Urban Design, although it deals with the built form of the city, is in many ways a social science, a soft science. The intent of urban design is usually to produce a social outcome of people walking, shopping, enjoying or just engaging with their community in a positive way. Social outcomes are sometimes hard to measure, making it difficult to prove that design of urban spaces is really adding value, or having an effect one way or the other.
There are theories about how the spatial attributes of our environment affect us, as living organisms and as reasoning beings. In terms of design, normative theory is business as usual, doing things in the way that they have been done previously and has been shown effective, up to the present. Sometimes a practice is continued for so long that the original reason for it has been forgotten, and may no longer exist. We look for evidence that something works for us or not: Personal experience, or anecdotal evidence; a body of observations, which are often used in the next level of evidence; scientific theory and testing of hypotheses. It's a circular process - the new evidence is internally incorporated, informing intuitive logic. It then becomes the new standard, the new normative theory, until the process rolls around again.
In the soft sciences, it's easy for investigations to dissolve into indeterminate squishiness. I don't really understand qualitative analysis that well; it seems to me that quantitative proofs are more convincing. The investigative model has to be developed thoughtfully and methodologically. Measures have to be developed, metrics determined, mathematical descriptors defined; this is probably the most difficult part. Not being mathematically inclined, I look for descriptors that others have been clever enough to define, trust that they are correct, then think of interesting things to do with them.
I wished to relate a social behavior to mathematical descriptions of the geometrical qualities of spatial fields, the hypothesis being that there would or would not be a statistically significant relationship. I wanted to prove that the design of space matters (or not). I chose opportunistic crime as the behavior, the dependent variable, because crime reporting is robust and crimes of opportunity are likely to be influenced by environmental factors. I geocoded the crime data into hot spots within census boundaries. I correlated it with a variety of other variables, one at a time, from census and GIS data that were readily available.
As might be expected, the significant demographic variable was income, with a strong negative relationship. No other variables were significant in direct correlation. Crime and income were transferred into multiple regressions with other variables. Of built form variables, building density was significant in relation to income and crime. It is very important to test for collinearity when introducing multiple variables, as they are often related to and influencing one another. Building density was the only significant variable at this second level, which looked at the regional scale of the census tract.
Geometric measures of isovists, or visual fields, were introduced in the next step. These tend to be very colinear and only one could be introduced at at time. To cut the telling short, statistically significant relationships were found. There are qualities of spatial form that do affect social behavior, or at least one behavior. Next I'd like to show a relationship to some more positive behavior, but as people don't report happy things as reliably as they do criminal activity, it would take a lot of hours of direct observation which I don't have the time or resources for.
There are theories about how the spatial attributes of our environment affect us, as living organisms and as reasoning beings. In terms of design, normative theory is business as usual, doing things in the way that they have been done previously and has been shown effective, up to the present. Sometimes a practice is continued for so long that the original reason for it has been forgotten, and may no longer exist. We look for evidence that something works for us or not: Personal experience, or anecdotal evidence; a body of observations, which are often used in the next level of evidence; scientific theory and testing of hypotheses. It's a circular process - the new evidence is internally incorporated, informing intuitive logic. It then becomes the new standard, the new normative theory, until the process rolls around again.
In the soft sciences, it's easy for investigations to dissolve into indeterminate squishiness. I don't really understand qualitative analysis that well; it seems to me that quantitative proofs are more convincing. The investigative model has to be developed thoughtfully and methodologically. Measures have to be developed, metrics determined, mathematical descriptors defined; this is probably the most difficult part. Not being mathematically inclined, I look for descriptors that others have been clever enough to define, trust that they are correct, then think of interesting things to do with them.
I wished to relate a social behavior to mathematical descriptions of the geometrical qualities of spatial fields, the hypothesis being that there would or would not be a statistically significant relationship. I wanted to prove that the design of space matters (or not). I chose opportunistic crime as the behavior, the dependent variable, because crime reporting is robust and crimes of opportunity are likely to be influenced by environmental factors. I geocoded the crime data into hot spots within census boundaries. I correlated it with a variety of other variables, one at a time, from census and GIS data that were readily available.
As might be expected, the significant demographic variable was income, with a strong negative relationship. No other variables were significant in direct correlation. Crime and income were transferred into multiple regressions with other variables. Of built form variables, building density was significant in relation to income and crime. It is very important to test for collinearity when introducing multiple variables, as they are often related to and influencing one another. Building density was the only significant variable at this second level, which looked at the regional scale of the census tract.
Geometric measures of isovists, or visual fields, were introduced in the next step. These tend to be very colinear and only one could be introduced at at time. To cut the telling short, statistically significant relationships were found. There are qualities of spatial form that do affect social behavior, or at least one behavior. Next I'd like to show a relationship to some more positive behavior, but as people don't report happy things as reliably as they do criminal activity, it would take a lot of hours of direct observation which I don't have the time or resources for.
Leave a comment