BSF - Supporting Statement Part A

CONTENTS (continued)

Page

Building Strong Families (BSF) Demonstration and Evaluation: Impact Study Second Follow-up

OMB Supporting Statement Part A

March 2009

1. JUSTIFICATION

The Administration for Children and Families (ACF) of the U.S. Department of Health and Human Services (DHHS) is undertaking the Building Strong Families (BSF) project. It is requesting an additional six months of clearance beyond the current expiration date of July 31, 2009 for the second of the following two data collection efforts:

• Data collection for the implementation analysis including: a guide for site visit interviews with program staff (Appendix A); a guide for focus groups with program participants (Appendix B); and a guide for brief phone calls with people who drop out of the program (Appendix C).

• A telephone survey of mothers and fathers (Appendix D, with supporting documentation in Appendix E).

With the exception of updates related to this continuation request, the remainder of this document is exactly the same as that submitted in support of the information collection that was approved on July 25, 2006 (OMB No. 0970-0304).

A1. Circumstances Necessitating the Data Collection

The goal of the BSF project is to learn whether well-designed interventions can help interested and romantically involved unwed parents build stronger relationships and fulfill their aspirations for a healthy marriage if they so choose. The BSF programs target parents before, or around the time of, their child’s birth, and provide instruction and support to help couples develop the relationship skills that research has shown are associated with healthy marriages. Ultimately, healthy marriage between biological parents is expected to enhance child well-being. ACF has contracted with Mathematica Policy Research, Inc. (MPR) and its subcontractors¹ to support the development of such interventions and to determine their effectiveness.

a. Background on the Building Strong Families Project

The BSF project originated from three bodies of research. The first body of research shows that, on average, children who grow up with their two married biological parents do better than those growing up in single parent households on a wide range of outcomes, including the likelihood of growing up in poverty, academic and behavioral outcomes, and the likelihood of the children themselves becoming single parents (McLanahan and Sandefur 1994).

The second body of research comes from the landmark study, The Fragile Families and Child Well-being Study (http://crcw.princeton.edu/fragilefamilies). The study follows about 3,700 unmarried couples who were recruited into the study shortly after the birth of their child between 1998 and 2000 in 20 large cities throughout the United States. Results from this study show that most unwed parents are romantically involved around the time their child is born, and anticipate marrying each other. Most unwed parents agree that it is better for children if their parents are married. Nevertheless, the study shows that only a small fraction of such couples are married a year later (McLanahan et al. 2001; Carlson 2002). The positive findings about the couples’ relationships and their aspirations for marriage suggest that there may be an opportunity for intervention around the time of the child’s birth.

The third body of research is on the effectiveness of programs that provide relationship education to married and engaged couples. A variety of programs prepare couples for marriage, strengthen the relationship of married couples, or prepare people for the stresses on their relationships when they become parents. Studies have shown these programs to be effective in improving couples’ marriages and reducing divorce rates (Markman et al. 1988; Markman et al. 1993; and Cowan and Cowan 2000).

There remain substantial gaps, however, in our understanding of how to strengthen the relationships of unwed parents and how to support those who choose to marry. Research on marriage and relationship skills programs has generally focused on their applications to married or engaged couples, and primarily middle‑class couples, rather than the low‑ income unmarried couples that constitute the BSF target population. Moreover, much of the research undertaken in these programs has been constrained by the problems of small sample size and sample attrition.

b. Overview of the Building Strong Families Program

To address this knowledge gap, the BSF project will demonstrate and evaluate interventions with unwed parents, starting around the time of their child’s birth. Building on this research, the first component of the BSF project was to develop a program model. Based on a conceptual framework for whether and how to intervene with unmarried and romantically involved parents having a baby, we developed detailed program guidelines for BSF programs (available at the BSF website, www.buildingstrongfamilies.info).

As described in the program guidelines, BSF programs have three components:

1. Healthy Marriage and Relationship Skills Education. The core, distinctive component of BSF programs is the provision of information to enhance couples’ understanding of marriage and the instruction in the relationship skills found in research to be essential to a healthy marriage. This instruction is provided in (usually weekly) group sessions with BSF couples.

2. Family Support Services. These include services to address special issues that may be common among low-income parents and that are known to affect couple relationships and marriage. They might, for example, help to improve parenting skills or provide linkages to address problems with employment, physical and mental health, or substance abuse.

3. Family Coordinators. These program staff assess couples’ circumstances and needs, make referrals to other services when appropriate, reinforce relationship and marriage skills over time, provide ongoing emotional support, and promote sustained participation in program activities.

BSF is intensive. The core component of BSF—the group instruction related to relationship skills and healthy marriage—requires up to 44 hours. It is typically provided over a sustained period of time, as long as five or six months. Program sites differ in the frequency and duration of time that couples meet with the family coordinators, but it may be as long as three years.

Couples are recruited for BSF either during pregnancy or shortly after the birth of their baby (up until the child reaches three months of age). To be eligible for BSF, a mother and father must be:

• Expectant biological parents or the biological parents of a baby three months of age or younger

• Age 18 or older

• Unmarried (or married since conception of the baby)

• In a romantic relationship

• Not involved in domestic violence that could be aggravated by participation in BSF (the BSF programs, working with local domestic violence experts and with input from MPR and ACF, will use program‑specific screening approaches)

• Available to participate in BSF and be able to speak and understand a language in which BSF is offered.

c. The BSF Program Sites

The BSF sites were selected through a process that involved both technical assistance and scrutiny of their implementation progress and capacity. The BSF project team provided information and technical assistance to a number of organizations and agencies interested in implementing the BSF model. The field was narrowed to seven sites that seemed the most promising, and those sites developed detailed plans for implementation. An intensive program design period helped these sites systematically consider and plan for such operational needs as recruitment sources, staffing structure, domestic violence screening, a management information system (MIS), and curriculum selection and training.

As each site completed its program planning, it moved into piloting the program. This pilot phase lasted between three and nine months, depending on the site. Throughout the pilot phase, each site’s operational progress was closely and regularly monitored by the research team, who also continued to provide assistance as needed.

At the end of the pilot phase, MPR assessed whether the site should be part of the evaluation. To be included in the evaluation, a site needed to meet three main criteria: (1) effective implementation of the BSF program in a way that was faithful to the program model; (2) demonstrated ability to recruit enough couples during the sample intake period to meet the sample size targets; and (3) ability to comply with the requirements of the evaluation, including administering the consent and baseline information forms. All seven sites in the pilot met the criteria and were chosen to be in the evaluation.

The BSF sites include: Atlanta, Georgia; Baton Rouge, Louisiana; Baltimore, Maryland; Orange and Broward counties, Florida; Marion, Allen, Miami, and Lake counties, Indiana; Oklahoma City, Oklahoma; and San Angelo and Houston, Texas. Five sites are located in urban areas. The San Angelo site is in a small city with a surrounding rural catchment area; and Miami County, Indiana is largely rural. The sites vary in a number of aspects, particularly the infrastructure in which BSF was implemented, the recruitment and referral sources, characteristics of the population served, and the chosen curriculum. Three of the sites built upon their Healthy Families programs, a nationally known intervention for preventing child abuse and neglect through intensive home visiting. The sites differed in terms of the host organization, the primary recruitment source, the race/ethnicity of the population served, and whether the couples were served prenatally, postnatally, or both (Table 1).

Table 1. Key Features of BSF Sites

Pilot Site	Host Organization	Primary Recruitment Sources	Predominant Race/Ethnicity Served	Timing of Recruitment
Atlanta, Georgia	Georgia State University, Latin American Association	Public health clinics	African American and Hispanic	Prenatal
Baltimore, Maryland	Center for Fathers, Families and Workforce Development	Local hospitals, prenatal clinics	African American	Prenatal and postnatal
Baton Rouge, Louisiana	Family Road of Greater Baton Rouge	Prenatal program for low-income women	African American	Prenatal
Florida: Orange and Broward counties	Healthy Families Florida	Birthing hospitals	African American and Hispanic	Postnatal
Indiana: Allen, Marion, Miami, and Lake counties	Healthy Families Indiana	Hospitals, prenatal clinics, WIC	African American, White	Prenatal and postnatal
Oklahoma City, Oklahoma	Public Strategies, Inc.	Hospitals, health care clinics, direct marketing	White	Prenatal
Texas: San Angelo and Houston	Healthy Families San Angelo and Houston	Hospitals, public health clinics	Hispanic and White	Prenatal and postnatal

d. Objectives and Overview of the BSF Evaluation

The goal of the BSF evaluation is to determine whether programs can help unwed parents develop stronger relationships and healthy marriages and thus enhance the well‑being of their children. To meet this goal, the evaluation has two main components: (1) documenting and analyzing program implementation; and (2) estimating the impacts of BSF on the lives of the parents and their children.

Documenting Program Implementation. An implementation analysis will examine the development and implementation of BSF programs in local sites. Documentation and analysis of program implementation within and across sites will help us understand what intervention led to estimated impacts and will inform policymakers and program sponsors on strategies for strengthening existing and future programs.

Broadly, we are interested in learning what happened during implementation, why and how it happened, and what can be learned for the future. The following areas will guide the data collection and analysis: program context, outreach and recruitment, operations, and participation. The research questions addressed by the implementation analysis include:

• What is the context in which programs are implemented? The overall purpose of gathering information on the program context of local sites is to develop a clear understanding of who, why, and what led local entities to implement the BSF program model. For instance, what were the circumstances that led to the development of a BSF program? What parties were involved in the planning process, and what resources were needed to carry out the planning and initial implementation? We also will examine the environment and setup of sites, such as whether BSF programs were built onto the infrastructure of pre‑existing programs, and if other marriage and family support services are available in the community.

• How are participants identified as eligible for BSF and then enrolled in the program? We will describe each site’s plan for outreach and recruitment and document any modifications. We will focus especially on what strategies are used to recruit fathers into the program, who may be more difficult to engage in relationship skills programs than mothers.

• How is BSF implemented? The study will detail the implementation of BSF at each site, using the core components of the BSF program (marriage skills instruction, family support services, and family coordinators) as a framework. The focus of this area is to determine how sites operationalize the BSF model guidelines in their communities, concentrating on initial implementation and daily operations of each component, such as training, staffing, and monitoring.

• To what extent do enrolled couples attend and complete BSF? This includes participation not only in the relationship skills groups, but also in recommended family support services, meeting with family coordinators, or other supplementary activities. We will explore the interest levels of couples assigned to the program group, their reasons for attending the groups, missing sessions, or ending their participation, and include an examination of how this might differ by such characteristics as gender and race/ethnicity.

The data needed to address these topic areas will be collected from five major sources. First, much of the information on program context will be gathered from interviews with program staff, such as program managers, intake workers, group facilitators, and staff members affiliated with referral sources for the sites. Second, to obtain data that are not filtered through program staff, we will conduct observations of program activities. Activities that are likely to be observed include BSF group sessions, intake assessments, and interactions between the family coordinators and the couples. Third, to acquire feedback from couples involved with BSF, we will conduct focus groups with those who have actively participated in the curriculum group sessions. Fourth, to learn about why some couples do not attend the group sessions, we will have brief phone interviews with those who dropped out of the program or never participated in groups. Last, we will use information from each site’s MIS, including quantitative data on such areas as group attendance and demographics of involved couples.

Estimating Program Impacts. The BSF impact analysis uses a rigorous experimental design with longitudinal data collection. In each of the BSF programs, couples are randomly assigned to either a program group that receives the BSF intervention or a control group that does not. The control group is eligible to receive other services available in the community.

When a couple is found to be eligible for BSF, the program services are explained to them. If they are interested in participating in the program, they would complete a consent form and a baseline information form (OMB clearance number 0970‑0273, expiration 03‑31‑2008). After the baseline forms have been completed, couples are randomly assigned to either the program (the intervention) or the control group. Couples assigned to the program group are offered BSF services; couples assigned to the control group do not receive BSF services.

We will conduct two follow-up surveys with both groups. The first follow-up survey is planned to occur 15 months after random assignment. Depending on when the family was recruited, the child the couple was expecting or had just been born when they were recruited (the BSF focal child) will be between 9 months of age (if the couple was recruited near the end of the first trimester of pregnancy) and 18 months of age (if the couple was recruited when the child was three months old) at the time of the first follow-up. The second follow-up survey is tied to the age of the BSF child rather than to the time of random assignment, and will occur when the child is about three years of age. Hence, this second survey may occur anywhere from 33 to 42 months after random assignment. At the time of the second survey, we will also conduct an in-home, direct assessment. These direct assessments will focus on child outcomes.

The impact analysis will address the following main questions:

• Does BSF change family outcomes? What is the impact of BSF programs on a wide range of family outcomes, including marital and relationship status, couple relationship quality, parenting^, household structure, family self-sufficiency, parent well-being, child social and emotional functioning, and child language development?

• Do BSF programs work better for some families than for others? Identifying the couples and families who benefit most will help programs improve and target services. We will examine whether program impacts vary by such factors as the demographic characteristics of couples (age, race, or ethnicity, for example), relationship quality at baseline, whether parents have children by other partners, or the “marriageability” of the parents (such as whether they are employed).

• What types of BSF programs work best? We will examine whether different program models have different impacts and whether program effectiveness depends on how they are implemented.

• How do the BSF programs work? If we find impacts of BSF on family outcomes, it will be important to identify the pathways by which BSF affects outcomes in estimating and interpreting the findings—for example, can changes in child outcomes be attributed to increases in marriage, improved relationship skills, better co-parenting, or other intermediate outcomes affected by BSF?

The outcomes that may be affected by BSF and the way they are expected to affect the couples and their child are illustrated in the conceptual framework for BSF (Figure 1). Outcomes of interest fall into four main categories:

• Services Received. While not common, other marriage and relationship skills education programs are available, as is marriage and relationship counseling. If BSF works, we would expect the members of the program group to receive more marriage and relationship education than members of the control group. Relative to control

group members, we also expect BSF program group members to receive more home visits (in sites where BSF involves home visits) and other support services, such as employment and education.

• Parents’ Relationship. The status and quality of the BSF parents’ relationship are key outcomes for evaluation. Relationship status includes whether the parents marry, remain in a romantic relationship, and whether they cohabit. The quality of the relationship includes happiness with the relationship, conflict management, friendship, supportiveness and intimacy, commitment and trust, fidelity, and domestic violence. It also includes how well the parents work together to “coparent” their children. Some BSF couples will inevitably have split up by 15 months after entering BSF and may form new relationships. We will ask about the status and (at the 36-month follow-up) the quality of those relationships.

• Family. BSF may affect many aspects of the family. In particular, it may affect parenting behaviors and father involvement (such as the quantity and quality of time spent with the child), the structure of the family (such as whether the child lives with his or her mother, father, or both), the self-sufficiency of the family (such as employment and income), and parent well-being (such as mental health and substance use).

• Child Well-Being. Important child outcomes include the economic resources available to the child, and his or her socio-emotional and language development. While most child well-being outcomes will be collected at the 36-month follow-up, the 15-month survey will collect information on the economic resources available to the child, including whether the child lives in poverty, and whether the child has health insurance.

A list of the individual-level outcomes and the corresponding data sources is provided in Table 2.

Table 2. Outcomes and their Data Sources

	Data Source
Outcome	15-Month Survey	36-Month Surveya	36-Month Direct Assessmenta
Services Received
Marriage and Relationship Skills Education Whether attended groups, workshops, or classes Number of hours spent in groups, workshops, or classes Whether usually participated with BSF partner Whether attended one-on-one or one-on-two sessions Number of hours spent in these sessions Whether usually participated with BSF partner	X
Home Visits and Other Support Services Whether received home visits, and number of visits Whether received parenting education Whether participated in job training Whether received employment-related services Whether participated in an education program Whether received mental health or substance abuse services Whether participated in anger management or domestic violence programs	X
Parents’ Relationship
Marital/Relationship Status Marital status of BSF parents at follow up (married, separated, divorced, never married) Whether still romantically involved Whether cohabiting Frequency of contact If applicable, when the relationship ended and the reason relationship ended If not married, whether engaged and have plans to marry Chances of marrying the BSF partner in the future Attitudes toward marriage	X	X
BSF Parent’s Relationship Quality Overall happiness with relationship Conflict management Interactions, communication, and time spent together Emotional and sexual intimacy and supportiveness Commitment and trust Fidelity Domestic violence from BSF partner and other partners	X	X
Coparenting Communication and problem solving between parents Trust in other parent’s parenting skills and judgment Work as a team for the child Trust in commitment of other parent to the child Satisfaction with responsibility (including financial) taken by other parent Recognition of the importance of the other parent in the child’s life	X	X
Relationship With New Partner Number of sexual relationships since random assignment Whether currently in a new romantic relationship Whether married to new partner Whether cohabiting with new partner Number and length of marriages since baseline Quality of relationship with new partner (36-month only)	X	X
Family
Parenting/Family Involvement Quantity and Quality of Time Spent with Child Whether father has had contact with child in past year Amount of time BSF parent spends with child Frequency of activities conducted with child (e.g., play games, change diapers) Observation of parenting behaviors at 36 months (yet to be specified) Material Support Whether paternity has been established Whether establishment was voluntary Child support (whether legal order, amount of order, amount paid) Informal child support (amount of cash and in-kind) Contribution of each parent to cost of raising child Stress in the Parenting Role Whether father and/or mother feel stress in their role as parents Discipline Whether either BSF parent spanks the child and frequency Whether new partner spanks the child and frequency	X	X	X
Family Structure BSF Child’s Living Arrangements Whether the child lives with mother, father, both parents, or someone else Number of months child lived with each BSF parent since baseline Number of months child lived with both BSF parents together since baseline Fertility Decisions Number of children born or conceived since BSF focal child Number of children born or conceived with BSF parent Household Structure Number of children who live with BSF parent Number of children who live with, and are the responsibility of, the BSF parent Number of adults in the household Number of persons in the household (36 months only) Relation of the adults in the household to focal child (36 months only)	X	X
Family Self-Sufficiency Employment and Earnings Whether currently working Number of months worked in the past year Hours worked per week in past month Earnings in past month/last month worked Public Assistance Amount of TANF received in previous month Amount of food stamps received in previous month Amount of SSI or SSDI received in previous month Amount of Unemployment Insurance received in previous month Family Income Own earnings Earnings from spouse or cohabiting partner Amount of child support received Amount of money received from friends and relatives Extent to which earnings from spouse/cohabiting partner is available to child Material Hardship Whether during the past year was unable to pay rent, mortgage, or utility bills Whether during the past year was evicted from residence Whether during the past year had their electricity or water service cut off Asset Accumulation Whether respondent owns a car, truck, or van Whether respondent owns his/her home	X	X
Parent Wellbeing Depression 12-item Center for Epidemiologic Studies Depression Scale (CES-D) Functional impairment as a result of depression Substance Use Frequency of binge drinking Functional impairment from drinking or drug use Criminal Activity Number of times arrested since baseline Whether incarcerated at followup Social Support Number of people available to take care of baby in an emergency Number of people to turn to if there is a need to borrow $100 Number of people who can provide emotional support Health Status Respondent report on general physical health Whether respondent has health insurance and, if so, whether public or private	X	X
Child Wellbeing
Economic Resources Available to the BSF Child Whether the child is in poverty Whether the child has health insurance and, if so, whether public or private	X	X
Social-Emotional Development Self-Regulation Social Competence Externalizing behaviors Internalizing behaviors Attachment		X	X
Language Development Receptive Expressive		X	X

^aThe current submission seeks approval of the instruments for the 15-month follow-up. A request for approval of the instruments for the 36-month follow-up will be submitted at a later date.

The study will also explore the contextual factors and background characteristics of the couple and their children. This information will be used to describe the population served by BSF, explore for which families BSF is most effective, and increase the precision of the impact estimates. While most of this information is being collected on the baseline information form, some information about the couple and their baby at baseline that can easily be recalled at 15 months after random assignment (such as the gender and birth weight of the baby) will be collected by the 15-month survey. A list of the contextual factors and background characteristics to be collected and their corresponding data sources is provided in Table 3.

e. Data Collection Activities Requiring OMB Clearance

Clearance is currently being requested for the data collection efforts related to the implementation analysis and the 15‑month survey.

Implementation Analysis Data Collection. Clearance is requested for three data collection efforts related to the implementation analysis:

1. Site Visit Interviews. In all seven evaluation sites, we will interview program staff, including: administrators, supervisors, group facilitators, outreach and intake workers, and family coordinators. Interviews will also be conducted with staff members affiliated with referral sources for BSF. We will conduct two rounds of site visits. Repeat visits will provide additional information on implementation strategies, as well as document change over time. The first round of site visits will occur in the second half‑year of each program’s operation (2006), and the second round will be made one year later in 2007.

2. Focus Groups. To acquire feedback from couples involved with BSF, we will conduct focus groups with those who have actively participated in group sessions. We expect to include about five couples in each focus group. Focus groups will be conducted in each site, during each of the two rounds of site visits.

3. Phone Calls with Program Dropouts. We will conduct brief phone interviews with those who dropped out of the program after two or fewer group sessions, or never participated in group sessions. These phone calls, designed to explore why couples did not participate fully in the program, will be very brief and semi‑structured; 12 individuals from each site will be interviewed for approximately ten minutes each. Timing of these interviews will coincide with the first site visit.

Table 3. Baseline Information and Data Sources

	Data Source
Item	Baseline Information Form	15-Month Survey
Socioeconomic and Demographic Characteristics
Age Gender Race/ethnicity Primary language Religiosity Whether completed high school Whether working at baseline Date last worked Total earnings in previous 12 months Receipt of public assistance Number of children with BSF partner Number of children with other partners	X
Citizenship, country of birth Length of time lived in the US Whether respondent grew up with his/her biological mother, biological father, both, or neither Whether biological parents of respondent were ever married		X
Couple Relationship at Baseline
Marital status and history Whether cohabiting Perceived likelihood of marriage with BSF partner in future Length of time knew partner before pregnancy Attitudes toward marriage Satisfaction with conflict management Intimacy (showing love and affection) Supportiveness Perception of fidelity Commitment Friendship	X
Child Characteristics
Whether pregnancy was unintended or mistimed Whether entered BSF prior to birth of baby	X
Birth date of baby Gender of baby Birth weight of baby Whether part of a multiple birth		X
Other Stressors and Supports
Mental health Social support	X
Age of first sexual intercourse Whether victim of sexual abuse as child Whether victim of physical abuse as child Whether incarcerated before baseline Whether convicted of a crime prior to baseline and length of longest sentence		X

The implementation analysis will also use program observations and assessment of data from each site’s MIS.

15‑Month Telephone Survey. Clearance is also requested for the first telephone survey of the mother and father at about 15 months after random assignment. Both the father and the mother in the couple will be interviewed separately by telephone. Interviewing will take place at MPR’s centralized telephone interviewing facility. In addition, field locating and interviewing using cellular phones will be used with sample members who initially cannot be contacted or successfully interviewed. Bilingual interviewers will complete interviews in Spanish when necessary.

A2. How, by Whom, and for What Purpose Information will be Used

The findings from the implementation analysis will inform program administrators wishing to develop a new BSF program or improve an existing BSF program. By describing the intervention and how it varies across sites, the implementation analysis will also be used to inform the impact analysis. It may provide insights into the contexts required for BSF to work, reasons for any differences in impacts across subgroups and sites, as well as reasons for the relative size of impacts on different outcomes.

The findings from the impact analysis will provide information on whether, for whom, and under what circumstances BSF works. This information will be used by policymakers, program administrators, program funders, and unwed parents.

A3. Use of Automated Electronic, Mechanical, and Other Technological Collection Techniques

The data collection for the 15‑month survey will use Computer Assisted Telephone Interviewing (CATI). The CATI system reduces respondent burden by automating skip logic and question adaptations that allow interviewers to progress from question to question without having to refer back to previous answers to questions to check whether a follow‑up question should be asked or phrasing should be adjusted to properly apply to a respondent’s circumstances. CATI minimizes interviewer error through control over the question logic, consistency checks, and probes, and it eliminates the need to call back respondents to obtain missing data since inconsistencies in responses are corrected during the interview process.

The CATI system facilitates survey tracking because of its capability to produce timely reports on screening and interview outcomes, yield rates, item nonresponse rates, and interviewer productivity. CATI improves interviewer supervision through the use of audio and video monitoring. The autodialer, linked to the CATI system, virtually eliminates dialing error and improves interviewer efficiency. The automated call scheduler manages interviewer assignments by scheduling and rescheduling calls to ensure that they are made according to the optimal calling patterns, that all appointments are kept, and that cases requiring special attention or fluency in other languages are routed to the appropriate interviewers.

A4. Avoiding Duplication of Effort

There is no similar prior or ongoing data collection being conducted that duplicates the efforts of the proposed data collection. The survey will not ask for any information that can be obtained through abstractions of existing records.

A5. Sensitivity to Burden of Small Entities

None of the respondents will be small businesses.

A6. Consequences to Federal Program or Policy Activities if the Collection is not Conducted or is Conducted Less Frequently than Proposed

The field of marriage initiatives is very young and little is as yet known about how to develop and implement effective programs targeting couples and their relationships. Through the Deficit Reduction Act of 2005, Congress provided $100 million to support demonstrations of healthy marriage services. Because the Building Strong Families project was begun earlier, the collection of information through the evaluation will provide the first findings about the challenges to implementation of such interventions and, importantly, the net impact of interventions designed to serve couples and improve their relationships and marriages. While not all healthy marriage demonstrations will target unwed parents, the lessons and impact findings from this study will nonetheless provide important, useful and timely information to the federal government and state and local agencies operating healthy marriage demonstrations. Early lessons and impact findings may allow for mid-course corrections within other demonstrations, as appropriate, or additional or new policy developments in this new field.

Further, failure to collect the implementation analysis data as proposed would make it impossible for the study team to provide the federal government an independent description and assessment of each of the BSF programs and their approaches. We would have lost the opportunity to document the evolution of site operations during the evaluation and provide lessons based on the experiences in these sites. In addition, without the implementation analysis, we would have limited ability to interpret the quantitative findings of the impact analysis, particularly if there is an unexpected result that would require further investigation into program context or approaches.

Failure to conduct the 15‑month survey as proposed would preclude estimation of the impacts of the program. Without this information, ACF would not be able to determine whether BSF is meeting its stated goals. Program MIS information is insufficient for estimating impacts because it does not provide information on couples in the control group.

It is important to conduct a survey 15 months after random assignment rather than wait until a later period such as the second follow-up planned at 36 months for three reasons. First, it will allow us to measure the relationship quality of the couple soon after they have participated in the program. It is possible that the impact on relationship quality will attenuate over time after the couple has left the program. Failure to collect this information fairly soon after program participation may cause us to underestimate the short‑term impact on relationship status and quality. This is important because even short‑term impacts on the parents’ relationship could affect child well‑being. Second, it will allow us to accurately measure other short-term outcomes, such as the receipt of marriage and relationship education and support services, that if asked at 36 months after random assignment would likely suffer from recall bias. Third, the response rate to a later survey would likely suffer if there has been no contact with the respondents for an extended period of time after random assignment.

A7. Special Circumstances

There are no special circumstances.

A8. Federal Register Announcement and Consultation

The request for comment on the proposed continuation of information collection activity and instruments was published in the Federal Register on December 4, 2008 (Vol. 73, No. 234, p. 73935). A copy of the first notice is provided in Appendix F.

The second Federal Register notice was published in the Federal Register, Volume 74, No. 30, p. 7444 on February 17, 2009. A copy of the 30-day notice is included in Appendix G.

a. Comments

There were no comments in response to the Federal Register notice.

b. Consultation Outside the Agency

During preparation of the data collection instruments, we have engaged the professional counsel of a large number of people. These consultants include experts in the study of marriages and relationships, child development and well‑being, program design, the needs of specific populations, and study design. They also include curriculum developers and program administrators. In June 2005, the BSF Technical Work Group met and provided feedback on our study design as well as our data collection plan. The experts consulted for BSF are listed in Table 4.

c. Unresolved Issues

None.

A9. Payments and Gifts to Respondents

We plan to offer compensation for participation in the focus groups and for response to the survey.

For the implementation study, we propose to offer $35 to participants in focus groups to reimburse respondents for their expenses. Participants are giving up substantial personal time and may incur expenses related to transportation and childcare. For these reasons, it is important to offer incentives commensurate with the expense associated with participation.

To secure sufficiently high response rates to the telephone survey, we propose to make a $25 incentive payment to all survey respondents. Singer and Kulka (2002), in a review of research on the use of incentives in survey, found that incentives are cost-effective, lowering the overall cost and burden of most surveys. Studies have also shown that incentives may reduce differential response rates and hence the potential for nonresponse bias (Singer and Kulka 2002). For example, there is evidence that incentives are effective at increasing response rates for people with lower educational levels (Berlin et al. 1992) and low-income and nonwhite populations (James and Bolstein 1990).

Evidence suggests that the incentive cannot be much lower than $25. An incentive experiment from the 1996 panel of the Survey of Income and Program Participation showed that a $20 incentive significantly increased response rates, while a $10 incentive had no effect

Table 4. Technical Work Group, Consultants, and Other Reviewers

Name	Affiliation	Telephone Number
Technical Working Group
Paul Amato	Department of Sociology Pennsylvania State University	814 865 8868
Thomas Bradbury	Department of Psychology University of California, Los Angeles	310 825 3735
E. Mark Cummings	Department of Psychology University of Notre Dame	574 631 3404
Lindsay Chase‑Lansdale	Institute for Policy Research Northwestern University	847 467 6906
Ron Haskins	The Brookings Institution	202 797 6057
Edwin Hernandez	Center for the Study of Latino Religion University of Notre Dame	574 631 8558
Linda Malone‑Colon	National Healthy Marriage Resource Center	202 659 9366
Ronald Mincy	Columbia University School of Social Work	212 851 2408
Consultants
Irv Garfinkel	Columbia University School of Social Work	212 854 8489
John Gottman	Relationship Research Institute University of Washington	206 832 0305
Sara McLanahan	Center for Research on Child Wellbeing Princeton University	609 258 4875
Robert Rector	The Heritage Foundation	202 608 6213
Anne Menard	Domestic Violence Resources Network	717 259 3674
Mathematica Policy Research
Alan Hershey	Mathematica Policy Research	609 275 2384
Barbara Devaney	Mathematica Policy Research	609 275 2389
Shawn Marsh	Mathematica Policy Research	609 936 2781
Robert Wood	Mathematica Policy Research	609 936 2776
Sheena McConnell	Mathematica Policy Research	202 484 4518
Kim Boller	Mathematica Policy Research	609 275 2341
Robin Dion	Mathematica Policy Research	202 484 5262
Peter Schochet	Mathematica Policy Research	609 936 2783

relative to those who received no incentive. Burghardt and Homrighausen (2002) found response rates for the third follow-up survey of youth in the National Job Corps Study were low with only a $10 incentive. When OMB approval was received to increase the incentive to $25, the response rate increased and the cost per completed interview was nearly 20 percent lower than those interviews conducted with the $10 incentive.

A10. Confidentiality of the Data

The data from the 15‑month survey will not be maintained with any information that would allow personal identification of the respondents. Respondents receive information about confidentiality protection when they consent to participate in the study and information about confidentiality will be repeated as part of the survey interviewers’ introductory comments. Respondents will be informed that all the information they provide will be kept strictly confidential and that the results of the study will be presented only in aggregate form. A certificate of confidentiality from NIH has been obtained.

All interviewers will be knowledgeable about confidentiality procedures and will be prepared to describe them in full detail, if needed, or to answer any related questions raised by respondents.

The following safeguards will be employed by MPR to carry out confidentiality assurances, as they are routinely.

• All employees at MPR sign a confidentiality pledge that emphasizes the importance of confidentiality and describes their obligations.

• Access to identifying information on study respondents is limited to those who have direct responsibility for providing the sample and maintaining sample locating information.

• Identifying information is maintained on separate forms and files, which are linked to the interview only by sample identification number.

• Access to the file linking sample identification numbers with the respondents’ identification and contact information is limited to a small number of individuals who have a need to know this information.

• Computer files are protected with passwords, and access is limited to specific users. Especially sensitive data are maintained on removable storage devices that are kept physically secure when not in use.

A11. Additional Justification for Sensitive Questions

It is not possible to avoid some sensitive questions in a study of a program designed to affect personal relationships. Table 5 describes the justification for the sensitive questions included on the 15‑month survey. Although these questions are sensitive, they are commonly, and successfully, asked of respondents similar to those who will be in this study. All sensitive questions have been successfully pretested. All respondents will be informed of the confidentiality of their responses and that they do not have to answer questions they feel uncomfortable answering. All data will be presented in aggregate form; no information about an individual will ever be reported.

Table 5. Justification for Sensitive Questions

Question Topic	Justification
Whether the BSF partner is the other parent of children born after random assignment (Question FS52)	This question will allow us to examine BSF’s potential impact on multiple partner fertility. Multiple partner fertility has been shown to have negative consequences for child well-being, reducing financial and other support from parents and increasing children’s exposure to unrelated adults, which can increase the risk of child maltreatment (McLanahan and Sandefur 1994; Radhakrishna et al. 2001; Carlson and Furstenburg 2006; Harknett and Knab 2005). This question has been used on follow-up surveys conducted as part of the Fragile Families and Child Wellbeing Study.
Whether respondent or their BSF partner have cheated; perceived likelihood of cheating in the future (Questions RR8-RR11)	Infidelity has been found to be a major obstacle to marriage for unwed parents (Edin and Kefalas 2005, Smock and Manning 2003). The BSF curriculum aims to address this issue by discussing the importance of fidelity and trust in building a healthy relationship. Several large surveys have included similar questions concerning infidelity, such as the Study of Marital Instability Over the Life Course, the Louisiana Fragile Families Study, and the Baseline Survey of Family Experiences and Attitudes in Florida.
Whether respondent has been physically or sexually assaulted by their BSF or other partner (RR14-RR16)	The BSF intervention aims to improve relationship quality and increase the likelihood that couples enter into a healthy marriage. A key characteristic of a healthy romantic relationship is one that is not marred by violence. These questions are drawn from the revised Conflict Tactic Scale (CTS2), the most widely used tool for measuring domestic violence in research studies (Strauss and Douglas 2004). The CTS2 has been well validated and shown to have good internal consistency (Strauss et al. 1996). Versions of these CTS questions have been used on many surveys, including the National Family Violence Survey, the National Violence Against Women Survey, and surveys conducted in six states as part of the ASPE-funded TANF Caseload Project.
Symptoms of depression (WB1-WB3)	Parental depression has been shown to have adverse consequences for child outcomes (Gelfand and Teti 1990, Downey and Coyne 1990). Given BSF’s ultimate goal of improving child well-being, the link between parental depression and child well-being makes this outcome of particular relevance. BSF may reduce depressive symptoms among participants by reducing stress and conflict in relationships. These questions represent the 20-item Centers for Epidemiologic Studies Depression Scale (CES-D), a widely-used measure with well-established good psychometric properties (Radloff 1997). The CES-D has been used as part of many large surveys, including those used as part of the Early Head Start Evaluation, the National Longitudinal Survey of Youth, and the Project on Devolution and Urban Change.
Alcohol and drug use (WB4-WB6)	Substance abuse and addiction can have major negative effects on the well-being of individuals and their families. If BSF improves relationship quality and stability, it may reduce substance abuse among participants. The question we include concerning binge drinking was developed by Henry Wechsler and is recommended as a screening tool by the National Institute on Alcohol Abuse and Alcoholism (Wechsler et al. 1995; Wechsler 1998). It has been used in several large national surveys, including the National Survey on Drug Use and Health and the Youth Risk Behavior Surveillance System. The two questions concerning functional impairment resulting from substance use come from Fragile Families surveys.
Family income (WW1-WW42)	Family income and poverty are important determinants of child well-being (Brooks-Gunn and Duncan 1997; Mayer 1997). BSF aims to enhance child well-being by improving the parental relationship and the likelihood that the parents remain together as a couple. Since two-parent families generally have higher incomes than single-parent ones, increases in family income may be an important avenue by which BSF improves child outcomes. Family income has been collected on many national surveys, including the National Survey of America’s Families and the Fragile Families surveys. The particular questions we use are drawn from the Work First New Jersey study, a large longitudinal study of welfare recipients.
Involvement with the criminal justice system (questions WB9-WB30)	Recent research suggests that a history of incarceration and involvement with the criminal justice system may be fairly common among fathers in the BSF target population (Western 2004). Parental incarceration has major negative effects on child and family well-being, reducing the financial and other support the parents can provide their children and families. BSF may reduce criminal involvement through its potential effects on relationship stability and quality. Similar questions have been included in other large national studies, such as Fragile Families survey and the National Job Corps Study.
Childhood history of sexual or physical abuse (BP7-BP8)	A history of physical and sexual abuse during childhood has been shown to reduce the likelihood of entering into and sustaining healthy relationships and marriages as an adult (Cherlin et al. 2004). This research also indicates that a history of childhood abuse is fairly common among low-income populations. For these reasons, those with a history of childhood abuse will be an important subgroup to examine as part of the BSF impact analysis. These two questions are from surveys conducted as part of Welfare, Children, and Families: A Three-City Study.
Age of first intercourse (BP9)	The BSF curriculum aims to build commitment and trust among unmarried couples with young children as a means of strengthening and preserving their romantic relationships. Individuals with a large number of sexual partners prior to entering the program may have difficulty establishing the necessary level of commitment and trust to build a healthy and lasting romantic relationship. Therefore, the number of sexual partners prior to random assignment is a variable of potential interest for subgroup analysis. Because of recall difficulties, however, asking about the number of sexual partners prior to random assignment on the 15-month follow-up survey is not practical. Therefore, we will ask instead about the age of first intercourse, which has been shown to be a good proxy for the number of sexual partners (USDHHS 1997). This question is drawn from the National Survey of Family Growth.
Number of sexual partners since random assignment (BP10)	Children who are exposed to the new romantic partners of their parents are placed at increased risk of abuse and other adverse outcomes (Radhakrishna et al. 2001). It is hoped that by increasing the likelihood that participating couples remain together, BSF will reduce the exposure that their children have to the new romantic partners of their parents. Therefore, the number of sexual partners since random assignment is an important variable to examine as part of the impact analysis. This question is drawn from the National Survey of Family Growth.

A12. Estimates of the Hour Burden of the Collection of Information

Table 6 presents the number of respondents, the number of responses per respondent, the average burden hours per response, and the total annual burden hours for the data collection.

Data for the implementation study will be collected from interviews of program staff and other key staff, focus groups with couples participating in the program, and phone interviews with couples who attended few or no group sessions. Data will also be collected by program observation and analysis of data from each site’s MIS. Much of the information will be collected during two rounds of site visits to all seven sites. The first round of site visits will occur in the second half‑year of each program’s operation (2006), and the second round will be made one year later in 2007.

Table 6. Annual Burden Estimates

Implementation Study
Instrument	Number of Respondents	Number of Responses per Respondent	Average Burden Hours Per Response	Total Burden Hours	Average Hourly Wage2	Total Annual Cost
Staff interview protocol	126	1	1.5	189	$19.30 (wage for “Social Workers”)	$3,648
Focus group protocol	70	1	1.5	105	$18.62 (wage for “All Occupations”)	$1,955
Telephone interview protocol (nonparticipants/dropouts)	84	1	0.17	14	$18.62 (wage for “All Occupations”)	$261
Impact Study
15-month survey (females)	1,434	1	0.91	1,305	$18.62 (wage for “All Occupations”)	$24,299
15-month survey (males)	1,434	1	0.83	1,190	$18.62 (wage for “All Occupations”)	$22,176
Estimated Totals				2,803		$52,339

The total burden estimate for the site visit interviews with program staff is 189 hours per year for two years. During each site visit, we will conduct approximately six interviews. Each interview may involve multiple respondents, ranging from two to nine people; we expect that on average there will be three respondents per interview. Each interview will last approximately one and one‑half hours. The total annual burden for these interviews will be 7 (sites) x 6 (interviews per site) x 3 (respondents per interview) x 1.5 hours = 189 hours.

The total burden for the focus group portion of the data collection effort is 105 hours per year for two years. The sample for each focus group is estimated to be 10 participants (5 couples). We estimate that each focus group will last approximately one and one-half hours, and that we will conduct a focus group in all seven sites during both rounds of visits. Thus, the annual burden of the focus groups will be 7 (sites) x 10 (participants per focus group) x 1.5 hours = 105 hours.

The estimate of total burden for the brief phone interviews with nonparticipants is 14 hours. We will contact couples who attended either no group session or less than two group sessions. These interviews will be very brief and semi‑structured; 12 individuals from each site will be interviewed for approximately ten minutes each. Timing of these interviews will coincide with the first site visit. Thus, the annual burden for the focus groups will be 7 (sites) x 12 (respondents) x 10 minutes = 840 minutes = 14 hours.

The 15-month survey will involve separate interviews with the mother and father of the focal child (who made the couple eligible for BSF) about 15 months after the couple was enrolled into the study. Each survey will be administered once to each respondent. Interviewing will start in fiscal year (FY) 2007 and is expected to be completed in FY 2009.

The mother and the father will be asked similar questions. However, the interview with the mother is slightly longer than the interview with the father because the mother is asked some questions that are not included in the father’s interview. Based on the pretests, we expect the mothers’ interview to last an average of 55 minutes and the fathers’ interview to last 50 minutes.

We will attempt to interview all 3,375 mothers and 3,375 fathers in the sample. Based on our experiences interviewing similar groups of respondents, we expect the response rate to be 85 percent. Hence, we expect to complete interviews with 2,869 mothers and 2,869 fathers.

The total burden of the mothers’ survey is 2,869 (respondents) x 0.91 hours = 2,610 hours over two years. The total burden of the fathers’ survey is 2,869 (fathers) x 0.83 hours = 2,381 hours over two years. The annual burden for the mothers and fathers survey together is 2,496 hours.

To compute the total estimated annual cost, the total burden hours were multiplied by the average hourly wage, according to the closest professional grouping in the National Compensation Survey, 2005. The total estimated annual cost is $52,339.³

A13. Estimate of Total Annual Cost Burden to Respondents or Record Keepers

None. Respondents will not incur any out-of-pocket costs. Telephone calls will be placed at the expense of the evaluation contractor, and respondents who wish to call the interviewers will be provided with a toll‑free number billed to the contractor.

A14. Estimate of Annualized Costs to the Federal Government

The total estimated cost to the federal government of this data collection is $4,985,014. This includes the cost of the implementation data collection and analysis, which includes developing protocols, conducting site visits, analyzing the data, and reporting the results. It also includes the cost of designing and administering the first follow-up survey, processing and analyzing the data, and preparing the first follow-up report. Table 7 shows the annualized costs to the federal government.

Table 7. Annualized Costs to the Federal Government

	Annualized Costs	Total
Implementation study data collection and analysis	$233,600	$1,168,002
15-month survey and analysis	$763,402	$3,817,012
Total	$997,003	$4,985,014

A15. Reasons for Program Changes or Adjustments

This is a request for continuation of an approved information collection. There is no change in burden.

A16. Plans for Tabulations and Publication and Schedule of Project

Our approach to addressing the research questions discussed in Section A.1d entails two complementary analytical methods: (1) implementation analysis, and (2) impact analysis.

a. Implementation Analysis

Data collected for the implementation analysis from each site will be first analyzed independently by site; for sites with multiple sub‑sites, we will focus on information that is common to all sub‑sites, although we will also highlight key differences at specific sub‑sites. We will describe particular aspects of the sites’ operations and structure, noting challenges the sites have encountered and how they have been managed. This level of analysis will include the “theory of change” as articulated by each program, including how the site expects to change participants’ behavior and attitudes. In these analyses, the multiple‑perspective approach will be applied, wherein information from interviews, site observations, and past visits are cross‑referenced to promote accuracy and consistency.

The analysis will then focus on combining experiences across sites and will be organized around issues, rather than sites. These issues may include various recruitment strategies, or determining which methods of recruitment were associated with a higher proportion of consenting couples. This does not imply that site context will be ignored in this analysis; we cannot argue that strategies would be universally applicable once implemented beyond the existing sites. However, we strive to capitalize on the variation within the sites for uncovering different approaches and practices.

Our approach to analysis will shape our reports. We will produce internal site‑specific memos following each site visit; these documents will focus on program structure, operations, key implementation issues, lessons for improving implementation, and outstanding implementation issues. In addition, after the first round of site visits, we will produce a report focusing on BSF implementation across sites. This report will include an overall assessment of BSF programs, analyzing the various approaches by comparing the experiences of the sites. The report will also provide the background to link the program’s structure and processes with the impact estimates of the evaluation. The results from the second round of site visits will contribute to further analysis, including an examination of the evolution of BSF programs over time. A chapter in the final report will draw on these results in detailing the implementation of BSF programs during the evaluation period.

b. Impact Analysis

The goal of the impact analysis is to compare observed outcomes for program participants with what those outcomes would have been without BSF. Because this counterfactual cannot be observed, we will use the experience of the control group as a measure of what would have happened to the program group couples in the absence of BSF. Random assignment of couples to a program and a control group ensures that the two groups of couples do not initially differ in any systematic way on any characteristic, observed or unobserved. Any observed differences in outcomes between the program and control group couples can therefore be attributed to BSF with a known degree of precision.

Depending on the outcome considered, the unit of observation for the analysis might be the couple (for marital status, for example), the individual (for parent well‑being, for example), or the child (for child well‑being, for example).

Differences of means or proportions in outcomes between the program and control group will provide unbiased estimates of the impacts of BSF. More precise estimates will be obtained using regression models to control for random differences in the baseline characteristics of program and control group members. In their simplest forms, these models can be expressed by the following equation:

(1) Y = X΄ß + δ P + e,

where:

Y is an outcome variable

X is a vector of control variables (including an indicator for each site)

P is an indicator that equals 1 for program group members and 0 for control group members

β is the vector of regression coefficient for the control variables

δ is the measure of the impact of BSF

e is a random disturbance term that is assumed to have a mean of zero conditional on X and P, and is interpreted as the unobserved factors that affect Y.

The statistical techniques used to estimate the regression‑adjusted impacts depend on the form of the dependent variable, Y. If the dependent variable is continuous, then ordinary least squares techniques will produce unbiased estimates of the parameter δ. However, if the dependent variable is binary—for example, whether the couple is married—then consistent parameter estimates can be obtained by using logit or probit maximum likelihood methods. If the dependent variable is censored or truncated—such as hours spent with the child in a given week—then tobit maximum likelihood or two‑stage procedures will be used.

Control variables in the vector X will include any variables that may affect the outcome that are not affected by the intervention. Hence, X could include the characteristics of the individual or couple for which data are collected on the baseline information form, including the status and quality of the relationship at baseline. X could also include baseline characteristics that can be easily recalled and will be measured using the 15-month survey (such as incarceration prior to random assignment.)

We will also estimate impacts for individual sites and for groups of sites. BSF may work better in some sites or circumstances than in others. For example, it is possible that BSF works better when it includes home visiting than when the family coordinator’s role is less intensive. Estimates of the impacts by sites or groups of sites will be obtained by introducing interaction terms in the regression model (1) that is the product of the program group indicator (P) and an indicator of membership in the site or group of sites of interest (Site_i):

(2) Y = X΄ß + δ P + γ_i* Site _i * P + e.

The estimated impact of BSF for site _i is given by (γ_i + δ).

Some people may benefit from BSF more than others. For example, it is possible that couples with a strong relationship at baseline gain the most from the BSF services. To address how BSF works for different subgroups of the population, impacts for key subgroups of the population will be estimated. Subgroups of interest include those defined by the following characteristics at baseline: demographic characteristics (such as age and race/ethnicity, and similarity in these characteristics within the couple); education; whether the baby was wanted or mistimed; relationship status and quality; whether the BSF focal child is the couple’s first child; whether either member of the couple has children with other partners; employment, income, and receipt of public assistance; mental health status; existence of social supports; and expectations of marriage.

Estimates of impacts by subgroup will be obtained by introducing an interaction term in the regression model (1) that is the product of the program group indicator (P) and an indicator of membership in the subgroup of interest (Sub_i):

(3) Y = X΄ß + δ P + θ_i* Sub _i * P + e.

The estimated impact of BSF for members of the subgroup of interest is given by (δ + θ_i ).

c. Publication Plans

We will prepare the following reports.

• Implementation Analysis Report. This report will present initial findings from the implementation analysis of the BSF programs after the first round of site visits. This report will describe the background and context for each site, describe the implementation of the BSF programs in each site, and illustrate the successes and challenges faced in the implementation of BSF. The report will be produced in late 2006.

• Interim Report on Program Impacts. An interim impact analysis report will present the impact estimates estimated using data from the 15‑month follow‑up survey. This report will be produced in 2010.

• Final Report. The final report will present the findings from the implementation and implementation analysis in addition to the impact findings using data from the 15‑month and the 36‑month follow‑up surveys. This report will be produced in 2011.

• Topical Papers and Research Briefs. These papers and research briefs will describe special topics of interest. The papers and briefs will be produced as requested by ACF.

d. Project Schedule

The project began in October 2002. The sites started pilot operations between February and September 2005. Intake for the evaluation began between July 2005 and November 2005, depending on the site. Intake in each site will last between 18 and 24 months, depending on the flow of couples into the program. Follow‑up surveys will be conducted 15 months after random assignment and roughly 36 months after random assignment. The study is currently scheduled to end in 2011. The proposed schedule for data collection, analysis, and reporting is provided in Table 8.

Table 8. Schedule for BSF Project

Activity/Data Collection	Time Period
Activity Study design and development of potential BSF sites Sample intake for first site Sample intake for last site	October 2002 - November 2005 July 2005 - July 2007 November 2005 – February 2008
Data Collection Baseline 15-month follow‑up survey 36-month follow‑up survey	July 2005 – February 2008 October 2006 – December 2009 October 2008 – March 2011
Analysis and Report Preparation Implementation analysis reports Interim report on program impacts Final report Special topics and research briefs	January 2005 – April 2009 February 2009 - May 2010 October 2010 - December 2011 As requested

A17. Reasons for not Displaying Expiration Date for OMB Approval

The expiration date will be displayed. The OMB number and expiration date will appear on the 15-month CATI questionnaire’s introductory screen.

A18. Exception to the Certification Statement

Exception to the certification statement is not requested.

1 MDRC, the Urban Institute, Decision Information Resources, and Public Strategies.

2 Average hourly wage, according to the closest professional grouping, in the National Compensation Survey, 2005. Bureau of Labor Statistics, National Compensation Survey, 2005. Retrieved on July 31, 2008 from http://www.bls.gov/ncs/ocs/

3 Source: Bureau of Labor Statistics, National Compensation Survey, 2005. Retrieved on July 31, 2008 from http://www.bls.gov/ncs/ocs/

2