, 32(2), 246–270.

80.

Zhou

(2023). Work trajectories and status attainment process: A study using sequence analysis. Journal of Chinese Sociology, 10(1), 1.

Family Background,Schooling,and Academic Achievement: Taiwan as an Illustrative Case

Abstract

Keywords

Introduction

Family Background and Academic Achievement

Effect of Schooling

Taiwanese Educational System in Brief

Methods

Data Source

Variables

Models

Results

Discussion

Conclusions

Implications

Caveats

Concluding Remarks

Footnotes

Acknowledgements

Author Contributions

Declaration of Conflicting Interests

Funding

Ethics Approval and Consent to Participate

Consent for Publication

ORCID iD

Data Availability Statement

Notes

References