Young Men Who Kill

AuthorRolf Loeber,Mark T. Berg,David P. Farrington
Published date01 May 2012
Date01 May 2012
DOIhttp://doi.org/10.1177/1088767912439398
Subject MatterArticles
Homicide Studies
16(2) 99 –128
© 2012 SAGE Publications
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1088767912439398
http://hsx.sagepub.com
439398HSX16210.1177/1088767912
439398Farrington et al.Homicide Studies
1Cambridge University, Cambridge, UK
2University of Pittsburgh, Pittsburgh, PA, USA
3Indiana University, Bloomington, IN, USA
Corresponding Author:
Mark T. Berg, Department of Criminal Justice, Indiana University, 311 Sycamore Hall, Bloomington, IN
47405, USA
Email: markberg@indiana.edu
Young Men Who Kill:
A Prospective Longitudinal
Examination From
Childhood
David P. Farrington1, Rolf Loeber2,
and Mark T. Berg3
Abstract
Prior research has revealed important insights about some factors which increase
the probability that individuals will commit murder; however, existing studies rely
on retrospective data from institutional samples and have not examined homicide
offending using data collected before the murder was committed. We use prospective
longitudinal data from the Pittsburgh Youth Study to examine homicide using factors
from multiple informants and developmental periods. Early risk scores showed whether
homicide offenders could be predicted at an early age. The study reveals early-life
factors that increase risk for perpetrating lethal violence and yields information on
the dose-response relationships between predictors and homicide.
Keywords
criminal careers, violence, developmental, prevention, youth/juvenile
Social scientists have long speculated whether it is possible to predict years in advance
who will commit lethal violence. Likewise, scholars and policymakers hypothesize
etiological factors which may differentiate those who murder from those who do not
(DeLisi, Hochstetler, Jones-Johnson, Caudill, & Marquart, 2011; Heide, 2003). Crimi-
nologists have amassed an impressive body of scientific evidence on the dimensions
Articles
100 Homicide Studies 16(2)
of homicide (see Rosenfeld, 2009a). Despite all of the attention paid to the develop-
mental predictors of violent conduct, a surprisingly small body of research has exam-
ined the predictors of homicide offending. Perhaps investigators have rarely broached
the subject because most offenders never murder. DeLisi (2001) reviewed some of the
integral data sets in the criminological literature, including the 1945 and 1958 Philadel-
phia birth cohorts, Sheldon and Eleanor Glueck’s 500 criminal careers cohort, and the
National Youth Survey. He documented that even these large sources of data, which at
times included entire birth cohorts, contained very few criminals who committed seri-
ous violent offenses including murder (DeLisi & Scherer, 2006). Prevailing crimino-
logical theories also implicitly assume that there is little need to develop a conceptual
explanation of homicide offending; yet this assumption lacks a strong empirical basis
(Heide, 1999, 2004).
Previous individual-level studies on youth homicide, although comparatively small
in number, have revealed important insights about the characteristics of homicide
offenders relative to the general population (Smith, 2000). Still, these studies rarely
investigate the predictive causal factors over the developmental life course. Moreover,
most studies on youth homicide contain various design limitations that curb their
potential contribution with regard to the causes of homicide. First, current studies are
restricted in scope, relying on small convenience samples of institutionalized offend-
ers while providing more information about prevalence than about causes (e.g., Bailey,
1996; Cornell, Benedek, & Benedek, 1987). Second, with some exceptions, prior stud-
ies have not used control groups and so it is unknown whether factors that define
young homicide offenders would apply to other groups such as violent youth or delin-
quents in general (e.g., Bender, 1959; Heide, 2003; van Soest, Park, Johnson, &
McPhail, 2003). Case-control designs lack adequate longitudinal information about
individuals who resembled homicide offenders at a younger age but who did not sub-
sequently commit murder. From these designs, it is not possible to specify the pro-
spective probability of becoming a homicide offender for those who do or do not
possess a given risk factor (Goodman, Mercy, Layde, & Thacker, 1988). Third, most
descriptive research has relied on retrospective information about putative causes of
homicide that is generally derived from official court records or third-party reports
(e.g., Busch, Zagar, Hughes, Arbit, & Bussell, 1990; Hagelstam & Häkkänen, 2006;
Lewis et al., 1988; Roberts, Zgoba, & Shahidullah, 2007; Shumaker &McGee, 2001;
Wolfgang, 1958). Several methodological problems exist with retrospective life-
course studies of homicide offenders (see Zagar, Busch, Grove, & Hughes, 2009).
Court records may provide incomplete or inaccurate archival information about pos-
sible risk factors. Moreover, retrospective data reported by “family members are often
biased by knowledge that the person has committed an offense, and may be character-
ized by incomplete or inaccurate recall of critical events and risk factors” (Heide,
2003, p. 25). Clinical data on murderers are likely to overstate the role of mental health
factors, while obscuring the contribution of behavioral and environmental predictors.
In short, the current body of research on homicide offenders is relatively sparse and
most existing studies, although informative, are descriptive rather than predictive and

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT