Could a Null-Hypothesis Model Bring Greater Clarity to Forensic Entomology?

green bottle fly

Lucilia sericata, also known as the common green bottle fly, is among the blow flies (family Calliphoridae) that are often the first insects to arrive at the scene of a death and are thus closely analyzed by medicolegal forensic entomologists. The factors affecting their arrival and their rate of colonization, however, are complex, and two experts in forensic entomology propose the adoption of a null-hypothesis model to reduce confusion and disagreement over terminology in the field. (Photo credit: Joseph Berger, Bugwood.org)

In its popular understanding, forensic entomology is the field of science that uses certain insects’ affinity for decomposition as a tool in death investigations. It is, of course, not as simple as TV crime dramas might lead you to believe, and practicing forensic entomologists know first-hand just how complicated the science is. Part of that stems from the inherent complexity of entomology in general, but it is also party due to a lack of agreement within the field itself about exactly how forensic entomology is practiced.

Michelle Sanford, Ph.D.

Michelle Sanford, Ph.D.

Disagreements over terminology in forensic entomology have flared up and lingered unresolved for many years, but a pair of entomologists are making the case for a new path toward resolving this lack of a clarity in the field. In their article, “Is PMI the Hypothesis or the Null Hypothesis?” published today in the Journal of Medical Entomology, Aaron Tarone, Ph.D., associate professor of entomology at Texas A&M University, and Michelle Sanford, Ph.D., forensic entomologist at the Harris County (Texas) Institute of Forensic Sciences, propose the use of a null-hypothesis framework throughout the profession and argue that “all terms can be responsibly presented in court if they are presented as null hypotheses with an appropriate list of assumptions.”

Aaron Tarone, Ph.D.

Aaron Tarone, Ph.D.

Tarone and Sanford answered a few questions for Entomology Today about the state of forensic entomology and the directions they hope the discipline could take in the future, based on their proposal. See their full essay promoting the use of null hypotheses in forensic entomology in the Journal of Medical Entomology, and read the lightly edited Q&A below.

Entomology Today: What led the two of you teaming up to write this Forum article for JME?

Sanford: A few years ago, I re-read an article that Paul Catts wrote in 1992, on problems in estimating the postmortem interval (PMI) using insects, and as I was doing more and more casework for the Harris County Institute of Forensic Sciences I realized we were still running into these issues. I was thinking a lot about how we were still encountering sources of uncertainty in the use of insects to estimate PMI, but I didn’t have a good solution. Aaron had a great idea for how we could organize our thought process and provide a framework for tackling these persistent research and validation issues in forensic entomology by using a null-hypothesis framework.

Tarone: Since before Michelle and I were graduate students, there has been an issue in the field that we are still dealing with as a research community. What frustrates me about what the way this issue has been handled is that our community has acquired a focus on terminology but not much of a clear strategy for systematically addressing issues like those noted by Catts in 1992. More importantly, it is abundantly clear that this approach has not addressed the core issues we grapple with, which is simply that entomological evidence must be interpreted conditionally due to the many assumptions made in casework.

As someone that was originally trained in genetics, this approach has been very frustrating to watch because there is an excellent example in forensic science of a highly complex problem effectively entering into the courtroom: DNA evidence. In the case of DNA evidence, the concept of Hardy-Weinberg equilibrium, which we note in our article, played a prominent role in the way the population genetics community recommended how to tackle the presentation of similarly conditional evidence in the courtroom. If population geneticists can do it, so can forensic entomologists, but not with our current approach. This forum article in the Journal of Medical Entomology was designed to encourage the field to think about whether we can learn from famous null hypotheses in biology to improve the way forensic entomology research and casework is implemented and presented.

Why is it critical to rise above disagreements over terminology, and how does operating in a null-hypothesis framework work toward that need?

Sanford: Terminology in any field is critical to mutual understanding, and the forensic entomologist must teach the court about their report using the terminology that best describes their findings. When a single organized approach is adopted, such as the null model described in this manuscript, we can provide consistent definitions to the court by agreeing on the assumptions that are being evaluated and disagree as experts on the more pertinent details of a case. Furthermore, the null-hypothesis framework organizes a research framework to attempt to tackle some of the longstanding sources of uncertainty in the field.

Tarone: Refining our definitions does not fix the challenges facing the field. As recognized by many of our colleagues, including one of our reviewers, there is no definition in forensic entomology that does not come with an exception. All terminology differs by the set of explicit or implicit assumptions associated with them (and in some cases their connotations), and these different terms may be more or less relevant on a case-by-case basis. There is limited empirical evidence to show definitive reasons for universally using one term over the other. All we have done is create an ever-expanding pool of terms over which we are going to argue about.

Also, it is clear that all of our colleagues appreciate this, but for varied reasons prefer to talk about what they do in different ways in a good faith effort to be as intellectually honest about their work as possible. This really shouldn’t be a problem, as long as there is an acknowledgement of assumptions associated with casework and research—but the issue keeps coming up. Because of this situation, it seems a more productive use of our time would be to develop the tests necessary to reject or fail to reject any of the terminology used in the field (in specific cases of interest) than to argue over terminology.

What work does the field of forensic entomology need to do to be able to work within a null-hypothesis structure?

Sanford: The first steps will consist of designing experiments within this framework that will address particular uncertainty issues in forensic entomology. I think a key component to this will be to build strong collaborations with agencies responsible for medicolegal death investigations to facilitate the validation of the research findings for use in casework in different geographic locations. Additionally, we should educate ourselves on the kinds of uncertainty that are used by forensic-sciences service providers so that we may also communicate our findings in a language common to forensic science.

Tarone: It is important to develop the tools to identify violated assumptions related to what we do. Ideally, the field would be 1) identifying assumptions, 2) identifying tests to determine when those assumptions are relevant to a particular case, 3) validating the effectiveness of such tests, and 4) determining consequences of violation of specific assumptions. The holy grail would be a test to determine when time of colonization (TOC) equals PMI or minimum PMI without testing each individual assumption. This kind of test would be akin to comparisons of allele frequency to genotype frequency under Hardy-Weinberg expectations.

Your proposal aims to bring the underlying assumptions that forensic entomologists make out into the daylight. What are some examples of those assumptions?

Sanford: We make many assumptions when calculating the TOC estimate that may influence its interpretation as an estimate of the PMI. One example is the assumption that the air temperature, typically obtained from the nearest weather station, accurately reflects the development temperatures of the insects on the body at the scene. However, we know that maggots will aggregate and alter their microclimate temperature, and we know that scene characteristics may differ between the weather station and the scene. We know that the possibility that these temperatures are different may introduce uncertainty to the estimate of TOC. This uncertainty leads to larger TOC estimate ranges, which in turn leads to uncertainty in the PMI estimate.

Another assumption that we have more recently acknowledged is that different populations of the same fly species may have different developmental responses to the same temperatures. However, we don’t have development data from local populations of many species and have difficulty in accounting for how this uncertainty impacts the interpretation of the TOC estimate for the PMI. Yet another assumption that is made is that the drugs and medications that the decedent may have ingested do not influence the development of the insects. There are some notable exceptions in certain illicit drugs and common medications, but we know that different species may respond differently and that these substances may differentially impact different life stages.

Tarone: There are not published rates of myiasis (colonization of live organisms by flies) that are relevant to human casework. All we know is that it can and sometimes verifiably does happen. Michelle published one of these examples. We do not know how to detect the instances in which it happened but discovery of the remains was too late to identify its occurrence. That is just one example of an assumption that, if violated, can cause some terms to be irrelevant to a particular case (but not necessarily all cases). A test to identify when myiasis has likely happened in cases when it is not obvious would be useful.

How would your proposed method help forensic entomologists in explaining levels of uncertainty in a courtroom?

Sanford: The null-hypothesis framework allows for the identification of sources of uncertainty in an estimate of TOC that may affect the interpretation of the PMI based on the insect data. With research and validation organized for the null model, the forensic entomologist may then be able to quantify and communicate the uncertainty in an estimate.

Tarone: Hopefully it will allow for more discussion of relevant assumptions, and their expected impacts, in casework and less distracting obsession with terminology.

What reaction do you expect to this proposal?

Sanford: I expect that there will be resistance, at first, because it appears to be a big change, and people tend to be hesitant to big changes. However, I would suggest that it is not so much a change to the way we are doing things than a change to the way we think about the process. We are already working to determine how well the TOC estimates the PMI but we are grappling with how we interpret and communicate the uncertainty present in these estimates. The null-hypothesis framework has the potential to facilitate the identification of uncertainty and, perhaps even more importantly, the ability to quantify and effectively communicate that uncertainty.

Tarone: The reviews of our manuscript and other conversations on this topic would suggest that our colleagues will be thoughtful about how to incorporate what we suggest. In the immediate future, I suspect resistance to the thought process because there are impractical aspects to what we suggest, as there are not many tests for assumptions that can be immediately implemented in casework. However, for me, this forum article is about the “long game” in forensic entomology research and casework.

In our article, we point to a historical example of a big change in evolutionary biology driven by Neutral Theory. It was initially considered impractical due to the inability to measure certain parameters. However, there were some enterprising and talented people in the field that found ways to develop tests for selection none the less. Now, those tests are commonplace in biology research. I think the best-case scenario for how the community would receive this proposal would be to follow that model. We can talk about what we do as a null hypothesis now and begin doing research and casework with this in mind. I suspect that if we do so, some enterprising colleagues will find effective ways to tackle challenges in the field and in several decades we could find ourselves positioned to more effectively address issues of uncertainty in casework.

Journal of Medical EntomologyRead More

Is PMI the Hypothesis or the Null Hypothesis?

Journal of Medical Entomology

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: