Example project - Introduction - A Basic Guide to Reproducible Research

This section will walk you through the example research project used to illustrate some of the practices throughout the guide. Here, we’ll move away from the concrete implementation and focus more on the research question, background, analysis, and results. If you want to look at the code, rerun the estimation, or even start out from the raw data files, visit the GitHub repository for instructions. To follow along with the analysis in a more interactive way, click on , which will allow you to reproduce the final steps in order to obtain the figures used on this website - directly in your browser without any setup needed. If you’re interested in how this works, head over to the corresponding section of the guide. To see a static version of the comparison, click on .

Now let’s dive right into the project and start off with an introduction on crime in Chicago and the crime prevention program I’m analyzing.

Crime in Chicago

Chicago notoriously suffers from a very high number of crimes, where many incidents happen in poor neighborhoods on the South and West Sides.¹ Figure 1 shows that even though there was a welcome downwards trend for a few years, for both property and violent crimes, this started to inverse around 2014/2015.

Figure 1 - Evolution of violent- and property-crime counts over time

The definition of violent and property crimes follows the National Incident-Based Reporting System (NIBRS) of the Federal Bureau of Investigation (FBI).
This figure is an approximate replication of Figure A.2 in McMillen et al. (2017)
Data source: Crimes - 2001 to present, Chicago Data Portal; accessed January 25, 2018

If we look at a more granular view of violent crimes, we see that the number of homicides in 2017 is on an even higher level than it was in 2006.

Figure 2 - Evolution of violent-crime counts over time by subcategories

The definition of violent crimes as well as the sub categories follows the National Incident-Based Reporting System (NIBRS) of the Federal Bureau of Investigation (FBI).
Data source: Crimes - 2001 to present, Chicago Data Portal; accessed January 25, 2018

To put the graveness of the situation into more graspable numbers: In 2017, on average, every 13 hours a person was murdered and every 2 hours and 30 minutes someone was shot.²

Let us change the perspective and look at the intra-year and intra-day variation of violent crimes. The top panel of Figure 3 shows violent crime counts partitioned both by years and months. Again, we can see the decline in crime counts until 2014. Furthermore, this figure shows that there is a clear seasonal pattern, where crime counts are relatively higher in the summer compared to the winter months. The bottom panel of Figure 3 depicts the intra-day variation by month. There is a strong increase in crimes starting in the late afternoon, when children are starting to leave their schools. Interestingly, this is also when property crime counts are notoriously high (not depicted in Figure 3).

Note: Figure 3 is interactive. If you want to see the intra-day variation in the lower part of the figure calculated with data only for one year, you can select a single year by clicking on a square corresponding to that year in the upper figure. The lower heatmap will adapt accordingly.

Figure 3 - Intra-year and intra-day variation in violent crimes

The definition of violent crimes follows the National Incident-Based Reporting System (NIBRS) of the Federal Bureau of Investigation (FBI).
Data source: Crimes - 2001 to present, Chicago Data Portal; accessed January 25, 2018

Chicago’s Safe Passage program

Chicago’s crime incidents regularly involve students. In 2012, for example, 29 current or former students of Harper High School were shot, and eight of them did not survive.³⁴

“[I] encountered kids at Harper [High School] who wake up and they wonder whether they’re going to make it out of school alive.” - Michelle Obama, 2013⁴

Due to Chicago’s crime problem and the threats many students face daily when going to and coming from school, the City of Chicago launched the Safe Passage program in 2009 for initially 35 schools. The program has the goal of keeping the students safe on their way to school, thereby enabling them to focus more on their studies.⁵

To ensure the safety of the children, guards, which are mostly residents from the respective neighborhoods, are posted along various routes to the participating schools, right before and after schooling hours.⁶ They are hired by local community organizations and trained in de-escalation strategies and safety protocols. However, they do not carry any weapons or have special authority.⁷

Since 2009, the Safe Passage program has been heavily expanded. The biggest extensions happened in the school years of 2013-2014 and 2014-2015, where in total more than 90 schools were added to the program. This happened in response to a large school closure of around 50 schools, which let to many students having to travel farther to school, and some of them were required to cross gang boundaries.

In the school year 2016-2017, more than 1,300 Safe Passage workers guarded around 75,000 students.⁸ The costs associated with the program for that school year amounted to $17.8 million.⁹ New Safe Passage routes are added almost every year and focus primarily on areas with especially high crime counts.

Motivation and research question

Because of the graveness of Chicago’s crime problem as well as the fact that a considerable amount of tax payer money is spent on the program, gaining more information on the effectiveness of the program is incredibly important. This analysis tries to answer the following question: Does Chicago’s Safe Passage program reduce crime along the guarded routes during operational hours?

What has been done before?

Officials from the City of Chicago have mentioned declines in crime from 20% up to 32% due to the program. They did not, however, publish any details on such an analysis.¹⁰^,¹¹^,¹² The question remains in how far this decline in crime, which we already partly saw in the figures above (at least up to 2014), is due to the Safe Passage program, and how much of it is caused by other factors.

A more sophisticated approach was taken by Daniel McMillen, Ignacio Sarmiento-Barbieri, and Ruchi Singh in their working paper “Do More Eyes on the Street Reduce Crime? Evidence from Chicago’s Safe Passage Program” from June 22, 2017.⁷ They find a reduction in violent crime counts along Safe Passage routes of around 14%. For property crimes, however, no significant effect could be detected. As the authors have so far made neither their code nor all of their data public, I took the opportunity to try to replicate one of the analyses done in their paper. You will be able to see a direct comparison of figures, tables, and the main results by clicking on the “launch binder” badge at the beginning of the page. Note, however, that the paper by McMillen et al.⁷ contains much more then what I replicate. You can find a more detailed overview of their paper, as well as other existing literature on the Safe Passage program, in the “Literature” section in the Appendix.

Through making my entire analysis publicly available, I aim to enable others to easily build upon and extend it, which will hopefully lead to a better understanding of how to implement similar programs and effectively fight crime. The results of my replication, as well as the methods used in the estimation, are presented in the next section.

References

Twitter Facebook Google+ LinkedIn

Example project - Introduction

Stefan Binder

Crime in Chicago

Chicago’s Safe Passage program

Motivation and research question

What has been done before?

References