Socio-economic gaps in educational aspirations: do experiences and attitudes matter?

ABSTRACT

We use detailed survey data linked to administrative records from secondary schools in England to investigate potential channels contributing to the socio-economic gap in post-compulsory educational aspirations. We investigate the role of experiences and attitudes including the provision of information, advice and guidance (IAG), bullying victimisation, locus of control and self-perception of academic potential. Our findings indicate a significant socio-economic gap in aspirations to stay in education, to follow the academic rather than the vocational route, and to attend university. We use decomposition analysis to show that the experiences we consider are not statistically correlated with the observed socio-economic gap while differences in attitudes explain up to 22% of the effect. The findings suggest that investing in self-esteem building and attribution training programmes within schools could contribute to equalising educational outcomes.

KEYWORDS:

• Educational aspirations
• socio-economic inequality
• experiences
• attitudes

JEL CLASSIFICATION:

• I21
• I24
• I26

1. Introduction

Educational aspirations are the goals that individuals set for themselves for the future, constituting the basis of investment in human capital and contributing to educational and labour market success (Erberber et al. 2015). Several recent studies reveal that low socio-economic status (hereafter SES) students are less likely to set high educational aspirations even when their academic potential would allow them to do so (Gutman and Akerman 2008; Hoxby and Turner 2015; Berrington, Roberts, and Tammes 2016; Anders 2017; Guyon and Huillery 2020). This behaviour is often attributed to external or internal constraints that could be influencing the decision-making process of students from disadvantaged SES backgrounds (Dalton, Ghosal, and Mani 2016). It is, therefore, important to understand whether inequalities in aspirations reflect differences in the way individuals set their goals or whether low SES students maximise their utility by aspiring to lower educational pathways given, for example, the experiences they face or differences in their attitudes.

In this paper, we study the factors associated with SES inequalities in post-compulsory educational aspirations by exploring potential channels that could be contributing to any observed SES gap. We apply Gelbach (2016)'s conditional decomposition to understand the association between SES and student experiences, including information, advice and guidance (IAG) and bullying victimisation, and attitudes including having an external locus of control and low self-perception of academic potential. To do so, we use data from the Longitudinal Study of Young People in England (LSYPE) linked to administrative records from schools. The linkage of the two datasets provides rich information on prior attainment and secondary school attended, and also allows us to observe the educational and professional trajectories of the individuals over a long period (from age 11 until age 25).

We find a significant SES gap in aspirations after accounting for important background factors and secondary school attended. We demonstrate that IAG is positively associated with aspiration formation, bullying victimisation is negatively associated only with aspirations to follow the academic rather than the vocational post-compulsory route, and having an external locus of control and a low self-perception of academic potential have a strong negative statistical association with educational aspirations. The decomposition analysis reveals that the experiences we consider are not associated with the SES gap in aspirations while differences in young people's locus of control and self-perception of academic potential explain up to 22% of the effect.

The paper provides three main contributions to the empirical literature that looks at the SES gap in aspiration formation. The majority of previous analyses focus on aspirations for university participation. Our first contribution is to consider aspirations of different kinds in a multi-dimensional perspective. In addition to university aspirations, we look at aspirations of whether to stay in education after completing the compulsory participation level at age 16 and, for those individuals aspiring to stay, aspirations of whether to follow an academic or a vocational pathway.

Second, we draw from the theoretical and empirical literature that associates SES with background characteristics to develop an empirical model that disentangles their contribution on the gap in aspiration formation, particularly focusing on individual experiences and attitudes that have been previously identified to be important for educational outcomes. Berrington, Roberts, and Tammes (2016) showed that there are large SES differences in prior attainment which act as a key mechanism for aspiration inequalities. Furthermore, differences in parental attitudes towards education, the level of parental engagement with schoolwork, the quality of parent–child relationship as well as disparities in parental investments, such as in private tutoring classes, are all important mediators (Smyth 2009; Berrington, Roberts, and Tammes 2016; Lazarides et al. 2016). Hoxby and Turner (2015) used data from the US, to show that the lack of information is a reason why low-income high-achieving students set lower aspirations than similarly achieving high-income students. Bettinger et al. (2012) suggested that combined assistance and information provision has a positive effect on college attendance, persistence, and aid receipt. The propensity to be bullied is found to have little difference by SES but to have important long-run effects on the educational outcomes of victims (Gorman et al. 2020). Finally, Guyon and Huillery (2020) identified that being ignorant of existing educational pathways and underestimating academic ability were the main drivers of social differences in aspirations between equally achieving classmates in France. We follow the approach of Guyon and Huillery (2020) and carry out a within-school investigation of the potential channels to take into consideration the decisive role of schools in shaping aspirations (Dupriez et al. 2012).

Our final contribution is to show evidence that student aspirations are highly correlated with educational progress. Indeed, from an economic perspective, educational aspirations would not really matter if they did not influence future outcomes. However, among the empirical literature, it is well established that educational aspirations play a significant role on post-compulsory education participation of young individuals and subsequently their labour market success. Within the English context, McCulloch (2017) identified that students with high aspirations made more progress in achievement and had a higher likelihood of attending university than those with low aspirations, all else equal. Khattab (2015) indicated that complete alignment between high aspirations and high achievement is the most important predictor of future educational behaviour. Recent quasi-experimental evidence from England showed that policies aiming to raise aspirations alone will not be sufficient to increase university enrolment of low SES students, if not coupled with policies to enhance young individuals' preparation for it (Rizzica 2020). We add to this literature by using high-quality data that allow us to take into account many important background factors and examine the relationship between aspirations and performance in high-stake examinations at the end of compulsory school, university participation and earnings and employment probability at age 25.

The remainder of this paper proceeds as follows. The next section provides institutional background information to contextualise and motivate the study. Section 3 explains the empirical strategy and Section 4 describes the data. Section 5 presents the main results and, finally, Section 6 concludes.

2. Institutional background and motivation

In England, the compulsory school curriculum is organised in blocks of years called Key Stages (KS). In the primary phase, students enter school at age 4–5 in the Foundation Stage, then move to KS1 at age 5–7. At the age of 7–8, students move to KS2 until the age of 10–11 where they leave the primary phase and go on to the secondary school of their choice, where they progress through KS3 until age 13–14. At the end of KS2 and KS3, students are assessed in three core modules, Maths, Science and English. Student attainment is recorded in terms of the scores achieved in the externally marked assessments, as part of the National Curriculum.¹ The students progress through KS4, at age 15–16, wherethey take the General Certificate of Secondary Education (GCSE) which used to coincide with the end of compulsory school. The GCSEs are high-stake public examinations taken by all school students usually in 9 or 10 subjects. Usually, 5 GCSEs at grades A*-C or equivalent is the required threshold to continue to the next educational level. Recently, the compulsory participation age rose to 18 but the cohort analysed in this study turned 16 before the policy change, and therefore, has not been affected.

Following the national GCSE exams, students enter post-compulsory education at KS5 which usually lasts between two and three years and is received in the form of upper-secondary academic or vocational qualifications. Finally, students have the option to attend higher education which is considered the optional final stage of ‘formal learning’ and is achieved through academic degrees or advanced vocational qualifications.

The literature generally finds big and statistically significant labour market returns to academic qualifications while, on the other hand, very low returns to vocational qualifications (McIntosh 2006; Dearden, Machin, and Vignoles 2009; McIntosh and Morris 2016, 2018). Hupkau et al. (2017) show that students who follow the academic path tend to have higher prior attainment and to come from families with higher SES backgrounds while at the same time, progression routes for students who do not undertake academic qualifications are not as well known, understood or preferred by students and employers alike. While the evidence on SES inequality in education is undoubted, any suggested policy implications should be highly dependent on the characteristics of the transmission mechanisms. For example, policy implications should take into consideration if the SES inequality stems from financial pressures (for example inability of buying adequate educational materials) or from parental investments, help and support (for example lack of sufficient parents' skills to help their children) (Chevalier et al. 2013). Also, it is important to understand when SES inequality starts producing its effects on educational outcomes, for example, in primary or secondary school.

In this paper, we investigate the factors mediating SES inequalities in aspirations, and in particular, we look at individual experiences andattitudes including IAG, bullying victimisation, locus of control and self-perception of academic potential (see Section 4 for details). The available data do not allow us to infer causality about these mechanisms contributing to the SES gap. Nonetheless, describing the association between experiences, attitudes, SES and subsequent outcomes sheds light on important factors to be considered for supporting students in their academic journey.

3. Empirical strategy

3.1. SES gap in aspirations

We start our analysis by investigating the relationship between SES and educational aspirations using the following equation: (1) $A_{i,s}={\text{β}}_0+{\sum }_{j=2}^3{\text{β}}_j\left(SE{S}_{i,s=j}\right)+{\delta }_{}{Z}_{i,s}+{\gamma }_{}‘X_{i,s}^{\prime }+{\eta }_s+{ϵ}_{i,s}$(1) where $A_{i,s}$ are the educational aspirations of individual i in secondary school s. $SE{S}_{i,s=j}$ is the SES tertile each individual belongs to, $‘Z_{i,s}^{\prime }$ represents prior attainment, $‘X_{i,s}^{\prime }$ is a vector of individual characteristics and family composition and ${\eta }_s$ are the secondary school fixed effects. Finally, ${ϵ}_{i,s}$ is a robust standard error clustered at the secondary school level.

We estimate the above equation sequentially, first presenting the raw SES gap in aspirations and then we proceed to examine if any identified effect persists when individual, family and school characteristics are taken in consideration.

3.2. Factors associated with the SES gap in aspirations

An influential literature on poverty traps argues that persistent poverty is driven by constraints that are either internal or external to the individual. Our primary aim is to test whether any identified SES gap in aspirations is statistically associated with such constraints. We do so by augmenting Equation (1(1) $A_{i,s}={\text{β}}_0+{\sum }_{j=2}^3{\text{β}}_j\left(SE{S}_{i,s=j}\right)+{\delta }_{}{Z}_{i,s}+{\gamma }_{}‘X_{i,s}^{\prime }+{\eta }_s+{ϵ}_{i,s}$(1) ) with the experiences and attitudes we consider, described in detail in the following section. The equation we estimate is the following: (2) $A_{i,s}={\text{β}}_0+{\sum }_{j=2}^3{\text{β}}_j\left(SE{S}_{i,s=j}\right)+{\delta }_{}{Z}_{i,s}+{\gamma }_{}‘X_{i,s}^{\prime }+{\theta }_{}‘C_{i,s}^{\prime }+{\eta }_s+{ϵ}_{i,s}$(2) where $‘C_{i,s}^{\prime }$ is a vector of experiences and attitudes including IAG , bullying victimisation, locus of control and self-perception of academic potential.

Our underlying assumption is that experiences and attitudes act as mediators of the SES gap in educational aspirations. By simply observing changes in the SES gap when adding experiences and attitudes in the model might produce results that are biased based on the order in which each covariate (or group of covariates) is added, if these are correlated with each other. To overcome this problem, we follow the procedure suggested by Gelbach (2016) to explore the statistical association between experiences and attitudes and the SES gap in aspirations. We first estimate the baseline model which includes the raw SES gap in aspirations. Then, we estimate the full model which includes all additional covariates, including the experiences and attitudes that are the focus of this paper. The difference in the coefficients of the SES gap from the baseline and the full model represents the total change.

Each variable's contribution to the total change can be computed as the multiplication of two coefficients. The first is the coefficient of each covariate in the full model. The second is the coefficient of SES, in a regression of SES on the respective variable. The sum of these multiplied coefficients over all variables equals the total change. For a more intuitive interpretation of our results, we consider five potential channels that SES could be associated with aspirations and group our covariates accordingly. These include: (1) prior attainment, (2) demographics and family composition, (3) secondary school attended, (4) experiences and (5) attitudes.

It is important to note that our results do not allow for a causal interpretation. Educational aspirations can be influenced by a myriad of factors and, therefore, our measures of experiences and attitudes might not fully capture the extent to which these mechanisms may matter in their formation. It is possible that part of the estimated effect captures other unobserved characteristics of the individual, different from those included in our model. As discussed in previous analyses following a similar approach, we cannot definitely eliminate omitted variable bias arising from unobservable factors by simply including additional observed variables and, in the absence of a randomised experiment, there is a limit to how far we can go in establishing causal relationships (Vignoles, De Coulon, and Marcenaro-Gutierrez 2011). At the same time, the variety of background factors included in the empirical analysis allows for a detailed description of the phenomenon which can help in deriving meaningful policy implications. In this way, the study aims to identify potential mediators of the SES gap in aspirations by quantifying the contribution of experiences and attitudes to the overall association of SES with aspirations.

4. Data

4.1. Description of the dataset

The study makes use of a very rich dataset, the Longitudinal Study of Young People in England (LSYPE). The LSYPE follows the lives of 15,770 young people born between 1989 and 1990, providing detailed information on the key factors affecting their educational progress and their transition following the end of compulsory school. The survey began in 2004 when young people were in Year 9 (age 13–14) of compulsory education. Following the initial survey, the cohort members were surveyed every year until 2010, when they were aged 19–20 and for a final interview in 2015, when aged 25. In addition to the young people themselves, parents or guardians were also interviewed in the first four waves of the survey in order to acquire a definite view of their households and their corresponding SES background.

The records of the cohort members were linked to the National Pupil Database (NPD), which is a longitudinal administrative dataset recording all school and college students in England. This link of the two datasets was particularly important for the needs of this study as we were able to gather records of the students' prior test scores alongside with information on the secondary school attended. The richness of the datasets allowed the exploitation of various factors, both at the individual and school level, that are associated with educational outcomes and eventually with aspirations. Table 1 presents the educational progression of the cohort.

Table 1. Education progression of the cohort.

Academic Year	Age	Event
1989–1990	0	Born
2000–2001	10–11	KS2 exams
2003–2004	13–14	KS3 exams
2005–2006	15–16	KS4 exams (GCSEs)
2007–2008	17–18	KS5 exams (post-compulsory education)
2008–2009	18–19	KS5 exams for continuing students
2008+	18+	Higher Education

4.2. Analytical sample

From the initial cohort of 15,700 young people, we considered only those participating in Wave 2 of the survey (N = 13, 539), when personality traits were measured for the first time. Other Waves of the survey were used to construct variables such as future outcomes (Wave 8), time-invariant characteristics (Wave 1) or to complete missing information. Given the requirement of very detailed information for the needs of this paper, various restrictions were adopted to get the final analytical sample resulting in a significant reduction of the sample size. Specifically, we have non-missing information on aspirations, SES background, prior attainment, secondary school attended, individual characteristics, family composition and experiences and attitudes for 8335 students (62% of Wave 2 participants).

The sample deduction is mainly due to missing information on the questions we used to construct the measures of individual attitudes. Since these measures are key to the purposes of this study, Appendix A provides the t-statistics for mean differences between the initial and analytical sample for all the main variables used in the analysis, as an indication of the representativeness of the sample. In some cases, the differences between the total and analytical sample are statistically significant but usually small in size. Overall, the issue of sample selection does not seem to be a cause for major concern as we do not observe any important differences in attitudes were most missing information has been observed and in our main variables of interest, which capture young people's educational aspirations.

4.3. Variables of interest

4.3.1. Educational aspirations

In the current literature on educational aspirations, there is no consensus with regards to the definition of the specific construct and especially with regards to what distinguishes aspirations from expectations. Marjoribanks (1997) described educational expectations as realistic aspirations that, unlike idealistic aspirations, are an outcome of self-reflection on social and economic opportunities available. While distinguishing between the two is beyond the scope of our analysis, throughout this paper, we adopt the definition of ‘educational aspirations’, as the information used to construct our variables is derived from questions where the respondents indicate their plans and hopes for post-compulsory education. These questions have been used by many scholars as a measure of future aspirations (e.g. Strand and Winston 2008; Gutman and Schoon 2012; Rose and Baird 2013; Khattab 2014, 2015).

We consider three types of educational aspirations to better reflect on students' plans and hopes. We gather this information from self-reported responses at age 15, that is the second wave of the survey.² First, we look at aspirations to stay in full-time education after completing the compulsory level at age 16. Those who expressed intentions to stay in education were asked where do they want to go, choosing between a sixth-form school or a college of further education. In England, while you can still study for academic qualifications at further education colleges, the courses offered tend to be more technical compared to school-based qualifications which tend to offer a range of more academic subjects (usually A-levels). We classify students who mentioned a sixth form as aspiring to an academic route and those mentioning a further education college as aspiring to a vocational route. Finally, the university aspirations of students have been constructed based on how likely the students were to apply to university. Those who responded that it is ‘very likely’ or ‘fairly likely’ were classified as aspiring to go to university while those who mentioned ‘not very likely’ or ‘not at all likely’ were classified as not aspiring to go to university.

4.3.2. Socio-economic background

The main variable of interest in this analysis is the SES group the student belongs to. Based on the idea that SES is a multi-dimensional construct and taking advantage of the richness of our dataset, we construct an SES index that combines several measures of family background including: (i) both parents' occupation, (ii) both parents' education, (iii) housing tenure (owned vs rented) and (iv) Free School Meal Eligibility.

We combine these measures using principal components analysis (PCA) to create a continuous SES index using the whole sample for whom information on these indicators was available (N = 11, 273).³ Therefore, throughout this study SES refers to the socio-economic position of a student relative to the whole LSYPE cohort. To facilitate the interpretation of the results across the whole distribution of SES, we divide the sample into SES tertiles on the basis of this index.

4.3.3. Measures of prior attainment

A key determinant of educational aspirations is prior attainment capturing a student's observable academic potential. We measure prior attainment using formal, externally marked, school tests taken at the end of primary school (KS2), at age 11, in three core modules, Maths, Science and English. We use average KS2 scores in standardised form, with a mean of 0 and a standard deviation of 1. The standardisation of these test scores reflects the ability position of each individual in the analytical sample compared to the total sample of students participating in the LSYPE. We also have available school tests (also externally marked) taken at age 14 (KS3) in Maths, Science and English, which we consider as a robustness check.

4.3.4. Other individual, family and school characteristics

We control for basic individual characteristics including gender, ethnicity, an indicator of special educational needs (SEN) and for family composition, including number of siblings and whether it is a single-parent household. We also apply secondary school fixed effects to account for unobserved school quality and to ensure that our estimates do not suffer from selection bias inherent in the school system. As a result, our contribution can be considered a ‘within-school’ analysis of aspirations and does not consider how different schools can shape such outcomes – something to be left for future research.

4.3.5. Experiences and attitudes

Holding prior attainment and background characteristics constant, our underlying assumption is that educational aspirations could still vary endogenously within schools, across SES backgrounds. We consider two potential channels, namely experiences and attitudes. Experiences are defined as parameters that are likely to be exogenous to individuals but could affect the formation of their aspirations. On the other hand, attitudes are related to characteristics, personality traits or interests that are potentially endogenous to each individual and could also be related to educational aspirations. Such attitudes are suggested to be the main drivers of persistent and perpetuating poverty traps (Dalton, Ghosal, and Mani 2016). In this paper, we examine whether individual experiences and attitudes are statistically correlated with educational aspirations and whether, when taken into consideration, can serve to mitigate any identified SES gap. This specific approach to the analysis represents the main contribution of our paper. It is important to clarify that we do not analyse these factors as ‘causal mechanisms’ but rather as mediators that might be statistically associated with the SES gap in aspirations.

Table 2 provides a summary of the experiences and attitudes we consider. Appendix B provides a detailed description of each variable's derivation from the survey questionnaire.

Table 2. List of experiences and attitudes.

Experiences
Information, advice and guidance (IAG)
Bullying victimisation
Attitudes
External locus of control
Low self-perception of academic potential

The first experience we analyse is associated with the provision of IAG regarding post-compulsory educational options. We consider differences between students who received a lot of IAG from teachers or family members in comparison with students who received a little IAG or no IAG at all. From the strand of theoretical literature on aspirations, Appadurai (2004) suggested that the stock of available experiences forms aspirations and Ray (2006) proposed that the flow of information and the availability of role models influence an individual's perception of existing options. From an empirical perspective, Hoxby and Turner (2015) used data from the US to show that lack of IAG is a reason why low-income students have lower aspirations than high-income students. Similarly, Guyon and Huillery (2020) using data from France identified that socially disadvantaged students are less aware of the top educational pathways suggesting that the options students consider are biased in a stereotypical way.

The second experience we consider is associated with students' bullying victimisation, defined as a repeated oppression, psychological or physical, of a less powerful person by a more powerful one (Farrington 1993). It can take many forms including social exclusion, name calling, being threatened with violence and being forced to hand over money or possessions. The consequences of bullying victimisation can be severe in terms of young people's mental well-being, attitudes towards school, educational attainment and even potential suicide risk (Smith 2004; Gorman et al. 2020). To the best of our knowledge, this is the first study to explore bullying victimisation as an experience associated with SES inequality in educational aspirations. Previous analyses on English students identified that low-achieving White girls from step- or single- families with SEN or other disabilities were the ones most likely to be bullying victims while no association with SES was identified (Green, Collingwood, and Rose 2010). It is, of course, acknowledged that students who are victims of bullying might be systematically different from students who are not, in terms of non-observable characteristics, such as students with low self-esteem. Although our very detailed empirical model accounts for a large diversity of such factors, we are aware that bullying victimisation could also capture the impact of attitudes that we might not address in our model.

Unlike experiences, attitudes can be interpreted as both the cause of poverty or its consequence (Dalton, Ghosal, and Mani 2016). We examine how individual attitudes are correlated with the formation of aspirations and whether these attitudes might act as mechanisms of SES inequality. We consider attitudes which could be argued to be inherent characteristics of the individual rather than choices they made as for example, engagement in positive activities like reading, or risk activities like smoking. Following this approach, we can rule out any potential reverse causality and describe a number of aspects of young people's self-concepts, motivations and behaviours, all of which have been previously linked to adolescent outcomes.

Although the LSYPE does not explicitly include information on the ‘Big Five’ personality traits that are usually adopted as a framework for such analyses, it includes a series of questions on locus of control and behaviours towards school which we use to construct our measures (Mendolia and Walker 2014, 2015). The first attitude we consider is associated with having an external locus of control. Locus of control refers to the extent to which individuals believe they are in control of events that affect them. Individuals with an external locus of control perceive their life as being controlled by environmental factors they cannot change, by chance or fate. Having an external locus of control is found to be negatively associated with behaviours and outcomes including educational performance and subject choices (Mendolia and Walker 2014). Second, we consider individuals' self-perception of academic potential based on responses on how good they believe they are in four school subjects. Guyon and Huillery (2020) showed that low SES students tend to underestimate their academic ability which in turn explains a significant amount of the SES gap in educational aspirations.

4.4. Descriptive statistics

The descriptive statistics of the analytical sample by SES tertile are provided in Table 3. The data show substantial differences in educational aspirations by SES, especially for academic and university aspirations. For example, 71% of the students in the highest SES tertile aspired to follow the academic route while only 53% of the students from the lowest SES tertle aspired to this route. Similarly, 78% of the highest SES students aspired to attend university while this proportion drops to 62% when considering students from the lowest SES group.

Table 3. Descriptive statistics.

SES Tertile	(1)	(2)	(3)	(4)
	Highest	Middle	Lowest	Total
	SES	SES	SES
	mean/(sd)	mean/(sd)	mean/(sd)	mean/(sd)
Educational aspirations
Aspires to stay in education	0.87	0.81	0.80	0.83
	(0.33)	(0.39)	(0.40)	(0.37)
Aspires to follow the academic route	0.71	0.58	0.53	0.62
	(0.45)	(0.49)	(0.50)	(0.49)
Aspires to attend university	0.78	0.67	0.62	0.69
	(0.41)	(0.47)	(0.49)	(0.46)
Experiences and attitudes
Received a lot of IAG	0.78	0.78	0.78	0.78
	(0.41)	(0.41)	(0.41)	(0.42)
Has been bullied	0.55	0.53	0.56	0.55
	(0.50)	(0.50)	(0.50)	(0.41)
External locus of control	0.39	0.42	0.44	0.41
	(0.49)	(0.49)	(0.50)	(0.49)
Low self-perception of academic potential	0.25	0.31	0.33	0.30
	(0.44)	(0.46)	(0.47)	(0.46)
Prior attainment
Average KS2 scores (standardised)	0.48	0.04	−0.30	0.10
Background characteristics
	(0.82)	(0.94)	(0.98)	(0.96)
Girl	0.47	0.49	0.51	0.49
	(0.50)	(0.50)	(0.50)	(0.50)
White	0.83	0.72	0.57	0.72
	(0.37)	(0.45)	(0.49)	(0.45)
Has special educational needs	0.15	0.17	0.17	0.16
	(0.35)	(0.38)	(0.38)	(0.37)
Number of siblings	1.39	1.43	1.71	1.50
	(0.91)	(3.32)	(3.67)	(2.82)
Single-parent household	0.00	0.20	0.53	0.22
	(0.03)	(0.40)	(0.50)	(0.41)
Observations	3,087	2,827	2,421	8,335

Note: The numbers presented in each column are the mean values of each characteristic for the analytical sample by SES tertile (Total N = 8, 335). The numbers in parentheses are the standard deviations. Observations for aspiration 2 are conditional on aspiring to stay in post-compulsory education (91% of the analytical sample).

Considering IAG provision and bullying victimisation we do not observe any significant differences among students of differing SES groups. On the other hand, there are some sizeable SES differences in students' self-perception of academic potential and locus of control. Lower SES students are more likely to believe little in their academic potential and that their behaviour will not lead to valued reinforcement. Such preliminary evidence suggests that individual attitudes are more likely to be associated with aspirations through a socio-economic channel than is the case for experiences, corroborating our approach to analyse the effect of these two groups of constraints individually.

Furthermore, there are significant SES differences in students' prior attainment. The average KS2 scores of students in the lowest SES tertile are significantly lower from students in the middle and highest SES tertile. Finally, students from ethnic minorities and from single-parent households tend to be over-represented in the lowest SES tertle.

5. Empirical results

5.1. Main findings

Our main empirical findings are presented in Tables 4–6 for each of the three educational aspirations we consider. The first column shows the raw SES differences in aspirations. Then, we examine the extent to which these gaps can be explained by differences in other observable characteristics, including prior attainment in column 2, basic individual characteristics and family composition in column 3 and the secondary school attended in column 4. Finally, column 5 presents the SES gap in aspirations when differences in experiences and attitudes are considered.

Table 4. SES gap in aspirations to stay in education.

	(1)	(2)	(3)	(4)	(5)
SES (reference: Highest SES)
Middle SES	−0.063***	−0.025**	−0.042***	−0.039***	−0.035***
	(0.010)	(0.010)	(0.010)	(0.011)	(0.011)
Lowest SES	−0.070***	−0.002	−0.045***	−0.046***	−0.040***
	(0.011)	(0.011)	(0.012)	(0.014)	(0.014)
Background characteristics
Average KS2 scores		0.087***	0.082***	0.078***	0.067***
		(0.005)	(0.005)	(0.006)	(0.006)
Girl			0.104***	0.108***	0.114***
			(0.008)	(0.009)	(0.010)
White			−0.163***	−0.127***	−0.117***
			(0.008)	(0.013)	(0.013)
Special educational needs			−0.052***	−0.059***	−0.053***
			(0.013)	(0.015)	(0.014)
Number of siblings			0.001	0.002	0.002
			(0.002)	(0.002)	(0.002)
Single-parent household			−0.008	0.001	0.001
			(0.011)	(0.013)	(0.012)
Experiences and attitudes
Received a lot of IAG					0.044***
					(0.011)
Has been bullied					0.005
					(0.008)
External locus of control					−0.025***
					(0.009)
Low self-perception of academic potential					−0.076***
					(0.011)
N	8335	8335	8335	8335	8335
Number of clusters	780	780	780	780	780
R-squared	0.01	0.05	0.12	0.22	0.23
School FEs				✓	✓

Notes: Standard errors clustered at the secondary school level and reported in parentheses.Significance levels: ${}^{\ast }p<0.10$, ${}^{\ast \ast }p<0.05$, ${}^{\ast \ast \ast }p<0.01$.

Table 5. SES gap in aspirations to follow the academic route.

	(1)	(2)	(3)	(4)	(5)
SES (reference: Highest SES)
Middle SES	−0.132***	−0.082***	−0.088***	−0.061***	−0.060***
	(0.013)	(0.013)	(0.013)	(0.013)	(0.013)
Lowest SES	−0.180***	−0.092***	−0.105***	−0.072***	−0.068***
	(0.016)	(0.016)	(0.018)	(0.019)	(0.018)
Background characteristics
Average KS2 scores		0.115***	0.115***	0.091***	0.078***
		(0.007)	(0.007)	(0.008)	(0.008)
Girl			0.021*	0.021*	0.032***
			(0.012)	(0.012)	(0.012)
White			−0.141***	−0.091***	−0.079***
			(0.016)	(0.018)	(0.018)
Special educational needs			−0.050***	−0.073***	−0.067***
			(0.017)	(0.018)	(0.018)
Number of siblings			−0.001	−0.000	−0.000
			(0.002)	(0.003)	(0.003)
Single-parent household			−0.051***	−0.044***	−0.039**
			(0.016)	(0.016)	(0.016)
Experiences and attitudes
Received a lot of IAG					0.004
					(0.012)
Has been bullied					−0.036***
					(0.011)
External locus of control					−0.013
					(0.011)
Low self-perception of academic potential					−0.086***
					(0.014)
N	7617	7617	7617	7617	7617
Number of clusters	764	764	764	764	764
R-squared	0.03	0.07	0.09	0.31	0.32
School FEs				✓	✓

Notes: Post-compulsory education participants only. Standard errors clustered at secondary school level and reported in parentheses.Significance levels: ${}^{\ast }p<0.10$, ${}^{\ast \ast }p<0.05$, ${}^{\ast \ast \ast }p<0.01$.

Table 6. SES gap in aspirations to attend university.

	(1)	(2)	(3)	(4)	(5)
SES (reference: Highest SES)
Middle SES	−0.117***	−0.051***	−0.080***	−0.072***	−0.066***
	(0.012)	(0.012)	(0.012)	(0.013)	(0.013)
Lowest SES	−0.167***	−0.051***	−0.121***	−0.118***	−0.108***
	(0.014)	(0.014)	(0.015)	(0.017)	(0.017)
Background characteristics
Average KS2 scores		0.149***	0.147***	0.140***	0.119***
		(0.006)	(0.006)	(0.007)	(0.007)
Girl			0.073***	0.083***	0.096***
			(0.010)	(0.011)	(0.011)
White			−0.285***	−0.211***	−0.194***
			(0.011)	(0.015)	(0.015)
Special educational needs			−0.064***	−0.074***	−0.064***
			(0.016)	(0.017)	(0.017)
Number of siblings			0.000	0.000	0.001
			(0.002)	(0.002)	(0.002)
Single-parent household			−0.013	−0.011	−0.010
			(0.014)	(0.015)	(0.015)
Experiences and attitudes
Received a lot of IAG					0.054***
					(0.012)
Has been bullied					0.008
					(0.010)
External locus of control					−0.053***
					(0.010)
Low self-perception of academic potential					−0.135***
					(0.012)
N	8335	8335	8335	8335	8335
Number of clusters	780	780	780	780	780
R-squared	0.02	0.11	0.20	0.30	0.32
School FEs				✓	✓

Notes: Standard errors clustered at secondary school level and reported in parentheses. Significance levels: ${}^{\ast }p<0.10$, ${}^{\ast \ast }p<0.05$, ${}^{\ast \ast \ast }p<0.01$.

Focusing on column 1, the findings reflect large and significant differences in educational aspirations by SES. On average, low SES students are 7 percentage points (p.p.) less likely to aspire to stay in education, 18 p.p. less likely to aspire to follow the academic route and 17 p.p. less likely to aspire to attend university than high SES students. The findings in column 2 indicate that a significant proportion of the SES gap is explained by differences in prior attainment, which, in any case, is also correlated with SES. When average KS2 scores are taken into consideration the SES gap becomes statistically insignificant for aspirations to stay in education. For aspirations to follow the academic route the gap is reduced to about half and for university aspirations it is reduced to about one third. Column 3 takes into account individual characteristics and family composition. Interestingly, the SES effect slightly increases compared to what we observed in column 2, but it is still considerably lower from the raw gap observed in column 1. When school fixed effects are included in column 4, the SES gap in aspirations to stay in education is essentially the same implying that differences in the types of schools that students attend do not explain the SES gaps in their aspirations. For aspirations to follow the academic route and to attend university, it is indicated that much of the SES gap can be attributed to factors associated with the secondary school attended, such as geographic location, teacher quality and peer effects. This is also reflected from the increase in the R-squared, particularly for aspirations to follow the academic route. The findings suggest that SES gaps in aspirations exist also within schools rather than being entirely driven by endogenous self-sorting of students across schools.

In summary, among schoolmates with similar prior attainment and background characteristics, students from low SES backgrounds set different aspirations for their future education. Thus, the factors associated with SES inequality are more profound and require further investigation.

Column 5 presents our most detailed specification, augmenting the within-school model with individual experiences and attitudes. Individuals who received a lot of educational IAG are more likely to aspire to stay in education and to follow the academic route. Bullying victimisation, on the other hand, is negatively associated only with aspirations to follow the academic route. Turning to individual attitudes, students with an external locus of control are significantly less likely to aspire to stay in education and to attend university. Finally, self-perception of academic potential seems to have the biggest impact on aspirations from all the mediators we consider. Young individuals who have a low perception of their academic potential are 8 p.p. less likely to aspire to stay in education, 9 p.p. less likely to aspire to follow the academic route and 13 p.p. less likely to aspire to attend university.

Overall, these results suggest that policies which, for example, would provide educational IAG and that would support individuals to maintain a higher self-esteem would be efficient in raising educational aspirations. While we interpret these correlations as providing a broad sense of the effect of IAG , bullying victimisation, locus of control and self-perception of academic potential on educational aspirations, we certainly do not claim that these estimates are causal. Nonetheless, paying attention to these factors constitutes a set of suggestions for positive policy implementations – as we discuss in our concluding section.

To suggest whether working on these factors would be also capable to reduce social inequalities in aspirations, we use Gelbach (2016)'s decomposition to quantify the specific contribution of five groups of covariates: prior attainment, demographic characteristics, secondary school fixed effects, experiences and attitudes. The outcome from this analysis is presented in Table 7. Decomposing the values of covariates on the change in SES coefficients reveals that differences in prior attainment account for most of the SES gap in all the educational aspirations we consider. Demographic characteristics are associated with the SES gap in aspirations to stay in education and to attend university but not for aspirations on route choice. On the other hand, the secondary school attended seems to be an important factor associated with the SES gap in aspirations to follow the academic rather than the vocational post-compulsory educational route, explaining over 50% of the effect. This is an interesting finding as it reflects how schools might encourage students to take specific post-compulsory routes, sometimes not reflecting the best match for them. This result comes to complement earlier findings from Maragkou (2020) suggesting that 50% of the SES gap in the match between student attainment and upper-secondary qualification could be explained by factors associated with the secondary school students from differing backgrounds attend.

Table 7. Conditional decomposition of the SES gap in educational aspirations.

	Post-compulsory educational aspirations
	Stay in education	Academic route	Attend university
Middle SES (reference: Highest SES)
Average standardised KS2 scores	−0.029***	−0.034***	−0.052***
	(0.003)	(0.004)	(0.004)
Demographics	0.013***	0.001	0.018***
	(0.004)	(0.004)	(0.004)
School FEs	−0.006	−0.035***	−0.007
	(0.004)	(0.008)	(0.005)
Experiences	−0.000	0.001	−0.000
	(0.001)	(0.001)	(0.000)
Attitudes	−0.005***	−0.004***	−0.010***
	(0.001)	(0.001)	(0.002)
TC	−0.027***	−0.071***	−0.051***
	(0.007)	(0.010)	(0.009)
Lowest SES (reference: Highest SES)
Average standardised KS2 scores	−0.052***	−0.060***	−0.093***
	(0.005)	(0.007)	(0.006)
Demographics	0.034***	0.002	0.046***
	(0.008)	(0.010)	(0.009)
School FEs	−0.005	−0.048***	−0.001
	(0.007)	(0.013)	(0.008)
Experiences	0.000	−0.000	0.003
	(0.000)	(0.001)	(0.001)
Attitudes	−0.007***	−0.007***	−0.013***
	(0.001)	(0.002)	(0.002)
TC	−0.030***	−0.112***	−0.059***
	(0.012)	(0.018)	(0.014)
N	8335	7617	8335
Number of clusters	780	764	780

Notes: Decomposition analysis based on Gelbach (2016). Post-compulsory education participants only. Standard errors clustered at secondary school level and reported in parentheses. Significance levels: ${}^{\ast }p<0.10$, ${}^{\ast \ast }p<0.05$, ${}^{\ast \ast \ast }p<0.01$. The basic model includes as regressors Middle SES and Lowest SES (Highest SES is the base category). KS2 scores include average scores from Maths, Science and English in standardised form. Demographics include gender, ethnicity, SEN status, number of siblings and single-parent household indicator. Experiences include IAG provision and bullying victimisation. Attitudes include having an external locus of control and low self-perception of academic potential.

Turning to the coefficient for experiences and attitudes which is the focus of this paper, the findings from the decomposition analysis suggest that IAG and experience of bullying victimisation are not associated with the SES gap in aspirations. This might not be surprising as we carry out a within-school investigation of the effect, therefore eliminating substantial variation in the provision of IAG (at least at the school level). The findings on bullying victimisation are also supporting the findings from Gorman et al. (2020), suggesting that the key determinant of being bullied is being different from those around you rather than the levels of any particular variable. Considering individual attitudes, the findings indicate a statistically significant association with the SES gap in aspirations. The magnitude of the coefficient ranges from a modest 6% for aspirations to stay in education and aspirations to follow the academic route, to a relatively substantial 22% for aspirations to attend university for the lowest SES students.

Taken together, the descriptive evidence emerging from the results of the empirical analysis is clear. The differences in aspirations between low and high-SES students, especially those related to attending university, are associated with early academic results, demographics and school attended – something already known in the literature. However, the analysis presented here adds important evidence about the correlation between individual attitudes (including locus of control and self-perception of academic potential) and the SES gap in aspirations. Given the importance of aspirations in shaping future outcomes these findings constitute new information to be considered when designing educational policies, as we suggest in the final section of the paper. To motivate further on the importance of aspirations, Appendix C presents and discusses evidence on the association between aspirations and future educational and professional outcomes, including performance in GCSE exams, university participation and employment and income at age 25.⁴ Finally, the lack of statistical correlation between SES gaps in aspirations and individual experiences does not imply that IAG provision and bullying victimisation are not relevant elements to be considered. It might be, for example, that families and schools already work substantially to mitigate the effects of these experiences for disadvantaged students, so that the differences in aspirations at age 15 are not extremely relevant anymore. The causal mechanisms of these relationships deserve further investigation (and more detailed data) in future research.

5.2. Robustness checks and additional analyses

Appendix D presents a series of robustness checks on our main estimates. In the first robustness check, we test the sensitivity of our results by considering students' performance in KS3 exams, undertaken a year before aspirations are measured, at age 14. In the second robustness check, we examine the sensitivity of our findings to estimating a conditional-logit model to take into consideration the dichotomous nature of our dependent variables. Inevitably, conditional-logistic regression results in a reduction of the sample size when no variation in aspirations can be observed within schools, corroborating using OLS regression as our main estimation method. Finally, given that from the initial cohort we only consider participants in Wave 2, we run a robustness check where we adjust survey weights to check the representativeness of the sample with the original cohort. In summary, all of these additional analyses do not change the pattern of our main estimations.

Moreover, while we used a composite measure of SES to take into account the multi-dimensional nature of this parameter, it is still interesting to see if specific factors are more important in the formation of aspirations. In Appendix E, we decompose the SES index and present our main findings when taking into consideration parental education in Table E.1 and parental occupation in Table E.2. We find that parental education and occupation are both statistically associated with aspiration formation, especially for aspirations to attend university. Precisely, we find that students whose main parents have no post-compulsory qualification are 10.5 p.p. less likely to aspire to go to university compared to students whose main parent has an academic degree. Also, students with main parents employed in unskilled occupations are 17 p.p. less likely to aspire to attend university compared to students with main parents in professional occupations.

6. Implications and concluding remarks

The findings presented in this paper are consistent with the literature establishing SES gaps in educational aspirations. We provide a comprehensive contribution towards understanding the importance of potential factors associated with the SES gap, including IAG, bullying victimisation, locus of control and self-perception of academic potential. The findings suggest that while both experiences and attitudes are correlated with educational aspirations, only attitudes are associated with SES gaps in their formation.

A number of policy consequences can be derived from the findings of this research. First, self-sorting of students across schools does not completely explain differences in their aspirations. Even in schools where students group together for any observable or unobservable reason, there are still differences in their aspirations, mainly associated with SES. This evidence calls for attention in all schools to design and implement instruments for raising the aspirations of disadvantaged students. Several instruments can be conceived in this regard. For example, tutoring and mentoring activities might become tools to be developed by all schools, and targeted primarily to disadvantaged students. Moreover, schools could create specific occasions in which external testimonials can bring their experiences to students, showing how they found their way to life satisfaction and success despite an initial disadvantage.

Second, aspirations and academic performance complement each other. Albeit difficult to establish the causal direction, these two dimensions are evidently correlated. In such a perspective, if teachers work for improving students' academic skills, this can contribute also to raise their aspirations which, in turn, will be associated with future outcomes – such as subsequent academic performance, educational choices and job market outcomes. This channel of action is particularly important as it is completely in the autonomy and core mission of schools. All activities that are directed towards improving the academic skills of students are welcome in this perspective. Indeed, improving academic results seems associated with subsequent higher levels of aspirations, and this contributes positively to the development of a more robust and wider ‘human capital’ for the students. Seen this way, improving academic results is not only an efficient use of resources but might also raise equity and equality of opportunities for disadvantaged students.

Third, working directly on aspirations would be worthwhile. The paper highlights several channels that are associated with aspirations which represent a potential agenda for action. For example, providing information, advice and guidance to disadvantaged and/or low-performing students could be an important aspect for improving their awareness and aspirations. This agenda extends to the relationships between schools, families, and external agencies. The set of positive experiences that can shape aspirations is, indeed, larger that the activities that can be developed by the schools alone. For example, widening the opportunities for disadvantaged families to attend cultural events and educational places (theatres, cinemas, and museums) can help students raising their aspirations for schooling. Or, universities can create ad-hoc programs for involving students in critical reflections about their future opportunities in the early stages of education (also beyond the school curricular and extracurricular activities). Or, non-profit agencies can develop student support initiatives that can be publicly and privately funded to assist students in their out-of-the-school time. Overall, the work on raising aspirations of disadvantaged students might be a ‘call for action’ directed to the society at large.

These implications are important for schools, suggesting that the role of teachers and principals should go beyond a narrow attention to academic results. These key decision-makers should, instead, consider their comprehensive potential impact on students' lives, which in turn is more likely to affect aspirations and educational decisions in future.

Acknowledgments

We are grateful to the Editor and two anonymous reviewers for providing relevant comments to improve earlier versions of this paper. We also thank seminar participants at the University College Dublin (October 2021) and participants at the Summer School on Socioeconomic Inequality (June 2021), the Association for Education Finance and Policy (AEFP) Conference (March 2021), the Essen Health Conference (May 2021), and the International School Choice and Reform (ISCR) Conference (January 2020). We are particularly thankful to Mark Berends, Steven McIntosh, Matteo Sandi and Leanna Stiefel for their useful suggestions. Access to the data has been kindly provided by the UK Secure Data Service (SDS) at the UK Data Archive. All errors are the authors' own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Since 2009, the externally marked exams at KS3 are replaced with teacher assessments but the cohort analysed in this study has not been affected by this change.

2 Missing observations on aspirations in Wave 2 of the survey (age 14–15) are replaced from recorded aspirations in Wave 1 (age 13–14). Further missing observations are replaced from recorded aspirations in Wave 3 (age 15–16). Interviews for Waves 1–3 were taken while individuals where still in compulsory education.

3 Missing information on second parent's education and occupation is replaced with missing dummies.

4 The measures on future outcomes are considered only when information was available for the analytical cohort.