Urban Welfare and Data Science: Going beyond

Socioeconomic Status in School Inequalities

dos Santos, RP1, Lemes, IL1
1: ULBRA - Lutheran University of Brazil, Canoas, Brazil, renatopsantos@ulbra.edu.br

Introduction
The uncontrolled growth of population and cities in Brazil segregates people in socio-
spatial terms and distributes urban public resources unfairly, threatening students
access to the structure of educational opportunities (de Queiroz Ribeiro, Koslinski,
Zuccarelli, & Christovo, 2016).
Despite the tantalising evidence for Brazilian students socioeconomic status being
one of the leading single predictors of their school achievements (dos Santos, 2017),
this measure does not account for the complexity of the real income (Harvey, 1971)
that shall be provided by the city and used collectively by the people, in terms of
material conditions of life, in the form of housing, transportation, and quality
education. Consequently, this concept of real income allows one to evaluate better
how the uneven distribution of urban conditions among social groups in the city
favours social inequalities.
The Observatrio das Metrpoles (Observatory of the Metropolises) developed the
Urban Welfare Index (IBEU) (de Queiroz Ribeiro & Ribeiro, 2013), having five
dimensions (D1 - D5), each composed of one to seven indices (Table 1), including
existence of public lighting, paving, sidewalk, wheelchair ramp, open sewage and
garbage accumulated around the homes, electricity and water supply, sanitation,
material of the walls of the houses, and number of residents per bedroom and
bathroom.

Table 1 Indices and dimensions of the Urban Welfare Index (IBEU)

D1 Urban Mobility: D4 Attendance of Urban
Collective Services:
1. Home-to-Work Displacement 10. Water supply
D2 Urban Environmental 11. Sewage treatment
Conditions:
2. Arborization around households 12. Electricity supply
3. Open sewage in the vicinity of 13. Waste collection
households
4. Accumulated garbage around D5 Urban Infrastructure:
households
D3 Urban Housing Conditions: 14. Public lighting
5. Subnormal clump 15. Paving
6. Household density: number of 16. Sidewalk
 people in the household per
number of dormitories
7. Resident/bathroom density 17. Curb / Guide
8. Household wall material 18. Road gully
9. Species of households 19. Wheelchair ramp
20. Street address identification
Source: de Queiroz Ribeiro & Ribeiro (2013)

Methodology
We expanded the study of de Queiroz Ribeiro et al. about Brazilian metropolises to
all 5,564 Brazilian municipalities to investigate the relation between the residential
segregation and segmentation and school inequalities.
Those indices and dimensions were studied as predictors of school achievements as
measured using their average ENEM (National High School Exam) 2015 outcomes
and IDEB (Basic Education Development Index) 2011 values, using Data Science
methods.
First, we evaluated1 correlation factors between IDEB and each of those indices and
dimensions, as well as the ENEM and socioeconomic status aggregated across
Brazil.
Subsequently, we analysed those relations across all the municipalities with
choropleth maps. Choropleth maps include one or more quantities of interest in
relative values (often percentages), such as demographic density or per capita
income, which are visually transcribed proportionally by means of patterns or
shadings by colours, according to a particular scale, and launched in the respective
geographic areas of occurrence. In the multicolour bivariate choropleth maps
(Murphy, 2016) we used here, two variables are superimposed in the map and
positive correlations between them are indicated by the colours ranging from green
(both variables have low values) to violet (both high) while colours such as red and
blue indicate negative correlations (one is high and the other is low or vice-versa).

Results
Correlation analysis indicates a moderate positive aggregate correlation between
IDEB and the Urban Housing Conditions (D3) and Urban Collective Services (D4)
dimensions of IBEU as well as with the socioeconomic status across Brazil (Table 2),
while students socioeconomic status correlates better with ENEM than with IDEB.

Table 2 Aggregate correlations with IDEB

Index Dim. r p
Resident/bathroom density D3 0.54 0.00E+00
Socioeconomic status 0.48 0.00E+00

1
The R code for calculations is available at https://github.com/RenatoPdosSantos/UrbanWelfare-
SchoolInequalities
 Household density D3 0.42 0.00E+00
Open sewage D2 0.42 0.00E+00
Road gully D5 0.40 0.00E+00
Waste collection D4 0.37 0.00E+00
Wall material D3 0.37 0.00E+00
ENEM 0.35 0.00E+00
Sewage treatment D4 0.35 0.00E+00
Electricity supply D4 0.32 0.00E+00
Address identification D5 0.30 0.00E+00
Paving D5 0.30 0.00E+00
Curb / Guide D5 0.28 0.00E+00
Water supply D4 0.19 0.00E+00
Accumulated garbage D2 0.19 0.00E+00
Public lighting D5 0.18 0.00E+00
Sidewalk D5 0.16 0.00E+00
Home-to-Work Displacement D1 0.14 0.00E+00
Arborization D2 0.10 7.28E-13
Subnormal clump D3 0.04 6.63E-03
Wheelchair ramp D5 0.03 4.18E-02
Species of households D3 0.01 4.79E-01
Source: This research.

Table 3 Aggregate correlations with

IDEB
Dimension r p
D3 Urban Housing 0.51 0.00E+00
Conditions
Socioeconomic status 0.48 0.00E+00
D4 Urban Collective 0.41 0.00E+00
Services
D5 Urban Infrastructure 0.39 0.00E+00
ENEM 0.35 0.00E+00
D2 Urban 0.34 0.00E+00
Environmental
Conditions
Source: This research.

In the analysis across all the municipalities, choropleth maps display many reddish-
and bluish-coloured regions on the map for some of the dimension of IBEU,
especially for the D1 dimension (Figure 1).
Figure 1 Choropleth map for Ideb D1.
Figure 2 Choropleth map for Ideb D2.
Figure 3 Choropleth map for Ideb D3.
Figure 4 Choropleth map for Ideb D4.
 Figure 5 Choropleth map for Ideb D5.

Conclusions
Our results suggest that some factor is missing in IBEU definition as it seems to not
work well as a predictor for part of the Brazilian municipalities.
Nevertheless, they also seem to confirm the neighbourhood effect (de Queiroz
Ribeiro, 2005) in the sense that the uneven distribution of urban conditions reduces
educational opportunities and engenders social inequalities that are incompatible
with a sustainable society.
References
de Queiroz Ribeiro, L. C. (2005). Segregao residencial e segmentao social: o
efeito vizinhana na reproduo da pobreza nas metrpoles brasileiras.
Cadernos Metrpole, n. 13, p. 4770.
de Queiroz Ribeiro, L. C., Koslinski, M. C., Zuccarelli, C., & Christovo, A. C. (2016).
Desafios Urbanos Democratizao do Acesso s Oportunidades Educacionais
nas Metrpoles Brasileiras. Educao & Sociedade, 37(134), 171193.
https://doi.org/10.1590/ES0101-73302016157093
de Queiroz Ribeiro, L. C., & Ribeiro, M. G. (2013). IBEU - ndice de Bem-Estar
Urbano. Rio de Janeiro: Letra Capital.
dos Santos, R. P. (2017). Impacto do ensino de Filosofia e Sociologia no
desempenho escolar: uma anlise a partir dos microdados do censo escolar do
Inep. Submetido a Pedagogia Em Foco, 12.
Harvey, D. (1971). Social Justice and the City. London: Edward Arnold.
Murphy, W. (2016). colorplaner: A ggplot2 Extension to Visualize Two Variables per
Color Aesthetic Through Color Space Projections.