The maps below show crime rates in each census block group. Red values indicate block groups with the highest number of crimes and blue values indicate the lowest number. These are exactly the same datasets and the block groups in both maps have exactly the same values for crime incidents, but they look radically different.
On the left hand side the map shows a high rate of crime near the downtown area, while the rest of the city looks relatively free of crime. On the right we see a more realistic pattern with higher values encircling the downtown and extending toward the northwest and northeast. Now the difference between the two is the way the values are symbolized.
Symbology and visualization will be reviewed in the next unit, but these two maps represent a very good reason why we need a quantitative technique for determining the pattern of crime in the city. Just by looking at these we come up with two very different conclusions in what is known as an observation bias- that is our judgments are influenced by the appearance instead of by the fact. So instead of relying on the symbology alone we can apply inferential statistical techniques to get a less biased representation of a phenomenon.
There are a lot of tools to use in inferential statistics, but also a lot of parameters that have to be defined before the tools can be used correctly and the situation is complicated further by the fact that there are sometimes no definitive answers on how to define these parameters, that is you might end up with more than one way to define a parameter
But there are a couple of way that feature can interact even on a small scale. In general the feature can interact either by being adjacent to each other as shown on the left, below, or by being within a certain distance as shown on the right. Interacting features are called a "neighborhood," and, so the conceptualization of spatial relation is essentially how we define the neighborhood.
