Educational aspirations among UK Young Teenagers: Exploring the role of gender, class and ethnicity

Large socio-economic differences in educational attainment and participation in Higher Education (HE) are seen in the United Kingdom (UK). Furthermore, improvements in attainment and in rates of progression to university have been much faster for most ethnic minority groups than for White children. Political rhetoric explains these differences in terms of a lack of aspiration, particularly among White, working-class boys. This paper extends recent work by examining the intersection of gender, class and ethnicity in their association with aspirations for higher levels of education among teenagers born in the late 1990s and early 2000s. We adopt a developmental-context approach using detailed information collected from teenagers and their parents within the United Kingdom Household Panel Survey. White boys from the lowest occupational class and from workless households have the lowest aspirations (around one-half have a positive aspiration for college or university) because the three elements – being White, male and working class – combine in an additive fashion to encourage lower aspiration. However, even though this figure is low, it is higher than the percentage of working-class boys who go to university. Thus, focusing on aspirations alone will not on its own reduce ethnic differences in HE participation. Class and ethnic differences in parental attitudes towards education, levels of parental engagement with their children’s schoolwork, and in the quality of the parent – child relationship act as important mediating factors. Key Stage 2 scores from the English National Pupil Database demonstrate that attainment is also a key mechanism through which parental class influences teenagers’ aspirations. The fact that the remaining unexplained differences in HE participation by ethnicity are increasing over time also suggests that these other factors are playing an increasingly important role in driving participation rates amongst ethnic minorities. Further research could usefully explore the specific factors that underlie these differences. between Indian and White teenagers, and the difference between male teenagers from professional backgrounds and male teenagers from working-class backgrounds. We conclude that all three dimensions are important. However, the multiple regression analyses find no evidence of an interaction effect between gender and class, or between gender and ethnicity, or between class and ethnicity – that is to say the effect of being male, white and working class appears to work in an additive way to affect the change of aspiring to college/university. In other words the impact of occupational class upon educational aspirations is similar for all ethnic groups corresponding to previous findings (e.g., Conolly, 2006; Rothon, 2007). White boys from the lowest occupational class and from workless households have the lowest aspirations (50% and 48% have a positive aspiration for college/university), this is because the three elements – being White, male, and working class – combine in an additive fashion to encourage lower aspiration. Our findings provide support then for the UK Government’s concern about low rates of application to HE from this group. backgrounds and have lower educational aspirations. These teenagers are retained in the analysis because they contribute information on the other covariates. However, we do not attempt to interpret the coefficients for this ‘not known’ group. Sensitivity analyses showed that the findings from the statistical-modelling results are unchanged whether or not this ‘not known’ group are included. 2 We tested for statistical significance between those from an intermediate-class background, those from a semi- routine or routine class background, and those from families where no parent is in employment by re-running the model with alternative categories as the baseline. Children living in a household where at least one parent has a semi-routine or routine job do not differ in positive aspiration to go to university when compared to children living in non-employed households.


Introduction
There has been a resurgence of interest among commentators, policy makers and academia into the mechanisms through which social inequality is reproduced in response to evidence showing continuing inequalities in educational opportunity, and declining social mobility in the United Kingdom (UK) (e.g., Blanden et al., 2005). Politicians have often blamed low aspirations (Roberts & Evans, 2012;St. Clair et al., 2013;Baker et al., 2014). Prime Minster David Cameron explained that promoting social mobility required helping people from poorer households and ethnic minorities to 'raise aspirations and get them to think they can get all the way to The fact that the remaining unexplained differences in HE participation by ethnicity are increasing over time also suggests that these other factors are playing an increasingly important role in driving participation rates amongst ethnic minorities. Further research could usefully explore the specific factors that underlie these differences.
Simultaneously, the gender gap in educational outcomes is still increasing, with boys falling behind. However, the size of the gender gap differs by socio-economic background and ethnicity, reflecting complex interactions between individual characteristics and context (Schoon & Eccles, 2014). It is, therefore, insufficient to examine the association between parental class, ethnicity and gender in isolation of each other (Strand, 2014).
In this paper we extend recent work and examine the intersection of gender, class and ethnicity in their association with aspirations for higher levels of education among UK teenagers. We adopt a developmental-context approach, similar to that used in previous studies of educational aspirations, expectations and outcomes (Schoon, 2010;Chowdry et al., 2011;Khattab, 2015). We analyse detailed data collected within a large, household survey to examine how parental and individual attitudes and behaviour mediate the relationship between gender/class/ethnicity and educational aspirations. The current UK context is much different to that even a decade ago, and our paper provides new insights in a number of ways: First, unlike for earlier cohorts, including those in the National Longitudinal Study of Young People in England (LSYPE), further education is becoming 'the norm' (Schoon, 2010) with considerably widened university participation across socio-economic backgrounds (Crawford & Greaves, 2015). Second, there is growing ethnic diversity among young people in the UK as a result of both increased rates of international migration during the past decade as well as family formation in the UK of previous migrants (ONS, 2012). Of particular note has been the growth of the Black-African population in the UK as a result of increased migration (primarily to London) from countries such as 730 A. Berrington et al. Somalia, Zimbabwe, Congo and Nigeria (Simpson, 2015). Related to both points, all ethnic minority groups are now more likely to go to university than White-British young peoplea very recent change, especially in regards to Black-Caribbean and Black-African students' changing participation (Crawford & Greaves, 2015). Finally, the new analyses are made possible by the availability of a nationally representative survey of sufficient size to enable analyses of minority ethnic groups, to sit alongside analyses undertaken on the Youth Cohort Studies and LSYPE (e.g., Schoon, 2010;Chowdry et al., 2011;Strand 2011Strand , 2014Anders & Micklewright, 2013;Tzanakis, 2014;Khattab, 2015).

Educational aspirations
By the time they reach secondary school age, most pupils have formed meaningful ideas about staying on in education and whether they are likely to apply to university (Anders & Micklewright, 2013;Croll & Attwood, 2013). Past research has examined what individuals aspire to, as well as what they expect to achieve in terms of educational ambition. The literature suggests that, although similar processes can lead to the formation of both, aspirations and expectations are not the same (Gorard et al., 2012), and aspirations, expectations and actual achievement may, or may not align (Khattab, 2014). Aspirations reflect teenagers' hopes or desires to reach a particular level of education and do not necessarily reflect specific socio-economic realities (Gorard et al., 2012). Thus aspirations represent idealistic preferences for the future and tap into underlying values (Bohon et al., 2006). Expectations are more concrete than aspirations since they incorporate individuals' evaluations of the likelihood of fulfilling themexpectations are 'what an individual believes will happen in the future' (Gorard et al., 2012, p. 13). Thus it is argued that expectations are more determined by 'the perceived structure of opportunity within society' (Khattab, 2015, p. 734). Khattab (2015) demonstrates that aspirations and expectations both have a positive effect on attainment, but when combined together they are mutually enforcing. The new UK household panel study only includes a question on aspiration and thus we are only able to focus on the development of educational aspirations.
While there is consistent empirical evidence in support of a positive relationship between educational aspirations and educational outcomes, there is debate as to whether this relationship is causal (Gorard et al., 2012;Baker et al., 2014). This is an important point since if aspirations positively influence attainment then policies to enhance the aspirations and beliefs of educationally marginalised groups would be useful. The status-attainment model sees aspirations and expectations as important links between socio-economic background and educational attainment. The effect of family background is explained by the educational ambitions significant others have of the individual, which they then internalise into personal goals. These goalorientations are motivational resources that influence subsequent attainment (e.g., Blau & Duncan, 1967). However, Bourdieu and others suggest that ambitions actually capture opportunity structures and 'habitus'individuals' conceptions of 'where they fit in the world' rather than motivational forces. It has also been suggested that aspirations for HE may also merely reflect school-level attainment (Croll & Attwood, 2013), highlighting the need to control for prior attainment where possible Educational aspirations among UK young teenagers 731 (Goldenberg et al., 2001). Nevertheless, since there are persistent socio-economic differences in aspirations, and aspirations continue to predict educational outcomes, net of socio-economic background and earlier attainment (Croll & Attwood, 2013;Khattab, 2015), it is important to explore how gender, class and ethnicity intersect in their relationship with aspiration, and how parental and individual attitudes and behaviour may mediate this relationship.

The intersecting roles of gender, class and ethnicity
Prior to the 1980s women tended to have lower levels of educational attainment and were underrepresented in HE. The early 1980s saw gender parity in university entrance, and more recently girls have overtaken boys in both their academic motivation and achievements (Schoon, 2010). Explanations include differences in gender socialization, gender discrimination and innate differences between genders in interests and skillsparticularly essentialist explanations that see boys' masculinity as incongruent with the 'feminised' school setting (Whitehead, 2003;Francis & Skelton 2005;Schoon & Eccles, 2014).
Socio-economic differences in HE participation, although having narrowed in recent years, remain substantialpupils from the highest socio-economic quintile group are around 40 percentage points more likely to go to university than those in the lowest socio-economic quintile group (Crawford and Greaves, 2015). For Coleman (1988), family background encompasses financial capital (e.g., income), human capital (e.g., parental education) and family social capital, defined as the relationships between children and parents. Such social capital 'gives the child access to the adult's human capital and depends both on the physical presence of adults in the family and on the attention given by the adults to the child' (Coleman, 1988, p. 111). In this formulation, social capital is viewed as the mechanism by which teenagers, and their parents' educational aspirations and expectations become aligned (Furstenburg, 2005). Bourdieu (1973) also argued that wealthier families tend to possess greater amounts of symbolic and material resources that they can draw upon to enable them to gain advantages for their members. Parents from higher social-class backgrounds socialise their children in ways compatible with educational success: for example, exposing them to legitimised culture, transmitting positive attitudes towards the value of education, helping them with their school work and engaging with their teachers and school. Recent evidence from the UK demonstrates the independent effect of parental attitudes, involvement and expectations for their children, in encouraging positive educational aspirations, expectations and outcomes Gorard et al., 2012;Khattab, 2015).
In the mid-2000s, school attainment was significantly lower among Black ethnic minority groups and those of Pakistani and Bangladeshi origin (Plewis, 2011). Subsequently there have been rapid increases in attainment, which have been greater for Black, Pakistani and Bangladeshi groups relative to the White group, despite their poorer average socio-economic position. Today, at around age 15/16 attainment is higher for most ethnic groups than for White teenagers, and a higher percentage of minority ethnic youth participate in HE than the White majority (Crawford & Greaves, 2015). It has been argued that children from minority ethnic groups are able 732 A. Berrington et al. to offset their, on average, poorer financial capital through 'family norms, values and networks, as well as a broader set of community values and networks which promote particular educational goals' (Shah et al., 2010(Shah et al., , p. 1112). According to Modood (2004) 'ethnic capitals' can be used to foster high ambitions and to enforce appropriate educational engagement. Evidence in support of ethnic capitals in the UK has previously been found by Strand (2014), who showed greater educational resilience and positive academic self-concept (e.g., completion of homework and lack of truancy), and by Khattab (2015) who showed that parental educational expectations were a key mediating factor in ethnic minority group aspirations and expectations. The relationship between ethnicity and recent immigration to the UK differs for ethnic groups, but the 'immigrant paradigm' thesis (Strand, 2014) also emphasises the selectivity of migrants and their tendency to put greater emphasis on education as a route to social mobility. Crawford and Greaves (2015) find children for whom English is an additional language, although not scoring particularly well at Key Stage 2 (KS2), are particularly likely to participate in HE, consistent with the 'immigrant paradigm thesis'.
It is important to recognise the intersection between these three dimensions (Bhopal, 2014;Strand, 2014). Recent evidence confirms the socio-economic gradient in both school attainment (Strand, 2014) and HE participation (Crawford & Greaves, 2015) is far steeper for White-British pupils, and much attention has been focused on the 'falling behind' of working-class White boys (House of Commons Education Committee, 2014). However, intersections between class and ethnicity are relevant for other ethnic groupsthe experience of working-class Pakistanis in the UK is also shaped, for example, by structural constraints, selective school systems and perceived racialized labour markets (Shah et al., 2010). Gender also shapes the experiences of south Asian men and women in ways different to that of the White majority (Bagguley & Hussain, 2014;Bhopal, 2014). We provide new insight into these issues by answering the following empirical research questions: How do aspirations for college/university relate to gender, class and ethnicity? To what extent are observed class and ethnic differences in aspiration mediated via parental educational attitudes and involvement, the quality of the parent-child relationship, or educational attainment? Does the association between parental background, attitudes and behaviour differ by gender or ethnicity?

Conceptual framework
The developmental-context approach, used extensively in the study of children's educational outcomes (e.g., Marjoribanks, 2002;Schoon, 2010;Chowdry et al., 2011;Strand, 2011;Gutman et al., 2014) accounts for the interplay between structure and agency over time and acknowledges that there are multiple and interlinked influences on individuals' lives (Bronfrenbrenner, 1979;Elder, 1998;Schoon & Eccles, 2014). Within this framework, the relationship between parental background and educational outcomes reflects both a direct relationship, but also an indirect relationship that works through more proximal factors such as parental attitudes and behaviour towards education, and the young person's attitudes and behaviour towards education (Chowdry et al., 2011). We refer to these indirect pathways as mediation. Parental background is a multi-dimensional concept (including parental occupational class, parental Educational aspirations among UK young teenagers 733 education and parental family structure). Parental education is included as a measure of human capital, but also identifies previous experience and knowledge of how HE works. While acknowledging that family structure as measured at a single point in time is a limited indicator of the teenager's experience of family dynamics, we might expect living within a single-parent family to be associated with lower educational aspirations due to other, generally unobserved factors such as downward economic mobility upon parental separation (Thomson & McLanahan, 2012;Hampden-Thompson & Galindo, 2015) and other disruptive life events such as moving from the family home into less expensive accommodation in a poorer neighbourhood (Amato, 2000). These parental-background factors work through (or are mediated by) more proximal transmission mechanisms to influence educational outcomes (Chowdry et al., 2011). These transmission mechanisms are similar to the social and cultural capitals discussed by Coleman (1988) and Bourdieu (1973) and include parental interest in their children's schooling, help with homework and engagement with the school. The quality of the parent-child relationship is also important since, according to Coleman (1988), weak relations between children and parents result in a lack of social capital. This said, the uncritical acceptance of family forms such as lone parenthood as being synonymous with a decline in social capital is problematic (see Edwards, 2004). The developmentalcontext approach assumes that experiences earlier in the life course influence later attitudes and behaviour. Thus, for a subset of the young people, for whom we have linked administrative data available, we additionally include educational attainment at the end of primary school, which has been shown to affect aspirations for college/ university among secondary school-age pupils (Croll, 2010).

Data and methods
The analyses are based on wave one of the United Kingdom Household Longitudinal Survey (UKHLS), an annual panel survey of over 30,000 private UK households, including an ethnic-minority boost for each of the five main ethnic groups: Indian, Pakistani, Bangladeshi, Black Caribbean and Black African (Lynn, 2009). The survey collects information on all household residents, and includes individual adult questionnaires for those over 16 and a youth, self-completion, questionnaire for those aged 10-15 years (McFall, 2012). Our analyses are based on responses of the 4,899 10-15 year-olds who completed the youth questionnaire. The UKHLS provides a unique opportunity to examine how teenagers' educational aspirations (collected within the youth questionnaire) are associated with their own, and their parents' characteristics. Moreover, the survey's large sample size, especially with the ethnic boost, allows exploration of whether these relationships differ by ethnicity. Sample weights are used to account for the complex multi-stage survey design and non-response in wave 1 (Lynn, 2009). Item non-response is dealt with by the inclusion of 'not known' categories as detailed in the description of variables below.

Aspirations for college/university
Responses to two questions are combined to measure aspirations for post-compulsory education. The first question asks 'At the moment, young people can leave school at 734 A. Berrington et al. age 16. What would you most like to do when you are 16?' Respondents could answer: 'get a full-time job', 'study full-time', 'get a job and study', 'do something else', or 'don't know'. All except those who answered 'get a full-time job' are then asked: 'Would you like to go on to do further full-time education at a college/university after you finish school?' The respondents then could answer 'yes', 'no', or 'don't know'. We define a positive intention for college/university education as those who respond 'yes' to the second of these questions. All others, including those who reported that at age 16 they wanted to get a full-time job (and hence were not asked the second question), and those who said 'no' or 'don't know' to the second question are deemed not to have a positive intention for college/university. Overall, two-thirds gave a positive intention for higher-level education. Consistent with Croll (2010) the percentage having a positive aspiration increased with age as they progress through secondary school. The outcome in our regression models is, therefore, dichotomous: whether or not the teenager has a positive aspiration for HE.

Independent variables
Individual characteristics include the teenager's gender, age (continuous in years), country of residence (England, Wales, Scotland, Northern Ireland) and ethnicity. Ethnicity is self-reported as one of 22 categories which have been grouped into: White, Black Caribbean, Black African, Indian, Pakistani, Bangladeshi, Mixed and Other, and Not known. 1 Among UK young adults the Mixed and Other groups are of growing significance, both numerically and in terms of their interest to policy makers. While we include the 'Mixed and Other' group in the analyses, we do not attempt to interpret findings for this group since they represent a wide range of ethnicities and not a clear and coherent group of respondents. We thus focus on the larger and clearly defined ethnic groups.
Parental Occupational Class is coded using the three-category National Statistics Socio-economic Classification (NS-Sec) (ONS, 2010): Managerial and professional; Intermediate, small account and own account workers and lower supervisory; and Semi-routine and routine occupations. Semi-routine and routine jobs best map onto what was once termed working class (Goldthorpe & McKnight 2006). Parental class is based on the highest class of either the mother or the father. For single-parent families, respondents are assigned the occupational class of their co-resident parent. Those living in a family where neither parent is employed, including where parents are unemployed or unable to work, are identified as living in a non-employed family. We are unable to identify class background for around 6% of respondents, largely due to incomplete responses to the adult questionnaire by the parent. Those not living with any parent are also included in this 'not known' group. Parental Education identifies those who have at least one parent with a degree-level qualification. As with occupational class, parental educational level is designated according to whichever parent has the highest level. Parental family structure identifies teenagers co-residing with both natural (or adoptive) parents; those living in a two-parent family where one parent is a step-parent; teenagers living in a single-parent family; and the minority (84 children) living in households containing neither parent (most of whom are living with other family members or foster parents.

Educational aspirations among UK young teenagers 735
Parental educational attitudes are identified by parents' responses to the question 'How important it was that their children completed A-levels (or Highers if in Scotland)'. The derived categories of response are: at least one parent says very important; at least one parent says important; at least one parent thinks it is not important; or not known (when neither parent has responded). The reports from parents on the frequency with which they help their children with their homework is coded as a binary variable parent helps on at least a weekly basis / less than weekly basis. The teenager was asked whether their parents attend parents' evening and whether their parents take an interest in how well they do at school. The responses to both questions are coded as binary: 'Always' and 'Less than always' Quality of parent-child relationship is captured by the teenager's relationship with their mother. We focus on the mother-child dyad since a significant proportion of young people are not living with their father. Frequency of quarrelling with their mother, and frequency with which they talk to their mother about important things is coded as: at least once a week versus less than once a week.
The extent to which the teenager is happy with their school is included to identify any association with school characteristics and educational aspirations. Teenagers rated their happiness with their school on a scale of 1 (very unhappy) to 7 (very happy) entered into the analysis as a continuous variable.
Educational attainment at age 10/11 is based on the combined total score for Maths and English at Key Stage 2, similar to the approach of Plewis (2011). UKHLS has thus far only been linked to the English National Pupil Database (NPD) (University of Essex, 2015) and hence attainment data are only available for a subset of our sample who attended English state schools, whose schools submitted data to the NPD, whose parents gave consent for data linkage, and where successful linkage could be made between UKHLS and the NPD. Consent for linkage was given in 76% of cases a figure lower than that seen for child cohort studies but one that reflects the broader social science nature of the household panel . Examination of consent rates shows that they do not differ systematically by parental class, or parental education but, consistent with previous research (Al Baghal et al., 2014;, they were significantly lower for ethnic minority groups (e.g., Pakistani 69%, Indian, Black Caribbean, and Black Caribbean 72%) as compared to White parents (78%). Therefore, linkage was less common for minority ethnic groups. Thus, the sample for whom attainment data are available is a select one; the reduced sample means that the final model lacks statistical power. Thus, results concerning the mediating role of attainment must be interpreted cautiously. Nevertheless, we believe that the analyses are informative for the White-majority population since we have found no systematic bias in linkage rates according, for example, to parental class, parental education or parental family structure.

Analytical approach
To address the first research question we undertake descriptive analyses of the overall association between ethnicity, class and higher educational aspirations among boys and girls (Table 1). The second research question is addressed first by crosstabulations of parental attitude and behavioural factors according to class and 736 A. Berrington et al. ethnicity (Table 2), and second through multiple logistic regression to estimate the factors associated with the odds of holding a positive aspiration for higher levels of education. By building up the model in a series of steps we can observe how the statistical associations between gender, class and ethnicity are mediated by the inclusion of parental educational attitudes and involvement, the quality of the parent-child relationship, or educational attainment (Table 3). Since educational attainment is only available for the sub-sample of English respondents with linked attainment from the National Pupil database, this analysis is shown separately (Table 4). To answer our third research question, we test the significance of all two-way interactions between gender, class and ethnicity and the remaining covariates in all of the models. Only two interactions are found to be statistically significant: One between gender and parental educational attitudes (Table 3), and one between gender and earlier educational attainment (Table 4). All analyses are weighted and adjusted to account for sample design, non-response and clustering at the primary sample unit level (McFall, Educational aspirations among UK young teenagers 737    Educational aspirations among UK young teenagers 739    Educational aspirations among UK young teenagers 741

Findings
Overall, 66% of 10-15 year olds stated positive aspirations for college/university (Table 1), but boys are significantly less likely to aspire (58%) compared to girls Notes: *=p < 0.10 **=p < 0.05 ***=p < 0.01 Other variables included in the analysis but not shown for reasons of clarity and space are: age; ethnicity; parental family structure; parental help with homework; frequency respondent confides in their mother; frequency respondent quarrels with their mother; respondent's happiness with their school.
Educational aspirations among UK young teenagers 743 (74%). A smaller percentage (66%) of White children hold aspirations for higher levels of education compared to all other ethnic groups. Aspirations are highest for Black-Caribbean (86%), Black-African (81%), Indian (82%) and Bangladeshi (78%) teenagers and the difference between these groups and the White majority are statistically significant. Within all ethnic groups girls are more likely to aspire than boys. However, Table 1 suggests that the gender difference is largest among White teenagers (17 percentage points, than for minority ethnic groups where the gap is smaller (between 4 and 8 percentage points). Indian boys appear more likely to have higher educational aspiration (79%) than Pakistani (70%) and Bangladeshi (75%), but the difference is not statistically significant due to the relatively small sample sizes. All boys in minority ethnic groups report higher educational aspirations than their White male peers and most tend to have higher aspirations than White girls. Aspirational differences according to parental occupational class are similarly large, especially for boys ( Figure 1). For boys, positive aspirations are much higher among those from managerial and professional backgrounds (67%) compared to those with intermediate (57%) or routine class backgrounds (50%). Among girls, those from the most advantaged class backgrounds are also more likely to hold high aspirations (80%), but the difference between this group and those in intermediate occupations (74%) is somewhat smaller. Furthermore, 69% of girls from routine backgrounds aspire to attend college/universitysubstantially more than the one-half of boys from Respondents in households where neither parent was employed have the lowest ambitions for college/university (48% boys and 63% girls). Next we examine the association between ethnicity and parental class and the factors hypothesized to mediate their relationship with aspirations for further and higher education (Table 2). While teenagers in our sample from the Indian ethnic minority were similarly likely to have at least one parent with a degree-level qualification as their White counterparts (around one-half); the percentage is a little lower among Black teenagers in our sample (44%) and significantly lower among those of Pakistani (29%) and Bangladeshi origin (10%). Unsurprisingly, differences in parental experience of HE according to class background are very largewith three-quarters of teenagers from professional and managerial backgrounds having at least one parent with a degree, as compared to two-thirds of those from an intermediate class background and one-half of those from a working-class background. Just one in five children living in non-employed families had at least one parent with experience of HE.
Past research has emphasized the more positive educational attitudes and engagement with school among ethnic minorities as an explanation for their greater educational resilience in the context of (often) lower socio-economic resources (Modood, 2004;Shah et al., 2010;Naidoo 2015). The answers provided both by parents and by the teenagers suggest there may be significant ethnic differences, and that differences according to ethnicity appear, for a number of the indicators, to be larger than differences according to class background. The biggest difference is seen in the response of parents as to the importance of A-levels (and Highers). While between 82% and 92% of Black and south Asian minority parents thought that A-levels were very important, only three-quarters of White parents did so. Black, Indian and Pakistani teenagers were also more likely than White teenagers to report that their parents were always interested in how well they were doing at school. Moreover over 90% of Indian and Pakistani teenagers reported that their parent(s) always attended school parents' evenings, with the proportion being a little lower for the remaining ethnic groups and White majority. No consistent ethnic difference is seen in the frequency with which parents help their children with homework.
Educational attitudes and involvement also differ by parental social class, but it appears that the main difference lies between those from a managerial and professional background and all other occupational class backgroundsmirroring to some extent the distribution of parental experience in HE. For example, 82% of parents from a managerial or professional-class background said that A-levels were very important compared to 72% of those in intermediate occupations, 71% of those in semi-routine and routine occupations, and 74% for those not in employment. Similar patterns are found for parental help with homework and parental interest in the child's schoolwork. Attendance at parents' evenings shows a strong linear relationship with class background such that 92% of children from a professional and managerial background said their parents always attended as compared to 85% of those from intermediate backgrounds, 82% of those from working-class backgrounds and 69% of those living in workless families.
The data in Table 2 show very large differences in family-structure distribution by ethnicity and by social class. Consistent with past research findings (Shaw, 2004), less than one-third of teenagers of Black-Caribbean origin are living with two natural Educational aspirations among UK young teenagers 745 parents, this rises to 60% of White teenagers and between 78-87% of south Asian teenagers. Cross-cutting these ethnic differences is a steep class gradient in the likelihood of living with two natural parents: from 76% of teenagers from professional and managerial backgrounds, to 23% of teenagers living in workless households. The frequency of quarrelling between teenagers and mothers is consistently lower among Black and south Asian teenagers, while a higher proportion of these groups report that they discuss important matters with their mother. Teenagers from professional and managerial backgrounds also report lower likelihoods of frequent quarrelling, but there are inconsistent class patterns in reported levels of confiding among all teenagers. In terms of predicting educational aspirations, these descriptive findings are consistent with the notion that the lower socio-economic resources of minority ethnic groups (especially the lack of parental education among Pakistani and Bangladeshi parents) may be offset to a certain extent by a higher emphasis on the importance of academic qualifications, an increased engagement in their children's schooling, increased family resources as measured by having two natural parents co-resident, and having higher quality parent-child relations. Moreover, these descriptive findings suggest that the observed lower educational aspirations of White children, and especially those from working-class backgrounds, might be mediated via less parental experience of higher education and lower engagement with school and homework. Table 3 shows the odds ratios for having a positive aspiration to attend college/university. Model 1 controls for the young person's characteristics: gender, country of residence, age and ethnic group. Model 2 additionally includes measures of parental socio-economic status (SES): occupational class, parental education and parental family structure. Model 3 includes parental attitudes towards and involvement in the teenager's schooling. Model 4 includes measures of the quality of the parent-child relationship, while model 5 includes the level of happiness that the child feels about their school. We then tested for all two-way interactions between gender and the remaining covariates, and between ethnicity and the remaining covariates. To gain more statistical power, we grouped the Black groups and the south Asian groups in order to test for these interactions. However, only one interaction was found to be significant at the 5% levelparental educational attitudes and gender. (It may be that other interactions exist but require a larger sample size to be statistically significant.) Model 6 in Table 3 presents this interaction model. The analyses containing attainment are based on a reduced sample, where the number of minority ethnic teenagers is too small to make inference for specific ethnic groups. However, the model can be interpreted as representing the associations found for the majority White population. Table 4 shows, for this reduced sample, the key covariates of interest that are significantly associated with aspiration for college/university.
When just the individuals' own characteristics are included (Model 1), the odds of holding positive aspirations for college/university increase with age, and are consistently lower for teenagers in Northern Ireland as compared to the rest of the UK. Boys are half as likely to aspire (odds ratio 0.48) compared to girls. Indian, Black-Caribbean and Black-African teenagers are more likely to have positive aspirations as compared to White teenagers. Pakistani teenagers are significantly less likely to report a positive aspiration as compared to Indians (odds ratio 0.56, p < 0.10). Model 2 shows a very significant, steep linear, social-class gradient in the likelihood of holding positive aspirations for college/university, which is independent of an additional strong positive association between having at least one parent with a degree and holding positive aspirations for college/university. Children from intermediate-class backgrounds have a significantly lower likelihood of aspiration (odds ratio 0.80) than those from managerial/professional-class backgrounds, but have a higher likelihood of aspiration as compared to those from semi-routine or routine backgrounds (odds ratio 0.59) and children who are living in a household with no working parent (odds ratio 0.50). 2 Although we found a bivariate association between living in a loneparent family or step-parent family and lower educational aspirations (results available on request), once parental occupational class and parental education are controlled, teenagers living in single-parent families and those living in a step-parent family are no less likely to aspire to college/university, although the small number living with neither natural parent are less likely to aspire.
When parental socio-economic status and family structure are controlled, the difference in the likelihood of aspiring to go to college/university between Indian and Pakistani teenagers narrows and is no longer statistically significant (odds ratio 0.68), whilst the odds ratio for Bangladeshi teenagers is actually greater than 1 suggesting that their lower observed propensities to aspire to college/university are associated primarily with having parents in lower occupational jobs and having parents with less HE experience. Once these differences in parental financial and human capital are controlled, young Bangladeshi's aspirations are as similarly high as young Indians.
Findings from Model 3 are consistent with previous research suggesting that parental educational attitudes and involvement in school are positively correlated with teenagers' aspirations and that these may act as 'transmission mechanisms' through which parental socio-economic status is associated with aspirations and outcomes. When these mediating variables are included the odds ratios for parental occupational class and parental education move closer to one (for example aspirations among teenagers from intermediate-class backgrounds are no longer significantly different to those from managerial and professional backgrounds). Furthermore, since parental educational attitudes and investment are found to be less supportive among the White population than Black and south Asian groups, (as discussed in relation to Table 2), when we enter parental educational attitudes and investment into the model, odds ratios for ethnicity for the White group moves slightly closer to one.
The association with aspirations is stronger for parental educational attitudes than for parental educational investment in their child's education. Net of other variables, teenagers whose parents reported 'A-levels/Highers qualifications were not important' were less than half as likely to report an aspiration for college/university. The odds of aspiring are significantly lower for teenagers who reported that their parents are not always interested in their school work (odds ratio 0.56, p < 0.01), and teenagers who report that their parents do not always attend parents evening (odds ratio 0.71, p < 0.01). Help with homework (as reported by parents) was not associated with aspirations. Since parental attitudes towards education differ according to ethnicity, when we enter parental attitudes into the model, odd ratios for ethnicity for the White group become closer to one. The quality of the parent-child relationship may also be positively associated with educational aspirations, with teenagers who quarrel less and who confide more in their mothers being more likely to hold positive Educational aspirations among UK young teenagers 747 aspirations for college/university (p < 0.01) (Model 4). Whether or not the teenager likes their school is also significantly (p < 0.01) associated with holding positive aspiration for college/university (Model 5) and the inclusion of this additional variable does not weaken any of the previously discussed associations suggesting that this may be capturing a different dimension.
Model 6 includes a statistically significant interaction between gender and parental educational attitudes. Compared to the reference category where at least one parent thinks A-levels/Highers are 'very important', the odds ratio for aspiring for higherlevel education is 0.60 for girls whose parents deem completion of this educational phase as 'not important'. For boys however, the odds are only 0.32 (0.53*0.60). Hence parental educational attitudes may be more important for boys in encouraging aspirations for college/university. Significant gender, class and ethnic differences in aspirations remain even when we have controlled for the observed differences in reported attitudes and behaviours. Table 4 shows results for the English sub-sample who have linked attainment data. The coefficients from Model 7 are those based on the final main effects model without attainment, while Model 8 additionally includes the KS2 exam score quartile. We find a significant interaction between attainment and gender, such that prior attainment is strongly positively associated with subsequent aspirations for boys, but less so for girlsa result that complements recent findings based on other UK cohorts (Schoon, 2010). The odds of holding a positive aspiration are 3.63 (1.33*2.73) higher for boys who scored in the highest quartile in the KS2 tests compared to those who scored in the lowest quartile. For girls, the odds were 1.33 higher. Including attainment into the analyses has an impact on the existing covariates in the model (compare Model 7 and Model 8 in Table 4). For example, the odds ratios for parental class and education are reduced (become closer to one), suggesting that by age 10/ 11 there are already large differences in attainment by parental SES. The coefficients associated with the other transmission mechanismsparental educational attitudes and investmentalso move slightly closer to one. We conclude that educational attainment is an important mediator of aspirations later on in the teenage years and that class differences in educational aspiration are (to a limited extent) mediated by attainment at age 10/11. However, there are still large and significant differences in aspiration according to parental class even when attainment is controlled. Net of other factors, the odds of having a positive aspiration for teenagers from semi-routine and routine class backgrounds are 0.59 those of a teenager from a professional and managerial background. The odds ratio is similarly low for teenagers from workless households (0.53).

Discussion
This paper both updates and extends existing research. The UK context in terms of the ethnic composition of the youth population, the increased diversification of the HE student body and an enhanced normalisation of extended educational trajectories is changing. The UKHLS provides the latest data source with sufficiently large sample size and rich parental and individual data in order to examine how ethnicity intersects with class and gender in their association with educational aspirations. By 748 A. Berrington et al. undertaking multiple regression analyses we have tested for intersectionality between these three dimensions. We also have tested whether the mediating pathways through which parental background factors are associated with educational aspiration differ according to gender and ethnicity, finding some support for the idea that the mechanisms do differ by gender.
Our conclusion is that, contrary to much political rhetoric, young people's aspirations for college/university participation remain high: just about two-thirds of young people aged 10-15 hold this aspirationa larger proportion than actually participate. The idea that there is a 'culture of low aspirations among young people', should therefore be questioned (Gorard et al., 2012;Rose & Baird, 2013;St. Clair et al., 2013;Baker et al., 2014). The objective of government should be to develop a greater understanding of the processes that delimit the achievement of aspirations. Furthermore, raising educational aspiration further still may be problematic from an employability perspective because the economy cannot accommodate more graduates without corresponding demand (St. Clair et al., 2013).
Despite overall high levels of aspiration, there are significant differences by gender, ethnicity and class, with higher aspirations found among girls, teenagers from professional and managerial backgrounds and teenagers from non-White ethnic groups. In contrast to earlier research which highlighted low aspirations of Black-Caribbean boys (Francis & Skelton, 2005), Black-Caribbean and Black-African teenagers born in the late 1990s and early 2000s have much higher levels of aspiration as compared to White teenagers, and Black-African teenagers have similar levels of aspiration to Indian teenagers. Thus in this sample, it is the White teenagers, who are falling behind. Given the high rates of Black-Caribbean /White unions, teenagers of Caribbean origin will increasingly identify themselves as mixed Black/White heritage (Simpson, 2015) suggesting that comparisons of ethnic differences over time may be misleading. Nevertheless our findings correspond with research noting similar trends of increasing HE participation among Black-Caribbean youth (Crawford & Greaves 2015).
The size of the differential according to ethnicity is of a similar magnitude to that across gender and class: there is around a 16 absolute percentage point difference in the likelihood of holding a positive aspiration according to gender, similar to the difference between Indian and White teenagers, and the difference between male teenagers from professional backgrounds and male teenagers from working-class backgrounds. We conclude that all three dimensions are important. However, the multiple regression analyses find no evidence of an interaction effect between gender and class, or between gender and ethnicity, or between class and ethnicitythat is to say the effect of being male, white and working class appears to work in an additive way to affect the change of aspiring to college/university. In other words the impact of occupational class upon educational aspirations is similar for all ethnic groups corresponding to previous findings (e.g., Conolly, 2006;Rothon, 2007). White boys from the lowest occupational class and from workless households have the lowest aspirations (50% and 48% have a positive aspiration for college/university), this is because the three elementsbeing White, male, and working classcombine in an additive fashion to encourage lower aspiration. Our findings provide support then for the UK Government's concern about low rates of application to HE from this group.
Educational aspirations among UK young teenagers 749 However, even though this figure is low, it is higher than the percentage of workingclass boys who go to university (Anders & Micklewright, 2013;Independent Commission on Fees, 2014). Thus focusing on aspirations alone will not on its own reduce ethnic differences in HE participation.
Once parental class and parental education are controlled, differences in educational aspiration by family type become smaller for the step-parent and no-parent group as compared to those living with two natural parents, but the association changes direction for those from single-parent families. In other words, all things being equal, children living with a single parent in the UKHLS sample have the highest odds of aspiring. This important finding, consistent with Hampden-Thompson and Galindo (2015), undermines persistent discourses that position single-parent families as morally reprehensible and damaging for children (McRobbie 2007;Allen & Osgood 2009). On average, lone parents tend to have lower financial and educational resources, and it is this which is associated with lower educational aspirations. As noted by Edwards (2004) household composition measured at a single point in time tells us nothing about the dymanics and relationships with extended family members and other persons outside the immediate family, which are also likely to have important influences on young people's attitudes and behaviour.
Class and ethnic differences in aspiration tend to work through more proximate mechanisms to influence aspiration (Chowdry et al., 2011). Our analyses have shown that parental attitudes are positively associated with their offspring's own aspiration over and above the effect of parental class or educational level, but that the statistical association of class and educational family background was weakened with their inclusion. We also find associations supporting the notion that parental involvement in their children's education can benefit children's aspirations. Children's reports of the frequency with which their parent(s) attend parents evening and the extent to which they felt their parents were interested in their school work were associated with more positive aspirations, although parental reports of the extent to which they helped their children with homework had a less consistent association. The different findings could relate to the fact that help with homework is reported by the parent, whereas parental engagement with school and parental interest in their children's schoolwork was reported by the teenager themselves. For the English, predominantly White, sub-sample, prior educational attainment is an important mediating factor between family background and aspirations. Thus it is important to control for prior attainment wherever possible. However, we also find that the inclusion of prior attainment did not significantly alter our substantive conclusions regarding the other mediating mechanisms.
We have found evidence consistent with an ethnic-capital effect (Modood, 2004;Modood & Khattab, 2015): parents of minority ethnic children had more positive educational aspirations, and were more likely to be reported by their children as interested in their school work and engaging with the school. Nevertheless, even when we control for these differences in parental attitude and behaviour, White teenagers continued to have lower educational aspirations suggesting that there are other, unobserved differences between ethnic minority groups not captured in the model. Future research is required to investigate influences outside the home, for example, from peers, or school [found to be important by Anders and Micklewright (2013)], or 750 A. Berrington et al. information about the wider social networks of minority ethnic groups that emphasise the importance of education that Modood (2004) and Feliciano and Lanuza (2015) suggest may help explain the higher aspirations of minority ethnic groups. There were two areas where the mediating mechanisms differed according to gender, class or ethnicitythey both involve gender: The positive effect of parental attitudes and of prior educational attainment was significantly more important for boys than for girls. These gender differences, consistent with other work (e.g., Whitehead, 2003;Schoon, 2010), require further research to understand why boys' aspirations can be more negatively affected by past poor academic performance, or why boys whose parents have less positive attitudes towards education have particularly low aspiration. This is important if policy interventions are to be implemented appropriately and tailored by gender.
While we have addressed important questions around educational aspirations in a much changed context, our study has some limitations. First, even though the UKHLS is by far the largest household panel survey ever conducted for the UK and has an ethnic boost, the sample sizes among certain minority ethnic groups of teenagers was relatively small. Second, the survey only collected information on aspirations, not expectations. We acknowledge that aspirations do not always translate into expectations or participation and urge the designers of the UKHLS to include additional questions in the subsequent sweeps of the youth panel on educational expectations, attainment and potential barriers to post-compulsory participation. The survey would then provide even more evidence to inform current government initiatives to reduce educational inequality and enhance social mobility. backgrounds and have lower educational aspirations. These teenagers are retained in the analysis because they contribute information on the other covariates. However, we do not attempt to interpret the coefficients for this 'not known' group. Sensitivity analyses showed that the findings from the statistical-modelling results are unchanged whether or not this 'not known' group are included.