A coalition of human services administrators and community advocates was puzzled why African Americans were disproportionately represented among families receiving Temporary Assistance for Needy Families (TANF). Some advocates were concerned that this happened because workers allowed African American families to languish in the program, instead of offering them intensive services. But our analytics showed that the survival curves for African American families and white families were indistinguishable. Then we modeled survival off TANF and found that African American families had much shorter spells off TANF than did white families, which pushed planning efforts in a wholly unexpected direction.
In another example involving TANF clients, immediately following reforms to federal work requirements, a county department was concerned about how few clients were meeting those requirements. Their starting assumption was that their clients were not complying. But our analytics proved that the department’s employees were not sending clients to assignments. This enabled the department to quickly direct their efforts toward worker training and supervision, with the result that their work participation metrics improved rapidly.