- Email: [email protected]
Does size really matter?
Available online at www.sciencedirect.com
The Social Science Journal 45 (2008) 330–344
Does size really matter? A reexamination of Sheldon’s somatotypes and criminal behavior Sean Maddan a , Jeffery T. Walker b,∗ , J. Mitchell Miller c a
Department of Criminology, University of Tampa, 401 West Kennedy Boulevard Tampa, FL 33606, USA Department of Criminal Justice, University of Arkansas - Little Rock, 2801 South University, Little Rock, AR 72204-1099, USA c Department of Criminal Justice, University of Texas - San Antonio, 501 West Durango Boulevard, San Antonio, TX 78207, USA b
Abstract In the mid-20th century, William Sheldon developed the idea of somatotypes in explaining different types of criminal behavior. Sheldon defined three body types believed to be associated with criminality: mesomorphs, who are athletically fit; endomorphs, who are overweight, and ectomorphs, who are characterized by fragility and thinness. Sheldon’s research implied that mesomorphic individuals were more prone to committing violent and aggressive acts. This research explores a new method of somatotyping using the Body Mass Index (BMI). What follows is an examination of prisoners from the State of Arkansas, using information provided by the Arkansas Department of Correction’s database on inmates. In particular, this paper examined prisoners’ physical characteristics such as age, height, and weight in comparison to the types of crimes that these prisoners committed. The findings of this paper show that the BMI is a useful alternative to traditional somatotyping techniques; the findings also suggest that a prisoner’s somatotype is associated with criminal patterns while being a meager predictor of criminality. Methodological and theoretical implications of this study are discussed. © 2008 Elsevier Inc. All rights reserved.
1. Introduction From the earliest days of criminological thought, researchers have sought biological causes of criminality. Because some of the earlier biological studies were disproved due to method∗
Corresponding author. Tel.: +1 501 569 3083. E-mail address: [email protected] (J.T. Walker).
0362-3319/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.soscij.2008.03.009
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
331
ological shortcomings or antiquated statistical techniques, a majority of criminologists today dismiss the biological research of criminality with relative ease. As Wilson and Herrnstein (1985, p. 80) wrote “many criminologists. . . are simply uneasy with biological and psychological concepts” involved with the causes of criminality due to their sociological training. In fact, “physical correlates of crime are often dismissed by most criminologists, for whom these are at best historical stages in the development of their subject”(Wilson & Herrnstein, 1985, p. 72). The last decade has seen a resurgence in the study of the links between biology and crime (Raine, 1993; Walsh, 2002). Under the mantle of the biological theories of criminality, the debate of whether or not criminals are characterized by a certain type of body build has long been debated in the field of criminology. Eysenck and Gudjonsson (1989, p. 22) wrote that the study of physique “is a complex subject, with many different typologies and methods of measurement.” In the 1940’s, William Sheldon (Sheldon, Stevens, & Tucker, 1940; Sheldon, Hartl, & McDermott, 1949) attempted to study the relationship between human physique and criminality. Although Sheldon was not the first to attempt such research, his method of doing so was more advanced than his predecessors. Sheldon put forth the idea of the somatotype; this is a personal score that is determined by different measurements taken from a human body. Sheldon posited three types of extreme somatotypes: mesomorphs who are athletically built, endomorphs who are overweight, and ectomorphs who are underweight. While several studies have been conducted on somatotyping and criminality over the last 50 years, overall interest in the subject has waned. This is due primarily to the stigma that has been associated with conducting research on the link between biological causes of crime in the past. Even though this is the case, it is hard to discard some of the findings from the somatotype literature, in spite of some methodological shortcomings. Research has consistently shown that mesomorphic physiques are more associated with criminal groups than with non-criminal control groups (Glueck & Glueck, 1956; Hartl, Monnelly, & Elderkin, 1982; Sheldon et al., 1949). This makes it difficult to simply dismiss the relevance of somatotypes as correlates to criminal behavior. The method by which a somatotype is measured is another factor leading to the lack of research on the link between somatotypes and criminality. Sheldon’s procedure (and reformulations of his method) required not only that subjects be naked, but that several pictures were taken of the naked subjects. Today, not only would a researcher have trouble finding subjects, but getting the research design approved by an Institutional Review Board (IRB) would prove unlikely. This paper attempts to further the literature on somatotypes and crime in two ways. First, since the actual process of somatotypic determination is comprised of numerous methodological and ethical considerations, the present work attempts to put forth a new, more refined procedure for measuring the morphologies of individuals, while at the same time making the process more reliable. Second, since there is no consensus as to a ranking of all crimes (Wolfgang, Figlio, Tracy, & Singer, 1985), this research attempted to examine whether a person’s somatotype was related to the crime an individual committed by analyzing initial violent and non-violent offenses. The next section of this paper explores the history of somatotyping and the empirical research surrounding the topic.
332
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
2. Literature review There have been numerous attempts to study the links between physique and criminality that have yielded different methods of measuring physique. These measurements ranged from the simplistic to the complex. For instance, Hooten (1968) separated height and weight into a three by three matrix, or nine subgroups, to see if there was a relationship between physique and crime; height was identified as short, medium, and tall, and weight was identified as slender, medium, and heavy. While a host of physique studies were carried out in Europe in the 1800s, one of the preeminent studies of physique, was completed by Kretschmer. Kretschmer (1925/1970), who most influenced Sheldon, studied 400 patients of every age and occupation in a hospital in an effort to link physique and character. Kretschmer’s key contribution to Sheldon’s research was the three types of physique he established: the asthenic, athletic, and pyknic forms. A person could contain characteristics from each of Kretschmer’s types, making for a blended physical type. The extreme asthenic type of physique has a “deficiency in thickness combined with an average unlessened length”(Kretschmer, 1925/1970, p. 22); this type is characterized as weak and frail. The extreme athletic type of physique is “recognized by the strong development of the skeleton, the musculature and also the skin” (Kretschmer, 1925/1970, p. 25). The extreme pyknic type of physique “is characterized by the pronounced peripheral development of the body cavities, and a tendency to a distribution of fat about the trunk, with a more graceful construction of the motor apparatus” (Kretschmer, 1925/1970, p. 30). Sheldon’s somatotypes would greatly resemble Kretschmer’s physical types. Drawing on this body of work, Sheldon et al. (1940) outlined the study of physique and the measurement of somatotypes in The Varieties of Human Physique. His theory was founded on the “assumption that it is possible to discriminate differences among human beings” (1940, p. 108). Sheldon decided on three physical types, adapted from embryology, which correspond approximately to Kretschmer’s: endomorphs (pycknic), mesomorphs (athletic), and ectomorphs (asthenic). These three types were selected because they illustrated the extreme cases of physique (Sheldon et al., 1940, p. 111). Sheldon et al. (1940, p. 5) defined endomorphy as the “relative predominance of soft roundness throughout the various regions of the body.” Mesomorphy means the “relative predominance of muscle, bone, and connective tissue” (1940, p. 5). Ectomorphy means the “relative predominance of linearity and fragility” (1940, p. 5); Sheldon noted that this form had the largest brain and central nervous system. To refine his method, Sheldon examined 400 male undergraduate students at the University of Chicago. In an effort at standardization, subjects were photographed from three angles (1940, p. 4): a front view, a side view, and a back view of the naked subject. The subject stood on a pedestal a certain distance away from the camera. From the 400 subjects, 4,000 photographs were amassed and examined. From these photographs, 17 measurements (dependent on the height to convert into ratio form) were taken from a body, and from this a three number scale was derived to conclude a person’s somatotype. The first number in the score reflects the amount of endomorphy in an individual, the second number represents the amount of mesomorphy in an individual, and the third number represents the amount of ectomorphy present in an individual; thus, an extreme endomorph would receive a score of 7-1-1, an extreme mesomorph would receive a score of 1-7-1, and an extreme ectomorph would receive a score
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
333
of 1-1-7. There are potentially 343 different identifiable somatotypes when using Sheldon’s classification procedure. Sheldon’s approach would be eclipsed by the work of Parnell (1958). Parnell’s method of somatotyping was considered to be more objective than Sheldon’s. This method of somatotyping emphasized the phenotype, not the somatotype. The phenotype is the body as it appears at a particular point in time (Parnell, 1958, p. 4). Because of this, Parnell indicated that his method for somatotyping was not a good variable for prediction purposes. Unlike Sheldon, Parnell labeled his physical types on the chart as Fat, Muscularity, and Linearity which correspond to the endomorph, mesomorph, and ectomorph, respectively. 2.1. Empirical evidence Research on the link between crime and human biology began in the 19th Century with the work of Lombroso, but it was Hooten who would first focus on physique and criminality. In the early 20th century, Hooten (1968) studied the physical characteristics of criminals in prisons across the United States. Hooten (1968, p. 9) wrote that it would be beneficial to examine the physical characteristics of large groups of delinquents to find out whether or not they were “physically homogenous”; his research was not to imply causality of crime, but merely to examine if there was a relationship between crime and physique. Although Hooten’s research suffered from the same sample selection bias as Lombroso’s work, his study was the first to examine the link between different kinds of physique and prisoners. Hooten (1969) examined 17,000 inmates from 10 states’ prisons. He measured physical characteristics (weight, height, and other measurements of the body), social variables (age, sex, race, etc.), mental variables, and morphological variables (moles, hair, skin color, and other various measures). Hooten (1969, p. 281) found that, in general, criminals tended to be smaller in height and weight, but recidivism decreased as both height and weight increased. Hooten also noticed that homicidal tendencies increased substantially with increments in height; on the other hand, tendencies to commit property crimes, burglary and larceny, decreased with increases in height. These results led Hooten (1969, p. 296) to conclude that “the variation in physique and body build [was] causally related to the nature of [the criminal] offense.” While this statement was a bit extreme, his findings certainly indicated at least a correlation between human physique and criminality. In 1949, Sheldon put his method of somatotyping to the test of studying crime for the first time. Sheldon followed the lives of a sample of 200 young men from the Hayden Goodwill Inn; because it was a social service agency, the sample consisted of youth with antisocial personalities as well as delinquent histories. This gave Sheldon’s study a comparison group of non-criminals on which to base his results. Although Sheldon examined many different sociological variables as well as biological variables, Sheldon’s (Sheldon et al., 1949, p. 805) chief finding among the criminal sample, in terms of somatotypes, was that mesomorphy was the most common somatotype. In essence, Sheldon concluded that delinquents were more inclined to being mesomorphically built. With the exception of a 30-year follow up study, this would also mark the last time that Sheldon’s method of somatotyping would be used in the study of the link between physique and crime.
334
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
In the 30-year follow-up of Sheldon’s research, Hartl et al. (1982) reexamined the 200 men whose biographies were presented by Sheldon, Hartl, and McDermott in 1949. Hartl et al.’s major finding was that future adult criminals differed from non-criminal subjects in the sample in terms of mesomorphy. As in Sheldon’s original study, criminals were more likely to have a mesomorphic build. In their discriminant function analysis on all of the data, Hartl et al. (1982) found mesomorphy to be the strongest discriminating variable. When conducting a multiple regression on criminal behavior, however, Hartl et al. (1982, p. 531) indicated that none of the variables relating to morphology, mesomorphy, endomorphy, or ectomorphy, were statistically significant. Other studies took place in the interim between Sheldon’s original study and the follow up study conducted by Hartl et al. Inspired by the work of Hooten and Sheldon, the Gluecks conducted an analysis on the relationship between physique and crime. The Glueck and Glueck (1951) compared the physiques between a sample of delinquents and a sample of non-delinquents. The Gluecks relied on a physical anthropologist to measure the somatotypes of their sample. Parnell’s method of somatotyping was used to study the physique of individuals in the Gluecks’ sample; the Gluecks merely interpreted the results. The Gluecks concluded that mesomorphy was more predominant among the delinquents and was the most “dominant component,” while the control group of non-delinquents contained no predominance of any single somatotype (1951, p. 196). Glueck and Glueck (1956) examined this data further by studying the physique of criminals in relation to other sociological variables and traits they had found. The Gluecks looked at the relationship between physique and several categories of variables that included neurological traits, intelligence, character, family, environment, and personality. While some of the variables in each category were found to be correlated with the different body types, the Gluecks concluded that there was no combination of physical traits and sociological traits which could predict delinquency in an individual. This was true even for the mesomorphs, who comprised the majority of the sample. Besides the Gluecks, two other studies utilized Parnell’s somatotyping procedure. Cortes and Gatti (1966) examined the relationship between a person’s physique and the need for achievement, or motivation. They examined 100 delinquent youths and a comparison group of 100 non-delinquents in a high school. In both groups, a significant and positive relationship was found between mesomorphy and motivation; they also found that a significant, but negative, relationship existed between ectomorphy and motivation. Cortes and Gatti (1966, p. 412) concluded that a relationship existed between mesomorphs and the desire to achieve. McCandless, Persons, and Roberts (1972) examined perceived criminal opportunity and body build among delinquent youth. McCandless et al. (1972) examined 500 adjudicated delinquent youth by both somatotyping the youth and administering a questionnaire to measure opportunity. McCandless et al. (1972) hypothesized that mesomorphs would be more likely to have committed more delinquent acts than either ectomorphs or endomorphs; however, their regression analysis lent no credence to this. Race was the only variable to have a significant relationship with perceived criminal opportunity (1972, p. 284). The biological approach to studying criminality has been much maligned. These studies have suffered from methodological issues, ranging from measurement to sampling, and conceptual
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
335
issues, ranging from un-testability to logic. One of the key problems with biological studies has been with regard to the types of variables included in analyses; most of the biological research, not including current work in this area has focused primarily on strictly biological variables. The other major problem for biological studies has been with regard to control groups. Most of the biological research has focused only on a single group; this has greatly decreased the worth of much of the biological research. Somatotyping, however, with its own set of limitations, fares better against these claims. There are several limitations to the research on delinquency and somatotypes. First, Sheldon’s method of somatotyping was very time consuming and expensive. It took 3–4 months to compare the data gathered from the three different pictures of each subject. This did not include the time it would take to analyze the data with other data in the study, in the case of criminology, delinquency and other structural variables. Second, and due to the time period in criminological development, little data was collected on females. Third, the study on the link between crime and biology, in general, and somatotyping, specifically, is that the results are either mixed or weak. Because of these findings, the focus shifted to sociological processes rather than biological differences across individuals. Sampson and Laub (1990) argued that this was further exacerbated by the early efforts and overreaction of sociologists to be the chief claims makers on the processes resulting in crime. All of the arguments against somatotyping miss one key point. The bulk of the evidence does suggest that criminals are more likely to have a mesomorphic build. While the relationship is a weak one, the relationship is consistent and it endures independent of most sociological variables (Glueck & Glueck, 1956). Until criminologists have determined what it is about mesomorphy that is related to criminality, work in this area should continue. While the current study does not overcome the sample selection bias of past studies, this research attempts to promulgate a new, more refined method of somatotyping. Unlike past research, this study centers completely around the issue of somatotyping and criminal behavior in presenting this new method. In the tradition of Parnell, this paper attempts to further the evolution of somatotyping by utilizing a new technique: the Body Mass Index. The Body Mass Index is a well established measure in the medical community; this is observed by a cursory review of the available literature (Anding, Suminski, & Boss, 2001; Feitosa et al., 2002; James et al., 2000; National Heart, Lung, and Blood Institute, 1998; Parsons, Power, & Manor, 2001). 2.2. The Body Mass Index The Body Mass Index (BMI), often referred to as the Quetelet Index, utilizes a person’s height and weight to gauge the total body fat in adults; it is an indicator of optimal weight for health and different from lean mass or percent body fat calculations because it only considers height and weight (National Heart, Lung, and Blood Institute, 1998). This measure correlates primarily to body fat and can be used on either males or females (but only for persons above the age of 15 or so, due to underage growth), unlike previous somatotyping techniques. The BMI is a “heterogeneous phenotype” (Feitosa et al., 2002, p. 72) or how the person’s physique is at a given point in time. In this sense, it is a measure of somatotyping more similar to Parsons’ method, which was based on physical phenotype as well.
336
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
A person’s BMI is calculated by dividing weight (in kilograms) by height (in meters) squared. Eq. (1) illustrates this formula: BMI =
wt (kgs) ht2 (m)
(1)
The BMI equation relies heavily upon the subject’s height. The relationship between the physique and height of an individual was discussed time and time again in the literature. Even Sheldon foreshadowed the importance of height and weight in measuring somatotypes; Sheldon et al. (1940, p. 805) wrote that with height–weight “norms. . . it will be possible to [create] a scale of height–weight standards for each different somatotype.” BMI values can range from one and up, making it a continuous, interval level variable. From the BMI measure it is easy to insert a somatotype label for an individual. According to Sheldon et al. (1940, p. 5), endomorphs have a “relative predominance of soft roundness throughout the various regions of the body” or in other words are overweight. According to the BMI, a person who is an endomorph receives a BMI score of 26 and above. A mesomorph has a “relative predominance of muscle, bone, and connective tissue”(Sheldon et al., 1940, p. 5), which is basically an athletically fit person. According to the BMI, a person who is a mesomorph will receive a BMI score of between 19 and 25. Ectomorphs, who have a “relative predominance of linearity and fragility” (1940, p. 5), are individuals who are underweight. According to the BMI, a person who is an ectomorph will receive a score of less than 19. To test the reliability of the BMI measure, we took Sheldon’s original sample data from Varieties of Delinquent Youth and compared the results of his somatotyping technique with the same sample using the BMI scale to determine the individual’s somatotype. Sheldon’s measures of different somatotypes did not preclude the possibility that an individual could include more than one somatotype. Unlike Sheldon’s measures, the BMI scale allows a person to have only one classification; hence, a person cannot be an endomorphic–mesomorph or a mesomorphic–ectomorph as appeared in Sheldon’s work. When this occurred, the BMI was considered to be the same as long as it correlated with one of the two labels in a particular case. Sheldon’s sample (N = 200) of somatotypes is compared with the BMI somatotypes in Table 1. Noting the values were similar for both measures of somatotyping, bivariate correlation analyses were completed to measure the relationship between the two types of ordinal level variables. All ordinal analyses of correlation and agreement indicated an association; for instance, Somer’s d had a value of 0.775 (p < .000) and Kappa had a value of 0.773 (p < .000). These analyses illustrate that the BMI is a reliable measure of somatotyping.
Table 1 Sheldon’s somatotypes compared with the BMI somatotypes Somatotype
Sheldon
BMI
Endomorphs (1) Mesomorphs (2) Ectomorphs (3)
81 (40.5%) 111 (55.5%) 8 (4%)
102 (51%) 97 (48.5%) 1 (0.5%)
Total
200 (100%)
200 (100%)
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
337
3. Method The primary question driving this research is: what is the relationship between human physique and violent criminality? The literature review indicated how previous criminologists tried to answer this question, i.e., Sheldon’s (Sheldon et al., 1940; Sheldon et al., 1949) and Parnell’s (1958) methods of somatotyping. The research hypothesis is that there is a relationship between different types of physique as measured by the Body Mass Index and different forms of criminal offenses (violent and property). 3.1. Data The data to be analyzed in this study was gathered from the Arkansas Department of Correction (ADC) for all prisons in the system dating from 1970 to 2000. A random sample of 5,000 male prisoners was drawn to evaluate the impact of the BMI on offending patterns. This data contained information from the prisoners at the time they began their respective sentences at a particular prison. Information on height, weight, the inmate’s crime, and the seriousness of the offense was obtained from this database. The fact that the information was gained as the inmate entered the system staves off any argument that a prisoner could become very athletic by working out in prison and thus growing into a mesomorphic build. 3.2. Dependent variables Prisoners in the ADC were sentenced for many different offenses. These offenses were aggregated into their respective Uniform Crime Reports (UCR) categories to make them easier to analyze; these were criminal homicide, forcible rape, robbery, aggravated assault, burglary, larceny-theft, motor vehicle theft, arson, and all other offenses were collapsed into one offense type labeled Type II Offenses. These offenses were then further aggregated to form the dependent variable of violent/nonviolent offenses for higher-level statistical analyses. Violent offenses consisted of criminal homicide, forcible rape, robbery, and aggravated assault. Non-violent offenses were comprised of burglary, larceny-theft, motor vehicle theft, arson, drug offenses, and all other Type II Offenses. This dependent variable is a dichotomous measure of violent and non-violent offenses (coded as non-violent = 0, violent = 1). It is hypothesized, based on prior research (Hooten, 1968; Hooten, 1969), that mesomorphs will be more likely to engage in violent criminality, whereas endomorphs and ectomorphs are more likely to engage in non-violent criminality. 3.3. Independent variables The primary independent variable of this research is the somatotype measure. As shown above, this is a more scientific measure of somatotyping that is positively related to Sheldon’s original method of somatotyping making it a reliable measure. For this study, the somatotype is measured in two ways. First, it is measured as strictly a Body Mass Index score (ranging from 0 to 100). However, the literature does not assume a linear relationship between somatotype and criminality; mesomorphs, which fall in between ectomorphs (smaller people) and
338
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
endomorphs (larger people), are considered more likely to engage in violent crime (Hooten, 1968; Hooten, 1969). As such, this research converted this interval level BMI measure into a partially ordered ordinal level variable. As was stated above, ectomorphs correspond to BMI values of less than 19, mesomorphs correspond to BMI values of 19–25, and endomorphs correspond to BMI values of 26 and above. BMI values can be comprised of values that have decimals. For the purposes of this research, the values of the BMI include the entire range of a number; for instance, a BMI score of 24 includes the values of 24.00–24.99. There was no rounding of BMI values. This new partially ordered ordinal variable was then converted to two dummy coded variables examining the ectomorphic inmates (other = 0; ectomorphs = 1) and endomorphic inmates (other = 0; endomorphs = 1) utilizing mesomorphs as the reference category. From the data, three other independent variables were examined in relation to the dependent variable: age, race, and educational level. Age was calculated by the subtraction of the offender’s date of birth from the date in which the offender began his prison sentence; this led to an interval level measure. Race was coded as a dichotomous variable (White = 0, non-White = 1) due to the fact that the overwhelming majority of the prison inmates in Arkansas from 1975 to 2000 were either White or African American. Education was measured as a dichotomous measure of whether or not the inmate had a high school diploma or GED equivalent (no diploma = 0; HS diploma/GED = 1). 3.4. Analytic strategy Several analytic techniques were utilized in this research. The first is univariate analyses of basic demographic features of the inmates. The second is a cross-tabulation of the different somatotypes with respect to UCR offenses to gain some further insight into offending patterns. Finally, the primary statistical technique incorporated by this research is a logistic regression analysis to assess the effects of physique on violent offending while controlling for demographic characteristics of offenders. The logit regression analysis discerned the impact of the BMI variable on violent offending. The findings of this analysis will be discussed in the next section. 3.5. Limitations There were several limitations to this study. First, the BMI is based exclusively on height and weight. In using the BMI as a measure of physique, the BMI does not distinguish between excessive fat and excessive muscle. Someone who is shorter, but is very bulked up in a muscular sense will have the same somatotype as someone who is short and extremely overweight. This measure of the somatotype thus suffers from a severe limitation. However, this limitation is similar to the one found with Sheldon’s original somatotyping procedure. Based on the threedigit number calculated by Sheldon, or Parnell, a person could be an endomorphic mesomorph or a mesomorphic endomorph. Thus, this limitation with the BMI as a measure of physique has been found in all somatotype research. Second, the subjects of this study came from the Arkansas Department of Correction. Only those criminals who were caught and were completely processed through the criminal justice system of Arkansas were present in this study. Criminals who were not detected or were
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
339
earlier removed from the justice system were naturally omitted from this study. In addition to those not caught, this research does not utilize a comparison group with which to evaluate findings. Thus, this study suffers from the same sample selection bias of past studies on the link between biology and criminality. This limits the generalizability of this study to those relegated to the prison system in Arkansas. However, this study does provide a portrait of prisoners’ somatotypes over a 30-year period in the Arkansas correctional system, which is an improvement over previous studies that only examined a prison over a single year. Further, the fact that the data came from the ADC meant that the research team was limited in the number of variables, sociological and biological, that could be used. While the lack of variables in this research is problematic, the focus is towards the impact of the biological to create a ground work for researchers in the future who have both sociological and biological indicators at their disposal. Third, inmates of the Arkansas Department of Corrections were measured in height and weight when they first came to their respective institutions. The criminal justice process could have made a gap in the time the offender committed the offense and the time in which the offender finally made it to the prison institution; this would give the offender time to either add or subtract from his or her weight which would influence the offender’s BMI. Fourth, while this research supported past research indicating that there was an overabundance of mesomorphs in the Arkansas prison population, the findings of this paper are again only applicable to Arkansas and not to other correctional systems in the U.S.
4. Findings The descriptive statistics for all of the variables in the analytic model are presented in Table 2. The majority of offenses that offenders were sent to prison for were non-violent offenses (burglary, theft, drug offenses, and other Type II Offenses). Table 2 also illustrates that the majority of inmates (68.6%) serving sentences in the Arkansas Department of Correction from 1975 to 2000 had mesomorphic physiques upon entering the correctional system. The majority of offenders were White (53%) and the mean age of offenders was closing on 32 years. Finally, the majority of inmates (62.8%) did not have a high school degree or GED equivalent. Table 3 shows cross tabulations of somatotypes and UCR offense types from offenders in Arkansas prisons. Results indicate that no matter what kind of offense is in question, the offender had a mesomorphic build ranging from 62 to 73% within UCR offense categories. Bivariate correlations indicated statistically significant relationships between all of the independent variables and the dichotomized dependent variable of violent and non-violent offenses. Age illustrated the highest association with the dependent variable (r = −0.218, p < .05). The interval somatotype measure (used as the actual BMI measure ranging from 1 to 100) showed a significant relationship (r = −0.09; p < .05) with violent and property crimes, albeit a small one. This was expected, however, since the theory does not suggest a linear progression of violent offending as one progresses from an ectomorph to a mesomorph to an endomorph; rather, mesomorphs are hypothesized to engage in more serious forms of crime which suggests a curvilinear relationship to crime. This, and the small relationship between the somatotype
340
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
Table 2 Descriptive statistics for all variables in the model Variable
Percent
UCR Offense
Homicide Rape Robbery Aggravated assault Burglary Theft Arson Drug offenses Type II offenses
16.2 9.1 19.1 1.6 19.8 10.1 0.0 0.3 23.8
Offense Type
Non-violent (0) Violent (1)
53.0 47.0
BMI
23.99
Somatotype
Ectomorph Mesomorph Endomorph
3.6 68.6 27.8
Race
White (0) Non-White (1)
53.0 47.0
Education
No degree (0) High school degree/GED (1)
62.8 34.8
Age
Mean
31.55
measure, was the rationale for coding the somatotype measure as two dummy variables; this allowed for the examination of each somatotype’s effect on violent/non-violent criminality separately. The mesomorphs served as the reference group. Although the relationships on the whole were relatively small, this analysis provided the basis for proceeding to the logistic regression analysis of the data. Table 4 presents the findings of the logistic regression testing the effects of the somatotype on the likelihood of engaging in violent or non-violent crime. Multicollinearity diagnostics indicated no severe levels of multicollinearity between the independent variables; the VIF, tolerance, and condition index number test were all in the appropriate ranges. Table 4 indicates that the model was significant, albeit it had only slight predictive power. The Nagelkerke R2 had a value of .073 (p < .05), which suggests that knowing the independent variables in the model only reduced the variance in explaining if an offender is imprisoned for a violent offense by 7%. The effects provided by the coefficients of the independent variables are of key importance. All of the independent variables in the model, with the exception of education, had statistically significant effects. Of the non-somatotype independent variables, race and age both had effects on violent offending. The impact of these variables was minimal as denoted by the odds column. In terms of the key independent variables utilized in this research, the dummy coded variables for ectomorphs and endomorphs with respect to mesomorphs, this logit model showed mixed results. The ectomorph dummy variable was not statistically significant in the model. This may be an artifact of the data as the number of ectomorphs in comparison to mesomorphs
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
341
Table 3 Cross-tabulations of somatotypes by UCR offense type Ectomorphs
Mesomorphs
Endomorphs
Total
Homicide
43 5.3% 23.8%
574 70.8% 16.7%
194 23.9% 13.9%
811 100%
Theft
13 2.8% 7.2%
334 73.1% 9.7%
110 24.1% 7.9%
457 100%
Aggravated Assaults
33 3.5% 18.2%
689 72.2% 20.1%
232 24.3% 16.7%
954 100%
Burglary
2 2.6% 1.1%
53 67.9% 1.5%
23 29.5% 1.7%
78 100%
Theft
40 4.0% 22.1%
704 71.1% 20.5%
246 24.8% 17.7%
990 100%
Arson
13 2.6% 7.2%
324 64.0% 9.5%
169 33.4% 12.1%
506 100%
All Other Offenses
37 3.1% 20.4%
750 62.2% 21.9%
417 34.6% 29.9%
1204 100%
Total
181/(100%)
3428/(100%)
1391/(100%)
5000/(100%)
and endomorphs was very minute (3% of the sample). The endomorph dummy variable was statistically significant in the model, albeit with very minor effect sizes (Odds = 0.868). Of more importance is the direction of the somatotypic dummy variables’ effects. According to the dummy variables, endomorphs were more likely to be incarcerated for non-violent offenses than mesomorphs. This was the same for ectomorphs as well even though this dummy variable failed to gain statistical significance. Mesomorphs were thus more likely to be incarTable 4 Logistic regression predicting violent criminality Variable Constant Endomorph dummy (endomorph = 1) Ectomorph dummy (ectomorph = 1) Age Education (HS degree = 1) Race (non-White = 1) −2 log likelihood Chi-square Nagelkerke R2 ∗
p < .05; two-tailed.
b
S.E. 1.440 −0.141 −0.029 −0.043 −0.106 −0.261
6424.92 272.68* .073
0.103 0.068 0.159 0.003 0.062 0.060
Wald
Odds *
197.20 4.276* 0.03 188.90* 2.88 19.56*
4.22 0.868 0.972 0.96 0.90 0.77
342
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
cerated for a violent offense than individuals with endomorphic and ectomorphic physiques. On the whole, the results of the logistic regression confirm our, and indeed Hooten’s, hypothesis that criminals with mesomorphic builds are more prone to violent offending. Conclusions based on these analyses will be discussed in the next section.
5. Discussion Raine (1993) asserted that “no factor linked to crime should be viewed in the anachronistic terms of genetics versus environment” (p. 204). To do this is both “divisive in disciplinary terms and overly simplistic” (p. 204). Raine (1993) further wrote that criminologists should move toward an integration of genetic and social factors for crime in an effort to examine how these two factors relate to one another. Although past research (until the mid-20th century) on the biological correlates of crime have been marred by poor methodological designs and a general un-acceptance by mainstream criminologists, and in accordance with new developments in biosocial and life course studies on the link between biology and crime, this research has attempted to contemporize a biological concept that has fallen to the wayside of criminological history: somatotyping. This research has advanced the study on somatotypes in two primary ways. First, due to the inherent difficulties with Sheldon’s (Sheldon et al., 1940; Sheldon et al., 1949) and Parnell’s (1958) methods of somatotyping, this study introduced a new method of somatotyping that was more methodologically and ethically sound. This new somatotype measure, based on the Body Mass Index (BMI), is highly correlated with the original method of somatotyping as presented by Sheldon, giving the new somatotype measure a high level of reliability. Thus, this new somatotypic measure can easily be used by criminologists in future studies of the biological correlates of crime. This is especially true when the data set analyzed has a greater availability of independent variables and the existence of a comparison group. Second, the current study examined the effects of particular somatotypes on violent offending patterns. The most important finding of this research was that somatotypes still accounted for some of the variation in violent and non-violent offending in a sample of male prisoners (N = 5000) taken in Arkansas from 1975 to 2000. While the somatotype showed effects, the impact was minor across the somatotype measures. The logistic regression analysis illustrated that the somatotypes had a statistically significant effect and added to the overall model. More importantly, the direction of the effects was in line with previous research on physique and criminality. Mesomorphic offenders were more likely to be in prison for a violent offense than their endomorphic counterparts; the direction was the same for ectomorphs, but this variable failed to gain statistical significance. While this research did indicate a minor effect of physique on criminal offending, on the whole, this research can also be interpreted as support for those criminologists who would discard biological determinants of criminality, specifically in relation to physique. Many criminologists simply dismiss any effect of physique, and other biological determinants of criminality, outright no matter what effects sizes are documented. As such, it will be incumbent on future criminologists and researchers to discern the efficacy of the use of the BMI as a measure of somatotypes in research projects.
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
343
This study has several strengths in terms of examining the links between somatotyping and criminality. Because the new method of somatotyping presented in this paper is based on the BMI scale, measuring a person’s somatotype is a more methodologically/scientifically sound process. It cannot be based on a researcher’s subjectivity, but on numbers provided in a BMI scale conversion chart or by the BMI calculation formula. This gives the study of somatotyping greater reliability. Second, and based on the first strength, each subject can only be one kind of somatotype. There is no mixture of somatotypes when using the BMI measure, thus reducing the researcher’s subjectivity when a case is not an extreme endomorph, mesomorph, or ectomorph. Third, by using the BMI, all races and both sexes can be easily measured, whereas, in the past, the majority of the subjects were White males. Somatotyping utilizing the BMI can be used to measure anyone except children and teenage adolescents who are still in the physical maturation process. Future research on the causes of criminality should consider utilizing the BMI when information about the offender’s height and weight are included in criminal justice databases.
References Anding, J. D., Suminski, R. R., & Boss, L. (2001). Dietary intake, Body Mass Index, exercise, and alcohol: Are college women following the dietary guidelines for Americans? Journal of American College Health, 49(4), 167–171. Cortes, J. B., & Gatti, F. M. (1966). Physique and motivation. Journal of consulting psychology, 30(5), 408–414. Eysenck, H. J., & Gudjonsson, G. H. (1989). The causes and cures of criminality. New York: Plenum Press. Feitosa, M. F., Borecki, I. B., Rich, S. S., Arnett, D. K., Sholinsky, P., Myers, R. H., Leppert, M., & Province, M. A. (2002). Quantitative-trait loci influencing body-mass index reside on chromosomes 7 and 13: The National Heart, Lung, and Blood Institute family heart study. American Journal of Human Genetics, 70(1), 72–82. Glueck, S., & Glueck, E. (1951). Unraveling juvenile delinquency. Cambridge: Harvard University Press. Glueck, S., & Glueck, E. (1956). Physique and delinquency. New York: Harper and Brothers Publishers. Hartl, E. M., Monnelly, E. P., & Elderkin, R. D. (1982). Physique and delinquent behavior: A thirty-year follow-up of William H. Sheldon’s varieties of delinquent youth. New York: Academic Press. Hooten, E. A. (1968). Crime and the man. New York: Greenwood Press, Publishers. Hooten, E. A. (1969). The American criminal: An anthropological study. Greenwood Press, Publishers. James, P. W., Astrup, A., Finer, N., Hilsted, J., Kopelman, P., Rossner, S., Saris, W., & Van Gaal, L. F. (2000). Effect of Sibutramine on weight maintenance after weight loss: A randomized trial. Lancet, 356(9248), 2119–2126. Kretschmer, E. (1925/1970). Physique and character: An investigation of the nature of constitution and of the theory of temperment. New York: Cooper Square Publishers, Inc. McCandless, B. R., Persons, W. S., III, & Roberts, A. (1972). Perceived opportunity, delinquency, race, and body build among delinquent youth. Journal of Consulting and Clinical Psychology, 38(2), 281–287. National Heart, Lung, and Blood Institute. (1998). Clinical guidelines on the identification, evaluation, and treatment of overweight, and obesity in adults: The evidence report. Washington D.C.: National Institute of Health. Parnell, R. W. (1958). Behavior and physique: An introduction to practical and applied somatometry. London: Edward Arnold Publishers, Inc. Parsons, T. J., Power, C., & Manor, O. (2001). Fetal and early life growth and Body Mass Index from birth to early adulthood in a 1958 British cohort: A longitudinal study. British Medical Journal, 323(7325), 1331–1335. Raine, A. (1993). The psychopathology of crime: Criminal behavior as a clinical disorder. New York: Academic Press. Sampson, R. J., & Laub, J. H. (1990). Crime in the making: Pathways and turning points through life. Boston: Harvard University Press.
344
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
Sheldon, W., Stevens, S. S., & Tucker, W. B. (1940). The varieties of human physique: An introduction to constitutional psychology. New York: Harper and Brothers Publishers. Sheldon, W., Hartl, W. M., & McDermott, E. (1949). Varieties of delinquent youth: An introduction to constitutional psychiatry. New York: Harper and Brothers Publishers. Walsh, A. (2002). Biosocial criminology: Introduction and integration. Cincinnati: Anderson Publishing Co. Wilson, J. Q., & Herrnstein, R. J. (1985). Crime and human nature. New York: Simon and Schuster, Inc. Wolfgang, M. E., Figlio, R. M., Tracy, P. E., & Singer, S. I. (1985). The national survey of crime severity. Washington D.C.: National Institute of Justice.
The Social Science Journal 45 (2008) 330–344
Does size really matter? A reexamination of Sheldon’s somatotypes and criminal behavior Sean Maddan a , Jeffery T. Walker b,∗ , J. Mitchell Miller c a
Department of Criminology, University of Tampa, 401 West Kennedy Boulevard Tampa, FL 33606, USA Department of Criminal Justice, University of Arkansas - Little Rock, 2801 South University, Little Rock, AR 72204-1099, USA c Department of Criminal Justice, University of Texas - San Antonio, 501 West Durango Boulevard, San Antonio, TX 78207, USA b
Abstract In the mid-20th century, William Sheldon developed the idea of somatotypes in explaining different types of criminal behavior. Sheldon defined three body types believed to be associated with criminality: mesomorphs, who are athletically fit; endomorphs, who are overweight, and ectomorphs, who are characterized by fragility and thinness. Sheldon’s research implied that mesomorphic individuals were more prone to committing violent and aggressive acts. This research explores a new method of somatotyping using the Body Mass Index (BMI). What follows is an examination of prisoners from the State of Arkansas, using information provided by the Arkansas Department of Correction’s database on inmates. In particular, this paper examined prisoners’ physical characteristics such as age, height, and weight in comparison to the types of crimes that these prisoners committed. The findings of this paper show that the BMI is a useful alternative to traditional somatotyping techniques; the findings also suggest that a prisoner’s somatotype is associated with criminal patterns while being a meager predictor of criminality. Methodological and theoretical implications of this study are discussed. © 2008 Elsevier Inc. All rights reserved.
1. Introduction From the earliest days of criminological thought, researchers have sought biological causes of criminality. Because some of the earlier biological studies were disproved due to method∗
Corresponding author. Tel.: +1 501 569 3083. E-mail address: [email protected] (J.T. Walker).
0362-3319/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.soscij.2008.03.009
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
331
ological shortcomings or antiquated statistical techniques, a majority of criminologists today dismiss the biological research of criminality with relative ease. As Wilson and Herrnstein (1985, p. 80) wrote “many criminologists. . . are simply uneasy with biological and psychological concepts” involved with the causes of criminality due to their sociological training. In fact, “physical correlates of crime are often dismissed by most criminologists, for whom these are at best historical stages in the development of their subject”(Wilson & Herrnstein, 1985, p. 72). The last decade has seen a resurgence in the study of the links between biology and crime (Raine, 1993; Walsh, 2002). Under the mantle of the biological theories of criminality, the debate of whether or not criminals are characterized by a certain type of body build has long been debated in the field of criminology. Eysenck and Gudjonsson (1989, p. 22) wrote that the study of physique “is a complex subject, with many different typologies and methods of measurement.” In the 1940’s, William Sheldon (Sheldon, Stevens, & Tucker, 1940; Sheldon, Hartl, & McDermott, 1949) attempted to study the relationship between human physique and criminality. Although Sheldon was not the first to attempt such research, his method of doing so was more advanced than his predecessors. Sheldon put forth the idea of the somatotype; this is a personal score that is determined by different measurements taken from a human body. Sheldon posited three types of extreme somatotypes: mesomorphs who are athletically built, endomorphs who are overweight, and ectomorphs who are underweight. While several studies have been conducted on somatotyping and criminality over the last 50 years, overall interest in the subject has waned. This is due primarily to the stigma that has been associated with conducting research on the link between biological causes of crime in the past. Even though this is the case, it is hard to discard some of the findings from the somatotype literature, in spite of some methodological shortcomings. Research has consistently shown that mesomorphic physiques are more associated with criminal groups than with non-criminal control groups (Glueck & Glueck, 1956; Hartl, Monnelly, & Elderkin, 1982; Sheldon et al., 1949). This makes it difficult to simply dismiss the relevance of somatotypes as correlates to criminal behavior. The method by which a somatotype is measured is another factor leading to the lack of research on the link between somatotypes and criminality. Sheldon’s procedure (and reformulations of his method) required not only that subjects be naked, but that several pictures were taken of the naked subjects. Today, not only would a researcher have trouble finding subjects, but getting the research design approved by an Institutional Review Board (IRB) would prove unlikely. This paper attempts to further the literature on somatotypes and crime in two ways. First, since the actual process of somatotypic determination is comprised of numerous methodological and ethical considerations, the present work attempts to put forth a new, more refined procedure for measuring the morphologies of individuals, while at the same time making the process more reliable. Second, since there is no consensus as to a ranking of all crimes (Wolfgang, Figlio, Tracy, & Singer, 1985), this research attempted to examine whether a person’s somatotype was related to the crime an individual committed by analyzing initial violent and non-violent offenses. The next section of this paper explores the history of somatotyping and the empirical research surrounding the topic.
332
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
2. Literature review There have been numerous attempts to study the links between physique and criminality that have yielded different methods of measuring physique. These measurements ranged from the simplistic to the complex. For instance, Hooten (1968) separated height and weight into a three by three matrix, or nine subgroups, to see if there was a relationship between physique and crime; height was identified as short, medium, and tall, and weight was identified as slender, medium, and heavy. While a host of physique studies were carried out in Europe in the 1800s, one of the preeminent studies of physique, was completed by Kretschmer. Kretschmer (1925/1970), who most influenced Sheldon, studied 400 patients of every age and occupation in a hospital in an effort to link physique and character. Kretschmer’s key contribution to Sheldon’s research was the three types of physique he established: the asthenic, athletic, and pyknic forms. A person could contain characteristics from each of Kretschmer’s types, making for a blended physical type. The extreme asthenic type of physique has a “deficiency in thickness combined with an average unlessened length”(Kretschmer, 1925/1970, p. 22); this type is characterized as weak and frail. The extreme athletic type of physique is “recognized by the strong development of the skeleton, the musculature and also the skin” (Kretschmer, 1925/1970, p. 25). The extreme pyknic type of physique “is characterized by the pronounced peripheral development of the body cavities, and a tendency to a distribution of fat about the trunk, with a more graceful construction of the motor apparatus” (Kretschmer, 1925/1970, p. 30). Sheldon’s somatotypes would greatly resemble Kretschmer’s physical types. Drawing on this body of work, Sheldon et al. (1940) outlined the study of physique and the measurement of somatotypes in The Varieties of Human Physique. His theory was founded on the “assumption that it is possible to discriminate differences among human beings” (1940, p. 108). Sheldon decided on three physical types, adapted from embryology, which correspond approximately to Kretschmer’s: endomorphs (pycknic), mesomorphs (athletic), and ectomorphs (asthenic). These three types were selected because they illustrated the extreme cases of physique (Sheldon et al., 1940, p. 111). Sheldon et al. (1940, p. 5) defined endomorphy as the “relative predominance of soft roundness throughout the various regions of the body.” Mesomorphy means the “relative predominance of muscle, bone, and connective tissue” (1940, p. 5). Ectomorphy means the “relative predominance of linearity and fragility” (1940, p. 5); Sheldon noted that this form had the largest brain and central nervous system. To refine his method, Sheldon examined 400 male undergraduate students at the University of Chicago. In an effort at standardization, subjects were photographed from three angles (1940, p. 4): a front view, a side view, and a back view of the naked subject. The subject stood on a pedestal a certain distance away from the camera. From the 400 subjects, 4,000 photographs were amassed and examined. From these photographs, 17 measurements (dependent on the height to convert into ratio form) were taken from a body, and from this a three number scale was derived to conclude a person’s somatotype. The first number in the score reflects the amount of endomorphy in an individual, the second number represents the amount of mesomorphy in an individual, and the third number represents the amount of ectomorphy present in an individual; thus, an extreme endomorph would receive a score of 7-1-1, an extreme mesomorph would receive a score of 1-7-1, and an extreme ectomorph would receive a score
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
333
of 1-1-7. There are potentially 343 different identifiable somatotypes when using Sheldon’s classification procedure. Sheldon’s approach would be eclipsed by the work of Parnell (1958). Parnell’s method of somatotyping was considered to be more objective than Sheldon’s. This method of somatotyping emphasized the phenotype, not the somatotype. The phenotype is the body as it appears at a particular point in time (Parnell, 1958, p. 4). Because of this, Parnell indicated that his method for somatotyping was not a good variable for prediction purposes. Unlike Sheldon, Parnell labeled his physical types on the chart as Fat, Muscularity, and Linearity which correspond to the endomorph, mesomorph, and ectomorph, respectively. 2.1. Empirical evidence Research on the link between crime and human biology began in the 19th Century with the work of Lombroso, but it was Hooten who would first focus on physique and criminality. In the early 20th century, Hooten (1968) studied the physical characteristics of criminals in prisons across the United States. Hooten (1968, p. 9) wrote that it would be beneficial to examine the physical characteristics of large groups of delinquents to find out whether or not they were “physically homogenous”; his research was not to imply causality of crime, but merely to examine if there was a relationship between crime and physique. Although Hooten’s research suffered from the same sample selection bias as Lombroso’s work, his study was the first to examine the link between different kinds of physique and prisoners. Hooten (1969) examined 17,000 inmates from 10 states’ prisons. He measured physical characteristics (weight, height, and other measurements of the body), social variables (age, sex, race, etc.), mental variables, and morphological variables (moles, hair, skin color, and other various measures). Hooten (1969, p. 281) found that, in general, criminals tended to be smaller in height and weight, but recidivism decreased as both height and weight increased. Hooten also noticed that homicidal tendencies increased substantially with increments in height; on the other hand, tendencies to commit property crimes, burglary and larceny, decreased with increases in height. These results led Hooten (1969, p. 296) to conclude that “the variation in physique and body build [was] causally related to the nature of [the criminal] offense.” While this statement was a bit extreme, his findings certainly indicated at least a correlation between human physique and criminality. In 1949, Sheldon put his method of somatotyping to the test of studying crime for the first time. Sheldon followed the lives of a sample of 200 young men from the Hayden Goodwill Inn; because it was a social service agency, the sample consisted of youth with antisocial personalities as well as delinquent histories. This gave Sheldon’s study a comparison group of non-criminals on which to base his results. Although Sheldon examined many different sociological variables as well as biological variables, Sheldon’s (Sheldon et al., 1949, p. 805) chief finding among the criminal sample, in terms of somatotypes, was that mesomorphy was the most common somatotype. In essence, Sheldon concluded that delinquents were more inclined to being mesomorphically built. With the exception of a 30-year follow up study, this would also mark the last time that Sheldon’s method of somatotyping would be used in the study of the link between physique and crime.
334
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
In the 30-year follow-up of Sheldon’s research, Hartl et al. (1982) reexamined the 200 men whose biographies were presented by Sheldon, Hartl, and McDermott in 1949. Hartl et al.’s major finding was that future adult criminals differed from non-criminal subjects in the sample in terms of mesomorphy. As in Sheldon’s original study, criminals were more likely to have a mesomorphic build. In their discriminant function analysis on all of the data, Hartl et al. (1982) found mesomorphy to be the strongest discriminating variable. When conducting a multiple regression on criminal behavior, however, Hartl et al. (1982, p. 531) indicated that none of the variables relating to morphology, mesomorphy, endomorphy, or ectomorphy, were statistically significant. Other studies took place in the interim between Sheldon’s original study and the follow up study conducted by Hartl et al. Inspired by the work of Hooten and Sheldon, the Gluecks conducted an analysis on the relationship between physique and crime. The Glueck and Glueck (1951) compared the physiques between a sample of delinquents and a sample of non-delinquents. The Gluecks relied on a physical anthropologist to measure the somatotypes of their sample. Parnell’s method of somatotyping was used to study the physique of individuals in the Gluecks’ sample; the Gluecks merely interpreted the results. The Gluecks concluded that mesomorphy was more predominant among the delinquents and was the most “dominant component,” while the control group of non-delinquents contained no predominance of any single somatotype (1951, p. 196). Glueck and Glueck (1956) examined this data further by studying the physique of criminals in relation to other sociological variables and traits they had found. The Gluecks looked at the relationship between physique and several categories of variables that included neurological traits, intelligence, character, family, environment, and personality. While some of the variables in each category were found to be correlated with the different body types, the Gluecks concluded that there was no combination of physical traits and sociological traits which could predict delinquency in an individual. This was true even for the mesomorphs, who comprised the majority of the sample. Besides the Gluecks, two other studies utilized Parnell’s somatotyping procedure. Cortes and Gatti (1966) examined the relationship between a person’s physique and the need for achievement, or motivation. They examined 100 delinquent youths and a comparison group of 100 non-delinquents in a high school. In both groups, a significant and positive relationship was found between mesomorphy and motivation; they also found that a significant, but negative, relationship existed between ectomorphy and motivation. Cortes and Gatti (1966, p. 412) concluded that a relationship existed between mesomorphs and the desire to achieve. McCandless, Persons, and Roberts (1972) examined perceived criminal opportunity and body build among delinquent youth. McCandless et al. (1972) examined 500 adjudicated delinquent youth by both somatotyping the youth and administering a questionnaire to measure opportunity. McCandless et al. (1972) hypothesized that mesomorphs would be more likely to have committed more delinquent acts than either ectomorphs or endomorphs; however, their regression analysis lent no credence to this. Race was the only variable to have a significant relationship with perceived criminal opportunity (1972, p. 284). The biological approach to studying criminality has been much maligned. These studies have suffered from methodological issues, ranging from measurement to sampling, and conceptual
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
335
issues, ranging from un-testability to logic. One of the key problems with biological studies has been with regard to the types of variables included in analyses; most of the biological research, not including current work in this area has focused primarily on strictly biological variables. The other major problem for biological studies has been with regard to control groups. Most of the biological research has focused only on a single group; this has greatly decreased the worth of much of the biological research. Somatotyping, however, with its own set of limitations, fares better against these claims. There are several limitations to the research on delinquency and somatotypes. First, Sheldon’s method of somatotyping was very time consuming and expensive. It took 3–4 months to compare the data gathered from the three different pictures of each subject. This did not include the time it would take to analyze the data with other data in the study, in the case of criminology, delinquency and other structural variables. Second, and due to the time period in criminological development, little data was collected on females. Third, the study on the link between crime and biology, in general, and somatotyping, specifically, is that the results are either mixed or weak. Because of these findings, the focus shifted to sociological processes rather than biological differences across individuals. Sampson and Laub (1990) argued that this was further exacerbated by the early efforts and overreaction of sociologists to be the chief claims makers on the processes resulting in crime. All of the arguments against somatotyping miss one key point. The bulk of the evidence does suggest that criminals are more likely to have a mesomorphic build. While the relationship is a weak one, the relationship is consistent and it endures independent of most sociological variables (Glueck & Glueck, 1956). Until criminologists have determined what it is about mesomorphy that is related to criminality, work in this area should continue. While the current study does not overcome the sample selection bias of past studies, this research attempts to promulgate a new, more refined method of somatotyping. Unlike past research, this study centers completely around the issue of somatotyping and criminal behavior in presenting this new method. In the tradition of Parnell, this paper attempts to further the evolution of somatotyping by utilizing a new technique: the Body Mass Index. The Body Mass Index is a well established measure in the medical community; this is observed by a cursory review of the available literature (Anding, Suminski, & Boss, 2001; Feitosa et al., 2002; James et al., 2000; National Heart, Lung, and Blood Institute, 1998; Parsons, Power, & Manor, 2001). 2.2. The Body Mass Index The Body Mass Index (BMI), often referred to as the Quetelet Index, utilizes a person’s height and weight to gauge the total body fat in adults; it is an indicator of optimal weight for health and different from lean mass or percent body fat calculations because it only considers height and weight (National Heart, Lung, and Blood Institute, 1998). This measure correlates primarily to body fat and can be used on either males or females (but only for persons above the age of 15 or so, due to underage growth), unlike previous somatotyping techniques. The BMI is a “heterogeneous phenotype” (Feitosa et al., 2002, p. 72) or how the person’s physique is at a given point in time. In this sense, it is a measure of somatotyping more similar to Parsons’ method, which was based on physical phenotype as well.
336
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
A person’s BMI is calculated by dividing weight (in kilograms) by height (in meters) squared. Eq. (1) illustrates this formula: BMI =
wt (kgs) ht2 (m)
(1)
The BMI equation relies heavily upon the subject’s height. The relationship between the physique and height of an individual was discussed time and time again in the literature. Even Sheldon foreshadowed the importance of height and weight in measuring somatotypes; Sheldon et al. (1940, p. 805) wrote that with height–weight “norms. . . it will be possible to [create] a scale of height–weight standards for each different somatotype.” BMI values can range from one and up, making it a continuous, interval level variable. From the BMI measure it is easy to insert a somatotype label for an individual. According to Sheldon et al. (1940, p. 5), endomorphs have a “relative predominance of soft roundness throughout the various regions of the body” or in other words are overweight. According to the BMI, a person who is an endomorph receives a BMI score of 26 and above. A mesomorph has a “relative predominance of muscle, bone, and connective tissue”(Sheldon et al., 1940, p. 5), which is basically an athletically fit person. According to the BMI, a person who is a mesomorph will receive a BMI score of between 19 and 25. Ectomorphs, who have a “relative predominance of linearity and fragility” (1940, p. 5), are individuals who are underweight. According to the BMI, a person who is an ectomorph will receive a score of less than 19. To test the reliability of the BMI measure, we took Sheldon’s original sample data from Varieties of Delinquent Youth and compared the results of his somatotyping technique with the same sample using the BMI scale to determine the individual’s somatotype. Sheldon’s measures of different somatotypes did not preclude the possibility that an individual could include more than one somatotype. Unlike Sheldon’s measures, the BMI scale allows a person to have only one classification; hence, a person cannot be an endomorphic–mesomorph or a mesomorphic–ectomorph as appeared in Sheldon’s work. When this occurred, the BMI was considered to be the same as long as it correlated with one of the two labels in a particular case. Sheldon’s sample (N = 200) of somatotypes is compared with the BMI somatotypes in Table 1. Noting the values were similar for both measures of somatotyping, bivariate correlation analyses were completed to measure the relationship between the two types of ordinal level variables. All ordinal analyses of correlation and agreement indicated an association; for instance, Somer’s d had a value of 0.775 (p < .000) and Kappa had a value of 0.773 (p < .000). These analyses illustrate that the BMI is a reliable measure of somatotyping.
Table 1 Sheldon’s somatotypes compared with the BMI somatotypes Somatotype
Sheldon
BMI
Endomorphs (1) Mesomorphs (2) Ectomorphs (3)
81 (40.5%) 111 (55.5%) 8 (4%)
102 (51%) 97 (48.5%) 1 (0.5%)
Total
200 (100%)
200 (100%)
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
337
3. Method The primary question driving this research is: what is the relationship between human physique and violent criminality? The literature review indicated how previous criminologists tried to answer this question, i.e., Sheldon’s (Sheldon et al., 1940; Sheldon et al., 1949) and Parnell’s (1958) methods of somatotyping. The research hypothesis is that there is a relationship between different types of physique as measured by the Body Mass Index and different forms of criminal offenses (violent and property). 3.1. Data The data to be analyzed in this study was gathered from the Arkansas Department of Correction (ADC) for all prisons in the system dating from 1970 to 2000. A random sample of 5,000 male prisoners was drawn to evaluate the impact of the BMI on offending patterns. This data contained information from the prisoners at the time they began their respective sentences at a particular prison. Information on height, weight, the inmate’s crime, and the seriousness of the offense was obtained from this database. The fact that the information was gained as the inmate entered the system staves off any argument that a prisoner could become very athletic by working out in prison and thus growing into a mesomorphic build. 3.2. Dependent variables Prisoners in the ADC were sentenced for many different offenses. These offenses were aggregated into their respective Uniform Crime Reports (UCR) categories to make them easier to analyze; these were criminal homicide, forcible rape, robbery, aggravated assault, burglary, larceny-theft, motor vehicle theft, arson, and all other offenses were collapsed into one offense type labeled Type II Offenses. These offenses were then further aggregated to form the dependent variable of violent/nonviolent offenses for higher-level statistical analyses. Violent offenses consisted of criminal homicide, forcible rape, robbery, and aggravated assault. Non-violent offenses were comprised of burglary, larceny-theft, motor vehicle theft, arson, drug offenses, and all other Type II Offenses. This dependent variable is a dichotomous measure of violent and non-violent offenses (coded as non-violent = 0, violent = 1). It is hypothesized, based on prior research (Hooten, 1968; Hooten, 1969), that mesomorphs will be more likely to engage in violent criminality, whereas endomorphs and ectomorphs are more likely to engage in non-violent criminality. 3.3. Independent variables The primary independent variable of this research is the somatotype measure. As shown above, this is a more scientific measure of somatotyping that is positively related to Sheldon’s original method of somatotyping making it a reliable measure. For this study, the somatotype is measured in two ways. First, it is measured as strictly a Body Mass Index score (ranging from 0 to 100). However, the literature does not assume a linear relationship between somatotype and criminality; mesomorphs, which fall in between ectomorphs (smaller people) and
338
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
endomorphs (larger people), are considered more likely to engage in violent crime (Hooten, 1968; Hooten, 1969). As such, this research converted this interval level BMI measure into a partially ordered ordinal level variable. As was stated above, ectomorphs correspond to BMI values of less than 19, mesomorphs correspond to BMI values of 19–25, and endomorphs correspond to BMI values of 26 and above. BMI values can be comprised of values that have decimals. For the purposes of this research, the values of the BMI include the entire range of a number; for instance, a BMI score of 24 includes the values of 24.00–24.99. There was no rounding of BMI values. This new partially ordered ordinal variable was then converted to two dummy coded variables examining the ectomorphic inmates (other = 0; ectomorphs = 1) and endomorphic inmates (other = 0; endomorphs = 1) utilizing mesomorphs as the reference category. From the data, three other independent variables were examined in relation to the dependent variable: age, race, and educational level. Age was calculated by the subtraction of the offender’s date of birth from the date in which the offender began his prison sentence; this led to an interval level measure. Race was coded as a dichotomous variable (White = 0, non-White = 1) due to the fact that the overwhelming majority of the prison inmates in Arkansas from 1975 to 2000 were either White or African American. Education was measured as a dichotomous measure of whether or not the inmate had a high school diploma or GED equivalent (no diploma = 0; HS diploma/GED = 1). 3.4. Analytic strategy Several analytic techniques were utilized in this research. The first is univariate analyses of basic demographic features of the inmates. The second is a cross-tabulation of the different somatotypes with respect to UCR offenses to gain some further insight into offending patterns. Finally, the primary statistical technique incorporated by this research is a logistic regression analysis to assess the effects of physique on violent offending while controlling for demographic characteristics of offenders. The logit regression analysis discerned the impact of the BMI variable on violent offending. The findings of this analysis will be discussed in the next section. 3.5. Limitations There were several limitations to this study. First, the BMI is based exclusively on height and weight. In using the BMI as a measure of physique, the BMI does not distinguish between excessive fat and excessive muscle. Someone who is shorter, but is very bulked up in a muscular sense will have the same somatotype as someone who is short and extremely overweight. This measure of the somatotype thus suffers from a severe limitation. However, this limitation is similar to the one found with Sheldon’s original somatotyping procedure. Based on the threedigit number calculated by Sheldon, or Parnell, a person could be an endomorphic mesomorph or a mesomorphic endomorph. Thus, this limitation with the BMI as a measure of physique has been found in all somatotype research. Second, the subjects of this study came from the Arkansas Department of Correction. Only those criminals who were caught and were completely processed through the criminal justice system of Arkansas were present in this study. Criminals who were not detected or were
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
339
earlier removed from the justice system were naturally omitted from this study. In addition to those not caught, this research does not utilize a comparison group with which to evaluate findings. Thus, this study suffers from the same sample selection bias of past studies on the link between biology and criminality. This limits the generalizability of this study to those relegated to the prison system in Arkansas. However, this study does provide a portrait of prisoners’ somatotypes over a 30-year period in the Arkansas correctional system, which is an improvement over previous studies that only examined a prison over a single year. Further, the fact that the data came from the ADC meant that the research team was limited in the number of variables, sociological and biological, that could be used. While the lack of variables in this research is problematic, the focus is towards the impact of the biological to create a ground work for researchers in the future who have both sociological and biological indicators at their disposal. Third, inmates of the Arkansas Department of Corrections were measured in height and weight when they first came to their respective institutions. The criminal justice process could have made a gap in the time the offender committed the offense and the time in which the offender finally made it to the prison institution; this would give the offender time to either add or subtract from his or her weight which would influence the offender’s BMI. Fourth, while this research supported past research indicating that there was an overabundance of mesomorphs in the Arkansas prison population, the findings of this paper are again only applicable to Arkansas and not to other correctional systems in the U.S.
4. Findings The descriptive statistics for all of the variables in the analytic model are presented in Table 2. The majority of offenses that offenders were sent to prison for were non-violent offenses (burglary, theft, drug offenses, and other Type II Offenses). Table 2 also illustrates that the majority of inmates (68.6%) serving sentences in the Arkansas Department of Correction from 1975 to 2000 had mesomorphic physiques upon entering the correctional system. The majority of offenders were White (53%) and the mean age of offenders was closing on 32 years. Finally, the majority of inmates (62.8%) did not have a high school degree or GED equivalent. Table 3 shows cross tabulations of somatotypes and UCR offense types from offenders in Arkansas prisons. Results indicate that no matter what kind of offense is in question, the offender had a mesomorphic build ranging from 62 to 73% within UCR offense categories. Bivariate correlations indicated statistically significant relationships between all of the independent variables and the dichotomized dependent variable of violent and non-violent offenses. Age illustrated the highest association with the dependent variable (r = −0.218, p < .05). The interval somatotype measure (used as the actual BMI measure ranging from 1 to 100) showed a significant relationship (r = −0.09; p < .05) with violent and property crimes, albeit a small one. This was expected, however, since the theory does not suggest a linear progression of violent offending as one progresses from an ectomorph to a mesomorph to an endomorph; rather, mesomorphs are hypothesized to engage in more serious forms of crime which suggests a curvilinear relationship to crime. This, and the small relationship between the somatotype
340
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
Table 2 Descriptive statistics for all variables in the model Variable
Percent
UCR Offense
Homicide Rape Robbery Aggravated assault Burglary Theft Arson Drug offenses Type II offenses
16.2 9.1 19.1 1.6 19.8 10.1 0.0 0.3 23.8
Offense Type
Non-violent (0) Violent (1)
53.0 47.0
BMI
23.99
Somatotype
Ectomorph Mesomorph Endomorph
3.6 68.6 27.8
Race
White (0) Non-White (1)
53.0 47.0
Education
No degree (0) High school degree/GED (1)
62.8 34.8
Age
Mean
31.55
measure, was the rationale for coding the somatotype measure as two dummy variables; this allowed for the examination of each somatotype’s effect on violent/non-violent criminality separately. The mesomorphs served as the reference group. Although the relationships on the whole were relatively small, this analysis provided the basis for proceeding to the logistic regression analysis of the data. Table 4 presents the findings of the logistic regression testing the effects of the somatotype on the likelihood of engaging in violent or non-violent crime. Multicollinearity diagnostics indicated no severe levels of multicollinearity between the independent variables; the VIF, tolerance, and condition index number test were all in the appropriate ranges. Table 4 indicates that the model was significant, albeit it had only slight predictive power. The Nagelkerke R2 had a value of .073 (p < .05), which suggests that knowing the independent variables in the model only reduced the variance in explaining if an offender is imprisoned for a violent offense by 7%. The effects provided by the coefficients of the independent variables are of key importance. All of the independent variables in the model, with the exception of education, had statistically significant effects. Of the non-somatotype independent variables, race and age both had effects on violent offending. The impact of these variables was minimal as denoted by the odds column. In terms of the key independent variables utilized in this research, the dummy coded variables for ectomorphs and endomorphs with respect to mesomorphs, this logit model showed mixed results. The ectomorph dummy variable was not statistically significant in the model. This may be an artifact of the data as the number of ectomorphs in comparison to mesomorphs
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
341
Table 3 Cross-tabulations of somatotypes by UCR offense type Ectomorphs
Mesomorphs
Endomorphs
Total
Homicide
43 5.3% 23.8%
574 70.8% 16.7%
194 23.9% 13.9%
811 100%
Theft
13 2.8% 7.2%
334 73.1% 9.7%
110 24.1% 7.9%
457 100%
Aggravated Assaults
33 3.5% 18.2%
689 72.2% 20.1%
232 24.3% 16.7%
954 100%
Burglary
2 2.6% 1.1%
53 67.9% 1.5%
23 29.5% 1.7%
78 100%
Theft
40 4.0% 22.1%
704 71.1% 20.5%
246 24.8% 17.7%
990 100%
Arson
13 2.6% 7.2%
324 64.0% 9.5%
169 33.4% 12.1%
506 100%
All Other Offenses
37 3.1% 20.4%
750 62.2% 21.9%
417 34.6% 29.9%
1204 100%
Total
181/(100%)
3428/(100%)
1391/(100%)
5000/(100%)
and endomorphs was very minute (3% of the sample). The endomorph dummy variable was statistically significant in the model, albeit with very minor effect sizes (Odds = 0.868). Of more importance is the direction of the somatotypic dummy variables’ effects. According to the dummy variables, endomorphs were more likely to be incarcerated for non-violent offenses than mesomorphs. This was the same for ectomorphs as well even though this dummy variable failed to gain statistical significance. Mesomorphs were thus more likely to be incarTable 4 Logistic regression predicting violent criminality Variable Constant Endomorph dummy (endomorph = 1) Ectomorph dummy (ectomorph = 1) Age Education (HS degree = 1) Race (non-White = 1) −2 log likelihood Chi-square Nagelkerke R2 ∗
p < .05; two-tailed.
b
S.E. 1.440 −0.141 −0.029 −0.043 −0.106 −0.261
6424.92 272.68* .073
0.103 0.068 0.159 0.003 0.062 0.060
Wald
Odds *
197.20 4.276* 0.03 188.90* 2.88 19.56*
4.22 0.868 0.972 0.96 0.90 0.77
342
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
cerated for a violent offense than individuals with endomorphic and ectomorphic physiques. On the whole, the results of the logistic regression confirm our, and indeed Hooten’s, hypothesis that criminals with mesomorphic builds are more prone to violent offending. Conclusions based on these analyses will be discussed in the next section.
5. Discussion Raine (1993) asserted that “no factor linked to crime should be viewed in the anachronistic terms of genetics versus environment” (p. 204). To do this is both “divisive in disciplinary terms and overly simplistic” (p. 204). Raine (1993) further wrote that criminologists should move toward an integration of genetic and social factors for crime in an effort to examine how these two factors relate to one another. Although past research (until the mid-20th century) on the biological correlates of crime have been marred by poor methodological designs and a general un-acceptance by mainstream criminologists, and in accordance with new developments in biosocial and life course studies on the link between biology and crime, this research has attempted to contemporize a biological concept that has fallen to the wayside of criminological history: somatotyping. This research has advanced the study on somatotypes in two primary ways. First, due to the inherent difficulties with Sheldon’s (Sheldon et al., 1940; Sheldon et al., 1949) and Parnell’s (1958) methods of somatotyping, this study introduced a new method of somatotyping that was more methodologically and ethically sound. This new somatotype measure, based on the Body Mass Index (BMI), is highly correlated with the original method of somatotyping as presented by Sheldon, giving the new somatotype measure a high level of reliability. Thus, this new somatotypic measure can easily be used by criminologists in future studies of the biological correlates of crime. This is especially true when the data set analyzed has a greater availability of independent variables and the existence of a comparison group. Second, the current study examined the effects of particular somatotypes on violent offending patterns. The most important finding of this research was that somatotypes still accounted for some of the variation in violent and non-violent offending in a sample of male prisoners (N = 5000) taken in Arkansas from 1975 to 2000. While the somatotype showed effects, the impact was minor across the somatotype measures. The logistic regression analysis illustrated that the somatotypes had a statistically significant effect and added to the overall model. More importantly, the direction of the effects was in line with previous research on physique and criminality. Mesomorphic offenders were more likely to be in prison for a violent offense than their endomorphic counterparts; the direction was the same for ectomorphs, but this variable failed to gain statistical significance. While this research did indicate a minor effect of physique on criminal offending, on the whole, this research can also be interpreted as support for those criminologists who would discard biological determinants of criminality, specifically in relation to physique. Many criminologists simply dismiss any effect of physique, and other biological determinants of criminality, outright no matter what effects sizes are documented. As such, it will be incumbent on future criminologists and researchers to discern the efficacy of the use of the BMI as a measure of somatotypes in research projects.
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
343
This study has several strengths in terms of examining the links between somatotyping and criminality. Because the new method of somatotyping presented in this paper is based on the BMI scale, measuring a person’s somatotype is a more methodologically/scientifically sound process. It cannot be based on a researcher’s subjectivity, but on numbers provided in a BMI scale conversion chart or by the BMI calculation formula. This gives the study of somatotyping greater reliability. Second, and based on the first strength, each subject can only be one kind of somatotype. There is no mixture of somatotypes when using the BMI measure, thus reducing the researcher’s subjectivity when a case is not an extreme endomorph, mesomorph, or ectomorph. Third, by using the BMI, all races and both sexes can be easily measured, whereas, in the past, the majority of the subjects were White males. Somatotyping utilizing the BMI can be used to measure anyone except children and teenage adolescents who are still in the physical maturation process. Future research on the causes of criminality should consider utilizing the BMI when information about the offender’s height and weight are included in criminal justice databases.
References Anding, J. D., Suminski, R. R., & Boss, L. (2001). Dietary intake, Body Mass Index, exercise, and alcohol: Are college women following the dietary guidelines for Americans? Journal of American College Health, 49(4), 167–171. Cortes, J. B., & Gatti, F. M. (1966). Physique and motivation. Journal of consulting psychology, 30(5), 408–414. Eysenck, H. J., & Gudjonsson, G. H. (1989). The causes and cures of criminality. New York: Plenum Press. Feitosa, M. F., Borecki, I. B., Rich, S. S., Arnett, D. K., Sholinsky, P., Myers, R. H., Leppert, M., & Province, M. A. (2002). Quantitative-trait loci influencing body-mass index reside on chromosomes 7 and 13: The National Heart, Lung, and Blood Institute family heart study. American Journal of Human Genetics, 70(1), 72–82. Glueck, S., & Glueck, E. (1951). Unraveling juvenile delinquency. Cambridge: Harvard University Press. Glueck, S., & Glueck, E. (1956). Physique and delinquency. New York: Harper and Brothers Publishers. Hartl, E. M., Monnelly, E. P., & Elderkin, R. D. (1982). Physique and delinquent behavior: A thirty-year follow-up of William H. Sheldon’s varieties of delinquent youth. New York: Academic Press. Hooten, E. A. (1968). Crime and the man. New York: Greenwood Press, Publishers. Hooten, E. A. (1969). The American criminal: An anthropological study. Greenwood Press, Publishers. James, P. W., Astrup, A., Finer, N., Hilsted, J., Kopelman, P., Rossner, S., Saris, W., & Van Gaal, L. F. (2000). Effect of Sibutramine on weight maintenance after weight loss: A randomized trial. Lancet, 356(9248), 2119–2126. Kretschmer, E. (1925/1970). Physique and character: An investigation of the nature of constitution and of the theory of temperment. New York: Cooper Square Publishers, Inc. McCandless, B. R., Persons, W. S., III, & Roberts, A. (1972). Perceived opportunity, delinquency, race, and body build among delinquent youth. Journal of Consulting and Clinical Psychology, 38(2), 281–287. National Heart, Lung, and Blood Institute. (1998). Clinical guidelines on the identification, evaluation, and treatment of overweight, and obesity in adults: The evidence report. Washington D.C.: National Institute of Health. Parnell, R. W. (1958). Behavior and physique: An introduction to practical and applied somatometry. London: Edward Arnold Publishers, Inc. Parsons, T. J., Power, C., & Manor, O. (2001). Fetal and early life growth and Body Mass Index from birth to early adulthood in a 1958 British cohort: A longitudinal study. British Medical Journal, 323(7325), 1331–1335. Raine, A. (1993). The psychopathology of crime: Criminal behavior as a clinical disorder. New York: Academic Press. Sampson, R. J., & Laub, J. H. (1990). Crime in the making: Pathways and turning points through life. Boston: Harvard University Press.
344
S. Maddan et al. / The Social Science Journal 45 (2008) 330–344
Sheldon, W., Stevens, S. S., & Tucker, W. B. (1940). The varieties of human physique: An introduction to constitutional psychology. New York: Harper and Brothers Publishers. Sheldon, W., Hartl, W. M., & McDermott, E. (1949). Varieties of delinquent youth: An introduction to constitutional psychiatry. New York: Harper and Brothers Publishers. Walsh, A. (2002). Biosocial criminology: Introduction and integration. Cincinnati: Anderson Publishing Co. Wilson, J. Q., & Herrnstein, R. J. (1985). Crime and human nature. New York: Simon and Schuster, Inc. Wolfgang, M. E., Figlio, R. M., Tracy, P. E., & Singer, S. I. (1985). The national survey of crime severity. Washington D.C.: National Institute of Justice.