Sociology Mind
2012. Vol.2, No.2, 148-152
Published Online April 2012 in SciRes (http://www.SciRP.org/journal/sm) http://dx.doi.org/10.4236/sm.2012.22019
Copyright © 2012 SciRes.
148
Young Child Homicide and Accidental Death Rates in the United
States, 1940-2005: Classification Issues in Mutually
Exclusive Events
Jack E. Riggs1, Gerald R. Hobbs2
1Department of Neurology, West Virginia University, Morgantown, USA
2Department of Stati s tics, West Virginia University, Morgantown, USA
Email: jriggs@wvu.e du
Received December 9th, 2011; revised January 10th, 2012; accepte d F e b ru ary 11th, 2012
Events that are mutually exclusive, collectively exhaustive, and have a constant sum must be negatively
correlated. This study examined whether non-motor vehicle accidental deaths and homicides in young US
children displayed a period of time when this rule governing mutually exclusive events was applicable.
Homicide and non-motor vehicle accident mortality rates in boys and girls, aged 1 to 4 years old, in the
United States from 1940 to 2005 were analyzed. Homicide mortality rates increased sharply in young
boys when the accident mortality rate dropped to about 25/100,000; and in young girls, when the accident
mortality dropped to about 18/100,000. This increase in child homicide mortality rates corresponded to a
time period when the sum of homicide rates and non-motor vehicle accident rates in these children were
relatively constant, making these rates of unnatural deaths mutually exclusive, collectively exhaustive,
and having a nearly constant sum. Homicide rates in young US children were relatively stable both before
and after this critical constant sum time period. These findings suggest that the increase in homicide rates
in young US children appear to have reflected the necessary negative correlation between mutually exclu-
sive, collectively exhaustive, and constant sum events, rather than an actual increase in societal violence
directed against young children.
Keywords: Accidental Death Rates; Child; Classification; Homicide Rates; Mutually Exclusive Events
Introduction
Unnatural deaths may be classified into four categories;
homicide, suicide, motor vehicle accident, and non-motor vehi-
cle accident. These four categories are mutually exclusive. That
is, the cause of each unnatural death can be attributed to only
one of these four categories. Young children, aged 1 to 4 years
old, do not commit suicide. Death due to a motor vehicle acci-
dent is generally obvious and for classification purposes, rela-
tively judgment-independent. Accordingly, for the purpose of
this study, young child motor vehicle accident deaths were
excluded. In this study, non-motor vehicle accident deaths are
referred to as accidental deaths. Therefore, this study examined
the rates of two mutually exclusive events; young child homi-
cides and young child accidental deaths. Each non-motor vehi-
cle accident unnatural young child death must be classified as
either a homicide or as an accident; it cannot be both and some
mechanism of classifying that death must be exercised in each
such young child death.
Events that are mutually exclusive, collectively exhaustive,
and have a constant sum must be negatively correlated. More-
over, the reliability of nominal data (such as whether the un-
natural death of a young child was an accident or a homicide)
that is based upon judgment can be questioned (Cohen, 1960;
Perreault & Leigh, 1989). Indeed, a prerequisite for measuring
interobserver agreement on such judgments is that the nominal
data categories must be mutually exclusive and exhaustive
(Cohen, 1960). The underappreciated rules governing the clas-
sification of mutually exclusive events was studied in
the classification of unnatural infant deaths as either accidental
deaths or homicides (Riggs & Hobbs, 2011). The consequences
of classification in those mutually exclusive events have impli-
cations for society, the victims, and in the case of homicide, the
accused perpetrators. In US infants, homicide rates increased
sharply only during the time period when the sum of infant
homicide rates and infant non-motor vehicle accidental death
rates were relatively constant (Riggs & Hobbs, 2011). During
that period, when these two events that are mutually exclusive,
collectively exhaustive, and had a relatively constant sum; an
increase in the rate of one of those two mutually exclusive
events was associated with a corresponding decrease in the rate
of the other event.
This purpose of this study was to examine whether non-mot or
vehicle accidental deaths and homicides in young US children
(aged 1 to 4 years) displayed a period of time when this rule
governing mutually exclusive, exhaustive, and constant sum
events was also applicable.
Data
This study utilized publicly accessible data provided through
the National Center for Health Statistics website
(www.cdc.gov/nchs). Young child (defined as a child aged
from 1 year old to less than 5 years old) homicide and accident
mortality rates (per 100,000) for both boys and girls in the
United States for the years 1940 through 2005 are shown in
Tables 1 and 2.
J. E. RIGGS ET AL.
Results
Young child homicide rates in the United States increased in
young boys from about 0.5 - 1.0/100,000 prior to 1960 to about
2.5 - 3.0/100,000 after 1980 (Figure 1) and in young girls from
about 0.5 - 1.0/100,000 prior to 1960 and to about 2.0 - 3.0/100,000
after 1980 (Figure 2). Homicide rates in both young boys and
girls incr eased s harp ly be tween 1960 and 1980 ( Figures 1 and 2 ).
To demonstrate the relative frequency of young child acci-
dent mortality and homicide rates, the annual ratio of young
child accident mortality rate to young child homicide rate in
both boys and girls was determined (Tables 1 and 2) and is
displayed in Figures 3 and 4. This ratio declined over 95 percent in
both young boys and girls between 1940 and 2005. Accident
mortality rates in young US children declined by over 85 per-
cent in boys and over 90 percent in girls between 1940 and
2005 (Tables 1 and 2).
Table 1.
Homicide (H) and accident (A) non-motor vehicle mortality rates (per
100,000) among young boys (aged 1 year to less than 5 years old) and
annual ratio of A/H in the United States from 19 40 through 2005.
Year H A A/H Year H A A/H
1940 0.6 42.1 70.2 1973 2.7 23.2 8.6
1941 0.7 41.0 58.6 1974 2.4 23.7 9.9
1942 0.6 42.7 71.2 1975 2.8 20.6 7.4
1943 0.5 48.8 97.6 1976 2.4 20.9 8.7
1944 0.5 45.1 90.2 1977 2.9 20.4 7.0
1945 0.7 41.2 58.9 1978 2.7 21.5 8.0
1946 0.7 38.8 55.4 1979 2.4 20.5 8.5
1947 0.7 36.2 51.7 1980 2.7 20.2 7.5
1948 0.6 33.7 56.2 1981 2.7 19.8 7.3
1949 0.8 30.7 38.4 1982 3.0 17.9 6.0
1950 0.5 29.5 59.0 1983 2.5 17.3 6.9
1951 0.6 28.0 46.7 1984 2.4 16.0 6.7
1952 0.8 28.3 35.4 1985 2.5 16.0 6.4
1953 0.7 27.2 38.9 1986 3.1 16.1 5.2
1954 0.6 25.5 42.5 1987 2.2 16.9 7.7
1955 0.5 25.0 50.0 1988 2.9 16.0 5.5
1956 0.7 24.9 35.6 1989 2.9 15.0 5.2
1957 0.5 23.6 47.2 1990 2.7 13.9 5.1
1958 0.7 24.3 34.7 1991 3.0 13.8 4.6
1959 0.9 24.3 27.0 1992 3.0 12.9 4.3
1960 0.7 24.3 34.7 1993 3.4 13.1 3.9
1961 1.0 23.9 23.9 1994 3.3 12.2 3.7
1962 0.9 24.0 26.7 1995 3.1 11.2 3.6
1963 1.2 24.7 20.6 1996 2.7 10.5 3.9
1964 1.3 24.6 18.9 1997 2.7 9.8 3.6
1965 1.2 26.4 22.0 1998 2.9 9.5 3.3
1966 1.2 26.7 22.3 1999 2.5 9.7 3.9
1967 1.1 26.2 23.8 2000 2.5 9.3 3.7
1968 1.5 24.6 16.4 2001 3.0 8.8 2.9
1969 1.8 23.9 13.3 2002 2.9 8.4 2.9
1970 1.9 23.9 12.6 2003 2.5 8.6 3.4
1971 2.3 24.3 10.6 2004 2.5 7.6 3.0
1972 2.0 25.2 12.6 2005 2.6 8.2 3.2
Table 2.
Homicide (H) and accident (A) non-motor vehicle mortality rates (per
100,000) among young girls (aged 1 year to less than 5 years old) and
annual ratio of A/H in the United States from 19 40 through 2005.
YearH A A/H Year H A A/H
19400.5 32.2 64.4 1973 2.3 15.6 6.8
19410.4 30.9 77.3 1974 2.0 14.2 7.1
19420.6 34.0 56.7 1975 2.0 14.6 7.3
19430.8 35.7 44.6 1976 2.5 13.3 5.3
19440.6 33.6 56.0 1977 2.4 13.2 5.5
19450.7 31.0 44.3 1978 2.3 14.1 6.1
19460.8 27.7 34.6 1979 2.6 12.7 4.9
19470.5 24.5 49.0 1980 2.2 12.9 5.9
19480.5 24.6 49.2 1981 2.4 11.8 4.9
19490.4 21.9 54.8 1982 2.5 11.3 4.5
19500.7 20.9 29.9 1983 2.0 11.4 5.7
19510.6 21.6 36.0 1984 2.4 9.8 4.1
19520.5 22.2 44.4 1985 2.4 9.9 4.1
19530.7 20.6 29.4 1986 2.3 10.9 4.7
19540.6 19.5 32.5 1987 2.4 9.9 4.1
19550.6 19.0 31.7 1988 2.3 9.5 4.1
19560.7 18.0 25.7 1989 2.5 9.0 3.6
19570.6 19.3 32.2 1990 2.4 8.1 3.4
19580.8 18.5 23.1 1991 2.6 9.3 3.6
19590.8 17.7 22.1 1992 2.5 7.7 3.1
19600.7 18.7 26.7 1993 2.5 8.3 3.3
19611.0 18.3 18.3 1994 2.7 7.5 2.8
19620.9 18.4 20.4 1995 2.6 7.2 2.8
19631.1 18.1 16.5 1996 2.7 6.5 2.4
19641.2 18.2 15.2 1997 2.2 6.2 2.8
19651.1 17.5 15.9 1998 2.4 5.9 2.5
19661.2 18.9 15.8 1999 2.4 6.5 2.7
19671.2 17.6 14.7 2000 2.1 5.9 2.8
19681.5 17.5 11.7 2001 2.4 5.3 2.2
19691.6 16.7 10.4 2002 2.5 4.8 1.9
19701.9 15.9 8.4 2003 2.3 5.2 2.3
19712.0 15.4 7.7 2004 2.2 5.0 2.3
19721.6 14.6 9.1 2005 2.0 4.6 2.3
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
1920 1940 1960 1980 2000 2020
Year
Homicide Rate
Figure 1.
Homicide rate (per 100,000) in US boys, aged 1 - 4 Years, from 1940
through 2005.
Copyright © 2012 SciRes. 149
J. E. RIGGS ET AL.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
1920 1940 1960 1980 20002020
Year
Homicide Rate
Figure 2.
Homicide rate (per 100,000) in US girls, aged 1 - 4 years, from 1940
through 2005.
0
20
40
60
80
100
120
1920 1940 1960 1980 2000 2020
Year
Accident Mortality Rate Divided By
Homicide Rate
Figure 3.
Ratio of accident mortality rate to homicide rate (per 100,000) in US
boys, aged 1 - 4 years, from 1940 through 2005.
0
20
40
60
80
100
1920 1940 1960 1980 2000202
0
Year
Accident Mortality Rate Divided By
Homicid e R a te
Figure 4.
Ratio of accident mortality rate to homicide rate (per 100,000) in US
girls, aged 1 - 4 years, fr om 1940 through 2005.
The sum of young child homicide rates and young child ac-
cident mortality rates in the US was relatively constant in
young boys from approximately 1960 to 1980 (Figure 5) and in
young girls from about 1955 to 1970 (Figure 6).
Plotting young child homicide rates versus young child acci-
dent mortality rates shows that young child homicide rates in-
creased sharply as accident mortality rates decreased below
certai n levels (Figures 7 and 8). In young boys, when the acci-
dent mortality rate declined below 25/100,000, the homicide
rate sharply increased and then became relatively stable as the
accident mortality rate continued to decrease below 20/100,000
(Figure 7). In young girls, when the accident mortality rate
declined below 18/100,000, the homicide rate sharply increased
and became relatively stable as the accident mortality rate con-
tinued to decrease below 12/100,000 (Figure 8). These periods
of sharp increases in young child homicide rates correspond to
the time periods when the sum of young child homicide rates
and accident mortality rates among US boys and girls were
relatively consta nt .
0
10
20
30
40
50
60
1920 1940 1960 1980 2000 2020
Year
Accident Mortality +
Homicide Rate
Figure 5.
Sum of homicide and accid ent mortality rates (per 100,000) i n US boys,
aged 1 - 4 years, from 194 0 th rough 2005.
0
5
10
15
20
25
30
35
40
1930 1940 1950 1960 1970 1980 1990 2000 2010
Year
Accident Mortality +
Homicide Rate
Figure 6.
Sum of homicide and accident mortality rates (per 100,000) in US girls,
aged 1 - 4 years, from 194 0 th rough 2005.
0.0
1.0
2.0
3.0
4.0
0.010.0 20.0 30.0 40.0 50.0
Accident Mortality Rate
Homicid e R a t e
Figure 7.
Homicide versus accident mortality rate (per 100,000) in US boys, aged
1 - 4 years, from 1940 thr ough 2005.
Copyright © 2012 SciRes.
150
J. E. RIGGS ET AL.
0.0
1.0
2.0
3.0
0.05.010.0 15.0 20.0 25.0 30.0 35.0 40.0
Accident Mortality Rate
Homicide Rate
Figure 8.
Homicide versus accident mortality rate (per 100,000) in US girls, aged
1 - 4 years, from 1940 thr ough 2005.
Discussion
Young child homicide is frequently related to child abuse
(Jenny & Isaac, 2006) or psychiatric dysfunction in a custodial
adult (Friedman et al., 2005; Nielssen et al., 2009). Head inju-
ries, drowning, and suffocation are the most frequent causes of
death due to homicide in young children (Reece & Sege, 2000;
Tung et al., 2006; Vanamo et al., 2001). Investigators have sug-
gested that the actual magnitude of fatal child abuse has been
underestimated (Crume et al., 2002; Herman-Giddens et al.,
1999). Beginning in the 1960’s to the 1980’s, the problem of
young child abuse and homicide became the focus of increased
societal attention (Adelson, 1961; Christoffel, 1984; Kempe et
al., 1962; Jason, 1983).
Similar to infants (Riggs & Hobbs, 2011), young child acci-
dent mortality rates in both boys and girls decreased over 85
percent in the United States between 1940 and 2005 (Tables 1
and 2, and Figures 3 and 4), reflecting improved societal injury
prevention and trauma management. Also similar to infants
(Riggs & Hobbs, 2011), young child homicide and accident
mortality rates were slightly higher in boys than in girls (Ta-
bles 1 and 2).
When classifying two mutually exclusive events, their rela-
tive natural frequency may be important. For example, if event
A and event B are potentially difficult to distinguish and event
A is much more frequent, there may be a tendency to bias clas-
sification towards event A. However, if event A becomes less
frequent and the sensitivity to recognizing event B is increasing,
there may be a tendency to bias classification towards event B.
This line of reasoning was suggested as contributing to the
observed increase in infant homicide rates in the United States
after 1980 (Riggs & Hobbs, 2011). In the case of young child
homicide, homicide rates in the US increased predominantly
during the period of time when the sum of young child homi-
cide and accident mortality remained relatively constant (Fig-
ures 1, 2, 5-8). This same phenomenon was also observed in
US infants. Infant homicide rates in the US increased during the
period of time when the sum of infant homicide and accident
mortality rates were relatively constant (Riggs & Hobbs, 2011).
Moreover, if the ratio of accident mortality to homicide mortal-
ity in young children is consistently declining (as shown in
Tables 1 and 2, and illustrated in Figures 3 and 4), during the
periods when the sum of accident mortality and homicide rates
are relatively constant and necessarily n eg a ti ve ly c or re l at ed , th e n
homicide rates that must go up. If accident mortality rates
where to go up during those periods, then the ratio of accident
mortality rates to homicide rates would have also had to in-
crease; and that did not occur.
These findings suggest that the logic rules governing two
mutually exclusive, collectively exhaustive, and relatively con-
stant sum events are consistent with and describe the observed
increased reported young child homicide rates in US boys and
girls. These logic rules may also explain why the influence of
declining young child accident mortality and increased societal
sensitivity to child abuse were not associated with continued
increasing young child homicide rates despite the fact that both
of these influences continued to occur after the critical time
periods when the sum of the rates of young child homicide and
accident mortality rates were relatively constant. Moreover, it
seems rather unlikely that societal violence in the United States
would have been increased against young children before 1980
(as observed in this study) and would have separately and dis-
tinctly increased against infants after 1980 (Riggs & Hobbs,
2011).
Epidemiological data is crucial in identifying and monitoring
societal problems. Classification mechanisms are routinely used
in the production of epidemiological data, and those classifica-
tion mechanisms can have significant societal implications and
consequences. This study suggests that the logic rules govern-
ing mutually exclusive, collectively exhaustive, and constant
sum events described the classification process of distinguish-
ing young child accidental deaths from homicides, similar to
what was observed in US i nfants (Riggs & Hobbs, 2011 ). More-
over, these findings suggest that the increase in the homicide
rates observed among young US children may have reflected
the necessary negative correlation between mutually exclusive,
collectively exhaustive, and constant sum events rather than an
actual increase in societal violence directed against young chil-
dren, similar to the pattern observed after 1980 among US in-
fants. These observations illustrate the potential hazard of fo-
cusing attention on the measurement of one mutually exclusive
event (the homicide rate) without also considering the influence
of the other mutually exclusive event (the accident mortality
rate). This changing classification propensity, as demonstrated
in this study, appears to emerge spontaneously in the epidemi-
ological classification process of events that are mutually ex-
clusive, collectively exhaustive, and that have a relative con-
stant sum. Moreover, this changing classification propensity ap-
pears to occur despite the best intentions and expertise of all
those individuals who are independently involved in the classi-
fication process of each such unnatural death in young children.
Child abuse and homicide is a serious societal problem that
cannot be tolerated. However, the inaccurate classification of a
young child death as a homicide, rather than an accident, also
has unacceptable and intolerable consequences.
This study only examined unnatural mortality classified and
reported as either homicide or non-motor vehicle accident mor-
tality. This study did not examine natural mortality due to ill-
ness. Consequently, no inferences can, or should, be derived
from this study regarding misclassification between natural and
unnatural deaths.
REFERENCES
Adelson, L. (1961). Slaughter of the innocents, a study of forty-six
homicides in which the victims were children. New England Journal
Copyright © 2012 SciRes. 151
J. E. RIGGS ET AL.
Copyright © 2012 SciRes.
152
of Medicine, 264, 1345-1349. doi:10.1056/NEJM196106292642606
Christoffel, K. K. (1984). Homicide in childhood: A public health prob-
lem in need of attention. American Journal of Public Health, 74, 68-
70. doi:10.2105/AJPH.74.1.68
Cohen, J. (1960). A coefficient of agreement for nominal scales. Edu-
cational and Psychological Measurement, 20, 37-46.
doi:10.1177/001316446002000104
Crume, T. L., DiG u isep p i, C., Byers, T., Sirotn ak , A. P., & Garrett, C. J.
(2002). Underascertainment of child maltreatment fatalities by death
certificates, 1990-1998. Pediatrics, 110, e18.
doi:10.1542/peds.110.2.e18
Friedman, S. H., Horwitz, S. M., & Resnick, P. J. (2005). Child murder
by mothers: A critical analysis of the current state of knowledge and
a research agenda. American Journal of Psychiatry, 162, 1578-1587.
doi:10.1176/appi.ajp.162.9.1578
Herman-Giddens, M. C., Bro wn , G., Verb iest, S. , Carlson, P. J., Hooten ,
E. G., Howell, E., & Butts, J. D. (1999). Underascertainment of child
abuse mortality in the United States. Journal of the American Medi-
cal Association, 282, 463-467. doi:10.1001/jama.282.5.463
Jason, J. (1983). Child Homicide spectrum. American Journal of Dis-
eases of Children, 137, 578-5 81.
Jenny, C., & Isaac, R. ( 2006). The relation between child death and
child maltreatment. Archives of Disease in Childhood, 91, 265-269.
doi:10.1136/adc.2004.066696
Kempe, C. H., Silverman, F. N., Steele, B. F., Droegemueller, W., &
Silver, H. K. (1962). The battered-child syndrome. Journal of the
American Medical Association, 181, 17-24.
doi:10.1001/jama.1962.03050270019004
Nielssen, O. B., Large, M. M., Westmore, B. D., & La ckersteen, S. M.
(2009). Child homicide in New South Wales from 1991 to 2005.
Medical Journal of Australia, 190, 7-11.
Perreault, W. D., & Leigh, L. E. (1989). Reliability of nominal data
based on qualitative judgments. Journal of Marketing Research, 26,
135-148. doi:10.2307/3172601
Reece, R. M., & Sege, R. (2000). Childhood head injuries, accidental or
inflicted? Archives of Pediatric and Adolescent Medicine, 154, 11-15.
Riggs, J. E., & Hobbs, G. R. (2011). Infant homicide and accidental
death in the United States, 1940-2005: Ethics and epidemiological
classification. Journal of Medical Ethics, 37, 445-448.
doi:10.1136/jme.2010.041053
Tung, G. A., Kumar, M., Richardson, R. C., Jenny, C., & Brown, W. D.
(2006). Comparison of accidental and nonaccidental head injury in
children on noncontrast computed tomography. Pediatrics, 118, 626-
633. doi:10.1542/peds.2006-0130
Vanamo, T., Kauppi, A., Karkola, K., Merikanto, J., & Rasanen, E. (2001).
Intra-familial child homicide in Finland 1970-1994: Incidence, causes
of death and demographic characteristics. Forensic Science Interna-
tional, 117, 199-204. doi:10.1016/S0379-0738(00)00408-4