Violent Acts and Violent Times: A Comparative Approach To Postwar Homicide Rates

Violent Acts and Violent Times: A Comparative Approach to Postwar Homicide Rates
Author(s): Dane Archer and Rosemary Gartner

Source: American Sociological Review , Dec., 1976, Vol. 41, No. 6 (Dec., 1976), pp. 937-963
Published by: American Sociological Association
Stable URL: https://www.jstor.org/stable/2094796
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Sociological Association is collaborating with JSTOR to digitize, preserve and extend
access to American Sociological Review
This content downloaded from

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
All use subject to https://about.jstor.org/terms
VIOLENT ACTS AND VIOLENT TIMES:
A COMPARATIVE APPROACH TO POSTWAR HOMICIDE RATES *
DANE ARCHER AND ROSEMARY GARTNER

University of California, Santa Cruz
American Sociological Review 1976, Vol. 41 (December):937-963
The idea that waging war might increase the level of domestic violence in warring
societies has occurred to many researchers. Discussions of this possibility have been limited
to a very small number of case studies-often as limited as the experience of a single
nation in a single war. A major obstacle to the general investigation of this question
has been the unavailability of comparative data on homicide rates. Over a three-year
period, a Comparative Crime Data File was assembled. The file includes time-series
rates of homicide for roughly 110 nations beginning in about 1900. Postwar rather than
wartime homicide rates were analyzed, since postwar data appear much less problematic
and are likely to be affected by artifacts in only a conservative direction. The homicide
data were analyzed to: (1) determine if postwar increases did occur and (2) identify
which of seven competing theoretical models appeared to offer the most adequate expla-
nation. The homicide rate changes after 50 "nation-wars" were compared with the
changes experienced by 30 control nations. The major finding of the study was that most
of the nation-wars in the study did experience substantial postwar increases in their
rates of homicide. These increases were pervasive, and occurred after large wars and
smaller wars, with several types of homicide rate indicators, in victorious as well as
defeated nations, in nations with both improved and worsened postwar economies,
among both men and women offenders and among offenders of several age groups.
Homicide rate increases occurred with particular consistency among nations with large
numbers of combat deaths. Using homicide and other data, it was possible to disconfirin
or demonstrate the insufficiency of six of the seven explanatory models.
During the Vietnam War, the murder Although the sudden increase in violent
and nonnegligent manslaughter rate in the crimes during the Vietnam War suggests
United States more than doubled. The inci- some theoretically interesting possibilities,
dence of these homicides was 4.5 (per more information obviously is required
100,000 inhabitants) in 1963 and 9.3 in before any general relationship between
1973 (F.B.I., 1963-73, Uniform Crime war and criminal homicide can be inferred
Reports). This rapid increase was some- -the case of a single nation during a
what more remarkable because it followed single war can only be a point of departure.
a period of almost monotonic decline in This paper attempts to address two con-
the homicide rate since record-keeping ceptually separable questions: (1) the
began in 1933. empirical question of whether, in general,
homicide rates do increase after wars and,
* This project and the development of the 110- if such increases are found, (2) the inter-
nation Comparative Crime Data File were sup- pretive question of which of seven theo-
ported by NIMH Grants Number MH 25881
retical models appears to provide the most
and MH 27427 from the Center for Studies of
Crime and Delinquency, by the Socity for the adequate explanation of these increases.
Psychological Study of Social Issues and by the The idea that war might foster crime and
Faculty Research Fund of the University of violence has occurred to many. Although
California. The authors are indebted to several
the first appearance of the idea is difficult
people for helpful comments and advice, includ-
ing Marshall Clinard, John Kitsuse, Thomas Pet-
to trace with certainty, Erasmus, Sir
tigrew, Robert Rosenthal, Thorsten Sellin, Mur- Thomas More and Machiavelli each specu-
ray Straus, Marvin Wolfgang and the hundreds lated that wars left a postwar legacy of
of people around the world who helped us during increased crime and lawlessness (Abbott,
the collection of the Comparative Crime Data
1918; 1927; Hamon, 1918).
File. Responsibility for the findings and interpre-
tations in this paper belongs, of course, to the Perhaps not surprisingly, much of the
authors alone. popular and scholarly writing on this ques-
937

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
938 AMERICAN SOCIOLOGICAL REVIEW
tion has appeared during or immediately Most research published in this period,
following wars. Shortly after World War I however, was concerned with the impact
ended, for example, several public figures of the First World War itself. There were
and writers suggested that the war had in- studies of this question in Austria (Exner,
creased crime. Among these were Winston 1927), France (Calbairac, 1928), Italy
Churchill, Clarence Darrow and Abbott (Levi, 1929), Czechoslovakia (Solnar,
Lawrence Lowell, then President of Har- 1929), Germany (Leipman, 1930), the
vard University (Abbott, 1927:213; Dar- United States, (Engelbrecht, 1937; Suther-
row, 1922:218; Lowell, 1926:299). The land, 1943) and England (Mannheim,
rich history of the idea that wars affect the 1941; 1955). The most influential of these
level of violence in postwar societies is researchers appear to have been Exner
reviewed elsewhere (Archer and Gartner, (1927), who attributed crime increases to
forthcoming). economic problems after wars, and Mann-
The end of the First World War also saw heim, who emphasized non-economic fac-
the appearance of some of the first schol- tors like "the general cheapening of all
arly work on the question. For reasons values; loosening of family ties; [and]
discussed below, the comparability of dif- weakened respect for the law, human life,
ferent case studies is complicated by a and property" (Mannheim, 1955:112).
number of factors. A number of these In the most rigorous study of this per-
early studies dealt with nineteenth century iod, Sellin (1926) compared postwar hom-
(and, in one case, eighteenth century) icide changes in five belligerent nations in
wars. Tarde (1912), Bonger (1916) and World War I and four nonbelligerent na-
Roux (1917) discussed changes in various tions. Sellin concluded that the warring na-
crimes in France following the Revolution tions did experience increases, although
of 1848 and in both France and Germany differences between the warring and non-
after the 1870 Franco-Prussian War. belligerent nations were not uniform.
Based on a review of six nineteenth cen- World War II renewed interest in this
tury case studies, Sorokin (1928:340-4) question. A number of discussions dealt
concluded that whether a war increased with crime changes during wartime years,
crime depended upon the success of the including Sellin (1942), Reckless (1942),
war, its popularity, whether it was fought Glueck (1942), Bromberg (1943) and
at home or abroad and other factors. von Hentig (1947). Wartime changes in
Based on legislation and published ac- nine Allied and neutral nations were also
counts of the era, Nevins (1924) con- surveyed by an international group of
cluded that the American Revolution had criminologists (Commission Internationale
produced an upsurge of horse-stealing and Penale et Penitentiaire, 1951). Postwar
highway robbery in many states. Abbott changes were examined for five countries by
(1927) reviewed prison records from 11 Lunden (1963; 1967) who compared the
states and concluded that the Civil War raw number of crimes committed after the
had greatly increased the number of men war with the number committed during the
single year prior to the outbreak of the
sent to prison, and that many of these
war. Lunden reported finding increases
men were veterans. Abbott concluded that
after the war and he attributed these to
the imprisonment of these veterans con-
social changes attendant upon war, includ-
tributed to the formation of a prison re-
ing social disorganization, increased mobil-
form movement in 1867; she quotes from
ity and disruption of community life.
a contemporary author:
There have been very few recent studies
"A man who has lost one arm in the de- of crime changes during wars. As part of a
fense of the nation working with the other study of American political turmoil during
at the convict's bench is not an agreeable the Vietnam War, Tanter (1969) noted
spectacle, nor do we like to see the com-
the increase in crimes of violence during
rades of Grant and Sherman, of Foote and
Farragut, exchange the blue coat of victory the war years. The increase prompted
for the prison jacket." (Abbott, 1927:234) Tanter (1969:436) to suggest the possi-

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
VIOLENT ACTS AND VIOLENT TIMES 939
ability that "as the war continues, it facili- (1918) reported that this practice is at
tates a state of 'normlessness' in which least as old as the Civil War and Mann-
traditional strictures against criminal acts heim (1965) found that it has occurred
lose their effectiveness." in other nations as well; (3) the tendency
This body of research is extremely diffi- of persons arrested during major wars not
cult to evaluate critically. Part of this diffi- to be convicted or imprisoned-Abbott
culty stems from differences of focus. (1918), Mannheim (1941) and von Hen-
Three of these differences, which have not tig (1947) reported evidence of a reduced
always been made explicit by individual re- willingness to prosecute offenders and a
searchers, have been between (1) re- tendency for wartime employers to pay the
searchers who have been concerned with fines of their workers rather than lose
crime rate changes during wars and re- them to imprisonment; (4) the fact that
searchers concerned with postwar changes, parolees too old for the armed services,
(2) researchers who have studied homicide unlike their peacetime counterparts, have
and researchers who have studied theft, easily found employment due to labor
violent crime or crime in general, undiffer- shortages created by wars-von Hentig
entiated by type and (3) researchers who (1947) found that many more jobs were
have been concerned with the direct effects found for New York State parolees during
of wars upon the behavior of returned vet- World War II than before it; (5) the
erans and researchers interested in the pos- wartime shortages of commodities often in-
sibility of more general effects upon the volved in crimes, particularly alcohol
civilian populations of combatant nations. (Abbott, 1918; Exner, 1927; Mannheim,
These differences of focus have had 1941); (6) changes in national systems of
important implications, since individual re- law enforcement, including a shortage of
searchers have tended to generate hypoth- policemen (Mannheim, 1965). The im-
eses appropriate to only one possible effect pact of all of these factors seems likely
of war-e.g., postwar homicides commit- to vary directly with the level of wartime
ted by returned veterans. These differences mobilization-crime rates will be more de-
among researchers also have made a com- pressed by these factors in large wars than
parison of their findings a formidable task, in small wars.
since they have frequently been discussing During some wars, there also may be
quite different phenomena. social changes that increase crime rates,
For several reasons, some aspects of including (1) the disruption of families by
the study of war and crime are more prob- conscription, long employment hours and,
lematic than others. For example, the inter- sometimes, evacuation (Mannheim, 1941);
pretation of crime rates during major wars (2) the existence of special crime oppor-
is severely complicated by the large number tunities like blackouts, bombed-out houses
of simultaneous social changes. During and other unguarded property (Mannheim,
major wars, crime rates are likely to be 1941:131); (3) the creation of new types
depressed by many of these social changes of crimes by special wartime regulations
including: (1) the massive removal of like rationing (Sutherland and Cressey,
young men from the civilian population 1960:208).
through conscription and enlistment-for Finally, other factors that have limited
example, von Hentig (1947) reported that the usefulness of wartime offense data
New York City shrank by three-quarters are (1) the confusion of crimes and acts
of a million people, mostly young males, of resistance in occupied nations (Mann-
during World War II, and Bennett (1953) heim, 1965); (2) a persistent belief that
estimated that when World War II broke "crime-prone" persons tend to enlist-as
out 82% of American men between 20 Tarde (1912) proposed; (3) changes in
and 25 enlisted or were drafted; (2) the the boundaries of warring nations (Suther-
premature release of convicts on the con- land, 1943); (4) the fact that record-
dition that they enter the armed forces keeping often has been interrupted during
rather than return to civilian life-Abbott wartime.

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
Because of all these simultaneous social (von Hentig, 1947:349). Since young men
and economic changes during a major war, are universally overrepresented in many
many researchers have concluded that the offense rates, particularly homicide (Wolf-
immediate effects of wars on offense rates gang and Ferracuti, 1967), massive losses
are extremely difficult to discern (von Hen- of young men remove from the population
tig, 1947; Sutherland and Cressey, 1960; precisely those who would be most likely
Mannheim, 1965). Some researchers (e.g., to commit these offenses in the postwar
Willbach, 1948) have completely aban- years.
doned hope of making meaningful com- A second factor, which would also oper-
parisons between wartime years and peace- ate in a conservative direction, involves the
time years. Despite the inherent and per- possibility that veterans arrested for or
haps insurmountable problems with the in- convicted of certain crimes would be
terpretation of crime rates during war years, treated with leniency because of their mil-
the pattern generally has been one of a itary service. Such a practice has existed
sharp drop during the initial year or years for some time-for example, Abbott
of a war, followed by a gradual return (1918) said that after the Civil War
late in the war to prewar levels (Mann- judges often pardoned first offenders who
heim 1965; Lunden, 1967). were veterans. Leniency toward arrested
Although there are also a number of veterans also has been cited by researchers
complexities in the interpretation of post- of the First World War, even in cases in-
war rates, this topic is more promising volving very serious offenses (Mannheim,
methodologically than the analysis of rates 1941: 119).
during wars. In general, this is because Leniency of this kind would affect some
many of the special characteristics of war- offense indicators more than others, and
time society diminish or disappear with the the indicators farthest removed from the
end of the war. The beginning of peace offense itself would be most affected. Data
generally (1) returns large numbers of on convictions and prison population,
young men to civilian society and reunites therefore, would be most influenced. Data
families; (2) eliminates tremendous war- on the number of offenses known to the
time labor needs; (3) restores law enforce- police, however, would be unaffected by
ment agencies to normal manpower levels; judicial leniency. This difference between
(4) ends war-related commodity shortages. types of crime indicators was emphasized
There may be some social and demo- in a discussion of wartime crime by Suther-
graphic changes, however, that could con- land (1956), who concluded that convic-
tinue to affect offense levels in postwar tion rates were potentially misleading (i.e.,
periods. Virtually all of these postwar too low) because of factors like judicial
factors would affect rates in a conservative leniency. At any rate, judicial leniency
direction-i.e., reduce levels of various (like the loss of young men) could only
offenses in the postwar years compared to operate in a conservative direction, so that
the prewar period. the observed rates of some offenses would
One of the most important of these post- understate the actual incidence of these
war factors, particularly after major wars, crimes.
involves changes in the age and sex struc- Although the data needed for studies of
ture of a nation's population due to the postwar changes are therefore relatively
number of young men killed during the interpretable, available studies on this ques-
war. These losses sometimes have been of tion have left important questions un-
enormous magnitude. For example, at the answered. For example, a critical issue
end of World War I, out of every 1000 which has not been adequately addressed is
men who had been between 20 and 45 whether only warring nations experience
at the outbreak of the war, 182 had died postwar increases or whether all nations
in France, 166 in Austria, 155 in Ger- (combatant and noncombatant) experience
many, 101 in Italy, and 88 in Britain comparable increases. With apparently a

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
single exception (the study by Sellin national war participation led him to con-
(1926) cited earlier), case studies of post- clude that, "For these reasons, the effect
war crime rate trends have not analyzed of wars on crimes is not a good theoretical
a "control" group of nations uninvolved in problem" (Sutherland, 1956:120).
war. Without such a controlled comparison In summary, studies of wartime or post-
based on a large sample of nations, it is war changes in crime rates have tended
impossible to separate the direct effects of to be inconclusive for a number of different
war from international crime rate trends reasons. Studies of crime changes in war-
which happen to include the warring time probably cannot go beyond simple de-
nations. scription, because of the large number of
One reason a strategy of controlled com- interdependent social changes which have
parison has not been attempted is simply occurred during most wars. Studies of
the unavailability of historical and com- crime changes in postwar years, while more
parative crime data. Time-series crime data promising, have been weakened in the
have not been accumulated in any cen- past by: (1) extremely small sample sizes,
tralized manner, and individual researchers often as small as a single nation during a
often have been limited to one or at most single war; (2) the absence of a control
a handful of nations. The problem has group of nations uninvolved in the war
been aggravated by variation in the types of studied; (3) different types of crime rate
indicators kept by individual nations indicators that may be differentially suscep-
(Wolfgang, 1967). It has been recognized tible to artifacts and therefore incompar-
for some time (Sellin, 1931) that arrest able; (4) the failure to operationalize
and conviction data are more subject to characteristics of different types of national
judicial and police discretion than data on participation in wars.
the number of known offenses. Since these
types of discretion could be important fac- THEORETICAL MODELS
tors in postwar societies (particularly in the
case of veterans suspected of some offense), A number of theoretical models have
the net effect of this variation in record- been advanced to explain crime rate
keeping is to reduce the comparability of changes during war years, during postwar
data kept by certain nations. Variation in years or both. Some indication of the va-
indicators, therefore, increases the difficulty riety of competing explanations is con-
of obtaining a sufficiently large number of tained in a comment by Sutherland (1956:-
combatant and control nations. 120-1):
A second limitation of studies of post- One theory states that war produces an in-
war changes has been their inattention to crease in crimes because of the emotional
differences in national war experiences, instability of wartime, and another states
that wars produce a decrease in crimes be-
since these differences are likely to produce
cause of an upsurge of national feeling.
dissimilar effects on crime rates-as sug- One states that crimes of violence increase
gested by Sorokin (1928), Mannheim in wartime because of the contagion of vio-
(1941) and Sutherland (1956). In part, lence, and another that they decrease be-
this omission has been an inevitable con- cause of the vicarious satisfaction of the
need for violence.
sequence of the extremely small samples
available. However, Sutherland (1956) We have identified seven theoretical
concluded that an examination of the ef- models which attempt to explain the effects
fects of different types of war participation of war on wartime and postwar rates of
was impossible -at the time he wrote, since various offenses, and we have tried to de-
variations in the experiences of individual rive expectations or predictions from each
nations in individual wars had not been of the seven models. There is some agree-
quantified. Since Sutherland believed that ment between the predictions implicit in
these differences could be important, the two or more of the theoretical explanations.
absence of empirical bases for comparing However, the seven models are sufficiently

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
distinct theoretically to justify a separate social disorganization model predicts that

discussion of each. postwar crime increases would occur as a
function of rapid social changes which any
1. Social Solidarity Model warring nation could experience. As used
Several researchers have suggested that by other researchers, this model predicts
wars increase social solidarity and, as a that crime rate increases would be con-
result, reduce crime rates. For example, fined to defeated nations.
Sumner wrote in his classic Folkways
(1906:12) that wars increased discipline 3. Economic Factors Model
and the strength of law. It has even been Economic factors often have been cited
suggested that habitual "criminals" are in- as a cause of wartime and postwar crime
fluenced by the exigencies of war. For ex- rate changes. Commodity shortages and
ample, Mannheim (1941:108) quoted a other economic changes were the main var-
1914 article from The Times of London: iables in Exner's (1927) theory concern-
"The criminal like the honest citizen is ing wartime property crime, and Mann-
impressed by the War conditions which heim (1941:205) also concluded that
make it every man's duty to give as little property offenses "invariably thrive during
trouble as possible." and after wars." The general relationship
The social solidarity model leads to the between economic changes and homicide
expectation that rates of crime would de- is, however, much less clear (e.g., Henry
cline during wartime compared to the pre- and Short, 1954; Radzinowicz, 1971).
war period. The social solidarity model Discussions of war-related economic
also predicts that postwar crime rates changes have reflected two quite different
should, with the passing of the temporary explanatory variables: (1) scarcities cre-
crisis, be comparable to prewar rates. ated by wars and (2) the general level of
employment and other economic indica-
2. Social Disorganization Model
tors. This second explanation reflects the
Wars may disrupt the established order fact that while wartime years frequently
of societies, and some researchers have have brought full employment, postwar
attributed crime increases to this disorgan- years often have not. If economic factors
ization. The concept of social disorganiza- are related to violent crimes, different hom-
tion has been used with considerable im- icide rate changes would be expected as
precision, and there are really two versions a function of the relative economic health
of this model. Some researchers use it to of prewar and postwar society.
describe anomic changes which could occur The economic factors model predicts a
in any warring nation, victorious or de- decrease in postwar property crime rates
feated: property losses, rapid industrializa- in cases where the postwar economy is
tion, changes in the labor force, sudden relatively better than prewar and an in-
population migrations, the breaking up of
crease where the postwar economy is rela-
families, etc. tively worse. The implications of this model
Most researchers, however, have used for homicide rates are, however, unclear.
the concept to describe the social and
psychological changes in a defeated nation 4. Catharsis Model
(Mannheim, 1965:595; Lunden, 1963; A persistent belief about wars has been
1967:77-97). Sutherland and Cressey the idea that wars substitute public violence
(1960:209) used this version of the social for private violence. Sorokin (1925:146)
disorganization model: ". . . postwar crime claimed that revolutions temporarily abated
waves are confined largely to countries "criminal murders" only to have them re-
which suffer rather complete disintegration turn to normal levels at the end of the
of their economic, political and social sys- revolution when legally sanctioned killing
tems as a result of the war." became impossible once again. Many other
As used by some researchers, then, the researchers have suggested that wars pro-

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
vide an outlet or catharsis for aggressive have become habituated, whether by in-
sentiments or "instincts" (e.g., Mannheim, dulging in antisocial or criminal behavior
1941:128). or by offering their services to the highest
The catharsis model predicts that crimes bidder."
of violence would decrease during wartime. 6. Artifacts Model
The implications of this model for postwar
Some researchers have suggested that
rates depend upon how long the cathartic
any changes its wartime or postwar crime
effects are thought to last. However, a
rates are due to various artifacts (e.g.,
corollary of the catharsis model might be
Reckless, 1942:378). Two examples of
the prediction that societies whose exper-
demographic artifacts are the depression
iences in war had been the most violent,
of wartime crime rates by conscription and
and therefore the most cathartic, would ex-
the depression of postwar rates by the loss
perience postwar decreases in violent
of men killed and maimed in the war. In
crimes.
addition, specific artifacts may operate for
5. Violent Veteran Model individual nations in a given war. For ex-
ample, American crime rates in the 1960s
At least since the American Revolution,
began to be influenced by the age structure
people have wondered whether war veter-
changes produced by the "baby boom" of
ans would be more likely than others to
the post-World War II period (President's
commit crimes of violence. In a famous
Commission on Law Enforcement and Ad-
charge to a Charleston grand jury in 1783,
ministration of Justice, 1967:25).
Judge Aedanus Burke said that contem-
It may be possible to control for some
porary violent crime was being committed
artifacts. For example, in the case of the
by men who had become accustomed by
effect of the "baby boom," data on the age
the Revolutionary War to plundering and
of persons arrested could be examined to
killing their enemies and who had since
see if age groups other than the "baby
turned upon their neighbors (Wecter,
boom" cohort experienced changes. The
1944:70). The essential idea of this model
artifacts model does not make any general
is that the experience of war may have re-
predictions, but suggests the need to inspect
socialized soldiers to be more accepting of
any observed crime changes to control for
violence and more proficient at it.
possible artifacts.
It has been suggested, for example, that
combat developed in some men an "appe- 7. Legitimation of Violence Model
tite for violence" (Hamon, 1918:355) and The seventh way in which war could
the "habit" of violent solutions to problems affect postwar homicide rates is through a
(Abbott, 1918:40), and attorney Clarence legitimation of violence. The central con-
Darrow (1922:218) attributed post-World cept of this explanation is that some mem-
War I crime increases to returned veterans bers of a warring society are influenced by
who had been "innoculated with the uni- the "model" of officially approved wartime
versal madness." The idea of the homicide- killing and destruction. During a war, a
prone veteran has also appeared as a fre- society reverses its customary prohibitions
quent theme in fiction (e.g., Remarque's against killing and instead honors acts of
The Road Back, 1931) . violence which would be regarded as mur-
This recurring idea has surfaced again derous in peacetime. Several researchers
in connection with veterans of the Vietnam have suggested that this social approval
War. The potential problems of these vet- or legitimation of violence produces a last-
erans are the subject of several articles in ing reduction of inhibitions against taking
a collection edited by Mantell and Pilisuk human life (Sorokin, 1925:139; Engel-
(1975), and Lifton (1970:32) makes the brecht, 1937: 188-90). This model can be
following prediction about these veterans: illustrated by this quote from a Reverend
"Some are likely to seek continuing out- Charles Parsons around the time of World
lets to a pattern of violence to which they War I:

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
When the rules of civilized society are sus-

minimize the idiosyncratic experiences of
pended, when killing becomes a business individual nations or of individual wars.
and a sign of valor and heroism, when the
wanton destruction of peaceable women
Inspection of a large number of cases
and children becomes an act of virtue, and would also maximize the chances of dis-
is praised as a service to God and country,covering differential effects of various types
then it seems almost useless to talk about of wars and various types of participation.
crime in the ordinary sense. (Parsons,
In addition, we believe that only a strategy
1917:267)
of controlled comparison between com-
If wartime killing does legitimate homi- batant and noncombatant societies can dis-
cidal violence in some lasting or general count the possibility that both experience
way, as this model suggests, then one would the same crime rate changes.
expect increases in violent crime in postwar
societies. In addition, since civilians and
METHODS AND PROCEDURES
soldiers alike could be influenced by this
legitimation process, this model predicts Comparative Crime Data
that homicide increases will occur among The major obstacle in this area of re-
both veterans and nonveterans. search has been the absence of a dependent
Having listed these seven theoretical variable. For this reason, our first task was
models and the predictions associated with to construct an inventory of crime rate
each one, it should be mentioned that an data with historical and comparative depth.
empirical test of their relative strengths Over a period of three years, we have col-
contains serious pitfalls. As discussed ear- lected a 110-nation Comparative Crime
lier, predictions concerning wartime crime
Data File with time-series rates of various
are probably not investigable. There also offenses for the period 1900-1970.
may be some difficulties in evaluating pre- The principal sources for the creation of
dictions concerning postwar crime. For ex- our data file have been: (1) correspon-
ample, it may be that more than one of dence with national and metropolitan gov-
the models operates after a given war, and ernmental sources in a large number of
their interaction could make a comparison societies; (2) examination of documents
of the models extremely difficult-i.e., the and annual reports of those nations which
effects of one model could mask the effects have published data on their annual inci-
of another. dence of various crimes; (3) correspon-
It is also interesting to consider what dence with other record-keeping agencies.
kinds of evidence would constitute discon- For many nations, of course, the main-
firmation of each theoretical model. In tenance of crime records did not begin as
some previous studies, crime rate changes early as 1900. Developing nations, for ex-
in one society have been treated as suffi- ample, generally have data only for rela-
cient evidence upon which to ground or to tively recent periods. However, several na-
discount certain explanations. For example, tions have crime data from the beginning
Mannheim (1941) based his major work of the century, and a few even have records
on a single nation during a single war- from the mid-1800s-a longer span than
England during World War I. Mannheim we had assumed could be studied. As
interpreted the fact that English homicide might be expected, the records for some
rates did not increase after that war to nations contain gaps due to national emer-
mean that violent crimes do not in general gencies and bureaucratic lapses.
increase after wars. The same single case There are now about 110 nations in our
was used by Sutherland and Cressey comparative data file. For most of the
(1960) as evidence for a refutation of the nations, we have data on the annual inci-
violent veteran model. dence of five types of offenses: homicide,
It is our belief that a fair test of the assault, robbery, theft and rape.
different predictions requires a large and As part of our collection of the 110-
heterogeneous number of cases. This would nation data file, we have reviewed the

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
literature on possible sources of unrelia- offense like homicide (offenses known to

bility and invalidity in official crime data,the police, arrests, convictions, incarcera-
and have used this literature to identify a tions, prison population, etc.), the most
research design that minimizes these prob- accurate is certain to be the number of
lems. The literature concerning these ques- homicides known to the police (Sellin,
tions is large and has been reviewed else- 1931; Ferinand, 1967; Mulvihill and
where (Kitsuse and Cicourel, 1963; Wolf- Tumin, 1969; Clinard and Abbott, 1973:-
gang, 1963; Biderman and Reiss, 1967; 22; Hindelang, 1974). This suggests that
Wheeler, 1967; Hindelang, 1974; Archer a conservative procedure in a cross-na-
and Gartner, 1975), and so will not be tional study would be to limit comparisons
dealt with extensively here. Most research to nations using the best indicators of hom-
on these questions has been done with re- icide. For analyses of trends, however,
spect to U.S. record-keeping practices, there is some evidence that different indi-
specifically the Uniform Crime Reports cators of homicide are highly correlated
(UCR) currently maintained by the FBI over time (Archer and Gartner, 1975).
(1962-1971). We have chosen a research design to
One of the main concerns has been the minimize the effects of these problems.
degree to which official figures underenu- There are three essential features of this
merate the actual incidence of offenses. A design: (1) comparisons will be limited to
number of victimization surveys have esti- homicide; (2) only longitudinal indices of
mated that the magnitude of the "hidden" trends within societies will be compared,
figure of unreported crime is large for cer-
rather than levels across different societies;
tain offenses (Ennis, 1967; Biderman, (3) the analysis will control for the var-
1967; Santarelli et al., 1974a; 1974b). Aiable validity of different indices of homi-
few researchers also have discussed the cide. This third feature will consist of a
interpretability of crime data from other data quality control procedure (Naroll,
countries (Verkko, 1953; Wolfgang, 1967; 1962) that analyzes the national homicide
Wolf, 1971; Clinard and Abbott, 1973). data in two stages. The first of these in-
Three general conclusions emerge from cludes all nations with any type of homi-
research on the possible impact of under- cide indicator (offenses known, arrests,
reporting: (1) underreporting appears to etc.). The second stage, however, will be
be more of a problem for data on the less restricted to only those nations with the
serious offenses and not a major problem best indicator-the number of homicides
for data on homicide (Ferdinand, 1967; known to the police.
Hindelang, 1974); (2) official crime data
from many other nations appear less sub- Records of Wars
ject to underreporting than American data Previous discussions of the possible ef-
(Verkko, 1953; 1956; President's Com- fects of wars on crime also have considered
mission on Law Enforcement and Admin- problems with the independent variable-
istration of Criminal Justice, 1967:20); i.e., the meaningful measurement of char-
(3) although comparisons of the level of acteristics of different wars. For example,
an offense across several societies do not one of the problems cited by Sutherland
appear justified because of national idio- (1956:120) was variation among wars:
syncracies in definition and reporting (cf.
Wolfgang, 1967), comparisons of trends Wars vary widely in many respects, and
the constituent elements have not been
within various societies seem valid be-
standardized, nor have their comparative
cause these idiosyncrasies are likely to weights in the total complex of war been
remain relatively constant within a society determined. Consequently, we do not know,
(Sellin, 1926:31). even approximately, how much more "war"
A second area of concern has been the is involved in one war than in another.
comparability of different indicators of the Since Sutherland wrote, efforts have
same offense. Of all the indicators of an been made to operationalize measures of

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
the kind he indicated had been lacking. fied for France for discrete wars in this
The most comprehensive analysis has been period.
published by Singer and Small (1972). Two indices of change have been con-
These authors reviewed wars between structed to compare prewar and postwar
1816 and 1965 and, for each war, recorded rates. The first of these is the ratio of the
the number of war dead, the length of each mean homicide rate of the five postwar
nation's participation, the size of each na- years over the mean rate of the five prewar
tion's standing army, etc. Although the years. This index can be expressed as a
Singer and Small study may not have oper- percent comparing the postwar homicide
ationalized all the variables of possible rate to the prewar level.
relevance to our study, it has provided in- The second index of change is a t-test
dices which can be used to compare some between the mean prewar homicide rate
of the ways different wars have affected and the mean postwar rate. This index is
individual nations. similar to the percent measure, but it also
takes into account the variance of the pre-
Procedures
war and postwar rates. The Mtest corrects
This study compares postwar homicide for the degree of fluctuation in the prewar
rates with prewar rates, but, for reasons and postwar periods. In the case of a coun-
discussed earlier, will not discuss wartime try with wildly erratic fluctuations, the t
rates.' The basic design involves a com- would tend to be conservative-i.e., would
parison of the mean level of the rate of diminish the importance of the change. In
homicide during a fixed prewar period withthe case of a country with very stable rates,
the mean level during a fixed postwar however, the index of a postwar increase
period. Somewhat arbitrarily, the length wouldof be enhanced.
both periods has been set at five years, al-
The t-test in this instance is best thought
though homicide data are only available
of as an effect size index, rather than as a
for fewer than five years in a few cases.
conventional test of statistical significance
This length was preferred over shorter
(Cohen, 1969). This is because a t-test
periods (e.g., one year) to minimize the
based on ten observations (five prewar
impact of special social forces in the single
years and five postwar years) is a low
year before and after a war and to reduce
power test, and differences have to be very
the effect of annual fluctuations. The five-
large to reach significance with samples this
year interval was chosen instead of longer
small. In addition, significance levels are
periods for pragmatic reasons-the history
potentially misleading when large numbers
of the twentieth century has not been char-
of t-tests are calculated. For these reasons,
acterized by long intervals of peace. Even
the t-test is used here only as an effect
using a five-year interval, some wars can-
size index which takes into account both
not be interpreted. For example, the
the means and variances of the prewar and
Korean War cannot be included in this postwar periods.2
design, because the five years preceding it
The two indices of change provide dif-
include the years immediately following
ferent kinds of information. The percent
World War II. Use of the five-year interval
measure provides an easily interpretable
also made it impossible to interpret some
index of the overall magnitude and direc-
smaller wars. For example, France was in-
tion of change. The t provides a rough
volved in overlapping and consecutive co-
index of how unusual the change is in
lonial wars for many years after World
terms of the general variability of the na-
War II. For this reason, no peacetime pre-
tion's homicide rate. Although the direc-
war and postwar periods could be identi-
1 Data from the Vietnam War is an exception 2 Indeed, since the number of prewar years and
to this general rule. For this war, the comparison postwar years are the same (five), t bears a
will be between crime rates in the prewar period linear relation to d, and effect size index sug-
and those in the wartime years. This procedure gested for comparisons of two means (Cohen,
was necessary for this war because postwar data 1969:18). In general, d=2t/N'< and, in this
was unavailable. case, d = .707t.

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
-4 o ON C o o- 0 o c Ch Ch t- 't o ~o
* ; * . * .4 *4 *4 * t. * *4 * *6 * .l *4 . * *
N 00 oo q t 0 en q h en en en en WI- * t t e >
8~~~~~~~~~1 6>t^mN ".4 "^_
2 ti E ; t-^ t w E _ = E E e ~~~~ w X
O &. 0 m 'z S a o^t
3 ,,! > - 00 > > > ,,: > W > ] e~~~~~~cd.-(
3 Q e I = 2 2 > z C aii +Q ew 3 =
x~~~~~~~c 11 1 -d I~ I X
| ~ ~~~~ ~ 04 z z g !F
sX~~~~~~~~~~~~~~~~~~e C; coo -o Z
b Ot Ot~~~~~vv

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
tion (positive or negative) of the two file were used to test specific predictions
measures is always identical, their magni- derived from the seven theoretical models
tudes can be very different. discussed earlier.
There were several steps involved in
using the Comparative Crime Data File. RESULTS AND DISCUSSION
First, homicide data for some nations had

Our analysis indicates that combatant
to be converted from raw numbers to rates. nations in the two World Wars were more
This was done using available demo- likely to experience homicide rate increases
graphic data (chiefly the U.N. Demo- than control nations. The two measures
graphic Yearbook 1955; 1963; 1970-3). (% and t) of homicide rate change for
For many nations, the data was already combatant and control nations are shown
in the form of rates per 100,000 inhabi- in Table 1.
tants. In control nations for these two wars,
Second, the tables in Singer and Small homicide rate changes were evenly dis-
(1972) were used to determine the dates tributed, with some nations decreasing,
of entrance and withdrawal for each nation some increasing and some remaining un-
participating in a war. This information changed. For the 25 "nation-wars," how-
was used to identify the five prewar years ever, postwar increases outnumbered de-
and the five postwar years. creases by 19 to six.3 Combatant nations
in these two wars, therefore, were more
Third, comparison or control nations not
likely to experience postwar increases than
participating in a given war were identified
were control nations.4
from the comparative file. For both World
War I and II, the control group consisted The increases experienced by some com-
of all the nations in the file for those years batant nations in Table 1 were very large.
which did not participate in these wars. In Italy following World War II, for ex-
The control nations for other wars were ample, the homicide rate more than dou-
chosen using procedures discussed below. bled (an increase of 133%). The large
effect size index (t) associated with this
Fourth, the two measures of change in
change (3.07) indicates that this increase
homicide rates (% and t) were calculated
was very unusual with respect to Italy's
for both participating and control nations.
The changes for individual nations then
3.The unit of analysis is the "nation-war"~-i.e.,
were grouped into three categories accord-
one nation in one war. The number of individual
ing to the magniture of the percent mea- nations in Table 1 is less than the number of
sure. Nations which changed downward units, since some nations were involved in more
more than 10% were categorized as de- than one war. In Tables 1 and 2 combined, the
total number of nation-wars is 50 and the total
creasing, nations which changed upward
number of different combatant nations in these
more than 10% were categorized as in- wars is 28.
creasing, and nations in which the changes 4Although most of the control nations in
had an absolute magnitude of 10% or less
Table 1 indicate no homicide increases, or only
were categorized as unchanged. Then the
small increases, there are two anomalous cases:
Finland and Thailand after World War I. Both
number of combatant and control nations
of these nations show large homicide increases
falling into each of these categories was after World War I, and the size of these increases
compared. prompted us to investigate further the classifica-
Fifth, using only combatant nations, an tion of these nations as controls. A letter to the
Finland Consulate revealed that Finland under-
effort was made to determine what types went an "internal" or civil war in 1918, and
of war participation produce the greatest Thailand is identified in the Encyclopedia
changes. Two of the variables examined Britannica as having sent troops to the Allied
were: (1) the number of men killed dur- cause during World War I. This information
ing the war, as a proportion of the nation's raises questions about the wisdom of classifying
these two nations as controls for World War I.
total population and (2) whether the However, since neither nation was listed as a
nation was victorious or defeated. combatant by Singer and Small (1972), they are
Finally, other data in the comparative retained as controls in Table 1.

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
V!W"oo o t " Il t- "I o _
to > P-4 m .-"s
z~~~~~~~
zD o *>^tu
z Xn z o 1, V-4t
oo en~r WI No o o c ^ . -tee -I
vo 0 f
a te V U5 _A.0 ena
Q ,lll ,llllllll i
X t~mt Oneeen"^00 en
q 8 g S 02 V V V~~~"1 cm

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
general variability.5 There were also cases only the population "at risk" for homicide
of modest increases that were still rela- (e.g., males over 15), the increases for
tively large when judged in terms of the combatant nations would be even larger
general variability of the homicide rate. than shown in Table 1 (Nettler, 1974:59).
For example, the homicide rate in Belgium The fact that combatant nations still
after World War I experienced a 24% showed increases greater than the control
increase, but the t associated with this in- nations, despite the conservative effect of
crease (5.36) suggests that the change was these two demographic changes, is there-
very big in terms of Belgium's general fore particularly striking.6
variability. For smaller wars, the differences are in
Finally, there was at least one case the same direction, although less pro-
where the percent increase was compara- nounced. For the Vietnam War, since post-
tively larger than the corresponding t-test. war data were unavailable, the measure of
For Norway following World War II, a change involves a comparison of wartime
fairly large increase in the homicide rate years with the prewar period. This war
(65%) was tempered by only a moderate did not involve total mobilization of the
t (1.90), indicating that Norwegian homi- combatant societies (except for Vietnam
cide rates were characterized by consider- itself), and therefore wartime rates were
able variation during this period. unlikely to have been distorted by major
Even though the difference between demographic changes. Data for six com-
combatant and control nations in Table 1 batant nations in this war are included in
is quite large, this comparison is a con- Table 2.
servative test. As discussed earlier, the All six combatant nations in the Viet-
large numbers of young men killed in these nam War experienced homicide rate in-
wars dramatically changed the age and sex creases, although three of these increases
structures of the populations of combatant were less than 10 percent. The selection of
nations. As a result, the postwar homicide control nations for this war posed some
rates for the combatant nations were based problems, since the 1960s are a thickly
on populations depleted of young men- covered period in our 110-nation file. The
the group most likely to commit homicide. control nations shown in Table 2 were
The analysis in Table 1 may be conserva- matched with the six Vietnam combatants
tive for a second reason as well; the birth on the basis of geographic proximity. Three
of postwar babies in combatant nations of the six control nations experienced in-
(particularly after World War II) tends creases, but the other three controls exper-
to reduce postwar homicide rates by inflat- ienced decreases.
ing the population denominator on which Although the number of combatant and
they are based. control nations for the Vietnam analysis is
The net effect of these two demographic small, there is a tendency toward more
changes is to reduce postwar homicide consistent increases among the combatants.
rates in the combatant nations. If refined There is also reason to think that post-
homicide rates were available based on Vietnam War homicide rates may be even
higher than the wartime rates used in
5 The increase is even larger when judged Table 2. In the U.S., for example, the
only in terms of the prewar variation. We calcu-
homicide rate continued to increase with
lated a z-score measure of change which was
the difference between the prewar mean and
postwar mean, divided by the prewar standard 6 The comparison is probably a conservative
deviation. For Italian homicide rate increases test for a third reason as well. That is, there may
after World War II, this z-score measure was be some ways in which even control societies
11.62-larger than the corresponding t of 3.07. are affected by large wars like the two World
This indicates that the war not only increased Wars. The idea that the crime rates of even neu-
the level of the homicide rate but also increased tral nations might be influenced by massive wars
the instability or variance of the rate. For all has been suggested by Lunden (1963:5). If
the nation-wars in the study, however, this z-score changes did occur in neutral nations, of course,
index of change was significantly correlated withthey would be reflected in inflated increases in
t (r = .66, p <.001). the control nations in Table 1.

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
0
u
cd
cl
cd
t0
cc 00 0 g ".4
cd 0-i
CIS
0 cis
4-4 .4 +5 0 C's
c5s" C'd O 'm 4 Q
E.-4 U 44
(D
cd
>c
co
z aA
0 aA
cod V-4 r-4 X
co
C'd
C's C'S
cd cd
C's
cd
C4
WI
C13 C4 t-
kn o
a-, C's
VI cd
0
u cd cd cd 0 cd cd
(D __l
4
w 0
>1
cd
k cd 0 4)
0 +_ cd
0-4 0.4 00 0-4 m U W4 01-41 04
cri
S
cd 19
cd u
4.0
(D 0
-cod 0
E-4 O z u z

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
each year of the Vietnam War. As a result, war data on homicide convictions could be
although the mean U.S. wartime rate was depressed by judicial leniency toward vet-
42% higher than the prewar rate (as erans. The possibility of this type of error
shown in Table 2), the rate in the last led Sutherland (1956) to conclude that
years of the war was much higher-e.g., the indicator of convictions was seriously
in 1973 the U.S. homicide rate was 107% misleading. To control for the variable
higher than the mean prewar rate. validity of different indicators, the analyses
The magnitude of the U.S. homicide rate of Tables 1 and 2 have been repeated using
increase during the Vietnam War was also only nation-wars for which the best homi-
very large when judged by the t measure cide indicator is available-the number of
of change. The t associated with this war offenses known. This data quality control
(2.58) indicates that the wartime increase procedure, which removes nations with
was very large when compared to the gen- indicators like arrests and convictions, pro-
eral stability of the homicide rate. duces the results shown in Table 3.
A similar pattern occurs in the other The results in Table 3 produce the same
small wars for which analysis is possible. conclusion as the analyses in Tables 1 and
The number of interpretable nation-wars 2; combatants were more likely to exper-
is constrained by: (1) the need for control ience homicide increases than controls. The
nations, (2) the need for a prewar and difference between combatants and con-
postwar five-year period of peace and (3) trols, therefore, remains even when only
the absence of homicide data for some so- the best homicide indicator is used. The
cieties for certain periods. Despite these general conclusion, which seems justified
constraints, the effects of an additional 11 on the basis of these comparisons, is that
small wars can be assessed and the results combatants are more likely than controls
of this analysis are included in Table 2. to experience homicide increases.
Because of the large number of possible There were some combatants, however,
control nations for some of these wars, the which did not experience postwar in-
controls shown in Table 2 were selected creases. This suggests that some types of
as the closest noncombatant neighbors of participation are more likely to produce
participating nations. increases than others, as a number of re-
The results of this analysis are consistent searchers have speculated (e.g., Sorokin,
with those for World Wars I and II. The 1928).
nations participating in these 12 wars were In an effort to assess one dimension of
somewhat more likely to experience in- participation, we operationalized the degree
creases than were the controls. For com- to which participation was mortal for a
batant nations in these small wars, those nation-i.e., costly in terms of lives lost
with increases outnumbered those with in battle. We divided the combatant nations
decreases by eight to four. into two groups: nation-wars with more
The data in Tables 1 and 2 can be than 500 battle deaths per million prewar
pooled for an overall test of significance. If population and those with fewer than 500
Table 1 and 2 are superimposed, the dif- battle deaths per million prewar population
ferences between the 50 combatant nation- (Singer and Small, 1972). This rough
wars and the 30 controls are easily signifi- classification provides one way of group-
cant (X2=9.54, p=.0088). The same ing nation-wars according to the "amount"
result is obtained if the combatant and con- of war experienced by each.
trol nations are contrasted on the difference These two groups of combatants can be
between their mean prewar homicide rate compared. If changes in the homicide rate
and their mean postwar rate (Mann- are a function of the level of a nation's war
Whitney U= 1006.5, z= 2.55, p =.0054). involvement, then one would expect hom.-
As discussed earlier, it is important in icide increases to be greatest for the na-
comparative crime research to control for tions with the greatest losses. A compar-
differences in the types of indicators kept ison of the changes in the two groups of
by individual nations. For example, post- nation-wars is shown in Table 4.

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
0 0~~~~~~~'
'~~~~4 c~~cd c
~~ '~~~ '~~~~'Z0 0 -4
c~o
co CPO,~~~~~
Z o0-
o ~~~~~~ b~~~~~~cIco e
0 ~ ~~~~~~~~~~~~~~~~~~~~~~~~ 0 k
0 ~~~~~~~~0
CU~~~~~~~~~~~~~~~~~~~~~~~~~~ .
0 '0~~~~ d
241 a L 'C 4 )
O eO. 0

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
The differences are in the expected di- the predictions derived from the seven
rection. Nations with large combat losses theoretical models discussed earlier. Al-
showed homicide increases much more fre- though a conclusive test of some models is
quently than nations with fewer losses. difficult, many of their predictions can be
This comparison is certain to understate inspected to see if they are consistent with
the actual differences between the two our findings or are disconfirmed by them.
groups. Nations with large combat losses
had greatly reduced numbers of young men1. Social Solidarity Model
and, if rates were available for men over One of the predictions of this model was
15, the differences between the two groups that homicide rates would decline during
would be even larger than that shown in wartime and then, at or soon after war's
Table 4. end, return to prewar levels. As the evi-
We interpret this result as an internal dence in Tables 1 and 2 indicates, this
validation since the nations most likely to explanation does not appear to be sup-
show homicide increases are precisely those ported: postwar homicide rates were gen-
nations which experienced the largest erally higher than prewar rates for com-
"amounts" of war. The analysis in Table 4 batant nations.
suggests an explanation for some of the
anomalous cases in Tables 1 and 2: nations 2. Social Disorganization Model
with only limited participation and losses in As discussed earlier, there are really two
war may not exhibit postwar homicide in- versions of this model. One suggests that
creases. any crime increases would be due to dis-
The effect of other dimensions of war ruptions all warring societies could exyer-
participation also can be assessed. For ex- ience: rapid industrialization, population
ample, nations can be compared on the movements, breaking up of families, etc.
outcome of the war. Combatant nations Since these social changes apply to most
were classified as victorious or defeated or all wars, this version of the model is
using Singer and Small (1972) and other not easy to test and may even be untest-
sources for more recent wars. Table 5 able. The second version of the social dis-
shows the effect of adding this variable to organization model predicts that homicide
the analysis in Table 4. increases would be confined to defeated
Both victorious and defeated nations ex- nations. This second version of the social
perienced homicide increases, although vic- disorganization model can be considered
torious nations were more prone to in- disconfirmed; as shown in Table 5, post-
creases than defeated nations. The results war increases were not limited to defeated
in Table 5 confirm our finding that wars nations and were actually more frequent
with heavy combat losses produce hom- among victorious nations than defeated
icide increases and indicate that these in- nations.
creases occur more often when these wars
are won than lost. 3. Economic Factors Model
In summary, combatant nations exper- An explanation derived from this model
ienced increases in homicide more often was that postwar increases if they occurred,
than control nations. This is the major would be a function of worsened economic
finding of this study. In addition, nations conditions. As a rough test of this explana-
with high battle deaths experienced hom- tion, we have compared homicide changes
icide increases more often than nations in two groups of nations: those with wors-
with fewer battle deaths. Finally, victorious
ened postwar economies (compared to the
nations experienced homicide increases prewar period) and those with improved
more often than defeated nations. The war postwar economies. We were able to find
most likely to produce an increased hom- prewar and postwar unemployment rates
icide rate, therefore, is a war which is both for only a portion of the nation-wars in
deadly and won. Tables 1 and 2. These nations were classi-
It is also possible to examine some of fied according to the direction of unem-

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
~~~~~ _~~~~0 e
~~~~~~~4)~~~~~~~~P.
4)U%.-
'- IN
0 C-)~~~~~~~~~-o,%,
0~~~~~~1 c
o -~~~5c
- OU 0 v
0~~~~
cd~ ~ ~ ~ ~ ~ ~~~~~~~~'04~
4)~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~~e
F-~~~~~~~~~~~~~~~~~~~~~0 ;

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
ployment rate change between the five pre-

N I I I 4)
war years and five postwar years. The hom-
icide changes experienced by these nations
are shown in Table 6, along with the
percent change in each nation's unemploy-
ment rate.
-~~~~~~~~~-
The results in Table 6 do not support
an economic explanation of postwar hom-
icide rate increases. Both nations with
4)
worsened economies and those with im-
proved economies experienced postwar
homicide increases. In fact, nations with
improved economies were slightly more
prone to homicide increases than nations
with worsened economies. This finding ap-
00 %n 0 00
pears to disconfim an economic explana-
N ~ 00e 1
tion of postwar homicide rate increases.
4. Catharsis Model
A prediction derived from this model
was that societies whose wartime exper-
iences had been the most violent would
experience postwar decreases in violent
crimes. This prediction appears to be dis-
confirmed by the results shown in Table 4:
nations with the most fatal experiences in
war were precisely those most likely to
show homicide increases.
CAA
5. Violent Veteran Model
A prediction of this model was that post-
war increases would be due to the acts of
veterans. Direct evidence of whether vet-
V 4
erans are overrepresented in the commission
of homicide is difficult to obtain, and re-
search on this question will not be re-
0
viewed here. However, we have found
some evidence that this model is not suffi-
cient to explain the observed increases.
Specifically, there is evidence that in-
CO~~~~~~~~~~~~~~~~~C
creases have occurred among groups who
could not have been combat veterans. Dur-
ing the Vietnam War, for example, U.S.
arrests for homicide increased dramatically
'4.4 ~ ~ CA0 4) C
for both men and women. The increases in
homicide arrests between 1963 and 1973
4) .~~ 4)'O. were 101% for men and 59% for women
(F.B.I., 1963-73, Uniform Crime Re-
ports). Homicide arrests also increased for
all age groups-including people over the
age of 45-as discussed below. For the
Vietnam War, therefore, this explanation
is inadequate since increases also occurred

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
00 as en 00 eq VI wIt It
CN Ch - en
v 0 ch 0;
.0
.0
cq cN en v en C4 Go 'tt lr m cq W 00
cq
cd Cd >1 Cd Cd
Cd 0 , u 0 -cc
Iwo
z z
1.4
424
c
C
00
Am
4;
v
'o
0 0
z z Cd
12. >
z z 0
0
Cd .1-4
Cdw w
0
It
m 0
eq
Cd 0
Cd
o Cd
0
Cd
00
4)
w
Cd
0
0
0
I.C3 > 'd 0 0 0 0

C o C)
v C!
0 0 vvv
0 4Qu o 0,
0
o
t

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
among people who could not have been nation-wars in the study; but there is also
combat veterans. evidence that this argument cannot suffi-
Evidence on this question also can be ciently explain even these Vietnam War
obtained for other wars. Although sex- increases. As shown in Table 8, U.S. hom-
specific data for homicide is not available icide arrest data, analyzed by the age of
for all nation-wars, it is available for sev- offenders, show that increases occurred in
eral. Data on offenses by men and women all age groups, not just the "baby boom"
in five nations after World War II were cohort.
published by the Commission Internation- All age groups in Table 8 show an in-
ale Penale et Penitentiaire (1951), and crease in arrests for homicide during the
data on an additional six nation-wars were Vietnam War. For example, the age group
obtained from our Comparative Crime of 25 or older could not have included the
Data File. Postwar changes in offense rates "baby boom" cohort until 1970. But as
are shown separately for men and women Table 8 indicates, the homicide rate for
for 11 nation-wars in Table 7. this age group as well as older groups
As Table 7 indicates, the postwar increased throughout the Vietnam War.
changes for women were comparable to The "baby boom" clearly did have an
those for men in these 11 nation-wars. effect on American homicide rates, as
Even if postwar increases do occur among shown by the increase in the number of
veterans, therefore, these increases are not arrests for those under 25. However, older
confined to them but occur among both cohorts also showed increases, indicating
sexes. The model of the violent veteran, that the "baby boom" argument is not a
while difficult to disconfirm, cannot be a sufficient explanation of U.S. homicide in-
sufficient explanation of postwar increases. creases in this period.
6. Artifacts Model 7. Legitimation of Violence Model

The major finding of our analysis runs Our findings are consistent with this ex-
counter to the effects of some artifacts. planation. This may mean either that the
Combatant nations experienced postwar in- model has some validity-that wars do
creases in homicide, despite large losses of tend to legitimate the general use of vio-
young men. The effect of this demographic lence in domestic society-or that we have
artifact seems to be masked or over- not identified a compelling and testable
whelmed by the size of the increases in prediction which can be derived from the
homicide these nations experienced. model. Merely showing that rival explana-
Another demographic artifact often cited tions are disconfirmed or insufficient does
as a factor in American crime increases not mean that the surviving theory is au-
during the 1960s (e.g., President's Com- tomatically the correct one. Although we
mission on Law Enforcement and Admini- have tried to formulate and test all the
stration of Justice, 1967) involves the theoretical models that seem suggested by
effects of the post-World War II "baby classical and recent discussions of the pos-
boom" on homicide rates a generation sible effects of wars upon violent crimes,
later. It is important to note that the "babyit is possible that an alternate explana-
boom" argument could not explain in- tion remains unseen. However, the legiti-
creases following any wars in our analysis mation model is the only one of the seven
other than Vietnam. In addition, only three discussed in this paper that is completely
of the six Vietnam combatant nations were consistent with our finding of frequent and
World War II combatants who could have pervasive postwar homicide increases in
experienced the maturation of a "baby combatant nations.
boom" cohort during the Vietnam War: If wars do act to legitimate violence,
the U.S., Australia and New Zealand. The it is interesting to speculate about the
"baby boom" could only be a factor in ways in which this effect could be medi-
these three nation-wars and cannot explain ated. Since this model suggests that the
homicide rate increases after the other 47 public acts of governments can influence

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
4_ W? - > tto N Os~ 0%
_,len XN
O cq en
e n lq "tW~om
ON oo ON wmwtw 6moe
0 oo en " t- lV4e 0;
9~~~~~~~W N6 , t N, { 0~O, M 4
N e V)00"14"1 t-M O e el 10
.~~~~~e l ^ t- t b- WI ON el t ^
"~~~~~~. I. I-- V. I .44 V--4 I- I- V- - 04

B O
" .o
on t Y o ~n en k 'R v-f 0 r- o r- 0 ON M o N B 0
O s u t O Ch O ~~~~00 00 04 - O DO WQ
t A38~~~~~~~~~~~~~~~~~t
.o A vD ^ b ? vD ? b vD ^. m oo o
e > t Y 0 t t Om eq " N tn eq F >q
co t.. t- 0 ; -,> C4 c ti W; N6 rW t- cq Noz
*~~~~~I
oo
00 0 v--l '- ' t- rnt % 0
o g r- W 0 kn '.o m m k o 6t ? t
o ~ ~~~~~~~~ c4 N en 't *- 10 VD 10 -- D bb I I 1 Z
_ W > > o o > OF o o o > o > > > o ^ ^~~~~~~~c

i a tH W?._ e~~~~~~~Or
F~~~~~~~~~~~~~~~~~~~~~~~~~~~c 4;1e

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
the private acts of individuals, it might is often explicit, in the form of official
be possible to investigate potential links be- propaganda about the justice of the na-
tween the two. If panel survey data were tion's cause, the inhumanity of the enemy
available for each year during several dif- and the manly virtues of those who kill
ferent wars, for example, one might be the most enemy soldiers. This ligitimation
able to see if public attitudes became generally is directed at both a nation's
progressively more accepting or approving soldiers and at people on the home front;
of violence with each year of a war. For but it may be more credible to civilians
example, one could measure attitudes than to combat soldiers with direct ex-
toward shooting a burglar, toward using perience of the realities of war.
force during an argument, toward spank- Wars provide concrete evidence that
ing a child, toward avenging an insult, homicide, under some conditions, is ac-
toward the death penalty, etc. ceptable in the eyes of a nation's leaders.
In the absence of such panel surveys, This wartime reversal of the customary
other indices of the impact of wars might peacetime prohibition against killing may
reveal potentially important links to indi- somehow influence the threshold for using
vidual behavior. As one example, Hug- homicide as a means of settling conflict
gins and Straus (1975) have recently in everyday life.
shown that fictional violence in children's
literature reaches a maximum during war SUMMARY AND CONCLUSIONS
years. Wars also may produce changes
in other depictions of violence like movies, The idea that waging war could increase
news stories, billboards and propaganda, the level of postwar homicide in com-
etc. Since most homicides grow out of dis- batant societies has occurred to many re-
putes among relatives and acquaintances, searchers. Speculation about the possible
perhaps wars simply increase the probabil- effects of wars on domestic violence has
ity that violence will be regarded as a reflected seven theoretical models: (1)
justifiable means of resolving such disputes. Social Solidarity Model, which posits a
Wars are also always accompanied by wartime decrease in domestic homicide and
casualties, and little is known about how a postwar return to normal levels; (2)
the scale of wartime killing might affect Social Disorganization Model, which pre-
members of warring societies. Wars could dicts postwar increases but mainly for de-
affect individual behavior simply through feated nations; (3) Economic Factors
the awareness that violent homicides (in Model, which attributes any postwar in-
the form of soldiers killed in combat) are creases to a worsened postwar economy;
occurring during a war-sometimes in (4) Catharsis Model, which predicts post-
staggeringly large numbers. This awareness war decreases as a result of wartime killing;
could be communicated through a variety (5) Violent Veteran Model, which attri-
of media but does not necessarily require butes any postwar increases to comb-at
television, since we have found homicide veterans; (6) Artifacts Model, which ex-
increases after early twentieth century wars plains any postwar changes by demo-
as well as after recent wars. graphic and other forces; (7) Legitimation
Homicides in war may be a potent in- of Violence Model, which predicts postwar
fluence on postwar domestic violence for increases as a result of the pervasive war-
another reason as well. Unlike other kinds time presence of officially sanctioned kill-
of publicized killing (accidents, violent ing.
films and television, crime stories, etc.), To provide an empirical test of the im-
war involves homicide legitimated by the pact of wars on domestic homicide, a
highest auspices of the state. During many Comparative Crime Data File was assem-
wars, the killing of enemy soldiers has bled with homicide data for roughly 110
been treated not merely as a regrettable nations. The homicide rates of combatant
and expedient measure, but instead as nations were analyzed to learn (1) whether
praiseworthy and heroic. This legitimation postwar homicide increases occurred and

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
(2) whether the evidence was consistent Academy of Political and Social Science
374:1-15.
with or disconfirmed any of the seven
Bonger, W. A.
theoretical models.
1916 Criminality and Economic Conditions.
Most of the combatant nations in the New York: Little, Brown.
study experienced substantial postwar in- Bromberg, W.
creases in their rates of homicide. These 1943 "The effects of the war on crime."
American Sociological Review 8:685--
increases did not occur among a control
91.
group of noncombatant nations. The in- Calbairac, M. G.
creases were pervasive and occurred after 1928 "Les repercussions de la Grand Guerre
large and small wars, with several types sur la criminalite en France." Etudes
Criminologiques 3:62-70.
of homicide indicators, in victorious as well
Clinard, M. B. and D. J. Abbott
as defeated nations, in nations with im- 1973 Crime in Developing Countries. A
proved postwar economies and nations Comparative Approach. New York:
with worsened economies, among both men Wiley.
and women offenders, and among offenders Cohen, J.
1969 Statistical Power Analysis for the Be-
of several age groups. Postwar increases havioral Sciences. New York: Academic
were most frequent among nations with Press.
large numbers of combat deaths. Commission Internationale Penale et Penitentiare
These findings indicate, first, that post- 1951 Les Effets de la Guerre sur la Crimi-
nalite. Berne: Staempfli.
war homicide increases occur consistently
Darrow, Clarence
and, second, that several theoretical ex- 1922 Crime: Its Causes and Treatment. New
planations are either disconfirmed by evi- York: Crowell.
dence on postwar changes or are insuffi- Engelbrecht, H. C.
cient to explain them. The one model 1937 Revolt against War. New York: Dodd,
Mead.
which appears to be fully consistent with
Ennis, P. H.
the evidence is the legitimation model, 1967 "Criminal victimization in the United
which suggests that the presence of author- States: a report of a national survey."
ized or sanctioned killing during war has U.S. President's Commission on Law
Enforcement and Administration of
a residual effect on the level of homicide
Justice Field Survey II. Washington,
in peacetime society. D.C.: U.S. Government Printing Office.
Exner, F.
REFERENCES 1927 Krieg und Kriminalitat in Oesterreich.
Vienna: Holder-Pichler-Tempsky.
Abbott, E.
Federal Bureau of Investigation
1918 "Crime and the War." Journal of
1962- Crime in the U.S. Uniform Crime Re-
Criminal Law and Criminology 9:
1971 ports. Washington, D.C.: U.S. Govern-
32-45.
ment Printing Office.
1927 "The Civil War and the crime wave of
Ferdinand, T.
1865-70." Social Service Review 1:
1967 "The criminal patterns of Boston since
212-34.
1849." American Journal of Sociology
Archer, D. and R. Gartner
73:84-99.
1975 "Some aspects of reliability and validity
Glueck, E. T.
in comparative crime data." Unpub-
1942 "Wartime delinquency." Journal of
lished manuscript.
Criminal Law and Criminology 33:
Forth- Murder and War: The Boundary be-
119-35.
coming tween Private Violence and Public
Violence. Hamon, A.
Bennett, J. V. 1918 Lessons of the World War. Tr. B. Mall.
1953 "The ex-GI in federal prisons." Pp. London: Fisher Unwin.
131-6 in Proceedings of the American Henry, A. F. and J. F. Short, Jr.
Correctional Association. 1954 Suicide and Homicide: Some Economic,
Biderman, A. D. Sociological, and Psychological Aspects
1967 "Surveys of population samples for of Aggression. Glencoe: Free Press.
estimating crime incidence." The Annals Hindelang, J. J.
of the American Academy of Political 1974 "The Uniform Crime Reports revisited."
and Social Science 374:16-33. Journal of Criminal Justice 2:1-17.
Biderman, A. D. and J. J. Reiss Huggins, M. D. and M. A. Straus
1967 "On exploring the 'Dark Figure' of 1975 "Violence and the social structure as
crime." The Annals of the American reflected in children's books from 1850

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
to 1970." Paper presented at the 45th quency and prostitution." American

annual meeting of the Eastern Socio- Journal of Sociology 48:378-86.
logical Society. Remarque, Erich Maria
Kitsuse, J. I. and A. V. Cicourel 1931 The Road Back. Boston: Little, Brown.
1963 "A note on the uses of official statis- Roux, J. A.
tics." Social Problems 11:131-9. 1917 "Ce que sera la criminalite apres la
Leipman, M. guerre." Revue Politique et Parliamen-
1930 Krieg and Kriminalitat in Deutschland. taire 91:27-42.
Stuttgart: Deutsche-Verlagsanstalt. Santarelli, D. E., C. R. Work and R. W. Velde
Levi, N. 1974a Crime in the Nation's Five Largest
1929 "Statistica criminale e riforma della Cities. U.S. Department of Justice Ad-
legislazione penale." Pp. 267 ff. in Scritti vance Report, April. Washington, D.C.:
in Onore di Enrico Ferri. National Criminal Justice Information
Lifton, R. J. and Statistics Service.
1970 "The veterans return." New York 1974b Crime in Eight American Cities. U.S.
Times, Nov. 8:32. Department of Justice Advance Report,
Lowell, A. L. July. Washington, D.C.: National
1926 Public Opinion in War and Peace. Cam- Crminal Justice Information and Sta-
bridge: Harvard University Press. tistics Service.
Lunden, W. A. Sellin, T.
1963 War and Delinquency. Ames, Ia.: The 1926 "Is murder increasing in Europe?" The
Art Press. Annals of the American Academy of
1967 Crimes and Criminals. Ames, Ia.: Iowa Political and Social Science 126:29-34.
State University Press. 1931 "The basis of a crime index." Journal of
Mannheim, H. Criminal Law and Criminology 22:
1941 War and Crime. London: Watts. 335-56.
1955 Group Problems in Crime and Punish- 1942 "War and crime: a research memoran-
ment. London: Routledge and Kegan dum." Committee on Research on
Paul. Social Aspects of the War, September.
1965 Comparative Criminology. London: New York: Social Science Research
Routledge and Kegan Paul. Council.
Mantell, D. M. and M. Pilisuk (eds.)
Singer, J. D. and M. Small
1975 "Soldiers in and after Vietnam." Jour-
1972 The Wages of War, 1816-1965: A Sta-
nal of Social Issues 31:1-195.
tistical Handbook. New York: Wiley.
Mulvihill, D. J. and M. M. Tumin (co-directors)
1969 Crimes of Violence. National Commis- Solnar, V.
sion on the Causes and Prevention of 1929 "La guerre mondiale et la criminalite
Violence. Vols. 11, 12, 13. Washington, en Tchechoslovaquie." Revue de Droit
D.C.: U.S. Government Printing Office. Penal 9:858-93.
Naroll, R. Sorokin, P.
1962 Data Quality Control: A New Research 1925 The Sociology of Revolution. Phila-
Technique. New York: Free Press. delphia: Lippincourt.
Nettler, G. 1928 Contemporary Sociological Theories.
1974 Explaining Crime. New York: McGraw- New York: Harper.
Hill. Sumner, W. G.
Nevins, A. 1906 Folkways: A Study of the Sociological
1924 The American States during and after Importance of Usages, Manners, Cus-
the Revolution. New York: Macmillan. toms, Mores and Morals. Boston: Ginn.
Parsons, Charles Sutherland, E.
1917 "The influence of war on crime." Pp. 1943 "Crime." Pp. 185-206 in W. F. Ogburn
266-8 in Proceedings of the American (ed.), American Society in Wartime.
Correctional Association. Chicago: University of Chicago Press.
President's Commission on Law Enforcement and 1956 "Wartime crime." Pp. 120-8 in A.
Administration of Justice Cohen, A. Lindesmith and K. Schuessler
1967 Task Force Report: Crime and Its (eds.), The Sutherland Papers. Bloom-
Impact-An Assessment. Washington, ington, In.: Indiana University Press.
D.C.: U.S. Government Printing Office.
Sutherland E. and D. Cressey
Radzinowicz, L.
1960 Principles of Criminology. Chicago:
1971 "Economic pressures." Pp. 420-42 in
Lippincourt.
L. Radzinowicz and M. E. Wolfgang
(eds.), Crime and Justice, Volume I: Tarde, G.
The Criminal in Society. New York: 1912 Penal Philosophy. Boston: Little,
Basic Books. Brown.
Reckless, W. C. Tanter, R.
1942 "The impact of war on crime, delin- 1969 "International war and domestic tur-

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00
SELECTION AND ALLOCATION WITHIN SCHOOLS 963
moil: some contemporary evidence." Wheeler, S.

Pp. 550-69 in H. D. Graham and T. 1967 "Criminal statistics: a reformulation of
Gurr (eds.), The History of Violence the problem." Journal of Criminal Law
in America. New York: Praeger. and Criminology 58:317-24.
United Nations Statistical Office Willbach, H.
1955- United Nations Demographic Year- 1948 "Recent crimes and the veterans." Jour-
1973 book. Department of Economic and nal of Criminal Law and Criminology
Social Affairs. New York: U.N. Print- 38:501-8.
ing Office. Wolf, P.
Verkko, V. 1971 "Crime and development: an interna-
1953 "General theoretical viewpoints in tional comparison of crime rates."
criminal statistics regarding real crime." Scandanavian Studies in Criminology
Pp. 47-75 in Transactions of the Wes- 3:107-21.
termarck Society. Copenhagen: Munks- Wolfgang, M.
gaard. 1963 "Uniform Crime Reports: a critical
1956 "Survey of current practice in criminal appraisal." University of Pennsylvania
statistics." Transactions of the Wester- Law Review 708:708-38.
marck Society 3:5-33. 1967 "International crime statistics: a pro-
von Hentig, H. posal." Journal of Criminal Law,
1947 Crime: Causes and Conditions. New Criminology and Police Science 58:
York: McGraw-Hill. 65-9.
Wecter, D. Wolfgang, M. and F. Ferracuti
1944 When Johnny Comes Marching Home. 1967 The Subculture of Violence. New York:
Cambridge: Houghton-Mifflin. Tavistock.
SELECTION AND ALLOCATION WITHIN SCHOOLS:

SOME CAUSES AND CONSEQUENCES OF
CURRICULUM PLACEMENT *
KARL L. ALEXANDER AND EDWARD L. MCDILL

Johns Hopkins University
American Sociological Review 1976, Vol. 41 (December):963-980
A multivariate "school process" model is evaluated to explore the antecedents and

consequences of curriculum enrollment. Selected characteristics of peer associates,
measures of academic achievement and three subjective outcome variables-educational
plans, self-conceptions of academic competence and intellectual orientations-are con-
sidered. The analysis, based on questionnaire and testing data for a sample of high
school seniors, identifies academic ability and status origins as important determinants
of curriculum placement. Curriculum enrollment and other school process variables,
including characteristics of peer associates and academic performance, contribute uniquely
to the explanation of subjective orientations and serve as important mediators of back-
ground influence. Evidence is provided for the substantial importance of curriculum enroll-
ment for both cognitive an noncognitive outcomes of schooling, as well as, by implica-
tion, for retention in school.
Despite the severely critical initial re- markably robust. In particular, the finding
sponse to the Equality of Educational Op- that educational outcomes largely are in-
portunity Report (Coleman et al., 1966), dependent of all school-to-school differ-
its general conclusions have proven re- ences has been borne out both in thorough
reanalyses of the EEO data themselves
* We are indebted to the Education and Work (Mosteller and Moynihan, 1972) and in
Group, National Institute of Education and the a substantial body of subsequent research
Center for Social Organization of Schools, Johns
(Alexander and Eckland, 1975; Hauser,
Hopkins University for support in the prepara-
tion of this paper and to Larry J. Griffin and 1969; 1971; Hauser et al., 1976; McDill
James B. Smith for their able research assistance. and Rigsby, 1973). These results now have

198.161.51.147 on Tue, 30 Jan 2024 19:56:31 +00:00

Violent Acts and Violent Times: A Comparative Approach To Postwar Homicide Rates

Uploaded by

Copyright:

Available Formats

Violent Acts and Violent Times: A Comparative Approach To Postwar Homicide Rates

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Violent Acts and Violent Times: A Comparative Approach To Postwar Homicide Rates

Uploaded by

Copyright:

Available Formats

Violent Acts and Violent Times: A Comparative Approach to Postwar Homicide Rates

Author(s): Dane Archer and Rosemary Gartner

Stable URL: https://www.jstor.org/stable/2094796

This content downloaded from

A COMPARATIVE APPROACH TO POSTWAR HOMICIDE RATES *

DANE ARCHER AND ROSEMARY GARTNER

American Sociological Review 1976, Vol. 41 (December):937-963

This content downloaded from

This content downloaded from

This content downloaded from

This content downloaded from

This content downloaded from

distinct theoretically to justify a separate social disorganization model predicts that

This content downloaded from

This content downloaded from

When the rules of civilized society are sus-

This content downloaded from

literature on possible sources of unrelia- offense like homicide (offenses known to

This content downloaded from

This content downloaded from

O &. 0 m 'z S a o^t

3 ,,! > - 00 > > > ,,: > W > ] e~~~~~~cd.-(

This content downloaded from

First, homicide data for some nations had

This content downloaded from

V!W"oo o t " Il t- "I o _

to > P-4 m .-"s

This content downloaded from

This content downloaded from

This content downloaded from

This content downloaded from

This content downloaded from

This content downloaded from

This content downloaded from

ployment rate change between the five pre-

This content downloaded from

I.C3 > 'd 0 0 0 0

This content downloaded from

6. Artifacts Model 7. Legitimation of Violence Model

This content downloaded from

4_ W? - > tto N Os~ 0%

"~~~~~~. I. I-- V. I .44 V--4 I- I- V- - 04

e > t Y 0 t t Om eq " N tn eq F >q

co t.. t- 0 ; -,> C4 c ti W; N6 rW t- cq Noz

_ W > > o o > OF o o o > o > > > o ^ ^~~~~~~~c

This content downloaded from

This content downloaded from

This content downloaded from

to 1970." Paper presented at the 45th quency and prostitution." American

This content downloaded from

moil: some contemporary evidence." Wheeler, S.

SELECTION AND ALLOCATION WITHIN SCHOOLS:

KARL L. ALEXANDER AND EDWARD L. MCDILL

American Sociological Review 1976, Vol. 41 (December):963-980

A multivariate "school process" model is evaluated to explore the antecedents and

This content downloaded from

You might also like