Many cities have opened up data on various forms of service requests, but it's difficult to interpret these data. If we see that part of a city has more calls for potholes, does that mean that there are more potholes or more people who will make service requests?

Have there been any studies that attempt to decompose number of service requests into the rate of the phenomena and propensity to call?

I am asking for data resulting from such studies, not for discussion about bias or how to avoid bias.

  • 1
    This seems off-topic. Can you expand your question to relate better to open data?
    – Kermit
    May 21, 2013 at 18:52
  • 2
    How do you suggest? This is about using open data, in this case, open data about service requests to cities. In Chicago, these kinds of data make up a good share of cities data portal.
    – fgregg
    May 21, 2013 at 18:54
  • 1
    At the very least, you should normalize the data by population ... which probably means refering back to your earlier question about crime statistics : opendata.stackexchange.com/q/381/263
    – Joe
    May 22, 2013 at 8:59

2 Answers 2


I understand your question and it does relate to open data. It seems like you have a piece of open data: municipal service requests (i.e. "fix the pothole in front of my house!"). Your followup question is a good one: given the number of service requests, are these people just cranky, or are there actually more needed items to be fixed in a certain area?

I searched a while for data on the responsiveness of public works departments in municipalities. Not surprisingly, this data is not collected yet in an open fashion that I could find. It is alluded to in annual reports, but even there, the numbers are sparse.

Some alternate possibilities:

  1. Try to get the total number of requests historically, before and after the open data request system opened up.

  2. Get the number of "potholes fixed" (public works improvements?) with and without an actual service request (this is harder I would imagine, and you will need to clarify what projects are and are not included).

  3. Get the number/type of service requests and overlay demographics (maybe old people call more, or maybe rich people, soccer moms, etc). A lot of this will depend on how the request is geo-located (by zip, by block group, etc)

As an aside, I think it might actually be a good thing to have some sort of "responsiveness" metric for public works departments; maybe you could make one. Every department seems to claim that it is responsive, but I was unable to prove this by looking at their annual reports since they didn't use numbers.


You would need to show evidence that the city of Chicago is reporting (unusually) more potholes than other service requests compared to other cities (ie. New York City). Once you have evidence, then you would need to approach the source of the data. Until that you can show that comparison, it's a subjective opinion that the data has (significant) bias.

Naturally, all data has some sort of bias. There is an acceptable tolerance for bias, but in this case, you would need to show a significant variance between service requests between cities.

  • 1
    I'm not asking about the differences between cities. I'm also asking whether anyone has done this analysis, not how to do it.
    – fgregg
    May 21, 2013 at 19:23
  • Which is why this question seems off-topic to me unless you can better relate it to open data.
    – Kermit
    May 21, 2013 at 19:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.