This article contributes to the sociological research of sentencing disparities. Using the dataset consisting of 1.5 million individual decisions of criminal courts of the Russian Federation in 2009-2010, the study focuses on the influence of socio-economic status of defendants on decisions to acquit, imprison as well as on the severity of punishment. The analysis uses multiple logistic and OLS regressions. It includes control variables related to the legally significant characteristics of the offence, offender, and process to separate legal and extra-legal factors in sentencing. The strength of regression coefficients in logistic regressions is estimated by means of calculating marginal effects at average values of the variables. The regression analysis shows strong and consistent social inequalities. The system of criminal repression is targeted mostly against socially marginal and lower status strata (prisoners, unemployed, manual workers) which constitute the absolute majority of defendants and are punished more harshly than representatives of the upper strata. Besides that, the study reveals another dimension of conflict: private entrepreneurs receive more severe punishments than public officials, especially for white-collar crimes. Another regularity revealed by the analysis is a consistently more lenient attitude to students: judges tend to give them non-prison sentences and when sentence to prison — assign a shorter term. To make initial assumptions about the ways these regularities are produced at the interaction level, the study uses interviews with judges and results of the original survey. Judges apply legal rationalizations to justify their bias against the unemployed, attributing to them a higher level of social danger. Other status biases are explained with reference to judges’ professional backgrounds and attitudes.

Keywords: social status, court decision, judicial bias, sociology of law, regression analysis