1 If a recipient had “L” (leave of absence, n=599) or an “I” (interim eligible, n=72) status in all listed months in 2005, they were not eligible for inclusion.
2 Several decision rules were adopted to address persons with multiple providers in any month or who changed provider types during the year: (i) If any individual appears in the Parent/Spouse group for any month in the year, they were classified as in the Parent/Spouse group, regardless of any other combination of providers. This corresponds to the assumption of “an intention to treat.” (ii) If the recipient-provider relationship is missing in all eligible months they were classified into the Non-Relative group. (iii) If an individual appears in a combination of Other Relative and Non-Relative groups during the year, they are classified as in the Other Relative group. This again assumes an intention to treat. (iv) If an individual has a combination of Other Relatives (e.g., minor/adult child, or Other Relatives) during the year, they are classified as having an Other Relative provider. If they had multiple Non-Relative providers in the year, they were classified in the Non-Relative provider group.
3 Additional county-level attributes were considered and tested in earlier analyses. These included alternative long-term care service supply measures like nursing homes, ICF-MR, adult care facilities, residential care facilities for the elderly; proxy measures for personal assistance labor supply like the per capita number of racial/ethnic minorities, and proxy measures for service demand like the per capita number of aged. These measures are associated with IHSS participation in counties, but they do have consistent associations with the choice of provider types for recipients in the IHSS program itself. For this reason they have not been used in the final models.