, ,

When Did Statistics Go Mainstream? – 9/8 GL Training

Join GovLoop and our partner IBM to learn how to use analytics to boost efficiency and cut costs in our next GovLoop training on 9/8 at 2pm – RSVP

****************************************************************************************************************************************

I remember the first time I got excited about statistics. I was in an introductory to statistics class and was taught how to use SPSS to find basic information with Census data.

I became hooked and from that point on I’ve always been a stats nerd.

In my master’s degree in Sociology, I analyzed social security data on disabled children to find interesting trends on how disabilities affected parental employment and how that affect varied across income levels and demographic behaviors. As GovLoop began to grow, it was awesome to peer into the web statistics as I could track a host of statistics over time and watch changes day-to-day.

I don’t know what happened but it must have been around 3 years ago when stats began to go mainstream.
-The book Freakonomics told compelling stories of using data to dispel myths and find correlations and causation.
-The use of stats began to show up in TV shows like “The Wire” where cops would debate statistics in the renown CitiStat program.
-The President fell in love w/ behavioral economics and hired a bunch of top leaders in the field and use the statistics and lessons as part of creating effective programs
-The NY Times began to write about the explosion of jobs for data scientists and even launched their own data visualization studio

It’s cool to see the trend start reaching government as there is so much potential to use data to make government more efficient.

-Analytics can tell trends in large benefits data like Medicare or unemployment benefits or Recovery funds to find trends of unusual and potentially fraudelent behavior.

-It would be great to analyze demographic data of citizens mapped to current utilization patterns of city services to model new ways to deliver services and in different formats.

Basically it’s a lot easier to accumulate data and make sense of it. So instead of making decisions based on hypothesis and gut, we can use analytical decision making process.

We are going to begin this discussion on statistics and analytics on 9/8 with a fun free 1-hour training with our partner IBM on how to use analytics to boost efficiency and cut costs.

Have any cool stories of how you use data analytics in your agency? Share below. Got questions you want us to address – let me know.

Leave a Comment

3 Comments

Leave a Reply

Profile Photo Bob Ragsdale

Great topic. There is so much untapped information out there. The real-world challenge is twofold. First, capturing the data and second, having a system that can organize and draw insights from the information.

An agency has to have the infrastructure in place to centralize the information it captures. If is all hanging out in spreadsheets and in folders on the hard drives of individual computers, analyzing it is going to be tough. In today’s environment there is really no reason not to be capturing and organizing this information in the normal execution of operations. Once in place, case management, workflow management and business process management systems do this as a matter of course.

A great example of this in practice will be at our upcoming 462 Conference for EEO professionals on September 7th. Tasha Richburg from the Centers for Medicare and Medicaid Services will be giving a presentation on how the CMS utilized the advanced reporting features in their EEO case processing software, icomplaints, to better fix on the cost of case processing, and to ultimately decrease their annual expenditure. If anyone is interested in attending this seminar they can register here: https://462registration.micropact.com/

Reply
Profile Photo Mark Hammer

Sadly, statistics is too often taught like some sort of catechism or articles of faith, rather than what it really is: a human invention to help represent and think through things. We can mistakenly view it as guiding the ship rather than being in service of where we want the ship to go.

As much as there might be an “explosion of jobs for data scientists”, the notion of a generic analyst makes about as much sense as the idea of a generic manager. There ARE core skills, but uiltimately, you have to know something about the content domain to know how to make the data dance gracefully. The numbers reflect ideas, and if yu have only numbers but no ideas, you don’t have much.

I think, as well, there is often a mistaken assumption that if you have someone who “knows stats”, and have a data set, insight is but moments away. It’s not. I tell people that data is like soup. It’s not enough to simply cook the ingredients until they are soft enough. Like soup, data has to acquire a flavour, and that can take some long slow simmering. Sometimes, even if you know the area and the data well, you can work with the data for months before its truths are finally revealed. Kudos to management for recognizing the value of analytics, but they need to learn patience too.

Moreover, they also need to recognize how easy it is for abstract numbers to corrupt or misdirect thinking. Ultimately, analytics is also bound up in performance measurement, accountability exercises, and such, and it is a struggle to avoid organizations being “managed-to-the-indicators”.

But you wanted some cool stories, right? Here’s mine.

Eight years ago, a former student of mine phones me up and wants to know if there is anything in the federal employee survey they could use to help them through a change-management exercise. His agency was blending with directorates coming from two other govrnment bodies to form a larger organization with a broader mandate.

As it happened, I had survey data for all the work units across the whole of government. I lined up the survey responses for his organization, against the work units they were to be merged with, and looked for cases where the expectations and experience of employees was different. I didn’t care about “better” or “worse”. The objective was to simply find differences. The rationale was that the merger would likely involve managers from one of the components eventually overseeing employees from the others. If there were any profound differences in their workplace cultures, conflict couldn’t be far away. Identifying such cultural differences, and apprising managers of them, so they can make anticipatory adjustments, would help to avoid such conflicts and facilitate the change.

I flagged about 4 or 5 areas where there were clear differences, one of them being what they construed as a “normal work week” (e.g., frequency of obligatory overtime), and another being rate of career progress. The survey itself had been devised as an accountability tool, but we were able to use the data in a proactive way to help shape the organization.

Reply
Profile Photo Michelle Mullin

In my line of work at EPA we pair statistics with spatial data to map potential contaminant plumes, exposure risks, etc. We also use it to help us decide where to take more samples, how many more samples are necessary, etc.

I’m unable to attend this particular webinar, but would be interested in future opportunities. Thanks for organizing this!

Reply