Career guidance to lessen inequalities in education: internal evidence from OECD–PISA 2018 data

ABSTRACT

The students’ socioeconomic background effect on the educational outcomes makes it more difficult to find real equality of opportunity within the educational system. In this paper, we analyse the relationship of career guidance at school with cognitive skills, and also with expectations and motivation for a sample of more than 188,000 15-year-old students in 29 different countries, using the OECD PISA 2018 data. We apply a propensity score matching procedure with a novel multilevel second stage. We provide novel evidence of the positive influence of career guidance on the variables of interest, but only for disadvantaged students.

KEYWORDS:

• Career guidance
• inequality
• education
• cognitive skills
• motivation

JEL CLASSIFICATION:

• A20
• J31

Disclosure statement

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

Notes

1 Moreover, avoiding a loss gives more incentive than having new and better experiences, favouring the most advantaged individuals and families to motivate and give school support (Keller and Zavalloni 1964; Müller 2014).

2 Analysing how the career guidance is provided is a difficult task, especially in an international context, especially because it is provided by people with different backgrounds (OECD 2004).

3 It has been impossible to use all OECD countries for availability of information reasons. The sample is formed by students from Australia, Switzerland, Chile, Colombia, Czech Republic, Germany, Denmark, Spain, Estonia, Finland, France, Greece, Hungary, Iceland, Israel, Italy, Korea, Lithuania, Luxembourg, Latvia, Mexico, Netherlands, Poland, Portugal, Slovak Republic, Slovenia, Turkey, and the United States.

4 The modal grade is the expected grade according to the students’ age and country.

5 Students were surveyed to report on their expected occupation at age 30, coding the responses into four-digit ISCO codes, and then recoding this into the ISEI index. See OECD (2018), for more information.

6 This motivation does not depend on any punishment or reward, i.e. it is the intrinsic motivation.

7 Students were asked to what extent they identified (“not at all true of me”, “slightly true of me”, “moderately true of me”, “very true of me”, “extremely true of me”) with the following statements about their ambitious learning goals: “My goal is to learn as much as possible”; “My goal is to completely master the material presented in my classes”; and “My goal is to understand the content of my classes as thoroughly as possible”. All statements were used to create the index through the application of a generalised partial credit model, where values correspond to warm likelihood estimates (with an average of 0 and a standard deviation of 1 across OECD countries). See OECD (2019), for more information.

8 See Figures 10–13 in the annex section for the differences in terms of distribution between these two groups of students.

9 Economically disadvantaged countries: Chile, Colombia, Greece, Hungary, Lithuania, Latvia, Mexico, Polonia, Portugal, Slovakia, and Turkey.

10 Economically advantaged countries: Australia, Switzerland, Germany, Denmark, Finland, Iceland, Luxembourg, Netherlands, and the United States.

11 The variables used in the logit estimations are gender, immigrant status, grade compared to the modal, age, socioeconomic background (ESCS index), pre-primary school attendance (ISCED 0), school type, socioeconomic background average, school size, location (located in a city or not), percentage of girls, and student–teacher ratio. See Section 2.1 and Table 1 for more information about the variables used. See Table 4 in the annex section for more information about the logit estimation.

12 This implies that individuals can be used more than once reducing the possible bias (Caliendo and Kopeinig, 2008).

13 See Figure 14 in the annex section for more information about the balance.

14 See Figure 15 and Table 9 in the annex section for more details about the matching procedure.

15 See Figure 16 and Table 11 in the annex section for more details about the matching procedure.

16 Linear regressions are used for reasons of simplicity and technical support.

17 It also happens in the unreal scenario of unobservables having twice the importance of observables (δ = 2).

18 See Figure 17 and Table 13 in the technical annex for more information.

19 See Figures 18–19 and Tables 14–15 in the technical annex for more information.

20 See Figures 20–21 and Tables 16–17 in the technical annex for more information.

Additional information

Funding

This work was supported by the Conselleria d'Hisendai Model Econòmic of the Valencian Government [grant number HIECPU/2019/6].