Unleashing Evidence-Based Management in Government

By training 100 mid-level staffers as 'Data Fellows,' New Jersey's child-welfare agency has found a formula for improving outcomes.
March 23, 2015
By John M. Kamensky  |  Contributor
John M. Kamensky is a senior fellow with the IBM Center for the Business of Government.

If evidence-based decision-making is all the rage in government these days, how do agencies develop the skills to conduct such analyses? And how do agencies create a culture to support this management approach so it isn't just another flavor-of-the-month fad?

New Jersey's child-welfare system may have cracked the code. A new report for the IBM Center for the Business of Government, by David Lambert and Julie Atkins of the University of Southern Maine, tells the story.

Faced with years of spotty performance in ensuring the safety and welfare of at-risk children, New Jersey's Department of Children and Families (DCF) felt good about its progress after a three-year upgrade in data and technology. By 2009, the report's authors write, "managers could now turn on their computers and see in real time the proportion of investigations completed within the target of 60 days in each of New Jersey's 10 areas and 47 local offices." Nevertheless, departmental leadership felt that performance had not measurably improved. In short, managers were not using the data to manage.

At about the same time, the U.S. Department of Health and Human Services (HHS) had funded a program to help state child-welfare agencies design and implement programs to improve system outcomes. New Jersey became part of HHS' initiative and developed an 18-month program to train 100 "Data Fellows" from different levels within the agency. "This initiative succeeded in changing the technical skills, attitudes and practice of a diverse group of workers," note the authors.

The success of the initial "Manage by Data" program has encouraged DCF to extend the program to two additional cohorts of departmental employees, using the department's own funds, since DCF's leadership realized this could not be a one-time investment.

New Jersey launched the development of its program in mid-2009, based in part on experiences piloted in five other states. While the department initially considered training all frontline employees, there was concern that this might result in superficial skill development. DCF ultimately chose to train 100 workers in more depth, targeting mid-level staff with at least 10 years with the department and from different areas of the state.

The DCF Fellows Program launched in mid-2010. The curriculum involved classroom work, coaching, group projects and special events. There were three phases of six day-long seminars covering how to be knowledgeable consumers of data, using data to understand and manage change, and how to put the diagnostic pieces together to improve outcomes.

Fellows' projects were based on challenges that bubbled up from area and local offices, and were in the context of the departmental strategic priorities. One group, for example focused on improving parent/child visitation. The Fellows found that children placed with kin were more likely to have parent visits (as expected) but that fewer visits were occurring than expected because they typically took place in forbidding-looking welfare offices. So the group recommended better training around how to make the visits more welcoming for the families.

The initiative has led to a number of changes in the way DCF conducts its operations. For example, Fellows found that offices that took the time to hold more frequent supervisor conference investigations with staff tended to save time and produced better results. In another case, Fellows learned that hotline call-screeners had highly variable rates of anonymous calls (ranging from 3 percent to 50 percent). By observing the best screeners, they were able to develop a script for use by all screeners that resulted in far fewer anonymous referrals.

More strategically, the effort gave Data Fellows the opportunity to exercise leadership and to serve as "data ambassadors" with their colleagues, regardless of their role within the department. It also contributed to the use of monthly "ChildStat" meetings at the departmental level as learning sessions to create culture around data. Area directors began to hold quarterly sessions on their own, comparing their performance with other area offices in order to improve their own.

New Jersey's experience holds lessons for other states. The performance movement over the past few decades has been successful in creating a supply of performance information. It has had trouble, however, in creating a demand for it by decision makers. The New Jersey program demonstrates the importance of engaging managers and frontline staff in establishing a model for data-driven management.