Note 1: Question order changes and children's government coverage - posted 9/28/05

The order in which Medicaid and State Children's Health Insurance Program (SCHIP) questions are asked in the 2004 panel is different from the 2001 panel. In the 2004 panel, respondents are first asked about SCHIP, then Medicaid, then other government health insurance. In the 2001 panel respondents were first asked about Medicaid coverage, then SCHIP, then other sources of government health insurance. Users should be aware that the question order differences are reflected in very different estimates between panels for the edited variable EMCOCOV. Users should not infer actual changes in the participation rates for these programs between the two panels.

Note 2: Identifying sources of government and private coverage - posted 9/28/05

SIPP defines Medicaid and Medicare as government insurance. SCHIP and other state health insurance programs are classified as Medicaid coverage. All other forms of health insurance and health care provisioning are broadly classified as private insurance. This allows users to specify non-Medicaid and non-Medicare health insurance sources (such as employer-provided insurance) and health care providers (such as military/VA health care) as government or private according to their own research needs. Users may use RPRVHI and RPRVHI2 to identify all types of insurance and health care other than Medicaid and Medicare.

Problem 1: Lower private health insurance rates for children - updated 10/5/06

After the initial 2004 Panel Wave 1 health insurance editing, it was found that many children whose parents had health insurance lacked coverage. This issue caused private insurance coverage rates for children age less than 15 to compare lower than the existing benchmarks of the 2001 Panel Wave 1, and the 2004 Panel Wave 2 preliminary data. The lower rates appear in both two parent and single-parent households. Specifically the lower rates are seen in two parent households where one or more parents indicate employer provided health insurance coverage, and in single family households when the single parent reports coverage from someone outside the household - and the person outside the households' coverage is provided by a current employer, work or business.

These findings may be due to a 2004 Panel instrument redesign of the health insurance core section, as well as to changes in the editing process. Because of the lower rates, coverage information for children in the previously released 2004 Panel Wave 1 SIPP files was suppressed. The suppressed variables and the values assigned to each were EHIMTH, EHIUNT1, EHIUNT2, EHIUNT3, EHEMPLY, RCHAMPM, RPRVHI, RPRVHI2, RCUOW58A, and RCUOW58B, a value of "-9." Suppressed variables RCUTYP58, AHEMPLY, and AHIMTH were assigned a value of "9."

To provide an indication of the extent of this issue comparisons between multiple instruments were conducted on married couple two parent households in which both parents were covered by employer provided health insurance (EHEMPLY=1), and where none of the children were covered by health insurance. The number of unweighted cases in which both parents were covered by private health insurance, but none of the children were covered in the SIPP 2004 Panel Wave 1 was 1,350 as compared to the SIPP 2001 Panel Wave 1 which had 278 cases. The preliminary 2004 Panel Wave 2 data showed 690 cases.

Single parent households who indicated private health insurance coverage from someone outside the household (ENONHH = 1), and the coverage was employer provided (EHEMPLY = 1) also had a higher than expected level of insured parent but uninsured children compared to benchmark data. SIPP 2004 Panel Wave 1 had 69 single parent household cases, compared to both the SIPP 2001 Panel Wave 1 which had 27 cases and the preliminary SIPP 2004 Panel Wave 2 data which has 23 cases.

Resolution of Problem 1:

Procedures were developed to assign private coverage to children whose data indicated no coverage. Applying the procedures to uninsured children increases the SIPP 2004 Panel Wave 1 estimate of children covered by health insurance from 58.67% before the procedure to 64.49% after. Applying the same procedures to the 2001 Panel Wave 1 also increases the number of children covered from 69.45% to 71.91%.

The procedures used to assign children private health insurance follow:

1. In two parent households where both parents indicate health insurance coverage and at least one parents coverage was employer provided, but no children less than 15 were assigned coverage by any type of private health insurance, then assign all children under age 15 in the household private coverage. Otherwise, if in a two parent household, at least one of the children under age 15 were assigned private health insurance coverage, but not all the children under 15 were assigned coverage, then make no change.
2. In single parent households where the parent indicated they were covered by private health insurance by someone outside the household (ENONHH = 1), and the private health insurance of the person outside the household was employer provided (EHEMPLY = 1), and where none of the children less than age 15 were covered by any type of health insurance, then assign health insurance to all children in the household less than 15, and indicate coverage by someone outside the household (ENONHH=1). Otherwise, if at least one child less than 15 in the one parent household has some type of health insurance, then make no change.
3. The allocated variables indicating whether a child had private insurance coverage each month of the reference period (AHIMTH1-4) are set to "5" for children assigned coverage based on 1. and 2. above in order to allow data users to identify instances where children have been assigned health insurance coverage.
4. The value for EHIOWNER is set to a "2," indicating coverage by someone else's plan, for children who have been assigned health insurance coverage with the above procedures.

Problem 2: Reasons for the lack of health insurance EHIRSN01-12 - posted 9/28/05

Due to an error in waves 1–5 of the 2004 panel, respondents are asked the reason(s) they did not have private health insurance only when they report being without health insurance for all 4 months of the wave. These reason(s) for the lack of health insurance are output as edited variables EHIRSN01-12 and allocation variable AHIRSN.

Resolution of Problem 2:

The error has been corrected for waves 6 and later. Beginning in wave 6, respondents without health insurance (other than Medicaid or Medicare) in one or more months of a wave are asked why they are/were without health insurance. These reasons for lacking private health insurance are output as the edited variables EHIRSN01-12 with the associated allocation variable AHIRSN.

Problem 3. SCHIP coverage for teens age 15–-19 - posted 9/28/05

Due to an error in waves 1–5 of the 2004 SIPP panel, respondents age 15-19 are inadvertently skipped over a question about SCHIP coverage. This question asks "At any time between ... and today were you covered by SCHIP, the State Children's Health Insurance Program, that helps families get health insurance?" Users should note that teens age 15–19 were skipped only when the teen was the first household respondent. When a guardian answers first, all children in the household for whom the guardian is responsible will have valid responses.

Resolution of Problem 3:

The error in waves 1-5 is corrected for waves 6 and later.

Problem 4. Medicare Coverage Waves 1 – 4 - posted 1/9/08

During Waves 1 - 4 of the 2004 Panel, the respondent's report of Medicare coverage was checked against their report of social security income and railroad retirement receipt. Despite a response of "yes" for Medicare coverage in the raw data, the edit process assigned a "no" value to Medicare coverage in the final data. The Census Bureau analysts suspect this was implemented panels ago to make sure people reporting Medicare coverage were eligible as determined by the U.S. Department of Health and Human Services on their official Medicare website: www.medicare.gov. Currently there are several ways a person may qualify to receive Medicare coverage.

Resolution of Problem 4:

Since SIPP does not allow a complete identification of eligibility with survey response, the edit specification was updated to allow respondents aged 65 or older, who indicated they had Medicare coverage, to be assigned Medicare coverage in the final edited data set (public use variables ECRMTH1-4=1). Prior to this Wave 5 specification update, in order for a respondent aged 65 or older to be assigned Medicare coverage, they also had to indicate they received either Social Security or Railroad retirement.

As a result of the code update, users should notice an increase in overall Medicare coverage in Wave 5+.