The factorial structure of the Personal Health Survey in normals and schizophrenics.

Administered the Personal Health Survey to five different groups: 74 incarcerated felons, 47 hospitalized alcoholics, 172 unmarried mothers, 51 college students, and 386 hospitalized chronic schizophrenics. On 191 of 200 items, the schizophrenics had higher base rates than any other group. Although some groups did not have large enough Ns for statistically adequate factor analyses, factorial studies also were done on the four control groups. Every group studied gave different factor patterns, with different orders of emergence of factors, different item loadings and heterogeneous composition of items. It is concluded that subgroups should be factored independently across time to determine population characteristics because overall population patterns do not make possible more than general predictions about subgroups. Individuals cannot be predicted from either overall or subgroup factorial characteristics.