The New Chance Observational Study is an in-depth examination of parenting behavior in 290 of the 2,322 families studied in the New Chance Demonstration.

Sample selected according to eligibility criteria: 16-22 years old; first gave birth age 19 or younger; not pregnant at time of enrollment; high school dropout (94% of sample) or high school graduate/holder of GED with reading scores below ninth-grade level (6% of sample); receiving ADFC (95% of sample) or economically disadvantaged according to Job Training Partnership Act (5% of sample); ability to speak, read, and understand English.

Random sample of 1,553 (67% of sample) in program group (who were allowed to enroll in New Chance) and 769 (33% of sample) in control group (who did not have access to New Chance services but may find alternate services in their communities).

"The quality of child care at the on-site day care centers was generally congruent with child care experts’ guidelines; moreover, the care was of higher quality than that typically provided by child care centers serving primarily low-income families"(xxxi).

"Phase II activities proved more difficult to implement, and were less uniform across sites, than the Phase I components"(xxxi).

"The majority of enrollees (89%) took part in one or more Phase I activities. 65% of GED earners participated in skills training or a work internship, the principal Phase II activities, but only 25% of the non-GED earners did so"(xxxii).

"Women in the experimental group received more services than women in the control group during the 18 months of follow-up. However, a very high percentage of the control group had participated in various activities, especially education programs"(xxxii).

Interim Cost-Benefit Findings:

"Sponsor agencies spent an average of $5,073 per experimental, excluding child care costs, operating New Chance. Child care costs amounted to an additional $2,573 per experimental"(xxxi).

Interim Impact Findings:

"By the 18-month point, a higher percentage of women in the experimental group than in the control group had obtained a GED certificate, and a higher percentage had earned credits toward a college degree"(xxxiii).

"The experimental and control groups had similar average scores on a test of literacy administered at the 18-month interview"(xxxvi).

"Women in the experimental and control groups had comparably high rates of births during the follow-up period. However, the experimental group reported a higher rate of pregnancies and a higher rate of abortions"(xxxvi).

"Women in the experimental group were more likely than those in the control group to be living with a partner or husband at follow-up, while women in the control group were more likely than women in the experimental group to the living with a parent or grandparent"(xxxviii).

"The program’s impact on subsequent pregnancy occurred only in conjunction with co-residence with a partner or husband at the time of the 18-month follow-up"(xxxviii).

"The program’s impact on GED attainment was sustained even among those with a post-random assignment pregnancy"(xxxix).

"With respect to the measured health outcomes for the mothers, there were no program effects"(xxxix).

"Levels of depression and stress were comparable among the experimental and control group women at follow-up. However, experimental group women were at an advantage with respect to two indicators of social support"(xxix).

"The home environments of children of experimental and control group members were largely similar, but children of experimental group members were living in home environments that were more emotionally supportive. Mothers in the experimental group also reported less authoritarian child-rearing attitudes"(x1).

"Children of experimental group members were more likely than children on control group members to have been in a non-maternal child care arrangement after random assignment, and were especially more likely to have used center-based child care. Children of experimental group members were also more likely to have been in a regular child care arrangement prior to age 1"(x1).

"Children’s health outcome were mostly comparable in the two groups"(x1i).

"Control group women were more likely than experimental group women to have been employed in the first few months after random assignment, but employment rates for the two groups converged after time. More than 40 percent of each group had been employed at some point during the follow-up period" (x1i).

"Control group members average higher cumulative earnings than experimental group members over the 18 months of follow-up. Over 80% of the women in each of the research groups were on welfare throughout the 18 months of follow-up; there were no substantial program impacts on ADFC receipt"(x1iii).

"Experimental group members were more likely than control group members to have been involved in a "skill building" activity (defined as employment or being in school or training), but the two groups were equally likely to be in such an activity at the end of the follow-up period"(x1iii).

“Mothers in the New Chance Observational Study sample were at particularly high risk in terms of parenting behavior.”

“Despite their similar economic circumstances and backgrounds, the mothers in the New Chance Observational Study showed variation in their parenting behaviors, and parenting behaviors were meaningfully related to the mothers’ background characteristics.”

“The New Chance Program was able to bring about positive changes in parenting behavior, even in a population burdened by economic stress and other serious difficulties.”

“Positive program impacts were found across parenting measures obtained in several different ways and from different informants.”

“Parenting behavior was an important predictor of specific child outcomes in this sample, as were variables reflecting maternal psychological well-being and the families’ larger social context.”

“Modest improvements in parenting behavior, in this context, did not suffice to bring about positive program impacts on child outcomes.”

“Observational measures add to the understanding of parenting behavior within this sample in multiple ways.”

“At the time of the 3½-year interview, the young women were, on average 22.4 years old, and most had children who were still toddlers. Contrary to the common stereotype of these young mothers as immobilized by—or content with—their circumstances, the evaluation found that over the 3½-year follow-up period the young women in the research sample—experimental and control group members alike—were moving forward in many ways. At baseline (that is, random assignment), fewer than 10 percent of sample members had a high school diploma or a GED; by the 3½-year point, almost half the sample had earned one of these credentials. Sixty-three percent of sample members did not work at all during the year prior to random assignment; in contrast, over half were employed at some point during the 12 months before the 3½-year interviews, and the large majority of those who worked did so for 30 hours a week or more. These rates of employment are surprisingly high given the young age of the mothers and the fact that most had very young children. Over the follow-up period, the proportion of sample members receiving AFDC dropped considerably (although the majority were still on the rolls at the 3½-year interview), the proportion of women who used a reliable method of birth control rose steadily, and fewer women were at risk of depression. Nevertheless, the large majority remained poor and on welfare after 3½ years” (3).

“Although experimental group members received more varied services in greater quantity than did their control group counterparts and received them sooner, the differential was not large, especially with regard to education- and employment-related services. This is partly because during the period of the demonstration many education and training programs were available in the New Chance communities, and members of the control group participated in these in unexpectedly high numbers. At the same time, because of absenteeism and early departures from the program, members of the experimental group received on average a much lower intensity and duration of services than had been anticipated, and many never participated in skills training, work experience, or job search—the activities in the program model most closely related to employment” (3).

“The New Chance evaluation is not, therefore, a test of extensive services compared with no services or minimal ones. Rather, the evaluation measures the effectiveness of a particular mix and level of services that were relatively easy for those in the experimental group to obtain against another mix and level of services that individuals in the control group could secure only if they displayed somewhat greater initiative” (3).

“The findings indicate that while experimental and control group members both advanced in many ways, experimental group members did not advance further than control group members in most respects. New Chance did boost participants' levels of GED receipt above those of the control group. The added services provided by the program, however, did not help participants secure skills training credentials, get and maintain employment, or reduce their rates of welfare receipt or subsequent childbearing relative to outcomes for control group members. The program did not improve their children's preschool readiness scores, and it had unexpected small but negative effects on participants' emotional well-being and their ratings of their children's behavior” (3).

“These results are puzzling, for MDRC observers judged all the sites to offer some high-quality services, and the large majority of young women in the experimental group said that they liked the program and benefited from it. It is likely that many factors, sometimes working in combination, account for the absence of impacts and for unanticipated impacts; different explanations may hold for different outcome areas. The possible factors include the slender differential in service receipt between experimental and control group members, the low absolute amount of services received by those in the experimental group, the possibility that some direct program effects produced unanticipated side-effects, and constraints on the magnitude of impacts imposed by larger social and environmental forces. It may also be that the program model itself was inappropriate for many young women” (4).

“These findings, unfortunately, are consistent with the results of other evaluations of programs serving young mothers on welfare who do not have a high school diploma or a GED, and the unsuccessful records of these programs highlight the importance of continuing to seek effective ways to assist these young women in improving their lives. But the impact results do not mean that the services New Chance provided (and that control group members received on their own) were of no value. Additional analyses were conducted to estimate the effects of service receipt for experimental and control group members together. While less definitive than the analyses under girding the impact estimates, the results suggest that young women who received more than 18 weeks of education were far more likely to earn GEDs than those who did not and that young women who received skills training and attended college earned higher wages than their counterparts who did not receive post-secondary education or training. These findings held true even after other differences between those who received more or fewer weeks of education, and those who attended training or college and those who did not, were controlled statistically” (4).

“Thus, the findings indicate that the combination and quantity of services that New Chance participants received, on average, did not result in improved outcomes vis-à-vis those achieved by control group members. But they also suggest that receiving adequate amounts of specific kinds of services can make a difference in the mothers' lives—a finding of considerable importance to program operators and policy makers” (4).

“The major impact of New Chance on observational measures of affective and behavioral qualities of mother-child relationships was that mothers assigned to the New Chance experimental group, in comparison with those in the control group, showed less coercive and punitive behavior toward their children during the Teaching Task session. This finding was true as an overall difference between the groups as well as within a wide range of baseline subgroups. It seems, therefore, that New Chance did reduce harsh treatment of children by their mothers”(16).

“There were no overall differences between the New Chance experimental group and the control group on more positive observational indices of interaction such as Mother’s Supportive Presence or Quality of Relationship. We did see significant differences within some of the subgroups that may have been indices of lower risk at baseline, such as having only one child at the time of random assignment. Within these lower-risk subgroups, mothers in the experimental group were rated higher on Quality of Relationship than were control group mothers. Although this finding must be interpreted cautiously due to the lack of an overall New Chance impact, it does suggest that mothers who were at lower risk at baseline may have benefited from New Chance not only through the reduction of punitive and coercive parenting behaviors, but also through an increase in positive parenting behaviors”(17).

“One of the major goals of the comprehensive New Chance Program was to have a positive impact on the children. We did not, however, find any overall child impacts at the time of the observational session. In fact, baseline subgroup analyses suggest that for some dyads, particularly those where the mothers faced substantial interpersonal demands at baseline, children of mothers in the New Chance experimental group showed less compliance than children of mothers in the control group. Again, subgroup findings must be interpreted cautiously due to the lack of an overall impact of the New Chance program on Child’s Compliance. This finding suggests that for some families the additional demands of the New Chance Program may have disrupted aspects of the mother-child relationship. Previous research has suggested that multiple maternal stressors may contribute to child behavior problems in the preschool years (Richman, Stevenson, and Graham, 1982) and in the elementary school years (Pianta, Egeland, & Sroufe, 1990). For a small group of families, it is possible that the intensive nature of the New Chance Program compounded the demands in their already stressful lives”(17).

“We found significant continuity of adaptation over time in the Observational Study sample, and within the New Chance experimental group in particular. Our analyses demonstrated that there is continuity of adaptation from the 21 month observation to the 42 month follow-up, with elements of the mother-child interaction predicting both later maternal reports of child behavior and teacher reports of child behavior. Our analyses also established that there is some evidence of significant continuity of adaptation within the New Chance experimental group but not within the control group. Because there are no actual intervention/control differences in child behavior at 42 months, we interpret this finding to indicate that what has changed is not child behavior, but the way in which mothers in the New Chance group perceive and report on their children’s behavior. This does not mean that the intervention effects on maternal report should be discounted as a methodological artifact. Changes in mothers’ perceptions of their children’s behavior may eventually lead to more sensitive interactions, and consequent changes in the relationships. This may be a small step toward improving mother-child relationships”(18).

“The overarching substantive consideration in asking whether to include observational work within this portfolio is the "value added" of the observational data. Such work could be undertaken as part of a freestanding study, in which all participants provide both self-reported data, perhaps enhanced by interviewer observations, and observational data. Or observational work could be embedded within a larger survey effort, as was the case in this study. In considering the embedded model, the "analytical leverage" provided by the observational data is a consideration. If the subsample participating in the observational component of the study is selected randomly, it is possible to generalize to the larger sample, and/or to the population from which it was drawn, through statistical weighting procedures. Other potential analytical strategies include using the richness of the observational measures to help in interpreting data from other sources, including assessing how much confidence to place in other measures that are potentially subject to different method effects. Finally, observational measures from a subsample may be used to augment quantitative survey findings.”

Researchers, however, should not expect to find a single decision rule for determining when observational research within a survey model is warranted. Every study is different. Survey-based observational work makes sense when it serves well-articulated analytical objectives and is conceptualized as an integral part of an overall research design. If resources are available and the potential added value of observational measures is clear for a particular study, the observational method is a valuable adjunct to more conventional survey approaches.”