Families: What Do We Know?
reasons exist for caring about what happened in New Jersey under the Family
Development Program (FDP), which was implemented in October 1992.
The program is New Jersey’s effort to experiment with welfare reform
under Section 1115 of the Social Security Act. At issue is the capacity of
public policy to alter fundamental behavior through economic incentives.
Whether this behavior could be shaped by moral incentives as well was not
deliberately tested, although this prospect cannot be dismissed.
The provision of greatest interest is that which precludes a
recipient of Aid to Families with Dependent Children (as it was then termed)
from receiving additional cash benefits for a child conceived while the recipient was on welfare, resulting in a
loss of cash benefits totaling $102 per month for the second child and $64 per
month for any additional children. The
impact of this alteration seems modest, because numerous other benefits for
the child remained intact, and alternative benefits to the parent were
instituted that served to offset the direct loss of cash benefits. If the point was to test the effect of an economic message,
the “signal” was hardly robust.
Nevertheless, a material incentive was applied to fertility
decisions. Gaining a clear picture of reproductive behavior and its
determinants is arduous under the best of circumstances. Reproduction is not
only private and subject to multiple layers of rationalization, it also is
subject to multiple intentions, forces, and evasions that are both individual
and social in character.
Whether reproduction is determined (or at least conditioned in its
timing, frequency, and ultimate disposition) by a rational calculus sufficient
for it to be apprehended by an econometric analysis is contestable.
Becoming pregnant is, needless to say, a complicated enterprise,
involving in its most common form at least two actors and originating in a
social interaction not always volitional and rarely free from pressures beyond
the material. Moreover, the subsequent disposition of the reproductive
product may take several turnings, only some of which are driven by
deliberation, and only some of which result in a live child.
Were fertility decisions to be in no measure amenable to rational
shaping, public policy would be forced to confront the disarray of
increasingly unconnected thrusts and parries between citizens and the state,
which seeks to apply sanctions. There are, nevertheless, grounds for regarding
an economic approach to the matter as potentially insightful. Importantly,
evidence does exist concerning whether and to what extent reproduction is
“rational” and may be conditioned by incentive pressures. Hence, if a
cost-benefit calculus enters into this complex dynamic, it may well be
possible to condition reproductive behavior by imposing gradients of
self-interest through the instrument of public policy.
Given that public policy decision makers may well find the New
Jersey results instructive, it seems prudent to stress for a moment the
impediments to our hope for conclusiveness.
Social science, though inherently inexact, may be arranged into
evaluative formats of various rigor and lucidity, a randomized case-control
study with true experimental design being regarded as a highly exacting
approach. A similar method was
originally (but unsuccessfully) brought to bear on the New Jersey question.
Eventually, two separate efforts were made to evaluate the effectiveness of
the program: A randomized experiment (with a treatment group subject to the
new provisions and a control) and a multivariate statistical analysis
constructed as a pre- and post-evaluation of the study population (Camasso
1998a,b). The researchers had to resort to the less methodologically precise
multivariate analysis because of manifest limitations in the implementation of
the original experimental design.
Such studies can yield insight under any number of minimal
methodological requirements. To disconfirm with confidence the null
hypothesis, two groups of well-matched subjects must exist, such that
potential confounders are held constant across the divide that segregates the
groups. The experimental group receiving the treatment and the matched control
group from which treatment is withheld must be maintained with a clear
distinction. Attrition must be
minimized, and other alterations in the composition of the groups must be
severely restricted or controlled for. If
the study is to provide a measure of change in conscious behavior under
treatment, then the experimental format must be clearly communicated to
participants, and they should be sequestered from influence external to the
treatment itself. Clearly,
material incentives cannot affect people who do not know
what the incentives are, nor can the features of behavioral
modification be understood if the conditions and consequences of that behavior
are ambiguous or overdetermined.
Moreover, even if the data do reveal disparities in the direction
sought (either between the experimental and control groups under one model or
between a pre- and post- evaluation of a population under the other),
isolating the specific cause requires careful analysis to determine the
particular effect attributed to the treatment itself.
Data derived from the exercise must be analyzed by statistical
techniques appropriate to the problem set, and we must ensure that differences
between groups reach significance.
Even when such minimal conditions are met, there is no assurance
that a particular thesis demonstrates external validity, nor can one move with
assurance from association to causation.
The best that can be said is that results were obtained that were not
inconsistent with a certain hypothesis and that the statistical analysis shows
that the results were not caused by random error under a certain probability.
That is, successfully carrying out all of the above steps satisfies the
minimally necessary, but by no means sufficient, conditions for being able to
declare an advance in our social knowledge.
Inevitably, claims regarding the New Jersey research will find
their way to the media and to policy decision makers. These claims should be
regarded with caution. In the absence of successfully satisfying the minimally
necessary methodological conditions,
however, to proclaim any finding would be to threaten the commission of a Type I
error. A Type I error, which concludes that an association exists—in this
case, between a treatment and a change in fertility—when in fact it does not,
should be regarded as a scientific failing more pernicious than its counterpart,
the Type II error, which is failing to note an actually existing association. An
improperly postulated association can mislead all subsequent investigation and
obscure genuine causes and effects, which might otherwise be pursued were it not
for the false lead.
We can now ask more specifically, what do we know with certainty
about the New Jersey outcome? Unfortunately,
little; by the minimal criteria outlined above, the New Jersey experiment was
deficient in nearly every regard. What
we seem to have is an inconclusive addendum to an already inconclusive
literature regarding welfare, fertility, and behavioral change.
As the critique by Rossi (see chapter X) and an earlier analysis
performed by the Congressional Research Service (CRS; Falk and Devere 1998)
ruefully conclude, both research designs brought to bear on the New
Jersey data failed to maintain fidelity to their internal principles. The randomized experiment suffered from contamination in the
separation of the treatment group and the control group, whereas the
multivariate statistical analysis was unable to purge artifacts of statistical
modeling (if not incompatibilities between the form of the data and the
capabilities of the analytic instruments).
In any event, one may question whether family-cap proposals have
been fairly tested. In circumstances in which the minimal methodological
conditions are not met, as we appear to have in the New Jersey situation, the
result is worse than mere confusion. We
risk an active misleading of understanding, a condition to which we should apply
Wittgenstein’s remonstrance found in the preface to the Tractatus:
“Whereof one cannot speak, thereof one must be silent.” (Wittgenstein 1922).
Limitations notwithstanding, lessons can be salvaged from the
attempt. We observe that fertility
rates and sexual behavior rates declined nationwide during the study period,
independent of experimental intervention. Moreover,
trends in welfare participation and fertility preceded
in many instances the implementation of policy measures and are not
adequately explained simply by association with robust economic growth (Rector
and Youssef 1999). It appears that social expectations, moral remonstrances, and
intangible cultural atmospherics all exercised some effect.
Declines in fertility seem affected both by contraceptive changes and by
a diminished number of sexual performances on the part of males as well as
females. (It is odd that the
analysis to date has neglected the behavior of males as actors in the fertility
arena.) The suggestion is that the
meaning of fertility behavior was being reformulated in certain populations.
If so, the literature offers support for a new approach.
The role of caseworkers in the New Jersey experiment should not be
overlooked. They provided, in some
instances, faulty information or used inappropriate discretion in assignment to
experimental groups, contaminating the experimental outcome.
This flaw suggests a wider challenge for reform efforts.
Welfare as an institution should be envisioned as both a policy
and a structured set of social relationships through which that policy is
realized. Under any particular
policy there will be specific terms and procedures, requirements, expectations,
and stipulations, all of which may be thought of as a content that may indeed
shift over time. The structure, on the other hand, is a specific kind of social
relationship or set of relationships nested within each other. That is, there
are regularized and ongoing social interactions in a welfare relationship that
are both the outcome of particular policies as well as the a priori frames
through which policies are realized. Thus, what New Jersey may well have
demonstrated is that changing the one while leaving the other unaffected is
futile. In particular, it appears
that “reforming” welfare policy as a content while leaving intact the
sedimented structures of the previous welfare relationships impedes reformation.
It therefore seems valuable to examine lessons from a separate
reform program constructed around home visitation by nurses, where we may view
the interactions as reflecting not only a distinctive policy but a novel
structure of relationships as well. As Loury argues (see chapter X), “of all
the programs reviewed in this chapter, home visitation by nurses is the only
effort that showed consistently significant success at reducing subsequent
births to participating welfare mothers.”
What worked? In
keeping with the approach of “human ecology” advanced by Olds and Kitzman
(1993), nurse visitors are encouraged to develop an empathic relationship with
the mother and other family members. The
qualitative aspects of the relationship with the visiting nurse differ from
those characteristic of a social welfare worker.
In Loury’s words, “the intervention is more intrusive, more
directive, and more unequivocal in the value judgments being communicated.
The authority of the health professional is invoked on behalf of the
clearly stated end of avoiding a repeat pregnancy . . . . The home-visitation
programs provided a greater number of unambiguous, normative messages that
becoming pregnant again is not desirable. . . . The directive approach says,
‘You shouldn’t have another baby, and here are ways to prevent it’”
As Loury concludes, “a pedagogic function in public
policy—showing citizens how to lead their lives better —may need to be
invoked.” Such an approach need not be exclusive of other incentives and
sanctions, nor should the qualitative and cultural dimensions of the welfare
relationship be dismissed.
Camasso, M. J.; Harvey,
C.; Jagannathan, R.; and Killingsworth, M. 1998a. A final report on the impact of New Jersey’s Family Development
Program. New Brunswick, NJ: Rutgers University.
Camasso, M. J.;
Harvey, C.; Jagannathan, R.; and Killingsworth, M. 1998b. A final report on the impact of New Jersey’s Family Development
Program. Results from a pre-post
analysis of AFDC case heads from 1990 to 1996. New Brunswick, NJ: Rutgers
Falk, G., and
Devere, C. 1998. Analysis of evaluations
of the New Jersey Family Development Program. Washington, DC: Congressional
Olds, D., and
Kitzman, H. 1993. Review of research on home visiting for pregnant women and
parent of young children. Home Visiting 3: 53-92.
Rector, R., and
Youssef, S. E. 1999. The Determinants of
Welfare Caseload Decline. Washington, DC: Heritage Foundation.
1922. Tractatus Logico-Philosophicus, London: Routledge & Kegan Paul.
Back to top