Flaws in the 2010 Winnipeg Police Annual Report – Part II

Dealing with the 9549 crimes listed as “Not District Specific”. 

In the previous post I discussed some of the major issues with the 2010 Winnipeg Police Annual Report.  They included the addition of two new statistical categories: one in the “Criminal Code by District  Table (Not District Specific)”, and in the “Criminal Code Offences by Month Table (Undertermined)” – small matter that the latter word is misspelled in the report.  As it turns out, the most glaring aspect of the report is in the math: crunch the city-wide crime numbers  as they are shown and compare 2010 to 2009 and you will come up with a 9 percent increase in crime as opposed to the 7 percent decrease that the Annual Report states.  What?!

What originally caused me to take a closer look at the numbers was the apparent disparity between the reported reduction at the district level (see column 1) .  The District numbers show reductions in the range of 20-30 percent and yet the city-wide reduction was listed as -7%.  Something’s wrong here.

The next step was to do the actual calculations using the District and city-wide data as listed in the 2009 and 2010 Annual Reports.  The results of those calculations are listed in column 2.  There is a significant disparity between the two columns – at this point I’m ready to suggest that the Police Service  pull the 2010 Annual Report off the website, scrap the whole mess and start over.

Another reason to start over:  Let’s look at the reported crimes listed under the new category of “Not District Specific”.  Without the help of an explanation, one is left to assume these are reported crimes in addition to those listed under the five Districts (when you add the numbers listed for the 5 districts and then add 9,549 from the “Not District Specific” category you arrive at 61,680 – the numbers listed in the 2010 Annual Report as the total number of crimes reported city-wide).

Although these  9549 crimes would be disproportionately distributed  (ie District 1 would have more than District 2),  for the purpose of the table below column 3 represents the percentage crime rate change if the “Not District Specific” crimes were equally distributed throughout the five districts).

Area

Column 1

As Reported*

Column 2

Actual**

Column 3

Projected***

City wide

-7

-8

+9

District 1

-20

-10

+4

District 2

-32

-14

+16

District 3

-23

-5

+10

East   District

-25

-7

+7

District 6

-26

-8

+13

*    These are the percentages as listed in the 2010 Winnipeg police Annual Report.

**  These are the percentages arrived at when you crunch the raw data in the 2009 and 2010 Winnipeg Police Annual Reports excluding the 9549 offences listed under “Not District Specific”.

*** These are the percentages when you factor in the 9,549 incidents listed in the “Not District Specific ” and evenly distribute them within the 5 Districts.

At this point I am not certain if the statistical wizards within the Winnipeg Police Service did a double count of some incidents or simply out smarted themselves with the introduction of the new categories.  I suppose the only way to find out is to ask.

I sent the following email to the Winnipeg Police  Service:

I have now had the opportunity to have a quick look at portions of the 2010 Annual Report and I have a few questions:

1.  The Criminal Code Offences by District table has a new column titled “Not District Specific” added.  Why was this column created? How are offences determined to be “not district specific”?  How are not district specific offences handled in terms of plotting them on Crimestat?

2.  The temporal table also had a column added under the heading “undertermined”.  I assume this was meant to say “undetermined” . Why was this column added?  How, for example, can there be 7,707 thefts in this category?  Does the WPS not require the complainant to provide dates in terms of when the offence took place?

3.  The 2010 Annual Report shows 61,680 as being the number of total criminal code offences reported to police and shows this to be a 7% reduction over 2009.   The 2009 Annual Report showed 56,427 .  Based on those two figures how was the 7% reduction calculated?

Thank you for your help.

I’m hoping to get a response in a more timely fashion than the release of the Annual Report – almost a year after the fact.  When (or if) I receive a response I will post it because this is important enough that it demands an answer.

Advertisements