murdoch research repository · study by song (2011), which examined substantial disparities in...

MURDOCH RESEARCH REPOSITORY

This is the author’s final version of the work, as accepted for publication following peer review but without the publisher’s layout or pagination.

The definitive version is available at

http://dx.doi.org/10.1080/13803611.2014.892432

Song, S., Perry, L.B. and McConney, A. (2014) Explaining the achievement gap between Indigenous and non-Indigenous students: An analysis of PISA 2009

results for Australia and New Zealand. Educational Research and Evaluation . pp. 1-21.

http://researchrepository.murdoch.edu.au/21398/

Copyright: © 2014 Taylor & Francis It is posted here for your personal use. No further distribution is permitted.

http://dx.doi.org/10.1080/13803611.2014.892432http://researchrepository.murdoch.edu.au/21398/

Explaining the achievement gap between Indigenous andnon-Indigenous students: an analysis of PISA 2009 results forAustralia and New Zealand

Steve Songa, Laura B. Perryb* and Andrew McConneya

aGeorge Fox University, Newberg, OR, USA; bMurdoch University, Perth, Western Australia

(Received 24 September 2013; final version received 27 December 2013)

This study investigates the relative roles of home and school variables in accounting forachievement gaps between Indigenous and non-Indigenous students in Australia andNew Zealand. Using data from the Programme for International Student Assessment[PISA] 2009, our findings show that achievement gaps between Indigenous and non-Indigenous students are associated with both home and school resources, not only interms of unequal allocations but also in relation to differences in the rates at whichhome and school affordances are converted into positive educational outcomes. In bothcountries, home resources accounted for more of the achievement gap than differencesin schooling resources. However, the achievement gap between Indigenous and non-Indigenous students is substantially larger in Australia than in New Zealand,apparently related to greater inequity in the allocation of school resources. We suggestthat education policymakers in Australia ensure a more equitable allocation of schoolresources between Indigenous students and their non-Indigenous peers.

Keywords: academic performance; PISA; Indigenous students

Introduction

In many countries, students from some ethnic minority groups have consistently lower edu-cational outcomes than their peers from ethnic majority groups. For example, in Europe,Turkish students (Simon, 2003; Song, 2011; Van Ewijk, 2011; Worbs, 2003) and Roma stu-dents (Gimenez Adelantado, 2003; Open Society Institute, 2007) and in the US, Latino andAfrican American students (Darling-Hammond, 2010; Fleischman, Hopstock, Pelczar, &Shelley, 2010; Losen & Orfield, 2002; National Center for Education Statistics, 2002)have consistently lower educational outcomes than their peers from dominant ethnicgroups. While it is normal and natural that educational outcomes vary between individuals,stable and substantial differences in educational outcomes between groups of individualsare a cause for concern. Such differences suggest that social and educational forces, pol-icies, and structures are systematically privileging some groups over others. Examininghow schools and education systems can be reformed for the purpose of reducing achieve-ment gaps between groups of students has therefore received considerable attention in theeducation research literature. International case studies can illuminate how context, cultural

© 2014 Taylor & Francis

*Corresponding author. Email: [email protected]

Educational Research and Evaluation, 2014http://dx.doi.org/10.1080/13803611.2014.892432

5

10

15

20

25

30

35

40

45

NERE892432 Techset Composition India (P) Ltd., Bangalore and Chennai, India 2/14/2014

mailto:[email protected] NoteAndrew is at Murdoch, not George Fox.The superscript for Andrew should be "b" not "a".

dynamics, and educational systems may be related to achievement gaps between ethnicminority students and their majority peers. This understanding can then lead to concrete rec-ommendations about ways to reduce achievement gaps.

Additionally, our motivation for this study is the severe poverty that many Indigenouspeople in both New Zealand and Australia continue to experience. On all major social indi-cators, Indigenous people in both countries continue to be at substantial disadvantage com-pared to their non-Indigenous peers. In Australia, for example, the life expectancy rate is upto 20 years less for Indigenous people than for non-Indigenous people (Oxfam, 2013). Indi-genous people are also 3 times more likely to be unemployed than non-Indigenous peopleand 3 times more likely to die of accident, violence, or self-harm (Australian Bureau of Stat-istics, 2009). Large inequalities in health and wellbeing between Indigenous and non-Indi-genous people also exist inNewZealand (Ministry of Social Development, 2010). Educationis strongly related to social development, poverty alleviation, and health and wellbeing(McMahon, 2002; Reimers, 2003). It is also a major tool for the empowerment of disenfran-chised people (Freire, 2002). Finding ways to improve the educational experiences and out-comes of Indigenous students in both countries is therefore a policy priority, and comparingthe experiences of two similar countries, as in this paper, can provide insight about ways toreduce educational inequity. To our knowledge, limited research has attempted to comparethe educational opportunities and outcomes of Indigenous populations in New Zealandand Australia, with some recent exceptions (e.g., Woods-McConney, Oliver, McConney,Maor, & Schibeci, 2013).

This study examines long-standing gaps in educational achievement between Indigen-ous and non-Indigenous students in Australia and New Zealand. It builds on a previousstudy by Song (2011), which examined substantial disparities in educational outcomesbetween Turkish and native students in Austria, Germany, and Switzerland. Turkish stu-dents show the lowest academic performance of any ethnic minority group in thesecountries. In Australia and New Zealand, the ethnic minority group that has the lowest aca-demic performance is Indigenous and Māori students, respectively. Like Song’s study, thispaper examines school and student factors that could contribute to our understanding ofacademic performance gaps between ethnic minority groups and their peers from thelarger society. By comparing the achievement of Indigenous students in two similarcountries, our purpose is to better understand how educational systems could be mademore equitable for the purpose of reducing academic performance gaps between groupsof students.

The present study comparatively examines relationships among student home back-grounds, school characteristics, and academic achievement to determine the extent towhich these variables are associated with gaps in educational outcomes between Indigenousand non-Indigenous students in Australia and New Zealand. Specifically, using the Oaxaca-Blinder decomposition, a variant of statistical methodology most often applied in laboureconomics (Blinder, 1973; Oaxaca, 1973), the study assesses the extent to which differ-ences in resources, both at home and school, contribute to explaining the achievementgap observed between these groups. Furthermore, the present study examines differencesin rates of return for the allocation of these resources. In other words, in explaining theachievement gap between Indigenous students and non-Indigenous students, the presentinvestigation asks the following research questions:

(1) How different are Indigenous and non-Indigenous students in Australia and NewZealand in terms of student resources (e.g., parental education, number of books

2 S. Song et al.

50

55

60

65

70

75

80

85

90

at home) and school resources (e.g., supply of instructional materials, quality of theteaching staff)?

(2) To what extent are differences in student and school resources associated withdifferences (gaps) in educational outcomes between Indigenous and non-Indigen-ous students in Australia and New Zealand?

(3) Further, and directly related to our use of the Oaxaca-Blinder decomposition, aredifferences in rates of return evident for student and school resources between Indi-genous and non-Indigenous students in Australia and New Zealand?

The undergirding rationale and intended practical significance of our choice of theOaxaca-Blinder decomposition method for analysing the data for this study is that edu-cational researchers and policymakers must be aware of not only unequal allocations ofresources among groups but also differences in the extent and/or efficiency with whichresources may be utilized. Importantly, this allows researchers and policymakers insightinto what resources are likely to work in redressing a social issue as well as providing ameasure of how well or poorly any particular resource may be utilized. Put another way,the simple provision of a resource, while important, does not in itself necessarily meanthat the resource is being utilized efficiently, or to its full potential. Theoretically, the under-utilization of a particular resource could be due to a variety of factors, including a lack ofawareness that it is available, a lack of understanding of how it could potentially be used, oran inability to take advantage of the resource because of poor access. For example, even ifafter-school tutorial services are provided, we need to ask if students and parents are awareof the programme, and if appropriate transportation is available for students if parents arenot able to provide it, as may be the case for many working-class families. Of course, wealso readily acknowledge that the Oaxaca-Blinder decomposition method cannot tell uswhy or how the resources are unequally allocated or unevenly utilized. Rather, theOaxaca-Blinder decomposition can show us that there is evidence of the existence ofsuch differences, and the strength of their associations with the policy issue being targeted(in this case, group differences in educational achievement). Such divides must be first seenbefore they can be investigated and perhaps remediated.

The article is structured in five parts. First, we provide an overview of the recent aca-demic performance of Indigenous students in Australia and New Zealand, as measuredin large-scale assessments, and briefly describe the educational contexts of the twocountries. Second, we describe the dataset used in the current study (Programme for Inter-national Student Assessment [PISA], 2009) and overview the analytic method used. Third,we describe the main findings of this secondary analysis. Last, we discuss the major find-ings of the analysis and offer several conclusions for consideration.

Performance of Indigenous students in Australia and New Zealand

Australia and New Zealand were purposefully chosen for this study. Despite obvious, dra-matic differences in their physical geographies, Australia and New Zealand are neverthelessco-located in the same region of the world (on either side of the Tasman Sea) and sharestrong commonalties in sociocultural histories and traditions (e.g., both are long-standingparliamentary democracies, with shared histories of British colonial rule and are currentmembers of the Commonwealth), in the proportions of their populations (about 20%)who are recent migrants, in systems of state-supported comprehensive secondary schooling,and in perennially high performance on comparative international assessments like PISA.

Educational Research and Evaluation 3

95

100

105

110

115

120

125

130

135

Typically, students in New Zealand and Australia have been among the top performerson international assessments. For example, and directly relevant to this secondary analysis,if the PISA results of the 15 top-performing countries are compared across time, there areonly 5 countries that have performed above the Organisation for Economic Co-operationand Development (OECD) average in all tests since the assessment began in 2000.These are Finland, South Korea, Canada, New Zealand, and Australia. In the most recentpublicly-released PISA study (2009), the top overall performing countries/regions are asfollows: Shanghai (ranked 1st in all categories in its first participation in 2009), Finland,South Korea, Hong Kong, Singapore, Canada, New Zealand, Australia, The Netherlands,and Estonia (Knighton, Brochu, & Gluszynski, 2010; OECD, 2010).

Although Australia and New Zealand as a whole have performed consistently well, therecontinues to be substantial gaps in educational attainment and in other life outcomes (e.g.,health, life expectancy, employment) between Indigenous and non-Indigenous peoples, withadvantages often going to the latter (Australian Bureau of Statistics, 2012; Sullivan, Perry,& McConney, 2013; Woods-McConney et al., 2013). Thus, some may argue that in Australiaand New Zealand, the overall high achievement has masked, to some extent, issues aroundequity in educational achievement for some groups. In both countries, for example, there arelongstanding and substantial disparities in achievement between ethnic groups. For NewZealand, PISA scores of Māori and Pacifika students are much lower than the average forPäkehä/European students; and although the gap has narrowed slightly for Māori, the gap isnot closing nearly fast enough for Indigenous students to catch up to their Päkehä/Europeanpeers (Telford, 2010). Specifically, students identifying as Päkehä/European (71% of all stu-dents) achieved an average reading score of 541 score points in reading literacy on PISA2009. In contrast, students identifying as Māori (19%) and Pacifika (10%) scored 478 scorepoints and 448 score points, respectively, considerably below theOECDmean (Telford, 2010).

For Australian Indigenous students, the picture may be even bleaker. As noted by Thomp-son and colleagues at theAustralianCouncil forEducationalResearch (ACER) in their analysisof PISA 2009 results, on average, Indigenous students lagged behind their non-Indigenouspeers by 82 points in reading literacy. This gap equates to more than one PISA proficiencylevel ormore than 2 years of schooling. Australian Indigenous students also performed signifi-cantly lower (57 score points) than the OECD average (Thompson, De Bortoli, Nicholas,Hillman, & Buckley, 2010). On the positive side, Indigenous Australians’ participation inearly childhood and primary schooling has improved: Retention rates in the final year ofhigh school (Year 12) have notably increased, and participation rates of 15- to 24-year-old Indi-genous students in vocational training approximately equate those of the total population(McRae et al., 2000). On average, however, substantial difficulties remain for Indigenous Aus-tralians, who in comparison to their non-Indigenous peers are less likely to attend preschool,have less access to secondary school in the communities where they live, have 2 to 3 timesthe rate of absenteeism of other students, leave school much younger than non-Indigenous stu-dents, and are less than half as likely to complete high school (McRae et al., 2000).

In summary, Australia and New Zealand have been perennially high-performingcountries in international assessments of educational achievement, including PISA. Inboth countries, however, significant challenges remain around equity of access and out-comes for Indigenous students.

Education context of Australia and New Zealand

Education systems of Australia and New Zealand share many similarities. Both systemsoffer comprehensive secondary education rather than the differentiated system of secondary

4 S. Song et al.

140

145

150

155

160

165

170

175

180

education common in Europe. New Zealand and Australia have similar years of compulsoryeducation, share educational philosophies based on progressivism and constructivism. InNew Zealand, because of its smaller size, education is the responsibility of the national gov-ernment, whereas in Australia, education remains the purview of state governments. Thismeans that in Australia, some education policies vary slightly by state. Nevertheless, theAustralian federal government plays a large role in education. For example, Australiahas a national curriculum (as does New Zealand), national standardized testing, andnational reporting of school performance indicators. The federal government also providesfunding to private schools and, to a lesser extent, public schools as well.

School choice is supported in both education systems. Within the public (government)sector, students may apply for admission to the public school of their choice, subject to theavailability of places. Schools that receive more enrolment requests than places must givepriority to students who reside within the school’s enrolment area (Australia) or home zone(New Zealand). These “over-enrolled” schools are typically in wealthy communities inlarge cities. Secondary schools may also select out-of-area students based on their academicperformance or demonstrated talent in the arts or sport.

School choice is also promoted in both countries by the public funding of non-government(private) schools, including faith-based schools. The funding mechanisms of non-governmentschools vary between the two systems. InNewZealand, non-government schoolsmay become“state-integrated” schools; in which case, they receive full public funding for their operatingcosts but cannot charge tuition fees. These are typically Catholic schools. Non-governmentschools may also remain “independent”, in which case they receive much less publicfunding but are able to (and do) charge tuition. Independent schools receive a per-pupilpublic subsidy based on the year level of the student, ranging from approximately $1,000New Zealand dollars (USD $8201) for primary school students to $2,200 (USD $1,803) forupper secondary students (New Zealand Ministry of Education, 2012a). In Australia, “state-integrated” schools do not exist. All private schools receive federal public subsidies, and allprivate schools charge tuition. Tuition can range from $7,000 Australian dollars (USD$6,345) at a Catholic school to $25,000 (USD$22,662) or more at a Protestant or non-denomi-national private school. To set these figures in context, the annual median household income inNew Zealand is $67,808 (USD $55,572) and $74,984 (USD $67,972) (Australian Bureau ofStatistics, 2013; Statistics New Zealand, 2013).

Accompanying these differences in the public funding of non-government schools aredifferences in enrolment patterns across school sectors. In New Zealand, approximately 4%of students attend a private school, 11% attend a state-integrated school, and 85% attend apublic school (New Zealand Ministry of Education, 2012b). In Australia, approximately67% of students attend a public school, and this number drops to 62% for students inhigh school (Australian Bureau of Statistics, 2006). These enrolment figures show thatthe proportion of high school students attending schools that charge tuition varies dramati-cally between the two countries: less than 4% in New Zealand versus 38% in Australia.

Method

Data

PISA is an internationally standardized assessment administered every 3 years to 15-year-oldsin schools. The survey was implemented in 43 countries in the first assessment round in 2000,41 countries in 2003, and57countries in 2006. For the 2009assessment, 65 countries and econ-omies are represented. Tests are typically administered to between 4,500 and 10,000 students in


185

190

195

200

205

210

215

220

225

20060040Inserted Textand

each country. Schools in each country are randomly selected by the international contractor forparticipation in PISA. At these schools, the test is given to students who are between age 15years 3 months and age 16 years 2 months at the time of the test, rather than to students in aspecific year of school. An average age of 15 was chosen because at this age young peoplein most OECD countries are nearing the end of compulsory education.

As a large-scale international assessment, PISA assesses students in three subjects –reading, mathematics, and science. Within each subject, the main focus is not to assesshow well students have mastered a specific curriculum, but to assess their ability toapply the knowledge and skills learned at school to solve real-life problems (OECD,2001). The scores are standardized to an international mean of 500 with a standard devi-ation of 100. The datasets used in this investigation of PISA 2009 comprise the following:1,143 Indigenous and 13,108 non-Indigenous students for Australia; and 833 Indigenous(Māori) and 3,768 non-Indigenous students for New Zealand. These groupings werederived from students’ responses to PISA’s student questionnaire, which asked studentsto indicate their ethnic background. For Australia, students who indicated Aboriginal orTorres Strait Islander ethnicity, and for New Zealand those students who indicatedMāori, were designated “Indigenous” for this analysis. All other students in the twosamples, regardless of ethnic background or immigrant status, were considered “non-Indigenous”. It should also be noted that Australia intentionally oversamples its Indigenousstudents, who constitute about 2% of the population, so that this subsample is large enoughin PISA to accommodate analyses that meet national reporting priorities. Differences in theproportions of male and female Indigenous students in the two countries, however, arelikely reflective of actual differences in the population.

Data analysis

Observed differences in educational outcomes between Indigenous and non-Indigenousstudents in Australia and New Zealand may be related to several factors. First, non-Indigen-ous students may have, as a group, higher proportions of highly educated parents and/ormore resources at school. Second, the relationships of different school and student-levelfactors with educational achievement may differ for Indigenous students as compared totheir non-Indigenous counterparts. In other words, the same characteristics (variables)may have different “rates of return” for Indigenous and non-Indigenous students, asmeasured by scores on standardized assessments, and reflected by substantial gaps in edu-cational outcomes between the two groups.

Analysis of suchobserveddifferences ingroupcharacteristics and associated academicout-comes can be performedusing a technique called theOaxaca-Blinder decomposition employedmost often in labour economics (Blinder, 1973; Oaxaca, 1973). The technique is simple toapply as it requires only coefficient estimates from linear regressions for the outcome variableand sample means of the explanatory (independent) variables. In essence, the Oaxaca-Blinderdecomposition compares ordinary least squares (OLS) regression models for multiple groupsto estimate differences in the association of each explanatory variables (e.g., parental educationlevel, gender, etc.) on the outcome variable in question (e.g., test scores).2

The classic Oaxaca-Blinder decomposition equation can be expressed as:3

�sn − �sT = bn �x′T − �x

′r

( )+ bn − br( )

�x′T

For the purposes of this study, �sn −�sT represents the gap in PISA test scores between Indi-genous and non-Indigenous student groups. The first term on the right side of the equation,

6 S. Song et al.

230

235

240

245

250

255

260

265

270

20060040Cross-Out

bn �x′n − �x

′T

( ), reflects mean differences in student and school resources between the two

groups, also referred to as “endowments”. The second term on the right hand side,bn − bT( )

�x′T , is referred to as “coefficients”, which represents differences attributed to

“rates of return” (i.e., the relationship between the dependent and independent variables)for each explanatory variable or characteristic. In labour economics, endowments arealso known as the “explained” portions of observed differences, whereas coefficients areoften referred to as “unexplained” portions of observed differences (or gaps in testscores, as in the current case).

A useful example to clarify the difference between endowments and coefficients can beseen in studies examining the relationship between access to clean water and health of poorand less poor children in some developing countries. Two studies in particular found thatchildren in some of the poor regions in India and Vietnam may be less healthy not onlybecause they have less access to clean, piped water (endowments or the “explainedportion”) but also because their parents tend to be less knowledgeable about how to maxi-mize health benefits from it (coefficients or the “unexplained” portion) (Jalan & Ravallion,2003; Wagstaff & Nguyen, 2004).

To avoid imprecise estimated coefficients due to large differences in sample sizesbetween Indigenous and non-Indigenous students, Neumark’s (1988) pooled sampleapproach was applied. The non-Indigenous group was then used as the reference group.This variant of the standard Oaxaca-Blinder model has the added advantage of using thesame coefficient estimates for weighting the explained portion of the decomposition,which allows for a more direct comparison between groups (Appleton, Hoddinott, &Krishnan, 1999; Rangvid, 2007).

Variables

In the current study, PISA 2009 reading scores served as the outcome variable. Wechose reading as an indicator of student achievement because it is a foundational skillthat is important for learning in all curriculum areas. Reading literacy is also commonlyused to compare levels of educational attainment and human capital across countries. Welist below explanatory (independent) variables used in our analysis. These variables wereselected because theory and empirical research have shown that they are related tostudent educational outcomes (OECD, 2010). All of these explanatory variables havebeen used in many secondary analyses (and the OECD’s primary analysis) of thePISA dataset using a variety of multivariate statistical methods, including regression-based Oaxaca-Blinder (see Lounkaew, 2013, for a recent example). Regression analysesoften use a mixture of ordinal and interval data, and this is considered “acceptable,appropriate, and quite useful” (Meyers, Gamst, & Guarino, 2006, p. 23) by mostresearchers.

Student-level variables

Parental education level. This variable was measured using the International StandardClassification of Education (ISCED). In this analysis, we used the higher of the parent edu-cation levels (HISCED). The categories were as follows: 0 = “Did not go to school”, 1 =“primary education”, 2 = “lower secondary education”, 3B/3C = “upper secondaryschool with entry into the labour market”, 3A and 4 = “upper secondary school providingentry into tertiary education”, and 5A/5B/6 = “tertiary education and/or beyond”.


275

280

285

290

295

300

305

310

315

Parental occupational status. This variable was measured using the International Socio-economic Index of Occupational Status (ISEI), based on a hierarchical coding scheme com-prising 390 different occupational categories provided by the International Labour Office.In this study, we used the higher of either the father’s or mother’s occupation (HISEI) tocapture the attributes of the occupations that convert parent’s education into income. Thevalue on the index ranged from 0 (lowest) to 90 (highest).

Gender. As a categorical variable, female was indexed as “1” and male as “2”.

Language. This categorical variable was labelled, 0 = “spoke the same test language athome” or 1 = “spoke a different language than the test language at home”.

Computers at home. The number of computers at home was set on a 4-point scale, from1 = “0 computers at home”, 2 = “one”, 3 = “two”, and 4 = “three or more”.

Books at home. The number of books at home was set on a 6-point scale, from 1 = “0 to10 books”, 2 = “11 to 25 books”, 3 = “26 to 100 books”, 4 = “101 to 200 books”, 5 = “201 to500 books”, to 6 = “more than 500 books”.

School-level variablesSchool community. Past research suggests that academic achievement varies depending

on whether students live in rural or urban settings (OECD, 2000). To assess the associationof the size of the communities in which participants receive their schooling, a categoricalvariable was created with 1 = “village” (less than 3,000 residents), 2 = “small town”(3,000 to 15,000 residents), 3 = “town” (15,000 to 100,000 residents), 4 = “city”(100,000 to 1,000,000 residents), and 5 = “large city” (greater than 1,000,000 residents).

School size. This variable measured the number of students enrolled at each participat-ing school.

Instructional materials. Each school principal was asked to assess how well or poorlytheir schools were resourced as far as instructional materials were concerned. The 4-pointscale ranged from 1 = “not at all” (i.e., “The school did not lack instructional materials”) to4 = “a lot” (i.e., “The school very often experiences lack of instructional materials”).

Computers. As a measure of how well schools were resourced in terms of their edu-cational resources/materials, the analysis included the extent to which schools wereequipped with computers. The 4-point scale options ranged from 1 = “not at all” (i.e.,“The school did not lack computers”) to 4 = “a lot” (i.e., “The school very often experienceslack of computers”).

Qualified teacher supply. As a composite variable, this measure was the average of teachershortages for each school in reading, math, and science. As with instructional materials andcomputers, the 4-point scale ranged from 1 = “not at all” (i.e., “The school did not lack tea-chers”) to 4 = “a lot” (i.e., “The school very often experiences lack of teachers”).

Proportion of teachers with advanced degrees. This variable assessed the prevalence of“highly qualified” teachers in the schools by examining the percentage of teachers in eachschool with the highest level of education (i.e., graduate degrees). This was calculated bydividing the number of teachers with graduate degrees by the total number teachers at eachschool.

Proportion of certified teachers. This variable was calculated by dividing the number ofcertified teachers by the number of all the teachers at each of the schools.

8 S. Song et al.

320

325

330

335

340

345

350

355

360

Student–teacher ratio. This variable was calculated by dividing the number of studentsby the number of teachers for each school.

Results

Descriptive analysis

Tables 1 and 2 present descriptive statistics that compare student- and school-level variablesfor Indigenous and non-Indigenous students in Australia and New Zealand. In bothcountries, on average, parents of Indigenous students had fewer years of formal educationand lower levels of occupational status. Indigenous students also reported fewer numbers ofbooks and computers at home. The two countries also differed in the use of the test languageat home, between Indigenous and non-Indigenous students. In Australia, non-Indigenousstudents were more likely to speak English at home than their Indigenous counterparts,although the opposite was the case for New Zealand. In terms of gender distribution,Australia had a slightly higher percentage of girls among Indigenous students thanamong non-Indigenous students, whereas the opposite was true for the New Zealandsample. As noted previously, these small differences in the proportions of male andfemale Indigenous students in the two countries are most likely reflective of actual differ-ences in the population.

Both countries showed varying levels of support and resources between Indigenous andnon-Indigenous students on several school variables. In Australia, for example, Indigenousstudents, on average, were more likely than their non-Indigenous peers to be enrolled inschools in less populous communities, attend smaller schools, and attend schools reportinggreater shortages of instructional materials, computers, and qualified teachers. Australian

Table 1. Student and school variables for Indigenous and non-Indigenous students in Australia,PISA 2009.

Indigenous Non-Indigenous

(n = 1,143) (n = 13,108)

Mean SD Mean SD

Student Academic PerformanceReading 435.13 101.98 514.46 95.58Student VariablesHISCED 4.44 1.14 4.83 1.17HISEI 46.32 15.22 53.83 16.03Gender (% male) 47.94 NA 49.37 NALanguage (% test language) 85.48 NA 89.08 NANumber of computers at home 2.74 0.89 3.24 0.81Number of books at home 3.03 1.39 3.73 1.37School VariablesSchool community 3.26 1.09 3.77 1.16School size 911.52 393.34 965.97 418.79Shortage of instructional materials 1.75 0.84 1.57 0.77Shortage of computers 2.23 0.84 1.94 0.91Shortage of qualified teachers 2.20 0.88 1.90 0.86Proportion of qualified teachers 0.95 0.16 0.96 0.14Proportion of certified teachers 0.99 0.09 0.98 0.09Student-teacher ratio 13.62 1.96 13.51 2.08

Note: Calculations were performed using weighted PISA 2009 data.


365

370

375

380

385

390

395

400

405

Indigenous students, however, were also more likely to matriculate in schools with slightlysmaller student–teacher ratios. In terms of existing proportions of qualified and certified tea-chers, observed differences between Indigenous and non-Indigenous students in Australiawere negligible.

Similarly, in New Zealand, Indigenous students were more likely than non-Indigenousstudents to be in schools located in less densely populated communities, attend smallerschools, and matriculate in schools reporting greater shortages of qualified teachers. Aswith Australian Indigenous students, Indigenous students in New Zealand were morelikely to matriculate in schools with slightly smaller student–teacher ratios. In terms ofshortages of instructional materials and computers, and proportions of qualified and certi-fied teachers, the schools attended by Indigenous and non-Indigenous students in NewZealand did not differ greatly.4

Regression analysis

Having assessed differences on student and school variables between Indigenous and non-Indigenous groups in Australia and New Zealand at a descriptive level, the next phase of theinvestigation analysed the relative roles of each of these variables using OLS regression. Onaverage, as shown in Tables 3 and 4, the results of this analysis suggest that among student-level variables, almost all were found to have meaningful relationships to readingachievement.

For Indigenous students in Australia, gender, language spoken at home, and the numberof books at home were variables showing the strongest relationships to reading perform-ance. On average, being male was associated with 39 fewer points, and speaking a language

Table 2. Student and school variables for Indigenous and non-Indigenous students in New Zealand,PISA 2009.

Indigenous Non-Indigenous(n = 833) (n = 3,768)

Mean SD Mean SD

Student Academic PerformanceReading 479.34 95.39 533.96 96.31Student VariablesHISCED 4.17 1.43 4.67 1.28HISEI 47.96 15.59 54.19 15.23Gender (% male) 52.43 NA 50.12 NALanguage at home (% test language) 90.61 NA 85.04 NANumber of computers at home 2.66 0.88 2.92 0.86Number of books at home 3.24 1.32 3.71 1.32School VariablesSchool community 3.33 1.20 3.72 1.16School size 1092.94 600.04 1279.87 619.61Shortage of instructional materials 1.56 0.68 1.53 0.66Shortage of computers 2.29 0.95 2.26 0.95Shortage of qualified teachers 2.07 0.85 1.92 0.85Proportion of qualified teachers 0.91 0.10 0.92 0.14Proportion of certified teachers 0.97 0.08 0.96 0.09Student-teacher ratio 14.55 2.69 15.10 2.13

Note: Calculations were performed using weighted PISA 2009 data.

10 S. Song et al.

410

415

420

425

430

435

440

445

450

other than English at home was associated with almost 83 fewer points in reading perform-ance. Number of books at home was positively associated with reading performance forAustralian Indigenous students, that is, students with more books at home tended to havehigher reading scores than their counterparts with fewer books. Similarly, for non-Indigen-ous Australian students, males trailed their female counterparts by almost 32 points, con-trolling for other variables. The number of books at home was also a significant, positivepredictor of reading performance for non-Indigenous Australian students, with a 15-pointincrease in reading performance associated with each unit increase in number of booksat home.

Table 3 shows that for Indigenous and non-Indigenous students in Australia, gender,language spoken at home, and the number of books at home were the strongest predictorsof reading performance. Controlling for other variables, being male was associated with a

Table 3. Student and school predictors for Indigenous and non-Indigenous Australian students inPISA 2009 reading.

Indigenous (n = 1,143) Non-Indigenous (n = 13,108)

Unstandardized Standardized SE Unstandardized Standardized SE

Student VariablesHISCED −1.22 −0.02 3.13 9.34*** 0.12 0.86HISEI 0.76** 0.13 0.25 0.83*** 0.15 0.06Gender −38.98*** −0.22 6.64 −31.90*** −0.18 1.75Language at home −82.69*** −0.20 14.13 −5.66 −0.02 3.69Number ofcomputers athome

9.47* 0.10 4.25 8.74*** 0.08 1.15

Number of booksat home

14.19*** 0.22 2.30 15.52*** 0.23 0.72

School VariablesSchoolcommunity

0.30 0.01 3.26 7.26*** 0.09 0.87

School size 0.03** 0.11 0.01 0.01** 0.03 0.01Shortage ofinstructionalmaterials

−4.22 −0.04 4.54 −3.50* −0.03 1.61

Shortage ofcomputers

−2.01 −0.02 4.62 2.26 0.02 1.22

Shortage ofqualifiedteachers

−2.67 −0.03 3.80 −6.43*** −0.06 1.19

Proportion ofqualifiedteachers

27.41 0.06 17.77 18.56*** 0.03 5.90

Proportion ofcertifiedteachers

−32.77 −0.02 38.31 −15.27 −0.02 9.93

Student–teacherratio

0.22 0.01 1.89 −1.26** −0.03 0.44

AdjustedR-square

0.22 0.24

Note: Calculations were performed using weighted PISA 2009 data. Standard errors were calculated at the schoollevel.*p ≤ .05. **p ≤ .01. ***p ≤ .001.


455

460

465

470

475

480

485

490

495

39-point gap in reading performance for Indigenous students, and a 32-point gap for non-Indigenous students. In terms of speaking the test language (English) at home, Indigenousstudents who spoke a language other than English at home lagged behind those who spokeEnglish at home by almost 83 points. This very large difference was not evident amongnon-Indigenous Australian students. For number of books at home, each additional unitincrease was associated with a 14-point increase in reading for Indigenous students and asimilar 15.5-point increase for their non-Indigenous counterparts. Also significant were par-ental education (HISEI) for Indigenous students and both parental education and occu-pational status (HISCED) for non-Indigenous students. These statistically significantrelationships suggest that Australian students with parents who had higher levels of edu-cation and/or occupational status also had higher reading performance scores.

Table 4 presents similar findings for New Zealand. Among the student-level variables,parental education, gender, and number of books at home showed the strongest relation-ships with reading performance for both Indigenous and non-Indigenous students. Control-ling for other variables, being female was associated with a 42-point advantage in readingfor Indigenous students, and a 39-point advantage for non-Indigenous students, respect-ively. Furthermore, as for Australian students, the number of books at home was stronglyrelated to reading performance on PISA for students in New Zealand. As to parental edu-cation levels, both groups showed that those students whose parents had more years of edu-cation had demonstrated higher reading performance. Each additional unit increase in thenumber of books at home was associated with an 18-point increase in reading performancefor non-Indigenous students, and a 15-point increase for Indigenous New Zealanders. Alsosignificant was the home language variable, where those who spoke English at home had60- and 37-point advantages over their counterparts, with Indigenous students showingthe larger gap.

For the school-level variables examined in this study, the number of variables associatedwith student performance varied widely between student groups. For Indigenous Austra-lians, only school size had a statistically meaningful relationship with reading performance;Indigenous students attending schools with larger enrolments had more positive outcomes.On the other hand, for non-Indigenous Australian students, all but two school-level vari-ables (i.e., reported shortage of computers and proportion of certified teachers) werefound to be significantly associated with reading performance. Specifically, non-IndigenousAustralian students who attended schools in more populous communities were enrolled inlarger schools, and reported more abundant resources in terms of instructional materials andqualified teachers had higher reading scores. These relationships were also more or lessreflected for Australian Indigenous students but fell short of statistical significance, possiblydue to the comparatively small size of the Indigenous sample.

For New Zealand, few of the school variables were found to be meaningfully related toreading achievement for both Indigenous and non-Indigenous student groups. As shown inTable 4, among Indigenous students, only shortages of qualified teachers and the student–teacher ratio produced statistically significant relationships with reading performance.These regression results suggest that Māori students who attended schools with more qua-lified teachers and higher student–teacher ratios tended to outperform their peers attendingschools with greater shortages of qualified teachers and smaller student–teacher ratios. Fornon-Indigenous New Zealand students, those attending larger schools and schools withfewer shortages of instructional materials had better reading performances.

Considered together, student-level and school-level variables explained between 22%and 29% of the variability in reading performance for Indigenous and non-Indigenous stu-dents in Australia and New Zealand. In the following section, we examine how these

12 S. Song et al.

500

505

510

515

520

525

530

535

540

variables differed in their allocation and “utilization” among these four student groups,organized by country and indigenous status.

Decomposition analysis

Using the Oaxaca-Blinder decomposition, this section examines differences in the pro-portional roles played by each of the student and school variables (i.e., allocation ofresources) as well as their coefficients (i.e., rates of return) for Indigenous and non-Indigen-ous students in Australia and New Zealand.

As can be seen in Table 5, student-level variables explained 34.7% of the achievementgap in reading between Indigenous and non-Indigenous students in Australia. Put another

Table 4. Student and school predictors for Indigenous and non-Indigenous New Zealand students inPISA 2009 reading.

Indigenous (n = 697) Non-Indigenous (n = 3.201)

Unstandardized Standardized SE Unstandardized Standardized SE

Student VariablesHISCED 2.90 0.05 2.90 4.71*** 0.07 1.32HISEI 1.23*** 0.21 0.25 1.22*** 0.20 0.12Gender −42.18*** −0.23 7.13 −38.91*** −0.21 3.04Language athome

−60.52*** −0.19 9.63 −37.08*** −0.14 4.84

Number ofcomputers athome

8.50* 0.08 4.37 4.83*** 0.05 1.86

Number ofbooks at home

15.49*** 0.22 2.84 18.48*** 0.27 1.25

School VariablesSchoolcommunity

−6.79 −0.09 4.00 −1.73 −0.02 1.63

School size 0.01 0.07 0.01 0.01** 0.07 0.01Shortage ofinstructionalmaterials

−3.92 −0.03 5.28 −11.10*** −0.08 2.54

Shortage ofcomputers

−0.93 −0.01 4.15 0.82 0.01 1.66

Shortage ofqualifiedteachers

−8.64* −0.08 4.23 −1.20 −0.01 1.88

Proportion ofqualifiedteachers

9.13 −0.01 38.99 −24.59 −0.04 11.49

Proportion ofcertifiedteachers

20.23 0.02 54.02 36.49 0.04 20.57

Student–teacherratio

4.08* 0.11 1.97 0.55 0.01 0.86

Adjusted R-square

0.27 0.29



545

550

555

560

565

570

575

580

585

way, if Australian Indigenous students were to have the equivalent advantages of student-related variables (e.g., number of books at home) as compared to their non-Indigenouscounterparts, their reading achievement as measured by PISA 2009 would benefit byabout 27.5 points. Similarly, school-related variables, considered separately fromstudent-related variables, explain about 18.5% of the gap between the two groups. Inother words, if Australian Indigenous students had the affordances of school-related vari-ables (e.g., proportions of qualified teachers) equivalent to their non-Indigenous counter-parts, their reading achievement as measured by PISA 2009 would benefit by 14.7points. Considered together, student- and school-level variables explained almost 44%(34.8 points) of the reading achievement gap between Indigenous and non-Indigenous stu-dents in Australia.

Specifically, as detailed in Table 6, when assessing differences in reading performancebetween Indigenous and non-Indigenous students in Australia in terms of their endowment(i.e., allocation of resources), parental occupation and education level, gender, numbers ofcomputers and books at home, size of the community in which the school is located, andshortages of qualified teachers were found to be statistically significant. These resultssuggest that if Indigenous students in Australia were afforded the equivalent value ofthese variables as enjoyed by their non-Indigenous peers, the benefit accrued wouldreduce the gap in reading achievement between Indigenous and non-Indigenous studentsby almost 44%. For the “rate of return” (unexplained) part of the decomposition analysis,parental education and occupational status, language spoken at home, and the size of thecommunity in which schools are located resulted in statistically significant associationswith the observed gap between Indigenous and non-Indigenous Australian students, favour-ing the latter group. For these and other student-level variables, the “rates of return” analysisshowed a 29-point disadvantage for Indigenous students, accounting for about 37% of thegap in reading achievement between Australian Indigenous and non-Indigenous students.

These findings suggest that in Australia, unequal allocation of resources and the ways inwhich resources or affordances translate to educational outcomes together account for about80% of the achievement gap in reading between Indigenous and non-Indigenous students.Additionally, it appears that the role of variables that are theoretically amenable to beingtranslated into more positive educational outcomes for students (i.e., rates of return associ-ated with parental education or occupational status) play a somewhat less substantial role(29 points) than the actual unequal allocation of resources (35 points) in explaining thegap in reading achievement between Indigenous and non-Indigenous students.

For New Zealand, the unequal allocation of resources in the student-level variablesexplained 35% of the achievement gap in reading between Indigenous and non-Indigenousstudents. Put another way, if the Indigenous students in New Zealand were to have theequivalent advantages of student-level resources (e.g., number of books at home) as

Table 5. Decomposed reading score differentials between Indigenous and non-Indigenous studentsin Australia and New Zealand, PISA 2009.

Home School Home & School

Explained Unexplained Explained Unexplained Explained Unexplained

Australia 27.49*** 34.49*** 14.67*** 49.19*** 34.79*** 29.08***New Zealand 19.32*** 32.90*** 4.11* 48.89*** 20.29*** 32.71***

Note: Calculations were performed using weighted PISA 2009 data.*p ≤ .05. **p ≤ .01. ***p ≤ .001.

14 S. Song et al.

590

595

600

605

610

615

620

625

630

compared to their non-Indigenous counterparts, their reading achievement as measured byPISA 2009 would benefit by about 19 points. On the other hand, school-related variables,considered separately from student-level variables, explain only about 7.5% (4 points) ofthe gap between the two groups. Considered together, home- and school-related variablesexplained just over 37% (20 points) of the reading achievement gap between Indigenousand non-Indigenous students in New Zealand.

As detailed in Table 6, for students in New Zealand, endowment differences related toparental occupation and education levels, language spoken at home, and the numbers ofcomputers and books at home were found to be statistically significant. For the “rate ofreturn” aspect of the decomposition analysis, only language spoken at home and shortagesof qualified teachers provided statistically significant contributions in accounting for thegap between New Zealand’s Indigenous and non-Indigenous students.

As a whole, the decomposition analysis suggests that for New Zealand, differences instudent resources explain substantially more (about 35%) of the gap in reading performancebetween Indigenous and non-Indigenous students than differences in school resources

Table 6. Decomposition of student and school variables related to the gap in reading achievementbetween Indigenous and non-Indigenous students in Australia and New Zealand, PISA 2009.

Explained Coefficients Z Coefficients Z

HISCED 4.31*** 6.08 2.09*** 3.10HISEI 6.86*** 7.82 7.06*** 5.84Gender −1.58* −2.09 1.08 0.92Language at home −0.30 −0.87 −2.09*** −2.70Number of computers at home 4.87*** 5.44 1.41** 2.72Number of books at home 11.05*** 9.13 7.75*** 6.57School community 5.74*** 3.25 −0.98 −0.92School size 1.41 1.24 2.19 1.67Shortage of instructional materials 1.00 1.15 0.57 1.12Shortage of computers −0.71 −0.75 −0.02 −0.17Shortage of qualified teachers 2.23* 2.17 0.55 1.10Proportion of qualified teachers 0.22 0.74 −0.10 −0.34Proportion of certified teachers 0.13 1.33 −0.03 −0.12Student–teacher ratio −0.46 −1.01 0.80 0.96Total 34.79*** 11.53 20.29*** 6.93Unexplained Coefficients Z Coefficients ZHISCED 46.38*** 3.58 7.77 0.63HISEI 3.38*** 0.29 −0.49 −0.04Gender 10.20 1.06 5.02 0.42Language at home 80.77*** 4.81 25.92* 2.18Number of computers at home −2.05 −0.16 −9.98 −0.75Number of books at home 4.01 0.54 10.05 1.03School community 21.58* 2.10 17.04 1.40School size −14.88 −1.53 −0.87 −0.09Shortage of instructional materials 1.29 0.15 −11.59 −1.58Shortage of computers 9.47 0.82 3.96 0.53Shortage of qualified teachers −8.03 −0.92 15.61* 2.07Proportion of qualified teachers −8.31 −0.46 −14.05 −0.69Proportion of certified teachers 17.39 0.46 15.62 0.78Student–teacher ratio −19.32 −0.70 −50.91 −1.82Total 29.08*** 8.00 32.71*** 9.67



635

640

645

650

655

660

665

670

675

20060040Sticky NoteThis Table is incorrectly reproduced from the manuscript. There needs to be an additional heading row that indicates that the first two (from left) columns of numbers are for Australia, and the next two are for New Zealand (see the layout for Table 6 in our manuscript).

20060040Sticky NoteTable layout is incorrect. This row should be in heading format, with horizontal line above and below (as for the first heading row). The text should not be bold. Please see the layout/format for Table 6 in the submitted manuscript.

(7.5%). This is in contrast to Australia, where the reading achievement gap between Indi-genous and non-Indigenous students appears to be explained in relatively equal portions bydifferences in home-based resources and differences in resources in the schools (Table 5).Furthermore, differences in the explained and unexplained portions of the decompositionanalysis indicated that for Australia, unequal allocation of resources plays a more substan-tial role than rates of return (i.e., the translation of affordances into educational benefit forstudents) in accounting for the gap in reading performance between Indigenous and non-Indigenous students. This was different in New Zealand, where the translation of affor-dances into educational benefit plays the more substantial role in accounting for the gapin reading achievement. For New Zealand, the ratio of explained to unexplained roles forstudent and school variables combined was about 2:3, whereas for Australia it was approxi-mately 1:1.

Discussion

Three important questions are posed in this study. First, how do student and school factorsdiffer between Indigenous and non-Indigenous students in Australia and New Zealand?Second, what factors explain the substantial difference between Australia and NewZealand in the size of the reading achievement gap between Indigenous and non-Indigenousstudents? Third, to what extent are student- and school-level resources differently convertedinto positive educational outcomes by Indigenous and non-Indigenous students in Australiaand New Zealand? Our analyses show that the answers to these questions relate to unequallevels or allocations of student and/or school resources between Indigenous and non-Indi-genous students. As well, our findings suggest different rates among the four groups for theconversion of resources into reading achievement. In this section, we interpret these find-ings and discuss potential implications for education policy and future lines of research.

In both countries, we observed that a large portion of the reading achievement gapbetween Indigenous and non-Indigenous students is associated with substantial inequitiesin student resources. On average, parents of Indigenous students have lower levels of edu-cation, lower occupational status, and fewer books and computers at home compared totheir non-Indigenous peers. In New Zealand, the regression and decomposition analysesshow that the reading achievement gap is associated almost entirely with student-leveldifferences in resources, favouring non-Indigenous students. In Australia, however, theregression and decomposition analyses showed that school variables played a muchlarger role in explaining the achievement gap between Indigenous and non-Indigenous stu-dents than in New Zealand. This is not surprising since our analyses also showed that schoolresources are allocated less equitably in Australia than in New Zealand.

These differences between the two countries are related to the size of the gap betweenIndigenous and non-Indigenous achievement. The achievement gap in reading betweenIndigenous and non-Indigenous students is large in both Australia and New Zealand.Our findings show, however, that the gap is substantially larger in Australia than in NewZealand (79 vs. 55 score points, respectively). Song (2011) reported similar results in hisstudy of the educational achievement gap between Turkish and native students inAustria, Germany, and Switzerland. Two studies, therefore, show that achievement gapsare smaller between socially marginalized ethnic minority groups and ethnic majoritygroups when resources are distributed more equally among schools.

Our analysis is not able to explain why school resources are allocated more equitably inNew Zealand than in Australia, but we can offer some possibilities. Socioeconomic segre-gation between schools is higher in Australia than in New Zealand (OECD, 2010). This

16 S. Song et al.

680

685

690

695

700

705

710

715

720

high level of segregation in Australia is associated with large differences in school resources,especially the ability to recruit and retain qualified and experienced teachers. Perry andMcConney (in press) have shown, for example, that principals of schools that enrol primarilystudents from low-socioeconomic backgrounds report that learning in their school is hin-dered by teacher shortages to amuch greater extent than principals of other schools. And Indi-genous students in Australia are concentrated in these low-socioeconomic schools (Rorriset al., 2011; Teese, 2011). Low-socioeconomic schools in Australia are often smallschools, even when located in large cities (Lamb, 2007; Perry & Southwell, 2013), and asour findings show, school size is related to academic performance for Indigenous studentsin Australia. Research from the US, which has a similar level of school segregation as Aus-tralia, has also shown that low-socioeconomic schools have more difficulty recruiting andretaining qualified and experienced teachers (see, e.g., Darling-Hammond, 2010; Guarino,Brown, & Wyse, 2011; Muijs, Harris, Chapman, Stoll, & Russ, 2009).

For educational policymakers, these findings suggest that school resources need to beallocated more equitably in Australia. This suggestion supports the work of Indigenousleaders in Australia who have called repeatedly for better schooling in Indigenous commu-nities (Hind, 2012). Further, as described earlier in this paper, the per-pupil funding ofschools is much more equitable in New Zealand than in Australia. Almost all schools inNew Zealand (96%) receive the same per-pupil funding from public authorities, and donot charge tuition fees. As noted by a recent report commissioned by the federal govern-ment, however, per-pupil funding varies widely across schools in Australia (AustralianGovernment, 2011). This is largely due to federal funding mechanisms that provide sub-stantial subsidies to private schools while also allowing them to charge tuition fees. Ascountless studies have shown, the inequitable allocation of resources usually privilegessocially advantaged groups (Berliner, 2001; Chiu & Khoo, 2005; Tate, 1997; Teese,2007). It is also likely that structurally inequitable funding in Australia is a driver of thehigh levels of social segregation noted above. Even if a low-socioeconomic school iswell resourced (and many of them are), it will remain a “hard to staff” school and likelyexperience all of the challenges that accompany that designation (Kahlenberg, 2001).

Our decomposition analysis also suggests that, overall, student-related resources aremore effectively translated into more equitable outcomes for Indigenous students in Aus-tralia than in New Zealand. This difference is not likely due to urban-rural residential pat-terns. Most Indigenous students in both countries live in urban communities or towns.Sixteen percent of Indigenous people in New Zealand live in communities with fewerthan 1,000 residents (Ministry of Social Development, 2010). Comparable census datafrom Australia show that 26% of Indigenous people live in “remote” and “very remote”communities (Australian Bureau of Statistics, 2077). This is a larger number than inNew Zealand, but “remote” also includes large but geographically isolated towns, suchas Alice Springs with a population of 24,000 people. Many of these towns have large popu-lations of Indigenous people. It is therefore likely that the number of Indigenous studentswho live in communities with less than 1,000 residents is similar in both countries. Thereasons behind this finding are probably multiple and could be related to differences ineconomic opportunities available to Indigenous people in both countries.

Additionally, the “rates of return” (unexplained portions of the decomposition analysis)associated with parental education and occupational status are stronger for Australian non-Indigenous compared to Indigenous students, with regard to reading achievement.Although the observation of this phenomenon is difficult to explain, and certainly requiresfurther examination, some insight may be found in the work of Coleman (1988), who


725

730

735

740

745

750

755

760

765

argued that social capital may be necessary for human capital to flow through in detectablefashion from parents to children.

Thus, although much more needs to be unpacked, it appears that Indigenous families inAustralia and more so in New Zealand are constrained in their ability to maximize poten-tially positive educational benefits of socioeconomic and occupational status. Furtherresearch about the cultural, linguistic, social, and economic dynamics that influence therelationship between socioeconomic status and educational outcomes for Indigenous stu-dents is warranted.

The limitations of our study should be noted. First, we are unable to establish causalrelationships among predictor and outcome variables since we are using a cross-sectionaldataset. We have described associations but are unable to say categorically that particularfactors cause particular outcomes. Second, the PISA dataset does not measure teacherexperience or teacher effectiveness. It is likely that we would be able to explain a largerpart of the Indigenous achievement gap within and between the two countries if we wereable to measure this feature of teacher quality.

Achievement gaps between ethnic minority and majority students are common through-out theworld, in large part due to social forces that are outside the control of education policy-makers. At the same time, however, education policies can either exacerbate or amelioratethese achievement gaps. This analysis has shown that the achievement gap in readingbetween Indigenous and non-Indigenous students is 1.5 times larger in Australia than inNew Zealand, and that this is accompanied by greater inequities in the allocation of schoolresources, especially shortages of teachers, in Australia. We therefore argue that educationpolicymakers in Australia should work to ensure a more equitable allocation of schoolresources between Indigenous and non-Indigenous students. Researchers and practitionersshould also examine how Indigenous families can be better supported in translating theirfamilial, cultural and social resources into strong educational outcomes for their children.

Notes1. Currency exchange rates as at December 5, 2013.2. The standard errors for the regression and Oaxaca-Blinder decomposition test were clustered at

the school level.3. For a more detailed and technical explanation of the Oaxaca-Blinder decomposition, refer to the

following articles: Blinder (1973), Neumark (1988), and Oaxaca (1973).4. With the exception of the proportion of certified teachers in New Zealand and proportion of qua-

lified teachers in Australia, all mean and proportional differences between Indigenous and non-Indigenous students in Australia and New Zealand differed significantly at p ≤ .05.

Notes on contributorsDr Steve Song is Assistant Professor of research methods and comparative/international education atGeorge Fox University in the US. His research interests have centred upon immigration, ethnicity,academic achievement, and other areas of the sociology of education.

Dr Laura B. Perry is Senior Lecturer of education policy, comparative education, and contexts of edu-cation at Murdoch University in Australia. Her research focuses on the structural and systemic factorsthat shape inequalities of educational opportunities, experiences, and outcomes.

Dr AndrewMcConney is Associate Professor of research and evaluation in the School of Education atMurdoch University. Andrew’s research interests include the secondary analysis of large-scale data-sets to inform educational policy and practice, and the evaluation of science, maths, and environ-mental education programmes.

18 S. Song et al.

770

775

780

785

790

795

800

805

810

20053485Sticky NotePlease add an "Acknowledgements" section with this text:"This study was funded in part by an Australian Research Council Discovery Project grant (DP1097057) awarded to Laura Perry."

20060040Sticky NoteSteve is Adjunct Professor, not Assistant Professor.

ReferencesAppleton, S., Hoddinott, J., & Krishnan, P. (1999). The gender wage gap in three African countries.

Economic Development and Cultural Change, 47, 289–312.Australian Bureau of Statistics. (2006). Australian social trends, 2006. Canberra, Australia: Author.Australian Bureau of Statistics. (2007). 4705.0 – Population distribution, Aboriginal and Torres Strait

Islander Australians, 2006. Canberra, Australia: Author.Australian Bureau of Statistics. (2009). 4714.0 –National Aboriginal and Torres Strait Islander social

survey. Canberra, Australia: Author.Australian Bureau of Statistics. (2012). The health and welfare of Australia’s Aboriginal and Torres

Strait Islander peoples, Oct 2010. Latest ISSUE Released at 11:30 AM (CANBERRATIME) 19/12/2012. Retrieved from http://www.abs.gov.au/AUSSTATS/[email protected]/lookup/4704.0Main+Features1Oct+2010

Australian Bureau of Statistics. (2013). 6302.0 – Average weekly earnings, Australia, May 2013.Retrieved from http://www.abs.gov.au/ausstats/[email protected]/Products/6302.0∼May+2013∼Main+Features∼Key+Figures?OpenDocument

Australian Government. (2011). Review of funding for schooling – final report. Canberra, Australia:Department of Education, Employment and Workplace Relations.

Berliner, D. (2001). Our schools vs theirs: Averages that hide the true extremes (No. CERAI-01-02).Milwaukee, WI: Center for Education Research, Analysis, and Innovation.

Blinder, A. S. (1973). Wage discrimination: Reduced form and structural estimates. The Journal ofHuman Resources, 8, 436–455.

Chiu, M. M., & Khoo, L. (2005). Effects of resources, inequality, and privilege bias on achievement:Country, school, and student level analyses. American Educational Research Journal, 42,575–603.

Coleman, J. (1988). Social capital in the creation of human capital. American Journal of Sociology,94, S95–S120.

Darling-Hammond, L. (2010). The flat world and education: How America’s commitment to equitywill determine our future. New York, NY: Teachers College Press.

Fleischman, H. L., Hopstock, P. J., Pelczar, M. P., & Shelley, B. E. (2010). Highlights from PISA2009: Performance of U.S. 15-year old students in reading, mathematics, and science literacyin an international context (NCES 2011-004). Washington, DC: U.S. Department ofEducation, National Center for Education Statistics.

Freire, P. (1992). Pedagogy of the oppressed (M. B. Ramos, Trans.). New York, NY: ContinuumPublishing Company.

Gimenez Adelantado, A. (2003). The education of gypsy childhood in Europe. Brussels, Belgium:European Commission.

Guarino, C. M., Brown, A. B., & Wyse, A. E. (2011). Can districts keep good teachers in the schoolsthat need them most? Economics of Education Review, 30, 962–979.

Hind, R. (2012). Anderson calls for schools to be made equal. Retrieved from http://www.abc.net.au/news/2012-10-24/anderson-on-bilingual-school-education-indigenous-schools/4331158

Jalan, J., & Ravallion, M. (2003). Does piped water reduce diarrhea for children in rural India?Journal of Econometrics, 112, 153–173.

Kahlenberg, R. (2001). All together now: Creating middle-class schools through public schoolchoices. Washington, DC: Brookings Institution.

Knighton, T., Brochu, P., & Gluszynski, T. (2010). Measuring up: Canadian results of the OECDPISA study. Ottawa, Canada: Statistics Canada.

Lamb, S. (2007). School reform and inequality in urban Australia: A case of residualising the poor. InR. Teese, S. Lamb, & M. Duru-Belat (Eds.), International studies in educational inequality,theory and policy (Vol. 3, pp. 1–38). Dordrecht, he Netherlands: Springer.

Lounkaew, K. (2013). Explaining urban–rural differences in educational achievement in Thailand:Evidence from PISA literacy data. Economics of Education Review, 37, 213–225.

Losen, D., & Orfield, G. (2002). Racial inequality in special education. Cambridge, MA: HarvardEducation Press.

McMahon, W. W. (2002). Education and development: Measuring the social benefits. Oxford, UK:Oxford University Press.

McRae, D., Ainsworth, G., Cumming, J., Hughes, P., Mackay, T., Price, K., . . . Zbar, V. (2000).Whatworks? Explorations in improving outcomes for Indigenous students. Canberra, Australia:Australian Curriculum Studies Association and National Curriculum Services.


815

820

825

830

835

840

845

850

855

http://www.abs.gov.au/AUSSTATS/[email protected]/lookup/4704.0Main+Features1Oct+2010http://www.abs.gov.au/AUSSTATS/[email protected]/lookup/4704.0Main+Features1Oct+2010http://www.abs.gov.au/ausstats/[email protected]/Products/6302.0~May+2013~Main+Features~Key+Figures?OpenDocumenthttp://www.abs.gov.au/ausstats/[email protected]/Products/6302.0~May+2013~Main+Features~Key+Figures?OpenDocumenthttp://www.abc.net.au/news/2012-10-24/anderson-on-bilingual-school-education-indigenous-schools/4331158http://www.abc.net.au/news/2012-10-24/anderson-on-bilingual-school-education-indigenous-schools/4331158

Meyers, L. S., Gamst, G., & Guarino, A. J. (2006). Applied multivariate research: Design andinterpretation. Thousand Oaks, CA: Sage.

Ministry of Social Development. (2010). The social report: Distribution of the population. Retrievedfrom http://www.socialreport.msd.govt.nz/people/distribution-population.html

Muijs, D., Harris, A., Chapman, C., Stoll, L., & Russ, J. (2009). Improving schools in socioeconomi-cally disadvantaged areas – a review of research evidence. School Effectiveness and SchoolImprovement, 15, 149–175.

National Center for Education Statistics. (2002). Status and trends in the education of Hispanics.Washington, DC: Author.

Neumark, D. (1988). Employers’ discriminatory behavior and the estimation of wage discrimination.The Journal of Human Resources, 23, 279–295.

New Zealand Ministry of Education. (2012a). Circular 2012/07 – private school subsidy funding2013. Retrieved from http://www.minedu.govt.nz/NZEducation/EducationPolicies/Schools/PublicationsAndResources/Circulars/Circulars2012/Circular201207.aspx

New Zealand Ministry of Education. (2012b). Education counts: Statistics. Retrieved from http://www.educationcounts.govt.nz/statistics/schooling/july_school_roll_returns/6028

Oaxaca, R. (1973). Male-female wage differentials in urban labor markets. International EconomicsReview, 14, 693–709.

Open Society Institute. (2007). Equal access to quality education for Roma: Vol. 1: Bulgaria,Hungary, Romania, Serbia. Monitoring report. Budapest, Hungary: Author.

Organisation for Economic Co-operation and Development. (2000). PISA 2000 technical report.Paris, France: Author.

Organisation for Economic Co-operation and Development. (2001). Knowledge and skills for life:First results from the OECD Programme for International Students Assessment (PISA) 2000.Paris, France: Author.

Organisation for Economic Co-operation and Development. (2010). PISA 2009 results: What studentsknow and can do – student performance in reading, mathematics and science. Paris, France:Author.

Oxfam. (2013). Indigenous Australia. Retrieved from https://www.oxfam.org.au/about-us/countries-where-we-work/australia/

Perry, L. B., & McConney, A. (in press). Mapping educational inequalities by school socioeconomiccomposition in Australia. In J. G. Ladwig & J. Albright (Eds.), On the within school stratificationof social inequities of learning and achievement. Rotterdam, he Netherlands: Sense Publishers.

Perry, L. B., & Southwell, L. (2013). Access to academic curriculum in Australian secondary schools:A case study of a highly marketised education system. Journal of Education Policy. Advanceonline publication. doi:10.1080/02680939.2013.846414

Rangvid, B. S. (2007). Living and learning separately? Ethnic segregation of school children inCopenhagen. Urban Studies, 44, 1329–1354.

Reimers, F. (2003). War, education and peace. Prospects, 33, 3–6.Rorris, A., Weldon, P., Beavis, A., McKenzie, P., Bramich, M., & Deery, A. (2011). Assessment of

current process for targeting of schools funding to disadvantaged students. Camberwell,Australia: Australian Council for Educational Research.

Simon, P. (2003). France and the unknown second generation: Preliminary results on social mobility.International Migration Review, 37, 1091–1119.

Song, S. (2011). Second-generation Turkish youth in Europe: Explaining the academic disadvantagein Austria, Germany, and Switzerland. Economics of Education Review, 30, 938–949.

Statistics New Zealand. (2013). New Zealand income survey: June 2012 quarter. Retrieved fromhttp://www.stats.govt.nz/browse_for_stats/income-and-work/Income/NZIncomeSurvey_HOTPJun12qtr.aspx

Sullivan, K., Perry, L. B., & McConney, A. (2013). How do school resources and academic perform-ance differ across Australia’s rural, regional and metropolitan communities? AustralianEducational Researcher, 40, 353–372.

Tate, W. F. (1997). Race-ethnicity, SES, gender, and language proficiency trends in mathematicsachievement: An update. Journal for Research in Mathematics Education, 28, 652–679.

Teese, R. (2007). Stuctural inequality in Australian education. In R. Teese, S. Lamb, &M. Duru-Bellat(Eds.), International studies in educational inequality, theory and policy (Vol. 2, pp. 39–62).Dordrecht, he Netherlands: Springer.

20 S. Song et al.

860

865

870

875

880

885

890

895

900

http://www.socialreport.msd.govt.nz/people/distribution-population.htmlhttp://www.minedu.govt.nz/NZEducation/EducationPolicies/Schools/PublicationsAndResources/Circulars/Circulars2012/Circular201207.aspxhttp://www.minedu.govt.nz/NZEducation/EducationPolicies/Schools/PublicationsAndResources/Circulars/Circulars2012/Circular201207.aspxhttp://www.educationcounts.govt.nz/statistics/schooling/july_school_roll_returns/6028http://www.educationcounts.govt.nz/statistics/schooling/july_school_roll_returns/6028https://www.oxfam.org.au/about-us/countries-where-we-work/australia/https://www.oxfam.org.au/about-us/countries-where-we-work/australia/http://www.stats.govt.nz/browse_for_stats/income-and-work/Income/NZIncomeSurvey_HOTPJun12qtr.aspxhttp://www.stats.govt.nz/browse_for_stats/income-and-work/Income/NZIncomeSurvey_HOTPJun12qtr.aspx20053485Highlight

20053485Sticky Noteremove "HE" in Perry and Teese refs.

20053485Highlight

Teese, R. (2011). From opportunity to outcomes. The changing role of public schooling in Australiaand national funding arrangements. Melbourne, Australia: Centre for Research on EducationSystems, University of Melbourne.

Telford, M. (2010). PISA2009: Our 21st century learners at age 15. Wellington, New Zealand:Comparative Education Research Unit, Research Division, Ministry of Education.

Thompson, S., De Bortoli, L., Nicholas, M., Hillman, K., & Buckley, S. (2010). PISA in brief, high-lights from the full Australian report: Challenges for Australian Education: Results from PISA2009. Melbourne, Australia: Australian Council for Educational Research.

Van Ewijk, R. (2011). Same work, lower grade? Student ethnicity and teachers’ subjective assess-ments. Economics of Education Review, 30, 1045–1058.

Wagstaff, A., & Nguyen, N. (2004). Poverty and survival prospects of Vietnamese children under DoiMoi. In P. Glewwe, N. Agrawal, & D. Dollar (Eds.), Economic growth, poverty and householdwelfare in Vietnam (pp. 313–350). Washington, DC: World Bank.

Woods-McConney, A., Oliver, M., McConney, A., Maor, D., & Schibeci, R. (2013). Science engage-ment and literacy: A retrospective analysis for Indigenous and non-Indigenous students inAotearoa New Zealand and Australia. Research in Science Education, 43, 233–252.

Worbs, S. (2003). The second generation in Germany: Between school and labor market.International Migration Review, 37, 1011–1038.


905

910

915

920

925

930

935

940

945

AbstractIntroductionPerformance of Indigenous students in Australia and New ZealandEducation context of Australia and New ZealandMethodDataData analysisVariablesStudent-level variables

ResultsDescriptive analysisRegression analysisDecomposition analysis

DiscussionNotesNotes on contributorsReferences

murdoch research repository · study by song (2011), which examined substantial disparities in...

Documents