Evolving approaches to the study of childhood poverty and education

ABSTRACT

Social scientists have conceptualised poverty in multiple ways, with measurement approaches that seek to identify absolute, relative, subjective, and multi-dimensional poverty. The concept of poverty is central in the comparative education field, but has been empirically elusive in many large, international educational surveys: these studies have not typically included measures that correspond to prevalent conceptualisations or measurement strategies in the poverty literature. In this paper, we contrast poverty conceptualisation and measurement in the poverty literature with socio-economic measures prominent in major international educational surveys. Disconnects between these approaches, and implications for understanding how the disadvantages of poverty in childhood are reflected in educational surveys, are considered. We discuss key challenges that continue to shape possibilities for incorporation of poverty-related concepts into educational surveys. We close with a set of recommendations and considerations.

KEYWORDS:

• Child poverty
• education surveys
• socio-economic status
• deprivation

Acknowledgements

This paper benefited from the insightful suggestions of the editor and three anonymous reviewers.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Emily Hannum is a professor of sociology and education at the University of Pennsylvania. Her research interests are poverty and child welfare, social stratification, and sociology of education. Current projects include studies of childhood poverty and inequality in China and minority youth in China’s educational system.

Ran Liu is a doctoral student in Sociology at the University of Pennsylvania. Her research interests include gender, education, and inequality. Her dissertation investigates how gender norms and stereotypes impact adolescents’ aspirations for STEM-related careers in East Asian countries.

Andrea Alvarado-Urbina is a doctoral student in Sociology at the University of Pennsylvania. Her research interests are education and migration in developing countries, with a focus on Latin America. She is currently working on a project on multicultural ethnic identities in the incorporation of immigrant students in the north of Chile.

ORCID

Emily Hannum http://orcid.org/0000-0003-2011-9984

Notes

1. Additional approaches not discussed here include subjective poverty, or falling below a subjective perception of ‘the amount of income it takes to barely get by’, and asset poverty, defined as wealth minus debt (Yoshikawa, Aber, and Beardslee 2012, 273).

2. Some scholars have proposed a ‘hybrid’ approach to setting poverty lines that is sensitive to changes in the general living standard, but less so than a purely relative approach (for a discussion, see Foster 1998, 335).

3. Although in its early years, LSMS studies were highly standardized, over time the growing number of low- and middle-income countries included in the project introduced the need for more increasingly customized designs.

4. However, asset-based measures also face challenges. For example, it has been argued that in middle-income countries, current asset indices may not be detailed enough to accurately reflect the quality and nature of household possessions. There has also been some concern that asset- and service-based measurements may capture social stratification better in urban than rural areas, leading to potential misclassification of rural households (Howe et al. 2012).

5. Others explore alternative ways to utilize these original measurements. For example, Sahn and Stifel (2000) employ factor analysis (FA) on DHS data in their earlier analysis of poverty in nine African countries. Booysen et al. (2008) employ multiple correspondence analysis (MCA) instead of PCA to track changes in poverty over a period of 10–15 years in 7 Sub-Saharan African countries.

6. The FAS has been used to study SES inequalities and relevant health outcomes in many HBSC studies. For example, children with lower FAS scores are found to consume more soft drinks and high-sugar foods (Mullan and Currie 2000; Inchley et al. 2001). High FAS scores have been associated with higher frequency of tooth brushing (Maes et al. 2006) and participation in physical activities (Mullan and Currie 2000; Holstein et al. 2004; Inchley et al. 2005). The FAS score is sometimes directly used as a continuous variable, but more frequently is recoded into groups at different affluence level to resemble the conventional categorization of SES groups (for example, Holstein et al. 2004; Inchley et al. 2005; Maes et al. 2006).

7. While PERCE only considered 3rd grade students; SERCE and TERCE included 3rd and 6th graders.

8. PERCE included 13 countries, SERCE included 16 countries and the Mexican State of Nuevo León, and TERCE considered 15 countries and the Mexican State of Nuevo León.

9. Furthermore, LLECE studies collect information about the ethnic identity and indigenous background of students and their families. For instance, the three studies ask about the language spoken at home, including indigenous languages among the alternatives. SERCE and TERCE include questions about language spoken by teachers, by parents, and by the students themselves, and TERCE also has a set of indicators of immigration status (child, mother, and father’s place of birth).

10. The school, principal, and teacher questionnaires provide information about the socio-economic status of the schools the students attend. These instruments gather data such as the average socio-economic status of students in each school, the average educational attainment of parents, and standardized measures of sanitation, infrastructure, and equipment (Spaull 2013).

11. There are additional problems with children as survey respondents. For example, some young children (in particular 4–7 years old) may be reluctant to answer questions, because they often assume that the adults already know everything, and they would want to avoid saying something wrong or stupid (Maccoby and Maccoby 1954).

12. According to Hill (1997), social work researchers have avoided using survey questionnaires when studying children, preferring methods that allow more flexibility and interaction with the child. Furthermore, methods and data gathering techniques ought to be selected keeping in mind the goal of ‘enhancing the willingness of children to communicate’; this often involves using techniques other than verbal communication, such as vignettes, written prompts, picture prompts, role playing, and drawing, among others (Hill 1997). More recently, Fargas-Malet et al. (2010) document the use of photography, drawings, ‘stimulus material’ or prompts, and diaries as means to stimulate children’s responses.

13. Not all research is completely consistent on this point: in the United States, a study using a national adolescent health survey indicated that child–parent concurrence on mothers’ education decreased with mothers’ education, but increased with household income (Ridolfo and Maitland 2011).

14. Looker (1989) reviews studies on disagreement and concludes that the students’ report on parental socio-economic status can be regarded as valid if the respondents are high school seniors, living with their parents, and reporting on characteristics that are salient to themselves.

15. Key lines of research on intra-household inequality have focused on gender. There has been growing attention to and debate about the ‘feminization of poverty’ (Chant 2008), which cannot be revealed and properly discussed using household-level measurements alone. Corsi, Botti, and D’Ippoliti (2016) argue in favor of using individualized measures of poverty by comparing the gender differences identified through two different measures. The first one is the At-Risk-of-Poverty Rate (ARPR), or the percent of adult population with a net equivalent income lower than 60% of the national median income; this measure uses households as units, assuming that within each household resources are equally shared (Corsi, Botti, and D’Ippoliti 2016, 85). On the other hand, the authors estimate the Financial Dependency Rate (FDR), or the share of men and women whose individualized income is less than 60% of the national median income. Corsi, Botti, and D’Ippoliti find that the gender gap is much larger when using FDR (gap between 20.6% and 22.8%) than when measuring poverty through ARPR (gap between 1.5% and 2.4%). This implies that relying on the assumption of equal sharing of households’ resources implies a serious risk of underestimating the true gender gap in poverty (Corsi, Botti, and D’Ippoliti 2016, 93).

16. In addition, studies of developed country have been including child welfare measures that are still uncommon in developing countries, such as physical and sexual abuse, teenage pregnancy, drug and alcohol abuse, and so on (White, Leavy, and Masters 2003).

Additional information

Funding

The authors would like to acknowledge fellowship support from the University of Pennsylvania Benjamin Franklin Fellowship Program to Ran Liu and from the Advanced Human Capital Program of the National Commission for Scientific and Technological Research (CONICYT) of Chile (Folio. 72150110) to Andrea Alvarado-Urbina.