The Outliers

In the process of a looking at a state as vast in size and as diverse as Texas, it became clear that some counties displayed unique characteristics. Based on the robust regression estimate of expected need, these counties appeared unusual, or as outliers, because the county’s actual service need was significantly different than the model would have predicted. These “outlier” counties showed an issue that was unique to their county, such as a higher than average census prevalence statistic indicating there might be a higher than expected amount of survivors of family violence in a county. This map highlights the unique component(s) that made the given county an outlier and the details of that county’s finding.

To understand this map a little better keep in mind that a positive residual reflects the likelihood of more need and a negative residual reflects a lower than predicted need. For example, in the outlier map you see that Hays County shows a positive residual under ‘HHSC Female’ of 3.72. This shows that there has been more need for services from a family violence program than the model would have suggested. Conversely, if you look at Orange County there is a negative residual of 6.72 indicating a lower number of Uniform Crime Reports on family violence than expected suggesting that targeted outreach to enhance law enforcement response might be warranted.

Click on each county for additional data

Counties with at least one standardized residual greater than 1.96 or less than -1.96 are noted as significant and are contained in this map set.