Unifying Alabama's Traffic Safety Efforts
     Working Together to Save Lives    
The Critical Analysis Reporting Environment (CARE) is a data analysis software package originally designed for problem identification and countermeasure development in traffic safety applications. Developed by the staff of the Center for Advanced Public Safety, CARE uses advanced analytical and statistical techniques to generate valuable information directly from the data. Although its primary use is still in traffic safety areas, CARE can be used to process any database, and its most recent applications have included databases in the areas of emergency medical services, medical, nursing data, a variety of questionnaires, and several criminal justice applications.

In 2003 the electronic citation (eCite) system was introduced in the state of Alabama, beginning with the Heflin weigh station. This system allows officers to utilize a license scanner, GPS device and laptop to write traffic citations quickly and easily from their vehicles. By 2007 this program was deployed to every state trooper, and it is now being rolled out to other law enforcement agencies throughout the state.

To view the CARE Dashboard, click here. For more information about eCite, visit caps.ua.edu/eCrash.aspx.
Most of your statistical information needs can be obtained right on line from the new CARE Dashboard, which was developed by the University of Alabama Center for Advanced Public Safety (CAPS).  Before getting on this site, please take the tutorials given on the http://www.caps.ua.edu/software/care/ CAPS page, which will explain the data and use of CARE and the Dashboard.  A PC (desktop) version of CARE is also available;  to download the desktop version and data, access the Download page of the CAPS website. If you would like to see details on the CARE software, it is available on the CARE page of the CAPS website. If you see any problem or need help, e-mail care@cs.ua.edu or call 205-348-7920.
eCrash System Deployed
The Alabama Department of Public Safety teamed with the University of Alabama Center for Advanced Public Safety (CAPS) to develop eCrash, the nation’s first totally paperless crash reporting system. Except for the reports provided to those involved in crashes, all other aspects of eCrash are paperless, from the officers’ entry of the data (in many cases in their vehicles), through the approval process and uploading to the crash records database in Montgomery. The use of eCrash for data entry in the officers' vehicles will keep them in the field where they can respond to emergencies; and moving the data entry to the field will make the data more accurate, timely, complete and consistent. About 95% of crash reporting agencies either use eCrash or submit their crash records electronically in eCrash format, which is now MMUCC compliant. The target date to get all agencies using the eCrash report is December 31, 2010.
Description of CARE Impact Output
The following IMPACT that compares Alcohol Crashes (red bars) vs. Non-Alcohol (blue bars) for the Day of the Week variable will be used to explain IMPACT outputs.

The above example IMPACT output for the alcohol related day of the week variable will be used for this explanation. The left (red) bars represent the current subset (Alcohol Related/DUI Crashes for this example), while the right (blue) bars represent the proportion of crashes that are in the subset being compared (in this case the complement, which is all crashes for which non-DUI causation was reported).

Note the largest Max Gain first output lists the “worst first” – all other things being equal. The first two numeric columns of the chart give the frequency and percentage for the subset (as defined by the current filter). In this case the subset is alcohol related crashes. So, Alabama had 1,718 alcohol crashes on Saturday for this particular time period, which were 25.478% of its total alcohol crashes. This is compared against the “Other” subset, which in this case is the non-alcohol crashes. This indicates that 14,650 non-alcohol crashes occurred on Saturday, which is 12.497% of all non-alcohol crashes.

To determine if there is an over-representation of alcohol crashes on Saturday, the proportions must be compared. In this example, 25.478/12.497 = 2.039, which is to say that there are more than twice the number of crashes than would typically be expected if Saturday alcohol crashes were the same as Saturday non-alcohol crashes. Note that the double bar chart for Saturday shows the same thing visually. A statistical test is performed, and the asterisk (*) on the 2.039* indicates that this is significant at the 99% level according to a statistical t-test that assumes a normal approximation of the binomial distribution. Statistical tests are not performed (no asterisk will ever appear) if either of the sample sizes used to compute the proportions (percentages) are less than 20.

Finally, the Max Gain (Maximum Gain) is the number of crashes that would be saved if we could somehow eliminate the over-representation. The total number of these crashes is 875 – notice that this is over half of the 1,718 alcohol crashes that occurred on Saturday. All other things being equal, this over-representation gives a measure of the maximum gain that could be expected if we could eliminate the particular (in this case “Saturday”) problem.