Project Director: Brian Ostrom, Ph.D. National Center for State Courts

1. Assessing Consistency & Fairness in Sentencing: A Comparative Study in Three States Project Director:Brian Ostrom, Ph.D.National Center for State Courts

2. What is the research goal? Examining consistency in sentencing across 3 alternative sentencing guideline schemes: Michigan, Minnesota & Virginia • Consistency can be defined: • Similarly situated offenders receive similar sentences • Increasing certainty and predictability • Reducing disparity Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

3. Why these 3 states? • These states represent 3 distinct approaches to structuring judicial discretion • Well-respected systems • Alternative design strategies • Voluntary and presumptive • Data is more readily obtainable • Minnesota: presumptive, determinate, and tighter ranges • Michigan: presumptive, indeterminate, and wider ranges • Virginia: voluntary and widest ranges Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

4. What is disparity? • General – when discriminatory factors play a role in sentencing • Minnesota – uniform and proportional sentences not effected by race or gender • Michigan – inconsistent and discriminatory sentences • Virginia—absence of appropriate and just criminal penalties Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

5. Who is similarly situated? • Similarly situated offenders • Minnesota – 60 grid cells, 60 similarity groups • Michigan – 256 grid cells, 256 similarity groups • Virginia—who is similarly situated? • Within a crime group • Worksheet A & C score Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

6. What is the study approach & what type of data analysis is used? • Empirical assessment of consistency in sentencing and how it relates to alternative sentencing guideline structures • Analysis • • Multivariate statistical analysis (various techniques) • Reviewing all other state guideline systems, and assessing impact of recent supreme court decisions • • Review and comment by commission and staff Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

7. What is the current status? • Databases are assembled for all 3 states • Statistical analysis is ongoing, VA was most recently added Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

8. Michigan - sample results • Major finding: Local legal culture (LLC) appears to have greatest impact on sentence variation. • Michigan has relatively large sentencing ranges – making it possible for LLC to have larger impact • Where you are sentenced plays a significant role in the sentencing outcomes Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

9. Michigan – more sample results • Using Circuit 3 (Detroit) as a baseline, 21 other circuits had significantly higher rate of imprisonment; 7 circuits had significantly lower rate of imprisonment • Out-state offenders more likely to get prison and, if so, for longer times, differences are not due to offense and offender factors • Results show that different sentences can be given while still complying with the guidelines. Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

10. Virginia Analysis Plan • Focus on these individual crime groups: • Assault • Larceny • Burglary • Fraud • Drugs • Robbery • Look at the decision making associated with • Worksheet A – to model prison/no prison decision • Worksheet C – to model prison sentence length decision Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

11. Virginia Initial Results -- Burglary • We explain less variation in sentencing in VA as compared to MI & MN (MN the most) • We find no evidence of disparity based on race or gender, or circuit • VA’s wider ranges make variation harder to explain Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

12. Virginia Initial Results -- Burglary • Prison/no prison decision (Worksheet A) • How well does the statistical model work? • 79% correct prediction • Model does 27% better than chance alone Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

13. Virginia Initial Results -- Burglary • Prison/no prison decision (Worksheet A) • Change in probability of state controlled sentence holding all else constant • Firearm: 33% • Felony Property Conviction (6-10): 21% • Felony Property Convictions (11+): 24% • Legally restrained other than parole: 17% • Parole/post-release: 25% Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

14. Virginia Initial Results -- Burglary • Prison sentence length (Worksheet C) • How well does the statistical model work? • 38% explained variance • Comparison to other states • Minnesota – 85% • Michigan - 64% Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

15. Virginia Initial Results -- Burglary • Prison sentence length (Worksheet C) • Estimated impact holding all else constant • Knife or firearm – increases sentence by 152% over what otherwise would get Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States

16. Virginia Initial Results Rate of prison sentences is similar for person crimes but higher for property and drug crimes What is current status of prison capacity? In general, do judges (or lawmakers?) perceive guideline ranges as wide? Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States