Transition into Adulthood Supplement

An Introduction to the TAS

Over the past several decades, the U.S. and other countries have seen a lengthening of the period between childhood and adulthood—the “transition into adulthood.” Youth no longer move quickly from secondary education into the labor force and independent economic living. Based on data from the Panel Study of Income Dynamics (PSID), less than 50% of individuals will form their own independent family unit before they reach their mid-20s.

Scientists are becoming increasingly aware that the period between the ages of 18 and 28 years are critical for life span development. It is during this period that major investments are made in education, crucial decisions are made regarding partnering and childbearing, and careers are planned and initiated. For PSID, this means that important educational and occupational transitions are often made while young adults are still dependent on their parents and are not primary respondents themselves.

The “Transition into Adulthood Supplement” (TAS) is part of PSID and may include individuals from the PSID Child Development Supplement (CDS). Although PSID collects some information about everyone who is a co-resident member of each family in the study, considerably more information is collected on the Reference Person and Spouse Partner (R-S/P). The TAS is intended to fill in the gap of time and information collected about youth experiences and the first interview as a PSID R-S/P.

The launching of TAS in 2005 was motivated by recognition that these years are marked by choices, changes, and transitions that have profound life-long consequences, but would be missed by the sample design of the PSID prior to 2005. To bridge this gap, TAS was initially tied to the Original CDS cohort, which began with children ages 0-12 in 1997, and was launched in 2005 when the oldest members of the Original CDS cohort reached 18 to 20 years of age. TAS has subsequently been conducted in 2007, 2009, 2011, 2013, and 2015. By the 2015 wave of TAS, all members of the Original CDS had reached adulthood and were eligible for at least one wave of TAS. In 2017, TAS was relaunched to capture information on the transition into adulthood all young adults in the PSID, not just those who participated in the Original CDS.

Based on current literature and theories guiding research on the adult transitional years, the TAS interview builds on the information collected from some of these young adults when they themselves were interviewed as children and adolescents in the CDS, and, at the same time, harmonized and coordinated with data to be collected on them when they are interviewed as adults in future waves of Core PSID.

The Panel Study of Income Dynamics is a longitudinal survey of a nationally-representative sample of U.S. families. Since 1968, PSID has collected data on family composition changes, housing and food expenditures, marriage and fertility histories, employment, income, wealth, time spent in housework, health, expenditures, philanthropy, and more. Over 100,000 people have ever participated in the panel, which includes up to seven generations within a family. PSID is the longest running panel on family dynamics, and is considered one of the most important data archives in the world. The PSID now is conducted biennially, primarily via telephone with data collection commencing in March and ending by December of odd-numbered years.

In 1997, PSID supplemented its main survey with collection of additional data on a cohort of 0-12 year- old children in the study and their parents. The objective of this supplement—the Original Child Development Supplement—was to provide researchers with comprehensive, nationally representative, and longitudinal data on children and their families with which to study the dynamic process of early human capital formation. Two additional waves of the Original CDS were conducted in 2002-2003 (CDS-II), when the children were 5-17 years of age, and in 2007-2008 (CDS-III) for children in the cohort who were under 18 years of age.

Within the context of family, neighborhood, and school environments, CDS gathered information about a broad array of developmental outcomes including (but not limited to) physical health, emotional well-being, cognitive skills, education achievement, and social relationships with family and peers. Each Original CDS child could have up to eight modules of data collected from three different family members (primary and secondary caregivers and the target child) and a school information source (teacher and/or school administrative data).

The Ongoing Child Development Supplement collects data on children’s health, development, and well-being within the children’s family and neighborhood context. In 2014, CDS was relaunched to collect data from all children in PSID households aged 0–17 years. Detailed information is collected on the same topics as in the Original CDS, including time diaries, assessments of reading and math skills, interviews with children’s primary caregivers and direct interviews with older children themselves. Because the CDS is a supplement to the PSID, an extensive amount of family demographic and economic data about the CDS child’s family is collected in PSID, providing more extensive family data than any other nationally-representative longitudinal survey of children and youth in the U.S.

Bridging the Gap Through the Original and Ongoing CDS, detailed information has been collected on participants during their childhood and adolescence. CDS youth will eventually become the future “active panel” of Core PSID when they move out of (or “split-off”) from their parents’ home and establish an independent household of their own. Under the current design of the TAS, CDS young adults will participate in TAS data collection until they reach age 28, regardless of whether they have become members of Core PSID. When they join Core PSID, they will participate in that study every other year from that point forward.

The Transition into Adulthood Supplement thus serves as a “bridge of information” between the rich data collected in the CDS on the years between birth and age 18 years, and the rich data collected in the PSID on the years after economic independence is established.

TAS Questionnaire Content

The TAS questionnaire comprises 10 sections, each of which represents a specific area of interview content. A summary of each section is provided below.

Section A: Community Engagement and Technology Use

Questions in Section A focus on involvement over the last 12 months in the community including volunteering and community service, group organizations, and sports participation, as well as the type of organization and frequency of participation. A question series on technology use asks about the access and ownership of cell phones, computers, tablets, and the internet. Frequency and type of technological use is also collected. Respondents living at home or away at college were asked all questions in this section; respondents living on their own were not asked questions about when they were widowed or when they were divorced, as these questions were asked in their 2017 Core PSID interview.

Beginning in TAS-2017: Questions A10a-A10e, which asked about internet use, were replaced by questions A17-A25, which gathered in depth information on the ownership of cell phones, smart phones, computers, tablets, types of internet use, and technological literacy.

Section B: Family Relationships, Personality, and Mental Health

Section B assesses the individual’s relationship with his or her parents. Respondents are also asked a series of questions which comprehensively assessed adult well-being in terms of emotional, psychological, and social well-being. Questions also included measures of self-rated levels of skill in areas such as leadership, intelligence, independence, confidence, and problem solving, as well as self-rated psychosocial measures about worries and discouragement.

The level of responsibility that the respondent assumes for living arrangements and money management including earning their own living, making rent or mortgage payments, paying their bills, and managing their personal finances is also assessed. Respondents were asked to rate their abilities to manage their money and solve day-to-day problems. Information about living arrangements during a typical school year and during the summer was also collected.

Respondents living at home or away at college were asked all questions in this section; respondents living on their own were not asked questions about when they were widowed or when they were divorced, as these questions were asked in their 2017 Core PSID interview.

Beginning in TAS-2017: Questions B6a-B6d about responsibility and C2d-C2f pertaining to worry were removed and questions B27a-B27k were added. These questions comprise the Rosenberg Self-Esteem Scale, a “10 item scale that measures global self-worth by measuring both positive and negative feelings about the self.”

Section C: Interpersonal Relationships

This section obtained information about the current marital and cohabitation status of the individual and subjective evaluations of all romantic/intimate relationships through questions about living arrangements, general satisfaction with relationships, time spent with partner, future expectations of relationship duration, and the likelihood of marriage and divorce. Information was collected on past, present, and future childbearing and fertility expectations, gender roles, biological/adopted child rearing/family values, and parenting skills and experiences.

Respondents living at home or away at college were asked all questions in this section; respondents living on their own were not asked questions about when they were widowed or when they were divorced, as these questions were asked in their 2017 Core PSID interview.

Beginning in TAS-2017: This section was revised substantially to align more closely with measures of cohabitation, marriage, and sexual behavior collected in the National Survey of Family Growth. Questions C4-C11 were added to capture detailed dating information on cohabitation and marriage, while questions C20-23, C32-C34, and C42-C43 were added to obtain information on sexual experiences and pregnancy.

Section D: Employment, Military Service, and Time Use

Section D collected detailed information about current employment status and all types of employment and money-earning activities for the previous two years. Measures included salary/wages, hours, experience, and size and type of the employer, reasons for being unemployed and/or not working, as well as the methods and frequencies of job hunting. Moreover, detailed information was collected about service in any branch of the Armed Services, and self-rated satisfaction with military service was obtained.

Information about how individuals spent their time during the past 12 months was collected including time spent on leisure activities, computer/internet use, and community engagement. Certain items from the CDS Primary Caregiver Child file were asked, permitting time-series analysis of activity patterns in organized arts and sport, TV watching, reading, and computer use.

Beginning in TAS-2017: Question series D9a-D9h were added in connection to the 2017 Core PSID, asking questions about hours, weeks, and overtime worked. In addition, questions D77-D81 and D112-D123 were added to both TAS-2017 and PSID-2017 which asked about time spent working, shopping, doing housework, caring for children, caring for adults, volunteering, doing educational activities, and doing leisure activities, as well as stylized time use measures obtaining information on the types of activities done during work.

Section E: Past Year Income and Financial Help

Information was collected on income earned during the previous calendar year from multiple sources, including unemployment compensation, workers’ compensation, dividends, interest, trust funds, child support, welfare, as well as financial help received from parents and other relatives for daily living expenses, larger monetary gifts, and inheritances.

Respondents living with their parents or away at college are asked all questions in these sections; respondents living on their own were only asked the latter questions pertaining to financial help, gifts, and wealth because their income and business holdings information was gathered in their 2017 Core PSID interview.

Beginning in TAS-2017: Section E was expanded to include more detailed questions about financial help received from parents or relatives for housing, education, vehicles, and living expenses. In order to reduce missingness, bracketed amounts were asked in each of these categories to obtain a more accurate value of these types of financial help.

Section F: Wealth

A series of questions estimating the net value of automobiles, stocks and bonds, checking and savings accounts, life insurance policies, and any other assets and investments is asked. Information is also collected about student loans, credit card balances, and other debts.

Respondents living at home or away at college were asked all questions in this section; respondents living on their own were not asked questions about when they were widowed or when they were divorced, as these questions were asked in their 2017 Core PSID interview.

Beginning in TAS-2017: In Section F, questions pertaining to the Great Recession were removed

Section G: Education

A key marker of the transition into adulthood is attainment of post-secondary educational degrees, which, in turn, feeds into work plans and career aspirations.

In Section G, information is gathered about the amount, dates, and location of education, starting with high school completion or GED attainment, high school GPA, and experience with college entrance exams. Respondents are asked if they had ever attended or are currently attending college and, if not, the reason for not attending.

Beginning in TAS-2017: This section was streamlined to coordinate with the education section in the Core PSID, asking highest grade of education, degrees, certifications, and licenses obtained, and standardized tests.

Section H: Health

Section H includes a measure of self-rated overall health and whether they have ever been diagnosed with a series of chronic illnesses/conditions such as asthma, diabetes, hypertension, cancer, any mental health condition, and learning disabilities. Age when first diagnosed and limitations on normal daily activities that resulted from each condition was asked. The section includes a short series of questions about psychological distress (K6) during the past 30 days. These questions are also asked in the Core PSID instrument.

In Section H, questions were asked about routine visits to the doctor and dentist, maintenance of a healthy body weight, and engagement in a number of lifestyle practices such as exercising, eating balanced meals, tobacco use, binge drinking, the use of illegal drugs or misuse of prescription medicines, and unprotected sex.

For respondents living on their own as PSID R-S/Ps, the first part of the section was skipped and started with health behaviors to avoid repeating questions that are collected in the Core PSID interview.

There is no Section I in the TAS-2017 Questionnaire.

Beginning in TAS-2017: Section H included a new retrospective childhood health calendar and follow-up questions on the effects of childhood health conditions on schooling and other activities. Conditions include but are not limited to: asthma, diabetes, cancer, high blood pressure, ear problems, headaches, and more. New questions on drug use were also added, including personal history and frequency of vaping.

In addition to childhood health, questions about parental mental and physical health during different stages of childhood were ascertained. A set of questions on Adverse Childhood Experiences (ACEs) were also included in TAS-2017, which collect information on childhood physical abuse, verbal abuse, sexual abuse, physical neglect, and emotional neglect. The ACEs questions are also asked in the 2014 Childhood Retrospective Circumstances Study.

Section K: Discrimination and Peer Influence

Section K includes questions addressing everyday discrimination, peer influence, assault, risky behavior, and encounters with the law. Day-to-day encounters with discrimination are measured by asking about frequency of experiencing specific types of discrimination. If any experience is endorsed as happening more than once a year, the perceived reason for the discriminatory experience is asked.

Peer influence is assessed using a set of questions about characteristics of friends with respect to school and work-related activities, community involvement, and general outlook and attitudes about the future. The frequency of engaging in dangerous and risky behaviors over the prior six months is assessed included fighting, damaging property, and drunk driving. Incidents of arrest, probation, and jail time were measured separately through questions on when and why the offense(s) occurred. Respondents are also asked about prior assaults, and the age at which an assault happened.

Beginning in TAS-2017: Section K added streamlined questions about the reasons behind discrimination in daily life.

Section L: Religious and Spiritual Beliefs, Race and Ethnicity

Section L assesses current religious preferences and the importance of religion and spirituality in the respondent’s life, as well as obtaining information on race, ethnicity, and locations of ethnic origin.

Beginning in TAS-2017: Questions on religious preference and religiosity were modified to mirror the questions in the 2017 Core PSID. The race and ethnicity question included new response options, specifically Middle Eastern or North African, and also included follow up items on origin for Hispanics, Asians, Middle Eastern or North Africans, and Native Hawaiian or Other Pacific Islanders. These items on race/ethnicity were asked of all TAS-2017 respondents and were new this wave.

TAS Sample

Eligibility

Age eligibility. Age eligibility at each wave is determined by birth year. Birth cohorts for each wave appear below. Participants are at least 18 years old and no older than 28 years in in the survey year.

TAS 2005: 1984-1987

TAS 2007: 1984-1989

TAS 2009: 1984-1991

TAS 2011: 1984-1993

TAS 2013: 1985-1995

TAS 2015: 1987-1997

TAS 2017: 1989-1999

Age Requirements. All potentially-eligible TAS respondents were identified and screened in the Core PSID interview. During the TAS interview, respondents were asked to confirm their date of birth. If, during the TAS interview, the interviewer learned that the respondent was under the age of 18, the interviewer was instructed to code the respondent as non-sample (age ineligible).

Active sample status. The young adult’s family participated in the same-year Core PSID interview (either through their own interview as Reference Person or Spouse/Partner or by identification as an “other family unit member” in a household interview).

Followable sample status. The young adult is a member of of the PSID Sample, meaning that they are a lineal descendant (natural or adopted) of individuals who were living in the original family unit at the time of the very first interview.

TAS Sample by Survey Year, Age in Years, and CDS Affiliation

Age in years/Survey year	2005	2007	2009	2011	2013	2015	2017	2019
28					X	X	X,Z	X,Z
27				X	X	X	X,Z	X,Z
26				X	X	X	X,Z	X,Z
25			X	X	X	X	X,Z	X,Z
24			X	X	X	X	X,Z	X,Z
23		X	X	X	X	X	X,Z	X,Z
22		X	X	X	X	X	X,Z	X,Z
21	X	X	X	X	X	X	X,Z	Y,Z
20	X	X	X	X	X	X	X,Z	Y,Z
19	X	X	X	X	X	X	Y,Z	Y,Z
18	X	X	X	X	X	X	Y,Z	Y,Z
Total	745	1,118	1,554	1,907	1,804	1,641	2,526	In process
From Original CDS (X)	100%	100%	100%	100%	100%	100%	48%
From Ongoing CDS (Y)							35%
No CDS Participation (Z)							17%
Response Rate	89%	91%	92%	92%	90%	87%	86%

TAS Sampling Weights

This chapter provides an overview of the sample weights for TAS-2017. There will be three weights for TAS-2017, a cross-sectional weight and two longitudinal weights. The cross-sectional weight accounts for unequal selection probabilities due to the PSID sample design, while the longitudinal weights are panel weights for any TAS Respondent who was interviewed in (1) the Original CDS cohort or (2) the Original or Ongoing CDS. The second longitudinal weight will be included in Release 2 of the TAS-2017 data.

TAS-2017 Cross-Sectional Weight

The TAS-2017 data are provided with a cross-sectional weight for every survey respondent to be used in data analysis and obtain unbiased estimates for population parameters. We describe in this chapter the construction of the TAS-2017 cross-sectional weight.

Weighting Methodology

The TAS-2017 cross-sectional weight was designed to account for the unequal selection probabilities, due to the original PSID sample design, and for differential eligibility and nonresponse. These weights were also calibrated to selected demographic variables of the target population to further mitigate any coverage and nonresponse error, and to improve the precision of survey estimates. We describe below the three main components of the TAS-2017 cross-sectional weights.

Base Weights

In order to account for differential selection probability and nonresponse during the PSID recruitment, we used as base weights for TAS-2017 the 2017 PSID Individual Longitudinal Weights. However, due to various eligibility criteria (including, for example, having to respond to both PSID 2013 and 2015), not every sample person at PSID 2017 that belongs to the TAS target population (individuals between 18 and 28 years old by December 31st, 2017) was eligible for TAS-2017. In order to account for this difference, the base weights of such ineligible cases were re-distributed across the eligible sampled individuals, so that the sum of the weights reflects, on average, the size of the target population.

Nonresponse adjustment

Unit nonresponse poses a threat to the quality of survey estimates as respondents and nonrespondents might differ in terms of the study outcomes, which can ultimately cause nonresponse bias in such estimates. If nonresponse follows a missing at random (MAR) mechanism (Little and Rubin, 20024), this nonresponse bias can be attenuated through certain statistical adjustments. To that end, a nonresponse weighting adjustment was performed over the TAS-2017 data using a response propensity procedure. In this approach, the weights are inversely proportional to estimate of the probability of response to the survey. These estimated probabilities of responding the survey, also referred to response propensities, are computed using a logistic regression model of the survey response indicator over a set of covariates available for both respondents and nonrespondents. In order to reduce nonresponse bias while not increasing sampling variance of the survey estimates, the covariates used in this adjustment should be correlated with both the survey response and the study outcomes (Little and Vartivarian, 2003 ). For this reason, the following TAS-2017 outcomes were selected to assist in this adjustment:

• Body Mass Index (BMI),
• Weeks of employment in previous year (WKSEMPPY),
• Weeks of employment in the past two years (WKSEMPPPY),
• Completed education of mother (MOCED),
• Completed education of father (FACED),
• Marital/cohabitation status (TAMS), and
• School enrollment status (EDSTAT)

Ideally, we would like to use these survey outcomes as covariates in the response propensity modelling. However, we only observe them for the survey respondents. Instead, we computed predictions for both nonrespondents and respondents for each of these survey outcomes using regression models over covariates available for every sampled individual, including:

• Census variables: block group or tract level variables from the Census Planning Database that can be appended to the sampling frame, such as percentage and medians of population, households and housing units by socioeconomic characteristics.

• Paradata: variables generated as a byproduct of the data collection itself, such as number of call attempts by survey mode (face-to-face, telephone, e-mail), indicator of refusal in previous waves.

These sets of covariates are typically available for every sampled element. There are a few cases in the TAS-2017 sample with missing values in some of these variables though. Therefore, as a first step in this nonresponse adjustment, we used regression-based single imputation to fill in the missing values in those variables. For the Census variables, the data were aggregated and imputed at the tract-level, such that elements within the same block group or tract received the same imputed values. All other variables were imputed at the element-level.

Next, we fitted a regression model for each of the seven selected outcome variables over the respondents’ data using all the covariates mentioned above. Given the large number of covariates in the models, we used the Lasso (Least absolute shrinkage and selection operator) for both variable selection and estimation in each of these models. Using these fitted regression models and the observed/imputed covariates for every sampled individual, we predicted the survey outcomes for both respondents and nonrespondents. These predictions can be seen as proxy summaries of the covariates correlated to the selected survey variables, thus satisfying one of the conditions for a successful nonresponse adjustment (Little and Vartivarian, 20036). The probability that a sample person was a respondent in TAS-2017 was estimated using a logistic regression model. The dependent variable for this response propensity model is Y=1 if the eligible sample person was a respondent in 2017 and Y=0 otherwise. The independent variables were the predicted values of the selected survey outcomes for numeric variables (BMI, WKSEMPPY, WKSEMPPPY, MOCED and FACED) and predicted probabilities for each category for the categorical variables (TAMS and EDSTAT). The estimated parameters and standard errors for this logistic model are reported in Table 1. For example, the results indicate that the odds of response were significant higher for those with higher chance of were never been married, not cohabiting.

To reduce variation in response propensity weights and lower the reliance on correct model specification of the logistic regression, 10 nonresponse adjustment classes were created based on deciles of the predicted response probability (propensity score stratification; Little and Rubin, 2002 estimated using the logistic model in Table 4. The inverse of the mean response probability for TAS-2017 eligible sample cases in each decile was assigned as the nonresponse adjustment factor for that weighting class. The final nonresponse-adjusted weight for TAS-2017 respondents was computed as the product of their base weights (2017 PSID Individual Longitudinal Weights adjusted by eligibility) and their weighting class nonresponse adjustment factor.

Calibration

As the final step in weight development, the nonresponse-adjusted weights are used as input in a calibration adjustment , in which the TAS-2017 sample weighted distribution are matched to population totals estimated from the ACS 2017 1-year PUMS data for individuals between 18 and 28 years by December 31st, 2017 on selected auxiliary variables. Similar to the nonresponse adjustment, if such variables are predictive of the survey outcomes, this calibration adjustment can reduce non-sampling biases (such as coverage and nonresponse) and improve the precision of the survey estimates. For this reason, we tested the main and interaction effects on the seven selected survey outcomes of the following variables:

• Sex (Male, Female)
• Race/Ethnicity (Hispanic, Non-Hispanic While Alone, Non-Hispanic Black or African American, Non-Hispanic Asian alone/AIAN/NHPI/Some other race alone, Non-Hispanic Two or more races)
• Family type and employment status (Married-couple family: Husband and wife in Labor Force; Married-couple family: Husband in labor force, wife not in Labor Force; Married-couple family: Husband not in Labor Force, wife in Labor Force; Married-couple family: Neither husband nor wife in Labor Force; Male householder, no wife present, in Labor Force; Male householder, no wife present, not in Labor Force; Female householder, no husband present, in Labor Force; Female householder, no husband present, not in Labor Force;)
• Region (Northeast, Midwest, South, West)
• Age (17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28)
• Household size (1, 2, 3, 4 or more)

Due to the high dimensionality that would have been created if we accounted for all the interaction terms, we decided to test only the two-way interactions. We then kept all the main effects (regardless of their level of explanation on the survey outcomes) and only the two-way interactions that were significant, at a 5% level, to at least four of the survey outcomes, which on this case were:

• Sex by Race/Ethnicity
• Sex by Family type and employment status
• Sex by Region

In order to avoid undue increase in the variability of the weights, the following calibration cells with small sample sizes were collapsed for the calibration procedure:

• Gender by Married-couple family: Neither husband nor wife in Labor Force (Family type and employment status)
• Gender by Male householder, no wife present, not in Labor Force (Family type and employment status)

The calibration adjustment was performed using a raking ratio (or iterative proportional fitting) method (Deming and Stephan, 1940 through a SAS macro developed by Izrael, Battaglia and Frankel, 2009. An advantage of this SAS implementation is that apart from running the raking procedure to adjust the weights to enforce the weighted sample distribution to match the population margins in the selected calibration dimensions, it also simultaneously trim the weights according to trimming parameters, in order to mitigate the increase of the sampling variance due to the weight variability. The final cross-sectional weight for TAS-2017 respondents was derived from the output weights of this calibration adjustment with trimming. Table 5 reports key summary statistics for the final TAS-2017 cross-sectional weight.

Table 5: Summary Statistics for the TAS-2017 Sample Weights
Description	Value
N	2,526
Minimum	394.34
Maximum	88,725.97
Mean	19,716.88
Standard Deviation	21,805.82

The TAS-2017 cross-sectional weight is stored in the variable TA171987.

TAS-2017 Longitudinal Weights

To account for differential probabilities of selection due to the original PSID sample design and subsequent attrition since CDS-I, the TAS-2017 data are provided with a longitudinal weight for original CDS-I participants. The construction of this TAS-2017 longitudinal weight is described in this section.

Sample Transition from CDS-I to TAS-2017

Most of the TAS-2017 respondents were all originally selected for CDS-I in 1997 when they were 0-8 years of age. CDS-I selected sample children between the ages of 0-12 in 1997, and those who were aged 9-12 in 1997 aged out of the TAS age eligible range by 2017. Of the 3,563 children who participated in the original 1997 CDS-I interview, 2,268 were projected to be eligible for participation in TAS-2017, based on their participation in the CDS, TAS, and Core PSID studies. For these 2,268 cases, Table 6 summarizes the final contact and interview dispositions in TAS-2017.

Among the projected eligible sample, which excludes deceased (n=7) and non-sample individuals (n=71), a total of 1,410 interviews were completed, resulting in a cumulative unweighted response rate of 64.4% (i.e., 1,410/(1,410+780) =0.644). See Chapter 5 for a description of the TAS-2017 wave-specific response rate (87%) and data collection procedures and outcomes.

Table 6: TAS-2017 Sample Disposition
Sample Count	Description
2,268	Total projected eligible
1,410	Completed TAS-2017 interview
780	Non-response
577	Non-response before the 2017 interview
172	Non-response in 2017
31	Difficult to access/outside of the U.S.
78	No longer eligible
71	Not a sample person
7	Deceased

Methodological Approach

Sample survey data are typically provided with weights designed to compensate for unequal probabilities of sample selection and non-response or data that is missing at random (MAR; Little and Rubin, 2002). These weights are inversely proportional to the probability that each observation is selected and, conditional on selection, that individuals respond to the survey questions. With longitudinal data, this joint probability at time t, where the study has started at t-1 or earlier, can be expressed as the following

P(St=1)=P(St-1=1)*P(Rt=1|St-1=1), (1)

where St is an indicator of participation in the study at time t and Rt is an indicator of response at time t. Under this quasi-random model of the survey response process, the probability of being a participant at time t is the product of the probability of participating in the previous period and the conditional “probability” of responding in the current period. Because the first term on the right-hand side of Equation (1) is proportional to the reciprocal of the weight in the previous period, the weight in the current period is a product of the weight in the previous period and the inverse of the probability of response (the second term on the right hand side of Equation (1)). We will refer to 1/ P(Rt=1|St-1=1) as the attrition adjustment factor.

To reduce variation in response propensity weights and lower the reliance on correct model specification of the logistic regression, nonresponse adjustment classes are created by grouping the probability of response (propensity score stratification; Little and Rubin, 20025) and then the inverse of the mean predicted probability of response of each adjustment class is used as the nonresponse adjustment factor for that class.

TAS-2017 Individual Longitudinal Weight for Original CDS-I participants

The TAS-2017 individual longitudinal weight for original CDS-I participants was designed to account for the differential attrition between the baseline CDS-I (in 1997) and TAS-2017, i.e. t=2017 and t-1=1997. Thus, the TAS-2017 longitudinal weight is a product of the CDS-I weight, i.e., the individual-level primary caregiver/child weight (stored in the weight variable named ‘CH97PRWT’), and the attrition adjustment factor.

To obtain the attrition adjustment classes, the probability that a sample person was nonresponse in TAS-2017 was estimated using a logistic regression model. The dependent variable for this nonresponse propensity model is Y=1 if the eligible sample person was a nonrespondent in 2017 and Y=0 if they were a respondent. The estimated parameters and standard errors for the logistic model of nonresponse attrition are reported in Table 7. For example, the results indicate that the odds of attrition between 1997 and 2017 were significantly higher among males as compared to females, white respondents as compared to non-white respondents, those in Northeast and South regions as compared to the West region, and significantly lower among older respondents, SRC sample as compared to non-SRC sample, and those whose reference person was male as compared to female reference persons, holding all else equal.

For the TAS-2017 attrition adjustment for original CDS-I participants, 10 nonresponse weighting classes were defined based on deciles of the predicted probability of CDS-I to TAS-2017 attrition estimated using the logistic model in Table 3. The inverse of the mean response probability for TAS-2017 eligible sample cases in each decile was assigned as the nonresponse adjustment factor for that weighting class. The final longitudinal weight for TAS-2017 respondents who were original CDS-I participants was constructed as the product of their CDS-I base weight and their weighting class nonresponse adjustment factor.

As the final step in weight development, the newly constructed TAS-2017 longitudinal weight was trimmed to reduce the influence of extreme weight values on the variances of sample estimates of population statistics. The cases with the weight values in the top one percent and in the bottom one percent of the weight distribution were assigned values corresponding to the 99th and 1st percentiles of the weight distribution, respectively. Table 8 reports key summary statistics for the final TAS-2017 longitudinal weight for original CDS-I participants.

Table 8: Summary Statistics for the TAS-2017 Sample Weights
Description	Value
N	1,410
Minimum	1.27
Maximum	88.34
Mean	21.70
Standard Deviation	18.96

To examine the properties of the TAS-2017 longitudinal weight, we compared weighted estimates for selected demographic, geographic, and socio-economic variables in the CDS-I data computed in two ways. The first set of estimates is based on the full CDS-I sub-sample that remained eligible for TAS-2017. The CDS-I weight was used to create these estimates for the full TAS-2017 sample. The second set of estimates is based only on the TAS-2017 respondent cases and employs the TAS-2017 longitudinal weight that adjusts for longitudinal nonresponse among the eligible cases in the TAS-2017 wave of data collection. The results are provided in Table 9 and show that the distributions of the selected characteristics are similar in the appropriately-weighted TAS-2017 eligible sub-sample of CDS-I original respondents and in the TAS-2017 interview sample, suggesting that the attrition adjustment for the TAS-2017 weight compensates for potential attrition bias for variables included in the analysis. It is important to note, however, that this comparison does not necessarily rule out the possibility of selection bias associated with other characteristics of the respondents.