As of June 1995, the federal government had granted to 28 states waivers from current federal laws and regulations, allowing states to test a wide range of welfare reform approaches. This study, conducted by the National Center for Children in Poverty, examined the evaluation plans of 21 waiver experiments in 17 states, focusing on the impact of welfare reform on family income, parenting stress behaviors, and children's access to services. For each impact study, recipients of Aid to Families with Dependent Children were randomly assigned to either an experimental group participating in the welfare reform approach or a control group. The major findings indicated that: (1) evaluators are collecting a great range of child-related data, but few variables were common among many states, reducing the chances for cross-state learning to inform national policy; (2) some waiver evaluations may lead to new information regarding children's education, particularly for states evaluating "learnfare" approaches; (3) some evaluations track potential harms to children related to family dysfunction, foster care placement, and substance abuse; (4) the waiver evaluations will not be able to demonstrate the ways welfare reform affects children's growth and development because very few variables directly measure such changes; and (5) variables in some evaluation plans related to children's services, but few related to quality services or to outcome measures indicating the effectiveness of those services, especially for child care and child health. (Appendices contain background information on selected states, research questions, and children's well-being variables. Contains 24 references.) (KB)