Measurement of Income in NCHS Surveys Diane M

Measurement of Income in NCHS Surveys Diane M. Makuc NCHS Data Users Conference July 12, 2006 Centers for Disease Control and Prevention National Center for Health Statistics

Outline of Presentation • Why collect income in health surveys? • • • NCHS surveys with income data Measurement of income Data issues and limitations

Why collect income in health surveys? • • Income affects health and health care • Evaluate health policies to improve health / health care for low income population Monitor socioeconomic disparities in health • Healthy People 2010 • Congressionally mandated reports

Income and Health Status Income Health behaviors Education Environment/ Living conditions Health status Health insurance/ Health care Psychological/ Psychosocial factors Early life experiences Human biology/ genetics

Percent of Persons with Untreated Dental Caries by Povert Status Ages 6 -17 years Ages 18 -64 years <100 100 -<200 <100% 100 -<200% 200+% 1971 -74 200+ Year 1988 -94 1999 -02 1971 -74 1988 -94 Year Data Source: National Health and Nutrition Examination Survey 1999 -02

NCHS surveys that collect income data • • National Health Interview Survey (NHIS) • National Survey of Family Growth (NSFG) National Health and Nutrition Examination Survey (NHANES)

NCHS surveys that collect income data State and Local Area Integrated Telephone Survey (SLAITS) • National Survey of Early Childhood Health (NSECH) • National Survey of Children with Special Health Care Needs (NSCSHCN) • National Survey of (NSCH) Children’s Health • National Asthma Survey (NAS)

Income and related data collected by NHIS, NHANES, NSFG • • Individual earnings amount – NHIS only Sources of family income - all • For each family member - NHIS / NHANES • Total family income amount - all • Program participation - all • For each family member - NHIS / NHANES • Home ownership – NHIS only

Total family income amount questions: NHIS and NHANES • Exact total combined family income amount in previous calendar year (NHIS) / last 12 months (NHANES) • Followup questions for nonresponders to exact amount • 2 categories (< $20, 000; > $20, 000) • 44 categories (< $1, 000…. ; > $75, 000)

Total family income amount questions: NSFG • 14 categories of total combined family income • • (<$96…. > $1442 in week or <$417… > $6250 in month or <$5000…> $75000 in year) Followup questions for nonresponders to 14 category question • 3 categories – questions depend on previous responses • Is income <$20, 000? ; if no, is it >$50, 000?

Total family income amount questions: SLAITS • Exact amount in previous calendar year for family (NSCH, NSECH) or for household (NSCSHCN, NAS) • Followup questions for nonresponders • 15 income categories - questions depend on previous responses e. g. Is income < $20, 000? If yes, is it <$10, 000? If yes, is it > $7, 500? • 9 categories of poverty ratios - questions depend on previous responses (NSCH, NSCSHCN).

Income data collection method • Computer assisted personal interview in household (CAPI) • NHIS and NHANES • Audio computer assisted self-administered interview (audio CASI) • • NSFG Computer assisted telephone interview (CATI) • SLAITS - all 4

Income variables on public use files: % missing (unweighted) and number of income groups NHIS (2005) 29% -- 11 groups NHANES (2003 -4) 6% -- 11 groups NSFG (2002) 8% -- 14 groups NAS (2003) NSCH (2003 -4) 23% -- 11 groups 9% -- 8 groups NSCSHCN (2000 -2) 15% -- 9 groups NSECH (2000) 11% -- 8 groups

Factors that may affect income nonresponse • Data collection mode - Income question is self - administered in NSFG • Survey content – NHANES respondents receive physical exams and lab tests • Question wording – SLAITS uses “bracketed” questions for nonresponders to exact income • Survey response rate - NHIS has higher survey response rate (86% in 2005) than other surveys

Imputation of family income amount • NHIS - multiple imputation of missing income values • Available on public use files about 3 months after initial file release • NSFG – single imputation of missing income values • NSCSHCN – single imputation of low income status for uninsured with missing income values

Data issues in analyzing income and health • Surveys may exclude some of the lowest income persons who are likely to have health problems • Household surveys exclude homeless and institutionalized • Telephone surveys exclude those without any phone and those with only cell phones

Data issues in analyzing income and health • Responses to health questions (particularly subjective health measures) may be influenced by income level. • Income in a single year has limitations as a measure of financial resources • It may underestimate the effect of income on health. • For older people, wealth may be a more important measure of financial resources.

Summary • Income data are difficult to collect resulting in high nonresponse in many surveys. • Differences in income response among surveys indicate that improvements may be possible. • Imputation of missing values may reduce bias associated with high nonresponse.

For more information See NCHS website //http: //www. cdc. gov/nchs/