Study of Substance Abuse Case Management Program

Temporary Assistance for Needy Families and Safety Net recipients whose application questionnaires indicated possible substance abuse were referred via an electronic scheduling system to New York City’s Human Resources Administration (HRA) for more detailed assessments. When the SACM program had available slots, the automated scheduling system referred applicants to the program. Otherwise, applicants were referred to services as usual. The referral process compromised randomization because new SACM slots always became available in the morning. Thus, in instances of systematic differences between applicants who applied for appointments in the morning and those who applied in the afternoon, the assignment mechanism would not generate equivalent groups. For example, those eager to engage in treatment and return to work could be more likely to show up for scheduling early in the morning. In addition, staff might have known that the probability of assignment to the SACM group was higher in the morning and thus encouraged participants who, in their opinion, needed extra help to show up earlier rather than later. The study authors argue that such probabilities are remote but later present evidence of differences between the groups and acknowledge that the assignment process was not entirely random. Consequently, the integrity of the random assignment process was compromised, and the study was evaluated as a quasi-experimental comparison group design.