The purpose of this website is only to host research that I have completed on the performance of photo ticketing systems in Arizona. The purported reason for employing photo ticketing is to improve safety. This can be measured by number of crashes and number of injury and fatality crashes. A system that does not reduce the total number of crashes or the number of injury and fatality crashes does not improve safety.

To judge the performance of photo ticketing systems, many municipalities have police perform a simple before and after tally of crashes at monitored intersections. This approach is too simplistic, as it avoids and ignores proper statistical analysis, bias, additional factors that affect the number of crashes at a given location. Such considerations might be:

  1. Selection bias: Locations chosen for photo enforcement presumably have an above average number of crashes. In many cases, the number of crashes is only temporarily above the average, and will normally revert (lower) back to the average without taking any action. Thus a reversion-to-the-mean should be considered so as not to attribute a return to normal crash levels to photo ticketing.

  2. Additional factors: It is proven and logical the crashes correlate to traffic volume. More traffic = more crashes. Traffic volume MUST be considered so as not to attribute an change in crashes due to a change in traffic volume to photo ticketing. From 2008 to 2013, Arizona saw a reduction in vehicle mileage, and this time period is coincident with the installation of many photo ticketing systems. Many municipalities have attributed reductions in crashes over this time period to photo ticketing rather than the decline in traffic volume. Additional factors may be improvements to intersections, adjustments to light timing, or other local factors.

  3. Local, regional, national trends: There are small and large trends that will have a more global impact on the number of crashes, injuries, and fatalities. Such examples might be safer cars, an increase in seat belt usage, a reduction in drunk driving, or changes in drunk driving laws that result in less impaired driving on a local or state level.

  4. Comparison to control intersections: To best analyze performance, a monitored intersection should be compared to a similar unmonitored intersection.

Due to the limitations of being just an average citizen, I am only able to implement some of these items. Regardless, it is more definitive and reliable than the simple tallies published by police agencies. Using ADOT's statewide crash databse and mileage numbers, I can compare the data at a given location with that of the entire city and the entire state which provides a pretty good analysis when done over a 10 year time period. A reduction in crashes or injuries would be indicated by a noticeable divergence from city or statewide trends, and that clearly isn't present in the data.


Latest Updates Title
Oct 19, 2015 Effectiveness of Red Light Cameras in Tucson, AZ