Variables in the four datasets are classified into the broad categories of conviction offense severity, prior record, offense seriousness, grid cell type, habitual/modifiers, departure, and extra guideline variables.The 2004 Michigan Sentencing Outcomes Data (Part 1) contains a total of 118 variables including type of disposition, prison sentence, length of prison sentence, log of prison length, statutory maximum, log of statutory maximum, 6 prior record level variables (Level A-Level F), 6 offense level variables (Level I-VI), 8 crime class variables (Class A-Class H), straddle cell, prison c