|
In 1995, as part of a third-party
evaluation of OOR, it was decided to compare two cohorts of OOR
completers with control groups, using the computerized records of what
is now known as the Board of Probation and Parole for control group
information and for status verification of both the experimental and
control groups. The results of this study, reported in the final report
of the evaluation in 1995, gave strong indications of program impact on
recidivism. It was decided, therefore, to conduct a longer-range study
to confirm the outcomes over a longer time frame and provide a broader
base for conclusions and projections of the cost-effectiveness of the
OOR program. This study, which covers the release-years of 1990 through
1994, is the result of that decision.
Methodology
There are two major issues related to
evaluation of correctional education; how to measure recidivism, and how
to be sure that the results can be attributed to the program under
study. To answer the first of these, we have the PBPP Research and
Statistics Division computer files, which provide either master or
history file information on each releases. To answer the second, we
developed a "blind" technique for creating a control group
closely matched to the experimental (OOR) group for each year. Groups of
matching size and characteristics were created, and their outcomes
analyzed for significance using chi-square.
Outcomes
The experimental (OOR) groups
outperformed the control groups for each of the five years, by varying
amounts. For the combined cohorts, the chi-square was zero, indicating a
100% chance of validity. For three of the five years, the reading was
under .05, the accepted scientific standard for proof of validity. For
the other two years, it yielded probabilities of about 14/1 and about
6/1 that the results were statistically significant. Percentage
differences ranged from 11% in 1992 to 19% in 1993 and 1994, with an
average percentage difference, always in favor of the experimental
group, of 16% per year.
Conclusions
The numbers tell the story of a clear
difference in outcomes between OOR participants and a carefully-selected
matching group with no known confounding variables. Sixteen of every
hundred participants are more successful three to seven years after
release than their counterparts, with no observed distinctions by race,
age, type of crime, or release locations. With an average cohort size
(the number actually found in the file for each year) of 64.2, this
translates into almost exactly 10 actual persons each year whose futures
are affected positively by the program.
Projections
Based on figures provided by
OOR from PA statistics, the average cost of incarceration in the state
for the period from 1990- 1997 is estimated at $23,496 and rising at the
rate of about 5% per year. One way to project the value of OOR would be
to multiply this average by the total of 250 inmate-years
"saved" since 1990 (10 x 7 plus 10 x 6 plus 10 x 5 etc.). The
savings, just since 1990 and just for the incarceration cost, would
total $5,874,000, and 1.6 times the total state money (about 3.7
million) used by OOR during the period. It is reasonable to conclude
that the social costs in Pennsylvania are at least equal to the
incarceration cost and probably substantially larger. If we use the
one-to-one ratio, however, the seven-year savings jump to about $11.7
million, or over three times (3.18) what the state has spent on the OOR
program during that period. |
