Research note: A critique of Potas and Walker's “Sentencing the federal drug offender” — some statistical and methodological issues
| Published date | 01 December 1983 |
| Date | 01 December 1983 |
| DOI | 10.1177/000486908301600410 |
| Author | Austin Lovegrove |
| Subject Matter | Original Articles |
248 (1983) 16
ANZJ
CRIM
RESEARCH NOTE: A CRITIQUE OF POTAS AND WALKER'S
"SENTENCING
THE
FEDERAL DRUG OFFENDER" -
SOME STATISTICAL AND METHODOLOGICAL ISSUES
Austin Lovegrove*
The
principal aim of this project (Potas and Walker, 1983) is to examine
the
feasibility of deriving sentencing guidelines empirically. Since it is now clear
that
some lawyers acknowledge there is unjustifiable disparity in sentencing and are
contemplating the use of more detailed sentencing statistics as a remedial measure
(eg, Law Reform Commission, 1980), this research project is extremely timely and
important.
The
authors set themselves the tasks
of
developing ageneral methodology for
deriving sentencing guidelines by empirically relating case facts and sentences
and
for measuring the extent
of
disparity.
The
analysis was restricted to 90 persons who
since the beginning of 1978 had been convicted under s
233B
of the Customs
Act
1901 (Cth) of importing or possessing heroin and as a consequence
had
been
sentenced to imprisonment. However, the authors' intent is to develop a
methodology applicable, with minor modifications, to the study of sentencing
offenders convicted of any offence. Unfortunately,
the
authors' analysis has serious
statistical flaws and, consequently, their major conclusions cannot be regarded as
valid.
Moreover, before researchers present their own analysis it is necessary
that
they
review past research on relevant matters. In this instance, it is essential
that
past
criticisms of sentencing guidelines are answered and
that
the use of guidelines in
Australia is justified.
Both
of these aspects of the report require revision. I shall deal with each
matter
in turn.
Sentencing Guidelines
In
order
to explain the technical
error
which undermines the validity of this
research it is necessary to give a generalized account of the critical points of
the
methodology.
1.
The
authors coded information on case (offence and offender) characteristics
taken
from transcripts of the sentencing judgments. This source of information was
supplemented by
data
collected on offender characteristics for
another
project. It
appears
that
the former source included
"hard"
data (eg, criminal record)
and
"soft"
data
(eg, whether the judge thought the offence was premeditated). Factors
which were statistically or intuitively related were combined, and those factors
which were rarely cited during sentencing were omitted.
The
resulting 28variables,
and several additional variables, were factor analysed and nine factors were
extracted.
2. Multiple regression was used to determine the combination of factors yielding
the
best prediction
of
sentence;
the
nine factors
were
able to
account
for
about
30%
of
the
variance.
*Criminology
Department,
University of Melbourne.
To continue reading
Request your trial