The first red-light camera in Collier County, Florida, was installed near the county’s busiest intersection in April 2009, a year after the cameras were approved by a divided county commission.

In a little over a year, the cameras were snaring red-light violators at 19 spots near 10 intersections.

The cameras were always touted as safety devices that would change driving habits for the better, reduce accidents and save lives. But, as in hundreds of other communities nationwide, Collier’s cameras were installed during the Great Recession, when government coffers were running low. This led to the inevitable accusations that the real purpose of the cameras was to raise revenue.

The red-light cameras were turned off and the program ended in late February 2013 after an election changed the dynamics of the county commission. During the nearly four years the cameras were active, they generated more than 62,000 citations and nearly $4 million in fines.

But the Naples Daily News wanted to know: Did the cameras actually make the intersections safer?

In late 2014, when I became a projects reporter, it had been more than a decade since I really worked with Excel spreadsheets. So that November, I attended an IRE/NICAR workshop and the IRE/NICAR boot camp two months later in Columbia, Missouri. The training gave me the skills necessary to run basic Excel formulas, sort and filter data, and insert pivot tables, all of which I used for this project.

To start, I wanted to look at a large enough time period to get an idea about what had happened at these 10 intersections before the cameras were installed, while they were activated and after they were removed. I reached out to our county’s transportation department, which provided crash data from each intersection from January 2005 through November 2014 (the most recent at the time). I combined the data from each intersection into one master file of about 6,000 crashes.

I also requested data from three intersections where I was told by a former traffic engineer they had considered installing cameras but never did. I used these intersections as a control group to compare to the intersections with cameras.

In all, there were well over 100 columns in the data, most of which I did not need for my analysis. The columns that were most useful were event crash date; event node description (name of the intersection); event direction from intersection; and event distance from intersection (many of the crashes had occurred near but not directly in the intersection).

I also looked at impact type (angle, front-to-front, front-to-rear, etc.). The crashes most closely associated with red-light running and red-light cameras were angle crashes — cars run a red light and get T-boned —and rear-end crashes, which some say increase when cameras are installed because drivers hit the brakes before entering an intersection to avoid a ticket. Unfortunately, anywhere from 10 percent to 25 percent of crashes each year did not include data on impact type, so I was unable to do a complete analysis. Of those that did include impact type, there had been no increase in rear-end crashes after the cameras were installed, as some opponents had predicted.

Because the red-light camera issue had become controversial and overtly political over the years, rank-and-file staffers at the county’s transportation department wouldn’t comment on the record. However, I did have sit-down meetings with them on two occasions to discuss my methodology and findings. Among their suggestions was to not include any crashes further than 250 feet from an intersection. I spoke on the record with the manager of the red-light-camera project, who no longer works for Collier County.

In addition to the crash data, I also compiled 10 years of traffic counts for each intersection approach using county transportation spreadsheets — a time-consuming process. Using the crash and traffic-count data, I put together about a dozen tables: crashes per month and per year at intersections with and without cameras; impact type by year; annual crash rates per million vehicles per intersection; total crashes by approach, etc.

A couple of years after the cameras were installed, supporters of the program pointed to lower year-over-year crash numbers as proof the cameras were working. But stepping back, we found that the trend toward decreasing crash numbers was already well established before the cameras were ever installed.

From 2005 to 2008, the years before the cameras were first installed, total crashes had dropped 34 percent at the 10 intersections in question. They reached a 10-year low in 2010, before starting to tick up again in 2011 and 2012 — before the cameras were removed. Crashes at the individual intersections followed the same trend: decreasing before the cameras were installed, bottoming out in 2009 and 2010, and then increasing.

We found the same trend at the three intersections we reviewed that never received cameras.

A more likely explanation for the changes in crashes over the years: traffic flow. As fewer vehicles drove on county roads during the Great Recession, there was a corresponding decrease in crashes. And when the economy improved and more vehicles were on local roads, crashes began to increase. There was little fluctuation in crash rates over the 10-year period.

Lastly, to look specifically at crashes that involved red-light-running, we asked the sheriff’s office to provide data on all the red-light-running crashes its deputies had worked over the 10-year period. The data showed that red-light-running crashes worked by the sheriff’s office peaked in 2006, and then declined, reaching a low of 78 in 2009. Looking at the individual intersections, some of the intersections had few red-light-running crashes in the first place, which begged the question: Was there even a problem at those intersections to begin with?



Ryan Mills has been with the Naples Daily News since 2006. Over the last nine years, he has covered cops and crime, elections and state government. He spent about three years as an editor before switching hats again and taking on a role as a projects reporter. He has also worked for The Birmingham News, Panama City News Herald, Miami Herald and Rosemount Town Pages.