How Streaming (Tracking) in Eighth Grade Mathematics Reinforces Racialized Social Class Inequalities in Aotearoa New Zealand

ABSTRACT

In Aotearoa New Zealand stark social class inequities persist between Māori (Indigenous) and Pacific people and the Pākehā (New Zealand European) majority. These inequities are apparent in domains including education, income, health, and incarceration. The article explores the relationship between streaming (tracking) and historically rooted ethnic inequalities in one diverse urban setting. Drawing survey, assessment, and administrative data from 450 eighth-grade students across three multicultural secondary schools, we ask how school mathematics reinforces or disrupts social-class divisions between majority Pākehā and minoritized Māori and Pacific students. Students entering secondary school imagined their future careers in ways that were already strongly differentiated by race, class, and gender. Tracking students into racially stratified mathematics classes reinforced such inequalities through a self-reinforcing interaction between aspirations and mathematics achievement.

In Aotearoa¹ New Zealand, stark social class inequities persist between Māori (Indigenous) and Pacific people on the one hand, and the Pākehā (New Zealand European) majority on the other. These inequities are apparent in the domains of education, income, health, incarceration, and others. In this article, we examine the extent to which streaming (tracking) in eighth-grade mathematics reinforces or disrupts social-class divisions between majority Pākehā (New Zealand European) and minoritized Māori (Indigenous) and Pacific² students, in Aotearoa New Zealand. In doing so, we empirically investigate two questions:

1. How are social class, mathematics achievement, and stream (track) allocation racialized on entry into secondary school?

2. To what extent does mathematics achievement mediate the relationship between parental social class and student job aspirations?

Regarding the first question, Pākehā students, compared to Māori and Pacific students, come from more-privileged social-class backgrounds, have higher mathematics achievement, and are allocated into higher streams. Regarding the second question, mathematics achievement has a small mediating effect on the relationship between parental social class and student job aspirations—students with higher mathematics achievement are more likely to aspire to upward occupational mobility. We discuss these empirical results in the context of previous studies that show that streaming exacerbates mathematics achievement inequalities (Hodgen et al., 2023; Linchevski & Kutscher, 1998; Oakes, 2005; Wiliam & Bartholomew, 2004).

Reflecting the language most prevalent in the study setting, we use the term streaming to describe the allocation of students into stable class groupings with relatively homogeneous prior attainment, a practice also known as tracking, setting, or attainment grouping. In Aotearoa New Zealand (henceforth Aotearoa), as in many other nations, racialized privilege and marginalization has a complex overlap with socioeconomic or social-class privilege and marginalization. Pākehā settlers and their descendants, on average, experience better outcomes that Māori and Pacific New Zealanders across a wide range of domains including income, health, and educational achievement (Rashbrooke, 2014).

Scholars in the field of sociology of education have argued for quite some time that schools cannot remain neutral in terms of the intergenerational reproduction of class inequalities; schools can perpetuate and/or challenge the status quo (Bourdieu & Passeron, 1990; Oakes, 2005; Reay, 2012). In this article we examine the phenomenon of tracking through a contained case study: eighth-grade (Year Nine, age 13–14) mathematics in Aotearoa. The empirical section highlights (1) racialized class inequalities on entry into secondary school and (2) the racial stratification of mathematics classes via streaming in the first year of secondary school. This analysis supports the argument that streaming largely reinforced rather than challenged racialized class inequalities.

Historical context of race and social class

We begin by providing a necessarily simplified overview of the historical processes that contributed to racialized inequalities in this study’s setting, Twin Tides City (a pseudonym). Recognizing that colonization, settlement, and immigration have regional specificities and that land and landscape shape wealth and poverty, we focus on Twin Tides City as a specific case rather than attempting to summarize the development of racialized class inequalities throughout Aotearoa. This context is necessary to our argument because if streaming were represented simply as a result of variation in current academic achievement, we neglect the historical injustices and intergenerational processes that influence academic achievement.

Twin Tides City borders a turbulent waterway that divides it from another significant portion of land within the wider region. When describing the geological structure and strategic significance of that region to its Indigenous inhabitants prior to British colonization, Davidson and Leach (2002) explained:

The significance of [that maritime region] in pre-European times as a “bridge”, rather than a “barrier” is clear not only from Māori oral narratives but from archaeological evidence of the transport in both directions of valuable resources such as greenstone … argillite, and obsidian (Leach, 1978). Control of traffic across the [turbulent waterway] would always be important … The shores of the [turbulent waterway] are “characterized by considerable diversity of structure and sediments” (Harris, 1990, pp. 25–47) … the shore of the inner [waterway] is dominated by a rocky coast of greywacke or associate rock … A narrow and, in places, almost non-existent coastal platform is broken by [the twin harbours: from which the City has gained its name].

(pp. 257–259)

Until the early 19th century, the Twin Tides area likely had a small and transient population until the arrival of a Northern iwi (tribe), who, armed with muskets, drove other iwi away and established control (Ballara, 2003; Boast, 2007; Mawer, 2022; Parliamentary Council Office, 2014). As such, today only that iwi are legally recognized as mana whenua, holding customary territorial rights to the area (New Zealand Government, 2022; Parliamentary Council Office, 2014). By the 1840s, this iwi had established 12 pā, or settlements, in the area, with only two of these settlements still occupied by them today (Parliamentary Council Office, 2014). British whalers were the first newcomers drawn to the Twin Tides area. Some established permanent shore stations and intermarried with the local iwi women (Ballara, 2003; Boast, 2007; Parliamentary Council Office, 2014).

Meanwhile, Bishop Hadfield noted in his memoirs that a Māori boys’ school was already operating at one of the pā sites as early as 1825 (Barrington & Beaglehole, 1974; Gibson, 2022). This school, like other mission schools, was most likely committed to “civilizing” these boys and preparing them for the role of cheap agricultural or maritime labor in a future British colony (Barrington & Beaglehole, 1974; Simon, 1998). In the late 1830s, a British land-purchasing agent sought to purchase the Twin Tides area due to its strategic position (Boast, 2007). His plans for colonization were critiqued by Karl Marx in the first volume of his classic text Capital: A Critical Analysis of Capitalist Production (Hastings, 1990; Marx, 1957). Marx claimed that the plans would leave Māori a landless proletariat. By 1839, this wealthy land agent and his company alleged that the local iwi had agreed to sell the Twin Tides area, resulting in a stand-off between the iwi and the newly arrived British settlers who believed they had purchased the lands legitimately from that company. However, following the signing of Te Tiriti o Waitangi (Treaty of Waitangi) in 1840, a Crown commission of inquiry found that most of this company’s land purchase was invalid (Boast, 2007; New Zealand Government, 2022; Parliamentary Council Office, 2014).

The Crown commission’s findings were not well received by the settlers or the influential owners of that company. The two most influential Rangatira (chief/s) of the area were increasingly portrayed by the settlers’ newspapers as a threat to the company’s notions of progress (Boast, 2007, New Zealand Government, 2012; Parliamentary Council Office, 2014). The arrival of an ambitious British governor general, determined to expand British sovereignty and his own reputation, further fueled tensions (Boast, 2007; Parliamentary Council Office, 2014). By 1846, the new governor had orchestrated the abduction of the area’s leading Rangatira (King, 2003; New Zealand Government, 2022) and military skirmishes left the second most powerful Rangatira displaced (Boast, 2007, Parliamentary Council Office, 2014).

By April 1847, the eight remaining Rangatira were effectively coerced into signing a deed of sale of 69,000 acres for a paltry fee, and only three reserves (totalling 10,000 acres) were set aside for the mana whenua tribe (Boast, 2007; New Zealand Government, 2022). Meanwhile, when the British army built its barracks in 1846 (next to an iwi settlement and trading station), a school was provided for children of the regiment. This reflected a pattern of early schooling being often associated with the growth of British military installations elsewhere (Barrington & Beaglehole, 1974; Manning, 2009). A small British settler community grew and, by 1926, most of the mana whenua iwi reserves had been alienated. Most notably, the iwi initially gifted 500 acres to the New Zealand settler parliament for the founding of a school that might serve their community.

When no school was developed, the tribe unsuccessfully sought to have the land returned in 1877 (New Zealand Government, 2012; Parliamentary Council Office, 2014). Later, in 1948 and 1960, the New Zealand government

seized several hundred acres of mana land … under the public works legislation for general housing purposes. Over time, the application of the native land laws led to [Twin Tides City] reserves being partitioned into smaller [economically unproductive] subsections. (Parliamentary Council Office, 2014, np)

By the early 21st century, the mana whenua iwi had become a landless proletariat (Parliamentary Council Office, 2014), thus, largely fulfilling much of Marx’s (1957) prophetic assessment of the land agent’s supposedly humanitarian plan to reproduce a stratified capitalistic society in Aotearoa (Hastings, 1990; Marx, 1957). Mana whenua iwi experiences of schooling from the mid-19th to mid-20th centuries also reflected the settler parliament’s wider assimilationist and pacification policies for Māori (Barrington & Beaglehole, 1974; Carlyon, 2022; Penetito, 2010; Simon, 1998; Walker, 1987). The children of the mana whenua iwi were exposed to a limited curriculum that largely equipped them to become obedient manual laborers—unlikely to be capable of protecting their economic base (lands) from further alienation in the courts (Barrington & Beaglehole, 1974, Kelsey, 1997; Penetito, 2010; Simon, 1998; Walker, 1987).

Much of this land alienation would occur after World War II, when government economic planning transformed the area into a satellite city—of largely state houses—to service a prosperous neighboring city (Keith, 1990). Improved rail and road links reduced travel time between these places, greatly contributing to Twin Tides’s population growth into the 1960s. The industrialization of parts of the city in the late 1960s reignited the thirst for cheap labor, precipitating another influx of low-skilled laborers (Keith, 1990). The first wave of these workers often came from rural Māori communities located elsewhere in Te Ika-a-Māui (Keith, 1990). Further workforce shortages in the late 1960s triggered the rapid arrival of waves of immigrant laborers from Polynesian nations such as Samoa, Tonga, and Fiji into “working-class” suburbs and jobs similar to those of the Māori (Elise, 2000), resulting in a “brown proletariat” (King, 2003, p. 404). The local (monocultural) schooling system was poorly equipped to deal with this new student demography because the largely Pākehā teaching workforce did not reflect the rapidly changing demography of the communities it served (Cumming, 2022; Elise, 2000; Fallon, 2018).

An economic recession following the 1973 OPEC oil crisis soon triggered growing unemployment levels (Elise, 2000; Kelsey, 1997), which left the predominantly “brown proletariat” (King, 2003) of Twin Tides city vulnerable to unemployment and discriminatory immigration policies—such as the infamous “dawn raids” by police on households of Pacific peoples deemed to be “overstayers” (Elise, 2000). Lower-skilled workers came under further attack from the neoliberal economic reforms of 1984 onward, which further illuminated the stark socioeconomic divides existing between those schools located in “have” and “have not” suburbs around Aotearoa (Elise, 2000; Fallon, 2018; Kelsey, 1997). As a result, intergenerational poverty now impacts many of the student families affiliated with the schools central to this article (Cumming, 2022; Elise, 2000; Fallon, 2018).

Framing racialized social-class inequalities

We use the historical context outlined above to point toward several interrelated factors that we will use to operationalize racialized social-class inequalities in the empirical sections of this article. There are multiple ways to conceptualize and measure social class, however, rather than adopting one definition or following a particular theorist, we take the pragmatic approach of accepting multiple approaches to social class and, therefore, to the measurement and description of social-class divisions. This approach reflects a commitment to “theoretical reflexivity” (Pomeroy, 2017)—that is, adopting theoretical approaches that are most appropriate to the focus of a particular investigation as opposed to viewing theoretical approaches as ideological or philosophical starting points.

Some sociologists (Bourdieu, 1984; Goldthorpe, 2010) have emphasized parental occupation and education as key indices of social class, reflecting a theoretical orientation that frames class in terms of labor-market position, distinguishing for example between “professional” and “manual” work. A related body of work considers young people’s aspirations or “imagined futures”—often in terms of future occupations within a social class structure—as a mechanism for the intergenerational reproduction of social class (Baillergeau & Duyvendak, 2022; Mazenod et al., 2019). While such models have not been completely superseded, other scholars have proposed models that consider the classed experiences of those not in paid work, and the “psycho-social” and emotional dimension of the way social class is experienced (Reay, 2006; Sayer, 2005; Sennett & Cobb, 1972). While educational attainment is not always used as a measure of social class, numerous studies have explored the relationship between social class and educational attainment, generally and in terms of specific disciplines, including mathematics (Noyes, 2009; Quaye & Pomeroy, 2022). Given these considerations, in this study we use parental occupation and education levels, students’ stated job aspirations, and students’ mathematics attainment as partial but complementary windows into students’ current experiences of social class and the potential for future class mobility.

The historical context aids in providing an additional layer to the way in which social class and racialized social-class inequalities should be interpreted in the context of Twin Tides City. One way is that it shows that specific racialized historical processes, such as land dispossession, racist school systems, and job insecurity for “low skill” workers actively produced current social-class positions of students in the Twin Tides area. Historical education and labor policies have actively deterred Māori and Pacific peoples from accessing well-paid professions and higher education and from succeeding in dominant terms in traditionally “academic” disciplines such as mathematics. Conversely, European settlers benefitted from the economic base of cheaply acquired land, were the founding members and gatekeepers of highly paid professions and institutions of higher education in the new colony, and had their culture reflected in the official schooling system. It is against this geographical and historical backdrop that we now analyze the experiences of eighth-grade students in mathematics in their first year of secondary school, with particular attention to the practice of streaming (tracking).

Streaming itself has been the subject of extensive research spanning many decades—scholars have discussed the interrelated dimensions of this complex range of practices (Domina et al., 2019), as well as its effects on self-confidence (Francis et al., 2020), emotional and behavioral outcomes (Papachristou et al., 2022), and peer relationships (Hargreaves et al., 2021). Many researchers have argued against streaming on a range of social-justice grounds, including that streaming exacerbates social-class and gender inequities (e.g., Archer et al., 2018; Jaremus et al., 2022; Oakes, 2005). Of key importance to this article’s argument is the strong consensus in existing research (e.g., Hodgen et al., 2023; Linchevski & Kutscher, 1998; Oakes, 2005; Wiliam & Bartholomew, 2004) that streaming tends to exacerbate achievement differences. In other words, in streamed contexts, there is a relative achievement advantage to being in a top stream and a relative achievement disadvantage to being in a low stream. Therefore, this study will add to this body of knowledge by providing a historically grounded and statistically based analysis of how streaming interacts with racial inequalities in Aotearoa.

Method

In this study we draw from data from a mixed methods study of socioeconomic, ethnic, and gender inequalities in eighth-grade mathematics in Aotearoa (Pomeroy, 2016, 2021). A previous publication provides a more detailed description of the study methodology than space permits here, including scale validation and missing-data analysis (Pomeroy, 2016). In the following section we describe the study setting and the data-collection methods used in the current analysis.

Setting and participants

Given the focus of the original study on intersectional inequalities, schools (and classes within schools) were recruited with the aim of maximizing the socioeconomic, ethnic, and gender diversity of the student sample. Therefore, the aim was to recruit schools that served communities across the spectrum of socioeconomic privilege and disadvantage, and which had significant cohorts of Māori, Pacific, and Pākehā (New Zealand European) students. In keeping with these aims, the study took place in three coeducational (mixed-sex) state secondary schools in and near Twin Tides City. Despite their relatively close geographical proximity, the schools drew on socioeconomically and ethnically contrasting communities. Queens College was a Grades 8–12 (Aotearoa Year 9–13) school located in a relatively affluent, majority Pākehā, commuter suburb. Robinson College was a Grades 8–12 school situated near and enrolling students from both very affluent, majority Pākehā and very poor, majority Māori and Pacific communities. In other words, Robinson had an unusually high degree of within-school socioeconomic and ethnic stratification. St. Edmunds College was “state integrated” (i.e., a state-funded school with a religious special character) and drew students from a socioeconomically disadvantaged, majority Pacific and Māori community. The schools also had contrasting sizes—Queens was the largest, followed by Robinson, then St. Edmunds, and this is reflected in the number of eighth-grade classes that participated in the study (nine, seven, and three, respectively, see Table 1).

Table 1. School and participant demographics. Ethnicity is based on administrative data.

School	Students in study	Classes	Māori $\mathit{n}$/%	Pacific $\mathit{n}$/%	Pākehā $\mathit{n}$/%	Asian $\mathit{n}$/%	SES (M/SD)
Queens	245	9	55/.22	34/.14	123/.50	29/.12	54/17
Robinson	161	7	50/.31	42/.26	53/.33	12/.07	46/19
St. Edmunds	69	3	15/.22	44/.64	4/.06	5/.07	40/16

The three schools all used varying degrees of streaming. The extent of streaming depended on school policy and size (see Table 2 and Figures 1–3). St. Edmunds College had only three eighth-grade classes, one “top” class and two “mixed” classes, based on an overall assessment of academic ability (see Figure 1). Robinson had eight eighth-grade classes, seven of which participated in the study, and that were streamed based on mathematics assessments into two “top,” four “middle,” and two “bottom” mathematics classes (see Figure 2). Queens had 12 eighth-grade classes, nine of which participated in the study, and that were tightly streamed by mathematics achievement into around five identifiable levels (see Figure 3).

Figure 1. Ethnicity by stream at St. Edmunds.

Figure 2. Ethnicity by stream at Robinson.

Figure 3. Ethnicity by stream at Queens.

Table 2. MA = mathematics achievement.

	Set	Students (n)	MA (M/SD)	Māori %	Pacific %	Pākehā %	Asian %	SES (M/SD)
Queens	Streamed	34	71.2/10.9	11.8	5.88	61.8	17.6	59.1/17.1
	High	32	56.1/9.55	18.8	9.38	59.4	9.38	57.0/17.6
	Mid-high	27	53.5/11.6	11.1	7.41	70.4	7.41	56.7/16.9
	Mid-high	30	44.1/11.5	16.7	23.3	46.7	10.0	56.9/17.7
	Middle	28	41.5/8.34	17.9	21.4	50.0	10.7	51.5/15.4
	Middle	28	40.4/12.2	46.4	10.7	21.4	21.4	46.0/13.3
	Low	25	39.6/9.78	24.0	16.0	56.0	4.00	51.8/15.8
	Low	16	36.0/9.86	31.3	18.8	37.5	12.5	51.8/18.5
	Low	25	32.0/9.13	32.0	16.0	40.0	12.0	46.7/15.7
Robinson	High	30	59.4/7.01	26.7	3.33	46.7	16.7	61.3/13.5
	Middle	27	41.0/5.80	22.2	33.3	40.7	3.70	40.9/20.3
	Middle	25	38.8/5.41	28.0	24.0	44.0	4.00	43.0/17.6
	Middle	29	37.8/4.77	31.0	24.1	34.5	6.90	39.5/15.4
	Low	25	23.5/6.36	40.0	48.0	8.00	0.00	42.7/17.8
	Low	25	23.8/7.91	40.0	28.0	20.0	12.0	39.2/16.6
St. Edmunds	High	26	40.0/15.8	23.1	57.7	11.5	7.69	43.9/14.7
	Mixed	21	29.0/12.5	14.3	66.7	0.00	14.3	37.0/19.6
	Mixed	22	28.0/5.61	27.3	68.2	4.55	0.00	39.0/14.1

Scores are standardized against a national age-normed mean of 50 and standard deviation of 20.

Data collection

The wider study included student questionnaires (n = 451), student interviews (n = 40), teacher interviews (n = 15), mathematics test results, classroom observations, and school administrative data including class lists and the gender and ethnicity provided on students’ enrollment forms. This section provides a detailed account of the quantitative methods that generated the data analyzed in the current article—namely, mathematics achievement data, student job aspirations, and student social class and ethnicity. We have presented data based on interview data in other manuscripts (Pomeroy, 2016, 2021).

Mathematics achievement data

Students in all schools sat a mathematics test in the first term of Grade 8. Queens and St. Edmunds students sat the Mathematics Progressive Achievement Test (PAT). The PAT has nine possible results, Stanine³ One (lowest) to Stanine Nine (highest). Stanines are derived by comparing the number of correct items with a nationally representative sample of the same age. A stanine is analogous to a percentile, but with a population divided into nine equal groups rather than 100, so a student at Stanine Five is in the “middle ninth” of eighth-grade students in Aotearoa for mathematics. Robinson students sat the Mathematics e-asTTle test, which produces results that link students’ progress to the eight levels of the New Zealand Curriculum (Ministry of Education, 2007). Although the two tests are not perfectly commensurate, both can be mapped onto national eighth-grade norms. To provide a single mathematics achievement metric that was broadly comparable across the three schools, the PAT and e-asTTle were both standardized according to a scale on which the national (not sample) mean was 50 and the national standard deviation was 20 (for further details, see Pomeroy, 2016).

Student questionnaires: Social class, ethnicity, and job aspirations

Students completed a pen-and-paper questionnaire during the first school term under the supervision of their mathematics teacher and Pomeroy, our lead author. When students were absent, teachers provided them with the questionnaire at a later date. Six percent (28/477) of students enrolled in participating classes did not attempt the questionnaire due to absence and five percent (24/477) were rejected due to large volumes of missing data or implausible response patterns. Overall, 425 student questionnaires, representing 89% of eighth-grade students enrolled in participating schools, were included in the analysis.

Social class

As discussed, there is no single definition of, or metric for, social class. However, it was strongly desirable to include a student-level socioeconomic metric of some sort in the study, as in Aotearoa social class is often either absent from analyses of equity or measured using very crude school-level indices that ignore the considerable within-school variation in socioeconomic status. This study utilized the New Zealand Socio-economic Index 2006 (NZSEI-06, Milne et al., 2013). The NZSEI-06 is based on data from the 2006 national census, and a “returns to human capital” model. It is based on the theory that

there is a relationship between cultural capital (i.e., education) and material rewards (i.e., income), and that this relationship is mediated through occupation. More simply, the “returns to human capital” model views occupation as the means by which one’s education is converted into income. Thus, differences in occupation are likely to represent differences in life chances and opportunity, and on this basis occupation can be used to stratify individuals according to socio-economic status. (Milne et al., 2013, p. 12)

The NZSEI-06 uses occupation as the primary indicator of SES and assigns values to occupations on a scale of 10 (lowest) to 90 (highest)—for example, legal professionals (80), mechanical engineering trades workers (45), and cleaners and laundry workers (14). Where information on occupation is unavailable, the NZSEI-06 has suggested values on the same scale assigned to highest educational qualifications—for example, doctorate (75), bachelor’s degree (61), and no school qualifications (35). Education levels are considered a less accurate indicator of SES than occupation.

Questionnaires asked students to list the main occupation and highest education level of one or both parents or carers, and Pomeroy manually applied the NZSEI-06 coding manual to classify all responses. Each student’s assigned parental SES value was determined by the highest of their listed parents or carers, with priority given to occupation-based, over education-based, scores. Not all responses were straightforward to code. For example, “works at gym” could be a personal trainer, cleaner, office administrator, or human resources manager; such responses were coded as “low validity.” Only responses that corresponded clearly with a category on the NZSEI-06 were included in the quantitative analyses using the SES variable. Of the 449 students who completed an otherwise valid questionnaire, 352 (78%) were assigned a usable or “high validity” parental SES value. A missing-data analysis showed that students with “high validity” SES values had higher mathematics attainment (M = 45.15, SD = 15.78) than students with missing or “low validity” SES values (M = 38.48, SD = 13.31), t(404) = 3.28, p < .001. Given the strong correlation between SES and mathematics achievement, this result suggests that low-SES students may be slightly underrepresented in the quantitative socioeconomic analyses that follow in the Results section.

Job aspirations

The questionnaire included the open response item What job would you most like to do when you finish school? Responses to this item, which had a response rate of 70%, were coded in two ways. Firstly, where possible, the jobs listed were mapped to the NZSEI-06 using the same process applied to parental jobs—we call this variable “Aspired SES.” Secondly, open responses were thematically grouped into generic categories such as “professional” (e.g., lawyer, nurse, marine biologist); “trade” (e.g., plumber, builder); and “service” (e.g., flight attendant, hairdresser). There were also bespoke and more stereotype-based categories specific to the themes of this research—for example, “maths jobs” (accountant, physical scientist) and “physical jobs” (police, armed forces, professional sport). Each job category was represented as a “dummy” (yes/no) variable for each student. While there is no single and completely objective system for coding jobs into categories, this study used categories that aligned closely with its focus as a sociological analysis of mathematics learning. For example, “professional” jobs mostly required higher education and had relatively high values on the NZSEI-06. “Trades” and “service” jobs have traditional “working class” associations and are typically strongly gendered masculine and feminine, respectively. The “maths jobs” and “physical jobs” were based on a wider analysis of the perceived dichotomy between intellectual and physical pursuits (Pomeroy, 2021).

Ethnicity

School administrative data provided ethnicities for all students enrolled in participating classes. All students were listed with a single ethnicity: Māori, Pacific, NZ European (Pākehā), Asian, or Other. Parents had the option to list multiple ethnicities on enrolment forms, however, students were then assigned a single ethnicity based on a priority system Māori > Pacific > Asian > NZ European (Ministry of Education, 2015). For example, if a parent listed Māori and NZ European on their child’s enrolment form, the child would be represented as Māori in school statistics. Therefore, the administrative ethnic data was limited in at least two ways: (1) it did not allow multiple ethnicities to be included, and (2) it systematically overrepresented some ethnicities and underrepresented others, relative to the actual data entered in enrollment forms. To mitigate these limitations, the questionnaires included an open-response item: What is your ethnicity (e.g., Samoan, Pākehā)? Pomeroy also explained verbally to the students that they could report more than one ethnicity if they felt that more than one was a significant part of their heritage or identity. Ninety-five percent of students provided an interpretable response to this item, which was coded into four separate “dummy” variables: Pākehā, Māori, Pacific, and Asian, so that multiple ethnicities could be recorded. According to this self-report data 99 students (23%) identified with more than one ethnicity. To confirm the robustness of results related to ethnicity, we ran key statistical analyses using both school administrative and student self-report ethnicities. Unless otherwise mentioned, the substantive results were consistent regardless of which data source was used.

For this study, we chose to conduct ethnicity analyses using only three ethnic groupings—namely, Māori (n = 104), Pacific (n = 123), and Pākehā (n = 150). The Asian category has been excluded from the analysis firstly because of its much smaller size (n = 39) and secondly because of its lack of ontological coherence as a meaningful descriptor. While any ethnic category oversimplifies complex and sometimes fluid identities, the Asian category was particularly problematic because it included, for example, Chinese New Zealanders, whose ancestors have lived in Aotearoa for several generations and who may speak only English, and recently arrived Southeast Asian refugees. Similarly, we did not report results for the “Other” ethnic category because of its very small size (n = 9) and catch-all nature. By including only three ethnic categories in the analysis, we risk appearing to gloss over the highly multicultural composition of contemporary society in Aotearoa. We acknowledge this limitation but see it as preferable to making claims that our data cannot support regarding other ethnic groups and suggest that with a different and larger sample, such analyses might be fruitful.

Statistical methods

We used a variety of statistical methods to address questions related to students’ job aspirations, namely: (1) are student demographics associated with aspirations in different job categories? (2) are student demographics associated with the SES-value (according to the NZSEI-06) of these jobs? We quantitatively measured the effects of ethnicity, parental SES, gender, and maths achievement on students aspired job categories and their aspired SES.

We compared mean SES between ethnicities using t-tests with Šidák correction for multiple comparisons.

Aspired job categories are analyzed with a series of logistic regressions—e.g., aspires to physical job—yes/no. Covariates include parental SES, ethnicity, and gender. In Wilkinson notation,

$\mathrm{Aspiration}\sim \mathrm{Parental}\mathrm{SES}+\mathrm{Ethnicity}+\mathrm{Gender}$

Multinomial regression does not work in this case because students may have multiple aspirations. We do not attempt to correct for multiple comparisons, as there is no standard practice. We leave the reader to interpret the uncorrected p-values.

We model the quantitative SES-value (according to the NZSEI-06) of students imagined futures with a multivariate linear regression. Covariates include parental SES, ethnicity, and gender:

$\mathrm{Aspired}\mathrm{SES}\sim \mathrm{Parental}\mathrm{SES}+\mathrm{Ethnicity}+\mathrm{Gender}$

Finally, we implement two models of mediation, both using Preacher-Hayes bootstrapping, as this is robust to OLS normality assumptions—first, a simple mediation to examine how parental SES mediates the relationship between ethnicity and aspired SES. This is represented as the system of regressions,

$\mathrm{Parental}\mathrm{S}\mathrm{ES}\sim \mathrm{Ethnicity}$

$\mathrm{Aspired}\mathrm{S}\mathrm{ES}\sim \mathrm{Parental}\mathrm{SES}+\mathrm{Ethnicity}$

and second, a serial mediation to examine how maths attainment mediates the relationship between ethnicity, parental SES, and aspired SES (while parental SES continues to mediate ethnicity and aspired SES):

$\mathrm{Parental}\mathrm{S}\mathrm{ES}\sim \mathrm{Ethnicity}$

$\mathrm{Maths}\mathrm{Attainment}\sim \mathrm{Parental}\mathrm{SES}+\mathrm{Ethnicity}$

$\mathrm{Aspired}\mathrm{S}\mathrm{ES}\sim \mathrm{Maths}\mathrm{Attainment}+\mathrm{Parental}\mathrm{SES}+\mathrm{Ethnicity}$

Results

Results are presented in two sections. The first section describes how different measures and proxies for social class are racialized on entry into secondary school. The second section reports the racialization of students’ experiences of learning mathematics in the first year of secondary school and in particular the practice of streaming. Taken together these two sets of results provide a platform for a discussion about the ways in which the schools in this study reinforced or challenged the racialized social class inequalities in the communities in and around Twin Tides City.

Racialized class divisions on entry into secondary school

This section presents evidence concerning the social-class differences observed between and within three ethnic groups: Māori, Pacific, and Pākehā. As previously discussed, we operationalize social class in several different ways so as to provide a multidimensional account of racialized class inequalities on entry into secondary school.

Socioeconomic status (SES) as measured by the NZSEI-06

Of the 352 students whose questionnaire responses could be used to derive a value on the NZSEI-06 (Milne et al., 2013), the highest-SES racial group was Pākehā ($\mathit{M}=54.2$, $\mathit{SD}=16.4$), followed by Māori ($\mathit{M}=46.1$, $\mathit{SD}=16.2$), and finally Pacific ($\mathit{M}=43.7$, $\mathit{SD}=19.3$). Pairwise two-tailed $\mathit{t}$-tests (with Šidák correction) show that both Māori and Pākehā ($\mathit{t}\left[202\right]=3.77$, $\mathit{p}<0.001$), and Pacific and Pākehā ($\mathit{t}\left[165\right]=4.29$, $\mathit{p}<0.001$) have significantly different mean SES values. This result should be interpreted in the context of the wide socioeconomic variation within each ethnic group (see Figure 4). In other words, the parents of this study’s participants reflected the national pattern of Pākehā overrepresentation and Māori and Pacific underrepresentation among those with higher-education qualifications and well-paid, professional jobs.

Figure 4. Socioeconomic distributions by race for Pākehā, Māori, and Pacific students.

Job aspirations

This section analyzes student responses to the survey item “What job would you most like to do when you leave school?”

We conducted a series of logistic regression analyses with SES, ethnicity, and gender as predictor (independent) variables and job categories as the outcome (dependent) variable in each analysis (see Table 3). In other words, we investigated the extent to which SES, ethnicity, and gender could be used to predict whether a student would list a certain job category as desirable for their future. Professional jobs have a strong theoretical alignment with and are sometimes used as an indicator of privileged social-class status, therefore, this category was of particular interest. There was a highly significant positive correlation between SES and professional aspiration (SES coefficient = 0.0309; odds ratio = 1.03; p < .001). After controlling for SES, Pākehā students were the most likely and Māori students least likely to list a professional aspiration, although the racialized differences did not reach statistical significance.

Table 3. Demographic predictors of student job aspirations.

	Covariates	Coefficient	Odds Ratio	$\mathit{p}$-value
Professional Jobs	SES	0.0309	1.03	$1.42\times {10}^{-5}$
	Female	0.247	1.28	.307
	Māori	−0.259	0.772	.374
	Pacific	−0.106	0.899	.723
Trades Jobs	SES	−0.0461	0.955	.0141
	Female	−3.13	0.0439	$2.70\times {10}^{-3}$
	Māori	0.632	1.88	.309
	Pacific	0.251	1.29	.713
Service Jobs	SES	−0.0235	0.977	.0387
	Female	–	–	–
	Māori	0.213	1.24	.636
	Pacific	0.198	1.22	.668
Maths Jobs	SES	0.0211	1.02	.0775
	Female	−1.29	0.274	$2.37\times {10}^{-3}$
	Māori	−0.199	0.819	.659
	Pacific	−1.27	0.282	.0518
Physical Jobs	SES	$-7.04\times {10}^{-3}$	0.993	.369
	Female	−1.23	0.293	$5.25\times {10}^{-6}$
	Māori	0.759	2.14	.0169
	Pacific	0.408	1.50	.246

Trades jobs and service jobs are traditionally working-class job categories that are gendered masculine and feminine, respectively. As predicted, there was a statistically significant negative relationship between SES and trades aspiration (SES coefficient = −0.0461; odds ratio = 0.955; p < .05) and between SES and service aspiration (SES coefficient = −0.0235; odds ratio = 0.977; p < .05). Trades and service aspirations had the predicted relationship with gender; boys were more likely than girls to list trades jobs as an aspiration (female coefficient = −3.13; odds ratio = 0.0439; p < .001) and only girls listed service jobs as an aspiration. After controlling for SES and gender, Māori students were most likely and Pākehā students least likely to list a trades or service aspiration, although the racialized differences did not reach statistical significance.

Two less conventional job categories are worth mentioning at this point. Physical jobs were jobs that Pomeroy judged to require or have discursive associations with bodily skill, fitness, or appearance—for example, sports professionals, dancers, mechanics, police officers, and military jobs. Maths jobs included computer programming, accounting, physical science, and other fields that would require or benefit significantly from university-level mathematics. These two categories were assembled based on the stereotypical and discursive association between Whiteness and intellectual ability, and between Indigeneity and “physical” prowess in settler colonial contexts (Hokowhitu, 2008; Pomeroy, 2021). Maths jobs were not significantly related to SES but were strongly gendered masculine (female coefficient = −1.29; odds ratio = 0.274; p < .01). Pākehā students were the most likely and Pacific students least likely to aspire to a maths job, with the difference between Pākehā and Pacific students around the conventional cut-off for statistical significance (Pacific coefficient = −1.27; odds ratio = 0.282; p = .052). The stereotype of White men being mathematical was reflected in students’ imagined futures. Similar to maths jobs, physical jobs were not significantly related to SES and were strongly gendered masculine (female coefficient = −1.23; odds ratio = 0.293; p < .01). However, Māori students were the most likely and Pākehā students least likely to aspire to a physical job, with the difference between Māori and Pākehā reaching statistical significance (Māori coefficient = 0.759; odds ratio = 2.14; p < .05). In other words, the stereotype of Māori and Pacific boys and men being physically capable was reflected in students’ imagined futures.

To further examine the implications of racialized class inequalities, we constructed a multivariate linear regression analysis with aspired SES as the outcome variable and SES, gender, and ethnicity as predictors (see Table 4). SES significantly predicted aspired SES (SES coefficient = 0.255; p < .001), gender was not related to aspired SES, and ethnicity did not have a statistically significant relationship with aspired SES.

Table 4. Aspired SES regression results.

Aspired SES ~ Parental SES + Ethnicity + Gender (OLS)
Covariate	Coefficient	p-value
Parental SES	0.255	$5.77\times {10}^{-6}$
Māori	−0.817	.735
Pacific	2.20	.379
Female	1.20	.533
Parental SES ~ Ethnicity (OLS; Preacher-Hayes bootstrapping)
Covariate	Coefficient	95% CI
Māori	−7.41	−12.3; −2.52
Pacific	−8.18	−13.9; −2.49
Maths Achievement ~ Parental SES + Ethnicity (OLS; Preacher-Hayes bootstrapping)
Covariate	Coefficient	95% CI
Parental SES	0.209	.101; .315
Māori	−6.43	−10.4; −2.41
Pacific	−7.40	−11.9; −2.82
Aspired SES ~ Maths Achievement + Parental SES + Ethnicity (OLS; Preacher-Hayes bootstrapping)
Covariate	Coefficient	95% CI
Maths Achievement	0.237	.109; .377
Parental SES	0.218	.108; .333
Māori	1.11	−3.05; 5.46
Pacific	3.84	−1.78; 9.59

However, given previous results, we have evidence to support that SES, and thus aspired SES, depends on ethnicity. We tested this hypothesis with a simple mediation analysis, wherein SES mediates the relationship between ethnicity and aspired SES. We find that Māori (Māori coefficient = −7.41; p < .05) and Pacific (Pacific coefficient = −8.18; p < .05) students have significantly lower SES values than Pākehā students. So, while ethnicity does not significantly correlate with aspired SES directly, SES (SES coefficient = 0.268; p < .05) is significantly positively correlated with aspired SES. Thus, ethnicity correlates with aspired SES via parental SES: Māori and Pacific students have lower SES parents and aspire to lower SES jobs.

Given that schools cannot influence the SES of their parent community but do have a role in supporting their students’ academic achievement, it seemed important to assess whether mathematics achievement had any direct relationship with aspired SES after controlling for SES. Therefore, we introduced mathematics achievement into the mediation model above, in addition to SES, as a mediator of aspired SES. This turns our simple mediation model into a serial mediation model. Mathematics achievement is significantly associated with parental SES (SES coefficient = 0.209; p < .05) and ethnicity (Māori coefficient = −6.43; p < .05. Pacific coefficient = −7.40; p < .05). Students of higher SES parents tend to achieve higher scores on maths tests, while Māori and Pacific students tend to achieve lower scores on maths tests relative to Pākehā students (see Figure 5). As in our previous model, ethnicity does not directly predict aspired SES but does so indirectly through parental SES (SES coefficient = 0.218; p < .05). Of importance, we find that aspired SES is significantly positively correlated with maths achievement (MA coefficient = 0.237; p < .05). In other words, students with higher mathematics achievement—typically higher SES Pākehā students—tend to have job aspirations with higher SES values.

Figure 5. Mathematics attainment by SES and ethnicity.

As we will argue in more detail in the discussion, this result suggests that any process that racializes mathematics achievement has the potential to disrupt or reinforce the intergenerational reproduction of racialized class inequalities.

Mathematics achievement

Mathematics achievement itself is an educationally relevant outcome that is closely related to social class and its reproduction across generations. International comparative studies of school achievement have shown that Aotearoa has a notably high “socio-economic gradient” for academic achievement (Caygill et al., 2013; May et al., 2013)—that is, social class is a stronger predictor of school achievement in Aotearoa than it is in most other jurisdictions. Mathematics achievement also functions as a gatekeeper to a wide range of higher-education pathways and professions. Therefore, mathematics achievement is likely to be related to the potential for upward social mobility.

Discussion

Our purpose in this article was to investigate the extent to which streaming (grouping by perceived academic ability) in eighth-grade mathematics reinforces or challenges racialized social-class inequalities in Aotearoa. The results presented above fell into three main categories. We briefly synthesize each of these sets of results here, then discuss the implications in terms of this purpose.

The first and most extensive set of results were those that described racialized social-class inequalities on entry into secondary school. These results compared Pākehā, Māori, and Pacific students on a range of indicators related to social class. According to a standard metric based on parents’ occupations and educational credentials, Pākehā students possessed more social-class privilege than Māori and Pacific students. Turning to students’ stated job aspirations, Pākehā students were more likely than Māori or Pacific students to aspire to well-paid professional jobs requiring university education and to jobs stereotypically associated with mathematics. Conversely, Māori and Pacific students were more likely than Pākehā students to aspire to service and trades jobs and to other “manual” jobs stereotypically associated with bodily rather than intellectual capabilities. Overall, Pākehā students aspired to jobs that were higher on a standard metric of SES than the jobs that Māori and Pacific students aspired to. Finally, Pākehā students had higher mathematics achievement than Māori or Pacific students. While there was wide social-class variability within each ethnic group, overall, the results present a bleak picture of racialized social-class inequalities between students entering secondary school. Furthermore, the close relationship between parental social class and students’ stated social-class aspirations suggests an intergenerational component of the observed inequalities (Quaye & Pomeroy, 2022).

The second set of results moved beyond raw ethnic comparisons to more-nuanced aspects of the racialized social-class inequalities. In particular, both parental SES and mathematics achievement were strong predictors of SES aspirations. In fact, in statistical models that included SES and mathematics achievement, ethnicity was not directly a significant predictor of aspirations. Instead, the racialized inequalities in students’ job aspirations were closely related to their mathematics achievement and parents’ unequal present circumstances, strongly influenced by historically rooted privilege and oppression in settler colonial configurations of power. Māori and Pacific students were more likely to aspire to high-SES jobs when their parents also had such jobs, a finding that also came through strongly in the student interviews (for an analysis of the interview data see Pomeroy, 2021). This suggests that if racialized social-class inequalities are reconfigured across generations—for example, through downward social mobility of Pākehā or upward social mobility of Māori—then the new (more equal) configuration is likely to persist.

Mathematics achievement also predicted students’ social class aspirations, independently from SES or ethnicity. Therefore, mathematics achievement is one factor over which schools have some influence that can intervene in the intergenerational reproduction of racialized social-class inequalities. This implies that school structures or processes that strongly support the mathematics achievement of Māori and Pacific students could potentially challenge intergenerational inequalities. Conversely, school structures or processes that support the mathematics achievement of Pākehā students relative to other students would reinforce intergenerational inequalities.

The third set of results (see Table 2 and Figures 1–3) described the racial stratification of the streamed mathematics classes in each school. In all schools, Pākehā students were overrepresented in the higher streams and Māori and Pacific students were overrepresented in the lower streams, although the degree of racial stratification varied between schools. Given that there are enduring racialized achievement inequalities in Aotearoa and that streaming involves allocation of students into classes based on some measure of achievement, we would argue that the observed racial stratification of streamed mathematics classes is an unintended but predictable consequence of streaming. As long as racial achievement inequalities persist, schools must accept that streaming by achievement means streaming by race.

To be clear, our empirical findings taken in isolation do not support conclusions about the extent to which streaming reinforces or disrupts racialized social-class divisions. To address this question, our findings must be considered alongside previous research on streaming and mathematics achievement. As noted earlier, there is a strong consensus that streaming confers a relative achievement benefit to high-stream students and a relative achievement disadvantage to low-stream students (e.g., Hodgen et al., 2023; Linchevski & Kutscher, 1998; Oakes, 2005; Terrin & Triventi, 2023; Wiliam & Bartholomew, 2004). Applied to the racially stratified classes in the current study, this implies that the relative achievement benefit of being in a high stream is conferred disproportionately on Pākehā students and the relative achievement disadvantage of being in a low stream is conferred disproportionately on Māori and Pacific students. On this basis, we argue that streaming reinforces rather than challenges racialized inequalities in mathematics achievement. Furthermore, given mathematics achievement’s capacity to predict the social-class position that students imagine for themselves in the future, reinforcing racialized inequalities in mathematics achievement very likely also reinforces the intergenerational reproduction of racialized social-class inequalities. The suggested role of streaming in reinforcing racial inequalities in both mathematics achievement and social class seems all the more plausible given that stream allocation tends to be racially biased, with White students overallocated to high streams and students of color overallocated to low streams relative to achievement (Connolly et al., 2019; Oakes, 2005).

Limitations

In this study we only addressed streaming as a potential mediator of racialized social-class inequalities in schooling. While we consider streaming an important practice to scrutinize, we acknowledge that many other school-related factors not considered in this study may reinforce or challenge racialized social-class inequalities. A second limitation derives from our reliance on test data, survey data, and administrative data in our analysis. Using such data enabled us to conduct the statistical analyses that formed the basis of the Results section; however, it meant that we had little to contribute to discussion of the lived or affective experience of streaming. Finally, due to sample-size constraints, we were unable to include nesting (students within classes, classes within schools) in our statistical models, which limited our ability to detect whether any of the results reported differed between the three participating schools.

Conclusion

In this article we have sought to address the question of whether streaming (grouping students by perceived academic ability) challenges or reinforces racialized social-class inequalities in Aotearoa. We summarized some of the key historical processes which, over time, have actively produced racialized and intergenerational social-class inequalities in a particular location in Aotearoa. Against this backdrop we analyzed social-class inequalities between Pākehā students, on the one hand, and Māori (Indigenous) and Pacific students, on the other. On all measures used (parental occupation and education, student job aspirations, and mathematics attainment) Pākehā students as a group occupied higher social-class positions than Māori or Pacific students, confirming previous findings in Aotearoa (Rashbrooke, 2014). Within each ethnic group, students with higher mathematics attainment were more likely to aspire to high-SES jobs, even after controlling for parental SES. This finding flagged mathematics attainment as one potential “lever” for challenging racialized social-class inequities. However, all schools practiced “streaming,” allocating students into classes based on tests taken at the start of Grade 8, such that “top” classes were disproportionately Pākehā and “bottom” classes were disproportionately Māori and Pacific. Extensive prior research shows that streaming confers a relative achievement benefit on students in top streams and a relative achievement disadvantage on students in low streams. On this basis we have argued that streaming in eighth-grade mathematics in Aotearoa reinforces racialized social-class inequalities. This outcome is in direct contradiction to a wide range of policy initiatives aimed at reducing such inequalities (e.g., Ministry of Education, 2018, 2020) and the legal requirement for schools to ensure “equitable outcomes for Māori students” under the Education and Training Act (New Zealand Government, 2020). Perhaps not surprisingly, Māori educational organizations are leading a movement to end streaming in Aotearoa (Tokona te Raki, 2023).

Acknowledgments

Diane Reay and Kenneth Ruthven supervised the original study on which this article is based. The Woolf Fisher Trust and Cambridge Trusts funded the original study on which this article is based.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

David Pomeroy

David Pomeroy is a senior lecturer in mathematics education at the University of Canterbury. He uses approaches from sociology and gender studies to examine the relationship between who we are (in particular our social class, ethnicity, and gender) and which school subjects we like and do well in. As a former secondary mathematics teacher, he is especially interested in educational inequalities in mathematics learning and in ways of teaching that enable more students to experience curiosity, joy, struggle, and success in mathematics. His current research has two main streams: (1) collaborating with secondary mathematics departments that are transitioning to mixed “ability” classes to make the transition successful for students and teachers; (2) working with Maths Craft NZ to better understand the educational potential of craft-based mathematics learning to make mathematics more accessible and rewarding.

Liam Gibson

Liam Gibson is a PhD candidate at Te Whare Wānanga o Waitaha University of Canterbury. He is studying gender and ethnicity gaps in academia using mathematical and statistical models.

Richard Manning

Dr. Richard Manning is a former secondary school teacher (1991–1999) who previously worked as a researcher for the Department of Māori Affairs and the Iwi Transition Agency (1989–1990). He has served as the education advisor at the New Zealand State Services Commission (2005–2006) and as a treaty claims researcher and inquiry facilitator for the Waitangi Tribunal (2006–2007). Richard has extensive experience as a teacher educator (1999–2003, 2008 to present). He is currently a senior lecturer at the University of Canterbury Faculty of Education, teaching in the Treaty of Waitangi Education Programme, primary social sciences curriculum area, and Pasifika education courses. Richard advises senior management teams of numerous New Zealand government agencies, corporations, and schools on a wide range of Treaty of Waitangi issues. He currently coleads a Marsden Fund research project titled Ngā Hanganga Mātua o te Whakaako Hitori: Critical Pedagogies for History Educators in Aotearoa New Zealand. Richard has also worked with Indigenous and non-Indigenous colleagues in Australia, Canada, and the United States to develop comparative studies in Indigenous education, particularly the ethical teaching of Indigenous histories.

Notes

¹ Aotearoa is used in the Māori language and increasingly in New Zealand English to refer to the modern state widely known as New Zealand.

² Pacific in this article refers to people of many Pacific Island nations, including Samoa, Tonga, and Fiji, and their descendants living in Aotearoa New Zealand.

³ A stanine is analogous to a percentile, but with a population divided into nine equal groups rather than 100.