Problems With Quantification
Most experts agree that the number of serial murders and of serial murderers is unknown (Egger 1990a; Gresswell and Hollin 1994; Hickey 1997; Holmes 1989; Holmes, Hickey and Holmes 1991; Jenkins 1994; Maxfield 1989; O'Reilly-Fleming 1996). It has been estimated that between 10 and 500 serial killers are active at any time in the United States (Egger 1990a; Kiger 1990; O'Reilly- Fleming 1996). In Canada, estimates range from 5 to 30 (Ratner 1996). The variation in these estimates can be attributed to a variety of problems with data sources: arbitrary definitions; small samples; samples biased toward only known/apprehended serial killers; and samples relying upon secondary sources such as biographies or newspapers. These alternative data sources have been used primarily because official data are not reliable.
For example, the FBI collects data from law enforcement agencies across the United States and publishes it in the Uniform Crime Reports(UCR). The Supplemental Homicide Report(SHR), part of the UCR, provides additional information about victims, offenders and circumstances. The intention is to reflect all criminal offenses that come to the attention of the police.
The data, however, are incomplete and unreliable. First, because reporting is voluntary, the information is incomplete (Kiger 1990; Williams and Flewelling 1987). Second, there may be organizational pressures within particular police jurisdictions not to alarm the public about the possible existence of a serial killer in that area. This may prevent reporting and/or effect homicide classification procedures (Kiger 1990; Williams and Flewelling 1987). Third, homicide data records only those crimes known to the police. Missing persons and undiscovered bodies are excluded.
The importance of this third observation is that many unidentified bodies are found each year and there have been attempts to link the number of unidentified bodies to serial murder. Annual estimates in the United States place the number of unidentified dead between 4,000 and 5,000. Some authors claim that many of these are victims of serial killers. For example, Radford (1992) estimates that most the 3,500 unidentified female victims of murder each year in the United States are victims of serial murder. This is an unwarranted claim which Jenkins has clearly denounced:
The problems with this statistic were, or should have been apparent. If, in fact, 3,500 females fell victim to serial killers each year during the mid- 1980s, this would have accounted for some 70 percent of all murders of women and girls. Apart from being implausible as it stands, this figure would also leave little room for the domestic murders that, according to the femicide literature, are believed to be so endemic a problem (1994:142).
The actual proportion of the unidentified dead who are the victims of serial murder is unknown (Hickey 1997: Egger 1990a; Jenkins 1994: Kiger 1990). In the United States, it is not mandatory for coroners to report the discovery of unidentified bodies to a centralized data bank. Also, coroners need no formal training for the position. Their reports to the FBI may contain unreliable information about the cause and time of death (Kiger 1990). Further, cause of death cannot be determined from badly decomposed bodies, which are then unlikely to be counted as homicides. Sometimes, bodies may be found where cause of death can be classed as homicide, but the victims were killed in previous years. These victims are not likely to be submitted as additions to a revised homicide count for that year (Kiger 1990). Moreover, many victims of a serial killer may not be classed as homicides at all. They may be assumed to have died of natural causes (eg. nursing home and hospital patients).
As a result, homicide rates are underestimated and the consequent analyses are inaccurate (Jenkins 1994; Kiger 1990; Williams and Flewelling 1987). UCR/SHR data reflect only the number of deaths labelled as homicide by the police. They do not necessarily reflect the actual number of deaths in a given year. At this point, numerical estimates of serial killers and victims based on this data are questionable. Thus, it is unwarranted to assume that serial killing is on the rise as has sometimes been claimed. It is also inaccurate to attribute most stranger or unknown homicides to serial murder. This limits our ability to understand the nature and extent of the problem, and or course, inflated estimates provoke fear of a serial killer epidemic.
A further complication arises in the inability to use existing data sources to make cross cultural or cross national comparisons. Although serial murderers are known to exist in most countries, there have been few attempts at cross national comparisons. Cross national analysis is impeded by definitional problems, differential reporting patterns, and data inaccessibility. Thus, more effort must be put into improved reporting standards (a point which will be discussed below) to allow for systematic comparison if we wish to advance understanding.