I first started questioning their information back in January, when they were issuing glowing press releases about their 2007 numbers (and I was getting hooked on HBO's police drama "The Wire"). I compared their mapping data with news reports about a couple of people who'd been murdered and died on the scene on Beacon Hill in 2007, and the two did not correlate. At the time, I e-mailed SPD community liaison Mark Solomon about it, but he could not help, so I decided to let the numbers bake for a while and revisit them later.
After looking at this stuff again today, I question the numbers in SPD's official reports (more on that later), and I totally mistrust everything their website's mapping feature says. I think what's happening is that the report numbers are getting mangled by the software (or by the data entry people).
If you go to "My Neighborhood Crime Statistics Map" and select "View by Crime Type: Homicide" (but no particular neighborhood or address), it will say "Loading" then bring up a graph at the bottom called "Homicide Aggregated by Month for all Census Tracts." If you hand-add all the numbers from each month in 2007, you'll come up with a total of 20.
But if you click on the gray "1997-2007 by Year" tab right above that graph, you'll see that the 2007 number is higher than 20 -- it looks like it's 24. (And the 24 figure is echoed in the mayor's January 17, 2008, press release "Crime in Seattle hits 40 year low," which says "Last year, Seattle experienced 24 murders, a 25 percent drop from five years ago.")
Why shouldn't the sum of each of the months in 2007 equal the 2007 total? I can think of two likely explanations. One, maybe the police found four homicide victims whose dates of death they could not determine and they therefore do not fall under any specific month. Two, maybe this website feature just happens to be screwed up. (Or both.)
But there's more fishy stuff too.
On September 10, 2007, the Seattle Post-Intelligencer reported that my neighbor was shot and killed in his front yard in the 2100 block of S. Graham Street on September 9, 2007. And I don't think they just made that up -- I personally saw news cameras and people in tears in front of the house the next morning. So let's assume that this incident really did occur. And the story says "Seattle firefighters tried to save the man, but he died at the scene"; let's also assume that this is true.
Graham Street is the boundary of two census tracts: 104 to the north and 110 to the south. Since I know exactly where this man lived, I know that his murder took place on the south side of the street, in census tract 110. Seattle firefighters witnessed his death of gunshot wounds on the scene on September 9, 2007, then that should, without question, place him on the September 2007 homicide list for Census Tract 110. Let's take a look and see what the map feature says.
The month-by-month part of the SPD map feature says there were no homicides in September (or any other month) 2007 in Census Tract 110:
(Before you suggest that this murder may have been mislabeled in Census Tract 104, let me inform you that there are exactly 0 homicides listed for Census Tract 104 for all of 2007, in either the map feature or the official reports.)
And if you go directly to the SPD's "Offenses Reported by Census Tract of Occurrence Report for September 2007," dated 12/3/07, you'll see that the SPD did in fact list a homicide for September 2007 in Census Tract 110 (the only murder in the city that month):
And it's this particular discrepancy that leads me to believe the SPD's "My Neighborhood Crime Statistics Map" feature is totally screwed up and can't be trusted.
Another Beacon Hill murder from last year is even more confusing.
On Sunday, December 16, 2007, the Seattle Post-Intelligencer reported that police were investigating an apparent murder-suicide:
Seattle police are investigating what they describe as a murder-suicide, in which a man apparently stabbed his mother to death Saturday [December 15] before killing himself.
Officers were called about 5:15 p.m. to a house in the 8600 block of Beacon Avenue South, across Interstate 5 from Boeing Field, where they found a woman bleeding and unresponsive, police spokesman Jeff Kappel said. The woman, believed to be in her 70s, was pronounced dead at the scene.
Police searched several surrounding blocks after finding nobody else in the house, he said.
Kappel said officers eventually found the woman's 52-year-old son face down in the street in the 8500 block of 37th Avenue South.
"When they turned him over, they found multiple stab wounds to the chest," Kappel said of the man, who died at the scene. "Officers found a knife a few yards away."
A week later, the P-I did a follow-up story and reported that the King County Medical Examiner's Office "ruled [the mother]'s death a homicide and her son's death a suicide."
OK, so if you look up the 8600 block of Beacon Avenue South (near the intersection with South Cloverdale Street), you'll see that this medical-examiner-ruled homicide, in which the victim was pronounced dead at the scene in the middle of the month, occurred well within the boundaries of Census Tract 117. Because it didn't occur on the edge of a tract, this case seems even more clear-cut than the last one. Let's take a look at the map feature for Census Tract 117 for December 2007.
It says there were no homicides:
OK, not so surprising, given that we already know that map is totally screwy.
Except this time the official report corroborates the information in the map, claiming a total of 0 homicides for Census Tract 117 for December 2007.
If the P-I (and Seattle Times) stories are true, I just don't understand how that figure could possibly be right. (Does it not count as a homicide until someone is charged or convicted? That wouldn't make any sense. If so, this one will never make it onto the books since the apparent perpetrator is also dead.) And if it isn't right, then that also means that the mayor's proclamation that "Last year, Seattle experienced 24 murders, a 25 percent drop from five years ago" also isn't right.
I mean, I understand the site's disclaimer that "Data contained at this location is generally not reviewed for legal sufficiency" and that their "completeness or currency are not guaranteed." But, still, the data shouldn't look like total shit, should they?
I guess it's possible that all this fishy info makes sense if you happen to be someone who understands exactly how the SPD reports its homicides, and why they post different information in their reports than they do on their maps. (Though I suspect that some of this information is just plain wrong.) In any case, the information doesn't make sense to an ordinary citizen who's spent a dozen hours studying it. So I have no idea why the City of Seattle would choose to direct regular people there for information about crime in their neighborhoods. Good luck actually getting it.