The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.
Survey Methods
The Snapshots of America's Families are based on information obtained from the 1997 National Survey of America's Families. We used probability sampling methods to select households in 13 states and in the balance of the nation. The resulting sample is representative of the noninstitutionalized, civilian population of persons under age 65 in the states studied and in the nation.
Before administering the interview, we screened households to determine eligibility. In households with children, we selected up to two children at random, one under age 6 and one between the ages of 6 and 17. Information about the children and the household was obtained from the adult in the household who was most knowledgeable about the health care and education of the children on whom the survey questions focused. In some households with children, adults under age 65 who had no children were also surveyed. In households without children, one or two nonelderly adults were selected randomly for interviewing. We collected data from February to November using computer-assisted telephone interviewing (CATI) technology. The sample consists of households with and without telephones. In households without telephones, cellular phones were provided to complete interviews.
The overall response rate was 65 percent for interviews about children and 62 percent for interviews about nonelderly adults. We weighted responses to the interviews to provide estimated values appropriate to the individual states and the nation. The weights adjust for design features of the sample, including oversampling low-income households and the study states, as well as nonresponse and undercoverage. Standard errors, which provide a measure of the precision of the estimates, were calculated using replication methods to account for the complexity of the sample design. Missing responses were imputed for questions regarding selected demographic characteristics, home ownership, affordability of housing, education, employment, earnings, income, and health.
Precision of Estimates
As in any other survey, the estimated values reported here have an implied degree of precision. For each estimate, we define a 95 percent confidence interval, meaning that we are 95 percent certain that the actual value in the population is within a given range of the estimate, where that range depends upon the standard error of the estimate. For example, in one we estimated that 65 percent of low-income adults were employed at the time of the survey in 1997. The standard error for this particular statistic is 0.66 percentage point. This implies that we are 95 percent certain that the true value is within 1.3 percentage points of 65 percent, that is, between 63.7 and 66.3. Because sample sizes are smaller for the states, confidence intervals are larger for state estimates than for national estimates. For example, we estimated that 70 percent of low-income adults in Colorado were employed. The standard error for this particular calculation is 1.5 percentage points, implying that we are 95 percent certain the true value is within 3 percentage points of 70 percent, or between 67 and 73.
An assessment of whether a state average is equal to the national average must take into account the imprecision of both the state and national estimates. To test for equality, the difference between the state and national averages was compared to the standard error of the difference. We applied a .05 significance level for a two-tailed test. State averages that are statistically different from the national average are indicated in the text as being above or below the national average and are printed in color in the tables. At this significance level, the difference between the state average and the national average will show up as statistically significant 5 percent of the time, even when it is not. For each indicator, if the national average were no different from any of the 13 state averages, there would still be a 50-50 chance that one of the 13 comparisons would be reported as statistically significant.
Finally, standard errors vary somewhat by state. It is therefore possible that not all states with similar estimates will differ from the national average to a statistically significant extent. Measures of statistical significance depend upon the difference between the two values being compared and the standard errors of the two estimates, which are affected by sample size.
Rounding
Estimates have been rounded to the nearest tenth in the tables and to the nearest whole number in the text and charts. Estimates that were originally at or above 0.50 have been rounded up, and estimates that were below 0.50 have been rounded down. An estimated value between x.45 and x.49 is rounded up to x.5 in the tables and down to x in the charts and text.
Timing
The majority of values in the Snapshots are point-in-time estimates of family characteristics at the time the survey was conducted in 1997. A notable exception is that income classifications (below the poverty level, below 200 percent of the poverty level, or above 200 percent of the poverty level) are based on 1996 information. Any other time frame used to calculate estimates is noted in the appropriate Snapshot.
Definition of Terms
Unless noted otherwise in the section on individual Snapshots, the following terms were used throughout:
| Adult |
A person between the ages of 18 and 64. |
| Child |
A person under the age of 18. |
| Higher Income |
Family income above 200 percent of the federal poverty level. |
| Low Income |
Family income at or below 200 percent of the federal poverty level. |
| Married |
Individuals who are legally married to someone living in the same household. |
| Parent |
An individual who identifies himself or herself as the adult in the household most knowledgeable about the child. In 95 percent of cases, this adult is the biological, adoptive, step-, or foster parent of the child; in 77 percent of cases, this adult is the child's biological, adoptive, step-, or foster mother. |
| Poverty |
At or below the federal poverty level, which is an annual income that varies by family size and composition. The following are examples of the federal poverty level in 1996:
| One adult with no children |
$8,163 |
| One adult with one child |
$10,815 |
| One adult with two children |
$12,641 |
| Two adults with no children |
$10,507 |
| Two adults with one child |
$12,629 |
| Two adults with two children |
$15,911 |
|
For more detailed statistical information on estimates (including information on actual numbers and standard errors), visit our Web site at http://newfederalism.urban.org.
Notes on Individual Snapshots
Authors' notes and acknowledgments for individual Snapshots, when appropriate, are listed below.
A-1 Poverty among Nonelderly Americans
Total cash income is the sum of income from the following sources: earnings from main job, earnings from self-employment, earnings from another job or jobs, Social Security, child support, foster care payments, Supplemental Security Income, Aid to Families with Dependent Children, General Assistance, pension or annuities, unemployment compensation, workers' compensation or veterans' payments, Emergency Assistance, money from friends or relatives, interest or dividends, rental income, and other income. New data on family relationships may alter the poverty classification of less than 0.3 percent of the households in the sample.
The authors would like to thank Margaret Weant for her review.
A-2 Poverty among Children
The National Research Council's recent study on measuring poverty recommends moving toward a measure that includes noncash resources available to families and deducts work-related (primarily child care) and health expenses. See: National Research Council. 1995. Measuring Poverty, A New Approach. Washington, D.C.: National Academy Press.
The authors would like to acknowledge Sean Williams and Ariel Halpern, who constructed the family structure variables, as well as Margaret Weant for her review.
A-3 Employment
Parents include all adults living with a biological, adopted, step-, or foster child under the age of 18 (available data do not provide distinctions among types of children), but exclude those whose child or children live outside the home.
The author would like to thank Daniel McKenzie, who calculated all estimates for this fact sheet, and Pamela Loprest, Greg Acs, and Sheila Zedlewski for their review.
A-4 Affordability of Housing
The authors would like to thank Thomas Kingsley for providing background materials.
A-5 Food Concerns and Affordability
Though data were collected at the family level, the findings are reported for individuals. Each person was represented equally and received the food indicator values reported for her or his family. Income was also measured at the family level and attached to individuals in this fashion.
Treated as a scale (0, 1, ... etc. positive indicators of food issues), the indicator variables have both internal validity (alpha = .865) and external reliability (i.e., high correlations with other measures of financial hardship).
The citation in the text is to 1997. Household Food Security in the United States in 1995: Summary Report of the Food Security Measurement Project. Alexandria, Va.: U.S. Department of Agriculture, Food and Consumer Service, Office of Analysis and Evaluation. This report also developed a five-way classification of U.S. households by their level of food security, a concept quite different from the measures reported here. However, the USDA classification was based in part on the current indicator variables (plus 14 more that were not on NSAF). An indicator-by-indicator comparison shows that the two data sources are closely aligned in every case. (The USDA data came from a special supplement to the April 1995 Current Population Survey.)
The authors would like to thank Anna Kondratas for help in designing and interpreting the food affordability measures and Martha Burt and Jennifer Ehrle for performing validity and reliability tests on the data in scale format.
B-1 Health Insurance Coverage of Children
B-2 Health Insurance Coverage of Nonelderly Adults
B-3 Confidence in the Ability to Get Children Medical Care
B-4 Children and Nonelderly Adults with No Usual Source of Health Care
B-5 Health Status of Nonelderly Adults and Children
The data presented in Snapshots B-1 and B-2 show a lower percentage of children and nonelderly adults being uninsured than reported through the Census Bureau's Current Population Survey (CPS). Although there are many differences between NSAF and CPS, two fundamental differences relate to the surveys' approaches to measuring insurance coverage. First, CPS measures insurance coverage during the calendar year prior to the survey (which occurs in March), while NSAF measured insurance coverage at the time of the survey. As a result, the CPS uninsurance rate would not seem to be directly comparable to the statistics reported in the Snapshots. However, analysis by many researchers of the responses to the CPS questions suggests that the CPS uninsurance rate is closer to an estimate of the proportion of people who lack insurance coverage at the time of the survey (see Lewis, K., M. Ellwood, and J.L. Czalka. 1998. Counting the Uninsured: A Review of the Literature. Washington, D.C.: Urban Institute). If this latter interpretation is valid, then the CPS uninsurance rate should be consistent with the statistics in these Snapshots. Although this issue does not explain why NSAF estimates of the uninsurance rate are lower than those of CPS, the conceptual difference between NSAF and CPS needs to be clear.
The second difference is likely to be more relevant. CPS asks a series of questions about insurance coverage and then assumes that any person not designated as being covered through any type of health plan is uninsured. NSAF used a series of questions similar in wording to CPS (with the exception of the time frame) but added a question that verifies whether people who appear not to have coverage are, in fact, uninsured. A substantial number of respondents used this opportunity to designate a particular type of coverage for those who initially appeared to be uninsured. Were it not for this verification question, the proportion of people lacking health insurance at the time of the NSAF survey would have been slightly greater than the uninsurance rate published by the Census Bureau. A more detailed analysis of the implications of this verification question will be available shortly.
In families with two parents, information on adults' health insurance coverage, health status, access to health care services, and utilization of services during the past year was obtained by randomly selecting either the respondent or the respondent's spouse or partner as the focus of the questions. Therefore, in 50 percent of cases the respondent is the focus of the questions, and in 50 percent of cases the respondent's spouse or partner is the focus.
The authors wish to thank Bahar Fadillioglu, Beth Kessler, and Aparna Lhila for their assistance with data.
C-1 Family Structure
In this Snapshot, family structures have been defined as follows: two-parent—child living with two biological or adoptive parents; blended—child living with one biological or adoptive parent whose spouse has not adopted the child; one-parent—child living with one biological or adoptive parent who may or may not be living with another adult or adults; no-parent—child living with either no parents or adults other than his or her parents.
The text refers to recent initiatives to promote marriage. Representatives David M. McIntosh (R-Indiana) and Jerry Weller (R-Illinois) are sponsoring the Marriage Tax Elimination Bill, which would eliminate the penalty married couples can face when pushed into a higher tax bracket.
The authors would like to thank Sean Williams for writing the base code for the family structure variables.
C-3 High Engagement in School
A scale to measure a child's engagement in school was developed by Jim Connell and Lisa Bridges at the Institute for Research and Reform in Education. Respondents were asked to describe whether the child cares about doing well in school, only works on schoolwork when forced to, does just enough schoolwork to get by, and always does homework. The response categories were "all of the time," "most of the time," "some of the time," or "none of the time." Answers were summed to create scale scores ranging from 4 to 16. A score greater than or equal to 15 indicates high engagement in school.
The four scale items were selected from a larger set of items contained in the parent report version of the Rochester Assessment Package for Schools (RAPS-P). The RAPS-P has been administered to approximately 5,000 parents in diverse school districts. The selection of the four items was based on analysis of 1,600 parents of children enrolled in grades 2 through 8 in the Rochester City School District. Selection criteria were determined using flags for threshold values for poor attendance, suspension from school, grade retention, failing grades in at least two core subject areas, and low standardized test scores. Selected questions had significant correlations with a number of flags, significant associations with at least two of the five flags, and low to moderately significant interitem correlations (.20 to .38). Using NSAF data (without weights), the alpha coefficient for these items is .76.
References relating to school engagement include Connell, J.P., M.B. Spencer, and J.L. Aber. 1994. Educational risk and resilience in African-American youth: Context, self, action, and outcomes in school. Child Development 65: 493-506; and Wellborn, J., and J.P. Connell. 1987. RAPS-S: Rochester assessment package for schools student report. Unpublished manuscript. Rochester, N.Y.: University of Rochester.
The authors wish to thank Chris Botsko and Suzanne LeMenestrel for assisting in the development and implementation of this measure, Carol Emig and Kathryn Tout for reviewing this text, and Sharon Vandivere for data production assistance.
C-4 Reading and Telling Stories to Young Children
Relevant references include Reese, E. 1995. Predicting children's literacy from mother-child conversations. Cognitive Development 10: 381-405; Duncan, G.J., J. Brooks-Gunn, W.J. Yeung, and J.R. Smith. 1998. How much does childhood poverty affect the life chances of children? American Sociological Review 63 (June): 406-23; and Zill, N., and E. Wolpow. 1991. School readiness: Examining a national goal. Early Childhood Education, (May): 14-16.
The authors would like to acknowledge Brett Brown, Carol Emig, and Kathryn Tout for reviewing this text and Sharon Vandivere for assistance with data production.
C-5 Participation in Extracurricular Activities
Relevant references include Brooks-Gunn, J., and F.F. Furstenberg. 1987. Continuity and Change in the Context of Poverty. In The Malleability of Children, J.J. Gallagher and C.T. Ramey, eds. Baltimore, Md.: Brooks, 171-87; National Commission on Children. 1991. Beyond Rhetoric. Washington, D.C.: Government Printing Office, 233-38, 243-44; and Moore, K.A., J. Manlove, D.A. Glei, and D.R. Morrison. 1998. Nonmarital school-age motherhood: Family, individual, and school influences. Journal of Adolescent Research 13(4): 433-57.
The authors wish to thank Chris Botsko and Suzanne LeMenestrel for assisting in the development and implementation of this measure, Brett Brown, Carol Emig, and Kathryn Tout for reviewing text, and Sharon Vandivere for data production assistance.
C-6 Behavioral and Emotional Problems in Children
This set of questions on behavioral and emotional problems of children was developed for the National Health Interview Survey (NHIS). Respondents for all children were asked to indicate whether the child doesn't get along with other kids, can't concentrate or pay attention for long, and has been unhappy, sad, or depressed. Respondents for 6- to 11-year-olds were asked to indicate whether the child feels worthless or inferior, has been nervous, high-strung, or tense, or acts too young for his or her age. Respondents for 12- to 17-year-olds were asked whether the child has trouble sleeping, lies or cheats, or does poorly at schoolwork. The possible response categories were "often true," "sometimes true," or "never true." The answers were summed to create scores ranging from 6 to 18. A score less than or equal to 12 indicates greater behavioral and emotional problems.
The NHIS Mental Health Indicator consists of problem items that were identified as providing the best means of discriminating between demographically similar children who were referred or not referred for
mental health services. The NHIS indicator distinguishes items further by age and sex. The NSAF scale distinguishes by age but not by sex. Using NSAF data (without weights), the alpha coefficient for these items is .73 for the scale used for children age 6 to 11 and .75 for the scale used for children age 12 to 17.
A relevant reference to this topic is Downey, G., and J.C. Coyne. 1990. Children of depressed parents: An integrative review. Psychological Bulletin 108: 50-76.
The authors wish to thank Brett Brown, Carol Emig, and Kathryn Tout for reviewing this text and Sharon Vandivere for assisting with data production.
D-1 Parental Participation in Volunteer or Religious Activities
A reference relevant to this topic is Bronfenbrenner, U. 1970. Two Worlds of Childhood. New York: Russell Sage Foundation.
The authors wish to thank Brett Brown, Carol Emig, and Kathryn Tout for reviewing this text and Sharon Vandivere for assisting with data production.
D-2 Parental Aggravation
The four-item parental aggravation scale was adapted from a component of the National Evaluation of Welfare-to-Work Strategies (NEWWS), the evaluation of the Job Opportunities and Basic Skills (JOBS) program. The aggravation scale in the NEWWS study used a 10-point response scale, whereas NSAF used a three-response category scale to accommodate a telephone interview. Respondents' answers were summed to create a scale score ranging from 4 to 16. A score less than or equal to 11 indicates high aggravation in parenting. Using NSAF data (without weights), the alpha coefficient for these items is .63.
A relevant reference to this topic is McGroder, S. Parenting among single, low-income, African-American mothers with preschool-age children: Patterns, predictors, and developmental correlates. Doctoral dissertation, The Pennsylvania State University.
The authors wish to thank Brett Brown, Carol Emig, and Kathryn Tout for reviewing this text and Sharon Vandivere for assisting with data production.
D-3 Mental Health of Parents
This five-item mental health scale (MHI-5) was constructed for the Medical Outcomes Study (MOS) by selecting the five items that best predicted the summary score for the 38-item Mental Health Inventory. Parents were asked how often in the past month they had been a very nervous person, felt calm and peaceful, felt downhearted and blue, been a happy person, and felt so down in the dumps that nothing could cheer them up. The answers to the MHI-5 are calibrated to a 100-point scale. A score of 67 or less indicates poor mental health.
Research has shown the sum of the five items used in the MHI-5 (without weights) to correlate at .95 with the 38-item mental health scale. Further, in a study comparing the MHI-5 scale with other scales measuring mental health status, the five-item scale was as good as the MHI-18 and the GHQ-30 and superior to the SSI-28 in detecting most significant disorders, including major depression, general affective disorders, and anxiety disorders. Using NSAF data (without weights), the alpha coefficient for these items is .81.
Relevant references include Hall, L.A., D.N. Gurley, B. Sachs, and R.J. Kryscio. 1991. Psychosocial predictors of maternal depressive symptoms, parenting attitudes, and child behavior in single-parent families. Nursing Research 40: 214-20; Moore, K.A., M.H. Zaslow, M.J. Coiro, S.M. Miller, and E.B. Magenheim. 1995. The JOBS evaluation: How well are they faring? AFDC Families with Preschool-Aged Children in Atlanta at the Outset of the JOBS Evaluation. Washington, D.C.: U.S. Department of Health and Human Services; Ware, J.E., and D.C. Sherbourne. 1992. The MOS 36-Item Short-Form Health Survey (SF-36). Medical Care 30(6): 473-81; Berwick, D.M., J.M. Murphy, P.A. Goldman, J.E. Ware, A.J. Barsky, and M.C. Weinstein. 1991. Performance of a five-item mental health screening test. Medical Care 29(2): 169-76; and Stewart, A.L., R. Hays, and J.E. Ware. 1988. The MOS Short-form General Health Survey, reliability and validity in a patient population. Medical Care 26(7): 724-35.
The authors wish to thank Brett Brown, Carol Emig, and Kathryn Tout for reviewing this text and Sharon Vandivere for assisting with data production.
Latest Research
Browse by Topic:
Printer-friendly summary
