Monday, September 6, 2010

Why Discrepancy between Neighbourhood Watch and E-Watch Info from Police?

Recently there were three garage break-ins (Inkster, Polson and Machray).
Neighbourhood Watch/CAPTAIN  messages for which were sent out by WPS Community Services.   However, they didn't appear on the E-Watch for the corresponding period.    Here's the explanation, provided by WPS  for discrepancies between the two data sets:

Aug 30/10 14:21 From: Project Leader Winnipeg Police Service
What I’m saying is that you have 2 different sets of information, but correct in the context they are provided.  Use the union of both sets to get a fuller picture of what is happening in the area of your interest.  If you want to know why one is on one list and not the other, I can provide you the info when I get a chance, but the vast majority will be one of the 3 reasons I sent out the last time.

1)    Type has changed, will affect both lists.
2)      B&E Other, will not be on E-Watch but will be on CAPTAIN
3)      B&E Res not processed by crime analysis due to lack of investigative report or no relevant information, will not be on E-Watch but will be on CAPTAIN

Aug 30/10 14:20 From: Project Leader Winnipeg Police Service

This list is evolutionary, based on the latest information we have about a particular occurrence.  The CAPTAIN list is a snapshot of what was reported since the last time CAPTAIN was run.  As well, the two list differ in what they report as no B&E Other is published on E-Watch.  As from the last list 9 out of the 13 differences you reported were due design differences between the 2 lists. I’m sure the differences you see this week are due to these design differences.



Aug 30/10: 14: 18 From: Project Leader Winnipeg Police Service
I did some viewing of the B&E data that we have and noticed a few things.  There are quite a few incidents that start of as a B&E and end up as something else, such as mischief, intoxicated person, dispute, robbery, theft, and there is a  number of incidents that start of as something else, such as theft, damage, graffiti and even assault that are later reclassified as B&E.  That is the biggest reason why incidents will appear on one list and not the other.  Another reason is that the captain query asks for residential and ‘other’ B&Es.  Type ‘other’ includes detached garages or lockers (B&Es to attached garages are classified as residential). I don’t believe that the e-watch list includes the B&E ‘other’ type. 

The captain query uses a time range based on the creation time of the incident to capture the data .  This is the only way to ensure we get all the data and don’t send duplicates, as we don’t keep track in the database of which incidents are sent through captain.  Each morning the query runs and asks ‘What incidents have been created since the last time this query was run that matches the captain criteria’.  This means that no matter what happens to the incident after the query runs it will never be re-selected.  The e-watch asks ‘What incidents that have been verified and occurred between these dates match the e-watch criteria’. 

There are at least 3 database systems involved that produce the 2 lists.  Entry is mostly done for the purpose of dispatching vehicles to the scene, the officers then take notes and update the reports.  Basic data is transferred from the dispatch system to the police records and reports system. The officers enter their reports and may re-classify the report if they feel that upon speaking to the complainant, the initial classification was incorrect. The captain program captures the data at this point.  Depending on the timing, the officer may or may not have written the report and updated an incident that is being selected for captain. A subset of the data is then transferred to the crime analysis system. The data selected is based on modification date, so that nothing matching the criteria is ever missed.  Here the data is put into a much more standardized and detailed format so that it may be analyzed for patterns.  The e-watch list is produced from this data.


No comments:

Post a Comment