User:Leecca01/sandbox

Criminology and MAUP
Crime data are sometimes collected as points in space, but in order to perform some analyses point data need to be aggregated to areal units. The areal unit chosen for aggregation can vary. Criminology researchers and crime analysts often use jurisdictional administrative boundaries to frame their analyses, which poses a problem; the units are not suitable for research because they delimit the full geographic distribution of crime. Administrative boundaries tend to follow physical landscape features without concern for grouping similar populations. These boundaries typically do not change over time, even if the area they encompass has changed. In all social science disciplines, researchers want to minimize variance within the areal units and maximize the variance across the areal units for the variables in the analyses.

MAUP can occur when no relationship is shown between adjacent regions. Nearest Neighbor Analysis can mitigate this problem but cannot be applied if neighbors are outside of the mapped area or if data from these areas are unavailable. Data can also be aggregated to reduce the effects of MAUP. For instance, when analyzing crimes in cities in the U.S., metropolitan boundaries should be used because police only have authority for those areas. When a single jurisdiction is used for analyzing crime, the crimes that occurred in the neighboring jurisdiction will not be available to contribute to the analysis. When data from the surrounding jurisdictions are included, the analysis will determine that the offender is further from the center of that jurisdiction.

Within the jurisdiction, changing areal units to aggregate changes how point data are grouped, which changes the descriptive statistics. When a larger geographic unit is used, variables will become more statistically related to each other from regression to the mean. Point data should be aggregated to several different areal units to see how the spatial statistics results are affected in order to choose the best areal unit for analysis. When comparing the results of analysis from several geographic unit aggregations the units that provide the minimal amount of variation should be used, otherwise the data will be skewed and the distribution non-normal. Additionally, explanatory variables should be chosen based on the level of aggregation because they may not be significant at different aggregations.

When data are aggregated into areal units, the modifiable areal unit problem will always occur. Edge effects can influence the outcome of an analysis when data from adjacent areas are unavailable, and different aggregations and areal units can impact how data are grouped, which affects the distribution. Criminologists must mitigate these problems by considering the way data are aggregated and how different spatial units impact their analyses.