Presenting GC Data Comparisons to Laypersons to Understand Potential Courtroom Implications

In forensic investigations, chromatography is regularly used to characterize a sample’s components to provide a chemical pattern to compare with known references, which is often presented to individuals without specific training in analytical chemistry. While comprehensive two-dimensional gas chromatography (GC×GC) has become popular in forensic research for analyzing samples such as fire debris samples, drugs, chemical threats, and human remains detection, among others, new methods are developed in forensic research regularly, and these methods are likely to challenge our view of what may be increasingly complex to convey through scientific communication. Katelynn Perrault Uptmor, Assistant Professor of Chemistry at William & Mary and associates investigated a pool of individuals’ ability to observe differences in images for non-chromatographic photographs, one-dimensional gas chromatography (1D GC) chromatograms, and comprehensive two-dimensional gas chromatography (GC×GC) contour plots in order to identify whether comparative observations between two outputs were facilitated or hindered when observing GC chromatograms compared to GC×GC contour plots, using photographs as a control. Katelynn spoke to LCGC International about their findings, and the paper that resulted from it.

Your paper (1) set out to answer the question “Is GC×GC data too complicated for the public to observe differences compared to conventional 1D GC data?” What motivated this question in the first place?

I have attended and coordinated several events (focus groups, discussions, and conferences) where I’ve discussed the use of GC×GC for future forensic applications. I have heard a lot of conjecture about how GC×GC would or would not fit into forensic workflows. On one hand, a lot of the GC×GC community feels that the data generated would be a huge benefit to the legal community. However, many forensic chemists have remarked that the technique is too novel for routine use and too complex for judges, juries, and lawyers to understand well. I wanted to see if we could collect primary data on what people see when they observe GC×GC output to help better inform these conversations. We were largely intrigued by the idea of playing a “spot-the-difference” type game with participants, seeing if one type of plot would be easier to compare than others. If you’ve ever played one of these games in a newspaper or magazine, you’ll probably appreciate the thought process that goes into making a visual comparison between two images. Once we started looking at literature, we also quickly realized that there is no primary literature available on how 1D GC data is viewed by laypersons either, so there was the potential to collect some foundational preliminary data in this area.

You state in the paper that GC×GC has been a new technique proposed as having potential for use in forensic analysis. What benefits does the technique offer that are superior to 1D GC?

1D GC has been a staple in forensic laboratories for a long time and is a gold standard for chemical analysis. When 1D GC is used to target a single analyte, or even just a few analytes, it is a very powerful tool. However, some existing and emerging forensic applications rely on nontargeted analysis–the need to characterize everything that exists in a sample, rather than one single component. GC×GC has higher separation capacity and detectability for these more complex scenarios and provides the ability to physically separate many analytes from one another without increasing analysis time. Some examples where this might be helpful would include applications in drug screening for unknown analytes or trying to identify a complex ignitable liquid residue (ILR) within a fire debris sample.

What challenges arise in explaining GC×GC results to non-experts, such as jurors in a courtroom, and how can further research improve the effectiveness of this communication?

Due to the additional components and difference in viewable output, I think many have grappled with whether GC×GC is too challenging to explain to people in a courtroom environment. However, one could also argue that many types of complex chemical instrumentation is presented in a courtroom environment on a regular basis, such as genetic analyzers, scanning electron microscopes with energy dispersive X-ray, and liquid chromatography with tandem mass spectrometry, to name a few. I feel that we can address this complexity through research and support the effective presentation of GC×GC when it is adopted into routine practice.

What type(s) of forensic samples are the most difficult to analyze and explain to the non-expert?

I don’t think one type of evidence is more difficult than others, but mixture samples tend to present unique challenges due to the multivariate nature of the data. It is challenging to monitor and identify many components in a complex mixture, and even more so, to compare those minute differences between samples. Forensic science often involves comparative analysis of one sample to another to determine whether an evidence sample could have originated from a known source, so determining subtle or small differences between samples is important. When dealing with a forensic “trace”, two of the things that are regularly challenged in the courtroom are indirect transfer and calibration. Indirect transfer occurs when material is transferred from one place to another, but indirectly through an intermediate person or object. For example, if I shake hands with someone who has discharged a firearm and they transfer gunshot residue to my hands, this would be considered indirect transfer. Calibration is often raised as a challenge in the courtroom since quantitative values can carry additional legal consequences, for example, as with the legal limit for alcohol in your bloodstream when driving.

Tell us a little bit about your participant pool and how you determined who to survey.

We started with the idea of using people from our university environment and eventually expanded that pool to one of the larger public-school systems, to capture a broader audience. Exclusion criteria were used so that anyone responding to the survey would be eligible for jury duty, since we thought this was a good representative portion of the public. We also excluded anyone who was currently studying towards or held a degree in the natural sciences, to avoid issues with those who might have directly learned about gas chromatography techniques and potentially skew the results.

Were there any challenges involved with putting together the image bank the participants reviewed in the survey?

The first author on the paper, Clarissa Camara (an undergraduate researcher in my group), came up with a workflow for creating images with different levels of difference, and for how to build and code a survey we could collect data from. We developed several coding systems to track the construction of the survey and to determine the best way to ask the questions. We decided to deploy an initial preliminary study to a small group of participants to gather feedback about how clear the activity was, and that was extremely helpful in altering the survey to make it more user-friendly. Through that exercise we ended up changing the way response selections were phrased, and came up with a practice activity that people would work through before starting the activity for themselves.

What difficulties did you encounter in your work?

Recruitment was initially challenging as we did not offer incentives and being that the university we started recruitment at was relatively small (~2000 students), we did not realize that we’d need a much larger pool. One of the co-authors on the paper, Cynthia Cheung, had the idea to work with the community colleges in our area. Brainstorming with our Institutional Review Board (IRB), we came up with altered recruitment strategies to expand our possible participant pool and were able to exceed the initial number of responses that we initially targeted (50), finally collecting 70 participant responses for a total of 3150 individual comparisons.

Briefly summarize your findings, and the conclusions you came to after reviewing these findings.

This study investigated individuals’ ability to observe differences in images for nonchromatographic photographs, 1D GC chromatograms, and comprehensive GC×GC

contour plots. The goal was to identify whether comparative observations between two outputs were facilitated or hindered when observing GC chromatograms compared to GC×GC contour plots, using photographs as a control.Participants were presented with a “spot-the-difference” survey via an online form where they were asked to compare two images side-by-side and answer questions about difficulty and similarity for each comparison. Each participant conducted 15 comparisons in each category for a total of 45 comparisons. We foundthat participants (n = 70) were generally very confident in their ability to compare all three categories, indicating a low difficulty level for all comparison types. We also found that participants were effective at determining across all categories whether images were completely the same vs. had some level of difference between them. Individuals said that comparisons of identical images were harder, but they were generally effective at determining whether the two images were identical. These results support that GC×GC output can be implemented in expert testimony without challenges over traditional one-dimensional techniques.

Were there any particularly interesting comments or feedback received from the participants?

We did not collect open remark feedback from participants, but in hindsight, it would have been great to include an open response box for them to remark on how they felt about the activity. I think we were most surprised by the high level of confidence that people had in performing image comparisons. Most people felt that the exercises were very easy to complete, even if their responses were incorrect. It is interesting to think about how that confidence might impact how juries make decisions.

You stated that this was one of the most *out-of-the-box* research studies my group has ever conducted, but it was a lot of fun to learn about how to conduct this type of research. Can you elaborate on those points?

I have conducted studies with human subjects before on exhaled breath and was relatively familiar with the process of obtaining IRB approval to work with subjects. However, this was the first time I conducted work for social/behavioral research using an IRB protocol, and I learned a lot about the measures that must be taken to protect individuals. My students are used to doing research in the chemistry lab, not deploying virtual surveys online, so this was also a shift in mindset. I think in the future it would be very helpful for us to collaborate with individuals from a psychology department to improve and further the study design. We consulted psychologists informally, but a formal collaboration would be beneficial.

How do you imagine the results of your study can/will be applied?

I think this work helps to inform how we think of GC×GC as a routine analysis tool that could be presented in a courtroom environment. We dispelled the myth that this is “too complex” for laypersons to understand, especially because of the high confidence they showed in comparing GC×GC output. We also dispelled the idea that there is a large benefit to heat maps for visualization, since no one performed significantly better comparing GC×GC data compared to regular photographs. I personally think that these preliminary results help to reinforce the idea that GC×GC has become a regular tool in the analytical chemist’s toolbox, and that presenting this type of data in the courtroom faces all the same challenges that any other technique would when being implemented for the first time in a new application.

Are there any next steps in this research?

We’re currently brainstorming new ideas for how to continue this work and looking at partnerships to strengthen the implementation. I am interested in comparing contour plots, surface plots and 1D representations of 2D plots. I am also interested in looking at color scheme differences in contour plots. I’ll be presenting the results of this work at the 16^th Multidimensional Chromatography Workshop in Liege, Belgium on Feb 3-5, 2025 and looking to hold discussion with experts in the field about future directions.

Reference

1. Camara, C.; Cheung, C.; Perrault Uptmor, K. A. Observation of Chromatographic Differences by Non-Specialist Viewers for One-Dimensional Gas Chromatography and Comprehensive Two-Dimensional Gas Chromatography Output.Forensic Chem. 2024, 41, 100620. DOI:10.1016/j.forc.2024.100620

Katelynn Perrault Uptmor is an Assistant Professor of Chemistry at William & Mary. Her research specializes in the application of comprehensive two-dimensional gas chromatography for nontargeted applications related to life, health, disease, and death. She has mentored numerous undergraduate researchers as part of her teaching and research program and holds the title of Henry Dreyfus Teacher-Scholar. Dr. Perrault Uptmor was also recently named the 2025 LCGC Emerging Leader in Chromatography.