Any outliers that the outlier analysis could not locate may result in suboptimal block group level model estimates due to suboptimal input UCR data. For example, if a law enforcement agency supplied the FBI incomplete data and the analysis used to detect incomplete and erroneous data are not 100 percent accurate, then the resulting modeled data may be less accurate. Additionally, a large-scale criminal event, such as a riot, may skew the accuracy of average, local crime statistics. As a result of incomplete or skewed UCR data, some block group crime statistics were entirely modeled using imputation. This means that this subset of block groups did not benefit from the macro level UCR crime data and were imputed based only on the crime statistics of similar block groups. Similar block groups were determined based on geodemographic analysis.
Context CrimeIndex is produced by Precisely data scientists. This data product is available for all the US and Puerto Rico block groups. While the Context CrimeIndex data is directionally accurate, the data should be used and interpreted “as is”; crime is often random and sporadic, and a low crime index score does not – in any way - guarantee that crime may not occur.
Data in this product are modelled estimates of how crime indices might roll out at a small spatial scale based on documented crime reporting and incidents. Due to updates in the source data, additional source data used, and changes made to methodologies with each version, it is not recommended that users compare the Context CrimeIndex data with previous versions. For further clarification, please contact Precisely client support .