4 Modelling neighbourhood support in Northern Hesse
4.4 Agent‐based social simulation setup for the case of neighbourhood support
4.4.6 Input Data
Lifestyle
Leading Traditional Mainstream Hedonistic
Desired in‐degree 15 5 5 10
Rewiring Probability 0.2 0.05 0.1 0.2
Probability to link to:
Leading 0.8 0.6 0.6 0.5
Traditional 0.0 0.3 0.1 0.0
Mainstream 0.0 0.1 0.3 0.0
Hedonistic 0.2 0.0 0.0 0.5
Table 7. Expert rating of lifestyle network preferences. Whereas members of leading and hedonistic lifestyles have far reaching networks and thus are assigned a high rewiring probability, people of traditional lifestyles do not. Data is based on Schwarz (2007). Table was adapted from Holzhauer, Krebs, and Ernst (2011)
One of the basic sources of baseline homophily is the geographical space individuals are embedded into. In order to reflect the characteristics of the spatial composition in the network initialisation, we define opportunity sets as the sets of agents located within a certain radius around a focal agent A. Depending on the type of an agent B from A’s opportunity set we construct a network link from B to A with the type specific probabilities shown in Table 7. This process is repeated for all agents from the opportunity set. The opportunity set is stepwise extended up to a maximum radius (depending on the local population density) or until the maximum in‐degree of the focal agent is reached. Then, in order to account for inbreeding homophily, network links are rewired with the type specific probabilities specified in the second row of Table 7. For this global rewiring process the opportunity set is extended to the complete agent population, i.e. rewiring introduces some long‐distance links that are unbiased by local restrictions.
For further details of the described process of spatial and social network initialisation, refer to Holzhauer, Krebs, and Ernst (2011, 2012).
socio‐empirical dataset that is used for agent initialisation. An overview of the datasets was given previously (Krebs et al., 2011). Here, we extend especially the description of the socio‐
empirical dataset.
4.4.6.1 Climate and Weather
The weather data driver is based on a climate projection generated by the regional climate model CLM (Climate Local Model; Hollweg et al., 2008) based on the scenario A1B of the IPCC (Intergovernmental Panel on Climate Change). In contrast to statistical models CLM is dynamic in that it allows for the emergence of entirely new weather conditions compared to historical weather data. For this reason and others all projects in the KLIMZUG‐Northern Hesse network use the CLM data as a climate reference (Matovelle, Simon, & Rötzel, 2009b, Matovelle, Simon, & Rötzel, 2009a).
The full dataset comprises air temperature, wind speed, rainfall, snowfall, and cloud cover with a temporal resolution of 1 or 3 hours. We use the air temperature data stream to obtain daily maximum temperatures for the investigated region. Figure 20 shows the predicted maximum daily air temperatures as spatial means for the entire region. Heat days with a maximum air temperature of 30° are marked in red; numbers quantify the count of heat days in one year. We observe a high variability in the numbers of heat days per year as well as the fact that heat days are only expected from May to September.
Figure 20. Weather sequence for the target region and the selected climate scenario in daily resolution Source: Krebs, Holzhauer, and Ernst (2011).
4.4.6.2 Spatially Referenced Socio Demographic Data
The described climate data mainly reflects the physical context in which agents make their decisions in the model. Data may straight forward be integrated in an ABSS by allowing agents to perceive temporally varying weather conditions which in turn allows them to assess the effective importance of neighbourhood support at any given point in time. It is far more challenging to initialise the agent population in way that it reasonably reflects the social situation characteristics found in the target area. A possible approach to structuring the target population is the concept of sociological lifestyles (Bourdieu, 1984) that clusters individuals typically in relation to their attitudes, values and orientations.
We apply the Sinus‐Milieus® (Sinus Sociovision, 2007) that are commonly used in commercial market research, but also in environmental research (Gröger, 2011; Schwarz
& Ernst, 2009). Sinus‐Milieus® group individuals or households along the classical dimension of social status given by income and education, and supplement this grouping by a second dimension that reflects social value orientations like tradition, modernisation and re‐
orientation. Figure 21 displays the classification of the ten Sinus‐Milieus® for Germany:
Establisheds are self‐confident and think in terms of success and feasibility, while Modern Performers are the young and unconventional elite. Postmaterialists have liberal and postmaterial values, intellectual interests. The old German educated class finds itself in the Sinus‐Milieu® Conservatives with humanistic values and cultivated forms. Traditionals prefer security and orderliness, while GDR‐Nostalgia believe in socialist visions of solidarity and justice. The modern mainstream aims at professional and social establishment and is very status‐oriented, while consumption‐materialists feel socially discriminated and aspire to the consumption patterns of the Mainstream. Experimentalists are very individualistic and see themselves as lifestyle avant‐garde. Pleasure seekers have a low social status and refuse to accept the expectations of a performance‐oriented society.
Figure 21. The ten Sinus‐Milieus® for Germany and their aggregation to four milieu groups.
Milieus located in the upper region of the diagram are characterised by higher levels of education, more income and belong to upper occupational groups. From left to right milieus increase in their degree of modernisation and individual innovativeness. Source: Sinus Sociovision (2007) adapted.
Figure 21 also shows that there are smooth transitions between the individual milieus.
Therefore, in order to stress distinctions between milieus it is common to cluster the ten Sinus‐Milieus® into four different milieu groups (Gröger, 2011; Schwarz & Ernst, 2009) as displayed in Fig. 3: leading lifestyles, mainstream lifestyles, traditional lifestyles, and hedonistic lifestyles. Additionally, for the context of neighbourhood support we expect only minor differences between milieus within one group especially when used in an ABM.
To make the Sinus‐Milieus® usable in a spatial context Sinus Sociovision and Microm®
(Micromarketing Systeme und Consult GmbH ‐ Microm Consumer Marketing http://www.microm‐online.de) merged the Sinus‐Milieus® with the MOSAIC database.
MOSAIC is a geodemographic segmentation system, i.e. a multivariate statistical
classification technique for inferring if the individuals of a population fall into different groups by making quantitative comparisons of multiple characteristics. The basic assumption is that the differences within any spatial neighbourhood should be less than the differences between neighbourhood groups.
The merged dataset of spatially referenced socio demographic data for 2007 and temporal extrapolations until 2030 is available to the project. For the target region, the extrapolations do not show dynamics that can be expected to have major influence in the context of neighbourhood support. Therefore, we use the 2007 empirical data as the socio‐empirical base for the results presented in this paper. The geographical reference units are so‐called market cells that comprise several hundred households. For each of the market cells the dataset provides the number of households belonging to each of ten different Sinus‐
Milieus®. Again, for the use in the model we cluster data obtaining four milieu groups. In the following we refer to milieu groups simply as lifestyles.