OXFORD REVIEW OF EDUCATION

2019, VOL. 45, NO. 6, 769–785

https://doi.org/10.1080/03054985.2019.1620717

Family income effects on mathematics achievement: their

relative magnitude and causal pathways
a b
Gary N. Marks and Artur Pokropek
a
Directorate of Government, Policy and Strategy, Australian Catholic University, Fitzroy, VIC, Australia;
b
Educational Research Institute (IBE), Warszawa, Poland

ABSTRACT KEYWORDS
This study examines the influence of family income on student PISA; mathematics; family
achievement in mathematics utilising data from the parents’ ques- income; cross-national
tionnaire for nine countries participating in the OECD’s 2012 PISA differences; home resources
study. It finds non-trivial effects for family income that were consis-
tently larger than, or comparable to, the effects of more commonly
used measures of socio-economic background: parents’ education
and occupational status. Its effects ranged from weak (Hong Kong
and Macau) to moderate (Belgium, Chile and Portugal) and were
often stronger than the effects of parents’ education and occupa-
tional status. There was no evidence that the effects of family income
were mediated through cultural resources, wealth (indexed by
household possessions) or educational resources in any country.
This suggests that the effects of family income reflect largely unmea-
sured aspects of families rather than resources in the home. In
Belgium and Portugal, much of the effect of family income could
be attributed to prior knowledge and skills in mathematics.

Introduction
Socio-economic status (SES) is generally understood as a multi-dimensional concept. Bradley
and Corwyn (2002) understand SES as capital: higher SES families have greater access to
financial capital (material resources), a greater amount of human capital (education), and more
social capital (useful social connections). Similarly, Buchmann (2002) posits three processes
responsible for the association between SES and educational performance: the transmission of
human and financial capital, and cultural status. However, most empirical studies do not
include measures of financial capital such as family income or family wealth. If the aim of
the research to understand the mechanism that accounts for the relationship between SES and
academic performance, the absence of measures of financial capital is not defensible. In some
contexts, SES-based disparities in performance may stem mainly from cultural factors, but
others may be due to parents’ ability to buy educational success for their children or more
simply from poverty.
In this article, we ascertain the importance of economic resources, measured by family
income, on students’ academic performance. Intuitively, family income is a likely source of
inequalities in academic performance, but there much less academic literature on income

CONTACT Gary N. Marks [email protected]

© 2019 Informa UK Limited, trading as Taylor & Francis Group
770 G. N. MARKS AND A. POKROPEK

effects on achievement than for the effects of parents’ education and occupation. There are at
least two reasons for this situation: most studies of student achievement do not collect data on
family income since this would involve interviewing parents, and academic research has
tended to focus on cultural explanations, for example, Bourdieu’s theory of cultural capital.
According to the economic resources approach, economic inequality is the prime
generator of differences in student performance. High-income parents can send their
child to a high-fee private school, buy a residence in the wealthy catchment area of
a high-performing public school and, in some countries, employ personal tutors. In
addition, high-income parents can buy resources helpful for their children’s education
such as reference books, dictionaries, computers, iPads and employ tutors. In contrast,
low-income families struggle to buy the essentials for their child’s education: textbooks,
reading books and other educational resources; and possibly live in crowded and sub-
standard accommodation that is detrimental to their children’s education.
An alternative explanation is that low income inhibits parents’ abilities to be good
parents. Financially stressed parents do not have the time or the emotional energy to
support their children’s education and be responsive to their needs. A second variant
focusing on role models argues that low-income parents develop social norms and beha-
viours (e.g. authoritarian parenting, low educational expectations) that are harmful to
children’s early cognitive development and school education (Mayer, 1997, pp. 48–54).
Family income is significantly related to student achievement. Sirin’s (2005, p. 434)
meta-analysis, of mainly small studies conducted in the US between 1982 and 2000,
found that the average effect size (the adjusted correlation coefficient) for the bivariate
relationship between family income and student achievement was around 0.29, compar-
able to that for parents’ education (0.30) and father’s occupational status (0.28). This
contrasts with White’s (1982, p. 470) earlier meta-analysis in which family income had
clearly stronger correlations (r = 0.32) with student achievement than parental education
(r = 0.19) or parental occupation (r = 0.20).
However, while the observed correlations are moderate, family income has small or only
negligible causal effects on student achievement. For the US, Mayer (1997, pp. 90–91) reported
a standardised effect of 0.10 (which was not statistically significant) for family income on
children’s test scores, net of parents’ education, mother’s ability measured by her score in the
Armed Forces Qualification Test (AFQT) and other factors. Mayer (1997, pp. 90–91) estimated
the ‘true effect’ of family income on test scores controlling for all stable parental characteristics
that affect income, such as their educational attainment. The effects were small: a standardised
effect of 0.13 for family income on children’s Peabody Picture Vocabulary Test scores, 0.07 for
mathematics scores, and a negative effect on reading scores. Orr (2003, pp. 291, 293) reported
no effect for family income (averaged over five years) on mathematics achievement, net of
father’s occupational status, mother’s education, mother’s AFQT score, family wealth and
demographic variables. According to Blau (1999, p. 267 model 9) an increase of $10,000 (in
1979 dollars) in permanent income (that is, income averaged over five years), increased test
scores by only 12% and 14% for mathematics and reading, net of mother’s ability and other
factors. Blau (1999, p. 261) concludes that ‘Family background characteristics play a more
important role than income in determining child outcomes.’
For the UK, Brown et al. (2011, p. 49) conclude that, when controlling for mother’s
ability, ‘parents’ income and educational attainment, appear unrelated to their children’s
age 7 test scores’. Aughinbaugh and Gittleman’s (2003, p. 429) analysis of children’s test
 OXFORD REVIEW OF EDUCATION 771

scores in the US and the UK found that the effects of family income on test scores were
‘quite small, the largest being 8.2% of a standard deviation’ net of control variables
including mother’s ability.
There are few studies on the effects of family income on achievement in other
countries. A small South African study concluded there was no relationship between
families’ financial resources and student achievement (Cherian & Malehase, 1998).
Analysing student achievement in Denmark using the Organisation for Economic Co-
operation and Development (OECD)’s Programme for International Student Assessment
(PISA) study, Humlum (2011, p. 994) noted that the effects of family income were small
and statistically insignificant. Even a substantial change in permanent income of 100,000
Danish Krone (equivalent to about $US15,000) was associated with a difference of only
2.6 PISA score points on a measure with a standard deviation of 96. Analysing student
achievement in Australia as students progressed from Year 3 to 9, Marks (2016) found
only weak bivariate effects for comprehensive measures of family income and only trivial
and statistically insignificant effects when taking into account stable differences
between students.
The weaker than expected effects of family income on student achievement in the US
and other wealthy countries may be because in rich countries there is less variation in
family’s educational resources or school resources. The incidence of families without the
means to buy textbooks, desks, computers and other educational resources would be
lower in richer countries. Across countries, the extent of educational inequality is
negatively associated with the average years of education and per capita GDP
(Thomas et al., 2001). School differences in resources are smaller in higher income
countries (Heyneman & Loxley, 1983). Intergenerational correlations in education are
high in Latin American countries which also tend to have large income inequalities
(Hertz et al., 2007; World Bank, 2013, p. 36).
Studies on student achievement in the OECD’s PISA study have generally emphasised
the importance of cultural factors and educational resources. In a 30-country study,
a measure of cultural possessions comprising ‘books in the home’ and cultural posses-
sions in the home such as works and art and classic literature accounted for 25% of the
effect of socio-economic background on students’ test scores (Marks et al., 2006). Barone
(2006) concluded that cultural capital provides a relevant, but far from exhaustive,
account of PISA performance and its explanatory power was similar across countries.
Analysing data from the 2012 PISA study, Pokropek et al. (2015) conclude that access to
cultural and educational resources is at the root of socio-economic inequalities across
the world. Standard rounds of the OECD PISA study did not collect data on family
income and argue that economic resources are covered by other variables. According to
the PISA 2014 technical report, ‘Family assets are believed to capture wealth better than
income because they reflect a more stable source of wealth’ (OECD, 2014, p. 316). In the
PISA study, ‘family assets’ does not mean financial assets as in household studies of
wealth or net worth, but ‘home wealth possessions’ which is a scalar measure of the
presence or absence of, or the number of, a variety of household possessions (see text
below on measure of home resources). However, the OCED’s PISA study has responded
positively to the possibility of family income effects by offering a parental questionnaire
as a national option.
772 G. N. MARKS AND A. POKROPEK

The exception to the conclusion that family income effects are small is Chmielewski
and Reardon’s (2016) analysis of family income effects in the 2001 Progress in
International Reading Literacy Study and the 2006, 2009 and 2012 PISA studies. They
calculated the gaps in standardized achievement between the 90th and 10th percentiles
and regress that measure on country characteristics, concluding that the achievement
gap in income is positively associated with educational differentiation and national
levels of poverty and inequality, and negatively associated with curricular standardisa-
tion. However, their focus was solely on family income, not considering that other
aspects of the home environment – e.g. parental occupation and education – which
are also (often more strongly) associated with student achievement. Nor does
Chmielewski and Reardon’s (2016) study consider the whole distribution of family
income focusing only on the difference between two percentiles.
So, there is little consensus in the literature on the magnitude of the effects of family
income on student achievement. Most studies conclude its effects are small, while other
studies emphasise its importance (Heckman, 2008; Reardon, 2011). If family income has
effects on student achievement, they should be mediated in some way or other, through
cultural resources, by a superior home environment or specific educational resources that
help with students’ education. Interestingly, mother’s ability accounted for much of the
effects of income in Blau’s (1999, p. 266) study commenting: ‘The large reduction in the
effect of current income on child outcomes caused by adding AFQT suggests that ability
and income are highly correlated, and that ability matters more than current income for the
child outcomes.’ With PISA data, it is not possible to evaluate if the effects of family income
can be attributed to parental ability, psychological stress, parenting or low expectations.
However, we can test if the effects of family income are mediated through cultural factors,
wealth (as indexed by household possessions) or educational resources in the home.
The purpose of this study is to address these issues. First, it estimates the effects of family
incomes vis-à-vis the effects of measures routinely included in analyses of socio-economic
inequalities in student achievement: parents’ education and parents’ occupational status.
This will establish if the effects of income are non-trivial and if the effects are larger in poorer
countries. Second, this study measures the extent to which family income effects on
achievement are mediated by cultural resources, the physical environment of the home
(home resources) and educational resources. Third, the study estimates the effects of family
income net of proxies for prior mathematical knowledge and skills.

Materials and methods

Data
The analyses are based on data derived from the OECD’s 2012 PISA study. The PISA 2012
surveys were conducted in 34 OECD and 31 partner countries and economies on two-
stage stratified representative samples of students enrolled in lower secondary or upper
secondary institutions and aged between 15 years 3 months and 16 years 2 months.
Since the primary focus of this paper is the influence of family income, and data on
family income was only available through the parents’ questionnaire, so the criteria for
countries to be included in this study are:
 OXFORD REVIEW OF EDUCATION 773

(1) They participated in the PISA parents’ questionnaire option.

(2) The parental income data is distributed across categories.
(3) There was only a limited amount of missing data.

Of the 11 countries and administrative regions that participated in the PISA parents’
questionnaire option, the Italian data had no data on parental income, and in the
Mexican data two-thirds of parents had incomes in lowest of six categories. Therefore, the
data analysed for this study were from the nine remaining countries: Belgium, Chile, Croatia,
Germany, Hong Kong China, Hungary, Korea, Macau China, and Portugal. From this point on,
the special administrative regions of Hong Kong and Macau are referred to as countries.
The PISA study sets very high standards for sampling, translation, measurement of the
cognitive test domains and the validity and reliability of the measures derived from the
student, parent and school questionnaires. Details on the sample design, response rates,
weighting procedures, quality assurance, questionnaire items, scale construction, and
statistical analysis are provided in the PISA technical reports (OECD, 2011, 2014, 2016).
Between 1999 and 2015 there were nearly one thousand peer-reviewed articles using
PISA data or discussing its findings (Hopfenbeck et al., 2018).

Measures
In the 2012 PISA assessments, the main domain was mathematical literacy. PISA test
scores are highly predictive of subsequent high-stakes educational and labour market
outcomes (Fischbach et al., 2013; Marks, 2009; OECD, 2010). Item Response Theory (IRT)
modelling was used to create scores standardised at an OECD country mean of 500 and
standard deviation of 100. Students’ scores are in the form of five plausible values rather
than a single score. Details on IRT procedure for PISA are available from the PISA
technical report (OECD, 2014, pp. 214–256).
In the parents’ questionnaire, the question relating to family income was, ‘What is
your annual household income?’ (PA07Q01). The six response categories are defined as
follows (OECD, 2013a, p. 253):

(1) Less than half the median household income.

(2) Between half the median household income and less than three-quarters of the
median household income.
(3) Between three-quarters of the median household income and less than the median
household income.
(4) Between the median household income and less than five-quarters of the median
income.
(5) Between five-quarters of the median household income and less than one-and
-a-half of the median household income.
(6) More than one-and-a-half of the median household income.

In the parents’ questionnaire, the actual amounts in the local currency replaced the
italicised phrases.
774 G. N. MARKS AND A. POKROPEK

The distribution of the categories and the students’ mean scores in mathematics and
standard deviations are presented in Table 1. It is clear from this table that students’
mean mathematics scores increase incrementally with family income.
The parents’ questionnaire collected data on father’s and mother’s occupations. The
occupational codes were used to create International Socio-Economic Indices (SEI) of
occupational status for both parents. The highest of father’s or mother’s SEI score was
used. SEI measures ranged from 11 to 99. The PISA consortium has used the highest
value for the construction of its composite measure since the first PISA study in 2000
(OECD, 2014, p. 351). Using the highest value reduces the amount of missing data since
sizable proportions of students’ mothers do not work and sizable proportions of stu-
dents do not live with their fathers. Details on the construction of this index are available
(OECD, 2013b, pp. 132–136; 2014, p. 132, 137).
There were fewer and broader response categories for parents’ education in the
parent questionnaire than the corresponding question in the student questionnaire. In
the parents’ questionnaire, the categories ISCED 5A (degree) and 6 (higher degree) were

Table 1. Mean mathematics test scores by household income category by country.

Country Country
income category No. of cases Mean score Std. error income category No. of cases Mean score Std. error
Belgium 2738 514.53 2.14 Hungary 3762 477.04 3.19
1 315 484.38 7.35 1 908 423.65 4.16
2 472 504.38 5.78 2 992 460.34 3.66
3 472 522.60 5.21 3 721 491.24 4.30
4 375 558.93 6.00 4 419 513.42 5.22
5 326 572.77 5.50 5 279 517.99 7.31
6 778 603.18 3.81 6 443 548.14 9.32
Chile 6277 422.63 3.07 Korea 4935 553.77 4.58
1 1030 484.38 7.35 1 840 514.20 5.46
2 1259 504.38 5.78 2 702 532.99 5.00
3 621 522.60 5.21 3 806 546.92 4.80
4 499 558.93 6.00 4 776 557.29 4.77
5 421 572.77 5.50 5 606 571.53 5.23
6 2447 603.18 3.81 6 1205 591.35 6.69
Croatia 4445 471.13 3.54 Macau 5002 538.13 0.96
1 527 435.08 4.88 1 991 450.56 4.14
2 1502 452.67 3.80 2 1091 482.85 5.08
3 1216 474.39 3.82 3 865 501.40 5.43
4 632 497.27 5.07 4 618 528.46 5.02
5 279 520.12 7.65 5 467 532.11 5.03
6 289 530.12 9.46 6 970 562.48 4.19
Germany 2111 513.53 2.88 Portugal 4279 487.06 3.81
1 277 469.85 7.86 1 1445 450.56 4.14
2 298 508.26 6.82 2 867 482.85 5.08
3 336 530.23 6.52 3 509 501.40 5.43
4 287 531.93 6.10 4 355 528.46 5.02
5 262 553.48 6.12 5 297 532.11 5.03
6 651 570.79 4.40 6 806 562.48 4.19
Hong Kong 4401 561.24 3.22
1 691 530.63 5.15
2 849 544.39 4.45
3 717 551.87 4.35
4 500 563.05 4.43
5 334 573.24 6.49
6 1310 592.49 4.88
Note: Entries opposite Country are the number of non-missing cases for household income and corresponding mean
score and standard error of the mean. MHI = Median Household Income. Categories: 1; 0.5*MHI ≤2< 0.75*MHI;
0.75*MHI ≤3<*MHI; 0.75*MHI ≤4< 1.25*MHI; 1.25*MHI ≤5< 1.5*MHI; 6≥ 1.5*MHI.
 OXFORD REVIEW OF EDUCATION 775

combined whereas in the student questionnaire they were separated. The parents’
questionnaire did not include ISCED categories 1 (primary school) and 2 (lower second-
ary school). Therefore, it was decided to use the data from the student questionnaire on
parents’ education using the already constructed variable (OECD, 2014, pp. 133–134),
highest level of parents’ education (HISCED) based on the International Standard
Classification of Education (ISCED).
The seven ordinal categories for the measure of the highest level of parents’ educa-
tion were:

(1) None
(2) ISCED 1 (primary elementary education)
(3) ISCED 2 (lower secondary education)
(4) ISCED 3B, C (upper secondary education providing direct access to the labour
market or to ISCED 5B programmes)
(5) ISCED 3A (upper secondary education providing access to ISCED 5A and 5B
programmes), and ISCED 4 (non-tertiary post-secondary)
(6) ISCED 5B (non-university tertiary education)
(7) ISCED 5A (university-level tertiary education) and ISCED 6 (advanced research
programmes)

The measures of cultural possessions, home resources and educational resources were
derived from the student questionnaire data. There were fewer missing data for these
measures than in the parents’ questionnaire. The 2012 PISA technical report (2014, pp.
314–319) details the constructed measures and the individual items in the student
questionnaire scales designated as cultural possessions (CULTPOS), home resources
(WEALTH) and educational resources (HEDRES).
The cultural possessions measure is based on items asking if the household pos-
sesses: classical literature, poetry books and works of art. The home resources measure
comprises the following yes-no items: ‘a room of your own’, ‘a link to the Internet’, ‘a
DVD player (or equivalent)’; three country-specific items; and the number of mobile
phones, televisions, computers, cars and bathrooms in the household. ‘Home resources’
is a more accurate description of the measure than the PISA study moniker ‘Wealth’. The
educational resources measure comprises whether the student has, or access to, ‘a desk
to study at’, ‘a quiet place to study’, ‘a computer for school work’, ‘educational software’,
‘books to help with school work’, ‘technical reference books’ and ‘a dictionary’.
We included two control variables related to learning outcomes: students’ grade and
time spent on mathematics in school. It was necessary to include both variables in the
analyses because of their importance to mathematics performance and possible correla-
tions with family income. Students in higher grades have learnt more mathematics, been
exposed to more difficult mathematical concepts and show higher scores. Similarly, the
more time spent learning mathematics should relate to higher scores. They serve as proxy
measures for prior mathematical knowledge and skills and may control for school selection
effects in that students from higher income families go to better schools. They may account
for some of the effects of family income. ‘Grade’ was centred on the modal grade for the
country. One year of schooling is equivalent to an average of approximately 40 score points
(OECD, 2013b, p. 69). The distributions of grade by country are available (OECD, 2013b, pp.
776 G. N. MARKS AND A. POKROPEK

154–155). Learning time in mathematics was computed by multiplying students’ responses

on the number of minutes on average in their mathematics class by number of mathematics
classes per week (OECD, 2013b, p. 136). Since the distribution of learning time was highly
positively skewed, it was transformed by taking the natural log.

Statistical methods
The countries vary in the amount of missing data. The main source of missing data is family
income, and the problem is most severe in Belgium and Germany. Missing data were handled
by Full Information Maximum Likelihood (FIML) Estimation (Allison, 2012; Baraldi & Enders,
2010). The FIML technique produces estimates in parametric models for incomplete data; it
assumes missing data are missing at random (MAR) and has a multivariate normal distribution.
FIML is reasonably robust to violations of the distributional assumptions (Baraldi & Enders,
2010). The means, standard deviations and correlations for the variables analysed after
adjusting for missing data using the FIML are presented in Table 2. Corresponding results
before adjustments are presented in Appendix Table A1. The FIML adjusted mean mathe-
matics scores for each country are almost identical to the mean scores reported by the OECD
(2013b, pp. 15, 174). The method tended to reduce the mean for family income for countries
with larger amounts of missing data but produced little or no change for the means and
standard deviations for the other variables and the correlations between variables.
Regression models were used to estimate the effects of the predictor variables on
students’ mathematics scores with FIML using the Structural Equation Modelling routine
in the Stata 14.2 statistical software.
The PISA’s two-stage sampling design involves first sampling schools and then
students. This complicates the analysis since students are sampled with unequal prob-
abilities and sample units are not sampled independently since they are clustered within
schools. To address these issues, we followed the OECD’s recommendations by:

(1) using PISA sampling weights for obtaining correct point estimates since students
do not have equal probabilities of selection (OECD, 2014, pp. 132–138);
(2) fitting five models for each of the five plausible values and then aggregating
resulting estimates using the Rubin rule (OECD, 2014, p. 147);
(3) using balanced repeated replication (BRR) weights with Fay’s adjustment to
calculate standard errors (OECD, 2014, pp. 138–140).

Three regression models were estimated for each country. The initial model comprised
three indicators of socio-economic background: family income, parents’ education and
occupational status. The second model adds the measures of cultural, home and
educational resources to assess the extent that the effects of family income are
mediated by these factors. Grade and learning time are added in model 3 to control
for prior mathematical knowledge and skills.

Results
Table 3 presents the estimates from the analyses of PISA mathematics scores from three
models. The metric (unstandardised) estimates are interpreted as the expected change
 OXFORD REVIEW OF EDUCATION 777

Table 2. Means, standard deviations and correlations after adjusting for missing data.
N Mean Std. Math Inc Ped Occ Cult Home Edur Gr
Belgium
Maths ability 8597 514.53 102.26
Household income 2738 3.54 1.80 0.45
Parents’ education 8129 4.84 1.24 0.29 0.46
Occupational status 8153 50.92 22.12 0.43 0.54 0.53
Cultural possessions 8287 −0.32 0.99 0.28 0.28 0.29 0.35
Home possessions 8503 0.08 0.80 0.17 0.35 0.22 0.24 0.18
Educational resources 8474 0.09 0.95 0.31 0.26 0.20 0.25 0.39 0.22
Grade 8483 −0.47 0.68 0.60 0.34 0.20 0.31 0.20 0.16 0.26
Learning time 4821 5.31 0.12 0.34 0.22 0.16 0.20 0.12 0.08 0.10 0.07
Chile
Maths ability 6856 422.63 80.75
Household income 6277 3.28 1.89 0.46
Parents’ education 6665 4.04 1.54 0.39 0.57
Occupational status 6499 40.43 21.49 0.42 0.63 0.63
Cultural possessions 6750 −0.21 0.96 0.23 0.27 0.26 0.26
Home possessions 6776 −0.59 1.01 0.39 0.62 0.50 0.53 0.29
Educational resources 6773 −0.49 1.01 0.24 0.36 0.32 0.29 0.37 0.50
Grade 6856 −0.27 0.71 0.38 0.14 0.12 0.09 0.06 0.11 0.14
Learning time 3733 5.88 0.22 −0.03 0.00 −0.02 0.00 0.00 −0.02 −0.01 −0.01
Croatia
Maths ability 5008 471.13 88.47
Household income 4445 2.92 1.35 0.32
Parents’ education 4978 4.56 1.11 0.23 0.44
Occupational status 4769 44.92 20.66 0.37 0.55 0.57
Cultural possessions 4823 −0.22 1.02 0.26 0.30 0.30 0.35
Home possessions 4996 −0.45 0.74 0.12 0.35 0.23 0.25 0.24
Educational resources 4986 0.05 0.90 0.15 0.15 0.14 0.16 0.34 0.24
Grade 5008 0.20 0.40 0.12 0.02 0.00 0.03 0.02 0.00 0.01
Learning time 3296 4.94 0.12 0.44 0.25 0.17 0.27 0.18 0.12 0.13 −0.05
Germany
Maths ability 5001 513.53 96.30
Household income 2111 3.72 1.81 0.34
Parents’ education 3936 4.24 1.59 0.29 0.36
Occupational status 4034 50.26 20.65 0.39 0.51 0.52
Cultural possessions 4151 0.08 0.99 0.24 0.26 0.31 0.34
Home possessions 4244 0.20 0.80 0.11 0.33 0.13 0.15 0.12
Educational resources 4235 0.19 0.87 0.24 0.23 0.18 0.22 0.43 0.21
Grade 5001 0.27 0.67 0.46 0.18 0.19 0.19 0.13 0.10 0.16
Learning time 2551 5.24 0.09 −0.13 −0.10 −0.05 −0.08 −0.02 −0.02 −0.05 −0.09
Hungary
Maths ability 4810 477.04 93.62
Household income 3762 2.84 1.62 0.42
Parents’ education 4692 4.44 1.41 0.36 0.45
Occupational status 4471 45.39 20.86 0.45 0.51 0.61
Cultural possessions 4717 0.16 1.02 0.40 0.34 0.36 0.38
Home possessions 4764 −0.54 0.78 0.19 0.40 0.34 0.30 0.24
Educational resources 4760 −0.10 0.92 0.32 0.27 0.29 0.27 0.47 0.36
Grade 4810 0.06 0.63 0.37 0.17 0.12 0.18 0.19 0.14 0.23
Learning time 3072 4.97 0.07 0.08 0.01 0.02 0.03 0.03 0.01 0.04 −0.05
Korea
Maths ability 5033 553.77 99.08
Household income 4935 3.64 1.80 0.27
Parents’ education 4983 4.94 1.20 0.23 0.37
Occupational status 4939 53.24 18.13 0.23 0.36 0.43
Cultural possessions 4950 0.27 0.94 0.25 0.32 0.29 0.26
Home possessions 5029 −0.70 0.62 0.15 0.47 0.27 0.28 0.35
Educational resources 5017 −0.10 0.96 0.30 0.28 0.24 0.22 0.50 0.36
Grade 5033 −0.06 0.24 0.09 −0.01 −0.03 −0.02 0.00 0.00 0.02
Learning time 3232 5.32 0.08 0.30 0.14 0.11 0.10 0.13 0.12 0.12 0.21
(Continued)
778 G. N. MARKS AND A. POKROPEK

Table 2. (Continued).
N Mean Std. Math Inc Ped Occ Cult Home Edur Gr
Hong Kong
Maths ability 4670 561.24 96.31
Household income 4401 3.69 1.88 0.24
Parents’ education 4477 3.37 1.57 0.24 0.51
Occupational status 4307 45.98 21.05 0.23 0.55 0.60
Cultural possessions 4573 −0.13 0.98 0.18 0.31 0.36 0.33
Home possessions 4579 −0.96 0.69 0.13 0.50 0.39 0.43 0.33
Educational resources 4577 −0.27 0.97 0.21 0.24 0.31 0.26 0.45 0.31
Grade 4670 −0.41 0.68 0.30 0.22 0.16 0.17 0.11 0.17 0.12
Learning time 2881 5.56 0.06 0.02 0.01 0.03 −0.01 0.05 0.04 0.06 0.10
Macau
Maths ability 5335 538.13 94.50
Household income 5002 3.28 1.79 0.18
Parents’ education 5252 2.97 1.57 0.11 0.36
Occupational status 5195 44.70 16.96 0.11 0.42 0.48
Cultural possessions 5289 −0.40 0.98 0.21 0.17 0.28 0.19
Home possessions 5316 −0.53 0.84 0.02 0.37 0.26 0.30 0.23
Educational resources 5312 −0.31 1.04 0.24 0.18 0.26 0.19 0.44 0.26
Grade 5335 −0.82 0.90 0.55 0.16 0.16 0.13 0.17 0.10 0.24
Learning time 3474 5.59 0.06 0.20 0.03 0.01 0.04 0.07 0.03 0.05 0.13
Portugal
Maths ability 5722 487.06 93.95
Household income 4279 2.88 1.93 0.46
Parents’ education 5572 3.16 1.96 0.35 0.54
Occupational status 5468 42.16 21.45 0.42 0.63 0.71
Cultural possessions 5578 0.05 1.01 0.30 0.35 0.40 0.40
Home possessions 5642 0.23 0.91 0.26 0.42 0.35 0.40 0.33
Educational resources 5637 −0.12 0.89 0.28 0.28 0.31 0.29 0.43 0.33
Grade 5209 −0.58 0.77 0.60 0.33 0.24 0.31 0.22 0.22 0.26
Learning time 3230 5.59 0.13 0.15 0.11 0.08 0.10 0.07 0.08 0.07 −0.03
Note: The number of cases for the FML adjusted data equals the number of cases for average mathematics score.

in students’ scores in mathematics for a one-unit change in the predictor variables. The
standardised coefficients allow comparison of the magnitude of effects.
In the initial model, the largest effect for family income for a one category increase is
found for Belgium (16 score points) and the smallest for Hong Kong (6 score points).
Together the three indicators of socio-economic background explain about one quarter
of the variance in students’ mathematics scores in Belgium (24%), Chile (25%) Hungary
(26%) and Portugal (24%). The variance explained by the initial model in the Asian
countries is much lower: Korea (10%), Hong Kong (8%) and Macau (4%).
The effects of family income are much the same or only slightly smaller than in the second
model compared to the initial model. The metric coefficients were 16 score points for both
models for Belgium, 12 for the initial model and 10 for the second model for Chile, 11 and 10
for Croatia, 10 and 9 for Germany, 7 and 7 for Hong Kong, 14 and 13 for Hungary, 10 and 9 for
Korea, 8 and 9 for Macau, and 15 and 14 for Portugal. Therefore, the effects of family income
are not mediated through cultural, home or educational resources.
The third model adds grade and time spent on learning mathematics at school which
markedly reduced the effects of family income in two countries; the unstandardised effects
declined from 16 to 9 for Belgium and from 14 to 9 for Portugal. In the other countries, there
were moderate declines from 10 to 7 for Croatia, 9 to 7 for Germany, 7 to 5 for Hong Kong,
and from 9 to 6 for Macau. For Chile, Korea and Hungary there were only small declines.
 OXFORD REVIEW OF EDUCATION 779

Table 3. Effects of socio-economic background indicators and grade on mathematics achievement.

Country/Variable B Beta Z-test p-value B Beta Z-test p-value B Beta Z-test p-value
Belgium (n = 8597)
Household income 16.24 0.28 10.94 0.00 15.74 0.28 10.93 0.00 8.71 0.15 6.98 0.00
Parents’ education 1.49 0.02 1.01 0.31 0.14 0.00 0.10 0.92 −0.16 0.00 −0.14 0.89
Occ. status (SEI) 1.25 0.27 14.70 0.00 1.04 0.22 13.00 0.00 0.58 0.13 8.33 0.00
Cult. possessions 6.75 0.07 4.86 0.00 5.07 0.05 4.00 0.00
Home possessions −3.72 −0.03 −2.01 0.04 −4.09 −0.03 −2.85 0.00
Educ. resources 17.61 0.16 11.44 0.00 8.63 0.08 6.35 0.00
Grade 69.80 0.46 31.43 0.00
Learning time 70.56 0.24 14.34 0.00
R-Square 0.24 0.28 0.51
Chile (n = 6856)
Household income 12.36 0.29 10.91 0.00 10.16 0.24 8.13 0.00 8.88 0.21 7.68 0.00
Parents’ education 6.66 0.13 7.40 0.00 5.40 0.10 6.14 0.00 4.43 0.08 5.78 0.00
Occ. status (SEI) 0.60 0.16 7.38 0.00 0.52 0.14 6.50 0.00 0.55 0.15 6.99 0.00
Cult. possessions 6.22 0.07 4.82 0.00 6.56 0.08 5.20 0.00
Home possessions 6.89 0.09 4.96 0.00 7.03 0.09 5.09 0.00
Educ. resources 0.80 0.01 0.61 0.54 −1.67 −0.02 −1.37 0.17
Grade 36.22 0.32 17.57 0.00
Learning time −3.66 −0.02 −1.11 0.27
R-Square 0.25 0.26 0.36
Croatia (n = 5008)
Household income 10.80 0.17 6.48 0.00 10.21 0.16 6.19 0.00 7.14 0.11 4.50 0.00
Parents’ education 0.07 0.00 0.05 0.96 −1.17 −0.01 −0.77 0.44 −0.44 −0.01 −0.30 0.77
Occ. status (SEI) 1.17 0.27 10.61 0.00 1.05 0.25 10.50 0.00 0.78 0.18 8.68 0.00
Cult. possessions 10.45 0.12 5.74 0.00 8.11 0.09 5.15 0.00
Home possessions −4.89 −0.04 −2.33 0.02 −4.65 −0.04 −2.31 0.02
Educ. resources 6.06 0.06 3.74 0.00 3.95 0.04 2.57 0.01
Grade 28.86 0.13 9.84 0.00
Learning time 90.84 0.35 11.43 0.00
R-Square 0.15 0.17 0.30
Germany (n = 5001)
Household income 9.50 0.18 6.07 0.00 8.83 0.17 5.59 0.00 7.20 0.14 5.22 0.00
Parents’ education 5.90 0.10 5.14 0.00 4.90 0.08 4.34 0.00 2.85 0.05 2.84 0.00
Occ. status (SEI) 1.18 0.25 9.94 0.00 1.06 0.23 8.83 0.00 0.91 0.20 8.49 0.00
Cult. possessions 4.97 0.05 2.99 0.00 4.84 0.05 3.16 0.00
Home possessions −2.85 −0.02 −1.30 0.20 −3.96 −0.03 −1.94 0.05
Educ. resources 12.73 0.12 5.58 0.00 8.43 0.08 3.77 0.00
Grade 52.73 0.37 21.40 0.00
Learning time −21.18 −0.07 −2.24 0.03
R-Square 0.19 0.21 0.34
Hong Kong (n = 4670)
Household income 6.62 0.13 5.72 0.00 6.91 0.14 6.23 0.00 5.10 0.10 4.69 0.00
Parents’ education 7.23 0.12 4.93 0.00 5.53 0.09 3.79 0.00 5.30 0.09 3.65 0.00
Occ. status (SEI) 0.41 0.09 3.66 0.00 0.39 0.08 3.55 0.00 0.33 0.07 3.00 0.00
Cult. possessions 4.51 0.05 3.15 0.00 4.49 0.05 3.26 0.00
Home possessions −9.27 −0.07 −3.36 0.00 −11.09 −0.08 −3.93 0.00
Educ. Resources 12.20 0.12 6.93 0.00 10.92 0.11 6.26 0.00
Grade 34.81 0.25 16.33 0.00
Learning time −6.49 −0.02 −0.69 0.49
R-Square 0.08 0.10 0.16
Hungary (n = 4810)
Household income 14.48 0.25 9.48 0.00 12.72 0.22 8.65 0.00 11.86 0.21 8.83 0.00
Parents’ education 6.07 0.09 3.82 0.00 3.46 0.05 2.42 0.02 4.29 0.06 3.56 0.00
Occ. status (SEI) 1.16 0.26 11.18 0.00 0.98 0.22 9.80 0.00 0.85 0.19 8.82 0.00
Cult. possessions 17.07 0.19 10.47 0.00 15.28 0.17 9.57 0.00
Home possessions −8.44 −0.07 −3.26 0.00 −9.07 −0.08 −4.34 0.00
Educ. resources 12.39 0.12 7.04 0.00 7.92 0.08 4.75 0.00
Grade 38.31 0.26 13.12 0.00
(Continued)
780 G. N. MARKS AND A. POKROPEK

Table 3. (Continued).
Country/Variable B Beta Z-test p-value B Beta Z-test p-value B Beta Z-test p-value
Learning time 26.72 0.08 3.42 0.00
R-Square 0.26 0.31 0.38
Korea (n = 5033)
Household income 10.32 0.19 9.29 0.00 8.87 0.16 7.71 0.00 7.90 0.14 7.58 0.00
Parents’ education 9.69 0.12 5.42 0.00 7.24 0.09 4.28 0.00 6.59 0.08 4.26 0.00
Occ. status (SEI) 0.61 0.11 5.65 0.00 0.49 0.09 4.90 0.00 0.46 0.08 4.62 0.00
Cult. possessions 7.09 0.07 3.87 0.00 5.87 0.06 3.35 0.00
Home possessions −11.63 −0.07 −3.57 0.00 −12.76 −0.08 −4.11 0.00
Educ. resources 21.16 0.21 9.98 0.00 20.00 0.19 9.94 0.00
Grade 16.41 0.04 1.68 0.09
Learning time 80.24 0.23 6.75 0.00
R-Square 0.10 0.15 0.21
Macau (n = 5335)
Household income 8.41 0.16 9.18 0.00 8.84 0.17 10.03 0.00 6.13 0.12 8.12 0.00
Parents’ education 2.74 0.05 2.61 0.01 −0.54 −0.01 −0.60 0.55 −1.96 −0.03 −2.49 0.02
Occ. status (SEI) 0.15 0.03 1.57 0.12 0.16 0.03 1.76 0.08 0.07 0.01 0.87 0.39
Cult. possessions 11.98 0.12 7.14 0.00 8.45 0.08 5.17 0.00
Home possessions −13.42 −0.12 −6.86 0.00 −12.47 −0.11 −8.18 0.00
Educ. resources 16.53 0.18 12.29 0.00 8.22 0.09 6.61 0.00
Grade 51.29 0.49 53.80 0.00
Learning time 48.25 0.12 8.84 0.00
R-Square 0.04 0.10 0.35
Portugal (n = 5722)
Household income 15.49 0.32 16.41 0.00 14.21 0.29 16.52 0.00 9.34 0.19 11.10 0.00
Parents’ education 2.64 0.06 2.62 0.01 0.95 0.02 1.00 0.32 1.60 0.03 1.94 0.05
Occ. status (SEI) 0.78 0.18 7.57 0.00 0.66 0.15 6.60 0.00 0.36 0.08 3.97 0.00
Cult. possessions 7.52 0.08 4.40 0.00 5.48 0.06 3.67 0.00
Home possessions 0.93 0.01 0.55 0.58 −1.39 −0.01 −0.92 0.36
Educ. resources 11.57 0.11 7.10 0.00 3.90 0.04 2.74 0.01
Grade 59.52 0.49 26.71 0.00
Learning time 33.54 0.13 5.42 0.00
R-Square 0.24 0.26 0.47

These analyses indicate that in Belgium and Portugal, much of the effects of family income
could be attributed to prior knowledge and skills in mathematics.
In the final model the estimates for family income are quite similar, ranging between
5 and 10 score points, with Hungary being an outlier (12 score points). In most of these
countries, due to the sizable correlation between grade, learning time and test scores,
the variance accounted for over 40% in Belgium and Portugal and over 30% in Chile,
Croatia, Germany, Hungary and Macau. In Hong Kong and Korea, the variance explained
is substantially smaller, 16% and 21%, respectively.
In several countries, the standardised effects of family income, net of the other
factors in the final model, remain non-trivial: around 0.20 for Chile, Hungary and
Portugal; around 0.15 for Belgium, Germany and Korea; and close to 0.10 for
Croatia, Hong Kong and Macau. In no country is the standardised effect of family
income substantially smaller than that for parents’ education or occupational
status. In most countries, the standardised coefficient for family income was the
largest or second largest of the coefficients for the six social status variables in the
regression analysis.
 OXFORD REVIEW OF EDUCATION 781

Conclusions and discussion

This study demonstrates that family income has non-trivial effects on student perfor-
mance in a range of countries. Although its effects could not be considered large with
standardised effects ranging from about 0.10 to 0.20, its effects were typically larger
than, or at least comparable to, the effects of parents’ education and occupational
status. This finding suggests income is relevant to studies of student achievement and
cannot be proxied by parental education and occupation, and home resources.
The effects of family income tend to be weaker in Hong Kong and Macau, but it cannot be
concluded that its effects are weaker in Asian countries, since its effects for Korea are
comparable to that for Croatia and Germany. In this group of countries, there is no indication
that the effects of family income are weaker in richer countries since its (net) effects in the
richer countries (Belgium and Germany) are not substantially smaller than its effects in
Croatia, Hungary and Portugal. The strong effect of family income for Chile is consistent
with stronger parental education and income inequalities in Latin American countries.
The main limitation of this study is the extent to which it can estimate causal effects.
The data does not include variables that correlate with family income which may
account for its effects. We do not have information on prior achievement, parental
intelligence, and aptitude. Although the precise mechanism by which income influences
on mathematics achievement cannot be elucidated with these data, the main conclusion
of this study, that family income has non-trivial effects on student performance in
a range of countries, even when controlling for a range of other influences, is in our
opinion unexpected and important.
It has been hypothesised that the effects of family income on achievement are
mediated by resources in the home since higher-income households can afford greater
resources, and poor households’ lack of resources impedes their children’s education.
However, in all the countries examined here, there is no evidence that the effects of
family income are mediated by cultural, home or educational resources. Therefore, the
explanation for the effects of family income involves something other than household
resources; possibly parental stress, parental ability, attitudes to education or parenting
style. Prior knowledge and skills in mathematics accounted for some of the effects of
family income in almost all countries, and for Belgium and Portugal substantially so. In
most of these countries, the net effects of family income on mathematics are non-trivial,
but small with standardised effects below 0.20. The literature review cited several studies
that found very weak effects of family income in studies that included mother’s ability
and Mayer’s (1997, pp. 90–91) estimates of the ‘true’ effects of income on achievement
were also small. Measures of parental stress, attitudes and values, parenting style and
parental ability are required to elucidate the nature of the association between parental
income and student achievement.

Disclosure statement
No potential conflict of interest was reported by the authors.
782 G. N. MARKS AND A. POKROPEK

Notes on contributors
Gary N. Marks is currently a Principal Fellow in the School of Political and Social Sciences at the
University of Melbourne. He has been involved in large-scale research projects: the Longitudinal
Surveys of Australian Youth, Household Income and Labour Dynamics of Australia and the OCED’s
Programme for International Student Assessment. He has published in a range of research areas
including school leaving, cross-national differences in student performance, the school-to-work
transition, school effectiveness, household wealth, income dynamics, youth and adult labour
markets, cognitive ability, behavioural genetics, and over-time changes in social stratification.
Artur Pokropek is a Professor at the Institute of Philosophy and Sociology of the Polish Academy of
Sciences and at the Educational Research Institute in Warsaw. The main areas of his research
are sociology of education, statistics, psychometry and research methodology. Currently, he is
working on social stratification, various aspects of compatibility in large-scale cross-country
surveys and modelling responses biases. He was involved in PISA and PIAAC realisation in
Poland. He worked and studied at Educational Testing Service (Princeton, USA) and EC Joint
Research Center (Ispra, Italy). He is the author of several dozen articles in international journals
such as Structural Equation Modeling, Journal of Educational Psychology, and Sociological
Methods & Research.

ORCID
Gary N. Marks http://orcid.org/0000-0002-7380-5243
Artur Pokropek http://orcid.org/0000-0002-5899-2917

References
Allison, P. D. (2012). Handling missing data by maximum likelihood Retrieved from http://www.
statisticalhorizons.com/wp-content/uploads/MissingDataByML.pdf
Aughinbaugh, A., & Gittleman, M. (2003). Does money matter? A comparison of the effect of
income on child development in the United States and Great Britain. Journal of Human
Resources, 38(2), 416–440.
Baraldi, A. N., & Enders, C. K. (2010). An introduction to modern missing data analyses. Journal of
School Psychology, 48(1), 5–37.
Barone, C. (2006). Cultural capital, ambition and the explanation of inequalities in learning out-
comes: A comparative analysis. Sociology, 40(6), 1039–1058.
Blau, D. M. (1999). The effect of income on child development. Review of Economics and Statistics,
81(2), 261–276.
Bradley, R. H., & Corwyn, R. F. (2002). Socioeconomic status and child development. Annual Review
of Psychology, 53, 371–399.
Brown, S., McIntosh, S., & Taylor, K. (2011). Following in your parents’ footsteps? Empirical
analysis of matched parent-offspring test scores. Oxford Bulletin of Economics and Statistics,
73(1), 40–58.
Buchmann, C. (2002). Measuring family background in international studies of education:
Conceptual issues and methodological challenges. In A. C. Porter & A. Gamoran (Eds.),
Methodological advances in cross-national surveys of educational achievement (pp. 150–197).
Washington, DC: National Academy Press.
Cherian, V. I., & Malehase, M. C. (1998). Relationship between family income and achievement in
English of children from single- and two-parent families. Psychological Reports, 83(2),
431–434.
Chmielewski, A. K., & Reardon, S. F. (2016). Patterns of cross-national variation in the association
between income and academic achievement. AERA Open, 2(3), 1–27.
 OXFORD REVIEW OF EDUCATION 783

Fischbach, A., Keller, U., Preckel, F., & Brunner, M. (2013). PISA proficiency scores predict educa-
tional outcomes. Learning and Individual Differences, 24(April), 63–72.
Heckman, J. J. (2008). Role of income and family influence on child outcomes. Annals of the
New York Academy of Sciences, 1136, 307–323.
Hertz, T., Jayasundera, T., Piraino, P., Selcuk, S., Smithy, N., & Verashchagina, A. (2007). The
inheritance of educational inequality: International comparisons and fifty-year trends. The B.E.
Journal of Economic Analysis & Policy, 7(2), 10.
Heyneman, S., & Loxley, W. (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.
Hopfenbeck, T. N., Lenkeit, J., El Masri, Y., Cantrell, K., Ryan, J., & Baird, J.-A. (2018). Lessons learned
from PISA: A systematic review of peer-reviewed articles on the programme for international
student assessment. Scandinavian Journal of Educational Research, 62(3), 333–353.
Humlum, M. K. (2011). Timing of family income, borrowing constraints, and child achievement.
Journal of Population Economics, 24(3), 979–1004.
Marks, G. N. (2009). Accounting for school-sector differences in university entrance performance.
Australian Journal of Education, 53(1), 19–38.
Marks, G. N. (2016). The relative effects of socio-economic, demographic, non-cognitive and
cognitive influences on student achievement in Australia. Learning and Individual Differences,
49, 1–10.
Marks, G. N., Cresswell, J., & Ainley, J. (2006). Explaining socioeconomic inequalities in student
achievement: The role of home and school factors. Educational Research and Evaluation, 12(2),
105–128.
Mayer, S. E. (1997). What money can’t buy: Family income and children’s life chances. Cambridge,
MA: Harvard University Press.
OECD. (2010). Pathways to success: How knowledge and skills at age 15 shape future lives in Canada.
Paris: Organization for Economic Co-operation and Development.
OECD. (2011). PISA 2009 Technical Report. Paris: Organisation for Economic Co-operation and
Development.
OECD. (2013a). PISA 2012 assessment and analytical framework. Paris: Organisation for Economic
Co-operation and Development.
OECD. (2013b). PISA 2012 results: Excellence through equity giving every student the chance to
succeed (Vol. II). Paris: Organisation for Economic Co-operation and Development.
OECD. (2014). PISA 2012 Technical Report. Paris: Organisation for Economic Co-operation and
Development.
OECD. (2016). PISA 2015 Technical Report. Paris: Organisation for Economic Co-operation and
Development.
Orr, A. J. (2003). Black-White differences in achievement: The importance of wealth. Sociology of
Education, 76(4), 281–304.
Pokropek, A., Borgonovi, F., & Jakubowski, M. (2015). Socio-economic disparities in academic
achievement: A comparative analysis of mechanisms and pathways. Learning and Individual
Differences, 42(August), 10–18.
Reardon, S. F. (2011). The widening academic achievement gap between the rich and the poor:
New evidence and possible explanations. In G. J. Duncan & R. Murnane (Eds.), Whither oppor-
tunity? Rising inequality, schools, and children’s life chances (pp. 91–116). New York: Russell Sage
Foundation.
Sirin, S. R. (2005). Socioeconomic status and academic achievement: A meta-analytical review of
research. Review of Educational Research, 75(3), 417–453.
Thomas, V., Wang, Y., & Fan, X. (2001). Measuring education inequality: Gini Coefficients of Education
(2525). Washington, DC. Retrieved from http://elibrary.worldbank.org/doi/abs/10.1596/1813-
9450-2525
White, K. R. (1982). The relationship between socio-economic status and academic achievement.
Psychological Bulletin, 91(3), 461–481.
World Bank. (2013). World development indicators 2013. Washington DC. Retrieved from http://
databank.worldbank.org/data/download/WDI-2013-ebook.pdf
784 G. N. MARKS AND A. POKROPEK

APPENDIX

Table A1. Means, standard deviations and correlations, raw data.

N Mean Std. Math Inc Ped Occ Cult Home Edur Gr
Belgium
Maths ability 8597 514.53 102.27
Household income 2738 3.78 1.77 0.43
Parents’ education 8129 4.86 1.24 0.29 0.48
Occupational status 8153 51.62 21.83 0.42 0.52 0.51
Cultural possessions 8287 −0.31 0.99 0.28 0.26 0.28 0.34
Home possessions 8503 0.08 0.80 0.17 0.31 0.21 0.22 0.18
Educational resources 8474 0.09 0.95 0.31 0.21 0.19 0.24 0.39 0.22
Grade 8483 −0.45 0.67 0.59 0.28 0.20 0.29 0.19 0.15 0.25
Learning time 4821 5.33 0.34 0.32 0.25 0.14 0.18 0.10 0.07 0.07 0.07
Chile
Maths ability 6856 422.63 80.76
Household income 6277 3.29 1.89 0.46
Parents’ education 6665 4.05 1.54 0.39 0.57
Occupational status 6499 40.84 21.43 0.42 0.63 0.63
Cultural possessions 6750 −0.21 0.96 0.23 0.27 0.27 0.26
Home possessions 6776 −0.59 1.01 0.39 0.62 0.50 0.53 0.29
Educational resources 6773 −0.49 1.01 0.24 0.36 0.32 0.29 0.37 0.50
Grade 6856 −0.27 0.71 0.38 0.14 0.12 0.09 0.06 0.11 0.14
Learning time 3733 5.88 0.47 −0.03 −0.01 −0.02 −0.01 0.00 −0.02 −0.01 −0.01
Croatia
Maths ability 5008 471.13 88.48
Household income 4445 2.90 1.35 0.31
Parents’ education 4978 4.56 1.11 0.23 0.44
Occupational status 4769 45.37 20.57 0.36 0.55 0.57
Cultural possessions 4823 −0.21 1.02 0.26 0.31 0.30 0.35
Home possessions 4996 −0.45 0.74 0.12 0.35 0.23 0.25 0.24
Educational resources 4986 0.05 0.90 0.15 0.16 0.14 0.15 0.34 0.24
Grade 5008 0.20 0.40 0.12 0.01 0.00 0.03 0.02 0.00 0.01
Learning time 3296 4.94 0.34 0.44 0.25 0.17 0.27 0.18 0.12 0.12 0.12
Germany
Maths ability 5001 513.53 96.31
Household income 2111 3.90 1.80 0.33
Parents’ education 3936 4.30 1.58 0.29 0.37
Occupational status 4034 50.91 20.64 0.39 0.51 0.52
Cultural possessions 4151 0.10 0.99 0.24 0.26 0.30 0.34
Home possessions 4244 0.21 0.80 0.11 0.32 0.12 0.14 0.12
Educational resources 4235 0.20 0.87 0.24 0.21 0.17 0.22 0.42 0.21
Grade 5001 0.27 0.67 0.46 0.17 0.18 0.20 0.12 0.10 0.16
Learning time 2551 5.24 0.30 −0.13 −0.08 −0.05 −0.08 −0.02 −0.01 −0.04 −0.04
Hungary
Maths ability 4810 477.04 93.63
Household income 3762 2.81 1.63 0.43
Parents’ education 4692 4.45 1.41 0.36 0.46
Occupational status 4471 46.14 20.61 0.43 0.51 0.60
Cultural possessions 4717 0.17 1.02 0.40 0.34 0.36 0.36
Home possessions 4764 −0.54 0.78 0.19 0.40 0.35 0.27 0.24
Educational resources 4760 −0.10 0.92 0.32 0.27 0.29 0.24 0.47 0.36
Grade 4810 0.06 0.63 0.37 0.17 0.12 0.16 0.19 0.14 0.23
Learning time 3072 4.98 0.27 0.08 0.01 0.02 0.03 0.03 0.00 0.03 0.03
(Continued)
 OXFORD REVIEW OF EDUCATION 785

Table A1. (Continued).

N Mean Std. Math Inc Ped Occ Cult Home Edur Gr
Korea
Maths ability 5033 553.77 99.09
Household income 4935 3.65 1.80 0.27
Parents’ education 4983 4.94 1.19 0.23 0.37
Occupational status 4939 53.38 18.09 0.22 0.35 0.42
Cultural possessions 4950 0.27 0.94 0.24 0.32 0.29 0.26
Home possessions 5029 −0.70 0.62 0.15 0.47 0.27 0.27 0.35
Educational resources 5017 −0.09 0.96 0.30 0.28 0.24 0.21 0.50 0.36
Grade 5033 −0.06 0.24 0.09 −0.01 −0.03 −0.02 0.00 0.00 0.02
Learning time 3232 5.32 0.28 0.30 0.15 0.11 0.10 0.13 0.12 0.12 0.12
Hong Kong
Maths ability 4670 561.24 96.32
Household income 4401 3.69 1.88 0.24
Parents’ education 4477 3.38 1.57 0.24 0.51
Occupational status 4307 46.21 21.00 0.23 0.55 0.60
Cultural possessions 4573 −0.13 0.98 0.18 0.31 0.36 0.32
Home possessions 4579 −0.96 0.70 0.13 0.50 0.39 0.43 0.33
Educational resources 4577 −0.26 0.97 0.20 0.24 0.31 0.26 0.45 0.31
Grade 4670 −0.41 0.68 0.30 0.22 0.16 0.17 0.11 0.17 0.12
Learning time 2881 5.56 0.24 0.02 0.01 0.03 −0.01 0.05 0.04 0.06 0.06
Macau
Maths ability 5335 538.13 94.50
Household income 5002 3.28 1.78 0.18
Parents’ education 5252 2.97 1.57 0.11 0.35
Occupational status 5195 44.81 16.95 0.11 0.41 0.48
Cultural possessions 5289 −0.40 0.98 0.20 0.17 0.27 0.19
Home possessions 5316 −0.53 0.84 0.02 0.37 0.26 0.30 0.23
Educational resources 5312 −0.30 1.04 0.23 0.17 0.26 0.18 0.44 0.26
Grade 5335 −0.82 0.90 0.55 0.16 0.16 0.13 0.17 0.10 0.23
Learning time 3474 5.59 0.24 0.20 0.03 0.01 0.04 0.07 0.03 0.05 0.05
Portugal
Maths ability 5722 487.06 93.96
Household income 4279 2.95 1.93 0.46
Parents’ education 5572 3.18 1.96 0.35 0.55
Occupational status 5468 42.64 21.40 0.41 0.63 0.70
Cultural possessions 5578 0.06 1.01 0.30 0.35 0.40 0.40
Home possessions 5642 0.23 0.91 0.26 0.42 0.35 0.40 0.33
Educational resources 5637 −0.11 0.89 0.28 0.27 0.30 0.29 0.43 0.33
Grade 5209 −0.52 0.75 0.58 0.29 0.22 0.28 0.19 0.21 0.23
Learning time 3230 5.60 0.36 0.15 0.12 0.09 0.11 0.08 0.08 0.07 0.07
Copyright of Oxford Review of Education is the property of Routledge and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.