January 12, 1997
DON'T BE MISLED BY PHOTO RADAR STATISTICS...
A QUICK REFERENCE GUIDE TO KEEP IN YOUR FILES

VANCOUVER -- From the government's point of view, photo radar can only be 
successfully implemented if it gains broad community acceptance ("successful 
implementation" equals suppressed public objection to paying tickets).  SENSE 
believes Victoria is attempting to sway public support through the use of 
questionable statistics that suggest photo radar works, but may not represent 
real safety improvements caused by photo radar.

As members of the media, you probably receive photo radar statistics from the 
police, ICBC, government, or community speed-watch groups, but may not have 
adequate time to give the information the scrutiny it requires.

Here's a primer on how to quickly assess the worthiness of the stats you're 
being asked to accept as fact: 

PRE-EXISTING TRENDS...
* Simple year-to-year comparisons are meaningless: "there were 50 crashes 
last year and only 35 this year," could be referring to anomalous years 
(e.g. 25, 28, 26, 50, 35).  Get a time-series of at least five years to 
ensure that the years were representative.  If possible, present the 
information in a graph.

* Time series also show historical trends... e.g. 80, 74, 68, 50, 35 -- 
perhaps the rate was naturally falling?

* Similarly, percentages ("crashes dropped 30% over last year") can be 
misleading for the above reasons.

CONSISTENCY...
* Are the changes representative of all photo radar locations?  Have you 
been provided with the statistics for all the photo radar sites, not just 
ones which best support their message?

* Have there been similar crash reductions at locations not using photo 
radar?  -- are there control sites, and are the control sites valid (i.e. 
do the control sites likewise meet the conditions on this page)?

ENSURE THAT YOU ARE COMPARING APPLES TO APPLES...
* Is the change explained by alterations to the engineering environment? 
(e.g. improved intersections, left-turn bays, better signals or signage, 
more lanes, road widening, etc.)

* Have traffic volumes remained constant?  (e.g. has an alternative road 
diverted traffic?)

* Are these just speed-related crashes, or do they mingle other crash causes 
which may explain the variance?

* Are police responding to and reporting crashes using the same criteria?  
(e.g. have they reduced responses to only serious crashes?, have they 
changed the reporting of speed-related crashes?)

* Have there been other factors that have affected crash rates?  (e.g. 
Counterattack, weather, economy, etc.)

SIGNIFICANCE...
* Finally, is the change statistically significant?  Are the changes outside 
expected random variation?

                                   - 30 -