While these studies provide useful descriptive information on maternity group homes and their residents, they suffer from several limitations that reduce the usefulness of their findings. There are a number of gaps in the breadth and depth of knowledge collected in specific substantive areas. In addition, these studies have suffered from a number of methodological issues, the most serious of which is the lack of a comparable control group.
The literature on maternity group homes provides limited information on a wide range of substantive topics, but more comprehensive data have been collected on only a few topic areas. Most studies either provide a brief look at of a number of different homes (the SPAN directory, for example, which catalogs almost one hundred different homes but provides only basic data on their characteristics), or a more intensive examination of a single maternity group home or network (for instance, the evaluation of Massachusetts' Teen Living Program). Thus, knowledge of some substantive issues related to maternity group homes is based on data collected from only a small minority of homes, which may not be representative of other homes nationwide. Some specific topics areas are not addressed in most studies; for example, there is little information on domestic violence or substance abuse outcomes in the research on maternity group homes. Few studies were able to explore longer-term outcomes of former residents due to short follow-up periods, or, in many cases, no followup at all after residents left the homes. In addition, few studies explored the implementation of maternity group homes, resulting in limited information on the challenges faced and lessons learned as staff operate maternity group home programs.
Besides substantive gaps, studies of maternity group homes have suffered from the following methodological drawbacks:
Lack of comparison or control group. In order to assess program effectiveness, a study needs to be able not only to track the outcomes of program participants, but also to determine how they would have fared in absence of the programs. To accomplish this, evaluations must include an appropriate comparison group and be able to control for differences between participants and comparison group members. The ideal approach is an experimental design with a control group; if that is not feasible, a well-constructed comparison group can serve the same purpose. However, no studies in the literature on maternity group homes focused on developing a comparison group and collecting the same data on comparison group members as participants. As a result, it is difficult to interpret findings on outcomes. Many existing studies did not provide any comparisons for outcomes of maternity group home residents at all. When comparisons were drawn, statistics on program residents were typically compared to external data on the general population of teenage mothers or to other similar populations, which most likely differ from program residents in meaningful ways. In addition, the external data sources to which studies compared the outcomes of maternity group home residents likely used different data collected methods, measures of key outcomes, follow-up time periods, and so on. A few studies compared characteristics of the same individuals at different points of time, but this method can capture changes in their outcomes over time that would have occurred anyway, with or without the program intervention.
Small sample sizes. Program capacity, combined with average length of stay, determines how many teens can enter a maternity group home during a given intake period. As discussed in Chapter II, most maternity group homes have capacity to serve fewer than a dozen residents at a time. This limited capacity results in small sample sizes for most studies of maternity group homes, which may be one reason for the lack of more rigorous studies. The sample sizes of the largest studies included in this review range from 80 to 199 families. These numbers were achieved by pooling data across a number of different homes within a maternity group home network and/or using an intake period of up to four years (Fischer 2000; Sawyer 2000; and Collins, Stevens, and Lane 2000). Most other sources in the maternity group home literature do not state the sample sizes on which reported outcomes are based, although extrapolating from the capacity of the programs they discuss suggests that they are likely to be considerably smaller than those reported above.
Sample attrition. Another problem with many of these studies is one of sample attrition. Accurate measures of characteristics at followup require high response rates among former participants, since the outcomes of respondents may differ from those that did not respond, resulting in nonresponse bias, in addition to a reduction in statistical power. But tracking participants after they leave a program can be difficult, particularly when the program is targeted to a population with a history of housing instability. Maternity group home residents all change residences when they exit the program, and, as discussed above, many move into temporary housing. Multiple changes of address complicate tracking, and inability to track participants can lead to lower response rates. In the few studies that tracked participants after they left maternity group homes, response rates at followup ranged from 55 percent (in a study whose followup occurred as long as four years after program exit) to 65 percent (Fischer 2000; and Collins, Stevens, and Lane 2000).(4) Maternity group homes were unable to provide researchers with contact information for some residents, including those who had unlisted numbers or had moved out of the state or gone underground due to domestic violence.