National Evaluation of Welfare-to-Work Strategies: Data

• Overview of NEWWS
• Information About the Data
• How to Access the Data Files

Overview of NEWWS

The Department of Health and Human Services (HHS) undertook a study of the effectiveness of welfare-to-work programs: the National Evaluation of Welfare-to-Work Strategies (NEWWS) is a study of the effectiveness of eleven mandatory welfare-to-work programs in seven locales: Atlanta, Georgia; Columbus, Ohio; Detroit and Grand Rapids, Michigan; Oklahoma City, Oklahoma; Portland, Oregon; and Riverside, California. Program impacts were evaluated by comparing outcomes for a randomly assigned experimental group subject to program requirements with outcomes for control groups. As part of NEWWS, the effects of two approaches to preparing welfare recipients for employment were compared in three sites (Atlanta, Grand Rapids, and Riverside). In one approach, the human capital development approach, individuals were directed to avail themselves of education services and, to a lesser extent, occupational training before they sought work, under the theory that they would then be able to get better jobs and keep them longer. In the other approach, the labor force attachment approach, individuals were encouraged to gain quick entry into the labor market, even at low wages, under the theory that their work habits and skills would improve on the job and they would thereby be able to advance themselves. Data from all eleven sites is available.

The evaluation used a random assignment design to get reliable results. Sample members were followed for five years from the time they entered the study. Comprehensive data on economic outcomes, including information on quarterly Unemployment Insurance-reported earnings and monthly Temporary Assistance for Needy Families (TANF) and Food Stamp payments was collected. A broad range of data was collected through surveys including data on educational attainment, family composition, housing status, wage progression, employment, child care, depression, and total family income. In addition, effects on the well-being of the children of the mothers in the study were evaluated. Four types of child outcomes were measured: cognitive development and academic achievement; safety and health; problem behavior and emotional well-being; and social development. Assessments in each of these areas were compared across research groups two and five years after the mothers entered the survey sample.

In three of the sites (Atlanta, Grand Rapids, and Riverside) two distinct types of welfare reform strategies were compared: a labor force attachment program and a human capital development program. A broad range of data has been collected in these three sites through maternal surveys including data on educational attainment, family composition, housing status, wage progression, employment, child care, maternal depressive symptoms, total family income, children's school readiness, school achievement, and behavioral outcomes. Sample members were followed for five years from the time they entered the study. Two-year and five-year follow-up data are available. The following files are available:

• 2-year full impact sample,
• Two-year client survey,
• Two-year child outcomes study survey,
• Two-year literacy and math test scores,
• Five-year full impact sample,
• Five-year client survey, and
• Five-year child outcomes study and teacher surveys.

Each file contains linked survey and administrative data. In all eleven sites, administrative data are available that include comprehensive, longitudinal data on economic outcomes, including information on quarterly Unemployment Insurance reported earnings and monthly AFDC/TANF and Food Stamp payments.

Several reports will be very helpful in understanding the NEWWS data, research methods, and results.

• In particular, Chapter 2 of the report "Evaluating Two Welfare-to-Work Program Approaches: Two Year Findings on the Labor Force Attachment and Human Capital Development Programs in Three Sites" (prepared by Gayle Hamilton, Thomas Brock, Mary Farrell, Daniel Friedlander, and Kristen Harknett, MDRC, 1997, Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Department of Education) provides detailed information on the research design, samples, and data sources.
• In addition, Chapter 1 and Chapter 2 of "Impacts on Young Children and Their Families Two Years after Enrollment: Findings from the Child Outcomes Study" (prepared by Sharon M. McGroder, Martha J. Zaslow, Kristin A. Moore, and Suzanne M. LeMenestrel, 2000, Washington, DC: U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation and Administration for Children and Families; and U.S. Department of Education) also provide an overview of the project and information on research design, samples, and data sources for the Child Outcomes Study.

In addition to the reports mentioned above, see [N.B. the links below are to the executive summaries of the reports]:

• "Evaluating Alternative Welfare-to-Work Approaches: Two-Year Impacts for Eleven Programs" (prepared by Stephen Freedman, Daniel Friedlander, Gayle Hamilton, JoAnn Rock, Marisa Mitchell, Jodi Nudelman, Amanda Schweder, and Laura Storto, MDRC, 2000, Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Department of Education). This report presents impacts for all eleven programs.
• The full report on the LFA-HCD sites, titled "Evaluation of Two Welfare-to-Work Program Approaches: Two-Year Findings on the Labor Force Attachment and Human Capital Development Programs in Three Sites" (prepared by Gayle Hamilton, Thomas Brock, Mary Farrell, Daniel Friedlander, and Kristen Harknett, MDRC, 1997, Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Department of Education) presents results from Atlanta, Grand Rapids, and Riverside.
• Separate reports are available for Detroit, Oklahoma, and Portland:
  • These are titled "Implementation, Participation Patterns, Costs, and Two-Year Impacts of the Detroit Welfare-to-Work Program" (prepared by Mary Farrell, MDRC, 2000, Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Department of Education);
  • "Oklahoma City's ET&E Program: Two-Year Implementation, Participation, Cost, and Impact Findings, and Implementation, Participation Patterns, Costs" (prepared by Laura Storto, Gayle Hamilton, Christine Schwartz, and Susan Scrivener, MDRC, 2000, Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Department of Education); and
  • "Two-Year Impacts of the Portland (Oregon) Welfare-to-Work Program" (prepared by Susan Scrivener, Gayle Hamilton, Mary Farrell, Stephen Freedman, Daniel Friedlander, Marisa Mitchell, Jodi Nudelman, and Christine Schwartz, MDRC, 1998, Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Department of Education).

Executive summaries of these, and many other, reports are available on the NEWWS web site.

Information About the Data

ASPE and ACF are making data from NEWWS available to researchers. Data for the two-year impact reports, as well as the five-year final impact report, are available. Data are available at the National Center for Health Statistics (NCHS) Research Data Center (RDC). ASPE has established an interagency agreement with NCHS to make the data files available at the NCHS Research Data Center. The files are organized around the reports from NEWWS: two-year impact reports, the adult basic education report, and the final report.

To protect confidentiality, data that could be used to identify individual sample members were deleted or grouped into broader categories. These changes (which are summarized below) primarily affect sample members' background characteristics data and outcomes measured from administrative records, although some survey responses were also changed. Entries in the codebooks indicate when measures have been deleted or modified.

Background Characteristics: The data files include the year but not the specific date in which a sample member was randomly assigned; and the sample member's "site" (a county or a group of counties) but not the specific location of her welfare-to-work office. Other measures grouped into larger categories, include: sample member's educational attainment, and her scores on literacy and math tests administered at random assignment.

Administrative Records: Welfare and food stamp payment data originally collected by month are summed into quarterly totals. Measures of total earnings, welfare payments, and food stamp payments have been rounded to the nearest $100. Relative dates of earnings and public assistance payments are available (Quarter 1 = quarter of random assignment), but not actual dates.

Survey Responses: Relative dates of survey outcomes are available (Month 1 = month of random assignment). Dollar amounts of reported monthly income have been rounded to the nearest $50.

Two-Year Impacts Report File

Full Impact Sample – these files are available in the zipped archive [259KB]

• Data codebook with information on each variable. This file contains the data codebook, a comprehensive description of the raw data. For each variable on the data file the codebook provides the following information: The name of the variable and its starting and ending columns on the ASCII file, the variable's SAS informat (5.0 is the default), the variable label, a brief description of the variable and any issues with the variable. (Note: The data are stored in ASCII, so informats can be changed to conform to any statistical or econometric software package.) The codebook also displays the N, NMISS, MEAN, Standard Deviation, Minimum, and Maximum for each variable. For variables with 51 values or less, a frequency distribution is listed. The frequencies show the value labels. These may be copied and, if necessary, modified to be read into SAS or any other statistical or econometric software package.
• SAS variable label statements for each variable of the data file. Labels can be modified to conform to syntax rules of any statistical or econometric software package.
• Table of sample sizes. Displays sample sizes for full impact sample, 2-year Client Survey respondents, 2-year Child Outcomes Study survey respondents, Private Opinion Survey respondents, and sample members who completed literacy and math tests at baseline. The table also shows random assignment dates.
• Annotated impact tables of key outcomes calculated with administrative data. Variable names are listed on the right side of the tables.
  • Table 1, Program Impacts on Employment and Earnings in Years 1 and 2
  • Table 2, Program Impacts on Employment Stability and Earning $10,000 or More in Year 2
  • Table 3, Program Impacts on AFDC Receipt and Payments
  • Table 4, Two-Year Impacts on Employment, Earnings, AFDC/TANF, Food Stamps, and Combined Income for AFDC-FGs
• Memos:
  • Background information on the research design for NEWWS, the research groups in each site, and the random assignment design.
  • Background information on administrative records source data and additional outcomes measures created from administrative data.
  • Background information on baseline test scores.
  • Background information on estimating program impacts.
• Selected Characteristics of Sample
• These files are available in the zipped archive [259KB]

2-Year Client Survey Files – these files are available in the zipped archive [250KB]

• General introduction to the data and documentation of the NEWWS Files (6-CD set) and offers suggestions on how to use the files.
• Data File. This is the ASCII data file with responses to sections of the survey administered to all respondents (N = 9,675). It includes questions on child care for employment and a limited number of child outcomes. (More detailed information on these subjects can be found in the Child Outcomes Study Survey file.) The file also contains outcome measures created from survey responses and information on administration of the survey: interview date, mode (phone/in-person), and length. Each sample member has 18 records of data. The file is sorted by each sample member's randomly generated IDNUMBER and record number.

IMPORTANT !!!:

1. Survey responses must be weighted to make the findings generalizable to the Full Impact Sample. The variable that weights the responses is called FIELDWGT and is included on the file. See S2SAMPLE and SIMPMEMO for details.
2. Respondents' background characteristics and administrative records data are stored on the Full Sample Impact file.
3. Read the 5-year data file, select SRV2RESP=1 (2-Year Survey respondent), and merge by IDNUMBER to get add these records to the survey file.

• Data codebook with information on each variable. This file contains the data codebook, a comprehensive description of the raw data. For each variable on the data file the codebook provides the following information: The name of the variable and its starting and ending columns on the ASCII file, the variable's SAS informat (5.0 is the default), the variable label, a brief description of the variable and any issues with the variable. (Note: The data are stored in ASCII, so informats can be changed to conform to any statistical or econometric software package.) The codebook also displays the N, NMISS, MEAN, Standard Deviation, Minimum, and Maximum for each variable. For variables with 51 values or less, a frequency distribution is listed. The frequencies show the value labels. These may be copied and, if necessary, modified to be read into SAS or any other statistical or econometric software package.
• The SAS input statement (variable name, starting column, length, informat) that will read the data file. To run this input statement in SAS you must add the word INPUT to the top line and a semicolon to the bottom line. This statement can be modified to input the data into any statistical or econometric software package.
• SAS variable label statements for each variable on the data file. Labels can be modified to conform to syntax rules of any statistical or econometric software package.
• Descriptive statistics for all variables.
• Table of sample sizes. Displays sample sizes for full impact sample, 2-year Client Survey respondents, 2-year Child Outcomes Study survey respondents, Private Opinion Survey respondents, and sample members who completed literacy and math tests at baseline. The table also shows random assignment dates.
• These annotated tables display key outcomes of use and cost of child care for employment.
  • Table 1
  • Table 2
  • Table 3
  • Table 4
  • Table 5
  • Table 6
• Annotated table displays child outcomes.
• Annotated table displays key outcomes of degree receipt.
• Annotated table displays key outcomes of household composition.
• These annotated tables display key outcomes of medical coverage and non-cash benefits:
  • Table 1
  • Table 2
  • Table 3
  • Table 4
• These annotated tables display key outcomes of participation in employment-related services, attitudes toward work and welfare, and sanctions:
  • Table 1
  • Table 2
  • Table 3
• These annotated tables display key outcomes of employment:
  • Table 1
  • Table 2
  • Table 3
  • Table 4
• These annotated tables display key outcomes of respondent and household income:
  • Table 1
  • Table 2
  • Table 3
• Memos:
  • Background information on sample selection, response rates, and weighting.
  • Background information on how outcome measures were created from survey data.
  • Background information on estimating program impacts with survey data.
• These files are available in the zipped archive [250KB]

2-Year Child Outcomes Study (COS) Files – these files are available in the zipped archive [151KB]

• Guide to using the files.
• Data file. This is the ASCII data file with responses to sections of the survey administered to COS respondents (N = 3,018). The file includes both the original survey items from the COS modules of the Two-Year Client Survey and created variables for Child Outcomes Study: covariates used in impact calculations, subgroups, child outcomes, targeted intervening outcomes, and non-targeted intervening outcomes. Each sample member has 14 records of data. The file is sorted by each sample member's randomly generated IDNUMBER and by record number.

IMPORTANT !!!

1. Survey responses must be weighted to make the findings generalizable to the Full Impact sample. The variable that weights the responses is called FIELDWGT and is included in the Two-Year Client Survey file. See C2COVER and SIMPMEMO for details.
2. Respondents' additional background characteristics and administrative records data are stored on the Full Sample Impact file. Responses to additional survey questions are stored on the 2-Year Client Survey file. These data should be merged (by IDNUMBER) for the most accurate and complete analysis of program effects.

• Data Codebook. This file contains the data codebook, a comprehensive description of the data. For each variable on the data file, the codebook provides the following information: The name of the variable and its starting and ending columns on the ASCII file, the variable's SAS informat (5.0 is the default), the variable label, a brief description of the variable and any issues with the variable. (Note: The data are stored in ASCII, so informats can be changed to conform to any statistical or econometric software package.) The codebook also displays the N, NMISS, MEAN, Standard Deviation, Minimum, and Maximum for each variable. For variables with 51 values or less, a frequency distribution is listed. The frequencies show the value labels. These may be copied and, if necessary, modified to be read into SAS or any other statistical or econometric software package.
• SAS Input Statement. This file contains the body of a SAS input statement (variable name, starting column, length, informat) that will read the data file. To run this input statement in SAS you must add the word INPUT to the top line and a semicolon to the bottom line. This statement can be modified to input the data into any statistical or econometric software package.
• Descriptive statistics on all variables.
• SAS variable label statements for each variable of the data file. Labels can be modified to conform to syntax rules of any statistical or econometric software package.
• Memos:
  • Additional background information on survey responses and outcome measures created from these data.
  • Contains SAS code that shows how covariate and subgroup measures were created.
• Output:
  • Shows control group means on the Bracken raw score and the maternal reports of the child being in very good or excellent health within each site and comparing across sites. Shows data in Table 5.1 of the COS 2-year impact report.
  • Shows control group means on employment variables within each site and comparing across sites. Shows data in Table 8.1 of the COS 2-year impact report.
  • Shows impacts on the Bracken raw score. Shows data in Table 6.2 of the COS 2-year impact report.
  • Shows impacts on the whether the focal child was in very good or excellent health. Shows data in Table 6.2 of the COS 2-year impact report.
  • Shows impacts on the mean Bracken score for the subgroups based on education/literacy risks. Shows data in the COS 2-year impact report's Appendix Table 7.1.
  • Shows impacts on the measure of whether the focal child was in very good or excellent health, for the subgroups based on the total number of composite risks. Shows data in the COS 2-year impact report's Appendix Table 7.11.
• Sample sizes for full impact sample, Two-Year Client Survey respondents, 2-year Child Outcomes Study survey respondents, Private Opinion Survey respondents, and sample members who completed literacy and math tests at baseline. The table also shows random assignment dates.
• These files are available in the zipped archive [151KB]

2-Year Client Survey Files – these files are available in the zipped archive [422KB]

• General introduction to the data and documentation of the NEWWS Files with suggestions on how to use the files.
• Data codebook with information on each variable. This file contains the data codebook, a comprehensive description of the raw data. For each variable on the data file the codebook provides the following information: The name of the variable and its starting and ending columns on the ASCII file, the variable's SAS informat (5.0 is the default), the variable label, a brief description of the variable and any issues with the variable. (Note: The data are stored in ASCII, so informats can be changed to conform to any statistical or econometric software package.) The codebook also displays the N, NMISS, MEAN, Standard Deviation, Minimum, and Maximum for each variable. For variables with 51 values or less, a frequency distribution is listed. The frequencies show the value labels. These may be copied and, if necessary, modified to be read into SAS or any other statistical or econometric software package.
• Data file. The ASCII data file with responses to sections of the survey administered to all respondents (N = 9,675) is available through the (NCHS) Research Data Center (RDC). The data file includes questions on child care for employment and a limited number of child outcomes. (More detailed information on a subset of 3,018 of these subjects can be found in the Child Outcomes Study Survey file.) The file also contains outcome measures created from survey responses and information on administration of the survey: interview date, mode (phone/in-person), and length. Each sample member has 18 records of data. The file is sorted by each sample member's randomly generated IDNUMBER and record number.

IMPORTANT !!!:

1. Survey responses must be weighted to make the findings generalizable to the Full Impact Sample. The variable that weights the responses is called FIELDWGT and is included on the file. See N2PSSAMP and N2PS_IMP for details.
2. Respondents' background characteristics and administrative records data are stored on the Five-Year Full Sample Impact file.
3. Read the data file, N5PI1809.TXT, select SRV2RESP=1 (2-Year Survey respondent), and merge by IDNUMBER to add these records to the survey file.

• The SAS input statement (variable name, starting column, length, informat) that will read the data file. To run this input statement in SAS you must add the word INPUT to the top line and a semicolon to the bottom line. This statement can be modified to input the data into any statistical or econometric software package.
• SAS variable label statements for each variable on the data file. Labels can be modified to conform to syntax rules of any statistical or econometric software package.
• Descriptive statistics for all variables.
• Table of sample sizes. Displays sample sizes for full impact sample, 2-year Client Survey respondents, 2-year Child Outcomes Study survey respondents, Private Opinion Survey respondents, and sample members who completed literacy and math tests at baseline. The table also shows random assignment dates.
• These annotated tables display key outcomes of use and cost of child care for employment.
  • Table 1
  • Table 2
  • Table 3
  • Table 4
  • Table 5
  • Table 6
• Annoted table displays child outcomes.
• Annotated table displays key outcomes of degree receipt.
• Annotated table displays key outcomes of household composition.
• These annotated tables display key outcomes of medical coverage and non-cash benefits:
  • Table 1
  • Table 2
  • Table 3
  • Table 4
• These annotated tables display key outcomes of participation in employment-related services, attitudes toward work and welfare, and sanctions:
  • Table 1
  • Table 2
  • Table 3
• These annotated tables display key outcomes of employment:
  • Table 1
  • Table 2
  • Table 3
  • Table 4
• These annotated tables display key outcomes of respondent and household income:
  • Table 1
  • Table 2
  • Table 3
• Memos:
  • Background information on sample selection, response rates, and weighting.
  • Background information on how outcome measures were created from survey data.
  • Background information on estimating program impacts with survey data.
• These files are available in the zipped archive [422 KB]

2-Year Child Outcomes (COS) Files – these files are available in the zipped archive [171KB]

• Contents of the COS files and documentation and important information about calculating impacts with the data set.
• Data file. The ASCII data file with responses to sections of the survey administered to COS respondents has N = 3,018 is available through the (NCHS) Research Data Center (RDC). The data file includes both the original survey items from the COS modules of the Two-Year Client Survey and created variables for Child Outcomes Study: covariates used in impact calculations, subgroups, child outcomes, targeted intervening outcomes, and non-targeted intervening outcomes. Each sample member has 14 records of data. The file is sorted by each sample member's randomly generated IDNUMBER and by record number.

IMPORTANT !!!:

1. Survey responses must be weighted to make the findings generalizable to the Full Impact sample. The variable that weights the responses is called FIELDWGT and is included in the Two-Year Client Survey data file. See C2PCOVER and N2PS_IMP for details.
2. Respondents' additional background characteristics and administrative records data are stored on the Full Impact Sample data file.
3. Most outcomes for COS respondents and a few additional outcomes for focal children are stored on the 2-Year Client Survey data file . These data should be merged (by IDNUMBER) with outcomes from the COS file for the most accurate and complete analysis of program effects.
4. Scores on literacy tests administered to a subset of all COS respondents during the 2-year interview are stored on the Two-Year Literacy and Math Test Scores file. Respondents who were also in the COS did not take the math test.

• Data codebook for each variable on the data file. For each variable on the data file, the codebook provides the following information: The name of the variable and its starting and ending columns on the ASCII file, the variable's SAS informat (5.0 is the default), the variable label, a brief description of the variable and any issues with the variable. (Note: The data are stored in ASCII, so informats can be changed to conform to any statistical or econometric software package.) The codebook also displays the N, NMISS, MEAN, Standard Deviation, Minimum, and Maximum for each variable. For variables with 51 values or less, a frequency distribution is listed. The frequencies show the value labels. These may be copied and, if necessary, modified to be read into SAS or any other statistical or econometric software package.
• SAS input statement. This file contains the SAS input statement (variable name, starting column, length, informat) which will read the data file. To run this input statement in SAS you must add the word INPUT to the top line and a semicolon to the bottom line. This statement can be modified to input the data into any statistical or econometric software package.
• Descriptive statistics for all variables.
• SAS variable labels for each variable of the data file. Labels can be modified to conform to syntax rules of any statistical or econometric software package.
• Memos:
  • Additional background information on survey responses and outcome measures created from these data.
  • The SAS code that shows how covariate and subgroup measures were created.
• Output:
  • Shows control group means on the Bracken raw score and the maternal reports of the child being in very good or excellent health within each site and comparing across sites. Data from Table 5.1 of the COS 2-year impact report.
  • Shows control group means on employment variables within each site and comparing across sites. Data from Table 8.1 of the COS 2-year impact report.
  • Shows impacts on the Bracken raw score. Data from Table 6.2 of the COS 2-year impact report.
  • Shows impacts on the whether the focal child was in very good or excellent health. Data from Table 6.2 of the COS 2-year impact report.
  • Shows impacts on the mean Bracken score for the subgroups based on education/literacy risks. Data from the COS 2-year impact report's Appendix Table 7.1.
  • Shows impacts on the measure of whether the focal child was in very good or excellent health, for the subgroups based on the total number of composite risks. Data from the COS 2-year impact report's Appendix Table 7.11.
• Tables:
  • Displays sample sizes for full impact sample, Two-Year Client Survey respondents, 2-year Child Outcomes Study survey respondents, Private Opinion Survey respondents, and sample members who completed literacy and math tests at baseline. The table also shows random assignment dates.
• These files are available in the zipped archive [171 KB]

5-Year, Public Use Files

5-Year Full Impact Files

The file F5README.TXT contains a description of each file that was. Download zipped archive file, p5fullsamp.zip [256 KB]

5-Year Client Survey

The file S5PREADME.TXT contains a description of each file that was created. Download zipped archive file, survey5p.zip [190 KB]

5-Year Child Outcomes Study (COS)

The file C5PREADME.TXT contains a description of each file that was created. Additional files are included with updated information: C5COVER.TXT, C5PCOVER.TXT, C5README.TXT, and Chage169.txt. Download zipped archive file, COS5p.zip [185 KB]

How to Access the Data Files

ASPE has entered into an interagency agreement with the National Center for Health Statistics (NCHS) to facilitate placement of NEWWS files at the NCHS Research Data Center (RDC). To gain access to the data, a researcher will have to apply by submitting a structured research proposal, sign an agreement not to disclose individual identities, and come to a Research Data Center in person. Applications for use of data will be reviewed and research use of the data will be monitored to prevent breaches in confidentiality and inappropriate disclosure of data. For example, the RDC does not allow researchers to print out individual observations or cells with fewer than five observations. To support the ongoing maintenance of the restricted access files at the Research Data Center, NCHS charges researcher access fees.

If you have additional questions, please contact Laura Chadwick at laura.chadwick@hhs.gov.