Table 2, Table 3 and Table 4 present breakdowns of NMES population estimates for number of mental retardation facilities, set up beds, current residents and residents with mental retardation and related conditions for the 3 facility groupings (type of operation, ICF-MR status, facility size). Discrepancies between these population estimates and what might be expected based on other data sources were noted in Part 1. Again, most significantly, the estimated 64,936 persons with mental retardation and related conditions in facilities with 15 or fewer residents was substantially less than the 104,189 persons with mental retardation that states reported in 1986 (the year in which the NMES sample frame was established). Only a small part of the discrepancy (10% or less) can be attributed to the decision to eliminate all the facilities of 1 and 2 residents that were in the sample frame. Discrepancies between NMES estimates and state reports of the number of large facilities, private and government operated, and the number of people living in them, is much less, and within the normal expectations of differences due to sampling. Indeed, had the total resident variable been available for categorizing facilities by size rather than requiring the use of "set up" beds, the difference between the state reports of 147,719 persons with mental retardation and related conditions in facilities of 16 or more residents and the NMES estimate of 153,619 persons with mental retardation and related conditions in facilities of 16 or more residents, would have been even smaller.
Grouping 1: Facility operation. Table 2 presents summary statistics on facilities by type of operation. The primary limitations in the NMES population estimates are among the private facilities and most specifically the small private facilities. The estimated number of such facilities and residents is considerably below the numbers known and reported by state mental retardation/developmental disabilities agencies. NMES estimates of government operated facilities and large private facilities are generally similar to what states reported for mid-1986, about the time the sample frame was developed (Lakin, Hill, White, & Wright, 1987).
Grouping 2: ICF-MR certification status. Table 3 presents summary statistics on facility groupings by ICF-MR/non-ICF-MR status. The NMES estimates of populations of both large and small ICF-MR facilities are quite similar to the statistics reported by states at the time the NMES sample frame was being developed (Lakin et al., 1987). States reported that on June 30, 1986 they had 20,890 residents with mental retardation and related conditions in small ICFs-MR. The NMES estimated 21,077. For the same date, states reported 122,925 persons with mental retardation and related conditions in large ICFs-MR. The NMES estimated 118,084. As noted before, non-certified facilities appear substantially underestimated In NMES when compared to state reports, with almost all of the discrepancy being in the smaller (15 or fewer resident) facilities.
Grouping 3: Facility size. Table 4 presents summary statistics on facility groupings by size ("set up beds"). Again, the problems with the NMES sample and population estimation appear generally limited to the smaller facilities. Whereas NMES estimated that in early 1987 11,054 facilities of 15 or fewer residents had 64,935 people with mental retardation living in them, states reported on June 30, 1986 that they had 27,205 facilities with 104,189 people living in them (Lakin et al., 1987). The population estimates from NMES indicate that the average size of facilities with 15 or fewer "set up beds" was 5.9 residents. Using state reported data of June 30, 1986, the average size of facilities with 15 or fewer residents (including facilities serving one or two residents, which were excluded from NMES) was about 2.4 residents.