While there have been a number of important studies of maternity group homes, there is much left to be learned. Early studies have provided useful descriptive data on some maternity group homes and their residents, but far less is known about many other homes. In addition, limitations in the methodology used in these studies severely limit what conclusions can be drawn about effectiveness. A rigorous evaluation of maternity group homes could address many of these gaps in the research.
- Increasing substantive depth and breadth. The literature on maternity group homes could be greatly advanced by studies that are both broader and deeper, examining a wide range of topics across a number of different homes. In particular, additional research on such topic areas as implementation and operation issues and longer-term participant outcomes would be useful. Substantive areas that are not addressed by most studies could be explored more extensively in a larger number of maternity group homes. Even topics that are commonly addressed in the existing research, such as education, could be investigated more thoroughly.
- Using control or comparison group. A rigorous evaluation could address the most serious methodological limitation of past studies by identifying an appropriate comparison group, then following comparison group members as well as program participants and collecting the exact same data for both groups. The most rigorous method for selecting a comparison group is through a random assignment design, whereby all applicants to the program being studied are randomly assigned to either the treatment or the control group. If random assignment is not feasible, either because of ethical considerations in the absence of excess demand or other practical issues, other methods can be used to identify as similar a comparison group as possible.
- Increasing sample sizes. To accurately measure the effectiveness of maternity group homes, future studies will need to ensure adequate sample sizes to generate sufficient statistical power. This could be achieved by focusing on only large maternity group homes; by pooling data across a number of similar homes, perhaps within a network; by extending the sample intake period; or by a combination of these. However, care must be taken in implementing any of these suggestions. Large homes may differ from small homes in other ways than size, so results of a study on large homes may not be representative of all maternity group homes. Pooling data requires identification of homes that are similar enough to be considered as one. Extending intake periods could be complicated by changes over time in the maternity group home programs or the populations they serve.
- Reducing sample attrition. To reduce the problems caused by sample attrition, evaluations can implement procedures to both increase response rates at followup and use statistical methods to adjust for any nonresponse bias. Rather than rely on busy program staff to assist in contacting former residents, evaluators could track study participants themselves, requesting contact information from them directly before they leave the maternity group home; incentives can be provided to encourage former participants to respond once they are located.
As MPR's design study progresses, we will refine these recommendations and develop a design for a future evaluation of maternity group homes. The next steps in the design study include calling a number of homes to collect basic information on their characteristics, conducting site visits to assess the evaluability of select homes, developing questionnaires, and drafting data-collection and evaluation plans. The design effort will culminate in a final report discussing the feasibility of a rigorous evaluation of maternity group homes. By addressing gaps in past studies, such an evaluation could contribute a great deal to what is known about the effectiveness of maternity group homes.
1. This study included only former residents who had lived in a maternity group home for at least 30 days. Including residents with very short stays in the computation would result in a lower mean.
2. Other types of public assistance followed the same pattern. At intake, 52 percent were receiving food stamps, 72 percent Medicaid, and 38 percent WIC. At exit, these percentages had risen to 89 percent, 52 percent, and 87 percent, respectively. But all had fallen by followup, when 52 percent were still receiving food stamps, 66 percent Medicaid, and 14 percent WIC.
3. About 14 percent of children in Teen Living Program homes had asthma, compared to 11 percent of children in the general population nationwide (Reich 1996; U.S. Department of Health and Human Services 2004).
4. Response rates are likely to be even lower in studies that did not report them on provide other details on methodology.