The first step is arming ourselves with as many useful databases as possible and keeping them up-to-date. We've bought all the relevant data NICAR has to offer, such as FBI Uniform Crime Report and fatal accident reports. We've also negotiated with a number of different state agencies for court records, auto accidents data, state police trooper activity reports, concealed weapons permits and autopsy statistics on causes of accidental deaths, suicides and homicides.
But, perhaps, the most crucial data, such as incident reports, we have came from local police departments.
This is the heart and soul of what we will be reporting on the weekly Public Safety page.
Using the police incidents data, we can look at almost any kind of crime big or small in our area. We get quarterly updates allowing us to track trends in our area long before the FBI releases its incident reports.
The data includes information about when and where crimes happened, including block numbers and street addresses, Census tracts, and whether the incident happened in a home or in an alley. Some of the databases also include information about stolen property, suspects, victims and any weapons used.
From data to news
The next step is to mine our growing body of data for stories. We look for one-hit stories in the data - a profile of the state's most prolific writer of traffic tickets, for instance.
But more importantly, we use the data to track crime rates, clearance rates and the changing faces of crime.
We look at which of the eight major crimes (murder, rape, robbery, aggravated assault, burglary, larceny, auto theft and arson) are increasing or decreasing. And we look for regional trouble spots.
We use the statistics to add context to daily stories by adding a paragraph to reports telling readers what the neighborhood's crime rate is and how it compares to the city rate. We can also include numbers on how prevalent the crime - whether it is murder, burglary or vandalism - is in this area. Yearly status reports on crime can also be turned quickly. Using both the FBI and our local incident data, we can track the frequency of major crimes in our area back to 1980. Often, these analyses lead us to bigger stories a recent series on violent crime in Portsmouth, for example.
Reporting crime in Portsmouth
Portsmouth's rate of violent crime had climbed every year since 1993. By 1995, Portsmouth reached it highest violent crime rate ever, with 1,480 violent crimes per 100,000 residents. It had also achieved the dubious distinction of being the 14th most murderous city in the nation.
Last year, the violent crime rate dropped 29 percent. Having tracked the crime rate for years, we were able to break the news. Our databases allowed us to give readers a comprehensive package about the drop and reality of violent crime in Portsmouth.
Here's how we did the analysis. The Portsmouth incident reports database includes the following fields: case number, offense, report date, incident date and time, block number, street address and Census track where the incident occurred, weapons used, property stolen and the race and sex of victims.
First, I added a field in order to categorize the crimes to fit the Uniform Crime Reporting definitions of murder/manslaughter, rape, robbery and aggravated assault. Then, grouping by that field, I did a citywide count of those four crimes.
Using the city's population, I used those counts to calculate a violent crime rate for the city.
Looking at the four crimes individually allowed us to see that huge decreases in murders and robberies were responsible for the plummeting crime rate.
It also showed readers the frequency of each violent crime.
Next, we wanted to see if the drop in violent crime was spread throughout the city or in specific areas of the city. So, we grouped by Census tracts and used a where clause to count only the four violent crimes. Using population statistics from the 1990 Census and those counts, we calculated a violent crime rate for each Census tract.
We did the same for 1995, and then calculated the percent change in violent crime for each Census tract.
What we found was that the most crime-ridden areas showed the biggest decreases in crime. The few areas where crime had increased in 1996 were still relatively low-crime areas.
By mapping the data, we were able to show readers how the crime rate in their neighborhoods stacked up to other areas of the city.
With the bulk of the analysis done, we ranked the four violent crimes and found robberies to be the most common.
Grouping by block number and street and using a where clause to narrow the field to only violent crimes, we located the 10 most dangerous city blocks.
Grouping by the weapon used, we found that feet and hands were the most common weapons used in the commission of a violent crime.
Last year's drop in violent crime in Portsmouth does not make a trend, and we clearly stated that fact in the third graph of the story.
But it may be an early signal that the city has reason to hope - a signal computer-assisted reporting helped us recognize.