Cognitive bias in DNA analysis

DNA mixture interpretation is subjective in nature and like most human beings, DNA examiners are prone to cognitive bias. Cognitive bias happens when scientists unknowingly get affected by certain information which then affects the choices they make as part of DNA interpretation.

In a study conducted by Itiel E. Dror and Greg Hampikian, 17 DNA examiners were asked for their interpretation of DNA mixture in a decided rape case, where the accused was convicted on the basis of DNA evidence and his co-accused’s testimony, who turned approver. Out of the 17 examiners, only one of them returned the same result as the DNA experts in the case i.e. the accused could not be excluded as the source of DNA mixture obtained from the victim’s vaginal swab. 12 examiners concluded that the accused could be excluded as the source of DNA and 4 said that the results were inconclusive. It was found that examiners that did not have any contextual information disagreed with the initial mixture DNA interpretation results.

This shows that contextual information like details surrounding the arrest of the accused, their past criminal record or access to the accused’s reference DNA profile before interpreting the evidence sample could bias the examiner. If the electropherograms for a mixed DNA profile are interpreted alongside the reference samples of suspects, it is likely that experts may fall prey to cognitive bias and observe patterns in the crime scene samples similar to that in the reference samples of the accused. It is important to note that cognitive bias occurs unknowingly and therefore scientists need to practice certain techniques to minimise the possibility of bias. Experts while determining the quality of the data and assigning alleles may make different choices based on the information available to them during their decision making process. If there are no standards and guidelines for experts to interpret this data, the results from experts can vary depending on the information they receive.

A practice of linear sequential unmasking (LSU) has been found to minimise bias in this process. LSU is a process by which the scientists examine the crime scene samples first and only once their examination is complete (including calculation of statistical significance of various genotypes) do they proceed onto the examination of reference samples. Best practice guidelines developed by leading experts in the field, now provide recommendations on to prevent cognitive bias and include LSU as essential to this process.

The Scientific Working Group on DNA Analysis Methods (SWGDAM) has published guidelines for interpretation of autosomal STR typing in forensic laboratories. These guidelines clearly state that comparison with the reference profile should be made only after the raw data of the crime scene sample has been examined and potential genotype pairings determined. Apart from this, the SWGDAM guidelines require detailed documentation of all assumptions and interpretations made by the DNA examiner, along with supporting data. This ensures transparency to the subjective process as well as the possibility of a future review by other scientists.

Similarly in the UK, the Forensic Science Regulator (FSR) has published a guidance document on cognitive bias effects relevant to forensic science examiners. The document provides details on the effect of cognitive bias and mitigating strategies that examiners can follow during casework. The guidance document provides multiple recommendations like the interpretation of profiles independent of the reference samples, another examiner carrying out repeat interpretation of the results, carrying out analysis and interpretation in blind (i.e. without any knowledge of the case), transparency of all records, etc.

Therefore, while reviewing a case, it is important to note the information that the DNA expert had access to before conducting the examination and its impact on the interpretation of results. Lawyers should ascertain if the scientist followed any of the recommendations from the above guidelines and verify the records to examine the possibility of bias in their case.