Exposing rats to repeated unpredictable stressors is a popular method for modelling depression. The sucrose preference test is used to assess the validity of this method, as it measures a rat`s preference for a sweet solution as an indicator of its ability to experience pleasure. Typically, if stressed rats show a lower preference compared to unstressed rats, it is concluded they are experiencing stress-induced anhedonia. While conducting a systematic review, we identified 18 studies that used thresholds to define anhedonia and to distinguish "susceptible" from "resilient" individuals. Based on their definitions, researchers either excluded "resilient" animals from further analyses or treated them as a separate cohort. We performed a descriptive analysis to understand the rationale behind these criteria, and found that the methods used for characterizing the stressed rats were largely unsupported. Many authors failed to justify their choices or relied exclusively on referencing previous studies. When tracing back the method to its origins, we converged on a pioneering article that, although employed as a universal evidence-based justification, cannot be regarded as such. What is more, through a simulation study, we provided evidence that removing or splitting data, based on an arbitrary threshold, introduces statistical bias by overestimating the effect of stress. Caution must be exercised when implementing a predefined cut-off for anhedonia. Researchers should be aware of potential biases introduced by their data treatment strategies and strive for transparent reporting of methodological decisions.