For 1997 to 1999, those who did not answer the initial question affirmatively are coded NIU (not in universe) for follow-up questions on insurance characteristics. Beginning in 2000, the response "no insurance" was included as a valid response to the initial question. Thus, for 2000 forward, the universe for follow-up questions on insurance coverage type and details was "all persons" (including those with an affirmative response to HINONE or "no insurance").

In essence, for 1976 to 1996, Medicaid was defined as a health insurance program that provides medical assistance to persons in need. The Field Representative's Manuals for 1997 forward defined Medicaid as "a medical assistance program that provides health care coverage to low income and disabled persons. The Medicaid program is a joint federal-state program which is administered by the States."

1997 forwardInterviewers first asked, "Are you covered by health insurance or some other kind of health care plan?" For individuals who had an affirmative response, the interviewer asked, "What kind of health insurance or health care coverage do you have?" Respondents selected the appropriate response from a flashcard listing various types of insurance coverage, including Medicaid. The options on the card were as follows:

• Private health insurance (2004 forward)

• Private health insurance plan from employer or workplace (1997-2003)

• Private health insurance plan purchased directly (1997-2003)

• Private health insurance plan through a State or local government program or community program (1998-2003)

• Medicare (1997 forward)

• Medi-Gap (1997 forward)

• Medicaid (1997 forward)

• CHIP (Children's Health Insurance Program) (1999 forward)

• Military Health Care/VA (1997-2003)

• Military Health Care (CHAMPUS/TRICARE/CHAMP-VA) (1997 forward)

• Indian Health Service (1997 forward)

• State-sponsored health plan (1997 forward)

• Other government program (1997 forward)

• Single Service Plan (e.g., dental, vision, prescriptions) (1999 forward)

• No coverage of any type (2000 forward)

Respondents could pick more than one type of insurance, and interviewers were instructed to mark all that applied.

During the course of data editing, the NCHS discovered many errors in the responses to questions about insurance coverage. This might be shown, for example, by a mismatch between the verbatim name of an insurance plan and the type of insurance coverage the person selected from categories on a flashcard. Most researchers studying only the period from 1997 forward should use the back-edited variables such as HIPUBCOVE to analyze public insurance coverage. HIPUBCOV, which reflects respondents' original replies without back-editing, is more suitable for studying public coverage both before and after 1997.