jeancochrane
commented
7 years ago Hey @jeremiak!
Thanks for taking the time to leave some thoughts here. "Low crime" was by far the most challenging priority to score, so it's great to hear your ideas about how we could have done it better. I think you've hit on a really interesting question, so I'm going to go ahead and dump a bunch of thoughts about it. Feel free to engage with any part that interests you – and keep in mind that these are my own personal opinions, and not the opinions of Crain's or anyone else who was involved in this project.
The scores were indeed generated based on CPD's reported crimes dataset from 2015, which to my knowledge does include white collar offenses. That being said, if I'm interpreting your question correctly, it's still really important: police reports involve a whole host of biases, including that the data will overemphasize crime in aggressively policed neighborhoods and underemphasize crimes that tend to go undereported, like "white collar" crime – but also sexual assault, violence against undocumented folks, and crimes committed by police officers themselves.
Given all of this, I think you're right that "police activity" is probably a more accurate name for what we're measuring than "low crime." But how often do people consider police activity, as opposed to safety or crime, when making a decision to buy a home? I wonder whether it's ever a primary factor, or if it's secondary to the idea of "safety" that we're trying to get at. (And that's genuine wondering, not rhetorical! I'd love to see some research about this.) Plus, there's the sticky consideration of whether "police activity" is a good or bad thing. I could imagine some of our users thinking it's good, and getting confused (from a UX perspective) when it largely correlates with what they perceive as "high-crime" neighborhoods. I'd also be worried about misleading users who really do care about finding low police activity neighborhoods, but for reasons not captured by the data – I'm thinking in particular of black homeowners, who may reasonably desire to find a neighborhood where they can feel that they won't be harassed by police.
Since what we're trying to represent is something like "neighborhood safety," what do you think would be a more accurate methodology for producing a score than using CPD reports? We made a decision to use reported crime rates given the time and resources available to us, but I don't think it was a perfect solution by any means. An interesting alternative model might be something like the Project on Human Development in Chicago Neighborhoods, which collected survey data from residents on how safe they felt their neighborhood to be. Maybe there are other priorities that would help provide better context for crime rates, too – I could imagine that reporting the density of neighbor social networks (something like a "neighborliness" score) would help put perceptions of safety into perspective, for example.
Feel free to leave as many or as few thoughts as you want. We probably won't substantially change the methodology for this project, but I'd still love to hear your input! And of course, if you'd like to push back on my thoughts RE: "police activity," you're more than welcome.
Hey all, this project is pretty darn cool and really easy to use! I think its a cool idea to allow somebody to rank the aspects of a neighborhood that are important to them and then find a home so they can join that community.
However, I'm not sure the "Low crime" button makes sense in this context. I propose using something like "Low police activity" since it seems like you're using data that is generated by the Chicago Police Department and not the judicial system.
A follow up question: does the definition of crime here include "white collar" offenses like embezzlement and fraud or is it strictly "street" offenses?
Anyway, cool project and happy to see that the code is open!