United States. Congress. House. Committee on the J.

Taking Back Our Streets Act of 1995 : hearings before the Subcommittee on Crime of the Committee on the Judiciary, House of Representatives, One Hundred Fourth Congress, first session, on H.R. 3 ... January 19 and 20, 1995 online

. (page 44 of 51)
Online LibraryUnited States. Congress. House. Committee on the JTaking Back Our Streets Act of 1995 : hearings before the Subcommittee on Crime of the Committee on the Judiciary, House of Representatives, One Hundred Fourth Congress, first session, on H.R. 3 ... January 19 and 20, 1995 → online text (page 44 of 51)
Font size
QR-code for this ebook

— Some 19,000 cases led to convictions.

— Under 10,000 of those ultimately convicted received non-custodial sentences (probation,
fines, etc.).

— About 9,500 of those ultimately convicted were sentenced to prison or jail.

— Most of those sentenced to prison or jail served only a fraction of their sentenced time
in coiifinement.

What such figures depict is a justice sorting svstem hard at work, with each set of actors
in the system— from citizens who are victimized by crime choosing whether or not to report it to
the police, to corrections officials deciding whether to recommend a prisoner for
parole-exercising discretion.

Conceptually, the main question is not the extent to which given laws accovmt either for
the growth in the correctior^s population or for the change in the composition of that population
(e.g., the steady increase in the fraction of the population that consists of drug offenders). Nor, for
that matter, is the key question whether given population trends are likely to continue.

Rather, the key question is how well is the svstem sorting, and according to what correctional
supervision and cost-effectiveness criteria!

Try the following thought experiment. You just arrived in America. You have neither a
personal stake nor any intellectual interest in the evolution of the nation's penal laws. In this
capacity, all you would really want to know is how many and what types of offenders are
currently under what types of correctional supervision. For example, are the prisons crowded
with non-violent offenders without any prior criminal records? Are the intensive supervision
programs handling all the offenders they can safely and cost-effectively handle? Whether behind
bars or in the community, do offenders receive the drug treatment they require?


As such hard-to-answer questions imply, sorting is a means of calculating tradeoffs and
allocating the system's limited human and financial resources in a way that is (or aspires to be)
sodcdlv optimal . An cuialogy to health care is helpful. Last year we heard a great deal of debate
about the nation's health-cetre system, which now consumes nearly 16% of Gross National
Product (GNP). Despite all this spending, in the course of a year some thirty-seven million
Americans will at some point be without either catastrophic or routine health insurance coverage.
Many sick people are iU-treated, under-treated, or not treated at all. All the various plans for
remedying the situation and enhancing cost-effectiveness were essentially alterriative strategies for
changing the way that the system sorts present and potential patients.

By the same token, most penal reform proposals can be understood as alternative
strategies for changing the way that the system sorts present and potential clients. Each strategy
embodies a di^erent set of tradeoff, some of them easier to specify, quantify, and calculate than


The New Consensus on Crime Policy:
Planting the Seeds /4

For example, imagine that, possessed by the ghost of Draco, we decided that in any given
year we wanted to catch and "throw the book at" every criminal in America. Imagine that this
desire translated into a fixed prison term of, say, ten years, with absolutely no exceptions, no
pleas, no "good time," no early release, and no parole. And imagine that the long arm of the law
grew long enough to achieve our goal.

What tradeoffs might be embodied in this approach? There might be social benefits in the
form of millions of crimes averted through simple incapacitation effects. And, emotionally, our
Draconian souls might fee! satisfied. But the social costs of detecting and incarcerating tens of
millions of criminals would be astronomical.

Obviously, America has chosen a different sorting strategy, one that embodies a much less
Draconian tradeoff between public protection and the legitimate desire for retribution, on the one
hand, and the drain on the public's purse, on the other . Again, most crimes are neither detected
nor punished (let alone punished with long, fixed prison terms) by the justice system. This has
been true throughout the recorded history of the nation's justice system, in "get-tough" periods as
well as in "go-easy" ones.

Presumably, while few Americams are possessed by the spirit of Draco, neither are mciny
of them advocating the thoroughgoing decriminalization of ail or most crimes, from first-degree
murder to spitting on the sidewalk. Conceptually, therefore, the key thing to understand is that
most penal reform options can affect how present and potential clients of the system are sorted,
and with what social consequences, only on the marg ins.

To illustrate this point about making social tradeoffs on the margins, consider plea
bargaining. Plea bargaining remains unpopular with the general public. The process is little
understood outside of professional or academic criminal justice circles. It is rarJced among the
"seamy" aspects of contemporary justice systems.

Viewed differently, however, plea bargaining is a sorting device that weighs the estimated
social costs of expending scarce human and financial resources on prosecuting a given offender,
on the one hand, against the estimated social benefits of doing so, on the other. There are, to be
sure, some perverse and unintended consequences of plea bargaining: some offenders get
"under-prosecuted," others get "over-prosecuted." By and large, however, plea bargaining can be
viewed as a potentially efficient (if not often an efficient-looking) way of sorting. The plea process
hcis become thoroughly institutionalized (if not favorably publicized and promoted). Were the
process to stop suddenly, there is absolutely no way that either the courts or the correctional
institutions and programs could avoid being flooded with even more cases than they now handle.

Many people, including not a few prosecutors, make both principled (e.g., just desserts)
jmd practical (e.g., public safety) objections to plea bargaining. But there is a gap between what
people say they want (e.g., everyone prosecuted "to the full extent of the law" and "maximum
public safety") and what people's actions over time say about what they want. Economists use the
term "revealed preferences" to refer to the preferences embodied in how people actually behave
as opposed to what people say their preferences are. With respect to criminal justice, sodety's
revealed preferences are clear: we do not want to detect and punish, let alone "throw the book
at," each and every person who commits whatever sort of crime. To do so would cost more in
human and financial resources than most people would actually be willing to spend.


The New Consensus on Crime Policy:
Planting the Seeds /5

Thus, plea bargaining is a widely-practiced sorting strategy. It hinees on the exercise of
prosecutorial discretion and embodies social tradeoffs on the margins . If every potential criminal
defendant were a cold-blooded murderer, then the typical marginal offender would be one whom
most of us would want prosecuted with no legal holds barred. And, indeed, the entire class of
such offenders is normally prosecuted fuUy. But once all the cold-blooded murderers and other
"clear-cut" cases have been taken in hand, the marginal offenders— those offenders regarding
whom the decision of whether or not to prosecute, on what charges, and how vigorously remains
an "open" or "close" call (and hence a somewhat risky use of prosecutorial discretion)— are likely
to be property offenders, small-time drug dealers, and certain classes of situationally violent but
not extremely dangerous persons.

Likewise, about 72% of the over 5 million persons under correctional supervision on any
given day in this country are on probation or parole . Obviously, having offenders in the
community means that some of them will commit crimes which could have been averted by
incarceration. But the social benefits of community-based corrections may or may not outweigh
the social costs . Again, the real issue concerns sorting offenders and finding the optimal sanction
for the marginal offenders.

Reasonable hearts and minds are bound to differ on what sorts of sodal tradeoffs are best.
But reasonable people cannot differ over whether tradeoffs must be made: every
sanctioning/sorting option involves both social costs and sodal benefits, every option has
diminishing marginal returns, and every option needs to be assessed and administered in
deference to the best available estimates of its marginal value .

Honest proponents of any given sentencing option are not free intellectually to calculate
its benefits but ignore its costs, present and potential. Honest opponents of any given option are
not free intellectually to calculate its costs but ignore its benefits, present and potential. This holds
equally for expanding (or contracting) prison capadty, expanding (or contracting) intermediate
sanctions, expanding (or contracting) drug treatment, and every other option that one can name
or imagine . Politically, ideologically, and rhetorically, tradeoffs can be avoided, masked, or
buried. Conceptually, however, they can only be acknowledged, weighed, and evaluated.

In sum, most criminal justice sorting strategies are not inherently good or bad, and the
actors in the system can use their discretion well or ill, reflectively or reflexively, in self-interest or
in the public interest. In order to arrive at a reasonable conclusion about the marginal value of
any given strategy, one needs to know a great deal about the characteristics of the present and
potential clients being sorted, the experience of other jurisdictior^s that have tried such strategies,
ttnd other factors relevant to specifying, quantifying, and calculating tradeoffs . Unfortunately,
reliable knowledge about the sodal costs and benefits of competing sentencing policy strategies
remains scarce.


It is even harder to find reliable information that would enable one to evaluate competing
crime polides in terms of their effectiveness.

Too often, the terms "effectiveness" or "cost-effectiveness" are used as synonyms for each
other and for "effidency." But there are some important conceptual lines here that ought not to be


The New Consensus on Crime Poliqr:
Planting the Seeds /6

blurred, and some important conceptual issues related to conventional justice system effectiveness
measures, mainly recidivism rates and crime rates, that ought not to be ignored.

By definition, an effective policy or program is one that achieves its goals . There are two
types of goals: operational and non-operational. An operational goal is an image of a desired
future state of affairs that can be compared unambiguously to an actual or existing state of affairs.
A non-operational goal is an image of a desired future state of affairs that caimot be compared
unambiguously to an actual or existing state of affairs. For example, "rehabilitating offenders" is a
non-operational gocd; "getting all inmates classified as having drug problems into structured
treatment progran\s" is an operational goal.

A cost-effective policy or program is one that achieves operational goals with the "lowest
possible" doUar expenditures . For example, if progreim A achieves the same goals in the same
way that program B does, but achieves them for half the money, then A is more cost-effective

The concept of efficiency is broader than the concept of cost-effectiveness. An efficient
policy or program is one that maximizes desired outputs with a given set of valued inputs, or
that minimizes the valued inputs needed to achieve any given level of desired outputs . For
example, some types of staff deployment measures within a prison may be more efficient than
others because they minimize the human and financial resources necessary to achieve any given
set of operational goals (e.g., no escapes, no assaults, reduction in need for overtime pay, etc.).
Likewise, some types of prisoner classification methods may be more efficient than others because
they minimize the human and financial resources necessary to provide care and custody to the
sentenced population (e.g., not making unnecessary use of scarce and expensive
maximum-security beds to hold prisoners whose institutional and post-institutional behavior
would be virtually the same were they held in medium- or even minimum-security facilities).

Within sentencing policy studies, recidivism rates and crime rates have been the dominant
effectiveness measures. But there are at least two major sets of problems with these measures .

Recidivism rates are measures of the proportion of offenders who, either during or
following some period of correctional supervision, return to crime and get either rearrested, or
reconvicted, or both. There is no one "best" recidivism measure . And any honest look at
recidivism measures breeds an appreciation for the conceptual and empirical complexities

Take a hypothetical group of offenders. A year after their release from supervision, X
percent are rearrested, and .5X are reconvicted. But over half are reconvicted for crimes less
serious than the ones for which they were previously convicted. Two years out, 1.5X of the group
have been rearrested, but still only .5X have been reconvicted. And three years out, 1.6X have
been rearrested, and .6X have been reconvicted, but all the additional reconvictions between yeiirs
one and three are for crimes less serious than the ones for which they were previously convicted,
and less serious on average than the crimes for which the earlier recidivists in the group were

What is the "best" or "truest" measure of recidivism for this group, and how, if at all, can
one relate it back to the custodial or supervisory experiences (conditions of confinement.


The New Consensus on Crime Policy:
Planting the Seeds /7

participation in this or that program, quality or intensity of community-based supervision, etc.) of
the group's members? There is no obvious answer, either theoretically or in most real-world

Also, there is something faintly ridiculous about measuring the efficacy of sentencine
policies or programs in terms of their long-term post-release or post-supervision effects . A
popular textbook on penology makes this point as follows: "Schools do not follow up their
graduates to see if they slip back into ignorance or fail to hold a job after leaving school. Yet, in
prison treatment programs, success is based on some post-program assessment of behavior.'" As
a number of analysts have noted, similar problems attend any attempt to measure the
performance of law enforcement policies or programs via such conventional measures as crime
rates, arrest rates, and emergency response times .' As the BJS-Princeton Study Group concluded:
"Few police officers believe that their work solely determines crime rates in their jurisdiction. Few
corrections officials believe that what they do chiefly determines recidivism rates. Likewise, most
criminal court judges, prosecutors, pubUc defenders, and other justice practitioners know from
experience that the prevalence and severity of crime depend mainly on factors affecting
individuals long before most are taken into custody ... To evaluate the performance of police
departments, correctional agencies, and other key components of the justice system exclusively in
terms of crime rates and recidivism rates may cause observers to overlook other important
contributions of the system's day-to-day performance."^

Whether given policies or programs are judged successful depends on what types of
effectiveness, cost-effectiveness, or efficiency measures are used in evaluating them . For example,
judged solely by recidivism measures, the best experimental studies cast question marks over the
social value of intensive supervision programs. But judged according to other criteria, and viewed
as steps on an intermediate-sanctions ladder, the programs are not dear-cut failures.*

'Richard Hawkins and Geoffrey Alpert, American Prison Systems (Englewood CMk, NJ:
Prentice-Hall, 1989), p. 198.

George L. Kelling, "Measuring What Matters: A New Way of Thinking About Crime and
Public Disorder," The City Journal (Spring 1992), p. 21-33.

^ohn J. Dilulio, Jr., Rethii\king the Criminal Justice System: Toward a New Paradipm
(Washington, DC: U.S. Bureau of Justice Statistics, December 1992), pp. 1, 4.

*Susan Turner and Joan PetersUia,'Tocusing on High-Risk Parolees: An Experiment to Reduce
Commitments to the Texas Department of Corrections," Journal of Reseiirch in Crime and
Delinquency, vol. 29, no. 1, February 1992, pp. 34-61; Joan Petersilia and Susan Turner, Intensive
Supervision for High-Risk Probationers: Findings from Three CaUfomia Experiments (Santa
Moruca, CA: Rand Corporation, December 1990); Joan Petersilia, "Community Corrections:
Measuring Success," paper for BJS-Princeton series, February 1993.


The New Consensus on Crime Policy:
Planting the Seeds /8


Much of the applied research or\ justice issues, especially the so-called program evaluation
literature, relies heavily on statistical analysis. Fundamentally, statistics is about drawing
inferences about populations from samples . As a rule of thumb, the more technically complicated
the statistical analysis of any given policy or program option, the less likely it is that the analysis
compares "apples to apples."

Indeed, the reason for many statistical techniques is that they enable one to make
meaningful compcirisons among and between "apples eind oranges," or to draw meaningful
ii^ferences about what a very small number of "apples " tells us about "all apples." Also, much of
statistical analysis is based on the assumption that "all other things are equal." But in policy
studies, ceteris are hardly ever paribus, and many of the things which the good analyst "holds
constant" for the purposes of arriving at some conclusion are things which in fact vary in complex

The beauty of experimental research and randomized research designs is that they
eliminate much, though by no means all, of the need for complicated statistical gymnastics, and
render conclusions that come close to "speaking for themselves." But very little of the research on
crime policy is experimented.

Public Philosophy

Even if there were a huge body of experimental research on each and every crime policy
issue of the day, that would not ensure agreement, either among scholars or among policymakers
and average citizens, about what policies ought to be adopted, how they should be funded, or
how best to administer them. Views on crime policy are mediated by competing public
philosophies. The "facts" never speak for themselves; competing public philosophies sort and
organize them.

Broadly speaking, positions on many crime policy issues are reflections or extensions of
views on the public philosophy of representative democracy. The public philosophy of
representative democracy has three main parts. First is the belief that, while average citizens lack
the time, energy, information, and interest to decide well on pubic policy, the people can make
reasonable choices among competing leaders or groups of leaders. Second is the idea that
representative institutions shoiild be structured in waj^ that make them politically and legally
resistant to temporary popular majorities but resporisive to persistent popular majorities. And
third is the hope that the size of the American republic, the variety of economic interests, the
diversity of religious opinions, the device of staggered elections, and other constitutional
contrivances will render persistent popular majorities rare and unlikely to threaten either
individual liberty or public order. In the language of Federalist Paper No. 10, the hope is that,
whatever the issue around which they form, persistent popular majorities virill transcend partial,
immoderate, or fleeting ("factional") interests and embody a true concern for "the permanent and
aggregate interests of the community," both national and local, both public and private.

There are thus at least three ways to evaluate, say, a crime policy such as Virginia's
no-parole policy. The first two are critiques from within the representative democracy perspective,
while the third is a critique from the direct democracy perspective. First, one can say that the


The New Consensus on Crime Policy:
Planting the Seeds /9

"proper guardians of the public weal" have abdicated their responsibility to "refine and enlarge
the public's views" by adopting a get-tough strategy which, whatever its actual social costs and
benefits, is calculated mainly to appease popular pressures. Second, one can say that the majority
sentiment in favor of a no-parole policy is that of a persistent popular majority which has been
unduly frustrated by lax parole policies in the past and are only belatedly getting their moral,
political, and constitutional due from government. Finally, one can protest that the issue has yet
to be presented directly to the voters who have both the constitutional writ and the requisite
understanding to make such choices without any filtering by elected officials, pro-incarceration or
anti-incarceration lobbyists, or other policy elites (including academic analysts).

The policy conclusions and implications one draws from any given piece of research on
crime vidll be affected by one's first-order assumptions, explicit or tacit, conscious or unconscious,
out front or buried, about the pubUc philosophy of representative democracy.


Likewise, there can be no escape from the first-order assumptions about free will,
determinism, causality, punishment, and personal responsibility that underlie any position on any
crime policy. Just as ^ere can be no "data-driven" view of genocide (e.g., the fact that the Nazis
killed millions of Jews), so can there be no "sLmply-coerced-by-the-data" position on, say, whether
persons convicted of a violent offense should be eligible for parole.

For example, the latest BJS data indicate that in 1992 most of the 431,279 admissions to
state prison were members of a radal minority, men under age 30, and entering directly from
court. About 75% of them were entering for a nonviolent offense. But murderers comprised 2% of
new court commitments, and some of these murderers were on parole at the time they committed
their latest crime.

What are the right policy implications of such data? Among other things, the answer
depends on how you understand the moral drama of race in American history; on your moral
concept of justice (egs., treat "like cases alike"-but "alike" how?; give people "what they
deserve"— but what of "mercy" and "mercy" being "more just than justice"?); on your moral
reasoning with respect to what degree of risk society ought to assume vis-a-vis persons who have
"offended" in the past ("violated life, liberty, and property in the past"?); and much, much more.

In particular, policy analysts are trained to look at the big picture, find the patterns, and
discount or ignore the outliers. But, morally, the "outliers" may be the kid who goes to prison
because of an overtly racist sentencing regime, or the lives that were cut short because of the tiny
fraction of aU parolees who murder. There is a moral argument behind the "racial politics" of
those who push for a Radal Justice Act and a moral argument behind the "victim politics" of
those who push for truth-tn-sentendng or no parole. The "data" indude the outliers, and they do
not tell how, morally, one ought to evaluate given polides, weigh "total" social costs and benefits,
or factor in uncertainties about who might win or lose the most from any particular policy.
Terming such arguments "ideological" or "sensationalistic" does not steal their moral force on the
one side or on the other. These moral arguments can be rejected or embraced, but they need to be
surfaced and confronted, not swept under anyone's favorite pseudo-sdentific rug with a broom of
"sdentific objectivity." Dedding what counts as "the data" and how to organize and interpret "the
facts" is unavoidably cm enterprise that involves moral reasoning and moral judgement.





Retiirniim to the Crime Scene

^^B cveral yean ago. in these pages, we cned co referee an acnmomous debace bc-
^^^^ cween cnminologists who insisted chat prisons "cost too much" and those who
^m responded that they "protect too Uttle." Our contention was that both sides of

Online LibraryUnited States. Congress. House. Committee on the JTaking Back Our Streets Act of 1995 : hearings before the Subcommittee on Crime of the Committee on the Judiciary, House of Representatives, One Hundred Fourth Congress, first session, on H.R. 3 ... January 19 and 20, 1995 → online text (page 44 of 51)