Seite - 165 - in Biodiversity and Health in the Face of Climate Change

165 it is relevant to take the functional role of the area within that ecosystem into account. For example, the redevelopment of a park may either be beneficial or det- rimental to this function. We will focus on the residential environment for a moment. This environment is defined in very different ways. In some studies it is defined as an administrative unit, such as a census tract or postcode area. In other studies, the residential area is defined as a buffer around the resident’s home. In the latter case, the buffer sizes that are used vary considerably, from 100 metres up to 3  km (Egorov et  al. 2017; van den Berg et  al. 2010). There are no clear rules for the most appropriate definition to use. However, using the boundaries of administrative units may be considered rather arbitrary from the perspective of a citizen’s lived experience. A very nearby green area that is located just on the other side of an administrative boundary may be as relevant as a green area within one’s own administrative area (which might even be located further away). Furthermore, administrative units may not all have the same size, which may introduce confounds. As for using buffer sizes, it may be argued that the optimal size depends on the mobility of the population, or the population segment, at hand. For example, when focusing on physical activity during outdoor play without adult supervision by chil- dren below the age of 10  years, then in many contexts using a buffer size of 1  km or more does not seem very sensible; parents usually do not allow their young children to play that far away from home on their own. Using a ‘wrong’ buffer size is likely to lower the strength of associations. If too large, irrelevant natural areas or natural elements are included; if too small, relevant natural areas/elements are ignored. 8.2.3 How to  Measure Distance? With regard to the use of distance in buffer approaches, there is also the issue of whether this should be Euclidean distance or network distance. Accessibility depends more on network distance than on Euclidean distance, since in the latter case barriers may prevent people from travelling in a straight line. However, net- work distances depend on the mode of transport. The network for travelling by foot may be quite different from that for travelling by car. Stairs, lawns and small alleys may be accessible or crossed by foot, but not by car. Incomplete networks can easily lead to an overestimation of network distance for some people, and in this way introduce a source of error. Nowadays, some researchers also take vertical distance into account (Jim and Chen 2010). A person living on the 20th floor of a high-rise residential building first has to get to the ground level, before getting out of doors (except for balconies and roof gardens, of course). When small buffer sizes are con- sidered appropriate, taking vertical distance into account may make a substantial difference. 8 Biodiversity in  the  Context of  ‘Biodiversity  – Mental Health’ Research