If you want to understand the artist, look at his work.
—JOHN DOUGLAS, former head of the FBI’s Behavioral Sciences Investigative Support Unit
In a macabre way, looking at the changing face of a serial killer is like looking at the changing face of America . . .
—ROBERT D. KEPPEL, criminal investigator
In 1991, the Jodie Foster movie The Silence of the Lambs, based on the Thomas Harris bestseller, claimed that the FBI had a center nestled deep in a forested region near Washington D.C., where specially trained FBI agents ensconced in underground chambers did nothing but think about and analyze hundreds of unsolved murder files and gruesome crime scene photographs from every part of the United States. The movie proposed that they attempt to enter the minds of unidentified killers to produce a profile—a hypothetical description of a suspect based on his psychology. That is precisely what the FBI does in real life—at Quantico, Virginia, about thirty miles outside the capital.
Quantico is a fortified U.S. Marine Corps base where the FBI training academy is located. The Behavioral Sciences Unit (BSU)* is based at the training facility and was originally established to train agents how to negotiate with and predict the behavior of hostage takers. It was in the BSU that the FBI’s system of analysis of serial killers and rapists was first developed in the late 1970s, primarily by FBI agents Robert K. Ressler, John Douglas, and Roy Hazelwood.
At Quantico, buried beneath the ground in a bomb shelter once built for postnuclear intelligence operations, the National Center for the Analysis of Violent Crime (NCAVC) is housed. FBI agents specialized in profiling offenders, known as criminal investigative analysts, study crime data from suspected serial homicides, kidnappings, terrorist acts, and other serious and violent offenses.
In most serial homicides, FBI agents do not actively participate in the investigation, secure evidence, or pursue the suspect—that is the responsibility of the local police agency. Nor is the FBI called in if serial homicides occur in different jurisdictions—that is a myth. The FBI analysts act in an advisory capacity, only at the request of a local police department that submits a standard, thirteen-page Violent Criminal Apprehension Program (VICAP) analysis report to the FBI. The data from the VICAP report is fed into a computer known as Profiler, and the output of the computer is then elaborated on by the analysts in the form of a profile before being sent back to the local police department. Although generically called profiling, the technical term is criminal investigative analysis (CIA).
FBI analysts sometimes travel to the scene of a crime or assign one of a team of specially trained local FBI agents, known as field profile coordinators, to work at the scene. The average FBI agent is fairly well educated—a university degree is required of recruits. The agents at NCAVC often hold postgraduate degrees in various social sciences—they are highly trained, highly experienced, and highly educated. They claim to be formidable “mind hunters” and “monster fighters.”
The NCAVC is a relatively new program, formally established by order of President Reagan in June 1984. In the wake of the serial killer “epidemic” described earlier, it took twenty-seven years to develop and implement the NCAVC from the time that an LAPD homicide detective named Pierce Brooks first thought of it.
Aside from fictional investigative profilers such as Arthur Conan Doyle’s Sherlock Holmes or C. Auguste Dupin in Edgar Allan Poe’s “The Murders in the Rue Morgue,” the use of psychological profiling of unidentified criminals is a relatively new technique. It was probably first used to profile one of history’s most notorious criminals, Adolf Hitler. During World War II, the OSS—the precursor of the CIA—commissioned psychiatrist W. C. Langer to construct a psychodynamic personality profile of Hitler with the purpose of anticipating decisions that Hitler might make under certain circumstances. Langer’s profile turned out to be relatively accurate, down to his prediction of Hitler’s suicide at the end of the war.198
Starting from 1940, for sixteen years, New York City was plagued by a series of bomb explosions in public places. Some thirty bombs had been planted, although not all had exploded. In 1956, the bombings began to escalate—the police were worried that it was only a matter of time before somebody would be killed. The individual was now leaving bombs inside movie theater seats, which he slashed open with a knife from underneath and stuffed the device inside.
From letters that the “Mad Bomber” left behind and sent to newspapers, it was obvious that he was a disgruntled former employee of Con Edison, New York’s electric power company—but there were records for thousands of such former employees and the police did not know where to begin their search. It was at this point that the New York City police turned for help to psychiatrist and New York State Assistant Commissioner for Mental Hygiene Dr. James A. Brussel.
After reviewing the police files, Brussel told the astonished police officers, “He’s symmetrically built. Perpendicular and girth development in good ratio. Neither fat nor skinny.” From the notes and the behavior, Brussels had concluded that the Mad Bomber was suffering from paranoia—an obvious deduction. From there, however, Brussels turned to his knowledge of Ernst Kretschmer’s study of some 10,000 patients in mental hospitals. Kretschmer found that 85 percent of the paranoiacs had what he described as an “athletic” body type. Thus Brussel concluded that if the Mad Bomber was a paranoiac, then indeed the chances were 17 in 20 that his build was symmetrical. (A recent study of serial killers, who are not all paranoiacs, concluded that 68 percent had height and weight in good proportion to one another.)199
As paranoia develops slowly and usually erupts in full force when the subject is near thirty-five years old, Brussel concluded that the Mad Bomber, after sixteen years of activity, must be now be middle-aged.
From the Mad Bomber’s neatly printed and orderly letters, Brussel deduced that he was a very neat and “proper” man. He was probably an excellent employee, always on time and always performing the best work. From the workmanship on the bombs, Brussel concluded that he was trained as an electrician or pipe fitter. He was probably always neatly dressed and clean-shaven, and was not involved in any disciplinary infractions at work. If he had been dismissed from work, Brussel concluded, it would be for medical reasons and not for work performance or disciplinary problems.
The language of his letters showed a fairly well-educated man but either foreign born or somebody who lived among foreign-born people—there was a certain European stiffness to his writing and a lack of American slang and colloquialisms. It was as if the letters were written first in a foreign language and then translated into English. Brussel felt that because of his choice of both a knife (to slash the movie seats) and a bomb, the Mad Bomber was probably a Slav, and by his regularity, probably a Roman Catholic—very likely Polish. His letters were posted from either New York or Westchester County. He was too smart, Brussel felt, to post letters from where he lived—he likely posted them on his way to New York from somewhere else nearby. Plotting a line from New York through Westchester County, Brussels arrived at Bridgeport, Connecticut, where he knew there was a large Polish community.
Looking at the hand-printed letters, Brussel saw that every letter was perfectly formed, as expected of a paranoiac who would make every effort to avoid any appearance of a flaw, except for one letter—the W—which was formed like a double U; to Brussel it symbolized a pair of female breasts. Brussel was troubled by this sloppy anomaly in the Mad Bomber’s otherwise neatly printed alphabet. Brussel was also struck by the Mad Bomber’s use of a knife to slash the movie theater seats, also a crude and sloppy action uncharacteristic of his ordered and neat personality. Brussel attributed sexual motives to the two factors and suggested that the Mad Bomber’s mother was probably dead and he lived either alone or with an older female relative; he was a loner, had no friends, and was single—but not a homosexual.
Finally, Brussel proposed that the Mad Bomber would be dressed conservatively and would avoid newer-style clothing. He told the police that when they arrested the Mad Bomber, he would probably be wearing a double-breasted suit and it would be buttoned.200
Armed with this information, the police again began to review the Con Edison company files for ex-employees who were skilled, were living in Connecticut with Polish-sounding names, might be in middle age, and were dismissed for medical reasons.
That’s how the name of George Metesky cropped up, a generator wiper who was involved in an accident at an electrical plant in 1931. Metesky filed for disability payments but doctors could find nothing wrong with him. He stopped coming to work and was dropped from the payroll. He filed for worker’s compensation in 1934, claiming that the accident gave him tuberculosis. After being turned down, he spent three years writing bitter letters to Con Edison. In 1937 he stopped writing, and in 1940 the first bomb exploded.
Metesky’s work record indicated that before the accident, he was an excellent employee—meticulous and punctual. He was a Roman Catholic and lived in Waterbury, Connecticut, in a house with his two older sisters. He was unmarried and both his parents were dead. He was known in the neighborhood as Mr. Think, because sometimes when he was greeted he would not respond, appearing to be in deep thought. Otherwise, he was considered an inoffensive loner. When the police questioned him, he quickly confessed that he was the Mad Bomber. When arrested, he was wearing a blue double-breasted suit—buttoned. Metesky was hospitalized in a psychiatric facility.
Following the Metesky case, police began to use the services of psychiatrists more often to assist them in their investigations. The problem, however, was that each psychiatrist based his profiles on his own knowledge and experience, and when the police pooled together several psychiatrists to assist them on the Boston Strangler case, they all came up with radically different profiles. There was no systematic approach to profiling other than that of each individual psychiatrist.
Former LAPD homicide detective Pierce Brooks is considered the modern pioneer of VICAP crime data collection and case linking. In 1957, Brooks, a former naval officer and blimp pilot, became convinced that two separate homicides in the Los Angeles area might have been committed by the same individual. Surmising that if one killer had murdered two victims, he might have killed three or more as well, Brooks spent weeks thumbing through old crime files and newspaper reports at the public library. Brooks realized that he needed a computerized database of unsolved homicide descriptions, a fairly astute idea in a time when most people were unfamiliar with computers and what they can do.
Brooks proposed that every police department in California have a computer terminal linked with a central mainframe that stored data on every unsolved homicide. Every department would have access to data on homicides in other jurisdictions. Aside from the fact that most of Brooks’s superiors didn’t know what a computer was, back in 1957 computers were huge elephants costing millions of dollars. Nobody took his proposal seriously. Brooks put the idea aside and went on with his career to become the head of the Los Angeles homicide unit, and later a chief of police in Oregon and Colorado during the 1960s and 1970s.
By the late 1970s, Robert Ressler, an FBI instructor in hostage negotiation at the BSU in Quantico, was beginning to feel restless. Realizing that taking a systematized approach and understanding of the psychology of hostage takers was vital to police responses, Ressler came to believe that the same applied to what was then called “stranger killings”—murders where the victim and perpetrator did not know each other. These very difficult-to-solve murders were on a dramatic rise in the 1970s. On his own time, Ressler and an associate from the BSU, agent John Douglas, began to interview convicted killers, asking them why they thought they had killed. When the FBI hierarchy heard about Ressler and Douglas’s visits to various prisons, they quickly put a stop to them, informing the agents that “social work” was not the FBI’s objective. Fortunately, it was not long before the old-style FBI leadership retired, and Ressler began to lobby for a research program into the minds of multiple killers.
Ressler and Douglas managed to reactivate the interview program and systemize the data that emerged from the encounters. Two important textbooks emerged out of those interviews, authored by Ressler and Douglas along with Ann Burgess, a professor of psychiatric nursing. These books are now considered the bibles of crime analysis: Sexual Homicide: Patterns and Motives and the Crime Classification Manual. At the same time, Douglas and Ressler began to train other FBI agents in the techniques of profiling an unknown subject—an “Unsub” in BSU terminology.
For Ressler and Douglas, however, the problem remained as to how to collect, analyze, and distribute the data on various unsolved homicides throughout the United States. While exploring the possibility of a U.S. Justice Department research grant, Ressler learned that Pierce Brooks had retired from active police work and had just received a small grant of $35,000 to revive and update his original 1957 proposal of an unsolved homicide computer data network. Brooks was now based at Sam Houston University in Texas, and Ressler went down to see him. The following year, Brooks and the FBI joined forces and got a $1 million research grant to design a data collection, analysis, and distribution system that became known as VICAP. While FBI profilers began working in the field around 1979, VICAP and profiling research and analysis was formalized in the NCAVC only in 1984. It had taken twenty-seven years for Brooks to see his idea transformed into reality.
While the thing itself has been around a lot longer, the expressions serial killer and serial murder came into popular use only in the mid-1980s. Prior to that, the terms mass murderer, lust murderer, chain murders, stranger killing, and recreational or thrill killing were used. Recently the term sequential predation has been gaining currency in academic literature. Nobody has precisely defined what the term serial killer means. The definitions of serial killers and serial murder vary from expert to expert and include the following:
The term serial killer might have been coined by Robert Ressler, who says that he felt the term stranger killings was inappropriate because not all victims of serial killers were strangers. Ressler says he was lecturing in England in 1974 at Bramshill, the British police academy, where he heard the description of some crimes as being in series—a series of rapes, arsons, burglaries, or murders. Ressler says that the description reminded him of the movie industry term for short episodic films shown on Saturday afternoons during the 1930s and 1940s: serial adventures. Each week, juvenile matinee audiences were lured back to movie theaters, week after week, by an inconclusive ending in each episode—a so-called “cliffhanger.” Ressler recalled that no episode had a satisfactory conclusion, and every time one ended, it increased rather than decreased the tension in the audience. Likewise, Ressler felt, serial killers increase their tension and desire after every murder to commit a more perfect murder—one that is closer to their fantasy. Rather than being satisfied when they murder, serial killers are instead agitated toward repeating their killing in an often-unending cycle. That, Ressler maintains, was a very appropriate description of the multiple homicides that he was dealing with and is the history behind the term serial killer.207 (Some dispute Ressler, claiming that others used the term earlier. Michael Newton says author John Brophy used the term in 1966 in his book The Meaning of Murder.208 Ann Rule credits Pierce Brooks with coining the term serial killer.209 But in the text of Rule’s 1980 seminal book on Ted Bundy, The Stranger Beside Me, which introduced the current concept of a serial killer, the term itself never appears. Bundy is still described as a “mass murderer.” It does seem, however, that Ressler at least introduced it into popular usage in the law enforcement community, even if others might have used the term previously.)
Today a local police department in the United States or Canada can fill out a VICAP form and submit it to the NCAVC for analysis of a series of unsolved murders, people missing under suspicious circumstances, or unidentified bodies. The VICAP data is then entered into the Profiler computer system. It is an artificially intelligent system, meaning that the computer program has been taught to reason like a human being. Of course, all artificial intelligence systems are limited by the number of rules by which the computer reasons. These rules, which were identified through years of interviews with serial killers and studies of homicide case data, are expressed as lines of instructions using “if-then” operators—for example, “if the victim is tied with duct tape, then the offender might be an organized personality; if the offender is an organized personality, then souvenirs of the murder might have been taken.” These lines of reasoning refer to each other and can be assembled in millions of different combinations by the computer, as would a trained human mind when reasoning out a problem. In 1990, Profiler was operating under 270 rules; however, every time the NCAVC research arm learns something new about the behavior of serial offenders, the information is programmed into Profiler in the form of a new line.
Profiler is not a substitute for human analysts—it is a tool. Its output is always analyzed and reviewed by a human profiler. Profiler is used to statistically pinpoint human conclusions. Sometimes Profiler and the human analyst disagree, and the data are reviewed—perhaps a mistaken piece of data was entered into the computer, or perhaps the human analyst forgot something in his or her assessment. Human error in the analysis can be identified, because unlike humans, a computer never forgets.
The FBI system of criminal profiling, called criminal investigative analysis (CIA), is essentially a six-step process. Step One, profiling inputs, involves gathering as much of the following data as possible: a complete investigative report, crime scene photographs, information on the neighborhood or area, the medical examiner’s report, a map of the victim’s movements before death, and background on the victim.
In Step Two, constructing decision process models, the basic nature of the homicide is established: whether the killing was a serial or single murder; whether the homicide was the primary objective of the offender or the result of another crime; whether the victim was a high-risk victim (prostitute or hitchhiker, for example) or low-risk, such as a middle-class married female. Other factors in building the model include whether the victim was killed at the scene or elsewhere and the nature of the location where the homicide might have taken place.
Step Three, the crime assessment, is the stage at which the FBI profiler decides whether the crime scene has the characteristics of an organized or disorganized offender, or a mix of both. This organized/disorganized dichotomy is both controversial in current debates on the effectiveness of criminal profiling (discussed later) and unique to the FBI system of profiling. It is primarily based on the already mentioned study of thirty-six sexual murderers conducted in the early 1980s. Factors such as whether the victim’s body was positioned, whether sexual acts were performed before or after death, insertion of foreign objects into the body, vampirism, mutilation, keeping of the corpse, age of the victim, and use of weapons or restraints all statistically characterize either an organized or disorganized killer. This categorization of offenders is the heart of the FBI approach.
According to this system, organized offenders have a tendency to:
As individuals, organized killers are likely to be of above-average intelligence, attractive, married or living with a spouse, employed, educated, skilled, orderly, cunning, and controlled. They have some degree of social grace and often talk and seduce their victim into being captured. They are more likely to appear physically attractive to their victim, and their victims are often younger. They are difficult to apprehend because they go to inordinate lengths to cover their tracks and are often forensically aware. Their family background is more stable, their father was more likely to have been gainfully employed, and they are of higher birth order, often the eldest of siblings. When they commit their crimes they are likely to be angry or depressed and under the influence of alcohol, and their offense is often triggered by some kind of personal stress such as divorce, loss of a job, or financial crisis. They are likely to follow the media reports of the crime and are prone to change jobs or leave town after the crime.
Disorganized offenders, on the other hand, are more likely to:
The disorganized offender is likely to be of lower birth order, come from a unstable family, and have been treated with hostility by his family. Parents often have a record of sexual disorders; the offender himself is sexually inhibited and sexually ignorant and may have sexual aversions. They are more likely to be compulsive masturbators. They often live alone and are frightened or confused during the commission of their crime. They commit the crime closer to their workplace or home than organized offenders and they are likely to kill victims they know and who are older than them. They often “blitz” their victims—use massive sudden force—to overcome and capture them. They are unlikely to stalk their victims for any length of time or to talk or seduce their victims into capture. Disorganized offenders are very likely to be heterosexual, be uneducated, and leave forensic evidence behind. They are unlikely to transport their victims from where they first encountered them to another location. Stress is not a major factor in triggering their crime, but significant behavioral changes may be evident in their daily conduct.210
A crime scene can also be mixed when there are multiple offenders of different personality types or when the offender is immature or undergoing psychological transformation midway in his killing career.
In Step Four of the FBI system, the criminal profile, the profiler offers a report describing the personality and possible physical and social characteristics of the unknown offender, listing objects that might be in the possession of the suspect, and sometimes proposing strategies for identifying, apprehending, and questioning him. The report can vary in length from a mere few paragraphs to pages, depending on how much information the profiler had to analyze.
Step Five, the investigation, depends on whether the report is accepted by the local police department and introduced into the investigative strategy. Using the profile data, police can narrow the list of likely suspects or adopt strategies likely to lead to an arrest. If additional crimes occur or more evidence or information is discovered, the profile can be updated and expanded during this stage.
Finally, Step Six is apprehension. The success or failure of a profile is compared to the identity of the actual offender in custody. The results are added to the variable database and the interpretive algorithms used by profilers are tuned, with the objective of further enhancing profiling accuracy in the future.
French criminologist Edmund Locard, in the early twentieth century, determined what is today called Locard’s Principle of Exchange: “Anyone who enters the crime scene both takes something of the scene with them and leaves something of themselves behind.” Locard was, of course, referring to physical evidence—fingerprints, hair, fibers, blood, soil, and so forth—but his rule also applies to methodological and psychological imprints left by the offender at the crime scene.
A criminal may leave one or both traces of the following behavioral processes: MO (modus operandi, or method of operation) and signature, the personal imprint of the offender. While every crime has an MO, not all crimes will have a signature, or at least, one immediately visible. Recognizing and differentiating signature from MO is one of the profiler’s essential tasks.
The MO is what the offender needs to do to commit his crime—break a window, cruise the highways looking for hitchhikers, pose as a fashion photographer. It is a learned and changing behavior—the repeat offender constantly alters and refines his MO to suit circumstances and to reflect things he learns. He may learn, instead of breaking a window, to seek out an unlocked door first; instead of using rope to tie a victim, the offender may discover that it is easier to use handcuffs. At its core, the MO is dynamic and changing, because it reflects outside circumstances that affect the offender’s behavior.
The signature, on the other hand, comes from within the offender and reflects a deep-rooted fantasy that he has of his crime—it is something that the perpetrator does for himself, regardless of the necessities of the crime. Police instructor Max Theil, when teaching investigators to recognize signature, says, “You should ask yourself what the offender did at that crime scene that he did not have to do.”
The signature is always static because it emerges out of a long history of deep-rooted fantasies that the offender has evolved long before he committed his first crime. He brings those fantasies with him to the crime scene. The essential core of the signature is always the same, but its expression can be evolving and changing, and extraneous circumstances can affect the character or even presence of the signature (for example, if the offender is interrupted during the commission of the crime and escapes before being able to carry out an act his fantasy might require).
Canadian serial killer Paul Bernardo’s MO, for example, changed during his career. His first victim was the teenage sister of his fiancée, Karla Homolka; he drugged her into unconsciousness with Homolka’s help on Christmas Eve and then raped her while Homolka videotaped. The girl died when an acute reaction to the drug combined with alcohol caused her to vomit and she choked to death while unconscious. His second murder was a crime of opportunity—he unexpectedly encountered his teenage victim, offered her a cigarette back at his car, and then overpowered her. His third victim was carefully stalked with an accomplice, Homolka, who helped him lure a fifteen-year-old girl to the car, where she was overcome by force. These were three different MOs that reflect three different sets of circumstances. Bernardo’s signature, however, remained the same: wholesome teenage girls as victims, all videotaped during their rapes, the same kind of content to his verbal interaction with the victims (called scripting)—“I’m the king.” “Call me master.” “Tell me you love me.” Those are signatures emerging out of things that Bernardo needed to do for himself, rather than for the purpose of the crime.
The dismemberment of one of the victims, for example, was something that Bernardo felt was necessary to conceal his crime—it was part of his MO. It has a dynamic learning curve: Bernardo’s next victim was not dismembered, because Bernardo learned that it was a disgusting task and concluded that it was not necessary. On the other hand, the dismemberment by California serial killer Edmund Kemper of his victims was solely for his own fantasy fulfillment—it had nothing to do with the necessity of hiding the bodies; it was all signature. Thus the same action, in this case dismemberment of a corpse by one offender, could be MO, while by another offender it could be signature. A vital stage of profiling involves telling the two apart.
Changes in MO are usually dynamically unpredictable and erratic; the offender may use a handgun with one victim but a knife with another. Changes are directed toward refining the ease with which the crime is committed. Changes in the signature are usually constant—they are directed toward elaborating, refining, or escalating the same core fantasy theme. Had he been able to continue killing, Bernardo probably would have kept his victims alive for longer periods of time before killing them, and would have introduced more evolved scenarios into his videotaped rape sessions. The direction an offender takes in his signature, therefore, is more predictable, and a successful profile in that area can significantly aid the police in apprehending the killer.
The profiler may also encounter deliberate alterations of the crime scene or of the victim’s body position. If these alterations are for the purpose of confusing or misleading the investigators, they are part of the MO and are called staging; if on the other hand, the alterations are for the fantasy needs of the offender, they are called posing.
A victim is sometimes posed to send some kind of message to the world—a victim’s body might be flung beside a No Dumping sign, or as in the case of Henry Lee Lucas, near the gates of a prison. Danny Rolling, the Gainesville Ripper left the head of one of his victims on a fireplace mantel, facing the door. Albert DeSalvo, the Boston Strangler, left his victims posed in positions exposing their genitals toward the door. The intention in both cases was to shock whoever discovered the bodies. William Lester Suff, convicted of killing thirteen women between 1988 and 1991, mostly street prostitutes in Riverside, California, left some of the bodies next to Dumpsters with their arms turned outward to expose needle track marks, clearly a message on the worth of the victims. Identifying and distinguishing staging and posing is part of the profiler’s task.
Later in this chapter we will see that there is a significant amount of criticism of the lack of scientific method behind the FBI’s profiling. But when it works, it works well. Some of the successful profiles have resulted in spectacular arrests.
The earliest cases of profiling by FBI Behavioral Sciences Unit agents were in the late 1970s. Robert Ressler was among the first agents to take the theories of profiling developed at the BSU out of the woods at Quantico and into the field of actual crime investigation.
On Monday, January 23, 1978, a local police department in a small town just north of Sacramento, California, called the FBI, asking for assistance with a homicide clearly outside the range of their usual experience. David Wallin had returned home at about six in the evening and found his pregnant twenty-two-year-old wife, Terry, dead in the bedroom.
Terry Wallin had apparently been attacked in the living room of the house as she was about to take out the garbage through the front door. There were signs of struggle all the way from the door to the bedroom. Two .22-caliber shell casings were found by the door. The murder had occurred some time before 1:30 that afternoon.
Terry had been shot four times through the head. The victim’s sweater, bra, and pants were pulled away from her torso, exposing her midriff. There were numerous knife wounds, but the principal wound was a cut extending from her chest to her navel. A portion of her intestines protruded through the wound while several other body organs had been taken out of the body cavity and cut. Several external body parts were missing. The killer had stabbed the victim in the left breast several times, and apparently moved and twisted the knife around. Animal feces were found stuffed in the victim’s mouth. A yogurt container had been used to collect the victim’s blood, and by the traces on the container, the killer had used it to drink the blood. Nothing appeared to have been stolen from the house.
Two days later, a disemboweled dog was found in the same neighborhood, shot dead by the same weapon used in the Wallin murder.
FBI agent Robert Ressler’s original profiling notes were as follows:
White male, aged 25–27 years; thin, undernourished appearance. Residence will be extremely slovenly and unkempt and evidence of the crime will be found at the residence. History of mental illness, and will have been involved in the use of drugs. Will be a loner who does not associate with females, and will probably spend a great deal of time in his own home, where he lives alone. Unemployed. Possibly receives some form of disability money. If residing with anyone, it would be with his parents; however, this is unlikely. No prior military record; high school or college dropout. Probably suffering from one or more forms of paranoid psychosis.211
The killer was obviously a “psycho” in the full clinical sense of the word—a paranoid psychotic or paranoid schizophrenic, somebody who was not in control of himself and suffering from delusions. (George Metesky, the Mad Bomber, was diagnosed as a paranoiac—a mental disorder distinctly different from a paranoid.)*
Ressler explained the reasons behind his profile as follows. The offender was white, because the victim was white and lived in a white residential area; statistically, most sexual attacks, he believed, are intraracial—white on white, black on black. Moreover, a black person would attract attention in a predominantly white neighborhood.
The disorganized mutilation of the victim indicated paranoid schizophrenia, a disease that often begins to slowly manifest itself around age fifteen and takes about ten years to become a full-blown psychosis. Thus the offender was probably around age twenty-five. The chances of his being older and not having committed similar acts previously would be slim.
Using Kretschmer’s body studies of mental patients, which psychiatrist Brussel used in his profiles, Ressler projected that the killer was thin. That would make sense, as an introverted schizophrenic would be so focused on his obsessions that he would skip meals—and would also ignore his personal appearance, not caring about neatness or cleanliness. Since nobody would want to live with an individual like that, Ressler surmised that the offender lived alone. His severe mental problems precluded him from finishing school, military service, or employment—he was probably surviving on a disability check. The fact that he was unemployed was further suggested by the time of the murder—midday on a weekday. Ressler felt that the murderer probably lived somewhere near the crime scene and walked away—his mind was too disordered to drive. If the killer owned a vehicle, it would be in a state of disorder—dirty, poorly maintained, filled with trash inside. As no other crimes with similar mutilation were on record, Ressler felt that the murderer had committed his first homicide—but that there would be more to come and that the killing would escalate in intensity.
The police traced Terry Wallin’s movements that Monday morning and determined that she had cashed a check at a shopping mall within walking distance to her house. Police speculated that she might have been followed from the mall.
Four days later, on a Thursday, the killer struck again less than a mile away from the Wallin murder scene. Neighbors discovered three bodies at about 12:30 in the afternoon—Evelyn Miroth, age thirty-six; her six-year-old son; and Daniel Meredith, a family friend, age fifty-two. Miroth’s twenty-two-month-old nephew, whom she was babysitting, was missing as well.
The males were all shot dead, but Evelyn Miroth was even more severely mutilated than Terry Wallin. There were two crossing cuts across her abdomen, through which her intestines protruded. Her body was covered in stab wounds, some particularly aimed at her face and anal area. Rectal swabs of her anus indicated the presence of copious amounts of seminal fluid. The baby’s pillow was stained in blood and a shell casing was found in the playpen. The bath was filled with bloodied water as well as brain and fecal matter. It appeared that the killer had washed the infant’s body afterward. A ring of blood on the floor left by a bucket-type container suggested that blood had been collected at this scene as well. Nothing had been stolen except for Daniel Meredith’s wallet. The killer had escaped in Meredith’s car, but it was found abandoned a short distance away with the keys still in the ignition.
In Ressler’s mind, the crime scene reinforced his first profile—Ressler stressed that the killer was severely mentally ill and was totally disorganized in his actions. Ressler felt that he had probably parked the stolen car near his house and walked home, where he lived alone. He might have been drinking blood from other sources before he began to murder, so Ressler suggested that the police begin an intense survey of the area of the two homicides, asking witnesses if they had seen an unkempt, gaunt individual with bloodstains on his clothing.
On Saturday morning, during the survey, a woman told the police that she had encountered such a person in the shopping mall near the scene of the first murder, about two hours before Terry Wallin’s death. He was a former classmate from high school and she was shocked by how gaunt and unkempt he had become. His mouth was caked in a yellow crust and there were bloodstains on his shirt. He had recognized her and asked her about her boyfriend, who had died in a motorcycle accident. When she got into her car, he attempted to get in with her on the passenger side, but the door was locked and she drove away, disturbed by how sick her former schoolmate had become. He was so sick that at first she didn’t recognize him and had to ask his name: Richard Chase.
The police immediately checked on Chase’s address—it was less than a block away from where Meredith’s car had been abandoned. When the police approached the twenty-eight-year-old Chase, he began to run and was quickly apprehended. He was carrying the .22-caliber handgun, which matched the one used in the murders, and in his back pocket was Daniel Meredith’s wallet. He was also carrying a box filled with blood-soaked rags.
As Ressler had projected, Chase’s truck was a mess, filled with old newspapers, empty beer cans, milk cartons, and rags. In a locked box, the police found a twelve-inch butcher knife and a pair of rubber boots caked in blood. Chase’s apartment, where he lived alone, was in complete disorder, covered in dirty clothing, some of which was stained with blood. There were food blenders with blood in them, and dishes in the refrigerator with human body parts. The partially consumed body of the missing infant was not found until months later, not far from where Chase lived. Police also found numerous cat and dog collars, matching descriptions of pets reported missing in the area.
Ressler’s profile had been almost correct—his only mistake was stating that Terry Wallin was the first victim; she was in fact the second. A few weeks earlier, some distance away from the other murders, Ambrose Griffin had been shot through the chest as he was taking groceries from his car. The .22-caliber bullet matched Chase’s weapon. Chase had fired from his truck and immediately driven away before any witnesses appeared on the scene.
The profile that the FBI constructed of Chase was instrumental in his capture. Armed with a specific description of a killer that nobody had seen, the police were able to narrow down the search for a suspect and capture him within five days of Terry Wallin’s murder.
Richard Chase’s story is somewhat sad—he was born a happy and sweet child in 1950 in a middle-income home. When he was twelve, however, Chase’s mother began to exhibit mental problems, accusing her husband of infidelity, poisoning her, and using drugs. In high school, Chase was an average student with a few girlfriends but no really close friends. His girlfriends remember that he attempted intercourse with them but could not sustain an erection.
At about the second year in school, his personality suddenly transformed, and he became “rebellious, defiant, and unkempt.” He would have been about fifteen then, when symptoms of paranoid schizophrenia often begin to first surface. (On the other hand, “defiant and unkempt” describes an entire generation during the 1960s—the hippies.) He began to hang around with “acidheads”—kids who took a lot of LSD—and in 1965 he was arrested for possession of marijuana. Again, Ressler was correct in his profile that the killer would have a drug record. But Ressler soberly reminds us, “Many in the public saw evidence for attributing Chase’s murders to the influence of drugs. I disagree. Although drugs may have contributed to Chase’s slide into serious mental illness, they were not a real factor in the murders; we have found that drugs, while present in many cases, are seldom the precipitating factor in serial murders; the true causes lie much deeper and are more complex.”
Chase managed to graduate from high school in 1969, but could not keep up with his studies in college or hold down a job. In 1972 he was arrested for drunk driving, and in 1973, after an incident at a party, he was charged with resisting police and carrying a concealed handgun without a license. He paid a $50 fine.
In 1976, Chase was put in a nursing hospital after he was found injecting rabbit blood into his veins. Because his parents could not afford mental treatment for their son, state conservators were appointed for Richard Chase. At the hospital, Chase was reported to be a “frightening” patient and known among the staff as Dracula. He often went out into the hospital garden and captured birds in the bushes. Afterward he bit off their heads and drank their blood. Two nurse’s aides quit their jobs rather than continue working around Chase.
Chase explained to staff at the hospital that his blood was turning to powder and he needed to drink blood to stave off death. When psychiatrists suggested that Chase’s problems could be controlled by drugs and that he should be released, the nursing staff at the hospital howled in protest. In 1977, Chase was released anyhow—into the custody of his mother, who had believed that her husband was trying to poison her. She rented Chase an apartment to live in.
People who had encountered Chase, who was by then twenty-seven years old, remembered that he seemed to live entirely in the past—he talked of events in high school as if they had happened yesterday. He also talked a lot about UFOs and a “Nazi crime syndicate” from high school that was still pursuing him.
In August 1977, Chase was driving near Lake Tahoe when police pulled his truck over in a routine stop. He was covered in blood and in his truck was a bucket of blood and several rifles. Police determined that the blood was bovine and let him go. In September, Chase had an argument with his mother and killed her cat. Twice in October, he bought dogs from the animal shelter for fifteen dollars each. In mid-November, he answered an ad in a local paper for puppies and managed to bargain for two puppies for the price of one. Police, meanwhile, were getting a barrage of missing pet reports in Chase’s neighborhood.
On December 18, after filling out a form stating that he had never been a mental patient or a criminal, Chase purchased a .22-caliber handgun. He drove around for several days shooting at people’s houses until he killed Ambrose Griffin.
On January 16, 1978, he set a garage on fire. On January 23, the day he killed Terry Wallin, he wandered through the neighborhood, entering any house where the door was unlocked. Several people called the police, but he was gone by the time they arrived. In one house that Chase entered, he defecated on a child’s bed and urinated in a clothes drawer. He rambled on this way, passing through the shopping mall where he encountered his former schoolmate and eventually arrived at Terry Wallin’s home.
Mumbling about UFOs following him, Chase was sentenced to death, much to Ressler’s regret, who felt that Chase was truly insane and did not know the nature of the acts he was committing. The death sentence, however, reflected the community’s fear that convicts like Chase might be released after being declared “cured” by psychiatrists—a fear that was well founded considering how many serial killers had records of being confined and released (such as Ed Kemper and Henry Lee Lucas). In 1980, Chase accumulated a surplus of depressants and took them all. He was found dead in his cell.
Psychotic serial killers like Chase can be both difficult and easy to capture—difficult because they strike like lightning with no rhyme or reason and are therefore highly unpredictable; easy because they make little effort to conceal their acts. They are, however, rare—only a small percentage of serial killers are truly suffering from psychosis.
Criminal profiling is not intended to “solve” a case or produce a definitive identity of the perpetrator. Profiling is a contributory tool that can filter out less likely scenarios and suspects or focus the investigation on fewer possibilities. It can be used to “trigger” witness recollections by providing a starting point. In the Chase case, when police asked witnesses specifically if they had seen an unkempt, gaunt individual, Chase’s name surfaced. The witness might not have come forward with his identity had police been more general in their survey.
Probably the first high-profile case FBI analysts became involved in was the Atlanta child murders of the early 1980s. Between July 1979 and May 1981, at least twenty-eight black children and youths, mostly males, were found dumped in various locations around Atlanta. Because there was some discussion as to whether these crimes were committed by a white supremacist group, eventually the federal government and the FBI became involved in the investigation. FBI agents John Douglas and Roy Hazelwood from the BSU came down to Atlanta to assist in profiling the killer.
They thought it was unlikely that a white supremacist group was committing the murders because the bodies were partially hidden, dumped in abandoned buildings or in the backs of vacant lots. A more public display of the bodies would have served white supremacists better. Because the crimes were committed in black neighborhoods, it was likely that the offender was black. A white killer would have stood out in the neighborhood and would have been noticed. Roy Hazelwood recalls driving through a neighborhood where some of the victims disappeared: “People walking on the sidewalk stopped. People mowing their lawns stopped. People stopped talking to each other and stared at us. It was like one of those E. F. Hutton ads on television . . . I knew there was no way a white serial killer could have moved through those neighborhoods without being noticed. I knew the guy had to be black.”212 Furthermore, statistically speaking, serial killers tend to kill victims within their own race.
There were no reports of any of the victims being forcibly abducted. The profilers concluded that the serial killer was using some kind of con to lure his victims away. He was intelligent, probably from a middle-class or upper-class background, and had some kind of occupation or hobby that might have been attractive to the young victims. He was probably in his midtwenties—not too old to frighten away the victims nor too young to be credible in his con. There was little evidence of sexual contact with the victims, and therefore the profilers suggested that he was sexually inadequate and probably single. He would be a police buff, might drive a police-type vehicle and own a police-type dog like a German shepherd, and might have already interjected himself somewhere into the investigation. He would be probably following media reports of the progress of the investigation.
Not long afterward, Atlanta police received an anonymous phone call from an individual claiming to be a racist killer and warning that he would kill more black kids. The caller hung up after giving police a roadside location to search for another body. Douglas had recently witnessed hoax messages while teaching in England during the Yorkshire Ripper investigation and remembered the impediment they had created to identifying that killer. He suggested that the police set up a telephone trace and then make a big show of searching the opposite side of the road from which the caller instructed them to search. Sure enough, the caller telephoned, telling the police how stupid they were in searching the wrong side of the road. His call was traced and he was quickly arrested. Again, we see that profiling is important not only in focusing on who the perpetrator might be, but also in filtering out who is not.
Incredibly, the next victim was found on the road that police pretended to search to flush out the hoaxer. Apparently the killer was following press reports and knew that he had not dumped any body there. Now he was going to show his superiority to the police.
John Douglas wrote that in February 1981, somebody in the medical examiner’s office released information that was widely carried by the media: that five victims had been linked to the same killer through hair and fiber evidence. “And something clicked with me. He’s going to start dumping bodies in the river,” Douglas wrote in his memoirs. “We’ve got to start surveilling the rivers, I said.”213
Local police and FBI agents staked out bridges for weeks over rivers in the Atlanta area, and in what has been called “The splash heard ’round the world” they identified Wayne Williams after seeing his vehicle on a bridge at 3:00 A.M. in the vicinity of the sound of a large splash. No body had been located at that moment, but a suspect was now identified. Several days later, police pulled from the river two bodies downstream from the bridge. On the basis of fiber and hair evidence and witness statements that placed the victims in Williams’s company, he was charged with the two murders.
Wayne Williams was a twenty-three-year-old black man who lived with his parents, both retired schoolteachers. He was an unrealized prodigy, establishing his own small low-powered radio station when he was fourteen. Although the station’s range did not extend beyond a few city blocks, he attracted a lot of press attention. He claimed to be a music producer and promoter and recruited young boys whom he promised stardom. Nothing came of his promises.
Williams monitored radio and police frequencies and sold freelance photographs of accidents to local newspapers. At one of the scenes of the child murders, Wayne Williams offered his services as a crime scene photographer. In 1976 he was charged with impersonating a police officer when his vehicle was found equipped with red and blue police lights beneath the grill and on the dashboard. He owned a German shepherd.
While he appeared genteel and harmless, FBI profilers coached the prosecution on how to get Williams to make angry outbursts in the courtroom during his cross-examination. Williams was never charged with the other child murders, but after his arrest they ceased. He was convicted of only two murders, neither of which exactly fit the pattern of the child murders. The victims were twenty-seven and twenty-one, and their cause of death was not clearly established. When state forensic evidence linking Williams to the two victims was discredited by the defense, the judge in a controversial decision allowed the introduction of fiber and hair evidence linking Williams to the other victims, even though he was not charged with their murders. Williams to this day proclaims his innocence.214
Looking over Wayne Williams’s description, it appears that the FBI profile that Hazelwood and Douglas submitted is incredibly accurate—right down to the ownership of a vehicle fitted with police lights and the dog. But just to show how difficult it is to sort fact from fiction in true-crime history, there is a radically different account of the FBI’s participation in the Atlanta child murders investigation. Robert D. Keppel, a professional investigator in more than fifty serial homicide cases and a consultant to the Atlanta child murder task force, remembers things differently upon his arrival in Atlanta in March 1980:
We couldn’t wait to hear what gems of wisdom would come from the BSU’s agents, most of whom were only self-proclaimed experts in murder investigations and had never investigated one lead in an actual murder case. The FBI were the kings of follow-up but couldn’t solve a crime in progress. Most homicide detectives knew this . . . The profile of the probable killer provided by the BSU mirrored the wishes of the community, that is, the killer was white.215
Moreover, the FBI surveillance of river bridges did not start until April 24, after four bodies had already been found in Atlanta-area rivers between March 30 and April 20.216 Perhaps bureaucratic intransigence can explain why the bridge surveillance was not initiated immediately after Douglas “clicked” but only after four bodies in a row had been pulled out from the rivers. One wonders, though, why neither Hazelwood nor Douglas mention in their autobiographies the earlier BSU profile suggesting that the killer was a white individual. Or did Robert Keppel get the story all wrong? Go figure it out.
In 1985, FBI profilers confronted a highly organized serial killer. On May 31, in rural South Carolina, seventeen-year-old Shari Faye Smith was kidnapped on her way home from school. She was driving, but at the gates of her home, she stopped to check the roadside mailbox. He car was found at the end of the driveway, keys in the ignition, motor still running and her purse on the seat. During the next four days, the family received numerous telephone calls from the kidnapper, who said to them, “Shari is now a part of me. Physically, mentally, emotionally, spiritually. Our souls are now one . . . Shari is protected and . . . she is a part of me now and God looks after all of us.”
A handwritten letter from Shari arrived several days after her kidnapping—it was her “last will and testament” and proclaimed her love for her family. It was dated June 1, 1985, 3:10 A.M. Reading the letter, the FBI was convinced that Shari was now dead.
The family kept receiving phone calls from the killer, but the calls were always placed from pay phones and never lasted long enough for the police to arrive in time at the location. The phones were dusted for prints, but the caller never left any behind. The voice was disguised as well, using an improvised electronic device.
Speaking to Shari’s sister Dawn on June 4, the killer told her how he had kidnapped and murdered Shari. He said that the whole thing had “gotten out of hand.” He was reading from some kind of written script, and at one point he made an error: “Okay, 4:58 A.M.—no, I’m sorry. Hold on a minute . . . 3:10 A.M., Saturday, the first of June, she handwrote what you received . . . 4:58 A.M. Saturday, the first of June, we became one soul.”
Giving Dawn precise directions, to within six feet of where Shari’s body was located, the killer told her, “We’re waiting. God chose us.” He then hung up.
Arriving at the described location, eighteen miles away from Shari’s home, the police found her body. The body was dressed and decomposition had occurred to the point where police could no longer determine how Shari had been killed or whether she had been sexually assaulted. Sticky residue from duct tape was found in Shari’s hair and around her face, but the tape itself had been removed.
FBI profiler John Douglas, who assembled the profile, believed it was highly likely that Shari was the killer’s first murder victim. He noted, however, that the removal of the duct tape was indicative of organized planning and that generally first-time murderers do not start out this organized. This meant that the killer was probably older and more intelligent than average. When he telephoned the family he read quickly from a script and got off the phone as soon as possible. His meticulous directions to the body indicated that he must have revisited the corpse numerous times after the murder for some kind of sexual gratification and ceased only when a “relationship” was no longer possible because of advanced decomposition. Douglas believed that the killer had had a brief failed marriage from the degree of cruelty in the mind games he played with the victim’s family. He probably had a criminal record for obscene phone calls or attacks on children, but would not attack prostitutes because he was intimidated by them. He was most likely living alone or with his parents and would be in his early thirties. Because he chose a body dump site to which he could return, he was probably a local man familiar with the geography of the region. Again from his detailed instructions and highly scripted phone contact, it appeared that he was meticulous, rigid, and orderly. His car would probably be well maintained, clean, and no more than three years old. The fact that he had committed the abduction in the middle of the afternoon in a rural residential district suggested a degree of controlled sophistication. The killer was using some kind of improvised variable-speed device to disguise his voice, which probably meant he was skilled in electronics or in electric trades and aware of the risk of forensic voice identification.
Two weeks exactly after Shari was kidnapped, the offender struck again, this time kidnapping nine-year-old Debra May Helmick from the front yard of her home. Her father was in the house, ten yards away. The kidnapping took place some thirty miles away from Shari’s home. Witnesses saw the man pull up in a car, speak to the girl, and then suddenly force her inside the vehicle and drive away. They attempted to follow the car but it lost them. John Douglas had little doubt that it was the same offender. Douglas ventured that he would now be displaying even more compulsive behavior. People around him might notice weight loss; he might be drinking more and not shaving regularly, and he would be eagerly talking about the murder, following television reports, and cutting newspaper clippings about the crime. He would be greatly enjoying his celebrity and sense of power over his victims and the community.
The police and the FBI felt that because he was closely following the case and was phoning Shari’s family, he might be flushed out into the open. The FBI had a local newspaper publish a story of how the Smiths were having a memorial service at Shari’s grave, and published pictures of Dawn putting a stuffed teddy bear at her sister’s graveside. FBI agents photographed the license plates of any cars that passed by the memorial service and staked out the grave for days in the hope that the killer would attempt to take the teddy bear as a souvenir. He was too smart to take the stuffed toy bait, but obsessed enough to drive by the memorial service: When he was arrested, his license number was among the cars that the FBI had noted passing the location.
The killer continued to call Dawn, telling her where the body of the nine-year-old girl could be found: “Debra May is waiting. God forgive us all,” said the killer. He warned Dawn that at some point he was going to come for her.
The FBI crime laboratory, in the meantime, had been analyzing the two-page letter that Shari had written before her murder. An Esta machine is an instrument that can identify microscopic impressions in paper made earlier by somebody writing on sheets of paper higher up in the pad. The device uncovered the traces of the first nine digits of a ten-digit telephone number—a number in Alabama. This narrowed the number down to ten possible numbers. It was a remarkable piece of luck that the number and its area code were written down, and that it was an out-of-state number, meaning that long-distance charge records would be available. The police began to look through the billing records of the ten numbers to see if any had received calls from the area around where Shari was kidnapped and murdered. One of the numbers belonged to a soldier stationed in Alabama who had received calls from a house located just fifteen miles away from where Shari had lived.
When the police arrived at the house, they found an elderly couple living there. The husband was an electrician by trade. The soldier in Alabama was their son and they frequently talked on the phone. The profile and the record of the elderly man, other than his being an electrician, made him an unlikely suspect. Moreover, when the murders had occurred, the couple had been visiting their son in Alabama. They had no idea how their son’s number could have ended up on the same pad from which came the sheet of paper that Shari had written her death letter on.
The disappointed police officers were almost ready to leave when they tried one more thing. They described to the couple the profile the FBI had drawn up of the hypothetical offender—did the couple know anybody who might fit the description?
Both the man and the woman responded immediately; that was the description of Larry Gene Bell, an assistant electrician who often worked for the man. He was in his early thirties, heavyset, and divorced from his wife. He was a very neat and meticulous worker. When they were away in Alabama, Larry house-sat for them—as a matter of fact, the man remembered, he wrote down his son’s number in Alabama on Larry’s pad just in case he needed to get in touch with them. When they came back from Alabama, Bell picked them up at the airport. They were surprised by his physical appearance—he was unshaved, had lost some weight, and seemed highly agitated. All he wanted to talk about was the kidnapping of the Smith girl.
A search of Bell’s apartment uncovered all manner of evidence linking him to the two girls. Witnesses who had seen the little girl kidnapped identified Bell as the culprit. Bell had numerous sexual offenses on his record, from several attempted kidnappings to making obscene phone calls. In 1986, he was sentenced to death.
Once again, the profile was essential not so much for the police to identify the offender as for potential witnesses to point out a possible suspect. As elaborate and evolved as the profiling process is, in the overall picture it is only one component in many different investigative procedures applied to tracking a serial killer. These basic police procedures could be as complex as laboratory analysis or as routine as surveying a shopping center, but without them, the psychological profile has nowhere to go.
The most valuable aspect of profiling is not that it identifies who the killer is, but helps in weeding out who it is not. Frequently police already have information on a killer’s identity but it is buried deep within masses of files, leads, and tips among other suspects. Profiling often helps police decide whom to look at first.
Many scientists challenge not only the reliability of the FBI’s two categorizations (three if you include “mixed”) but also whether those offender categories actually display the characteristics attributed to them. There simply have been not enough data or scientific tests to confirm the FBI’s system, which was first based on interviews with only thirty-six sexual murderers, not all of whom were serial killers.
For example, the FBI asserts that organized offenders were “likely to change jobs or leave town” after committing a murder. A University of Detroit Mercy study of the data the FBI used revealed some problems with that assumption. As the FBI says, no disorganized killer in their study left town or changed jobs, but out of ninety-seven homicides by organized killers, only eleven killers left town and eight changed jobs—and presumably the eight job changers are included in the eleven who left town. This means that a profiler who suggests that an unknown killer who appears organized has changed jobs or left town would be wrong 89 percent of the time.217
Since their introduction some two decades ago, the FBI’s profiling techniques have not been empirically substantiated by scientific testing methodology. One measure of scientific integrity is the ability of a hypothesis to survive attempts to disprove it—attempts to falsify it. This is a kind of reverse-engineering approach to scientific proof known as hypothetico deductivism. In his criticism of the scientific foundations of the FBI’s system of profiling, criminologist Damon A. Muller gives the example of water’s boiling temperature. By inductive reasoning, we assume that it is scientifically proven that water boils at 212°F because if we repeat this experiment thousands of times, water will always boil at that temperature.218 But we would be wrong. If you take water to a mountaintop at a higher altitude, it boils at a lower temperature. The boiling point of water depends on more complex issues than merely temperature. If instead of attempting to prove that water always boils at 212°F, we attempted to first disprove it, we would quickly emerge with a truer scientific assessment of the reliability of the hypothesis. Repeating the experiment thousands of times with the same result is not necessarily conclusive scientific proof. The ability of a hypothesis to survive disproval is a more reliable measure. For something to be scientific, it must present theories that are empirically testable—not just simply repeatable by experiment.
While the process is simple for testing the behavior of water, how does one reverse-engineer human homicidal behavior—how does one falsify and test the FBI’s theories of fantasy underlying homicidal behavior, for example? The problem is that the FBI has not fully explained the psychological mechanics motivating the offenders it attempts to classify. For example, the FBI has not identified the process by which a killer becomes either organized or disorganized except to suggest that acute mental disorders are more often present in the disorganized murderer. The FBI sort of says, “It works, but we don’t know why it works.” FBI profilers have commented that they do not care why offenders do what they do; they are sufficiently satisfied with the investigative value of knowing what they do, without worrying about the whys.
The FBI does not particularly want its profiling system to be scientifically systemized. Veteran profiler John Douglas states, “The key attribute necessary to be a good profiler is judgment—a judgment based not primarily on the analysis of facts and figures, but on instinct . . . Many, many factors come together in our evaluations, and ultimately, it comes down to the individual analyst’s judgments rather than any objective scale or test.”219
Thus, according to Douglas, the FBI profiling system is not based on any systematically scientific approach, but ultimately on a subjective hunch of the profiler. One would not tolerate this kind of approach in other police sciences such as DNA testing, fingerprint identification, and ballistics testing. But often police departments turn to the FBI for assistance when they come to an investigative dead end and find themselves in a “nothing to lose” situation. The scientific veracity of FBI profiling, therefore, remains largely unchallenged. When the profiles are correct, the FBI is praised; when the profiles are wrong, well, “no harm done.”
Gradually the FBI’s system of remote psychological profiling began coming under heavier scrutiny and criticism in the 1990s. On April 19, 1989, a gun turret mysteriously exploded, killing forty-seven crewmen on board the battleship USS Iowa. Navy investigators decided that Clayton Hartwig, one of the sailors who perished in the turret, deliberately caused the explosion. It was alleged that he was motivated by a frustrated homosexual fixation on a fellow crewman. The Navy Investigative Service (NIS) met with veteran FBI profilers John Douglas, Roy Hazelwood, and Richard Ault, asking them to conduct an “equivocal death analysis”—a type of psychological autopsy intended to determine a cause of death: Was the death accidental, suicide, or murder? This type of analysis works on the same principles and techniques as criminal profiling, from which it is almost indistinguishable other than that the subject is identified and often more is known about him than about an unknown killer.
John Douglas felt uncomfortable with the data the navy was providing and declined to participate in profiling Hartwig. Hazelwood and Ault plunged ahead, a decision they would come to regret. Using only the information that the NIS provided them, as was standard procedure with other investigative agencies making such requests, Hazelwood and Ault concluded that Clayton Hartwig was:
. . . a very troubled young man who had low self-esteem and coveted power and authority he felt he could not possess. The real and perceived rejections of significant others emotionally devastated him. This combined with the inability to verbally express anger and faced with a multitude of stressors had he returned from the cruise, virtually ensured some type of reaction. In this case, in our opinion, it was a suicide. He did so in a place and manner designed to give him the recognition and respect that he felt was denied him.220
Unlike police agencies, however, the navy subsequently issued a statement that the Iowa explosion was caused by the late Clayton Hartwig, and cited the FBI profile as primary evidence. Ault and Hazelwood made matters even worse when they testified to the veracity of their profile before both Senate and House committees reviewing the navy’s investigation of the explosion. John Douglas felt that Ault and Hazelwood botched the profile and worse, “We never testify to a profile. We use it as an investigative tool, and [Ault and Hazelwood] used it as tantamount to proof.”221
After the navy released its conclusions, there were many skeptical critics of the investigation, from the Hartwig family to CBS’s 60 Minutes investigative news program. The sailor who was supposedly the subject of Hartwig’s fixation asserted that he and Hartwig were the “best of friends . . . I think he loved me as a brother, but nothing sexual or anything.” There were allegations asserting that the navy was scapegoating Hartwig to cover up technical problems with its battleships. The matter became of interest to Congress, and both House and Senate committees looked into the question. A House Investigations Subcommittee started to ask how the FBI profilers concluded that the sailor committed suicide without any surviving witnesses or physical evidence. On December 21, 1989, Hazelwood and Ault were called to testify before the committee about their profile and the methods and practices of the BSU. It was not going to be a good day for the two FBI agents.
In opening the questioning that day, one of the congressmen stated, “Given the serious defects in the Navy investigation that we have uncovered in our previous hearing today’s testimony becomes even more crucial to the Navy’s case against Hartwig. If the psychological assessment compiled by the FBI is flawed, then exactly what is left with which to support its conclusions?” It was obviously the intention of the subcommittee to question the navy investigation by discrediting the BSU’s profile.222
Hazelwood and Ault arrogantly insisted to the incredulous congressmen that their profile was always definitive; they did not qualify it on scales of probability. They either committed to a position or said they did not know. Their profile “could be used or discarded or discounted. That is simply our opinion.”
Representative Les Aspin of Wisconsin asked, “You do not configure it as your job to put in qualifiers. The world is full of qualifiers, and you come down yes or no.”
“Yes, sir,” replied Ault.
Another congressman asked whether the interviews on which the FBI profilers based their analysis were bias-controlled. Hazelwood and Ault rejected that as a technique in academic research. They were criminal investigators, not researchers, they explained. When it was mentioned that one expert in forensic psychology questioned the validity of some of the witness interviews submitted to the FBI profilers because they were not conducted by clinicians experienced in interviewing people under stress, Ault snapped back, “First of all, I certainly appreciate that wonderful academic approach to a practical problem. It is typical of what we find when we see people who have not had the experience of investigating either crime scenes, victims, criminals, and so forth in active, ongoing investigations. Second, in the real world, you just can’t work that way with the NIS. So, we have to take the information as we get it.”223
Testifying before a Senate committee, Ault was asked if there was “any hard evidence, any evidence that would support the idea that Hartwig actually carried out this act”; he replied, “No, Sir, this opinion that we submitted is based on a half scientific, half art form.”
After Hazelwood and Ault concluded their testimony, a panel of clinical psychologists and a psychiatrist tore the FBI profiling program apart like a pack of wild dogs: “The FBI report which has been discussed substantially here this morning is incorrect as to form, as I understand what psychologists are able to do . . . flawed in terms of failing to present in a candid fashion the limits of the science . . .”
In its March 1990 report, the House committee was scathing in its dismissal of the BSU’s techniques:
Unfortunately, if there was a single major fault in this investigation, it was the FBI system for producing the Equivocal Death Analysis. These documents, most commonly generated after orthodox police investigations have reached no conclusions, are invariably unequivocal in reaching a conclusion. The analysis contains no comment on probabilities, no qualifications and no statement of the limitations of a posthumous analysis in which the subject cannot be interviewed . . .
Just as the FBI psychological analysis was key to the Navy investigation, so it was to the subcommittee inquiry. As a result, the subcommittee, using the professional services of the American Psychological Association, sought the opinions of 11 independent clinical licensed psychologists and one psychiatrist. Additionally, the subcommittee consulted independently with a psychiatrist and a psychologist who had some previous knowledge of this case. Ten of the 14 experts consulted considered the FBI analysis invalid. And even those who believed the analysis to be somewhat credible were critical of procedures, methodology, and the lack of a statement of the limitations of a retrospective analysis of this nature.
The FBI psychological analysis procedures are of doubtful professionalism. The false air of certainty generated by the FBI analysis was probably the single major factor inducing the Navy to single out Clayton Hartwig as the likely guilty party . . . The procedures the FBI used in preparing the Equivocal Death Analysis were inadequate and unprofessional . . . The FBI agents’ Equivocal Death Analysis was invalidated by ten of fourteen professional psychologists and psychiatrists, and heavily criticized even by those professionals who found the Hartwig possibility plausible . . . The FBI should review its procedures and revise its Equivocal Death Analysis format to address probabilities and make clear to consumers of such reports their limitations and speculative nature.224
Some press coverage took glee in portraying the BSU as unprofessional quacks. New York Daily News columnist Lars-Erik Nelson wrote, “Deep within the FBI there exists a unit of people who—without ever talking to you or anyone who knows you—are prepared to go into court and testify that you are a homicidal maniac.” He concluded with the hope that “any defense lawyer, any judge, any jury would laugh this kind of quack evidence out of court.” The Washington Post editorialized, “Long-distance personality analyses by experts who have never met the subject are almost always ludicrous. But when the subject is dead, cannot defend himself and is found to be guilty of a horrendous crime, opinions like this are deserving of contempt.”
It was revealed by the FBI’s own testimony that they had used Equivocal Death Analysis in more than fifty cases but obtained “satisfactory results” in only three.225 The failures of profiling in general were dredged up. In Los Angeles in the mid-1970s a panel of psychologists drew up a profile of the “Skid Row Slasher”—a serial killer preying on homeless people. The day after they issued a description of a tall, emaciated white male with stringy blond shoulder-length hair with latent homosexual tendencies and possibly some congenital defects, the real Skid Row Slasher dropped his wallet replete with ID near the scene of his latest attack. Vaughn Greenwood was a chubby, short-haired black man of medium height. Despite the evolution of the FBI’s profiling system since the Skid Row Slasher days, (a profile for which it was not responsible), the tendency since Iowa has been to question the real effectiveness of the FBI as Sherlock Holmes–like analysts.
Hollywood came to the rescue just in the nick of time. The February 1991 release of The Silence of the Lambs could not have occurred at a better moment for the besieged FBI profilers. The huge box-office and critical success of the Jodie Foster movie quickly refocused the news media toward positive promotional stories on the FBI’s Behavioral Sciences Unit. FBI agents Clarice Starling (Jodie Foster) and Jack Crawford (a character synthesized from Douglas, Ressler, and Hazelwood, portrayed by Scott Glenn) introduced the public to Quantico’s world of profilers, unsubs, and organized and disorganized offenders. Perversely, the suave serial killer Hannibal “the Cannibal” Lecter, portrayed by Anthony Hopkins, grew to even greater cult status, and by 2000 he would headline the sequel, Hannibal. Matchbook covers and small classified ads in detective magazines offered diplomas in “criminal profiling.” TV shows such as Profiler, Millennium, and CSI ranged in their portrayals of profilers from scientists to magicians with psychic powers.
The controversy over the lack of scientific method, however, has not died out. If profiling is more an art than a science, it means that the accuracy of the profile depends on the talent of the individual profiler. That becomes problematic because to accept the analysis of one profiler is to sometimes reject the talents of another—the process becomes highly antagonistic and career-threatening among profilers. Moreover, the lack of applied scientific method can lead investigators outside the profiling specialty to conclude that their analysis is as valid as or better than a profiler’s. As Representative Frank McCloskey put it during the Iowa hearings, “It would seem to me that in many ways politicians, teachers, or people in the insurance business would have similar outstanding psychological skills. I don’t know that that certifies them to make some of these judgments.”
The dependence on individual profiler talent is not conducive to the investigative cooperation that complex multijurisdictional cases often require. For example, the profile used by the Unabomb task force to investigate a series of bombings for which Ted Kaczynski would be eventually charged suggested that the serial killer was in his midthirties to early forties, perhaps with some college education but most likely a skilled blue-collar worker. FBI profiler Bill Tafoya dissented, suggesting that the killer probably had a graduate degree, perhaps even a Ph.D., in a “hard” science—something like electrical engineering or mathematics—and would be in his early fifties. Tafoya claims that tremendous pressure was put on him to “get with the program” and amend his profile to one consistent with the task force’s then current investigative strategy focused on blue-collar aircraft workers around age forty.226 When Kaczynski was arrested he was fifty-three years old and held a Ph.D. in mathematics, as Tafoya’s profile suggested.
Most recently, profiling has again come under question, over the speculations offered on the Beltway Snipers who terrorized the Washington, D.C., region in the autumn of 2002. Although it is unclear what the official FBI profile on the sniper was, former FBI profilers and private profiling consultants almost unanimously ventured that the sniper was an organized white loner, in his mid- to late thirties, killing for the “thrill” of it and not for political or ideological motives. Instead police ended up arresting two black suspects: forty-one-year-old John Allen Muhammad and seventeen-year-old John Lee Malvo, whose motives are still unclear. Despite the fact that several witnesses described “dark-skinned” suspects and an earlier witness reported two “Hispanic” men, police continued to focus on what is known as “RWG”—regular white guys. One Beltway Sniper task force officer responded to the witness reports, “Anybody could put makeup on . . . we don’t want anyone to give up on the fact that it could be a white guy.”227
Former FBI counterintelligence agent I. C. Smith explained, “First you start with the statistic that 80 to 90 percent of all (serial) shooters are white guys. Then you back out the guys happily married. That leaves the introspective loner types.” This is exactly the primary weakness of profiling: its basis in statistical probability. Indeed, the majority of snipers in the past have been white antisocial male thrill killers. But a profile collapses the moment an offender does not conform to the statistical norm. Since numerically serial snipers are still a relatively rare phenomenon, there is a high probability of a profile being invalidated by the appearance of a single anomaly. Criminologist Robert Keppler commented, “In a macabre way, looking at the changing face of a serial killer is like looking at the changing face of America . . . Because serial offenders exist at the margins of society, and not in the mainstream, even though they can camouflage themselves in the mainstream of their victim pool, it is at the margins where one sometimes has to look to find the truth about what is taking place in society.”228 While this might be less of a critical issue in a dead-end investigation, in a rapidly unfolding case like that of the Beltway snipers, an inaccurate profile could have misguided investigative efforts and jeopardized lives.
Researchers are only now beginning to explore the scientific veracity of criminal profiling. This is difficult, because most practicing profilers have refused to cooperate in controlled tests of their profiling abilities, no doubt fearing for their personal and professional credibility should they score unfavorably. Two recent studies on profiling, while not necessarily putting profiling itself under question, looked into the question of what kind of practitioners were most accurate in their profiles.
The first test, the results of which were reported in 2000, compared the profiling skills of a group of professional profilers (those known to have been retained by police departments to submit profiles), a group of psychologists with no training in forensic or criminal psychology, police officers with five or more years of experience, psychics, and sophomore science and economics students.229 The drawbacks of this study were that the group of profilers were not law enforcement officers, the way FBI profilers are, and that only five profilers volunteered to participate in the test. The other groups consisted of twenty to thirty-five volunteers each.
Each participant was given the same investigative file of a closed case for which the offender’s identity and pathology were known to the researchers, and asked to assemble profiles describing the offender’s physical characteristics, cognitive processes, offense behaviors, and social history and habits. They were then scored on how accurate their profile was compared to what the researchers knew about the actual offender.
The psychics scored at the bottom of the scale. The profilers scored at the top for most accuracy. The psychologists with no criminal training followed them. In some categories, such as offense behaviors and physical characteristics, the psychologists actually scored marginally better than the profilers. Notwithstanding the fact that only five profilers were participating in the test, it did indicate that professional profilers did best at what they say they do best—criminal profiling. That psychological insight was a primary factor in the accuracy of this type of profiling was confirmed by the fact that psychologists scored second best. What was surprising, however, were the results from the other groups. In terms of accuracy, students, then police, and finally psychics followed the profilers and psychologists. That untrained science and economics students scored better than police was troubling, because FBI profilers are law enforcement officers first, and profilers second. Moreover, the FBI stresses that the primary qualification of their profilers is a long career as a criminal investigator prior to subsequent training as a profiler-analyst.
In 2002, the same team of researchers tackled the issue of the effect of investigative experience on profiler accuracy. This time they gathered six groups: thirty-one senior detectives with a minimum of ten years investigative experience; twelve seasoned homicide detectives; nineteen trainee detectives, all with a minimum of ten years of general police duties; fifty police academy recruits with less than six weeks’ training; fifty police academy recruits with less than three weeks’ training; and thirty-one sophomore chemistry students. Each participant was given a similar task as in the previous test.230
The results were astonishing: Sophomore chemistry students outperformed all the police groups across the board. Among the police groups, police recruits scored higher than experienced homicide detectives and outperformed the other police groups on some of the questions. These results contradict the FBI’s assertions that investigative experience is the highest qualification that successful profilers can bring to the job.
The researchers suggest that paradoxically, the more experience an investigator has, the more that experience gets in the way of interpreting data for the purposes of profiling. Investigators, they propose, develop over the years a set of commonsense “heuristics”—impressions about criminals and crime that are based not necessarily on fact but on subjective experience and perceptions. Thus college students and police recruits with no such prejudicial experience produced more accurate profiles. The other factor that might be taken into account is the educational requirements of senior police officers at the time when they were originally recruited, which may account for their poorer performance in comparison to current police recruits. Formal education, it seems, is most valuable for effective profiling, instead of long police investigative experience. In that context, it should be noted that many FBI profilers hold MAs and Ph.D.s in behavioral sciences, which may account more for the BSU’s effectiveness than their investigative experience.
In the end, however, before jumping to conclusions, one needs to remember that these tests presented only one case for profiling, and the nature of that particular case could perhaps favor the performance of one group over another. Nonetheless, we can see that if profiling is based on statistical probability, the more experienced investigators are, the more likely they are going to base their profiles on their perception of probability, as was the case with the Beltway Sniper.
The FBI’s crime scene analysis approach is not the only kind of behavioral profiling system. There is also diagnostic evaluation (DE), an adaptation of largely Freudian psychotherapeutic theory to crime analysis that depends on the diagnostic conclusions of the individual profiler. This was what Dr. James Brussel was practicing in his profile of the Mad Bomber in the 1950s. DE is highly individualistic from practitioner to practitioner and not used often now as an investigative tool.
There is also investigative psychology (IP), used often by profilers in England, which is based on a collection of theories founded on the premise that a crime can be analyzed as an interpersonal transaction in which criminals are performing actions in a social context between themselves and their victims. IP was developed by David Cantor, an environmental psychologist, and first applied to the investigation of serial killer and rapist John Duffy, dubbed “The Railway Rapist” in London in 1986.
Cantor structured four broad approaches to creating an IP profile:
It is obvious that IP is similar to the FBI’s criminal investigative analysis in many ways. The primary difference is the FBI’s dualistic categorization of offenders as organized or disorganized.
Criminal investigative analysis and psychological profiling are not the only types of profiling available to analyze the patterns of a serial killer. The most recent studies in England are proposing a very simple so-called circle hypothesis, which suggests that serial offenders live within a circle whose diameter is equal to the distance between the offender’s two farthest offenses. On the average, this equals an area of approximately ten square miles. Various studies confirmed this hypothesis: Eighty-seven percent of forty-five serial rapists in the south of England resided in this range; 82 percent of serial arsonists and 70 percent of serial rapists studied in Australia; and 86 percent of 126 U.S. serial killers as well.231
The most recent advances in computer-driven geographic profiling of serial offenders has been made in Canada, where a radically new dimension has been added to profiling systems. A system called Orion, developed by former Vancouver police constable D. Kim Rossmo, operates on the principle of geoforensic profiling. What makes this system extraordinary is that it goes beyond identifying a hypothetical personality of the suspect—it ferrets out where the suspect lives and even his possible identity.
Kim Rossmo was fascinated by accounts of a particular type of rural crime that has been plaguing India since 500 B.C., called a dacoity. A dacoity is when a gang of five or more members of one village raids and robs another village. The dacoity is always committed at night and always when the moon is new, because with little electrical lighting in rural India, the new phase of the moon provided almost complete cover of darkness for the raiding party.
The Indian police developed a systematized geographic profiling system for capturing dacoity gangs. The police calculates the speed at which a person could travel on foot cross-country and plot a distance radius. Because farmers working in the fields at dawn would observe any strangers moving through their territory, the police knew that the gang had to get into the territory of their own village before dawn. Thus the probable village from where the gang came was quickly identified. As dacoity gangs followed a tradition by which they would not attack a member of their own caste, the police, once they identified the village, could further narrow down suspects by caste. From that point on, standard police investigative techniques could focus on the narrowed list of suspects.
Rossmo also listened to stories of his fellow officers who pursued suspects with dogs. They told him that suspects almost always did the same thing when attempting to flee on foot: Right-handed suspects, when turning, tended to turn left; when encountering obstacles, they moved to the right; when throwing away incriminating evidence, they threw to their right; and when hiding in large buildings, they kept to outside walls.
Rossmo studied cases in which geoforensic profiling was attempted. When Angelo Buono and Kenneth Bianchi were dumping bodies on hillsides in Los Angeles, the LAPD measured the distances between each site and then, using a computer, plotted circles on a map. The center of each circle represented victim availability; the circumference represented capacity; while the radius represented ability. Vectors drawn from the various sites narrowed down an area of about three square miles, which the Los Angeles Police then saturated with uniformed officers in an attempt to frighten off the serial killers. Later, when Buono and Bianchi were apprehended, it turned out that Angelo Buono’s residence and upholstery shop, where some of the victims were raped and murdered, was near the center of the plotted zone.
Rossmo’s Orion system takes into account not only geographical factors, but traffic flow factors, urban zoning, census data, and offender personality profiles to emerge with a highly probable offender location. Although Orion does not work with all types of personalities, it is highly effective with nontraveling serial offenders—rapists, bank robbers, child molesters, and killers who live in the same territory where they commit their crimes and do not change residences.
Orion mathematically plots zones that an offender is likely to strike in, known as jeopardies, taking into account that most criminals tend not to strike too close to their own home but not too inconveniently far away either. From the jeopardies, depending on the personality of the criminal as profiled, Orion begins to calculate his or her probable place of residence by measuring “decay” from each crime site—where a convenient crime site becomes too close to a possible residence of an offender. To be practical, the system needs several crime scenes to triangulate an accurate rate of decay—which is precisely why it is so powerful a tool with serial offenders. Orion becomes more than 90 percent accurate with six or more crime locations entered into the system.
When Orion went online in Vancouver, it quickly proved itself in catching a child molester who was attacking children in different parts of the city. The police had a vehicle description, but little else. After determining the traffic patterns at the times of the attacks, Orion plotted jeopardies, eliminating industrial, commercial, and other nonresidential areas. Based on census data and reflecting the probable socioeconomic background of the offender as identified by psychological profiling, Orion calculated a list of postal codes, in order of probability, where the offender likely resided. These postal codes were then filtered through motor vehicle registration and driver’s license files to match to the suspect vehicle described. A very short list of suspects was identified and subsequently investigated by standard police procedure, quickly leading to the arrest of the child molester.
As powerful and as automated as all these systems are, Profiler and Orion are no better than the information fed into them. The geographical targeting done by Orion depends on the quality of the psychological profile of the suspect, as different profile types behave differently in geographical space. Likewise, Profiler’s ability to analyze data is only as good as the artificial intelligence rules written for it by human programmers. Furthermore, if police on the scene do not recognize, note, and identify certain crime scene characteristics, there is nothing to input into Profiler. Finally, the output of these computer systems only provides high probability rates—it still remains up to human officers to interpret the data and fit them into their investigations in an effective manner.