The posts below are brief summaries of 14-week research projects designed and carried out by our student team. tPP plans to release the full studies as peer-reviewed publications in the future.
Gender, Jail, and Injustice: Gender Interaction Effects on Judicial Sentencing Rhetoric
Maddie Weaver & Alexandria Doty
The following report analyzes court documents that correspond with the Prosecution Projectís dataset as of mid-April 2019. There is a clear and statistically significant relationship between gender and sentence lengths, but there are no prevailing theories on the reasons behind this correlation. The purpose of this study was to find out why defendants that have similar backgrounds in all other coded variables still face different sentences, as decided by a judge through the federal sentencing guidelines. There is no written record of how a judge uses the guidelines or any other logic for that matter to show how they reach the sentence that they do. We wanted to find the patterns between how sentences are given and connect it with the rhetoric that judges use during sentencing hearings for both men and women. It was hypothesized that, when analyzing judgeís sentencing transcripts, there will be gender-based biases within the rhetoric.
The final sentencing hearing is focused on giving both the prosecuting witness(es) and the defendant a space to express the effects of the crime and the trial on their lives. Federal judges are elected or appointed to be impartial upholders of the law, but as shown by the creation of the Federal Sentencing Guidelines, impartiality is something not commonly achieved. Thus, we believed that we would be most likely to find consistent rhetoric used by judges by utilizing sentencing hearings. Our study looked at the transcripts published from these hearings, as well as other documentation related to sentencing such as memorandums and judgements, in order to collect data of the lexicon and rhetoric used.
The hearing transcripts from 54 women within tPP database were analyzed and then compared to a comparable sample of men to look for major differences or patterns that were present. We analyzed these documents through AntConc, a corpus linguistic analysis software that allows us to look into patterns present in large selections of text and found many interesting patterns.
One interesting finding was through another software we used, R, that showed menís memorandums being shockingly similar to womenís judgments when it came to the ratio of positive and negative words used throughout the document. Similarly, menís judgments were very similar to womenís memorandums ratio of positive to negative words. While all had significantly more negative than positive words, the menís judgments had the most positive to negative ratio, just exceeding the half and half point.
With more time and expertise of R and other sentiment analysis packages, this research could dive deeper into finding what these patterns describe. If bias is found with closer examination by more advanced sentiment analysis, judges could be trained and better understand their bias as well as the bias of the Federal Sentencing Guidelines that could be leading men to receive disproportionally longer sentencing lengths. This could then be examined not only for gender but also inclusive of variables like race.