ever, the estimates these data produce are somewhat dated, particularly because there is some expectation that acuity levels have been increasing in RCFs in most states. A more recent ASPE study has produced data on a national probability sample of assisted living facilities, residents, and staff (Hawes et al., 1999, 2000). Rough estimates of resident acuity levels in physical functioning (i.e., ADL) and cognitive status are available for all facilities (Hawes et al., 1999), while more detailed data (e.g., behaviors, falls) are available for residents in assisted living facilities that offer either high services or high privacy (Hawes et al., 2000).
Sampling decisions in residential care are complicated by the fact that there appears to be significant variation across states. Some, like North Carolina, have made aggressive use of RCFs to limit use of nursing homes. In these states, the resident mix is considerably more impaired physically and cognitively than in other states that have not pursued an aggressive expansion in the number of RCFs and in the level of care that may be provided in those facilities. In addition, there appear to be important differences in resident characteristics among different types of residential care facilities. For example, small homes are more likely to have residents with primary psychiatric diagnoses. Similarly, the assisted living sector of residential care is largely populated by persons who have higher incomes and are more racially homogeneous than residents in traditional RCFs. If these factors are expected to have an effect on the prevalence of abuse or neglect, this must be considered in designing the sampling plan of any study aimed at examining prevalence.
There does not appear to be sufficient information about staff risk factors from prior studies that would facilitate decisions about whether to oversample certain types of staff. For example, one study suggests that the belief that nursing home residents are childlike makes a staff member somewhat more likely to abuse residents, but there is no obvious way to initially select a sample based on this characteristic. However, the MEPS study provides general demographic data on nursing home staff, and the two ASPE studies also provide such data on staff demographics, training and education, and other characteristics.
My strong recommendation would be to depart from prior studies that interviewed only staff working during the shifts that field interviewers were on site. In practice, when in-person interviews have been used to collect data, this has meant that staff members who work during the morning and evening shifts during the week were overrepresented in staff samples. Night shifts and weekends, however, have a tendency to be short-staffed and to