Christa Polczynski Olson refused to provide the geocoding data so that we could verify the errors and insure that major crime is being properly counted.  The report header states "Bridgewater State College" and our emails to department leaders Harriet Hayes and Timothy Brazill have been unanswered.

Summary of Issues with papers published by IFC regarding crimes related to the relocation of the Community House Men’s Shelter

Primary Issues with “Crimes around the IFC Community House” IFC Paper

1. Every map in this paper shows that there is a hotspot of criminal incidents at the current shelter.  The author acknowledges in the conclusion that she cannot disprove that this crime was committed by shelter residents.
2. The paper does not conclude that the shelter is crime-free.  The paper merely shows that there are other significant hotspots of crime in Chapel Hill.  In fact, the shelter is a prominent hotspot for violent, sexual, property, non-violent, and ordinance crimes in the paper’s figures.
3. The star on 100 W Rosemary obscures the coloring of the map in figure 1, 2, and others such that the purported maximum density of crime is hidden.
4. There appear to be other possible issues related to problems that the author may have had when geocoding the addresses. We have asked her for the geocoding data so that we can verify it, but the author has refused to provide it.
   

Primary Issues with “Crimes around the Proposed Location for IFC Community House” IFC Paper

1. The paper places the Homestart shelter on Homestead Rd in the wrong location and fails to correlate the shown crime density on Homestead to Homestart.  The three references to crime densities "east of the Orange County Social Services Building on Homestead Rd." related to figures 4, 6, and 8 in are actually pictured to be *west* of the OCSSB and are almost certainly crime known to be at Homestart.
2. The paper is completely missing known Rainbow Heights crime.  This is likely one of several identified geocoding errors in the analysis.
3. The paper compares a suburban area to downtown areas, which is an invalid comparison.  In contrast, nccrime.us compares suburban neighborhood areas to each other.
4. The paper fails to document that the main identified sex crime density area is actually the police station itself, which should have been omitted from the results.  Nccrime.us documents quite clearly why the police station crime should be not be included in statistics.
5. The paper fails to consider the involuntary commitments at Freedom House and the risk of a facility that must routinely commit drug addicts to another institution against their will.
   

    

Flawed Premise

The premise of the IFC papers is that neighborhoods can continue to take risky facilities as long as they are not at the top of the crime hotspot list!

Do citizens agree that neighborhood crime rates equal to the downtown crime rate is an acceptable planning metric?  The chart below is from is a recent study down by the town:  36 percent of town residents surveyed feel Unsafe or Very Unsafe downtown compared to 1-8% in neighborhoods.









The nccrime.us Web Site

The nccrime.us website is the only website which contains detailed crime records and a comparison of suburban neighborhood crime for Chapel Hill.  The author of the paper trusted nccrime.us enough to use its incident and address data rather than doing an independent study directly from the police data.  nccrime.us shows 3 different reports relating to geocoding issues (due to poor address data in police data) and also allows visitors to verify neighborhood crime locations on a map.  The IFC papers do not provide any details about geocoding issues aside from stating that 13% of the addresses are known to be invalid (note that 13% invalid addresses could represent 20-30% or more of the incidents if these invalid addresses are incident hotspots).

Neither the town nor the police department provides information about crime in and around neighborhoods in Chapel Hill, nor do they provide the ability to search criminal incidents and arrests, nor do they aggregate crime statistics by address or place to show problem areas. In addition, this site allows users to search all crime, arrests, and police incident histories back to 2003 by street name, incident description, arrest description, day, month, year, etc.

The nccrime.us website does not provide a mechanism for reliable programmatic export of information.  Nor does nccrime.us enable the bulk export of the latitude and longitude of incidents, such as those which are visible in the suburban neighborhood map comparisons.  The geocoding, a process which maps addresses and locations to latitude and longitude, for the IFC papers was NOT provided by nccrime.us.  The geocoding for the papers was done by Christa Polczynski Olson, the author of the papers.  Thus, the author of the papers is responsible for any errors in the location of places and incidents.

There appear to be omissions in the data in at least one of these papers, so we emailed the author, Christa Polczynski Olson, to request the geocoding data and geocoding score since these have not been made available.  In contrast, please note that nccrime.us lists of all known geocoding issues found in the data provided by CHPD.