Define a list of specialized datasets to be available for download based on commonly held questions about crime

[LarryBafundo]

What types of datasets might we want to add to the downloads and documentation page?

We can't easily determine what is most interesting to the community based on user research, so we should start by making a few educated guesses based on the types of inquires CJIS receives. From there, we should consider how we can collect feedback from the community regarding what new datasets they would like to see added.

[LarryBafundo]

Here is a list of the most common data request CJIS receives:

· Gang related offenses - i.e. a recent request for gang related murder
· Offenses cleared by arrest and by exceptional means – type of exceptional clearance.
· Location of an incident – murders at a residence and the weapon used; sex assault at schools; church violence.
· Incidents around the holidays and incidents by time of day.
· Church violence
· Juvenile violence
· Elder abuse
· Child abuse
· Type of offense/type of injury/type of weapon
· Type of drugs seized
· Type of property stolen – looking for boat thefts

[LarryBafundo]

These could be potential starting points for new datasets, but i think this leaves out other perspectives that are likely to be useful for the general public - like handgun violence, specific types of drugs and their patterns, etc.

I think we should work with Cindy to develop a short-list of priority datasets and then think through how they should be defined and eventually added to the CDE as downloads. Since there's no clear answer as to which datasets are most important, this underscores the importance of providing users with a mechanism to submit feedback about the data.

[harrisj]

So, in the slides about NIBRS they mentioned that one of the challenges of NIBRS is that it isn't inherently categorical like SRS and these examples of queries show why. Some of them follow logical top-down queries (like homicides or assault) that might be further refined with filters based on the offenders or victims (ie, child abuse or narcotics seizures). But there are also the queries that basically searching for all incidents that match a filter regardless of what crime category they are in (like those around the holidays, or all crimes involving gangs. There is nothing in NIBRS that forces you to go in a specific hierarchical direction (starting with a crime then going down to states or such) like there is with SRS, which makes it interesting, but I think makes it easy to veer into wacky analysis like "here's everything that happened in America's shopping malls in 2015!" or "Here's where most victim injuries with aggravated assault happened" Of course there are useful types of things we could present here like "sex assault at schools" so it's something to explore

I think we should recommend the following specialized reports potentially:

1. Drug possession charges, with detailed information pulled from the nibrs_drug tables to document types and amounts. Similar for drug trafficking. This could be very interesting in WV, which is dealing with the opioid epidemic.
2. SHR or NIBRS homicide details. I haven't had a chance to see if there is a clean mapping between SHR and NIBRS details for homicides, but it might be a good example of what NIBRS can do vs. SHR if we had some fields provided by both, + extra only in NIBRS
3. SHR seems to have similar weapon categories to the SHR, but provides weapon information for other types of crimes so it can give a more holistic picture of how guns are used for robberies, rape and other categories that get obscured by looking only at homicides
4. I don't really see it in the CJIS list above, but I feel like that one thing NIBRS also gives you is a perspective on fraud crimes. It might be good
5. Child abuse and elder abuse are things this data can reveal (as well as animal cruelty since 2016), and I think we should explore some way to present that as a specialized file (should we just provide incidents or do some analysis)?
6. I've said already I'm not a fan of picking all crimes that match a certain criteria, but I do think it would be worthwhile to do that for everything with a bias_motivation (ie, a potential hate crime) just because that is a special type of data and is relatively small and coherent in an interesting and newsy way that some other broad queries wouldn't be
7. Robbery/burglary would be an interesting place to do some analysis of what gets stolen and where if we wanted to do some sort of summary report.

Ultimately, what I am curious about when thinking about specialized reports:

1. Are there any widely reported facts/myths about crime this data can provide insight into?
2. Are there any notable correlations between two dimensions that could be an interesting story (for instance, does a gun make it more likely that assault become homicide (it automatically becomes aggravated if a weapon involved) or looking at sex assault from the perspective of certain location types (schools, churches) or certain acquaintances.

[LarryBafundo]

let's try to tackle this while we're onsite next week