Mothers' Transitions from Welfare to Work and the Well-Being of Preschoolers and Adolescents

R. Blank, R. Haskins, Eds. The New World of Welfare (Brookings Institution, Washington, DC, 2002).

G. J. Duncan, R. E. Dunifon, M. B. Ward Doran, W. J. Yeung, in For Better and For Worse: Welfare Reform and the Well-Being of Children and Families, G. J. Duncan, P. L. Chase-Lansdale, Eds. (Russell Sage, New York, 2001), pp. 103–131.

A. Kalil, R. E. Dunifon, S. K. Danziger, in For Better and For Worse: Welfare Reform and the Well-Being of Children and Families, G. J. Duncan, P. L. Chase-Lansdale, Eds. (Russell Sage, New York, 2001), pp. 154–178.

Hofferth S. L., Smith J., McLoyd V. C., Finkelstein J., J. Soc. Issues 56, 747 (2000).

Smith J. R., Brooks-Gunn J., Kohen D., McCarton C., Child Dev. 72, 1512 (2001).

Smith J. R., Brooks-Gunn J., Klebanov P. K., Kyunghee L., J. Marriage Fam. 62, 808 (2000).

I. Ku, R. D. Plotnick, paper presented at the Institute for Research on Poverty, Madison, WI, December 2000.

Coley R. L., Chase-Lansdale P. L., Soc. Serv. Rev. 74, 380 (2000).

E. H. Peters, N. C. Mullis, in Consequences of Growing Up Poor, G. J. Duncan, J. Brooks-Gunn, Eds. (Russell Sage Foundation, New York, 1997), pp. 340–381.

Bianchi S., Demography 37, 401 (2000).

Waldfogel J., Han W., Brooks-Gunn J., Demography 39, 369 (2002).

P. Morris, A. Huston, G. J. Duncan, D. Crosby, D. Bos, “How welfare and work policies affect children: A synthesis of research” (Manpower Demonstration Research Corporation, New York, 2001).

L. M. Gennetian et al., “How welfare and work policies for parents affect adolescents: A synthesis of Research” (Manpower Demonstration Research Corporation, New York, 2002).

R. Moffitt, M. Ver Ploeg, Eds., Evaluating Welfare Reform in an Era of Transition (National Academy Press, Washington, DC, 2002).

The other components include an ethnography and an embedded developmental study. See Winston et al., Welfare, Children, and Families: A Three-City Study, Overview and Design Report (Johns Hopkins University, Baltimore, MD, 1999), available at www.jhu.edu/∼welfare. That document also describes the welfare policies of the three states, which ranged considerably in their time limits, work requirements, diversion policies, earnings disregards, and benefit levels.

Materials and methods are available as supporting material on Science Online.

Data on 430 infants and toddlers are not included here, owing to different measurement.

J. P. Shonkoff, D. Phillips, Eds. From Neurons to Neighborhoods: The Science of Early Childhood Development (National Academy Press, Washington, DC, 2000).

Graber J., Brooks-Gunn J., Dev. Psych. 32, 768 (1996).

In our data, just under 10% of women were working while on welfare, as should be expected in the period after 1996 when many women have combined welfare and employment. But when we disaggregated by child age group and omitted the infants and toddlers as described in (17), the absolute number of women making transitions into and out of this category was too small for analysis. Therefore, we omitted women in this category.

An alternative regression formulation to measure change is the fixed effects model, which is similar to our model except that the dependent variable is measured as the change in child outcomes. We do not use that model because of its assumption that changes in child outcomes are the same regardless of their initial values, which also implies that welfare and employment transitions have instantaneous effects on child outcomes. To the contrary, changes in child outcomes are very likely to depend on the starting point, and to adjust gradually and with a lag.

Maccoby E. E., Ann. Rev. Psychol. 51, 1 (2000).

Our data indicate that there was very little short-term cycling on and off welfare or employment before the two interviews [e.g., R. Moffitt, J. Roff, The Diversity of Welfare Leavers, Working paper 00-01, Welfare, Children, and Families: A Three-City Study (Johns Hopkins University, Baltimore, MD, 2000)]. Therefore, we did not test directly for the effects of such cycling.

The weighted mean total sample size across our four model specifications in Table 1 is 493. The sample sizes for each of the four estimated transition effects in Table 1—each of which represents a comparison of outcomes for those who made a transition and those who did not—are as follows: for the into employment transition, n = 83 moved into employment and n = 279 stayed unemployed; for the out of employment transition, n = 33 moved out of employment and n = 99 stayed employed; for the onto welfare transition, n = 30 moved onto welfare and n = 301 stayed off welfare; and for the off welfare transition, n = 83 moved off welfare and n = 80 stayed on welfare.

The probability of obtaining 80 insignificant coefficients at the 5% level is quite low if all 80 are independent. However, the tests are positively correlated to some degree, both across columns because different outcome measures often move together, and across rows because the four different panels shown in Table 1 are estimated on almost the same sample and with very similar specifications. Determining the exact probability of observing no significant coefficients in these 80 tests would be difficult, but that probability is higher than it would be if all of the tests were fully independent.

The weighted mean total sample size across our four model specifications in Table 2 is 800. The sample sizes for each of the four estimated transition effects—each of which represents a comparison of outcomes for those who made a transition and those who did not—are as follows: for the into employment transition, n = 32 moved into employment and n = 402 stayed unemployed; for the out of employment transition, n = 63 moved out of employment and n = 204 stayed employed; for the onto welfare transition, n = 32 moved onto welfare and n = 572 stayed off welfare; and for the off welfare transition, n = 102 moved off welfare and n = 94 stayed on welfare.

As with Table 1, it is possible that the 16 significant coefficients could have occurred by chance at the 5% level. Although the 16 coefficients represent more than 10% of all the effects estimated, the tests are correlated, and this raises the probability of a chance occurrence.

G. Becker, A Treatise on the Family (Harvard Univ. Press, Cambridge, MA, 1991).

The results are not shown but are available upon request from the first author.

Chase-Lansdale P. L., Pittman L. D., Future Child. 12, 167 (2002).

McLoyd V. C., Jayaratne T. E., Ceballo R., Borquez J., Child Dev. 65, 562 (1994).

We gratefully acknowledge the support of the National Institute of Child Health and Human Development, Office of the Assistant Secretary for Planning and Evaluation, Administration on Developmental Disabilities, Administration for Children and Families, Social Security Administration, National Institute of Mental Health, Boston Foundation, Annie E. Casey Foundation, Edna McConnell Clark Foundation, Lloyd A. Fry Foundation, Hogg Foundation for Mental Health, Robert Wood Johnson Foundation, Joyce Foundation, Henry J. Kaiser Family Foundation, W. K. Kellogg Foundation, Kronkosky Charitable Foundation, John D. and Catherine T. MacArthur Foundation, Charles Stewart Mott Foundation, David and Lucile Packard Foundation, Searle Fund for Policy Research, and Woods Fund of Chicago. We thank the other principal investigators of the Three-City Study, R. Angel, L. Burton, and W. J. Wilson, as well as J. Quane, C. O'Muircheartaigh, and the members of the MacArthur Foundation Network on the Family and the Economy, for their helpful insights.