Abstract
Maternal employment's association with young school-age children's academic and behavioural outcomes was examined in a sample of families currently or recently receiving welfare. The sample comprised 1,197 African-American mothers and their preschool-age children who were randomly assigned either to the ‘human capital development’ (HCD) programme, the ‘labor force attachment’ (LFA) programme or to the control group in the Atlanta site of the Child Outcomes Study, a substudy of the National Evaluation of Welfare-to-Work Strategies (NEWWS). Once observed selection effects were controlled, maternal employment — regardless of whether obtained under mandatory or non-mandatory circumstances — was largely unrelated to contemporaneous measures of five to seven year olds’ academic school readiness and behaviour. There was one exception: maternal employment secured through Atlanta's HCD programme significantly predicted fewer and/or less frequent antisocial behaviours even after observed selection factors were controlled. Implications for policy and future research are discussed.
La corrélation de l'activité professionnelle de la mère avec les résultats scolaires et comportementaux de ses jeunes enfants à l’école a été étudiée sur un échantillon de familles recevant ou ayant récemment reçu des aides sociales (welfare). L’échantillon était composé de 1,197 mères noires américaines et de leurs enfants à l’école primaire, qui avaient été réparties au hazard soit dans le programme du ‘Human Capital Development’ (HCD), soit dans le programme ‘Labor Force Attachment’ (LFA), ou soit dans le groupe de contrôle du ‘Child Outcomes Study’ du site d'Atlanta, une sous-étude du ‘National Evaluation of Welfare-to-Work Strategies’ (NEWWS). Une fois les effets de la sélection observée pris en compte, l'activité professionnelle des mères — qu'elle ait été obtenue dans des circonstances obligatoires ou non-obligatoires — a été déterminée être sans relation directe avec les mesures, sur la même période, des résultats et du comportement scolaire de leurs enfants de 5 à 7 ans. Une exception s'est dégagée: l'activité professionnelle des mères obtenue par le programme HCD d'Atlanta a été déterminée être un prédicteur, de façon significative, de comportements anti-sociaux moins fréquents ou en nombre réduit, même une fois les facteurs de la sélection observée pris en compte. Les conséquences sur les politiques et recherches futures sont exposées.
We would like to thank the National Institute of Child Health and Human Development (NICHD) for supporting this research (1 R01 HD38762). We also want to thank officials at the US Department of Health and Human Services for granting permission to use data from the Child Outcomes Study (COS) prior to their public release. The views expressed in this paper are those of the authors alone and should not be construed to represent the official views of the US Department of Health and Human Services. The NEWWS and the COS were funded by the Office of the Assistant Secretary for Planning and Evaluation and by the Administration for Children and Families — both in the US Department of Health and Human Services — and by the US Department of Education. The NEWWS was conducted by MARC, and the COS was conducted by Child Trends, under subcontract to MDRC. An earlier version of this paper was presented at ‘From Welfare to Work: What Happens to Infants and Toddlers When Single Mothers Exchange Welfare for Work?’, Washington, DC, 17–18 May 2001. This conference was organized by Natasha Cabrera, PhD (University of Maryland), Robert Hutchens, PhD (Cornell University, Department of Labor Economics) and Elizabeth Peters, PhD (Cornell University, Department of Policy Analysis and Management) and was sponsored by Cornell University School of Industrial and Labor Relations Pierce Memorial Fund, Cornell University Institute for Labor Market Policies, The National Institute of Child Health and Human Development Family and Child Well-being Network, and the Russell Sage Foundation.
Notes
1. At the time the NEWWS was designed, the Code of Federal Regulations (CFR), Title 45, Part 46, ‘Protection of Human Subjects’ did not require sample members’ informed consent to be randomly assigned or to have their administrative records accessed by evaluators, by virtue of their receiving public assistance. However, all study procedures, including random assignment, were approved by the US Office of Management and Budget, which assumed an IRB-type role in federally funded data collection activities. CFR has since been revised to require IRB review and approval of all study design and data collection procedures in federally funded data collection activities.
2. For greater detail on the JOBS programme and the design of the NEWWS, see Hamilton, Brock, Farrell, Friedlander, and Harknett (Citation1997). For greater detail on the design of and findings from the COS, see Moore, Zaslow, Coiro, Miller, and Magenheim (Citation1995), McGroder, Zaslow, Moore, and LeMenestrel (Citation2000), and Hamilton et al. (Citation2001).
3. Except for the Grand Rapids sample, in which about one-third of focal children had younger siblings at study entry.
4. See McGroder et al. (Citation2000) for a detailed presentation of two-year impacts on children in the COS, and Hamilton et al. (Citation2001) for a detailed presentation of five-year impacts on children in NEWWS and for young children in the COS.
5. In an effort to be parsimonious in the number of selection variables included in these analyses, we examined the degree to which each hypothesized selection variable did, in fact, predict both employment and the child outcomes. Not every hypothesized selection variable significantly predicted both employment and the child outcomes; however, each hypothesized selection factor did significantly predict either employment or at least one of the child outcomes. In the interest of interpreting identical models, we chose to keep all hypothesized selection factors in each model. A test for multicollinearity (using the ‘/collin’ option on the PROC GLM command in SAS) gave no indication that multicollinearity was a problem.
6. There were no significant differences in two-year employment rates across the three research groups. McGroder et al. (Citation2000) reported that Atlanta's LFA programme increased employment in the month prior to the two-year survey by two percentage points over control group levels (though this impact was significant at only the p < 0.10 level). The slight discrepancy between this finding and the non-significant difference in the present study derives from the slightly different sample and different covariates used in the present analyses.