Sentencing Guidelines Reanalysis

1
Sentencing Guidelines Reanalysis

• Overview of Methodology

2
Sentencing Guidelines Reanalysis 2002

• At its March 2002 meeting, the Commission
  approved a work plan for conducting a thorough
  reanalysis of the current sentencing guidelines.
• By examining sentencing practices under the
  truth-in-sentencing/no-parole system, the
  reanalysis will provide a more focused picture of
  Virginias experiences since the abolition of
  parole.
• Reanalysis will capture sentencing events from
  fiscal years (FY) 1997 through 2001.

3
Data for Reanalysis
Total Number of Cases 126,533
Supplemental data averages 23 of total cases per
year
4
Development of Methodology

• General methodological approach has been in place
  since 1987.
• Methodology was approved by the judiciary for
  development of Virginias first discretionary
  sentencing guidelines.
• Judicial Sentencing Guidelines Committee of the
  Judicial Conference of Virginia oversaw the
  development of the early guidelines.
• Judges approved the concept of discretionary
  guidelines that were descriptive of historical
  sentencing practices.
• Guidelines should be grounded in the historical
  incarceration rate.
• Guidelines ranges should represent the middle 50
  of historical sentences (extreme low and high
  sentences excluded).

5
Development of Methodology

• Judges felt that the most recent five years of
  available data would most accurately capture
  current judicial thinking.
• Using the five years of data minimizes
  year-to-year fluctuations and reduces the
  likelihood of spurious results in the sentencing
  models.
• Models were developed for each offense group by
  type of judicial sentencing decision.
• Prison in/out decision
• Probation/jail decision
• Prison sentence length decision
• Judiciary reviewed and approved the statistical
  techniques utilized by guidelines staff to model
  sentencing practices.
• The early sentencing guidelines system was
  developed by judges for judges.

6
Reanalysis Past and Present

• Prior to 1995, reanalysis was performed using the
  accepted methodology to periodically update the
  guidelines based on the most recent five years of
  data.
• This has not been done under truth-in-sentencing
  because five of years of sentencing data under
  the new system has only recently become
  available.

7
Historical Example Drugs
Total number of cases 24,901
8
Modeling the Prison In/Out Decision

• Logistic regression
• A statistical technique used to identify factors
  that best discriminate between two groups (e.g.,
  offenders sentenced to prison and offenders not
  sentenced to prison).
• Analyst can easily determine which factors are
  statistically significant.
• Results are in terms of the log of the odds
  (e.g., the odds of winning the state lottery)
• Interpretation of the effect of each factor is
  complex.
• Discriminant analysis
• A statistical technique used to identify factors
  that best discriminate among two or more groups.
• Determining which factors are statistically
  significant is more difficult than with logistic
  regression.
• Interpretation of the effect of each factor is
  more straight forward than with logistic
  regression.

Used to identify significant factors
Used to determine the indepen-dent effect of each
factor
9
Historical Example Drugs
10
Historical Example Drugs
11
In/Out Model to Worksheet Scores

• Worksheet scores are developed from the weights
  of factors in the model.

• Factor weights tend to be small because the
  in/out model (Section A) simply determines if an
  offender will be recommended for prison or not.
• Factor weights are adjusted so that the smallest
  score value will be at least one point.
• This process is referred as standardizing.
• The relationships among the variables remain the
  same.
• After standardizing, the factor weights are used
  to develop worksheet scores.
• Example prior incarcerations

Rounded to 2 points
12
DRUG PRISON IN/OUT DECISION
Worksheet
Worksheet
A
A
Type of primary offense Add Score
Other than listed below 1 3 counts
. 1 4 or more counts
. 3 Possess Schedule I or II
drug 1 count . 0 2
counts 4 3 or more
counts . 7 Sell, etc., ½
ounce to 5 pounds of marijuana for profit Sell,
etc., marijuana to inmate for accommodation 1
count . 3 2 or more
counts . 7 Sell, etc.,
Schedule I or II drug for accommodation 1 count
. 4 2 or more counts
. 7 Manufacture marijuana
not for personal use (any number of counts)
. 6 Sell, etc., more than 5
pounds of marijuana for profit (any number of
counts) .. 12 Sell, etc.,
Schedule I or II drug for profit (any number of
counts) .. 12 Sell, etc.,
Schedule I, II, III, IV drug or marijuana to
minor at least 3 years younger (any number of
counts) .. 12
Additional offenses Years Add Score
(including counts) at conviction, less than
2 0 22 27 . 5 with
maximum penalties totaling 2 5
.. 1 28 32 . 6
6 11 2 33 38
. 7 12 16 .. 3
39 or more ... 8 17 21
.. 4
Prior adult convictions Years Add Score
with maximum penalties totaling less than 3
0 3 19 1 20
44 . 2 45 or more ... 3
Prior adult incarcerations If yes, add 2
If total is 11 or less, go to work sheet B. If
total is 12 or more, go to work sheet C.
Prior felony drug convictions
Number Add Score
1 .. 0 4
.... 3 2 . . 1
5 or more .. 4 3
... .. 2
Legally restrained at time of offense Add Score
None . 0 Other than
parole, supervised probation or CDI
2 Parole, supervised probation or CDI 6
TOTAL SCORE
If total is 11 or less, go to work sheet B. If
total is 12 or more, go to work sheet C.
13
Setting the Prison Threshold on Section A
Historical Prison In/Out Decision
Sentencing Guidelines
14
Modeling the Probation/Jail Decision

• This stage of analysis includes only those cases
  that did not result in a prison sentence.
• The same statistical methods used to model the
  prison in/out decision are used to model the
  probation/jail decision.

15
Historical Example Drugs
16
Historical Example Drugs
17
DRUG PROBATION/JAIL DECISION
Worksheet
Worksheet
B
B
Type of primary offense Add Score
Other than listed below (any number of counts)
. 1 Possess Schedule I or II
drug 1 count . 2 2 or
more counts ... 4 Sell, etc.,
½ ounce to 5 pounds of marijuana for
profit Sell, etc., marijuana to inmate for
accommodation 1 count .
7 2 or more counts . 10 Sell
, etc., Schedule I or II drug for accommodation
1 count . 8 2 or more
counts . 9
Additional offenses Years Add Score
(including counts) at conviction, less than
1 0 28 34 . 5 with
maximum penalties totaling 1 7
.. 1 35 41 . 6
8 14 2 42 or more
... 7 15 20 .. 3
21 27 .. 4
Prior adult convictions Years Add Score
with maximum penalties totaling less than 1
0 1 10 1 11
21 . 2 22 31
..... 3 32 42 . 4
43 or more 5
Prior adult incarcerations If yes, add 2
Prior criminal misdemeanors If yes, add 1
Legally restrained at time of offense Add Score
None . 0 Other than
parole, supervised probation or CDI
1 Parole, supervised probation or CDI
2
TOTAL SCORE
See DRUG PROBATION/JAIL TABLE to convert score to
guidelines sentence.
18
Setting the Jail Threshold on Section B
Historical Probation/Jail Decision
Sentencing Guidelines
19
Drug Probation/Jail Sentence Table
20
Modeling the Prison Sentence Length Decision

• This stage of analysis includes only those cases
  that received a prison sentence.
• Ordinary Least Squares (OLS) Regression
• A statistical technique used to identify factors
  (e.g., weapon use, victim injury, etc.) that
  influence a response measure (e.g., sentence
  length).
• OLS regression assumes a linear relationship
  between predictor factors and the response
  measure.
• Results are calculated by minimizing the models
  prediction error.
• Analyst can easily determine which factors are
  statistically significant.
• Interpretation of the effect of each variable is
  straight forward.

21
OLS Regression - Example
22
Historical Example Drugs
23
Historical Example Drugs
24
Sentence Length Model to Worksheet Scores

• Worksheet scores are developed from the weights
  of factors in the model.

• For the sentence length model (Section C), the
  score represents months of incarceration.
• Example additional offenses

Rounded to 1 3 5 9 18 28 37
25
DRUG PRISON LENGTH DECISION
Worksheet
Worksheet
C
C
Type of primary offense Add Score
Other than listed below (any number of counts)
. 30 Possess Schedule I or II
drug 1 count . 19 2
counts ... 29 3 or more
counts . 39 Sell, etc., ½
ounce to 5 pounds of marijuana for profit Sell,
etc., marijuana to inmate for accommodation 1
count . 26 2 counts
... 41 3 or more counts
. 49 Sell, etc., Schedule I
or II drug for accommodation 1 count
. 35 2 counts
... 47 3 or more counts
. 66 Manufacture marijuana
not for personal use (any number of counts)
. 41 Sell, etc., more than 5 pounds of
marijuana for profit 1 count
. 84 2 or more counts .
120 Sell, etc., Schedule I or II drug for
profit 1 count . 59 2
counts ... 85 3 counts
... 106 4 or more counts
. 143 Sell, etc., Schedule
I, II, III, IV drug or marijuana to minor at
least 3 years younger (any number of counts)
. 91
Additional offenses Years
Score Years Score
Add Score
(including counts) at conviction, less than
1 ... 0 5. 5 each
having a maximum penalty of 1.. 1
10... 9 2. 2
20... 18
3. 3 30... 28
4 4 40 or more 37
Prior adult convictions Years
Score Years
Score Add Score
each having a maximum penalty of less than 4
.. 0 30 ... 4
4, 5, 10 1 40 or more
..... 6 20 3
Prior felony drug convictions Number
Score Number
Score Add Score
See DRUG PRISON LENGTH TABLE for guidelines
sentence range.
1 .. 5 4
.... 19 2 .. 9
5 ..... 24 3
.... 14 6 or more .. 28
Prior felony person convictions Number
Score Number
Score Add Score
1 .. 5 3
.... 14 2 .. 9
4 or more .. 18
Firearm in possession at time of offense
If yes, add 20
Legally restrained at time of offense
Add Score
None 0
Supervised probation or CDI 13
Other than parole, supervised probation or CDI
7 Parole .. 46
TOTAL SCORE
26
Recommended Sentence Range
Prison Sentences for Offenders Convicted of
Selling a Schedule I/II Drug (No Prior
Record) (1987 1991)
Months
Historical Sentence Guidelines Range
Actual Prison Sentences
27
Drug Prison Sentence Length Table
Sentence Range Midpoint
Sentence Range
28
Reanalysis and Revisions

• Sentencing models developed through statistical
  analysis will be presented to the Commission as
  they are developed.
• Input from Commission members is vital to guide
  and shape analysis.
• Experience and expertise of Commission members
  will be valuable to analysts as models are
  explored.
• The Commission, to date, has not made
  prescriptive adjustments to the guidelines.
• Prescriptive adjustments have been mandated by
  the General Assembly.
• Midpoint enhancements
• Nonviolent offender risk assessment
• Sex offender risk assessment

29
Status Update

• Reanalysis of murder, robbery, rape and sexual
  assault offense groups is underway.
• Supplemental data collection for rape and sexual
  assault cases has begun.
• Offense detail from PSI narratives
• Prior record detail from rap sheets
• Preliminary models will be presented at the
  September 9th meeting.

30
(No Transcript)