In my thesis I developed, based on the German Socio-Economic Panel, a procedure that calculates optimal neighborhood profiles for any subpopulation. The neighborhood profiles are based solely on physical and social infrastructure characteristics. The residential location needs of individual households were grouped so that households with a similar profile were combined. Conversely, it is also possible to calculate an optimal neighborhood profile for any predefined populations. The calculated profiles can be compared with the profiles of any given urban residential location and the percentage deviation from the optimal profiles of the populations are given. The developed procedure provides the possibility to analyze the environment that surrounds an individual. The method enables one to analyze not only one interrelation but provides the possibility to analyze – because of its openness – the meso-environment as a complex system (see Eckardt, 2009, p. 187ff) and therefore provides a better understanding of urban structures. This understanding can contribute to future urban systems, like the ones being developed e.g., by the Fraunhofer Society in the “Morgenstadt” project, by the SENSEable City Lab at MIT or the Zurich Future Cities Laboratory, so that the development of urban structures is not an end in itself but is fitted to the citizen’s needs. This study closes a data gap and contributes to the quantitative framework that Bettencourt and West (Bettencourt, Luis M. A. et al., 2010; Bettencourt, Luis M. A. et al., 2007) have in mind.
Problem and objective of the work
More and more people are moving to cities. According to the United Nations data set, in 2010 3.6 billion or 51.6 % of the world’s population already lived in in an urban agglomeration, so from an ecological point of view the city is a central issue for the future. The focus is in particular on the question of how future cities must be designed for a sustainable and livable life. Many authors argue for an increased urbanization (cf. Glaeser, 2012). Kent Larson argued in his presentation at the TEDx Boston 2012, that the cities should be structured in a way that all daily errands can be completed within a 2 kilometer radius. This raises the question of how such a district must be designed so that people feel comfortable. Not all people have the same ideas of what makes a good neighborhood. There are only a few studies about residential needs for German cities. Unfortunately these studies only address very limited local districts in selected cities. There is no study that addresses the German resident population. The dominant paradigm in Germany assumes that the requirement of the population for their neighborhood is rather uniform. However, a closer observation shows that not everyone is the same, but different people have some very different ideas about what makes a good neighborhood. The aim of this study was therefore to develop a methodology that can calculate the optimal neighborhood requirement profiles for any given subpopulation.
State of science
Keim (1979) and Matthiesen (1998) already showed that milieu differentiation in urban research can provide a deeper understanding of social structures and processes in urban agglomerations (see also Breckner 2003; Breckner 2004). In the last few years these opportunities were recognized in urban research and development (cf. Müller-Schneider, 1996) and taken into account (cf. Scheiner & Holz-Rau, 2007; Scheiner, 2005; Huss, 2006; Matthiesen, 2007, 99f; Breckner, 2009). Scheiner and Holz-Rau come to the conclusion in their study that the lifestyle in the residential location choice plays a crucial role (Scheiner and Holz-Rau, 2007). Becker also states that for the residential location choice the respective living conditions are of great importance (Becker, 2003). The meso-environment, which is dominated by the physical infrastructure and the various supply structures, is especially essential for a location decision. This is also supported by a study within the federal-state program “Social City” (cf. Böhme, Christa et al., 2003) where 88 % of respondents declared that deficits in the residential environment would be a major problem for the neighborhood. These deficits become even more dominant in neighborhoods where the residents depend on the meso-environment. For children and elderly people the structures in the immediate vicinity are of major importance. Therefore, Kabisch stated that in the context of urban redevelopment the site-specific residential needs, specifically those of the elderly, must be taken into account. However, other population groups have specific needs for the residential location as well. If the needs of the diverse residential groups are not taken into account, the whole process of urban renewal can be stalled (Kabisch, Sigrun et al., 2007). Therefore more and more cities explore the question, such as the Lucerne research project “housing preferences”, which factors influence people in their residence decision. In this study, the population is polled with quantitative and qualitative interviews about their residential location preferences (Delbiaggio, 2010). Schmitt, Jürgen et al. (2006) also rely on a comparative study of living orientations on qualitative interviews. All studies have in common that they only cover a very limited geographical area and therefore cannot make a representative statement about the residential preferences in general.
Based on a general milieu theory, we assume in this study the existence of social milieus and lifestyle types. The milieus were defined as groups of people that share the same lifestyle and outlook on life. Based on Bourdieu’s theory (Bourdieu, 1982; Bourdieu, 1991), which states “the social environment has a tendency to form the physical space, more or less strictly, resulting in the form of the distributional arrangement of actors and attributes. Therefore, all distinctions, according to the physical space, can be found in the objectified social environment”, we assume that the different ways of life can be identified in the residential location decision or, to be more precise, is important for satisfaction with the residential location (hypothesis 1). The district as a habitat is characterized in particular by the existing infrastructure. A residential location can therefore be represented by a characteristic infrastructure combination, which form a specific neighborhood (hypothesis 2). Based on Bourdieu’s assumption, the neighborhoods should not only differ by their composition in terms of occurring characteristics of the infrastructure, but also in the composition of the particular milieu structure in a neighborhood (hypothesis 3).
With the Socio-Economic Panel (SOEP) from the German Institute for Economic Research (DIW) (see Wagner et al., 2008) there is a longitudinal data source available from which different subpopulations can be extracted, and also contains every four years a questionnaire about the neighborhood and neighborhood structure of the respondents. Due to the high complexity of the information and the interdependence of the individual infrastructures there has been no attempt to evaluate this data so that for any subpopulation the optimal neighborhood profile could be calculated, in which the subjective satisfaction with the neighborhood reaches its maximum. To solve this problem the data has been projected with the singular value decomposition in a subspace of reduced complexity and analyzed by a customized correspondence analysis approach based on Faust (2005). The approach was originally developed for the analysis of affiliation networks, however, if we understand a neighborhood as the sum of infrastructure characteristics that are related to the inhabitants, we can describe the relationship as a bipartite graph, which can be analyzed using this method. The advantage of Faust’s approach in contrast to the canonical correspondence analysis is the use of a symmetric representation of the point values. This allows us, in contrast to the asymmetric representations, to perform an inter-dimensional analysis which is the prerequisite for a cluster analysis. In a second step distance measures were used to determine these infrastructures that have a special meaning in the particular dimensional combination. For this purpose, a significance horizon was calculated around the centroid of each dimensional combination with a radius of twice the average deviation of all infrastructures to the centroid. Infrastructures that are beyond this threshold were considered to be significant in the sense that these infrastructures were disproportionately important for the differentiation of the state space. Afterwards, the typical subpopulation groups of a neighborhood environment were isolated using a two step clustering method. In the first step the Ward’s algorithm was used to find the cluster centers which had been optimized in a second step using the k-means algorithm. Finally, the results of each dimensional combination were projected back to the original data set and the respective neighborhood profiles determined.
The results of the analysis that are described in this study show that a population can be separated based on the infrastructure combination in the meso-environment. Therefore this thesis provides representative quantitative evidence for the German population in cities for Bourdieu’s assumption that there is a connection between the physical and social space according to hypothesis (1) of this thesis. It was also possible to show that urban planning cannot be reduced to a few factors. All examined variables had a significant proportion of the variable combination of the analyzed matrix. This means that a single variable is not sufficient to localize a subpopulation in the physical space, but the meso-environment constitutes a high complex interdependent structure of infrastructure combinations. These infrastructure combinations are crucial to whether or not a particular population group lives in a location. It is therefore not enough to focus on individual characteristics, rather, as formulated in hypothesis (2), a specific setting of factors that are preferred in different ways by the various population groups. However, the setting doesn’t have a significant difference between the neighborhood types according to the micro environment variables. This means that every neighborhood type, as well as in each cluster, include all different forms of building structures, family structures, employment stages, occupational status and income levels. Also, no significant differences in the rent structure could be found. A residential location preference regarding the infrastructural supply in meso-environment can therefore, regardless of financial resources, in principle, be met in a German city. Looking at the neighborhood environment profiles of the different milieus and lifestyles, it could be shown that the neighborhood environment profiles differ significantly from each other. However, all milieus can be found in all neighborhood types. This may be due to the method used. The procedure covers all locations with the same infrastructure profiles in the meso-environment. However, the sites must therefore have no physical relation to each other. Instead the described neighborhood types represent a fundamental space typology of urban space in which the residents feel comfortable. Therefore, the assumption that was described in hypothesis (3) can partially be confirmed, as people from a similar life-world context prefer a similar residential district. It can therefore be assumed that there is no unified idea of a good neighborhood that meets the needs of all population groups, but that this idea is dependent on social and cultural factors. The results that were presented in this study represent another explanation for segregation and quantify the influence of the different infrastructures to the preferred residential area of the respective populations. Based on the given infrastructures, the procedure produced results that highly matched the SINUS milieu characteristics. It can therefore be concluded that the results of this analysis confirm the assumptions that segregation is mainly based on social and cultural differences. The presented method describes a way to calculate population-specific neighborhood profiles based on geographical and infrastructural characteristics. The result is a profile in the form of a vector which describes the geographical and infrastructural characteristics within walking distance of a location that are necessary for a specific subpopulation to be satisfied with the residential location. In reverse, this vector could also be used to evaluate residential areas in any German city in terms of their attractiveness for a subpopulation. Such a residential location preference map was developed as an example, based on the described method for the city of Dresden. For the residential location preference map of Dresden the infrastructures used were extracted from the open source project open street map OpenStreetMap (2012). Incomplete or missing information were georeferenced and added from additional sources. A shortest-path-tree analysis was applied to the dataset and the results afterwards compared with the neighborhood profiles of the different milieus. The result is a residential location preference map of Dresden, which specifies for each address point for which milieu this address is the best fit. The subpopulation representation for Dresden can be found here.
Approaches for further work
The results of this residential location preference map will be the subject of further research. The goal is to automate the process, so that for cities where sufficient OpenStreetMap information is available such a residential location preference map can be generated automatically. For this purpose, appropriate programming interfaces (APIs) for the used data sources and a uniform procedure need to be developed. The next step, therefore, is the implementation of the described approach into a R function that can handle any arbitrarily complex binary data matrix as input and produces as output the respective adjacent profiles. Besides the practical application of the results of this study, there are also some other issues which analysis would be interesting and important for the understanding of migration decisions in Germany. In this work, only the year 2004 could be considered, but it would also be interesting to analyze the changes over several years. For this, the survey results from 2009 are also available now. The next survey will be applied in 2014 and is expected to be available in 2015. To analyze the long-term development, the presented method would have to be expanded to include functions for longitudinal analysis. This would not only be interesting to see how the neighborhoods change over time, but also whether the results are stable over the years. Furthermore, only the neighborhoods of those residents were considered in this study, who were satisfied with their surroundings. It would also be possible to examine the households who are not happy with their surroundings, and to compare the neighborhood profile of these respondents with those of satisfied households. Moreover, it would also be possible to include various indicator variables such as satisfaction, size of city, etc. into the analysis. In addition, the model is not limited to the milieu environment, but can in principle be applied to any subpopulation which can be derived from the SOEP. It is also possible to analyze the the population based on other indicators e.g., health indicators. This will be explored in the context of a research project at the University of Stuttgart. The described method will be used to analyze the relationship between neighborhood structure and health-promoting behavior.
Becker, M. (2003). Lebensqualität im Stadtquartier. PhD thesis, Albert- Ludwigs-Universität zu Freiburg i. Br., Freiburg i. Br.
Bettencourt, L. M. A., Lobo, J., Helbing, D., Kühnert, C., and West, G. B. (2007). Growth, innovation, scaling and the pace of life in cities. Proceedings of the National Academy of Sciences of the United States of America, 104(17):7301–7306.
Bettencourt, L. M. A. and West, G. B. (2010). A unified theory of urban living. Nature, 467:912–913.
Böhme, C., Becker, H., Meyer, U., Schuleri-Hartje, U.-K., and Strauss, W.-C. (2003). Handlungsfelder integrierter Stadtteilentwicklung, chapter 5. Deutsches Institut für Urbanistik (Difu), Berlin.
Bourdieu, P. (1982). Die feinen Unterschiede. Suhrkamp Verlag.
Bourdieu, P. (1991). Physischer, sozialer und angeeigneter physischer Raum. In Wentz, M., editor, Stadt-Räume. Campus Verlag, Frankfurt a. M.
Breckner, I. (2003). Künftige Anforderungen an Wohnumfeld- und Freizeitqualitäten in den Städten von Nordrhein-Westfalen. Bericht für die Enquêtekommission “Zukunft der Städte in NRW” des Landtags Nordrhein-Westfalen. Technical report, Landtag Nordrhein-Westfalen, Hamburg.
Breckner, I. (2004). Eliten, Minderheiten und soziale Milieus als regionale Entwicklungsressourcen. In Hönekopp, E., Jungnickel, R., and Straubhaar, T., editors, Internationalisierung der Arbeitsmärkte, volume 282 of Beiträge zur Arbeits- und Berufsforschung, pages 209–230. Institut für Arbeitsmarkt- und Berufsforschung der Bundesagentur für Arbeit., Nürnberg.
Breckner, I. (2009). (Un-)wissen im Handeln urbaner Milieus. In Matt- hiesen, U. and Mahnken, G., editors, Das Wissen der Städte. VS Verlag.
Delbiaggio, K. (2010). Gründe für ihren Wohnorts- oder Wohnungswechsel.
Eckardt, F. (2009). Die komplexe Stadt. VS Verlag, Wiesbaden.
Faust, K. (2005). Using correspondence analysis for joint displays of affiliation networks. In Carrington, P. J., Scott, J., and Wassermann, S., editors, Models and Methods in Social Network Analysis, chapter 7, pages 117–147. Cambridge University Press, New York.
Glaeser, E. (2012). Triumph of the city. Penguin Group, New York.
Huss, E. (2006). Mosaic Milieus für München. Münchener Statistik, (4. Quartalsheft.):3–10.
Kabisch, S., Peter, A., and Bernt, M. (2007). Stadtumbau ost aus sicht der Bewohner. Informationen zur Raumentwicklung, 1:S. 37–47.
Keim, K.-D. (1979). Milieu in der Stadt. Ein Konzept zur Analyse älterer Wohnquartiere. Kohlhammer, Stuttgart.
Matthiesen, U. (1998). Die Räume der Milieus. Neue Tendenzen in der sozial- und raumwissenschaftlichen Milieuforschung, in der Stadt- und Raumforschung. edition sigma, Berlin.
Matthiesen, U. (2007). Wissensmilieus in hetrogenen Stadtregionalen Räumen Ostdeuschlands – zwischen Innovationsressourcen und kulturellen Abschottungen. In Koch, G. and Warneken, B. J., editors, Region – Kultur – Innpvation, pages 83–122. VS Verlag.
Müller-Schneider, T. (1996). Wandel der Milieulandschaft in Deutschland. Von hirarchischen zu subjektorientierten Wahrnehmungsmustern. Zeitschrift für Soziologie, 25(3):190–206.
OpenStreetMap (2012). Main Page — OpenStreetMap Wiki,. West Mid- lands.
Scheiner, J. (2005). Lebensstile, Standortbewertungen und Wohnmobilität. Technical report, Universität Dortmund.
Scheiner, J. and Holz-Rau, C. (2007). Travel mode choice: affected by objective or subjective determinants? Transportation, 34(4):487–511.
Schmitt, J., Dombrowski, J., Seifert, J., Geyer, T., and Murat, F. (2006). Einfamilienhaus oder City? Wohnorientierungen im Vergleich. VS Ver- lag, Wiesbaden.
Wagner, G. G., Göbel, J., Krause, P., Pischnerand, R., and Sieber, I. (2008). Das sozio-oekonomische panel (SOEP): Multidisziplinäres Haushaltspanel und Kohortenstudie für Deutschland. AStA Wirtschafts- und Sozialstatistisches Archiv, 2(4):301–328.