Since @PizaEric asked during my #crimcon panel and I gave him a short answer due to time constraints: A thread on how to choose a data dashboard/ data visualization tool.

Spoiler alert: It's NOT easy.

First, there is no "best" tool. It's like the "best" car. Best for whom? For what sort of driving?

Instead, of "best", it's best for what? Best for whom? Just like with cars, this is a combination of needs and preferences.

Second, the tool is just a tool. It matters less than basic design of your viz.

Don't think that a fancy new tool is going to make you an awesome data visualization designer. It won't. It'll just give you more ways to make crappy graphs.

If you're traditionally trained in crim/CJ, you probably have no training in how to design charts. I recommend @visualisingdata's book as a reasonable starting point, and Stephanie Evergreen's book is great too.

tl;dr: Make great charts first, then worry about dynamic charts.

But let's say you're past that and you want to do dashboards. Now you gotta do a needs assessment. This should be based on a holistic review both functional requirements and other needs.

"Functional requirements" is a fancypants way to say "what do you need this to *do*?"

Other needs is everything else. It include things like cost and integration with existing toolchains & workflows.

It doesn't matter how great a tool is if the licenses are too expensive for your organization, or if its products can't be injected into your website for reasons.

A word about cost: It ain't just money. Consider training costs (YOUR TIME). If you're already familiar with R, Shiny is going to be a lot easier to use than for someone (like me) who can barely get a hello world script to run in R.

For the Alaska Justice Information Center, we're heavily invested in @Stata, and moving away from that would incur change costs. While we love LOVE love Stata for data management and modeling, it's not a good viz tool.

So we've segmented our work into data stuff and viz stuff.

Our data stuff happens in Stata, and we do dynamic vizualization stuff in @tableau. We use @tableaupublic to publish our vizzes, in part because it is (relatively) easy to get vizzes on the web from Tableau.

Is that the only way to do this? NOPE. There's Microsoft's PowerBI, for which many orgs already have a license. And there's R, Python, and whatever language all the cool kids are using these days.

Always think of the audience. It doesn't matter how fancy your charts are if they don't tell the audience what they want/need to know.

And keep in mind that you don't have to dashboard ALL of the things.

As @LessCrime said in our session, you can scaffold your work for different audiences. To see an example, go to http://uaa.alaska.edu/ajic/homicide  , where you'll find a dashboard, one-page fact sheets, and a full narrative report on homicides in Alaska.