Roger Williams University
School of Justice Studies Faculty Papers School of Justice Studies
Bad moon on the rise? Lunar cycles and incidents
Joseph A. Schafer
Southern Illinois University Carbondale
Sean P. Varano
Roger Williams University, firstname.lastname@example.org
John P. Jarvis
Behavioral Science Unit, Federal Bureau of Investigation
Jerey M. Cancino
Texas State University-San Marcos
Follow this and additional works at: hp://docs.rwu.edu/sjs_fp
Part of the Criminology and Criminal Justice Commons, and the Legal eory Commons
is Article is brought to you for free and open access by the School of Justice Studies at DOCS@RWU. It has been accepted for inclusion in School of
Justice Studies Faculty Papers by an authorized administrator of DOCS@RWU. For more information, please contact email@example.com.
Schafer, Joseph. A., Sean P. Varano, John P. Jarvis, and Jerey M. Cancino. 2010." Bad moon on the rise? Lunar cycles and incidents of
crime." Journal of Criminal Justice38 (4): 359-367.
Bad moon on the rise? Lunar cycles and incidents of crime☆
Joseph A. Schafer
⁎, Sean P. Varano
, John P. Jarvis
, Jeffrey M. Cancino
Department of Criminology & Criminal Justice, Southern Illinois University Carbondale, Carbondale, IL 62901-4504, United States
School of Justice Studies, Roger Williams University, One Old Ferry Rd., Bristol, RI 02809, United States
Behavioral Science Unit, Federal Bureau of Investigation, Quantico, Virginia 22135, United States
Department of Criminal Justice, Texas State University-San Marcos, 601 University Drive, San Marcos, TX 78666, United States
Popular cultures in Western societies have long espoused the notion that phases of the moon inﬂuence
human behavior. In particular, there is a common belief the full moon increases incidents of aberrant,
deviant, and criminal behavior. Using police, astronomical, and weather data from a major southwestern
American city, this study assessed whether lunar cycles related with rates of reported crime. The ﬁndings fail
to support popular lore, which has suggested that lunar phase inﬂuenced the volume of crime reported to
the police. Future research directions examining qualitative rather than quantitative aspects of this problem
may yield further inform the understanding of whether lunar cycles appreciably inﬂuence demands for
© 2010 Elsevier Ltd. All rights reserved.
It is the very error of the moon;
She comes more near the earth than she was wont,
And makes men mad.
-Othello, Act V, Scene II
Western lore has long suggested a relationship between the
phases of the moon and various forms of aberrant, antisocial, deviant,
and criminal human conduct. Popular thinking, visual arts, drama, and
literature have expressed this notion well beyond the legend of
vampire, werewolves, witches, and warlocks. The root of the word
“lunatic”is “luna,”Latin for “moon”(Merriam-Webster Online
Dictionary, 2008). Even in contemporary times with considerable
advances in scientiﬁc knowledge, there was a continued belief that
lunar phases effect human behavior (Lieber, 1996; Rotton & Kelly,
1985; Rotton, Kelly, & Elortegui, 1986) including limited evidence of a
high rate of belief among mental health care providers (Vance, 1995).
An on-going debate concerned whether lunar effects could be
empirically veriﬁed; that is, whether “a disproportionate number of
deviant or abnormal episodes occur when the moon is full”(Culver,
Rotton, & Kelly, 1988, p. 684).
Policing, crime, and criminal justice have not been immune from
speculation concerning the lunar-crime relationship. In working
with police ofﬁcers on a wide range of research projects, the authors
frequently heard individuals express the belief that the quality and/or
quantity of activity were linked with full moons (see also Lieber, 1996,
p. 21). Such beliefs have been systematically veriﬁed through surveys
of police ofﬁcers (Rotton et al., 1986; Vance, 1995). The study reported
herein examined the lunar-crime relationship using ﬁve years of
policing, astronomical, and weather data from a major southwestern
American city. The results were inconclusive in supporting what
popular lore would suggest being true. With few exceptions the
moon's phase was not related with the level of crime and disorder
reported to the police, controlling for relevant weather conditions.
The limited signiﬁcant ﬁndings would tend to contradict the lunar
lore, suggesting that when lunar phase mattered it constrained, rather
than increasing, targeted forms of aberrant behavior.
How might lunar phase affect behavior?
The cultural lore concerning lunar effects is vague in describing the
speciﬁc ways and reasons for the moon's effect. Most lunar effect
studies had not gone into great depth in explaining how and why the
moon inﬂuenced human conduct. It simply noted that lunar effects
might have been expected. Scientiﬁc study of lunar effects might best
be grouped into two broad categories: (1) the notion that lunar cycle
inﬂuences certain opportunities for criminal acts and (2) the notion
that the moon inﬂuences human behavior in some measurable
manner. Prior research tended to frame inquiry in light of the latter
hypothesis—that lunar cycles inﬂuenced rates of aberrant behavior.
The former category had received less scholarly inquiry, perhaps
because it implied the effect of the moon was far less direct. Rather
than actually causing human behavior, this perspective suggests that
rational choices concerning criminality are inﬂuenced by a number of
Journal of Criminal Justice 38 (2010) 359–367
☆This paper was accepted under the Editorship of Kent Joscelyn.
⁎Corresponding author. Tel.: +1 618 453 6376.
E-mail address: firstname.lastname@example.org (J.A. Schafer).
0047-2352/$ –see front matter © 2010 Elsevier Ltd. All rights reserved.
Contents lists available at ScienceDirect
Journal of Criminal Justice
variables, including lunar illumination. Under this line of thinking, a
full moon does not actually cause offenders to commit criminal acts;
rather, the illumination provided by a full moon serves to facilitate or
hinder certain forms of criminal conduct. Rational offenders might
include lunar illumination in making assessments about their
behaviors. Just as a burglar would look for a property offering easy
entry and with limited guardianship, she/he would also be motivated
to avoid committing that crime in well-lit areas, such as those with
exterior illumination, street lights, or signiﬁcant ambient moon light.
To do so would be to risk detection by alert citizens or the police.
The latter category, in contrast, was the subject of much of the
existing social and behavioral scholarship. The physical sciences have
long-established the relationship between the moon and geophysical
conditions on earth, particularly the ebb and ﬂow of tides. This
relationship led some medical and social science researchers to
postulate the high concentration of water in the human body would
result in a similar gravitational pull that could have affected human
behavior (see Lieber, 1978b, 1996). Although most research followed
this line of thinking, the relationship was usually not made explicitly
clear. Research was focused on testing the existence of a lunar effect,
rather than explaining the actual reason for an observed effect. The
vagueness of the mechanisms driving the postulated lunar-behavior
relationships complicated efforts aimed at empirical veriﬁcation.
Studies of lunar effects
Social science research has studied lunar effects using a wide range
of methodologies, data sources, populations, and variables. The
resulting body of literature tended to suggest weak to non-existent
lunar effects, though a few vocal scholars argued improper method-
ologies were to blame for mixed ﬁndings (Cyr & Kaplan, 1987, 1988;
Lieber & Sherin, 1972; Templer & Veleber, 1980). As addressed below,
the debate over lunar effects was as much a debate over ‘proper’
methodology as a debate over actual inﬂuences. Some contended
improper methods resulted in null ﬁndings; others argued that
stray observed effects tended to disappear upon introducing control
variables into an analysis (Bickis, Kelly, & Byrnes, 1995; Durm, Terry, &
Hammonds, 1986; Kelly & Rotton, 1983; Rotton, Kelly, & Frey, 1983).
Studies generally failed to ﬁnd a lunar effect, though researchers
found varying associations between lunar phase and some categories
of crime (Purpura, 1979), aggression (Lieber, 1978a), violent incidents
in correctional settings (Pettigrew, 1985), volume of demand for
emergency room services (Blackmon & Catalina, 1973), suicide
attempts (Taylor & Diespecker, 1972), hospital admissions (Templer
& Veleber, 1980; Weiskott & Tipton, 1975), and calls to telephone
counseling services (Templer & Veleber, 1980; Weiskott, 1974). Most
studies employed few (if any) controls, used short time frames (i.e.,
four months, which only measured effects across four lunar cycles),
and achieved relatively weak statistical signiﬁcance, sometimes
ﬁnding effects on some dependent variables, but not on others. In
addition to ﬁnding a lunar effect during periods of full moons, studies
found effects of new moons—periods when the moon was not
reﬂecting the sun's light back toward earth, making it difﬁcult to
observe with the naked eye (Templer & Veleber, 1980; Templer,
Veleber, & Brooner, 1982).
Despite these positive associations, the majority of lunar research
failed to establish effects. This included failing to ﬁnd effects on:
aggressive acts in prisons (Atlas, 1984; Simon, 1998) and psychiatric
facilities (Durm et al., 1986; Quinsey & Varney, 1977), prison escapes
(Pettigrew, 1985), use of hospital psychiatric services (Bauer &
Hornick, 1968; Shapiro, Streiner, Gray, Williams, & Soble, 1970;
Walters, Markley, & Tiffany, 1975), incidents of suicide (Biermann
et al., 2005), use of telephone crisis centers (Wilson & Tobacyk,
1990), hospital admissions for dog bite injuries (Chapman &
Morrell, 2000), aggression among ice hockey players (Russell & de
Graaf, 1985; Russell & Dua, 1983), demands for police and ﬁre services
(Bickis et al., 1995; Frey, Rotton, & Barry, 1979), automobile accidents
(Campbell & Beets, 1978; Laverty & Kelly, 1998), suicide (Lester,
Brockopp, & Priebe, 1969), and homicide (Pokorny & Jachimczyk,
1974). Studies often achieved contradictory results on crime-related
outcomes, particularly calls for policing service and homicides. At
times, two sets of researchers examined the same data and achieved
opposite conclusions (cf. Kelly & Rotton, 1983; Templer et al., 1982)
due to inconsistency in conceptualizing and operationalizing key
study variables, as well as using different statistical controls and
Lunar effects on crime
Despite the volume of research testing for lunar effects, few
researchers employed crime-related outcomes (exceptions included
Cohn, 1993; Purpura, 1979). Where relationships were found lunar
inﬂuences were modest, particularly in contrast to temporal and
situational predictors. There are several reasons to hypothesize that
lunar cycles may inﬂuence crime, either due to direct effects on
human behavior or as a factor inﬂuencing rational offenders. First and
foremost, though not directly tested in this study, research has shown
that belief in lunar effects continued to be strong (Russell & Dua,
1983), including among police ofﬁcers (Rotton & Kelly, 1985; Rotton
et al., 1986; Vance, 1995) and tended to be associated with beliefs in
other paranormal phenomena (Rotton & Kelly, 1985). Ofﬁcers have
the beneﬁt of directly observing the aftermath of criminal incidents;
their experiences may provide them with unique insights into lunar
effects, or police culture may simply perpetuate false beliefs about
lunar effects. Second, studies of lunar effects on crime tended to use
limited offenses and short time frames, while failing to introduce
control variables. This study sought to overcome weaknesses in
prior studies by considering a long-time frame, multiple offense
types, and salient control variables. Third, moving beyond the beliefs
of police ofﬁcers, there was evidence that beliefs in lunar effects
were prevalent. Testing the veracity of these beliefs was of obvious
Is crime inﬂuenced by seasons or short-term variations in
One of the ﬁrst observations about the effects of weather on crime
was offered by Belgian statistician Adolphe Quételet (1842/1969),
who noted “during summer, the greatest number of crimes against
persons are committed and the fewest crimes against property; the
contrary takes place during the winter”(p. 90). For over 150 years
scholars tested and elaborated on this assertion, introducing broader
meteorological data into empirical models. The social contact
hypothesis suggests that during times of pleasant weather, aggressive
and hostile acts may be more common because there is an increase in
normal human interactions, which increases the opportunity for
interpersonal conﬂict. Within weather and crime research, scholars
attempted to sort out the social versus physiological effects of weather
on human behavior (Cheatwood, 1995).
More contemporary analyses of policing data from American
communities tended to ﬁnd that many types of personal and property
crimes were more common during periods of warm versus cool or
cold weather (Cheatwood, 1995; Cohn, 1990b, 1996; Cohn & Rotton,
2000; Hipp, Bauer, Curran, & Bollen, 2004; Rotton & Cohn, 2003) and
that demand for police services were greater during periods of
warmer temperature and longer hours of daylight (i.e., spring and
summer) (Cohn, 1996; Heller & Markland, 1970; LeBeau & Corcoran,
1990; LeBeau & Langworthy, 1986).
Research suggested this
relationship generally held true, though there were upper limits.
Rotton and Cohn (2000) found disorderly conduct and assault in
Minneapolis had an ‘inverted U-shaped’relationship; during periods
of extreme heat and cold these behaviors were less frequent. In
360 J.A. Schafer et al. / Journal of Criminal Justice 38 (2010) 359–367
reviewing early weather and crime research, Cohn (1990a) observed
research established stronger relationships for personal than property
assaults, burglary, collective violence, domestic violence, and rape
tend to increase with ambient temperature…The relationship
between heat and homicide is uncertain. High temperatures do
not appear to be correlated with robbery, larceny, and motor
vehicle theft. (p. 61)
When analyses accounted for temporal variables (i.e., time of
day, weekends, holidays, periods when public schools are not in
session), weather effects offered relatively modest explanatory effects
(Cheatwood, 1995; Cohn, 1993). The inﬂuence of temporal and
weather variables were presumably a result of changes in routine
activities of offenders and victims (e.g., Cohen & Felson, 1979), though
these effects might be trumped by situational variables (Rotton &
A signiﬁcant challenge surrounding the study of lunar effects on
human behavior is the imprecision of this cultural lore. What, exactly,
is the moon supposed to inﬂuence and how is that inﬂuence exercised?
Such basic questions hold tremendous implications for conceptual,
operational, and analytical decisions. Similar concerns are noted
in literature examining weather-crime relationships (Block, 1983,
How, for example, should lunar phase be deﬁned and
Also important to consider is whether the relationship is
based on the moon's visibility, which would suggest restricting an
analysis to nighttime hours and controlling for salient weather
If there is an effect on crime, is it the quality of crime
(i.e., offenders act more ‘crazy’than normal), quantity (i.e., the
volume of crime), or both? Will crime-related effects translate into
police records in an observable manner? The contention that existing
“methodological chaos has made impossible any consistency in results
in the evaluation of the lunar hypothesis”(Cyr & Kaplan, 1987, p. 391)
was well founded. While characterizing previous efforts to study
this alleged phenomenon as chaotic may be overly dramatic, such
difﬁculties should be expected when scholars sought to study a vague,
ill-deﬁned, and indistinct dimension of western cultural lore.
Decisions concerning proper data sources, variable structures,
operational deﬁnitions, and statistical tests, among other considera-
tions, were not straight-forward. Block (1983) noted a similar
assessment when testing for seasonal effects on crime. Her mono-
graph offered a number of insightful parallels to analyzing lunar
effects. For example, Block noted researchers were typically con-
strained by the quality and structure of the data at hand; even when
there was great control over data sources and variable structures,
there were few clear rules for conducting analyses. Given the folklore
origins of the lunar effect, researchers did not have a clear map
informing choices concerning the selection of dependent and
independent variables, or the analytical methods used to examine
data. This situation certainly contributed to the widely varying
methodologies and results observed across salient research literature.
The purpose of this study was to test the effects of lunar phase on
criminal and criminal and disorderly behavior through the construct
of methodological framework more rigorous than those found in
many prior studies. San Antonio, Texas, is located in the southwestern
United States and has a population of more than one million residents.
Using call for service (CFS) to the San Antonio Police Department
(SAPD) (e.g., 911) over a ﬁve-year period (2001-2005) as a measure
of crime and disorder, a time-series design was used to determine the
linear relationship between lunar cycles and levels of reported crime
and disorder. Calls for service that occurred during nighttime hours
(8 p.m. –3 a.m.) were grouped into seven categories: assaultive
violence, burglary, theft, drugs/vice, trafﬁc, other disturbances, and
aggregated CFS (the sum of the six categories). In so doing, the study
attempted to control for two competing perspectives about the
inﬂuence of lunar phase. First, did lunar phase itself appear to
inﬂuence activities reported to the police; did lunar phase produce a
quantitative shift in human behavior as measured by the volume of
select categories of calls to the police? Second, could lunar phase
create conditions that would enhance or diminish the rationality of
select offenses? The research design accounted for potential impacts
of land use issues by analyzing citywide trends, but also considered
trends in both San Antonio's entertainment area and non-entertain-
The analysis relied on ﬁve years (2001-2005) of data for police
calls for service and National Weather Service (NWS) records that
depicted San Antonio's weather patterns during the same period. As
state above, the purpose of this research was to determine if there was
a relationship between lunar cycles and nightly crime levels, a
perception that was embedded in the folklore shared by both police
personnel and citizens. The analysis ﬁrst considered descriptive
analysis of variables along with bivariate correlation analysis; this
was followed by multivariate and time series analysis on each of the
dependent variables to determine the impact of lunar cycles on levels
of crime and disorder.
CFS data represented instances of crime and disorder that were
received by the dispatch unit of the SAPD and are more typically
known as “911 calls”to most citizens. The data provided by SAPD's
crime analysis unit were limited to calls of direct jurisdiction of law
enforcement (i.e., requests falling under the jurisdiction of other
service providers were not included). Incidents of concern to law
enforcement typically involved alleged crimes or other incidents
requiring that an ofﬁcer be dispatched to a scene. During the time
period under consideration, SAPD received approximately 900,000
CFS per year; this resulted in approximately 225,000 completed police
incident reports per year. Annually, approximately 130,000 CFS
concerned the six disorder categories considered in this analysis.
Nightly counts of crime and disorder were computed for seven CFS
categories: assaultive violence, burglary, theft, drugs/vice, trafﬁc, other
disturbances, and aggregated CFS. Nighttime was operationalized as
reported incidents between 8 p.m. to 3 a.m. Using January 2, 2001 as an
example, total nighttime CFS were calculated for calls received
between 8 p.m. on January 1 and 3 a.m. on January 2, 2001. San
Antonio has a well-known and established entertainment district (the
Riverwalk) within its borders. The Riverwalk draws a high volume of
tourists and residents to restaurants, bars, hotels, convention facilities,
shops, and other entertainment venues. To account for possible land
use effects (see Brantingham & Brantingham, 1995; Robinson &
Rengert, 2006) CFS totals were computed for the entire city, the non-
entertainment area alone, and the entertainment area alone. This
strategy allowed for consideration of any possible interplay with land
use (see Table 1 for descriptive statistics).
The key independent variable, lunar phase, was a continuous level
variable that measured the where the moon was in its lunar cycle at
midnight during each date in the analysis. This data were publicly
available from the Astronomical Applications Department of the U.S.
361J.A. Schafer et al. / Journal of Criminal Justice 38 (2010) 359–367
Naval Observatory (see http://aa.usno.navy.mil). The variable ranged
from 0 (a new moon night) to 1.00 (a full moon night); this value was
based on the presumption of a clear sky and did not take into account
any weather conditions that might have obstructed the actual
visibility of the moon.
Each night constructed for the dependent
variable was paired with the respective lunar phase at midnight. In
other words, the dependent variable consisted of CFS counts for each
night (8 pm –3 am) in the ﬁve years under consideration; these
nights were paired with the lunar phase reported for midnight on that
same night. This approach represented a substantial advancement in
this area of study. Prior research tended to operationalize lunar phase
in dichotomous (full moon versus non-full moon periods) (see Atlas,
1984; Cohn, 1993; Purpura, 1979) or basic ordinal terms (full moon,
new moon, and “interphase”; new moon, ﬁrst quarter, full moon, last
quarter) (see Kelly & Rotton, 1983; Pokorny & Jachimczyk, 1974;
Templer et al., 1982). These approaches restricted the lunar phase to
nominal or ordinal variability; this analysis employed a more robust
ratio variable to represent lunar phase.
Control variables were included to account for possible seasonal,
temporal and weather effects that might have mediate lunar effects.
Winter accounted for seasonal ﬂuctuations in weather that might have
affected socialization patterns related with rates of crime. Winter was
a dummy variable where “1”indicated dates that fell between
December 1 and March 31. It was expected that crime patterns would
naturally decrease during winter months and then increase during
other months (see Cheatwood, 1988). The variable weekend, also a
binary variable, was coded as “1”to represent Fridays, Saturdays, and
Sundays. Social patterns that inﬂuence levels of crime and disorder
were expected to independently affect nightly crime patterns.
Additional control variables represented weather patterns and
were derived from the National Climatic Data Center (NCDC), a
division of the National Oceanic and Atmospheric Association. NCDC
records are based on data and observations collected by instruments
and human observers at NWS stations around the country. Data from
the NWS station located at the San Antonio International Airport were
used to represent weather. These controls were important because if
lunar cycles inﬂuenced the opportunity for crime through increased
visibility, such effects could have been diminished by inclement
weather that obscures such visibility. Weather also had the capacity to
inﬂuence social patterns that may have independently inﬂuenced
crime and disorder levels.
Sunset was a continuous variable that identiﬁed the ofﬁcial time of
sunset based on a twenty-four hour (military) clock. Time of sunset
may have inﬂuenced social patterns that also affected crime levels.
Nightly temp reﬂected the temperature recorded by the NCDS in
degrees Fahrenheit. The NCDS data included temperature observa-
tions taken at three hour increments thorough the day. For this
analysis mean nightly temperatures were computed using the 9 pm,
midnight, and 3 am observation periods. The temperature was
expected to be positively associated with crime levels. Finally,
precipitation was a continuous variable that measured precipitation
(in inches) during the twenty-four hours on each day. This measure
was a proxy indicator of the overall extent to which prevailing
weather patterns might have made criminal acts less favorable. Levels
of crime and disorder were expected to be negatively associated with
precipitation. See Table 1 for descriptive statistics.
Analysis and ﬁndings
To test for possible lunar effects on crime and disorder levels,
analyses of nightly crime trends was done while controlling for
temporal and weather variables that may mediate any apparent
effects. The analysis included univariate descriptive statistics that
describes each of the independent variables across the three land use
types (e.g., citywide, non-entertainment area, and entertainment area
only). In addition, bivariate and multivariate statistics were used to
determine the relationships between lunar cycles and crime while
controlling for other possible explanatory variables. The multivariate
analysis included time series design commonly referred to as
autoregressive moving average (ARIMA) technique or ordinary-least
squares regression. The exact analytical techniques used were based
on a preliminary analysis of time or season-related patterns that
required statistical control. A diagnostic tool available in the SPSS was
used to automatically determine if time-or-season-dependent trends
existed in each trend. When no such patterns were evident, simple
OLS analysis was used. When time-dependent trends were apparent,
time-series analysis was used to control for these patterns.
The descriptive statistics are included in Table 1. There were 1,826
nights included in the study period of January 1, 2001 to December
31st 2005. These nightly units comprised each unique observation or
unit of analysis. The statistics included in the top half of Table 1
provide summary measures of nightly crime totals (counts) for
aggregated crime and the six subcategories (assaultive violence,
burglary, theft, drugs/vice, trafﬁc, and disturbances) for all three areas
of the city. It is important to note that these crime classiﬁcations were
based on how events were classiﬁed in call for service (911) data and
not ofﬁcial police reports. This method of measuring crime and
disorder did not take into account if police failed to ﬁle a police report
or changed the classiﬁcation of an incident to another category than
originally classiﬁed in the call data. Thus, the categorizations reﬂected
those made by call takers after consultation with complainants/
victims. An advantage of this approach was removing the possibly
biasing effects of ofﬁcer beliefs; if ofﬁcers believed crime increased
during full moon phases they may have had a proclivity to categories
incidents as criminal and/or more serious in nature.
Descriptive statistics for nightly totals (n = 1826)
Citywide Totals Non-Entertainment Area Entertainment Area Only
Min. Max. Mean Std. Dev. Min. Max. Mean Std. Dev. Min. Max. Mean Std. Dev.
Totals 0 806 359.81 101.12 0 750 340.97 96.11 0 111 21.80 15.84
Assaultive Violence 0 72 19.38 9.51 0 65 17.62 8.73 0 47 4.55 5.84
Burglary 0 265 58.09 19.32 0 255 55.09 18.54 0 26 2.99 2.61
Theft 0 16 2.11 1.85 0 16 1.90 1.76 0 13 0.39 1.13
Drugs/Vice 0 32 10.97 5.61 0 32 10.14 5.31 0 15 0.83 1.31
Trafﬁc 0 84 23.45 9.29 0 83 22.07 8.70 0 11 1.38 1.62
Disturbances 0 599 226.91 74.14 0 573 216.24 71.10 0 50 10.66 7.52
Lunar Phase 0 1 0.50 0.35
Winter 0 1 0.33 0.47
Weekend 0 1 0.29 0.45
Sunset 1735 1938 1843.72 71.61
Nightly Temp 31 86 65.21 12.90
Precip 0 9.5 0.10 0.39
362 J.A. Schafer et al. / Journal of Criminal Justice 38 (2010) 359–367
A comparison of the absolute values of average nightly counts
across geographical areas was not valid because the areas included
different geographical sizes with different size populations at risk. The
data presented in Table 1 indicated that the average number of calls
for service reporting criminal situations was approximately 360 per
night with a range 806 reports (see citywide totals). The largest single
group of reported crimes/disorders was the “disturbances”category
for the citywide data, but also for the entertainment and non-
entertainment areas. This category was a very broad grouping that
included a wide-range of problematic behavior including disorderly
persons, loud parties, suspicious persons, etc. It amounted to a “catch
all”classiﬁcation for events that did not fall neatly into other
categories. The next largest categories were burglary and assaultive
violence for both the citywide and non-entertainment area totals. The
average nightly totals for assaultive violence (x̄= 4.55) were higher in
the entertainment area compared to other sectors of the city. This was
likely due to differences in land use patterns and the nature of
socialization patterns (e.g., high density of bars and other entertain-
The second grouping of variables included the principle indepen-
dent variables, lunar phase, and ﬁve additional variables that
controlled for seasonal (winter), temporal (weekend and time of
sunset), and weather (nightly temperature and precipitation) varia-
tions. The average lunar phase across the observation time was ﬁfty-
percent. The sunset time was measured on a military time scale that
ranges from 0-2400; the average time of sunset was 1843 hours, or
6:43 pm. The average temperature during the nightly observation
periods was 65 degrees (Fahrenheit). Finally, there was approximate-
ly .10 inches of daily precipitation during the study period, though an
appreciable range was noted on that variable.
Bivariate correlations are included in Table 2. At the bivariate level,
the data suggested there was no relationship between lunar phase
and crime. This relationship generally held for citywide disorder
patterns, but also for crime trends in non-entertainment and
entertainment areas with one exception. The relationship between
drugs/vice crimes in the entertainment district was signiﬁcant and
negative, although the coefﬁcient was relatively small. This relation-
ship suggested that while illumination was not a factor in crime, the
inﬂuence was dependent on land use patterns. It was also noteworthy
that many of the temporal, seasonal and weather control variables
were signiﬁcant across the data, largely in the expected directions.
Winter was a seasonal control variable; it was hypothesized that
due to changes in season-related socialization patterns, crime would
likely be lower during the winter. The signiﬁcant and negative
relationship across crime types and locations indicates that crime/
disorder was generally lower during winter months, even in such a
temperate climate. Weekend was a temporal control that depicted if
the observation period occurred on Friday, Saturday, or Sunday.
Related again to natural variations in socialization patterns linked to
day-of-the-week, a positive relationship was expected. That is, more
crime was expected on the weekends. The hypothesis was supported
by the signiﬁcant and positive relationships across crime types and
geographical locations, relationships that were greater than 0.7 in
several cases. The data presented in Table 2 also indicated a consistent
signiﬁcant and positive relationship between daily sunset and crime.
This indicated that crime increased as sunset was delayed. This ﬁnding
was interesting because it suggested levels of crime were positively
associated with levels of solar illumination.
Finally, the data also indicated strong relationships between levels
of crime/disorder and both average nightly temperature and levels of
precipitation. The signiﬁcant and positive coefﬁcients for temperature
indicated that crime increased during warmer days across geograph-
ical locations. Finally, a pattern emerged suggesting that when
signiﬁcant, the relationship was generally negative. This indicated
that crime decreased on rainy nights. Interesting, the relationship for
precipitation and crime was less apparent in the entertainment area
which might suggest that tourists were less deterred by bad weather.
The bivariate analyses indicated that while all of the control
variables were signiﬁcantly related to levels of crime and disorder,
lunar phase demonstrated little explanatory power. With the
exception of levels of drugs and vice in the entertainment area,
lunar phase showed no signiﬁcant relationships. Moreover, the
direction of the relationships was scattered across types of disorder-
ly/criminal behavior and geographical areas. Results of the bivariate
analysis seemed contrary to the lore of a lunar affect on crime and
disorder; in fact the weak drugs/vice ﬁnding in the entertainment
area is contrary to what lore would predict. To further test this
relationship, multivariate analysis (both ordinary least squares
regression and time series analysis) was used to understand the
relationships between lunar phase and crime while controlling for
other temporal, seasonal, and weather related conditions.
Time-series analysis was an appropriate analytical strategy for
regressing time-and-season dependent variables on independent
variables. The autoregressive integrative moving average model
(ARIMA) was developed by Box and Jenkins (1976) as a methodo-
logical strategy for removing trends from data. ARIMA analysis
involves an iterative model building process where the analyst
diagnoses the structure of the trend and then builds the appropriate
statistical model to these biasing effects that may distort statistical
relationships. ARIMA models have three structural parameters
that must be diagnosed and modeled, autoregressive (p); difference
(d); and moving average (q) parameters (see McDowall, McCleary,
Meidinger & Hay, 1980).
The ﬁrst step in constructing these models is to ensure trends are
stationary. Stationary refers to the degree to which the data series
ﬂuctuates around a ﬁxed mean. A nonstationary trend is represented
by a non-zero integer in the dterm. The differencing process involves
subtracting a value from the preceding k
observations with k
representing the dparameter. The autoregressive parameter reﬂects
the mathematical relationship between an observation and k
preceding values. Finally, the moving average parameter (q) is similar
Winter Weekend Sunset Nightly
City Trafﬁc -0.00 -0.27** 0.57** 0.26** 0.29** 0.05*
City Burglary 0.02 -0.16** 0.24** 0.25** 0.21** 0.14**
City Disturbance 0.00 -0.26** 0.79** 0.22** 0.26** -0.08**
City Assault -0.02 -0.23** 0.71** 0.19** 0.25** -0.07**
City Drugs/Vice -0.01 -0.32** 0.22** 0.31** 0.34** -0.09**
City Theft -0.01 -0.12** 0.15** 0.14** 0.11** -0.02
City Aggregated 0.00 -0.31** 0.78** 0.29** 0.32** -0.05*
NonEnt Trafﬁc -0.00 -0.27** 0.55** 0.26** 0.29** 0.06*
NonEnt Burglary 0.02 -0.17** 0.24** 0.25** 0.22** 0.14**
NonEnt Disturbance -0.00 -0.26** 0.79** 0.23** 0.26** -0.08**
NonEnt Assault -0.01 -0.23** 0.70** 0.19** 0.25** -0.07**
NonEnt Drugs/Vice 0.00 -0.31** 0.19** 0.31** 0.35** -0.09**
NonEnt Theft -0.00 -0.11** 0.12** 0.13** 0.11** -0.03
NonEnt Aggregated 0.00 -0.31** 0.78** 0.29** 0.32** -0.05*
Ent Trafﬁc 0.02 -0.11** 0.30** 0.10** 0.10** -0.01
Ent Burglary 0.01 -0.01 0.08 0.03 0.01 0.06
Ent Disturbance 0.01 -0.12** 0.33** 0.11** 0.10** -0.05*
Ent Assault -0.03 -0.10** 0.33** 0.09** 0.08** -0.02*
Ent Drugs/Vice -0.06* -0.10** 0.17** 0.09** 0.06* -0.02
Ent Theft -0.04 -0.05 0.13* 0.05 0.01 0.06**
Ent Aggregated -0.01 -0.13** 0.35** 0.12** 0.10** -0.03
363J.A. Schafer et al. / Journal of Criminal Justice 38 (2010) 359–367
to the autoregressive parameter except that it represents the affects of
past error that cannot be modeled with the autoregressive parameter.
The iterative process described above can be labor intensive and non-
precise as the analyst visually interprets graphs (correlograms) to
determine the best ﬁt for the data. SPSS 14.0, however, includes a time
series modeler that automates this process by running through a
series of iterations and identifying the best model for the data series.
This modeler is useful as it reduces the level of subject interpretation
associated with identifying the proper parameter estimates (see
Bushway & McDowall 2006) for a discussion on the challenges
associated with establishing causal effects in trend data).
Tables 3–5present the multivariate ﬁndings for the three
geographical areas (citywide, non-entertainment, and entertainment
Each table includes seven different models, one for each of the
crime categories; a description of the model types is located below the
categories. The ARIMA parameter estimates are included when
applicable and OLS indicates a simpliﬁed regression model was
used. As indicated above, these model types were determined through
diagnostic procedures available in the statistical software package
The citywide analysis is presented in Table 3. The discussion will
focus primarily on the relationships between lunar phase and not the
other control variables for reasons of brevity (seven categories of
crime across three geographical areas). Looking across the citywide
analysis, lunar phase was signiﬁcant only for the burglary model and
in the positive direction. This suggested that as lunar phase increased
(i.e., the moon became more full), so did levels of burglary. While the
lunar phase-crime relationship was positive across the other
categories (with the exception of assault and theft) none of these
reached statistical signiﬁcance. The relationships among the control
variables and crime levels largely remained unchanged in the
multivariate models, with the exception of sunset which was
relegated largely insigniﬁcant across the models.
The ﬁndings for the non-entertainment area are detailed in
Table 4. This area includes the entire city of San Antonio minus the
Riverwalk district. Similar to Table 3, these ﬁndings indicated that
lunar phase had no effect on levels of crime and disorder in San
Antonio. The relative impact of the sunset variable was also moderated
in these models; sunset was signiﬁcant in only the burglary and theft
models, exerting a positive relationship in both instances.
Finally, the ﬁndings for the entertainment area are presented in
Table 5. Looking across the seven models, it is apparent that the
seasonal, temporal, and weather-related control variables that
were so prominent in most of the bivariate relationships and the
prior two multivariate tables largely disappear in Table 5. This
suggested that after including the additional control variables and
detrending the data, crime and disorder levels in the entertainment
district were inﬂuenced to a lesser degree by these factors. The
one notable exception, not unexpectedly, was the weekend control
variable. That is, crime and disorder levels were signiﬁcantly greater
in the entertainment area on weekends net all other controls. This
may suggest that socialization patterns that explain the occurrence
of crime and disorder are largely “ﬁxed”regardless of the weather,
temperature, and season. It is also interesting to note that while the
relationships between lunar phase and crime levels were not
signiﬁcant across the models, there was one exception; the rela-
tionship between lunar phase and drugs/vice was signiﬁcant and
negative. This indicates that as lunar phase increased (i.e., the moon
trended toward a more full state), the reported levels of drugs and
Lunar lore is an established aspect of western culture, though the
speciﬁcs of this belief (i.e., how and why the moon might exert an
inﬂuence on human behavior) was imprecise. This study attempted to
determine whether lunar phase inﬂuenced disorderly and criminal
acts as indicated by the volume of incidents reported to the police
each night during a ﬁve year period in a major southwestern US city. If
a presumed lunar effect was supposed to inﬂuence humans through
some form of biosocial mechanism (Lieber, 1978b, 1996) the ﬁndings
reported here offered no real support of this supposition using the
dependent variable at hand. Prior studies were conﬁrmed here; there
was no clear evidence that lunar cycles had more than marginal (and
likely spurious) explanatory power in understanding levels of crime
and disorder. Although popular culture, folk lore, and even certain
occupational lore suggested the “freaks”come out during full moons
(Lieber, 1996; Rotton & Kelly, 1985; Rotton et al., 1986; Vance, 1995),
this phenomenon was not reﬂected in San Antonio police data as used
here. Though a small number of associations were noted between
lunar phase and various aspects of criminal and disorderly conduct,
this was a common element of prior research and could have been a
probable effect of the large number of associations under consider-
ation (i.e., a spurious ﬁnding). A Bonferroni correction to address this
concern could also have been performed to decrease the possibility of
this Type 1 error. The few statistically signiﬁcant associations that
were detected and the impact of other variables other than the lunar
phase suggest that such a procedure, while more accurate, would not
substantively impact these ﬁndings or conclusions.
This study relied on CFS, which captured the quantity of reported
incidents, but did not provide robust insights into qualitative aspects
of police encounters involving crime and disorder. CFS were good data
points as they were immune from ofﬁcer discretion, though not call-
taker discretion. It was common, however, for lunar research to focus
on the volume of select forms of behavior, such as violence and
aggression in prisons and psychiatric facilities (Lieber, 1978a, 1978b;
Pettigrew, 1985; Templer & Veleber, 1980; Weiskott & Tipton, 1975)
or calls to crisis centers (Templer & Veleber, 1980; Weiskott, 1974).
This was likely a partial function of convenience; quantitative/volume
indicators of behavior were far more convenient and inexpensive than
Multivariate analysis - City Level Analysis
Model Type Trafﬁc Burglary Disturbances Assault Drugs/Vice Theft Aggregated
ARIMA(0,0,8) (OLS) ARIMA(0,0,8) ARIMA(7,0,8) (OLS) (OLS) ARIMA(7,0,8)
B StdError B StdError B StdError B StdError B StdError B StdError B StdError
Sunset 0.009 0.006 0.067* 0.009 -0.008* 0.032 -0.003 0.004 0.007* 0.003 0.002* 0.001 0.050 0.046
Winter -1.738* 0.851 4.116* 1.460 -21.871* 4.701 -2.470* 0.656 -1.318* 0.412 -0.140 0.148 -23.848* 6.703
Weekend 11.372* 0.403 10.361* 0.927 126.698* 2.470 15.410* 0.630 2.719* 0.261 0.602* 0.094 190.794* 4.763
Nightly Temp 0.126* 0.024 0.154* 0.050 1.133* 0.139 0.138* 0.020 0.091* 0.014 0.003 0.005 2.002* 0.187
Precip. 1.008* 0.407 6.510* 1.067 -16.697* 2.282 -1.814* 0.362 -1.569* 0.301 -0.133 0.108 -13.978* 2.835
Lunar Phase 0.102 0.670 1.455* 1.194 2.214 3.633 -0.299 0.472 0.023 .0336 -0.006 0.121 4.083 5.144
Constant -5.031 10.434 -80.994* 16.757 139.067* 57.315 13.385 7.985 -7.638 4.722 -2.772 1.699 88.983 83.023
364 J.A. Schafer et al. / Journal of Criminal Justice 38 (2010) 359–367
qualitative indicators, which also suffered from methodological
concerns and limitations.
It is possible that arrest data might be a more accurate repre-
sentation of the quality of calls police handle. Based on this logic,
incidents occurring during times of full moon would be expected
to be more atypical and would result in a greater need for ofﬁcers
to intervene with an arrest. This logic, however, allows for the
introduction of ofﬁcer discretion. Given that lunar effects are
embedded in collective western beliefs (Lieber, 1996; Rotton &
Kelly, 1985; Rotton et al., 1986), affects observed in areas heavily
inﬂuenced by ofﬁcer discretion (such as the decision to arrest, ﬁle an
ofﬁcial report, or make referrals) could suffer from social contagion
(Purpura, 1979; Simon, 1998). If ofﬁcers accept that there are lunar
effects (Rotton et al., 1986; Vance, 1995) they may opt for differential
handling of situations based on lunar phase. In effect, the lunar effect
would become a self-fulﬁlling prophecy; dependent variables would
be inﬂuenced due to acceptance of the lore, rather than due to actual
citizen/offender behavior. For this reason, the initial classiﬁcations of
CFS represent a relatively innocuous indicator of situations brought to
the attention of the police, though they are subject to some social
contagion among 911 call takers.
The ﬁndings of this research, rather than conﬁrming a lunar-crime/
disorder relationship, corroborated earlier literature establishing that
crime varied temporally (Cohn, 1996; Heller & Markland, 1970;
LeBeau & Corcoran, 1990; LeBeau & Langworthy, 1986) and due to
weather (Cheatwood, 1995; Cohn, 1990b; Cohn & Rotton, 2000; Hipp
et al., 2004; Quételet, 1842/1969; Rotton & Cohn, 2003). Weekends
and nightly temperature were consistently and positively associated
with aggregated levels of crime, though there was variation based on
speciﬁc offense categories. Rates of aggregated crime were much
higher on weekends and were higher during periods of warmer
temperatures. During winter months and periods of greater precip-
itation, crime decreased across the city as a whole and within the non-
entertainment areas. Interestingly, winter and precipitation did not
inﬂuence aggregated crime within the entertainment district.
Substantive lunar effects on crime were not found in the data
analyzed here. If any interpretations regarding the few (and possibly
spurious) relationships were offered, a logical starting point might be
rational choice notions. The mechanisms of inﬂuence are likely a
function of some effects on the availability of suitable targets and the
ability of offenders to escape (or afford) adequate opportunities to
conduct their activities. The limited inﬂuence of lunar phase in the
two instances: (1) citywide burglary (positive relationship found
which might support lunar impacts) and (2) entertainment district
drugs/vice (negative relationship found which may be interpreted to
undermine lunar lore) may be suggestive of such perspectives. On the
whole, however, the data and models examined here were not able
to specify insights or details pertaining to these incidents that would
shed light on this contention. Further scholarly inquiry considering the
choices and decisions of these types of offenders might, for example,
question the role of ambient moon light in shaping offending choices
That noted, insistent believers in the existence of a lunar effect
may interpret this study differently and offer an alternative
possibility. One could review this study and contend that while this
lunar cycle research failed to ﬁnd effects on the volume of crime, the
possibility of effects on the varying character or quality of crimes
reported may remain. More precisely, lunar effects may shape not the
volume of crime encountered by police and other service providers,
but rather alter the nature of problems, individuals, and circum-
stances. As the data used in this study were aggregated and
quantitative, they offered limited insight into these possible aspects
of the nature and character of individual incidents and the motiva-
tions and/or rationales of offenders. For the lunar contention to be
adequately tested and explored further study employing greater
methodological rigor (as this study has employed in examining the
Multivariate analysis - Non-Entertainment District Analysis
Model Type Trafﬁc Burglary Disturbances Assault Drugs/Vice Theft Aggregated
ARIMA(0,0,8) (OLS) ARIMA(0,0,8) ARIMA(7,0,8) ARIMA(0,0,14) (OLS) ARIMA(7,0,8)
B StdError B StdError B StdError B StdError B StdError B StdError B StdError
Sunset 0.009 0.005 0.065* 0.009 -0.009 0.030 -0.006 -0.005 0.005 0.003 0.002* 0.001 0.044 0.045
Winter -1.547 0.799 3.956* 1.401 -20.723* 4.515 -2.341* 0.619 -1.076* 0.467 -0.080 0.141 -22.445* 6.519
Weekend 10.39* 0.381 9.878* 0.889 121.390* 2.376 0.351* 0.188 2.997* 0.329 0.462* 0.090 174.976* 5.079
Nightly Temp 0.119* 0.023 0.157* 0.048 1.100* 0.133 8.248* 0.909 0.099* 0.015 0.003 0.005 1.934* 0.180
Precip. 1.053* 0.39 6.149* 1.024 -15.588* 2.197 0.138* 0.019 -1.393* 0.273 -0.157 0.103 -13.147* 2.716
Lunar Phase 0.018 0.628 1.355 1.145 1.970 3.487 -1.684 0.334 0.206 0.295 0.021 0.116 4.018 4.975
Constant -4.848 9.759 -79.468* 16.069 134.465* 55.044 17.015* 8.286 -6.004 5.684 -2.931 1.623 89.851 81.004
Multivariate analysis - Entertainment District Analysis
Model Type Trafﬁc Burglary Disturbances Assault Drugs/Vice Theft Aggregated
ARIMA(4,0,7) (OLS) ARIMA(7,0,7) ARIMA(0,0,14) ARIMA(5,0,9) (OLS) ARIMA(7,1,13)
B StdError B StdError B StdError B StdError B StdError B StdError B StdError
Sunset 0.001 0.001 0.002 0.001 0.009 0.009 0.002 0.005 0.001 0.001 0.001 0.001 0.024 0.020
Winter -0.194 0.155 0.160 0.212 -1.038 0.713 -0.244 0.769 -0.202 0.155 -0.125 0.091 -1.636 1.644
Weekend 0.995* 0.078 0.483* 0.134 4.785* 0.578 4.033* 0.288 0.495* 0.065 0.328* 0.058 9.411* 1.223
Nightly Temp 0.004 0.005 -0.003 0.007 0.059* 0.017 0.032 0.019 0.003 0.004 -0.005 0.003 0.137* 0.037
Precip. -0.055 0.081 0.361* 0.155 -1.051* 0.248 -0.228 0.297 -0.093 0.070 0.185* 0.066 -0.776 0.510
Lunar Phase 0.087 0.104 0.099 0.173 .0256 0.328 -0.346 0.561 -0.206* 0.100 -0.119 0.074 -0.280 0.804
Constant -0.858 2.470 -1.526 2.429 -9.690 16.345 -2.819 9.958 -0.614 2.582 -0.646 1.043 -32.233 37.118
365J.A. Schafer et al. / Journal of Criminal Justice 38 (2010) 359–367
volume of crime) would be necessary in an effort to examine the
contours of any possible relationship between the quality of crimes
reported and lunar cycles. Only then could this question of the lunar
lore and crime be more fully answered.
1. See Rotton and Cohn (2000) and Cohn (1990b), among others, for a discussion
of theoretical and empirical relationships between weather, temporal variables, social
contact variables, and crime.
2. Block (1984) wrote that although some general seasonal trends could be noted,
they tended to be offense-speciﬁc and were not evident in all jurisdictions. Her
comprehensive review of data she analyzed, as well as various published and
unpublished research done by others, found ample variation in seasonal trends. Some
of this might have been attributed to variable data, operational deﬁnitions, and
3. Researchers have employed myriad methods of structuring the timing and
intensity of full moons relative to other periods of the lunar cycle. For example:
contrasting the day of the full moon with all other days in the cycle (Frey et al., 1979);
creating seemingly-capricious “windows”of time (i.e., three days) around the full
moon for contrast with the rest of the cycle (sometimes separating out windows of
time around the new moon) (Walters et al., 1975); breaking cycles into equal (Frey et
al., 1979; Quinsey & Varney, 1977; Taylor & Diespecker, 1972) and unequal (Frey et al.,
1979; Lester et al., 1969) periods of times; and, allowing lunar phase to vary from day
to day throughout the lunar cycle (Simon, 1998). The lunar lore is ambiguous
regarding whether the inﬂuence of the moon is truly restricted to full-moon periods
and whether a heightened full moon inﬂuence might be offset by a converse effect
during the new moon.
4. This consideration was another complexity created by the vague nature of the
lunar lore. Was the effect of the full moon mediated by its visibility due to daylight
and/or weather patterns? If the effect was thought to result from the actual intensity
of the moon's illumination, the answer may have been “yes.”If the effect was thought
to result from the moons ‘pull’on water in the human body, the answer might have
5. An anonymous reviewer pointed out the U.S.N.O. also reports moon rise and
setting times within archival data. It was noted using a sliding window of analysis
based on whether the moon was above the horizon might have been an alternative
approach within the analysis. Additional analyses not reported here, but available from
the lead author upon request, suggest the results were not substantively different with
the use of such a variable window of analysis. The original models (ﬁxed 8:00 pm –
3:00 am window of analysis) were retained for the analysis reported in this article,
though the alternative scheme was included in later portions of the discussion.
6. The analysis was also conducted using a measure of the moon's actual visibility
above the horizon net of atmospheric conditions that might preclude such visibility.
These models yielded no signiﬁcant differences in parameter magnitudes or
signiﬁcance levels. Since the same pattern of signiﬁcant and non-signiﬁcant results
was obtained, the initial results were retained and presented in this article. Those
seeking further information pertaining to these alternative modeling efforts can direct
inquires to the second author.
Atlas, R. (1984). Violence in prison: Environmental inﬂuences. Environment and
Bauer, S. F., & Hornick, E. J. (1968). Lunar effect on mental illness: The relationship of
moon phase to psychiatric emergencies. American Journal of Psychiatry,125,
Bickis, M., Kelly, I. W., & Byrnes, G. F. (1995). Crisis calls and temporal and lunar
variables: A comprehensive examination. Journal of Psychology,129, 701−711.
Biermann, T., Estel, D., Sperling, W., Bleich, S., Kornhuber, J., & Reulbach, U. (2005).
Inﬂuence of lunar phase on suicide: The end of a myth? A population-based study.
Chronobiology International,22, 1137−1143.
Blackmon, S., & Catalina, D. (1973). The moon and the emergency room. Perceptual and
Motor Skills,37, 624−626.
Block, C. R. (1983). How to handle seasonality: Introduction to the detection and analysis
of seasonal ﬂuctuation in criminal justice time series. Chicago: Illinois Criminal Justice
Block, C. R. (1984). Is crime seasonal? Chicago: Illinois Criminal Justice Information
Box, G. E. P., & Jenkins, G. M. (1976). Time series analysis: Forecasting and control
(Rev. ed.). San Francisco: Holden-Day.
Brantingham, P., & Brantingham, P. (1995). Criminality of place: Crime generators and
crime attractors. European Journal on Criminal Policy and Research,3(3), 5−26.
Bushway, S. D., & McDowall, D. (2006). Here we go again- Can we learn anything from
aggregate-level studies of policy interventions? Criminology & Public Policy,5,
Campbell, D. E., & Beets, J. L. (1978). Lunacy and the moon. Psychological Bulletin,85,
Chapman, S., & Morrell, S. (2000). Barking mad? Another lunatic hypothesis bites the
dust. British Medical Journal,321, 1561−1563.
Cheatwood, D. (1988). Is there a season for homicide? Criminology,26, 287−306.
Cheatwood, D. (1995). The effects of weather on homicide. Journal of Quantitative
Cohen, L. E., & Felson, M. (1979). Social change and crime rate trends: A routine activity
approach. American Sociological Review,44, 588−608.
Cohn, E. G. (1990a). Weather and crime. British Journal of Criminology,30,51−64.
Cohn, E. G. (1990b). Weather and violent crime: A reply to Perry and Simpson, 1987.
Environment and Behavior,22, 280−294.
Cohn, E. G. (1993). The prediction of police calls for service: The inﬂuence of weather
and temporal variables on rape and domestic violence. Journal of Environmental
Cohn, E. G. (1996). The effect of weather and temporal variations on calls for police
service. American Journal of Police,15(1), 23−43.
Cohn, E. G., & Rotton, J. (2000). Weather, seasonal trends and property crimes in
Minneapolis, 1987-1988: A moderator-variable time-series analysis of routine
activities. Journal of Environmental Psychology,20, 257−272.
Culver, R., Rotton, J., & Kelly, I. W. (1988). Geophysical variables and behavior: XLIX.
Moon mechanisms and myths: A critical appraisal of explanations of purported
lunar effects on human behavior. Psychological Reports,62, 683−710.
Cyr, J. J.,& Kaplan, R. A. (1987).Geophysical variables andbehavior: XLVI. Thelunar-lunacy
relationship: A poorly evaluated hypothesis. Psychological Reports,61,391−400.
Cyr, J. J., & Kaplan, R. A. (1988). Investigating the lunar-lunacy relationship: A reply to
Rotton and Kelly. Psychological Reports,62, 319−322.
Durm, M. W., Terry, C. L., & Hammonds, C. R. (1986). Lunar phase and acting-out
behavior. Psychological Reports,59, 987−990.
Frey, J., Rotton, J., & Barry, T. (1979). The effects of the full moon on human behavior:
Yet another failure to replicate. The Journal of Psychology,103, 159−162.
Heller,N.B.,&Markland,R.E.(1970).Aclimatological model for forecasting the
demand for police serv ice. Journal of Research in Crime & Delinquency,7,
Hipp, J. R., Bauer, D. J., Curran, P. J., & Bollen, K. A. (2004). Crimes of opportunity or
crimes of emotion? Testing two explanations of seasonal change in crime. Social
Kelly, I. W., & Rotton, J. (1983). Geophysical variables and behavior: XIII. Comment on
“lunar phase and accident injuries”: The dark side of the moon and lunar research.
Perceptual and Motor Skills,57, 919−921.
Laverty, W. H., & Kelly, I. W. (1998). Cyclical calendar and lunar patterns in automobile
property accidents and injury accidents. Perceptual and Motor Skills,86, 299−302.
LeBeau, J. L., & Corcoran, W. T. (1990). Changes in calls for police service with changes in
routine activities and the arrival and passage of weather fronts. Journal of
Quantitative Criminology,6, 269−291.
LeBeau, J.L., & Langworthy, R. H. (1986). The linkages between routine activities,weather,
and calls f or police service s. Journal of Police Science and Administration,14,137−145.
Lester, D., Brockopp, G. W., & Priebe, K. (1969). Association between a full moon and
completed suicide. Psychological Reports,25, 598.
Lieber, A. L. (1978a). Human aggression and the lunar synodic cycle. Journal of Clinical
Lieber, A. L. (1978b). The lunar effect: Biological tides and human emotions. Garden City,
NY: Anchor Press.
Lieber, A. L. (1996). How the moon affects you. Mamaroneck, NY: Hastings House.
Lieber, A. L., & Sherin, C. R. (1972). Homicides and lunar cycle: Toward a theory of lunar
inﬂuence on human emotional disturbance. American Journal of Psychiatry,129,
McDowall, E., McCleary, R., Meidinger, E. E., & Hay, R. A., Jr. (1980). Interrupted time
series analysis: Vol. 2.1 Quantitative applications in the social sciences. Newbury Park,
Merriam-Webster online dictionary. (2008). Retrieved from http://www.m-w.com
Pettigrew, C. G. (1985). Seasonality and lunacy of inmate behavior. The Southern
Pokorny, A. D., & Jachimczyk, J. (1974). The questionable relationship between
homicides and the lunar cycle. American Journal of Psychiatry,131,827−829.
Purpura, P. P. (1979). Police activity and the full moon. Journal of Police Science and
Quételet, L. A. J. (1969). A treatise on man and the development of his faculties.
Gainesville, FL: Scholar's Facsimiles and Reprints. (Original work published 1842).
Quinsey, V. L., & Varney, G. W. (1977). Characteristics of assaults and assaulters in a
maximum security psychiatric unit. Crime and Justice,5, 212−220.
Robinson, J. B., & Rengert, G. F. (2006). Illegal drug markets: The geographic perspective
and crime propensity. Western Criminology Review,7(1), 20−32.
Rotton, J., & Cohn, E. G. (2000). Weather, disorderly conduct, and assault: From social
contact to social avoidance. Environment and Behavior,32, 651−673.
Rotton, J., & Cohn, E. G. (2003). Global warming and U.S. crime rates: An application of
routine activity theory. Environment and Behavior,35, 802−825.
Rotton, J., & Kelly, I. W. (1985). A scale for assessing belief in lunar effects: Reliability
and concurrent validity. Psychological Reports,57, 239−245.
Rotton, J., Kelly, I. W., & Elortegui, P. (1986). Assessing belief in lunar effects: Known-
group validation. Psychological Reports,59, 171−174.
Rotton, J., Kelly, I. W., & Frey, J. (1983). Geophysical variables and behavior: X. Detecting
lunar periodicities: Something old, new, borrowed, and true. Psychological Reports,
Russell, G. W., & de Graaf, J. P. (1985). Lunar cycles and human aggression: A replication.
Social Behavior and Personality,13, 143−146.
Russell, G. W., & Dua, M. (1983). Lunar inﬂuences on human aggression. Social Behavior
366 J.A. Schafer et al. / Journal of Criminal Justice 38 (2010) 359–367
Shapiro, J. L., Streiner, D. L., Gray, A. L., Williams, N. L., & Soble, C. (1970). The moon and
mental illness: A failure to conﬁrm the Transylvania effect. Perceptual and Motor
Simon, A. (1998). Aggression in a prison setting as a function of lunar phases.
Psychological Reports,82, 747−752.
Taylor, L. J., & Diespecker, D. D. (1972). Moon phase and suicide attempts in Australia.
Psychological Reports,31, 110.
Templer, D. I., & Veleber, D. M. (1980). The moon and madness: A comprehensive
perspective. Journal of Clinical Psychology,36, 865−868.
Templer, D. I., Veleber, D. M., & Brooner, R. K. (1982). Geophysical variables and
behavior: VI. Lunar phase and accident injuries: A difference between night and
day. Perceptual and Motor Skills,55, 280−282.
Vance, D. E. (1995). Belief in lunar effects on human behavior. Psychological Reports,76,
Walters, E., Markley, R. P., & Tiffany, D. W. (1975). Lunacy: A type I error? Journal of
Abnormal Psychology,84, 715−717.
Weiskott, G. N. (1974). Moon phases and telephone counseling calls. Psychological
Weiskott, G. N., & Tipton, G. B. (1975). Moon phases and state hospital admissions.
Psychological Reports,37, 486.
Wilson, J. E., II, & Tobacyk, J. J. (1990). Lunar phases and crisis center telephone calls.
Journal of Social Psychology,130,47−51.
367J.A. Schafer et al. / Journal of Criminal Justice 38 (2010) 359–367