Forensic Science and the Scientific Method


The scientific method, a time-honored approach for discovering and testing scientific truth, does not and cannot work for the forensic sciences in its standard form because it does not work for past events. Past events cannot be observed, cannot be predicted or deduced from physical evidence, and cannot be tested experimentally. The forensic scientific method is a modified form of the scientific method that compares anamnestic evidence obtained by investigators with observable physical findings discovered at the crime scene, in the crime laboratory, or in the autopsy suite. This comparison verifies if witnesses or suspects are telling the truth about what they witnessed. The method is a powerful technique for determining the truth of past events. Unfortunately, a confused reliance on the standard scientific method for past events can lead to mistaken conclusions. The Shaken Baby Syndrome or Shaken Infant Syndrome hypothesis is one notable and regrettable example.


forensic science, scientific method, forensic scientific method, Shaken Baby Syndrome, Shaken Infant Syndrome, Whiplash Shaken Infant Syndrome, Shaken Impact Syndrome

Natural and physical scientists since the 17th century have learned most of what they know about the universe through the scientific method. This method continues to define the way scientists conduct the work of science. It has withstood the test of time.

The scientific method is “a method of research in which a problem is identified, relevant data are gathered, a hypothesis is formulated from these data, and the hypothesis is empirically tested1“ (Table 1). First, the scientist makes an observation or identifies a problem related to that observation. He or she then creates a hypothesis to explain the observation. The hypothesis then allows the scientist to make predictions. The scientist then tests the predictions by experiments or by further observations under controlled conditions.

Falsification is an important element of the scientific method. The scientist ideally attempts to disprove or falsify the hypothesis. If the hypothesis can be disproved, it can then be discarded so that the scientist can move on to a more accurate hypothesis. On the other hand, if experimentation or further observation confirms the hypothesis, this confirmation does not necessarily prove the hypothesis to be true. Other scientists are allowed to test and to attempt to falsify the hypothesis. Repeated confirmations of the hypothesis over time may result in the hypothesis becoming a theory. A theory is a general principle that scientists use to explain phenomena and predict events.

Forensic scientists have also been engaged in science for several centuries; however, the applications of forensic science have been distinctly different from the applications of the natural and physical sciences. Forensic science developed from societal concerns. Some of these concerns, including public health and safety, the identification of human remains and the administration of justice, are weighty and affect the well being of people living together in society. The purpose of forensic science has never been to understand the universe and how it functions. The purpose has always been to understand what happened and who is responsible for what happened.

Although natural and physical scientists pursue concerns that differ from those of forensic scientists, both groups may believe they have the scientific method in common. Nordby in his textbook on forensic science expresses that belief, but he provides no description of how the scientific method applies to forensic science. Rather, he states, “attempting to characterize reliable scientific methods, as if describing some lifeless nonexistent abstraction, remains doomed to failure. There simply is no such generalized abstraction available to describe2.” Nordby then presents a list of “reliable methods” to evaluate hypotheses presented in the courtroom (Table 2). This list makes no mention of the standard scientific method used by natural and physical scientists for centuries.

There is a reason for this confusion about forensic science and the scientific method. The scientific method as upheld by most scientists does not and cannot apply to the forensic sciences.

Forensic Sciences Study the Past and Not the Present

It may seem almost sacrilegious to suggest that forensic scientists cannot use the scientific method. After all, to suggest this may be to question whether or not forensic scientists are even scientists! If science is defined as “a branch of knowledge or study dealing with a body of facts or truths systematically arranged and showing the operation of general laws1“, then forensic science is a science. The methods employed by that science must differ from those of the natural and physical sciences. Why is this? It is because forensic sciences study the past and not the present.

Although the administration of justice very much involves present issues and concerns, most of the crimes or torts brought to the courtroom occurred in the past, mostly the recent past. Witnesses called to testify in court explain to a judge or jury what they saw, not what they are currently seeing. Tangible evidence admitted in court — evidence analyzed by a scientist or presented as documents — mostly point to events in the past. Although a suspect may possess a distinctive print pattern in his fingers in the present, his latent prints at a crime scene point to potentially legally culpable behavior in the past.

How does the past prevent the use of the scientific method? First of all, one cannot observe the past. Items in the past may be remembered by some, but they cannot be seen, smelled, heard, tasted, or sensed in any way. Observation is an activity in the present that requires the use of the senses.

Secondly, one cannot predict the past. Prediction is an activity in the present that looks to the future, not the past. An attempt to use the scientific method to determine what happened in the past would be “retrodiction.” “Retrodiction” is a neologism for good reason: science cannot “retrodict.” This will be explained subsequently.

Thirdly, one cannot design experiments or controlled observations to determine what happened in the past. Experiments or controlled observations might help one see if a situation is possible or not possible under a set of defined circumstances, but one cannot design an experiment that will replicate the complex variety of conditions that existed in the past — conditions that are often not known in full detail. An experiment or set of controlled observations also cannot provide information about the order and timing of past events.

Fourthly, a hypothesis confirmed by multiple experiments and observations in time may become a theory, but forensic science is not and should not be concerned with the formation of theories. Forensic science may use theories derived from the work of natural and physical scientists, but determining what happened and who is responsible are not scientific theories. Understanding events in the past — events that are not controlled, not replicable, and frequently complex — is the intent of the forensic sciences. Using theories to explain the specific events of any case will oversimplify the explanation. This type of error will be further explained subsequently.

The Fallacy of Scenario Building

Science can predict but it cannot “retrodict.” Events in the past (symbolized by “E” in Fig. 1) can lead to a set of observable findings in the present (“F”). The forensic scientist can predict reasonably how certain events can lead to specific observable findings. For example, a scientist or any other person can predict that someone shedding his blood at a crime scene can leave a specimen containing his or her DNA. If a person places the muzzle of a gun firmly against his temple and fires it, a pathologist can reasonably predict a gaping contact gunshot wound. Forensic pathologists and other forensic science experts use “consistent with” to explain how a hypothetical scenario presented in the courtroom can explain a set of physical findings. Predicting how events can lead to findings can be done reasonably, and it is done frequently.

What happens when one attempts to reason from findings to events (Fig. 2)? Any set of observable scientific findings can be consistent with numerous different scenarios of past events. For example, we may observe at autopsy a contact gunshot wound in the head of a decedent. Numerous different firearms, all of the same type or even different types, can cause a contact gunshot wound resembling the one seen at autopsy. Any man or woman in the world, including the decedent, shooting the firearm at any time day or night is capable of replicating that wound. Other scientific evidence may narrow the possibilities but not close to the extent needed to be certain. Even finding the fingerprints or the DNA of a particular man or woman on the firearm provides no proof that he or she fired a bullet from the gun that formed the contact gunshot wound.

It is tempting for the scientist to deduce the events of a crime from the crime scene findings. This activity, known as scenario building, is not only useless but also harmful. A person who does this will frequently find himself to be wrong when he learns more facts at a later date. Professor Alan Moritz, in his important paper describing mistakes made by forensic pathologists, calls this mistake “categorical intuitive deduction3.” Usually the experienced expert in pathology is the one often culpable of using this Sherlock-Holmes-style of reasoning, according to Dr. Moritz. A practitioner of “categorical intuitive deduction” does not realize that any number of situations can lead to the same findings.

Mystery writers, such as Sir Arthur Conan Doyle and Agatha Christie, write about protagonists who engage in this form of thinking. Sherlock Holmes himself provides an example when he infers Dr. Watson’s past military service:

Here is a gentleman of a medical type, but with the air of a military man. Clearly an army doctor, then. He has just come from the tropics, for his face is dark, and that is not the natural tint of his skin, for his wrists are fair. He has undergone hardship and sickness, as his haggard face says clearly. His left arm has been injured. He holds it in a stiff and unnatural manner. Where in the tropics could an English army doctor have seen much hardship and got his arm wounded? Clearly in Afghanistan4.

In reality, any and all of these observations could apply to a wide variety of individuals under a wide variety of circumstances. Some may possess military bearing without serving in any branch of the military, let alone the army. Some may develop a tan without visiting the tropics, let alone Afghanistan. Some may have haggard faces without having been sick or undergoing hardship, let alone the hardship specifically brought about by war. Someone may develop an injured arm for numerous reasons. Perhaps Sherlock Holmes was enhancing probabilities by observing all of these items together, but neither Sherlock Holmes nor anyone else can reasonably rule out other possibilities as being true or even plausible. Whatever the case, the author of the mystery novel — Arthur Conan Doyle in this case — acts in deus ex machina fashion to control the outcome of the detective’s inferences, presenting them in the story as always right. In real life, none of us has this advantage.

Regarding probability, some see the underlying role of forensic science in the courtroom to be the analysis of probability, particularly whether or not the scenario presented by the plaintiff or prosecutor is probably true5. Terms such as “preponderance of evidence” and “more likely than not” are merely expressions of statistical probability. Such expressions of probability intended for a population of cases are often misleading when applied to the facts of an individual case6. The fair administration of justice requires a more convincing approach than simply the play of probability.

The Forensic Scientific Method

There is a useful analytical process — a forensic scientific method (Table 3) — that addresses past events in a way that the scientific method could never do. It avoids scenario building and probability assessments, and it satisfies the legal and humanitarian requirement to judge someone as innocent until proven guilty. It also recognizes the ability of the forensic expert to reason from events to findings (Fig. 1) on the basis of his or her experience and expertise.

The forensic examiner, forensic pathologist or other consultant first obtains information about the incident of concern — the crime, the assault, the death, etc. The examiner learns the information from primary, “eyewitness” sources and makes initial assessments of the reliability of those witness accounts. The examiner anticipates the questions that others — family members, public health agencies, insurance agencies, law enforcement officers, attorneys, and the courts — will ask in the future. Then, the examiner performs scientific procedures — an autopsy, the retrieval of evidence from a crime scene, crime scene photography, or any other forensic science analysis such as toxicology or firearms examination — with a focus on finding answers to the anticipated questions.

Following the acquisition of physical evidence from procedures, the examiner then compares the anamnestic data — data obtained from a history or the memories of witnesses — to the physical evidence, essentially asking the question, “Are the physical findings consistent with the events related by witnesses?” If they are consistent, the witness or witnesses are reliable and truthful, even if and particularly if a witness is a suspect. If the witness accounts do not agree with the physical findings, then the witness or witnesses are not telling the truth. Some forensic pathologists use the term, “anamnestic-anatomic disharmony,” to describe this discrepancy — “anatomic” referring to the anatomic findings of a medicolegal autopsy7. During the analysis, the examiner may need to obtain more information until the quality and quantity of the information are sufficient to make assessments. Once the assessments are complete, the examiner then can offer opinions but only to a reasonable degree of medical or scientific certainty. The examiner must always keep in mind the limitations of science when rendering opinions.

This method is not new. Many well-trained, American forensic pathologists working in busy coroner and medical examiner offices have employed the same or a similar method to determine cause and manner of death for decades. Wright and Tate used a flow chart describing this method (Fig. 3) nearly four decades ago8. Except for some minor differences, this flow chart shares most of the characteristics of the forensic scientific method described in this article. It emphasizes the need to obtain anamnestic data and to correlate that data with procedural findings. Most importantly, it indicates that the hypothesis is to be generated from the anamnestic data and not the physical evidence.

The standard scientific method does not allow for the acquisition of witness information. Scientists using the scientific method generate hypotheses from the observations of physical evidence only. Once again, this does not work when dealing with past events. The forensic scientist does not form a hypothesis as a natural or physical scientist would. The witnesses and other anamnestic data provide the hypothesis, and the forensic scientist’s role is to test the veracity of that hypothesis.

Falsification and Verification:

The scientific method uses falsification, but the forensic scientific method uses both verification and falsification (Table 4). The strategy of falsification in the standard scientific method allows for other explanations for experimental results. It prevents scientists from closing their minds to other possibilities. The forensic scientific method must also use a careful approach in order to prevent an injustice. Allowing the witnesses, particularly the suspect, to shape the hypothesis and then verifying that hypothesis with further testing allows the benefit of the doubt to be with the suspect. This is compatible with the dictum of our society that allows one to be “innocent until proven guilty.”

Beyond the argument for fairness, verification is also a powerful strategy to determine the truth. This is because a suspect and most witnesses ideally are not allowed to know the forensic science findings before making statements. Also, most people do not know enough about science to form false statements that will perfectly match the findings. Armed with information from the crime scene, the crime laboratory and the autopsy suite, an interrogator may question a witness and evaluate the veracity of his or her statements without the witness being aware of the scientific findings or that such findings even exist. A sufficiently trained interrogator with the help of the scientist can easily detect a lie in this setting.

With some crimes, reliable primary witness information might not be available or obtainable. The prosecutor or plaintiff then has to rely on indirect or circumstantial evidence — much of it possibly from forensic science methods — to try a case. Proof in a case like this requires the prosecutor or plaintiff rather than the witnesses to form the hypothesis. The prosecutor then must reason more from the findings to the events rather than the opposite and must argue that a guilty verdict is correct based on high probability. The strategy used — either by the prosecutor or plaintiff as he or she considers whether or not to file charges or by the defense attorney during the trial — is falsification. All that is needed to falsify is to demonstrate convincingly how one plausible, alternate scenario can explain the physical evidence.

Cases relying on circumstantial evidence have other challenges. Arguments based on probability are subject to certain biases9. These biases come from the tendency to prove a hypothesis or scenario rather than to falsify it. Three forms of bias are particularly common. A tendency to select the first alternative that comes to mind consistent with the evidence — a bias known as satisficing — may prevent the analysis of other alternatives. Even when more than one alternative is considered, the limitations of the human mind to form and remember lists may limit the explanations to only those available to the imagination. The bias of availability limits the number of considered possibilities. Finally, if one scenario is considered and additional evidence shows it to be incorrect, the next scenario that is considered may not differ much from the first one. This is because of anchoring. The truth may not resemble either the first or second scenario at all, but the tendency of the human mind to anchor additional hypotheses to the first one will limit the considered possibilities.

The Limitations of Science:

Most in our society have a profound belief in science. Science in the courtroom is afforded greater respect than witness accounts. After a jury decision in a court case in Queens, New York, a juror explained his decision by remarking, “You can’t argue with science10.” The promulgation of television programs like CSI does little to alter that belief.

Those of us who work in the forensic sciences have a view opposite to this juror. Many of us are acutely aware of the limitations of science, particularly when that science is directed toward the recent past. Additional information in the form of new witnesses or other physical evidence can alter the interpretation of findings. Since many of the cases we consider have a high media profile, mistaken interpretations or even simple laboratory errors can have devastating consequences to one’s career and reputation.

The awful truth is that science is a human endeavor with definite limitations. This is not to say that witness accounts do not have limitations because they certainly do. A relatively recent example of this comes from the 2002 crash of American Airlines Flight 587 near the Kennedy Airport in New York City. Aircraft accident investigators noted that hundreds of witnesses saw “hundreds of different things,” relating numerous different accounts of a crash sequence that suddenly and unexpectedly took place within 93 seconds11. On the other hand, witnesses often provide much better, more detailed and more reliable data than scientists. History often trumps science, particularly if multiple reliable sources document that history.

Consider what happens when witness accounts are sparse or questionable. In these situations, scientists more often cause confusion rather than clarify the issues. Three of the most controversial and misunderstood topics in forensic pathology involve situations where witness accounts are non-existent or highly questioned. One example is the common situation where an infant is found unexpectedly dead after being put to bed. Sudden Infant Death Syndrome (SIDS) has baffled scientists and pediatricians for scores of years. With SIDS, there are no witnesses to give an account of what happened physiologically to the child. There are no electroencephalograms, telemetry, and measures of respiratory function to explain the death. Scientists do not fill the void left by the absence of a witness account; instead, hypotheses abound.

Two other controversial topics, the sudden and unexpected death of a person under police custody or the death of a child from physical abuse, are controversial because the public questions the truthfulness of the eyewitnesses — in many cases, the police or the parents. As with SIDS, scientists do not always provide clear and conclusive answers. Rather, scores of scientists line up and step forward with varying and diverse opinions. Many would be surprised how often professionally trained scientists disagree with each other.

Even technologically powerful DNA analyses are subject to limitation. Although these tests can provide exceedingly high probability ratios, giving the appearance of near certainty, analysts can make mistakes in sampling the evidence, performing the analysis, or applying the results.

The courts should use the forensic sciences to evaluate witness accounts rather than to replace them. The triers of fact in the court setting must understand the limits of science before reaching a verdict.

The Basis for Forensic Science Knowledge:

Dr. Joseph H. Davis, an American forensic pathologist and a long-time advocate of learning from the death scene, describes the 81,000 case files in the Miami-Dade County, Florida, Medical Examiner Department as chapters in a textbook for learning forensic pathology and training forensic pathologists12. Each file, with both anamnestic and anatomic data, contains unique sets of past events. Each provides the basis by which a pathologist learns to reason from the events of the past to the subsequent forensic and anatomic findings. In other words, the pathologist in training learns how the events can or cannot be “consistent with” the findings.

Each forensic pathologist or forensic scientist — in training or through practice — learns science through experience — both his or her experience and the experiences of others. Each previous case or analysis, if properly understood and interpreted, can contribute to the knowledge that the scientist applies to each new case. Also, knowledge gained from the natural and physical sciences, the experiences of others, and experiments performed in the present can be applied to any forensic science issue, provided that the information is used in the proper way and in the proper context.

As a gatekeeper for the proper application of science to a court case, the court evaluates the experience and training of the analyst before declaring him or her an expert. This is appropriate. It is also and perhaps even more appropriate for the court to evaluate the scientific method of the analyst. Do the opinions of the expert, particularly opinions that involve causation, go beyond a simple focus on the scientific data? Has the expert appropriately used the forensic scientific method, comparing witness and anamnestic data to the physical evidence? Without the proper method, the experience of the expert is not only useless but also potentially harmful.

The courts require a medical or scientific expert to express only opinions made to a reasonable degree of medical or scientific certainty. The courts in the United States have never specifically defined this term, leaving it somewhat to the discretion of the expert13. In the context of the forensic scientific method, reasonable certainty requires the scientist to reason using appropriate methodology — reasoning logically from hypothetical or real events to findings — and, having done so, to express opinions that he or she is certain are true.

Shaken Baby Syndrome — The Evolution of a Hypothesis:

The origin and development of the Shaken Baby Syndrome (SBS) provides an excellent example of what can happen when scientists and physicians apply the scientific method inappropriately to past events.

The SBS concept originated with radiologist, John Caffey. He presented a case series in 1946 of six battered babies with subdural hematomas and “traction changes” in the periosteums of long bones14. Subsequently in papers published in 1972 and 1974, he presented his hypothesis of the “Whiplash Shaken Infant Syndrome.” He believed that shaking was a common form of infant battery and the cause of intracranial and intraocular bleedings and long bone injuries in infants15 16.

To support his hypothesis, he presented 27 cases in his 1972 paper and added in 1974 a few more cases to the original 27 cases. Each case presentation was scantly detailed, and the witness information came mostly from confessions to a crime. Some of the cases involved only long bone or vertebral injuries without head injuries. One case involved a folk treatment for “caida de mollera” (sunken fontanel) where the infant, while held inverted by the ankles, was shaken vertically and slapped repeatedly in the soles of his feet. This situation differed markedly from the typical shaking scenario described in later years. Fifteen of the cases involved a nurse for infants — a heavy woman with “large hands” — who allegedly injured the infants over a period of eight years. Twelve of them never died but were “maimed.” Only two out of the three deaths had autopsy data.

From these observations and perhaps other observations that he did not mention, Caffey hypothesized that even innocent, socially acceptable and habitual practices of shaking were leading to permanent brain damage in thousands of children every day in the United States. He believed even mild shakings over time had doleful consequences to brain function.

Over time, Caffey’s hypothesis developed into the clinical entity later known as the Shaken Baby Syndrome (SBS) or Shaken Infant Syndrome (SIS). The entity has had many believers over the years, but there have been some who do not believe that children can be severely injured by shaking17. Among those who believe in it, the syndrome underwent changes not originally described in Caffey’s papers. One study discovered that more recent cases of SBS differed from Caffey’s original descriptions18. Modern forms now included cases where there was evidence of impact. These cases were also referred to as the “Shaken Impact Syndrome,” retaining the concept of shaking in the name. Although Caffey claimed, “Direct evidence of trauma through admission by the parent-assailant or the statement of a witness is rarely obtained or obtainable16,” more recent cases had more confessions to shaking than before. There were also now more male perpetrators than female perpetrators and more rib fractures than long bone fractures. Recently, several have moved away from the use of the SBS term, relying instead on more general terms that include both shaking and blunt head trauma, such as Abusive or Inflicted Head Injury19.

Some have studied the medical literature of SBS. According to Donohoe, much of the clinical literature prior to 1999 consisted of consensus papers or retrospective studies with considerable methodological constraints20. Leestma found in the literature 54 cases with admissions of shaking by perpetrators, but only 11 cases had no sign of cranial impact. The majority of the cases also had no individual case data21.

Experimental studies performed by or with biomechanical experts also generated controversy and mixed results. Two biomechanical studies provided data that did not support the concept that brain injury could be caused by shaking alone22 23, but one study indicated that brain injury from shaking alone is possible if the shaking is vigorous24. Another biomechanical study indicated that shaking severe enough to cause subdural hematomas would cause the cervical spine to fail mechanically25. Others, however, criticized this study for its design26.

Adding to the confusion, numerous observers in case reports and studies found retinal hemorrhage, an important condition for the diagnosis of SBS, associated with an enlarging number of conditions not related to child abuse27.

Shaken Baby Syndrome and the Scientific Method:

Dr. Caffey, true to the scientific method, invented a hypothesis to explain a set of physical findings that to him constituted a syndrome. He did not ask for the detailed accounts of witnesses, and he did not allow the witness accounts to form the hypotheses. Instead, he formed his own hypothesis from physical data, overlooking the fine detail from witness accounts in each case. “Shaking” became the general mechanism invoked in most pediatric head trauma cases, particularly those cases without visible external head injuries.

The standard scientific method cannot and does not work with past events, and it did not work for Caffey. Although he noted a set of clinical and radiological findings in 1946 and later, Caffey never observed the events that caused those findings. No one can observe past phenomena. He could only surmise that shaking was the only possible event in every single case capable of causing the deleterious lesions. Essentially, he selected one scenario out of the numerous possible scenarios available in widely varying cases.

The confessions of the accused did not satisfy the requirement for the detailed eyewitness testimony called for in the forensic scientific method. We have no way of knowing from the case studies presented by Caffey or many others whether or not shaking was mentioned to the accused prior to the confession. Confessions to infant shaking under those circumstances would not be surprising. Interrogators could easily lead suspects to believe that any event that included any kind of shaking, no matter how trivial, could cause deleterious consequences to the brain, particularly when Caffey and others promoted that belief. A confession in that setting would then be an expression of perceived guilt, regardless of whether or not an episode of shaking, no matter how brief or prolonged, caused any harm. Suspects could also use the shaking explanation as a way to keep from admitting the infliction of more socially unacceptable forms of head trauma. False confessions are well-known and well-documented phenomena, and numerous different social factors can lead to them during a police interrogation28. The increase in the number of confessions to shaking with time suggests that police interrogators introduced the idea to suspects more frequently with time.

In spite of the numerous experiments and studies on SBS, scientists and physicians can never determine from these studies the past events of any infant head injury case. Scientists conducting experiments and controlled observations can only use them to predict what could happen under a specific set of circumstances, not what actually happened in the case. In other words, experiments and controlled studies can only determine what is or is not possible, not what actually occurred. Over the years, experiments, particularly the biomechanical experiments, have added to our understanding of how the head and brain react to trauma. Even though they have added to our knowledge of head injury, not one of these experiments could determine the past events of any case. Only the forensic scientific method could do that with any reasonable certainty.

The forensic scientific method is not for the development of a theory, like what SBS has seemed to become over the years. Proper use of the forensic scientific method would not allow any such concept to be introduced into the analysis. Forensic scientists would need only an eyewitness account to compare with the findings, preferably one that is not encumbered by added notions or theories.

Unfortunately, the proponents of SBS today prefer to rely on shaking; however, unlike what is called for with the standard scientific method, there have been few attempts to falsify that hypothesis. Instead, the proponents have tried to promote and support it. Consequently, the biases of satisficing, availability and anchoring mentioned earlier in this paper have come into play. Infant shaking remains the first or nearly the first consideration in most pediatric head trauma cases (satisficing), and there has been little interest or capacity to consider other options (availability). Also, even though most of the cases over the years have evidence of cranial impact and even though recent biomechanical studies indicate falsification, many remain anchored to the shaking concept. Remarkably, even Duhaime and colleagues remained anchored to shaking, even though their biomechanical study did not support the shaking hypothesis. The final paragraph of their paper made this evident:

It is our conclusion that the shaken baby syndrome, at least in its most severe acute form, is not usually caused by shaking alone. Although shaking may, in fact, be a part of the process, it is more likely that such infants suffer blunt impact. The most common scenario may be a child who is shaken, then thrown into or against a crib or other surface, striking the back of the head and thus undergoing a large, brief deceleration22.

Conclusion — the Need for the Forensic Scientific Method:

Science is, or should be, about the truth. Truth in its abstract, spiritual or metaphysical forms is important, but science deals only with empirical truth. The scientific method has proven itself over time to be a reliable way to arrive at real, measurable, observable truth.

Past events, by their nature of being in the past, have passed from real to abstract. All that is in the past is now in the form of memory or record, if it is in any form at all. To arrive at the truth of past events, particularly truth that is real, measured, or observed, the anamnestic data must not be ignored or minimized. It must be coupled with the observations made at the crime scene, in the crime laboratory or in the autopsy suite. The scientific method without modification to allow for past events will only cause mistakes. Mistakes of this nature lead to injustice. Those of us in the legal and forensic science communities should never tolerate this kind of injustice or allow it to occur.

The Shaken Baby Syndrome is only one example of how a mistaken use of the scientific method for past events can lead to years upon years of mistakes and injustices. Many other examples could be cited, but there is not sufficient space in this paper to do so. We cannot calculate the injustices brought about by confused science. Some injustices in time may be remedied, but unfortunately most will not.

It is my hope that forensic scientists, law enforcement officers, the legal community, and others who scientifically analyze events from the past will uniformly embrace the modification of the scientific method for past events: the forensic scientific method.


The author thanks Joshua Renk for drawing the illustrations in figures 1 and 2 and for “cleaning up” the illustration in figure 3 from a faded photocopy.


1 Flexner SB, Hauck LC, editors. The Random House Dictionary of the English Language, 2nd ed, unabridged. New York: Random House, 1987.

2 Nordby JJ. Here we stand: what a forensic scientist does. In: James SH, Nordby JJ, editors. Forensic science: an introduction to scientific and investigative techniques. Boca Raton, FL: CRC Press LLC, 2003.

3 Moritz AR. Classical mistakes in forensic pathology (American Journal of Clinical Pathology, 1956). Am J Forensic Med Pathol 1981;2:299-308.


5 Kiely TF. Forensic science and the law. In: James SH, Nordby JJ, editors. Forensic science: an introduction to scientific and investigative techniques. Boca Raton, FL: CRC Press LLC, 2003.


7 Schaber B, Hart AP, Armbrustmacher V, Hirsch CS. Fatal pediatric head injuries caused by short distance falls [letter]. Am J Forensic Med Pathol 2002;23:101-3.

8 Wright RK, Tate LG. Forensic pathology: last stronghold of the autopsy. Am J Forensic Med Pathol 1980;1:57-60.

9 Heuer RJ. Psychology of intelligence analysis. Washington DC: Center for the Study of Intelligence, Central Intelligence Agency, 1999.

10 Neufeld PJ, Colman N. When science takes the witness stand. Sci Am 1990;262:46-53.

11 Wald ML. For air crash detectives, seeing isn’t believing. The New York Times 2002 June 23.

12 Davis JH. Medicolegal death investigation. In: Dolinak D, Matshes EW, Lew EO, editors. Forensic pathology: principles and practice. Amsterdam: Elsevier Academic Press; 2005.

13 Bradford GE. Dissecting Missouri’s requirement of “reasonable medical certainty.” J Mo Bar 2001;57(3):136.

14 Caffey J. Multiple fractures in the long bones of infants suffering from subdural hematoma. Am J Roentgen 1946;56:163.

15 Caffey J. On the theory and practice of shaking infants. Am J Dis Child 1972; 124:161-9.

16 Caffey J. The whiplash shaken infant syndrome: manual shaking by the extremities with whiplash-induced intracranial and intraocular bleedings, linked with residual permanent brain damage and mental retardation. Pediatrics 1974; 54:396-403.

17 DiMaio VJ. The “shaken baby syndrome” [letter]. N Engl J Med 1998;339:1329.

18 Lazoritz S, Baldwin S, Kini N. The whiplash shaken infant syndrome: Has Caffey’s syndrome changed or have we changed his syndrome? Child Abuse Negl 1997;21:1009-14.

19 Krous HF, Byard RW. Controversies in pediatric forensic pathology. Forensic Sci Med Pathol 2005;1:9-18.

20 Donohoe M. Evidence-based medicine and shaken baby syndrome, part I: literature review 1966-1998. Am J Forensic Med Pathol 2003;29:239-42.

21 Leestma JE. Case analysis of brain-injured admittedly shaken infants: 54 cases, 1969-2001. Am J Forensic Med Pathol 2005;26:199-212.

22 Duhaime AC, Gennarelli TA, Thibault LE, Bruce DA, Margulies SS, Wisner R. The shaken baby syndrome: a clinical, pathological, and biomechanical study. J Neurosurg 1987;66:409-15.

23 Prange MT, Coats B, Duhaime AC, Margulies SS. Anthropomorphic simulations of falls, shakes, and inflicted impacts in infants. J Neurosurg 2003;99:143-50.

24 Cory CZ, Jones MD. Can shaking alone cause fatal brain injury?: a biomechanical assessment of the Duhaime shaken baby syndrome model. Med Sci Law 2003; 43:317-33.

25 Bandak FA. Shaken baby syndrome: a biomechanics analysis of injury mechanisms. Forensic Sci Int 2005;151:71-9.

26 Margulies S, Prange M, Myers BS, Maltese MR, Ji S, Ning X, et al. Shaken baby syndrome: a flawed biomechanical analysis. Forensic Sci Int 2006;164:278-9.

27 Aryan HE, Ghosheh FR, Jandlal R, Levy ML. Retinal hemorrhage and pediatric brain injury: etiology and review of the literature. J Clin Neurosci 2005;12:824-31.

28 Conti RP. The psychology of false confessions. J Credibility Assessment Witness Psychol 1999;2:14-36.

#Table 1: Scientific method
1. Observation of event or problem related to event
2. Formation of hypothesis to explain event
3. Prediction of other events assuming hypothesis to be true
4. Testing of hypothesis and predictions through controlled experimentation or observation
5. Hypothesis either eliminated by falsification or tested further by others
6. If hypothesis supported by testing performed by others over time, it may become a theory
#Table 2: Features of reliable methods in forensic science, according to Nordby2
Help distinguish evidence from coincidence without ambiguity.
Allow alternative results to be ranked by some principle basic to the sciences applied.
Allow for certainty considerations wherever appropriate through this ranking of relevant available alternatives.
Disallow hypotheses more extraordinary than the facts themselves.
Pursue general impressions to the level of specific details.
Pursue testing by breaking hypotheses (alternative explanations) into their smallest logical components, risking one part at a time.
Allow tests either to prove or to disprove alternative explanations (hypotheses).
#Table 3: Forensic scientific method
1. Acquisition of primary witness and other anamnestic evidence
2. Anticipation of future questions
3. Acquisition of physical evidence
4. Comparison of consistency of alleged events (hypothesis) with physical findings, obtaining additional data as needed
5. Assessment only to a reasonable degree of scientific certainty, recognizing the limitations of science
#Table 4: Falsification and verification
Scientific method Forensic scientific method
Uses falsification Uses verification for witness evidence
Uses falsification for circumstantial evidence

#Figure 1

FIG. 1: Events (E) and Physical Findings (F) from the perspective of the events. Observable physical findings can be judged as “consistent with” or “not consistent with” a set of hypothetical or actual events if the events are stated or known.

#Figure 2

FIG. 2: Events (E) and Physical Findings (F) from the perspective of the findings. Observable physical findings may be “consistent with” numerous different sets of events, making it impossible to discern the actual events from the findings.

#Figure 3

FIG. 3: Originally from the article by Wright and Tate with the original legend: “Forensic pathologist’s approach to arriving at the answer to the question, ‘Why did he die?'” Used with permission.