ACF OPRE: National Study of Child Care of Low-Income Families

Chapter One: The Community Survey

The Community Survey is the second of the three major components of the National Study of Child Care for Low-Income Families. Completed in 2000, it was a one-time random-digit-dialing telephone survey of poor and near-poor families with working parents in each of the 25 study communities. It addressed the following questions:

What types of non-parental care arrangements and how many different child care arrangements do low-income families use?
What are the considerations that influence low-income families’ choice of non-parental child care?
What are low-income families’ perceptions about the availability of different kinds of child care in their community and the choices available to them?
What proportion of family income do low-income families spend on child care?
How does the presence or absence of a child care subsidy affect parents’ child care decisions?

Selection of the 25 Study Communities

For the National Study of Child Care for Low-Income Families, we used the county as our definition of a community, because of the availability of information at the county level from earlier child care studies conducted in a nationally-representative sample of counties.

Because the focus of this study was on low-income families and their child care needs and choices, rather than on families at all income levels, our goal was to select a sample of communities that would, in a broad sense, be representative of where low-income children and families live. Starting with the sample of 100 counties/county groupings used by the 1990 National Child Care Survey,2 we identified 80 with a 1993 poverty rate for children greater than 13.8 percent.3 These formed our sampling frame. When properly weighted, these 80 counties/county groups represent more than 90 percent of poor children in the United States. Our sample of 25 counties, selected randomly with probability of selection proportional to size, is a representative sample of the 80 counties/county groupings. The sample of counties is shown in Exhibit 1.1.

Exhibit 1.1

Selected States and Communities
State	Communities (Counties or County Groupings)
Alabama	Mobile
California	Los Angeles* Orange Riverside
Illinois	Cook
Indiana	Madison
Louisiana	Oachita
Massachusetts	Franklin*
Michigan	Wayne
Minnesota	Hennepin Itasca/Koochiching/Pennington
New Jersey	Union
New Mexico	Dona Ana Luna/Grant/Hidalgo
New York	Orange
North Carolina	Mecklenberg Alamance Johnston
Ohio	Hamilton
Tennessee	Shelby Hardeman/Fayette/Lake Lauderdale Marshall/Coffee/Bedford
Texas	Harris*
Virginia	Arlington
Washington	King*

* included in the in-depth study of family child care.

Selecting Families for the Community Survey

A random-digit-dialing (RDD) telephone survey was conducted in each of the 25 communities to screen, recruit and interview a total of 2,500 families – 100 in each community. In an RDD survey, blocks of telephone numbers for exchanges within the county are randomly selected, screened to exclude identifiable business numbers and then dialed. Blocks of numbers continue to be randomly selected and released until the desired quota for each county is complete. Interviewers dialed each number up to 10 times before the number was assigned “no contact” status. When the call was answered, the interviewer used a Screening Questionnaire to identify households eligible to participate in the survey. For this study, eligible households were defined as:

having children under age 13;
with family income below 200 percent of the Federal Poverty Level (FPL);
with a mother working or in school more than 20 hours a week;4 and
using some form of non-parental child care.

If respondents passed the first three eligibility screens, the interviewer asked about child care arrangements for a randomly selected child (if there was more than one). If the first selected child was in non-parental care, the survey was conducted at that point. Otherwise, the interviewer asked about the remaining children, in random order, until a child was identified as in non-parental care (at which point the survey was conducted) or it was determined that no children were in non-parental care (and the screening interview was terminated).

The survey was conducted over a 12-month period, beginning in August 2000 and ending in July 2001.5

This survey is the most recent in a series of surveys that have investigated parents’ use of child care, among other topics. Where possible, we have compared findings from this survey with the findings for the two most recent surveys on this topic. The first survey, from which child care data were analyzed and reported in “Who’s Minding the Kids?” (Smith, 2000), is the U.S. Census Bureau’s 1993 Survey of Income and Program Participation (SIPP).6 The child care module, conducted in fall 1995 was redesigned to collect information on all types of child care arrangements, not just the primary arrangement of employed mothers (the focus of earlier surveys). The respondents for the SIPP child care module were “designated parents” with children under 15 years of age. The goal of the module was: “to present a comprehensive view of the regular weekly experiences of children under 15 years of age.” As a consequence, child care was very broadly defined to include school, sports, lessons and clubs, as well as the range of settings typically classified as child care. The survey included families at all levels of income; the analyses distinguish between families in poverty and families not in poverty.

The second survey, conducted in 1997, is the National Survey of American Families (NSAF), a household survey conducted by the Urban Institute as part of Assessing the New Federalism,7 a multi-year effort to examine the devolution of social programs from Federal to state and local levels. Among many other topics, the survey investigated all regular child care arrangements for a nationally-representative sample of children under 13 years of age. The survey oversampled households with incomes below 200 percent of poverty and interviewed “the person most knowledgeable about the child.” Analyses of child care are presented in a series of reports that deal with child care for preschool-age children, child care for school-age children, and child care costs.

One frequently encountered problem in comparing survey findings is that questions may be asked differently and results reported in ways that make direct comparisons difficult. We were able to reanalyze NSAF data on low-income families’ use of child care arrangements to make comparisons more meaningful.

Both of these surveys differ from the Community Survey in a number of ways: first, and most importantly, they surveyed a nationally representative sample of families. The Community Survey surveyed low-income families living in communities with child poverty rates of 13.8% or higher that represented similar communities nationally. It was possible with the NSAF to calculate means and frequencies including and excluding counties with low levels of child poverty; comparisons between the full and restricted samples are presented in Appendix C. Second, because the Community Survey was concerned with the non-parental child care arrangements of working parents, it includes a substantial number of questions not included in the other surveys (or not presented in reports on child care use) about the hours and schedule of parental employment, use of child care subsidies, parental child care preferences and considerations, among other topics. Thus, for most of the topics discussed in this report, there are no comparisons available from the two surveys.

The Screening Questionnaire and the Survey

As we designed the Screening Questionnaire, we realized that, because of the number and type of questions that were asked to determine eligibility for the survey, the screening instrument itself could provide valuable information to answer a basic question, namely:

What are patterns of parental employment and child care arrangements in low-income families?

Answers to this question provide a context for the more elaborate discussion of non-parental child care that is the true focus of the survey.

Consequently, this report presents findings from analyses of two samples: a sample of low-income households with children under 13; and a smaller subset of those households with a mother working outside the home or in school and one or more children under 13 in a non-parental child care arrangement.8

The sample sizes that we are using ensure a fair amount of statistical precision. The screener sample of 6,160 low-income families with children under age 13 allows us to estimate a true proportion of 50 percent with a standard error of only 0.6 percentage points.9 The 95 percent confidence interval is +/-1.3 percentage points. In other words, there is only a five percent chance that, if the true proportion is 50 percent, our sample estimate will err by as much as 1.3 percentage points in either direction. The survey sample of 2,710 families allows us to estimate a proportion of 50 percent with a standard error of 1 percentage point, and a 95 percent confidence interval of 2.0 percentage points.

2 The National Child Care Survey was nationally representative study of 4,392 households with one or more children under the age 13 conduced in late 1989 and early 1990. the study consisted of a survey of parents in randomly selected households with children under age 13 (the Parent Study), a survey of individuals who provide child care in their own homes, a survey of child care providers used by the respondents in the Parent Study, a low-income substudy, and a military substudy (Hofferth et al.1991). (back)

3 We excluded counties with relatively small numbers of poor children. The study focuses on the impact of welfare reform and child subsidy policies on poor families and we felt that we would learn a little from the more affluent counties excluded. As documented in Appendix C, supplementary analyses of the National Survey of American Families (NSAF) suggests that the exclusion of counties with low levels of child poverty did not result in markedly different results. (back)

4 We were prepared to include father-only families, but the situation did not arise. (back)

5 Copies of the Screening Questionnaire and Survey instruments can be found in Volume 2 of this report. Appendix A provides information on weighting procedures and response rates. (back)

6 For information on the 1993 SIPP, go to www.bls.census.gov/sipp. (back)

7 For information on Assessing the New Federalism and the National Survey of American Families, go to www.urban.org/center/anf. (back)

8 Some differences between the two samples should be noted. The child considered in the screener analysis is not necessarily the focus child of the survey analysis. The screener analysis child is the first randomly selected child about whom child care information was collected, while the survey focus child was randomly selected from all children in the family that were in non-parental child care. This distinction was made because the screener sample is intended to represent all children in low-income families while the survey sample is restricted to children in non-parental care. (back)

9 Because the sample is nationally representative, it provides an unbiased estimate of the rate in the population from which the sample was drawn (the true proportion). It does not give an exact estimate of this rate, however, because of the random sampling variation. (back)

Table of Contents | Previous | Next