The judicial system frequently falls short of its claimed standard of impartiality, instead perpetuating racial and gender biases. While these shortcomings are well documented, their origins are less well understood. To get a better understanding of the emergence of these biases, we take a step back from the judicial system and focus on law students instead. We conduct a randomized control trial (RCT) in the introductory lecture of criminal law by randomizing a characteristic feature of the perpetrator in the final exam and analyzing the response behavior of the students. Thereby, we can determine whether biases are already observed in students starting law school. In follow-up projects we further aim to track the development of the biases throughout law school.
Experimental design
Our study consists of two main parts. First, participants complete an online questionnaire covering socio-demographics, motivations for studying law, preferences, challenges, societal opinions, and suggestions for improving the German law school system. The questionnaire is used for randomization verification and subanalyses.
In the subsequent phase, administered during the final examination of the law course, two distinct versions of the exam are assigned, each featuring randomized elements pertaining to a singular aspect of the characters in the exam. The data collection process is bifurcated into two rounds: the completion of the online questionnaire during the course lecture and the subsequent collection during the final examination.
Design details
Randomization method
To facilitate a rigorous randomization process, we will employ a computer-based random generator. This tool will be utilized to randomly assign individuals to either Exam A or Exam B, ensuring an unbiased and systematic distribution. More precisely, the random generator will specifically assign either Exam A or Exam B to each individual table within the examination room. Subsequently, the corresponding exam will be promptly placed on the designated table according to the random assignment.
Randomization unit
Treatment randomization is conducted at the individual level in our study.
Clusters
Our intended sample size for the study is 180 students, and it's important to note that our design does not involve clustering.
Observations
Our intended sample size for the study is 180 students, and it's important to note that our design does not involve clustering.
Arms
Our study employs a design featuring one treatment group and one control group. Students are randomly assigned to either the treatment or control group, ensuring that 50% of our sample, comprising 90 students, is allocated to the treatment group, while the remaining 50% is assigned to the control group.
Primary outcomes
The outcome of interest is the response behavior of the students. Specifically, we are interested in the critical parts of the students argumentation to characterize the criminal offense. Thereby, we are especially interested in variations of the assessment of specific features of the offense.
To conduct the analysis, we use text analysis. One approach to conduct the analysis would be to define different text libraries for plausible variations in the argumentation in cooperation with the law faculty. Thereby, we could track the use of words associated with the libraries and classify the students' string of argumentation according to the words they use.
An alternative approach for evaluating the exams would be to check how the students characterize the offense according to the facts of the case and how they align with the model solution.
Cited by
No citation data available for this registration yet.