Without back data there is no way to go forward.
You would have to start out by collecting data to know where to begin. Without back data there is no way to go forward. You would have to keep in mind that during the night, police have difficulty detecting the race of a suspect before they make a stop. The suggestion of the darkness that is the race distribution of drivers stopped during the day should differ from the racial distribution of drivers stopped at night if officers engage in racial profiling. Thus if travel patterns, driving behavior, and exposure to police are similar between night and day, then we can test for racial profiling by comparing the race distribution of drivers stopped during the day as to those race distribution of drivers stopped during the night. Limiting much of our analysis to stops occurring during the twilight period roughly between roughly 5 and 9 PM, we will not be able to test for differences in the racial distribution of traffic stops between night and day. I do feel that we would need to have someone from outside the department to be over the research making it fair and unbiased. Most law enforcement officers would maintain they are not biased, prejudiced, or using racial profiling in their policing methods.
We would also want to divide the high-discretion stops from the low-discretion stops because with the high-discretion stops officer has ample discretion to decide whether or not to make a stop. In a traffic perspective, this might involve a driver failure to signal a turn, or a vehicle with an almost, flat tires, license plate unlighted, or something hanging from the mirror, an example in the pedestrian situation would be when an