An Inferential Test for Expert Testimony

One can be reasonably certain if witness accounts of the past are consistent or not consistent with physical evidence in the present, but one cannot reliably surmise past events from physical evidence unless there is only one plausible explanation for that evidence.

The wordy statement above seems difficult to understand at first, but the concepts in it are really very simple. Any court of law can readily apply these concepts to evaluate the testimony of any scientist offered as an expert. Here is an explanation of this inferential test.

One can be reasonably certain if…

The courts require that the opinions offered by a scientist in the courtroom are made to a reasonable degree of scientific certainty or scientific probability—the exact terms may vary according to the state or local jurisdiction. Although feelings of certainty may be subjective and although a scientist may be wrong, application of this test nevertheless guarantees the possibility that the expert can be certain, and it will expose whether or not that expression of certainty is or is not logically valid. This is because this inferential test is based on and derived from deductive inference. Deductive inference is valid for certainty because it allows one and only one conclusion. The truthfulness or soundness of the inference, however, depends on whether or not the premises of the inference are correct.

The use of the word, probable, in any forensic opinion is more of an expression of uncertainty rather than certainty. It is a way of saying, “I think this is what it is, but I am not sure.” When there is more than one possible conclusion to an inference, the conclusion can no longer be certain, although it might be probable.

This inferential test allows that the expert can be reasonably certain of certain things, although one can never claim absolute certainty of anything due to the philosophical doctrine of fallibilism. The test also points out the limitations of the expression of probability for past events.

…witness accounts of the past…

The past no longer exists. It cannot be touched, seen, heard, or measured. The past remains with us only in the form of memory or record. Witnesses who were present to see or hear a past event may remember it and relate it to others. Witness accounts of past events—and in some cases, even the past events themselves—may also be documented in the form of a record such as a video recording, an audio recording, or a written statement. These memories or records may not always be entirely accurate, but they may be found to be reliable after comparing them to physical evidence.

This inferential test recognizes the importance of witness accounts preserved in memory or record. These items are really the only link to the complex and unique events of the past, and as such they should not be ignored or minimized by scientists or by the court.

…are consistent or not consistent with physical evidence in the present,…

The process here described involves a comparison of witness statements with physical evidence. Examples of physical evidence include hairs and fibers on clothing, stains with DNA, damage patterns in motor vehicles, and autopsy findings. They are the items that scientists can observe and measure.

The scientist sees if the witness accounts are consistent (they fit) or not consistent with (they do not fit) the physical evidence. Consider the process one goes through in buying clothing. A person tries on the clothing at the store to see if it fits the shape and proportions of his or her body. In a similar fashion, a scientist can compare witness evidence with physical evidence and say whether or not the two “fit” each other. If the two “fit” then there is a consistency. If the two do not “fit” then there is an inconsistency. This form of comparison utilizes deductive inference, allowing witness accounts and well-established scientific principles to serve as premises for predicting the physical evidence, thereby testing the witness accounts with the physical evidence.

For example, a pathologist can look at a gunshot wound in the head and tell something about the range of fire. If the witness gives an account of a gun fired 20 feet away from a victim and the gunshot wound in the head has the physical characteristics of a pressed contact gunshot wound, then the witness account of the distant range gunshot is not consistent with the physical evidence—the pressed contact gunshot wound. On the other hand, if the witness sees a person put a gun to his head and shoot himself, then the witness account is consistent with the pressed contact gunshot wound.

…but one cannot reliably surmise past events from physical evidence…

Sherlock Holmes, a fictional character in multiple stories written by Arthur Conan Doyle, advocated surmising past events from physical evidence. All detective and crime-fighting fiction since Sherlock Holmes use this kind of surmising. First, the detective surmises what happened in the past on the basis of “clues” usually in the form of subtle physical evidence. Later in the story, when the criminal is confronted with the story marvelously derived by the hero, the criminal immediately confesses to the crime, acknowledging the truthfulness of the conclusions of the supersleuth. The numerous detective and forensic science shows that fill the prime time slots on television these days use the very same plot device. Screenwriters apparently find great drama with this approach, but real life is not like the television shows or the detective novels. Unfortunately, numerous people in our society—including ones who should know better (forensic scientists, forensic pathologists, child abuse pediatricians, police officers, attorneys)—are persuaded that this approach works reliably.

Although this type of surmising might be useful as a heuristic (“rule of thumb”) that may help during the investigation of a case, it should never be the basis for any settled forensic scientific opinion expressed in the courtroom. This is because in real life, this kind of surmising does not work reliably. It is essentially affirming the consequent—a classic logical fallacy. For any set of physical findings, there may be and often is more than one possible explanation for the physical evidence. With multiple possible explanations, selecting one explanation over another would then be a matter of guesswork or speculation. It then logically follows that a scientist cannot on the basis of speculation express a reasonable certainty that his or her belief of what happened is the truth.

Other names for this type of thinking are scenario building or backward reasoning. It is a fallacy to believe that a scientist can manufacture a “story”—as if he or she was a witness to events in the past—on the basis of physical evidence. It is also a fallacy for an expert to surmise the commission of a crime on the basis of physical evidence. These activities cannot be done reliably, and opinions of this type should never be admitted as expert testimony. A scientist should not invent scenarios or surmise criminal activity, particularly when witness accounts are available for testing in the manner stated previously.

When witness accounts are not available, a scientist cannot and should not state what happened with any certainty. Without witness accounts, he or she can only state reliably what is possible, not what actually happened.

There is a universal experience we all share. That experience is known as “jumping to conclusions.” It can also be described as “prejudice.” Certain observations we make about individuals and situations in the present may lead us to conclude that we know the events and circumstances that led to the items we see. For example, we know that judging someone’s character by skin color or the way he or she dresses is wrong and frequently incorrect, but we have all done it. This is the problem with surmising past events. It leads to a form of conclusion jumping that we are prone to do as human beings, and it does not allow us to consider carefully the scientific limitations of our observations and conclusions.

…unless there is only one plausible explanation for that evidence.

The final phrase provides the only exception to surmising past events from physical evidence. Under certain circumstances, when the possibilities are limited to only one plausible explanation, that explanation can be stated with a reasonable degree of certainty. In classical logic, it represents the if and only if exception to affirming the consequent.

For example, a medical examiner filing a death certificate has to offer an opinion of the manner of death. A manner of death is a one-word description of the circumstances surrounding the death. There are only five manners: natural, homicide, suicide, accident, and undetermined.

If a body is discovered in a wooded area with numerous stab wounds to vital organs and multiple devastating blunt force head injuries, it would be reasonable to conclude a manner of homicide. One cannot envision any circumstance where natural, suicide, or accident could plausibly apply. Also, the nature of the injuries and the condition and location of the body are of such a nature to allow one to conclude reasonably that foul play has occurred, so an undetermined manner would not be suitable.

Note, however, that no scenario is proposed for how this homicide came to occur. No scientist can describe in any more detail the complex and unique chain of events that led to that death with any reasonable certainty. Only eyewitnesses can relate such events.

In a courtroom setting, this if and only if exception should not be allowed in expert opinion testimony. The jury should be the only ones allowed to apply this exception as an ultimate issue, once all of the evidence has been presented to them.

What is the practical bottom line of this inferential test for scientific testimony? I believe it can be summarized like this:

1. Before offering an opinion in any case, the scientist must first free his or her mind of any preconceived ideas and then read the statements of the witnesses who saw what happened. The scientist must compare those witness statements with the physical evidence.

2. If the witness statements in their entirety fit the physical evidence, those statements should be affirmed as consistent with the physical evidence.

3. If one or more portions of the statements do not fit the physical evidence, those portions should be considered as not consistent with the physical evidence.

4. If scenarios are offered during direct or cross-examination in the form of hypotheticals and the expert is asked to offer an opinion as to whether the scenario is consistent or not consistent with the physical evidence, the expert can offer that opinion with a reasonable degree of certainty.

5. The scientist should never invent a scenario that is not supported by witness statements to explain the physical evidence.

6. If scenarios not supported by witness statements are offered by other people, the scientist should speak of them only with terms of uncertainty. Terms of uncertainty include such words as: possible, probable, might happen, could happen, frequently happen, infrequently happen, suspicious for, cannot rule out, often associated with.

No court wants to be confronted with the consequences of improper scientific inference. I believe the courts should apply this inferential test to determine whether or not the opinions of scientific experts should be admitted as evidence. Doing so will prevent mistakes that lead to injustice.

An Inferential Test for Expert Testimony, now referred to as The Inferential Test, was written on April 5, 2009.