Ted G. Goertzel, Ph.D.

Professor of Sociology, Rutgers University

311 North 5th Street, Camden NJ 08102

office:(856) 225-6013; home office: (609) 953-1670; fax: (413) 793-2597

email: goertzel@camden.rutgers.edu

homepage: http://crab.rutgers.edu/~goertel

New Jersey's Family Development Plan (Goertzel and Hart, 1995), which denied a cash increase to mothers who conceived an additional child while on welfare, has been highly controversial since it was passed in 1992. Opposition came from three distinct groups: abortion opponents, feminists and welfare rights advocates. These groups have sometimes offered conflicting arguments, claiming both that the program would be ineffective in influencing birth rates and that it would cause an increase in abortions by mothers who could not afford to have another child. They were united, however, in the view that the reform would be harmful to poor women and children in New Jersey. It was passed, despite their opposition, thanks to effective advocacy by its sponsor, then state Assemblyman (now Senator) Wayne Bryant, an African-American democrat representing the city of Camden.

Policy makers and advocates, on both the state and the federal level, were eager to have evaluation results to bolster their arguments. A large-scale evaluation was funded, using both experimental and time series methods. The early results showed no difference between experimental and control groups, but later results showed some statistically significant differences. There were serious problems, however, with the integrity of the implementation of the experimental design which made those results less than definitive. The time series research, on the other hand, showed that trends in New Jersey were quite favorable to the reform: welfare case loads and births to welfare mothers were down. Indicators of child welfare, such as infant mortality and low birth rate infants, were positive. For a review of this research see Goertzel (1998).

As soon as the first recipients were effected by the "family cap," opponents filed lawsuits to attempt to have it overturned. The first lawsuit, filed in federal court, used the novel argument that the reform was an experiment and that therefore women should have to give informed consent. Federal law, however, explicitly exempts this kind of evaluation of changes in benefit programs from informed consent provisions. The lawsuit in federal court was dismissed with prejudice by the federal judge, who also dismissed a number of constitutional arguments. The opponents, however, reserved the right to bring a lawsuit in New Jersey state courts arguing that the reform violated privacy rights inherent in the state constitution. A central feature of that lawsuit has been the allegation that the abortion rate has increased as a result of the reform, presumably because women are forced to have abortions out of increased economic hardship. This claim has been based on the results of an evaluation study conducted by Michael Camasso and his associates at the Department of Social Work at Rutgers University in New Brunswick.

This lawsuit is still ongoing, and Peter Rossi and I have been serving as expert witnesses for the State of New Jersey in this matter. We argue that the evidence does not show an increase in the abortion rate as a result of the reform. I also filed an affidavit responding to claims made in affidavits filed by expert witnesses for the complainants in this lawsuit. I found the reports by these witnesses unconvincing. They relied upon their own expert judgments, although many of them were physicians without any expertise in evaluating social programs. None of them, not even the one sociologist (Kathryn Eden), took the trouble to examine relevant data which was available on trends in the well-being of poor families in New Jersey since the Family Development Act was implemented in 1993. Even when they wrote in 1997, there was a good deal of data available that should have led them to be much more cautious in their statements. When I wrote my affidavit in April, 1999, there were more data available and it was quite clear that the dire predictions they made in their affidavits simply had not come true.

They also failed to make use of data available from focus group interviews and sample surveys such as those which my colleagues and I conducted at the Forum for Policy Research at Rutgers University at Camden in 1993, 1995 and 1996 (listed on my Vita on my WEB site http://crab.rutgers.edu/~goertzel) and the one done by the Rutgers New Brunswick researchers under contract to the New Jersey Department of Human Services (Camasso, et al., 1998, pp. 77; a preliminary version of the same report was available in 1997).

Their affidavits relied largely on their expert judgment, but even in 1997 expert opinion about the likely consequences of welfare reform measures was much more divided than they acknowledged. By 1999, their arguments were seriously out of date because they failed to incorporate the lessons which welfare specialists had learned from the nationwide welfare reforms passed in 1996. The nation's leading experts in welfare research were no longer making sweeping predictions such as those in these affidavits, because the data simply had not borne out predictions of this sort made in 1996. The affidavits also showed little or no familiarity with the public policy process in New Jersey which led the authors to make unwarranted and inaccurate statements about the intent and implementation of welfare policy in New Jersey. This led them to policy recommendations which I believed to be inadvisable.

This dispute raises a number of interesting issues, both in the interpretation of statistical data and in the use of social science findings in public policy lawsuits. It has been my observation that the lawyers have a very limited understanding of statistical techniques and social science research methods. In depositions, much time is spent on issues that social scientists would regard as insignificant. The lawyers are trained to analyze texts, so they search through the documents in search of any turn of phrase that might be used to advance their case. Their primary goal is to discredit the witnesses on the other side by questioning their motives or by finding contradictions in their writings. The opposition lawyer carefully scoured all of my relevant publications, apparently looking for anywhere where I might have said something different from what I said in my affidavit. The examples she found, however, were useless because they referred to different groups or had no relevance to the case at hand. Unlike social scientists, the lawyers make no pretense of seeking objective, scientific truth. It simply never occurs to them. The legal emphasizes winning one's case over other values. In this case, they wasted a great deal of time seeking evidence of a conspiracy on the part of the State of New Jersey to cover up the true impact of the welfare reforms.

Since the economic and health indicators were strongly positive, the opponents fell back on the claim that the welfare reform had caused an increase in the abortion rate. Legally, they hoped to make the case that women were being unconstitutionally forced to have abortions because of the lack of an additional $64 in their cash grant. The time series data on abortions to welfare mother, however, do not bear out their claim. The following table shows
 
 

the quarterly abortion rates for AFDC mothers in New Jersey. These data come from the Camasso reports (1998). Just looking at the table, it is apparent that there was no significant change in the abortion rate before and after the FDP. Camasso and his colleagues, however, did a sophisticated multivariate analysis that concluded that there were more abortions after the FDP than there would have been if it had not been past. The essence of their argument was that the trend was down before the FDP and that it would have continued down without the reform.

I do not believe that the numbers justify this conclusion, nor did Peter Rossi who filed an affidavit and gave a deposition in the case. I think Camasso's results depend entirely on which period one uses as a bases for projecting the "counterfactuals." In the graph above, I have plotted two linear projections. One is based on the quarterly data from June 1991 to March 1993, the other on the quarterly data from March 1991 to March 1993. The difference is only one data point, but the result is entirely different. Using the first, the trend was up, which would imply that the welfare reform cut the abortion rate. Using the second, the trend was down, which would imply that the welfare reform increased the abortion rate. When a minor change in the data leads to such a sharp change in the results, the only accurate conclusion is that the data are not sufficient for us to know what the abortion rate would have been had there been no welfare reform. Social scientists have a terrible record as soothsayers, it is simply wrong to claim that we know what social trends would have been under different conditions.

Indeed, on a fundamental level, I believe the logic of multiple regression is suspect here. Regression analysis is used to "hold everything constant" except the independent and dependent variables. I call this the ceteris paribus fallacy - the fallacy of assuming, for purely methodological reasons, that "everything else would have stayed the same." In the real world, everything else doesn't stay the same, and policy makers know it. They enact policies based on their intuitive sense of the trends of the times, and they are judged by their results. For a social scientist to say, "my ideas would have worked if conditions had not changed," is no better than a physician saying, "my operation was a success but the patient died." In this case, the welfare reforms have worked remarkably well. Using convoluted statistical manipulations to deny this fact is grasping at straws.

It is impossible to predict the ultimate results of this lawsuit. It will probably go to the New Jersey Supreme Court, and the litigants are undoubtedly hoping that the justices will vote their ideology instead of the evidence. Of one thing we can be certain, however. By the time this case is decided, it will be behind the times. The family cap was important for a few years, from 1993 to 1996. After 1996, the federally mandated time limits eclipsed it as an issue. Under current policy, few mothers will remain on welfare long enough to be impacted by the family cap, and those that do will have much bigger things to worry about. The social science research supplied in this case is also rapidly going out of date. The experts for the litigants relied on conclusions from studies done before 1996, when the world was very different. The Camasso evaluation of the Family Development Plan was discontinued a year early by the Department of Human Services because conditions had changed so much that it was no longer relevant. New Jersey now has a "Work First" plan which mandates a rapid transition from welfare to work, and the Department is appropriately doing research aimed at facilitating that program.

Ironically, the case might have to be dropped because the women the advocates recruited to serve as the litigants-of-record in this case are rapidly improving, or at least changing, their lives. Several of them can no longer be found by the lawyers who are so eager to "represent" them. Several of the others are doing quite well in the transition from welfare to work. The legal process is simply too slow to keep up with the lives of New Jersey's rapidly progressing welfare mothers and former welfare mothers.
 
 

References:
 
 

Michael Camasso, et. al, "Final Report on the Impact of New Jersey's Family Development Program: Results from a Pre-Post Analysis of AFDC Case Heads from 1990-1996," (1968). A report submitted to the New Jersey Department of Human Services.
 
 

Goertzel, Ted (1998). "Why Welfare Research Fails," paper presented at the August, 1998 meetings of the Eastern Evaluation Research Society. Available online at: http://crab.rutgers.edu/~goertzel/fail2.html.
 
 

Goertzel, Ted and John Hart (1995). "The Politics of Welfare Reform in New Jersey," in The Politics of Welfare Reform (Donald Norris, ed.), SAGE Publications.