Accessibility to primary schools in Portugal: a case of spatial inequity?

Social inequalities have shown a tendency for being amplified by the distribution of socio-economic groups and of spatial amenities and disamenities. Given their commitment to principles of spatial equity, it is important to understand if public services fit into this pattern or, on the contrary, provide similar levels of accessibility to different socio-economic groups. This paper analyses the relationship between the location of primary schools and the spatial distribution of socio-economic groups in Portugal and it concludes that there is a systematic tendency for disadvantaged populations to have worse accessibility than advantaged ones, independently of spatial scales and types of territories. Las desigualdades sociales han mostrado una tendencia a ser amplificadas por la distribución de los grupos socioeconómicos y de las ventajas y desventajas espaciales. Debido a su compromiso con los principios de equidad espacial, es importante comprender si los servicios públicos se ajustan a este patrón o, por el contrario, proporcionan niveles de accesibilidad similares a los diferentes grupos socioeconómicos. En este artículo se analiza la relación entre la ubicación de las escuelas primarias y la distribución espacial de los grupos socioeconómicos en Portugal y se concluye que existe una tendencia sistemática a que las poblaciones desfavorecidas tengan peor accesibilidad que las más favorecidas, independientemente de las escalas espaciales y de los tipos de territorios. 社会的不平等は、社会経済的集団及び空間的な利便性/非利便性の分布によって増幅される傾向を示す。公共サービスが空間的平等性の原則を約束するものであることを考えると、公共サービスが、このパターンに適合するかどうか、あるいは反対に、異なる社会経済的集団に同じレベルのアクセシビリティを提供するかどうかを理解することが重要である。本稿では、ポルトガルにおける小学校の立地と社会経済的集団の空間分布の関連性を分析したところ、空間規模と地域のタイプとは無関係に、不利な集団は有利な集団よりも小学校へのアクセシビリティが悪くなる系統的な傾向があると結論づけられた。 Fast-paced territorial transformations have been a defining aspect of many European countries in the last decades. In particular, the polarization of the urban systems led to a growing concentration of the population in the major urban centres (Borja Frey Hall Stratmann, 2011), while more peripheral regions stagnate or decline (Bernt et al., 2012; Gomes, Silva, de Castro, Jaroszewska, 2019; Martinez-Fernandez et al., 2016; Turok Macintyre, Macdonald, Guo, Chang, Chen, Kenyon Fotheringham Openshaw, 1984) the scale at which spatial equity is assessed can also influence the outcomes (Tan a street network, available from OpenStreetMaps for the whole country; and socio-economic data, where the smallest units for which some of the Census 2011 data is available are the statistical subsections (equivalent to census tracts) of the Geographically Referenced Database (BGRI) of the National Statistical Institute (INE). The subsections are the basic unit of this analysis. i) Schools by coordinates for mainland Portugal ii) Street network i) DGEEC ii) OpenStreet Map The qualification index (QI) is calculated for each subsection, based on the following formula: where x are the years in each cycle of studies, y is the number of people in that cycle of study, and i is the statistical subsection. i) Spatially delimited statistical units ii) data for population by educational level for the Census 2011 i) BGRI ii) INE i) Schools by coordinates for mainland Portugal ii) Spatially delimited statistical units with population for the Census 2011 i) DGEEC ii) BGRI & INE As an indicator for socio-economic status, a qualification index (QI) was calculated for each subsection. The QI, which represents the average years of schooling of a given territorial unit's population, can be considered a good proxy measure for socio-economic status. In fact, in the European Union there is a strong relation between the average salary and the educational level of the workforce, and the income of those with higher education is approximately 50% higher than those with only secondary education (Eurostat, 2020). Similar findings were also made by Aisa, Larramona, and Pueyo (2019), showing a negative relation between education and poverty, and Eikemo, Bambra, Joyce, and Dahl (2008), who note that people with higher education perform better in health-related indicators. Regarding accessibility, finding the shortest path between two points on the street network, and connecting them to spatial units for which statistical data is available, is a non-trivial task. Many works in this field use centroids of the statistical units. Macintyre et al. (2008), for example, calculate the distances between resources and the centroids of administrative zones, while Mayaud, Tran, and Nuttall (2019) use an open-source routing engine to generate a time and travel matrix based on grid centroids. Roberts et al. (2020) use the real distances for some household locations, but for the most part also use centroids for the region to calculate the distances to hospitals in India. In this paper, the goal was to calculate distances for the entire country, in contrast with the analysis at the city scale made by those works, while also getting accessibility levels which are not restricted to a single centroid in each statistical subsection—this that the accessibility levels be in but not in ones with But the distance from subsections to schools, at the level, be The distance to the next primary school each (see which also from a public policy given that, with to a primary school in the where For this, was used to the network data by to a with a of a by (2019) Afterwards, & 2012) was used to the network as a as by the in and to calculate the distances from to the nearest school This was and in with the actual being the spatial to in the the points are The data in Table shows the and of the for the different spatial units. The of these is and the are not the since the subsections are the smallest spatial units and it is not to of central tendency or for The number of schools by is also not presented at this scale, because the not have For and parishes, showing central tendency or for the average distance of the subsections to schools, are also presented for the distance of the subsections at each of these spatial are and from a This is in with a more understanding of justice in accessibility, which is concerned not only with average but also with understanding the at the or the of the basic units changes as are As be the of the are at a tendency for units to more The for the QI, for example, is similar for the three scales years of but the of is higher for the subsections than for the parishes, and higher for the than for the The in of the the distances to primary As is shown in there is a negative correlation between the and the distance from the to the nearest This relation also holds when the as a with the to the different of the Portuguese education as shown in The between the average distance to the nearest school of and is of meaning that people between and years of schooling are on average to a primary school than those with less than one of schooling any level of cycle of basic to years of cycle of basic between and years of cycle of basic between and years of school level of between and years of higher education level, more than years of finding show that, in the spatial distribution of primary schools can be considered to fit into a deprivation where less advantaged socio-economic groups also have poorer accessibility while for it This is to be to the that, for people a socio-economic accessibility secondary to the and provided by locations the urban It also shows that, as in the differences in the accessibility levels between socio-economic groups necessarily, imply spatial or in service but can be to and a with the made it to understand if the different of accessibility can be into a and or if different of the statistical were using different for the The were by the following where a than and where the was which led to the of the of the The with the two which a of and of the for and municipalities, is with the level of to primary schools and has for the three different of distances and while has for the schools The of the analysis to and are the nearest school is with for parishes, and is to for it is to as the distance to primary schools and as the of analysing the of each in each spatial it is to two territorial for the is between the with higher levels of accessibility, and which have lower for the is between the of the country, which tends to have schools by and the were the is generally territorial patterns can be for as as for As by the in the of the distribution of the tends to be for the parishes, the that, at this level, there are in the accessibility the spatial scale the heterogeneity of the spatial The of the of of the distribution is a statistical since the following statistical analysis should use are the of for In this context, a was to a of from the of accessibility and, to use the for are important if are at different which is the (see & the use of the problems of of some when different of distance are For defining the clusters, different and the spatial units were using the means of and distances to measure the level of between This is one of the most and the most used et al., Table the central and the of the in the types of territorial units were This is the and the to was when the territorial patterns of the were analysed, the spatial of the is different for and more and of each is presented This which of the is by territories and with low levels of accessibility, by distances and of The in the are average that the distances are and the of the of schools are analysing the of the it is to different spatial for the different scales. The to this are the regions of the country, but with a in the The are also across the but with in the This the territories with a number of schools and accessibility levels and represents of the for and The of this are urban in the two regions and and in the For the parishes, the spatial pattern of this is more but to be in and of the The territories to this have a number of schools and distances to schools It is the including approximately of the spatial units considered in the analysis of the and of the and it have a spatial pattern but is more in the of the areas and, for municipalities, in the central This and that have many schools but the distances to those primary schools are also This of the and only of The spatial units to this are more in the of the and, for parishes, in the while for to be in the As was the different accessibility levels in the can be with territorial that patterns of land or the distribution of different population groups. But are the inequalities in the accessibility to primary schools a of these different territorial For this the was with accessibility levels for the different spatial at different scales. types of outcomes are an analysis of the relationship between the and the two of accessibility in the analysis an analysis of the of accessibility by the socio-economic groups, for the As shown in the of there is a strong and correlation between the years of schooling at a given and the distance to the nearest school by the of correlations are for parishes, where the correlation is while for it is the of schools there are meaning that the number of schools is regarding the socio-economic of different territorial units. Regarding the importance of different spatial shows that there is a in accessibility levels by socio-economic groups and, most that this relation holds for different types of The most groups are from schools and the more groups are in and for the two administrative units. a more analysis to that the scale has some on the relationship between accessibility levels and socio-economic groups. In fact, while for the different types of territories have the pattern of accessibility, for there are some differences between and, in some of the clusters, the differences between socio-economic groups are not as In particular, the differences between socio-economic groups are higher in the where the two of accessibility are more that territories where the to schools and the number of schools are average What can be from the accessibility to primary schools with the socio-economic status of the population at the for different types of territories and for territorial The finding it that there is a relationship between the socio-economic level of the population and the level of In fact, the distance to primary schools is with the level of schooling of the population by the qualification and this relation can be for municipalities, for parishes, and subsections of the whole finding is that the correlation holds for different territories by different levels of with the being from primary schools in low territories as as in the ones where are to the findings of Talen (2001), a inequality was found in the location of primary different types of territories and different scales and showing that the inequalities in accessibility exist in of this paper also ecological fallacy or modifiable areal unit But this to of in the accessibility to primary a Rawlsian their location shows of spatial some aspects of the analysis to a more differentiated the relation between accessibility levels and socio-economic is not since the people with the level of education to be at a distance from schools than those which are in educational for the number of schools by relation was found with the This means that the inequalities in the provision of primary schools are restricted to the distance to schools and not the in territories. there are some and in this the spatial of social groups is to of spatial but territories are and, to a it is that social aspects are in this This means that it is to between differences which territorial and those that from an unjust distribution of resources an analysis of the way the and different aspects of spatial inequalities for example, & 2019). the provision of public services, in also has an on the social of the areas which are to it. and for example, found that the to good schools is to higher the of an public good by to schools can, in this also a spatial of socio-economic groups. But the systematic way in which less advantaged people have poorer lead to that it is or of spatial planning In rural it be to the tendency for public services in central locations, which also to have a of the better For these types of the closure of many schools in the more locations in the last is to these spatial inequalities Marques, & 2020). In the more urban the inequalities can be considered the of a tendency for public and private services in central locations, according to a general planning which an of and some of the less advantaged areas as these there are some and in this which in particular, are it be to the different patterns of inequality more in to them more from the territorial This be by or according to other territorial such as population or in to accessibility using the as a proxy for socio-economic while being and by the is also a It be to this in understanding the way in which these different are to spatial accessibility (see, & or & for a more discussion of socio-economic accessibility was assessed in distances to the nearest that different of have different to as as different them provide an to this analysis. In this such travel or service be the socio-economic were only available for since been changes in the such as the of regions or the of and more The next Census data will to understand these but it is not that have a on the distribution of socio-economic at the local This is the of the research of two Spatial for and of urban are by the research unit on and Policy and by and by a a