Abstract
When examining group-based differences in sentence length, researchers must manage an invariably right-skewed dependent variable. While some count-based approaches can mitigate sentence length’s overdispersion, the federal measure is continuous, and therefore, it is common to log-transform sentence length before entering it into a (single- or multi-level) linear model. Using federal sentencing data from 2018 to 2022, we show that modeling sentence length in its logged form does not “solve” the issue it sets out to—it simply saddles estimates with an analogous issue. Moreover, by product of the logarithmic function’s non-linearity and racial/ethnic differences in sentence types, the log-transformation (1) produces racial/ethnic differences non-reflective of reality and (2) kickstarts a process through which racial/ethnic effects are differentially generated.
Get full access to this article
View all access options for this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
