Measuring the socio-economic background of students and its effect on achievement in PISA 2000 and PISA 2003

Abstract

One of the consistent findings of educational research studies is the effect of the students’ family socio-economic background on their learning achievement. Consequently, international comparative studies emphasise the role of socioeconomic background for determining learning outcomes. In particular, PISA results have been used to describe how different structures of the educational system can mediate the impact of socio-economic family background on performance with comprehensive systems generally providing more equity in educational opportunities. Given the cyclical nature of the PISA study, it is of interest to observe changes in the relationship between socio-economic background and performance. However, in order to measure the change in a relationship between two variables over time, one needs to rely on the same (or at least very similar) measures for both constructs. Furthermore, there are some concerns regarding the validity of reports on family background. This paper addresses the issue of measuring socio-economic background in the context of the OECD PISA study. It describes the computation of a composite index of “Economic, Social and Cultural Status” derived from occupational status of parents, educational level of parents and home possessions for the first two PISA cycles. It also shows differences in the relationship between socio-economic background and performance, both using single-level and multi-level analyses and compares and parent reports on occupation and education in order to explore the validity of these measures. Socio-economic status and performance Socio-economic status (SES) is an important explanatory factor in many different disciplines like health, child development and educational research. Research has shown that socioeconomic status is associated with health, cognitive and socioemotional outcomes (Bradley and Corwyn, 2002). In general, educational outcomes have been shown to be influenced by family background in many different and complex ways (Saha, 1997). For example, the socio-economic status of families has been consistently found to be an important variable in explaining variance in achievement. Socio-economic background may affect learning outcomes in numerous ways: From the outset, parents with higher socio-economic status are able to provide their children with the (often necessary) financial support and home resources for individual learning. They are also more likely to provide a more stimulating home environment to promote cognitive development. At the level of educational providers, students from high-SES families are also more likely to attend better schools, in particular in countries with differentiated (or tracked) educational systems, strong segregation in the system according to neighbourhood and/or clear advantages of private over public schooling (as for example in many developing countries). Socio-economic background measures have been used to control the effects of characteristics on performance dates: Coleman and others (1966) as well as Jencks (1972) claimed that schools were not major determinants of a child’s achievement, particularly when contrasted with the influences of family background on outcomes. When analysing school-level effects of social intake, the question arises how socioeconomic status should be measured at the level: Though some researchers argue against the use of aggregated SES from samples (see for example Sirin, 2005), alternative school-level measures like participation in free lunch programmes or neighbourhood SES census data are also viewed as inappropriate (see Hauser, 1994). Furthermore, school-level SES data from other sources would most likely be incomparable across countries. In view of the importance of including social intake as a variable in the analysis of performance, most national and international studies of educational achievement typically rely on aggregated data as school-level estimates of SES. Though there is a relative consensus that socio-economic status is represented by income, education and occupation (Gottfried, 1985; Hauser, 1994) and that using all three of them is better than using only one (White, 1982), there is no agreement among researchers which measures should be used in the analysis (Entwisle and Astone, 1994; Hauser, 1994). And whereas some argue that it is better to use a composite measure, others prefer to use single indicators for each component. Furthermore, there are different preferred ways of how to create composite SES measures (Mueller and Parcel, 1981; Gottfried, 1985). 1 Though it should be noted that other researchers (see for example Mayeske et al., 1972) have argued that social intake in conjunction with school-related variables should rather be viewed as school factors than student background factors. Differences in the use of measures have often led to quite different results regarding this relationship between SES and achievement (White, 1982; Sirin, 2005). As Keeves and Saha (1992, p. 166) point out it is clear that adolescents should not be asked about parental income. Alternatively, family wealth (as measured by assets) is often cited as an even better measure of resources than income (Bradley and Corwyn, 2002, p. 372). Based on the notion of assets as good measure of capital, Filmer and Pritchett (1999) propose using household assets to create a socio-economic index that can be validly used in cross-national research. In international studies additional caveats are imposed on the validity of background measures and the cross-national comparability of family background measures is an ongoing challenge for researchers in this area (see Buchmann, 2002). Clearly, in international comparative research requires the collection of comparable SES measures across countries. Student reports on parental education in international research have often suffered from high levels of non-response and a lack of comparability across countries. Recent studies (like PISA) have started to use the International Standard Classification of Education (ISCED) in order to enhance the comparability of data on parental education (OECD, 1999). Higher levels of non-response and the uncertain quality of student-derived data are the most salient concerns regarding the measurement of parental education. Whereas earlier IEA studies like the First International Mathematics Study (FIMS) collected data on parental occupation, later studies like the IEA Reading Literacy Study or TIMSS did not continue this practice but instead relied on reports on parental education and household items as measures of SES. The OECD PISA study uses the ISCO classification (ILO, 1990) to code open-ended responses on father's and mother's occupation which in turn are scored using the International Socio-economic Index of occupational status (SEI) to obtain socioeconomic measures (Ganzeboom, de Graaf and Treiman, 1992). International studies of educational achievement have often made use of reports on household items as measures of family capital (see Buchmann, 2002). Student reports on the number of books at home are often taken as proxies of SES in analyses of IEA data (see an example in Raudenbush, Cheong and Fotiu, 1996). However, there are concerns regarding the meaning of this variable in different cultural context, in particular in Asian countries. 2 Other measures of socio-economic background, typically not used in international educational research but also proposed in the literature, are associated with the concepts of social capital (Coleman, 1988) or cultural capital (Bourdieux and Passeron, 1977). One example is the quality of parent-child communication, which is reported to correlate with performance (see Howerton, Enger and Cobbs, 1993). Other examples include collecting data about or parent participation in, and preferences for, cultural activities as going to concerts, listening to music, or reading literature (see for example DiMaggio, 1982). 3 It should be noted that collecting data on parental occupation requires considerable additional resources in order to undertake the coding of open-ended responses which may explain the omission of this aspect in some educational studies.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 65 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.