Regions that cover a large geographic area are often challenged to produce an accurate point in time homeless count given the limitation of volunteers and the sheer amount of territory that needs to be covered in a relatively short period of time. To address this, Simtech Solutions has engaged the services of statistician Dan Treglia, PhD, to develop and deploy a framework that allows these large regions to conduct a count that produces reliable count estimates without having to canvas an entire area.
The process is referred to as geographic-based sampling, and the steps involved are as follows...
- Regional administrators establish the count areas and assign a count team leader to each area using region and team management in the Regional Command Center.
- Count team leaders use the Regional Command Center to identify the census tracts where it is likely that people experiencing homelessness will be found. Typically, feedback for this step is gathered from those who come in contact with the homeless on a regular basis such as street outreach workers and police officers. Additional information about this process can be found here.
- Regional administrators will begin the sampling selection by using previous PIT results in addition to feedback collected in the prior step to automatically designate areas for high probability.
- Regional administrators will then use the sampling feature to make low probability designations based on the remaining number of tracts. The sampling size is determined by the community based on the desired confidence level and confidence interval while taking into account the number of volunteers. Details on setting the sampling size can be found on this site.
- Maps are produced from within the Regional Command Center for each count team. All volunteers are able to view their assigned team's map from within the Counting Us mobile app.
- The homeless count is conducted using the Counting Us mobile app.
- Once the count is over, any necessary data cleanup is done by regional administrators using the survey management tools found within the Regional Command Center.
- Simtech staff are notified when the cleanup process is over. Once notice is received, statistically reliable point in time reports are produced for the entire region. As was done with the State of Connecticut, these estimates can be produced for regions of all sizes including statewide, by CoC, by county, etc..
- The point in time count results are submitted to HUDHDX.
- Celebrate! Your region just completed a highly accurate homeless count in far less time than it would have taken if you were to use paper!
The State of Connecticut used the geographic sampling features within the Regional Command Center in order to conduct a statewide count for 2017. The Final Methodology Report for Connecticut's Point in Time Count provides a detailed overview of how technology and statistical analysis can be used together to produce more accurate counts with far time and energy.