In order to deploy resources such as police presence most effectively, it behooves a city to know where criminal activity is most prevalent. Because of the impact on the City and its residents, of most concern are areas where shootings and homicides occur most frequently. This reasoning appears to be a driving factor in the Mayor’s Safety Plan released in late 2011: according to the limited city documents released for the plan, 90% of homicides and shootings occurred in 100 blocks. To date the city has published a low resolution map of the 100 blocks but has refused to specify blocks, regions or to provide any useful methodology (exact data used, procedures, assumptions). As a very high profile project, it is unusual for there to be so much secrecy around the exact location of this place-based effort.
As a trusted community intermediary and data resource we have received dozens of inquiries regarding which 100 blocks are the focus of the plan and the crime statistics used to support the selection of these 100 blocks. Since the city chose not to release the details of the blocks or their methods and data, we decided to conduct our own analysis to see if we could develop a similar 100 blocks model with available data.
In this document we share the results of our analyses along with tables and maps we produced and some observations of the implications of our findings. Additionally, we detail our methodology and will make the raw data used for this analysis available for others to replicate and confirm our results. We conclude this report with some lessons about public data that the 100 block example provides for policy makers and community.
Using Oakland Police Department (OPD) data on crime reports from 2007 through 2011, we built two models: the first model covered the entire five-year period and answered the question of which 100 blocks accounted for the most shootings and homicides in the City of Oakland; in the second model we focused only on the 2011 calendar year and identified the 100 blocks with the highest levels of shootings and homicides.
Within the two models we differentiated between two types of shooting offense categories. Our primary analysis includes more serious shootings involving individuals (PC245 A2, PC245 B, PC245 D, PC664/187 A, and PC245 C). We also developed a secondary model and summary statistics using the crime reports that included all types of shootings which includes shootings at vehicles, buildings, firearm exhibition, or firearm discharges (PC245 C, PC246, PC246.3, PC247 B, PC417 C, PC417.8, and PC417.3).
In the five year model (2007-2011), we determined that 17%, or 598 homicides and shootings, occurred within the top 100 blocks in the City. There were a total of 3601 total homicides and shootings in the City over this period.
In 2011, we found that 20% of homicides and shootings (165) were within the top 100 blocks in the City, compared to 845 homicides and shootings in the City overall. (Note that the 100 blocks we identified from just 2011 data varied from those identified using the 5-year data set.)
In our second model that included all shooting types the proportion of all shootings and homicides in our selected 100 blocks decreased by 2% for 2011 and remained the same for 2007-2011. This indicates that more serious shooting crimes are occurring in a slightly more concentrated area in 2011 than the wider set of shooting incidents.
Our expanded model shows that in 2007-2011, only 17% (1041) of total shootings and homicides occurred in the top 100 blocks, compared to 6112 in the City overall.
In 2011, 18% (268) of the total homicides and shootings occurred within the top 100 blocks, compared to 1515 in Oakland overall.
Click through the checkboxes in the layer box to display the 2011 and/or 07 – 2011 one hundred blocks.
How do our 100 Blocks match up?
By georectifying the maps provided by the City of Oakland in their Oakland Safety Plan document, we were able to roughly compare the areas selected by the City with those determined using our methods. A visual comparison of our selected blocks reveals some obvious discrepancies in results produced by the two approaches. Because the City has not been specific in how it has calculated it’s 100 blocks, it is very difficult to recreate an accurate method to test their approach.
Our single year and five-year models have several small areas of overlap with the City’s 100 blocks, yet there are also stark differences between the two regions. There are two large block clusters in the City’s published PDF that contain none of the top 100 blocks according to our methods; both of these are in East Oakland. There are also a number of block clusters in our model that are seemingly not part of the City’s chosen blocks. Most notably the blocks clustered in central East Oakland, adjacent to High Street and along Fruitvale Avenue, are not part of the City’s 100 blocks. This raises more questions with the City’s published maps as it is difficult to know what year or period of time was used to derive the 100 blocks they identify in their plan. Additionally, it is not possible to know exactly which of the “shooting” offenses were included in the city’s calculations resulting in their assertion that 90% fall homicides and shootings occur in 100 blocks.
The City has claimed that 90% of the homicides and shootings occur in just 100 blocks. In order to reach the 90% level of homicides and shootings in the City between years 2007-2011, we had to include 1,303 blocks out of the 6,560 total census blocks in Oakland. In 2011 we had to include 527 blocks out of 6,560 to reach 90% of homicides and shootings in the City. This suggests that homicides and shootings are more dispersed throughout the City and not as concentrated in a 100 block neighborhood as the City suggests.
Implications for the 100 Blocks Policy
According to our analysis, only 17-20% of homicides and shootings occurred in the top 100 blocks and our analyses identify somewhat different blocks than the city’s analyses. Below are some of the implications of this plan given our uncertainty of the model used to define it.
• Providing accurate, useful information to residents is not an optional part of governing. Sharing the details of this plan would empower and better engage residents, not drive a wedge between the public and city hall.
• Any time such a large intervention is planned the public should be fully aware of the reasons and details that led to the development of such a plan- Measure Y is a prime example of visible policy and services.
• Based on the published research from hot spot policing efforts in other urban cities, it should have been clearly communicated from the outset that focusing efforts to reduce crime in the most dangerous parts of the City would result in displacement of crime into nearby neighborhoods. This should have been addressed with either an on-the-ground effort to contain and control crime spread or with an analysis of the crime displacement resulting from this initiative. Such analysis uses a concept called the displacement quotient to measure the spill-over in crime, similar to what should have been done with the gang injunction analysis.
• With any serious initiative to reduce violent crime in a focused part of a city there should be a corresponding evaluation to help the public and other cities to learn if this intervention was effective or not. The Department of Justice requires such evaluations for its funded efforts and makes a point to publish these results so other cities can learn what works and what does not in reducing crime: these studies are published freely at http://www.crimesolutions.gov/ . Because the City has not published the data behind their model nor the exact block boundaries it renders the public powerless to judge the claims of the City regarding success of the initiative.
Some Lessons about Open Data and Transparency
Since we have been unable to obtain the 100 block details or specifics on the datasets used, we have opted to build our own 100 block model and publish the entire project for full public scrutiny- the raw data, the results and the models. This is how we expect public policy to be made in a modern city; with full transparency and accountability in decision making, and so we are publishing this short study as an example of how open, engaged government should look. City budgets, funding and other public records must be publicly available and so it is inconsistent to be secretive when it comes to the development of the data analysis used to base a huge new city project upon. Under California law (Public Records Act, Section 6252(c)), any data used by a government agency must be made public for the public to sufficiently evaluate the claims of government. We strongly believe in and support data-driven decision making in local government, however when the data are not disclosed, decisions cannot be evaluated and there is a breakdown in the democratic process.
If you’re inclined, or if you just love maps and data then we invite you to read our full methodology on how we build this model and derived all our figures.