Content uploaded by Paul Wesley Schultz
Author content
All content in this area was uploaded by Paul Wesley Schultz on Aug 25, 2015
Content may be subject to copyright.
Environment and Behavior
45(1) 35 –59
© 2013 SAGE Publications
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0013916511412179
http://eab.sagepub.com
412179EAB45110.1177/00139165114121
79Schultz et al.Environment and Behavior
© 2013 SAGE Publications
Reprints and permission:
sagepub.com/journalsPermissions.nav
1California State University, San Marcos, USA
2Action Research, Inc., CA, USA
3SUNY Plattsburgh, NY, USA
Corresponding Author:
P. Wesley Schultz, California State University, 333 South Twin Oaks Valley Road, San Marcos,
CA 92096, USA
Email: wschultz@csusm.edu
Littering in Context:
Personal and
Environmental Predictors
of Littering Behavior
P. Wesley Schultz1, Renée J. Bator3,
Lori Brown Large2, Coral M. Bruni2,
and Jennifer J. Tabanico2
Abstract
This article reports the results from a large-scale study of littering behav-
ior. Findings are reported from coded observations of the littering behavior
among 9,757 individuals at 130 outdoor public locations in the United States.
The focus was on littering behavior of any item, but a separate sample is
also reported on the littering behavior of only smokers. For smokers, the
observed littering rate for cigarette butts was 65%. Results from the general
littering observations showed that of all the disposal behaviors observed,
17% resulted in litter. Statistical analyses using multilevel modeling showed
that age (negatively) was predictive of individual littering. At the level of the
site, the presence of existing litter (positively) and the availability of trash
receptacles (negatively) predicted littering. Supplemental analyses showed
that among individuals who disposed of an item, distance to the receptacle
was positively predictive of littering. Implications for litter prevention strate-
gies are discussed.
Article
36 Environment and Behavior 45(1)
Keywords
Littering, field study, behavioral observation
Littering in Context: The Personal and
Environmental Determinants of Littering
Behavior
Litter is any piece of misplaced solid waste (Geller, 1980). This can range
from small items, such as cigarette butts or candy wrappers, to abandoned
automobiles, appliances, and even spacecraft. Most commonly, litter refers
to items that are discarded by an individual, but it can include any item that
is in an unacceptable location, regardless of the origin. This could not only
include the candy wrapper dropped on the ground but also the newspaper that
blows out of a trash can. The distinction here is between litter (the item) and
littering (the behavior). Although the exact percentage of litter attributed to
improper disposal behavior by individuals is unknown, there is evidence to
suggest that a large majority of litter is linked with individual disposals
(MSW Consultants, 2009). A recent analysis of the sources of litter along
roadsides attributed 70% to individuals (52% to motorists and 18% to pedes-
trians). In comparison, 21% came from unsecured loads, 5% from the vehi-
cles themselves (e.g., tires and vehicle debris), and 3% came from unsecured
containers in the nearby vicinity. Similarly, at transition points such as bus
stops, 88% of the small littered items were attributed to individuals, as was
90% of large items (69% to pedestrians and 21% to motorists). These find-
ings underscore the importance of the individual as a source of litter.
Litter poses a number of important environmental, social, and aesthetic
problems. As an environmental problem, litter is a substantial source of con-
tamination. Misplaced plastics, Styrofoam, paper, glass, and many other
commonly used consumer materials accumulate in the environment, posing a
number of harmful environmental consequences. The social problems related
to litter include safety hazards, fire hazards, human health hazards, and indi-
rect health hazards from bacteria, rats, roaches, and mosquitoes that are
attracted to litter. In addition, litter is predictive of changing crime rates in a
community (Brown, Perkins, & Brown, 2004), and there is experimental evi-
dence showing that the presence of litter results in an increase in other social
transgressions like theft (Keizer, Lindenberg, & Steg, 2008). There are aes-
thetic issues with litter, as there is near unanimous agreement that litter is
unsightly (Pandey, 1990). Indeed, the presence of litter in a residential com-
munity decreases property value, and litter in commercial areas reduces sales
Schultz et al. 37
and attracts fewer customers (National Association of Home Builders, 2009;
Skogan, 1990). Finally, there are the direct costs of litter cleanup, which con-
servatively tops US$11 billion annually in the United States (MSW
Consultants, 2009).
Given the myriad of problems that result from litter, it is not surprising
that a sizable amount of research has focused on understanding and prevent-
ing it. Litter was one of the first environmental problems to lend itself to
systematic behavioral research, with studies going back more than 40 years.
In an early 1968 study, Keep America Beautiful (KAB) reported on the
attitudes, beliefs, and self-reported behaviors among a large national sample
(Public Opinion Surveys, Inc., 1968). Subsequently, studies throughout the
1970s were used as a basis for creating litter prevention programs (Burgess,
Clark, & Hendee, 1971; Cone & Hayes, 1980; Geller, Winett, & Everett,
1982). In the section below, the three dominant approaches to understanding
litter and littering behavior are summarized.
Prior Studies of Litter
Who litters? One approach to understanding littering focuses on the demo-
graphic and personal qualities of the type of person who litters—the “litter
bug.” Although much of these data come from surveys in which people self-
report littering rates, a few studies have conducted observations (e.g., distrib-
uting a marked flyer or handbill under varying conditions and monitoring to
see which accumulate as litter). The widely accepted conclusions from these
studies are that littering is more common among males, younger adults, and
individuals living in rural communities more than cities. However, the
research results on these characteristics of the “litter bug” are far from con-
clusive and many studies have failed to find significant demographic predic-
tors (Beck, 2007; Finnie, 1973; Geller, Witmer, & Tuso, 1977). As a result,
there is little consistent evidence for demographic characteristics of the “litter
bug.”
How often do people litter? Given the volume of litter that accumulates
nationally and worldwide, it is important to understand the littering behavior
of individuals. One way to address this question is by watching the behavior
of individuals in public spaces (Geller et al., 1977; Heberlein, 1971). Although
only a handful of studies have utilized observational methods, the results are
instructive. An early study by Finnie (1973) reported observations of indi-
viduals in four outdoor spaces in Philadelphia as they ate hot dogs purchased
from street vendors. Of the 272 observed individuals, 91 littered the wrapper
(33%). Littering was more common in sites that were already littered and in
38 Environment and Behavior 45(1)
sites without trash receptacles. Similarly, Cialdini, Kallgren, and Reno (1991)
and Cialdini, Reno, and Kallgren (1990) placed flyers on the windshields of
parked cars and observed the percentage of individuals who littered. In one
illustrative finding, they found that 14% of the individuals littered when the
environment was litter free, whereas 32% littered into an already-littered
environment. In an interesting extension of these findings, Keizer et al.
(2008) found that participants were more likely to litter into “disordered” set-
tings (those with graffiti or fireworks or shopping carts left unreturned).
These findings illustrate the importance of understanding the role of the
physical context in facilitating or discouraging littering behavior, and similar
results have been reported in other studies (Williams, Curnow, & Streker,
1997).
Collected litter. By far, the most commonly used method for litter research
is to count and characterize the types of litter collected from different loca-
tions (KAB, 2007). Litter cleanups happen on a regular basis, including the
KAB (2007) Great American Cleanup, regular Adopt-a-Highway cleanups,
and the Ocean Conservancy’s International Coastal Cleanup. In addition,
states and local governments regularly conduct “litter surveys” to identify the
types and sources of materials found along roadways throughout the country.
These events remove millions of pounds of litter annually from roadways,
parks, shorelines, and natural areas worldwide. In the 2007 Coastal Cleanup,
the Ocean Conservancy collected 6 million pounds of materials, including
cigarette butts (1,971,551 or 27% of all collected items), food wrappers (10%
of collected items), caps and lids (9%), bags (8%), plastic beverage bottles
(7%), plastic utensils (5%), and glass beverage bottles (5%; Ocean Conser-
vancy, 2007).
The current study involved making unobtrusive observations of disposal
behavior of pedestrians at outdoor sites in which we simultaneously exam-
ined demographic characteristics of the participants as well as contextual
variables such as the presence, characteristics, and placement of receptacles.
As an extension of prior littering studies, the current work examined both
person-level and context-level predictors of observed littering behavior using
a multilevel modeling framework.
Current Project
Although there is a long history of research on litter and littering, a number
of fundamental questions remain to be answered. In the current article, the
results from a nationwide study of littering behavior are reported. This
research investigation had three goals: (a) to conduct an observational study
Schultz et al. 39
of littering behavior across a diverse sample of sites and locations; (b) to
develop a set of observational methodologies for observing littering (includ-
ing a modified protocol for observing smokers) that could be replicated over
time and in different locations; and (c) to utilize a multilevel approach in a
way that would allow for the simultaneous analysis of personal- and contex-
tual-level determinants of littering. At the level of the individual, we exam-
ined the effects of variables found to predict littering in past research: gender
(males littering more than females), age (younger littering more than older),
and distance from a receptacle (greater distance at the time of disposal pre-
dicting higher littering rates). We also explored new potential predictors,
including time of day, and whether being in a group might be associated with
lower littering rates because of social disapproval. At the level of the context,
past research led us to expect that littering would occur more often: in sites
that were high in existing litter, in sites with fewer receptacles, and in sites
with no existing signage about littering. We also explored several less widely
studied variables, including rural versus urban locations, cleanliness, land-
scaping, infrastructure, and the number of people within the location.
Method
Sites and Participants
During the spring of 2008, systematic observations of individuals were con-
ducted in a wide range of outdoor public locations across the United States.
The research design was developed as a multilevel model, with random
samples of individuals “nested” within site (see Raudenbush & Bryk, 2002).
At each location, random samples of individuals were selected, and their
behavior was unobtrusively monitored as they moved through the site. A
modified protocol was developed for monitoring the behavior of smokers,
which included the various means by which smokers typically dispose of
their butts.
Observations were conducted in 10 states (Arkansas, California, Georgia,
Illinois, Kentucky, Nevada, New Mexico, New York, Utah, and Vermont),
selected to represent a variety of regions across the country. Within each
state, an urban, rural, and suburban city was selected using U.S. Census sta-
tistics. Finally, within each of those cities, specific observation sites were
randomly selected from a list of all possible sites of each type: city center,
fast food, recreation, gas station, and rest stop. Three additional site types
were selected for observations of cigarette smokers: medical, bars/restaurant,
and retail.
40 Environment and Behavior 45(1)
The final data set included observations of 9,757 individuals from 130
locations: 86 general litter and 44 focused on cigarette disposal. Of these, 30
were recreational, 24 city center, 22 fast food, 12 retail, 12 bars/restaurants, 11
gas stations with convenience stores, 11 rest stops, and 8 medical facilities.
Procedure
Systematic observations were made by pairs of observers following a strict
protocol that was developed after considerable training. The protocol and
code sheets are available on request from the authors. On arrival at the
research site, the field team first defined the physical boundary of the obser-
vation area. This would be an area that the team could clearly observe from
an unobtrusive lookout point (e.g., in a public seating area or inside a parked
car). This lookout point was typically at the back border of the observation
area. This enabled the research team to remain forward facing (which was
necessary when they were inside a parked car) and increased their ability to
remain unobtrusive. These boundaries were typically areas of about 2,000
square feet that allowed for unobstructed and unobtrusive observations of
individuals. This observation area always included the most heavily traf-
ficked part of the site (e.g., the entrance/exits to the nearest building). Before
observing any participants, the research team used a detailed codebook to
record a variety of characteristics related to the setting.
Setting characteristics. The codebook provided a variety of categorical and
continuous measures for the research team to record for each setting. The
research team specified the site type—recreational, city center, fast food, retail,
bar/restaurant, gas station with convenience store, rest stop, or medical facility
(categorical measure); identified whether the location was categorized as rural,
urban, or suburban (categorical measure); recorded the time of day as before
noon, afternoon, or after 4:00 p.m. (categorical measure); rated the amount of
existing litter in the location from 0 = not at all littered to 10 = extremely lit-
tered (continuous measure); indicated whether or not each of 9 different types
of litter (e.g., paper, food wrappers, cans, bottles, etc.) was present or absent
(categorical); rated the amount of cigarette butt litter present from 0 = not at all
littered to 10 = extremely littered (continuous); counted the number of cigarette
butts in the observational area (continuous); rated the overall cleanliness of the
site from 0 = not at all clean to 10 = extremely clean (operationalized as free
from bad smells, litter, unkempt infrastructure, and objects that do not belong
in the location; continuous); judged the landscaping (operationalized as the
presence and care of foliage; continuous from 0 = not at all landscaped to 10 =
extremely landscaped); and rated the overall infrastructure from 0 = low
Schultz et al. 41
infrastructure to 10 = high infrastructure (operationalized as the placement of
physical objects within a location as a means to increase the aesthetics, walk-
ability, cleanliness, and landscaping of the area). This included planters, paved
walkways, benches, and trash receptacles (continuous). The research team also
recorded the number of trash receptacles for each of five receptacle types: trash
can, ashtray, ash/trash combination, dumpster, and recycling (continuous). For
analytic purposes, these were summed to produce a single score of the total
number of available receptacles. The team recorded whether or not there was
littering signage present (dichotomous, 1 = yes, 0 = no). Finally, they rated the
crowdedness of the location from 0 = not at all crowded to 10 = extremely
crowded, operationalized as the inability to move freely. It was defined for the
observational team as the combination of the number of people in the location,
given the features of the location (continuous). These measures of the setting
were made to examine the impact of contextual variables on participants’ litter-
ing behavior.
Participant characteristics. After recording the details of the setting, the
research team randomly selected a participant by taking the Nth person to
enter the space, with N determined using the crowdedness of the location and
ranged from 1 to 6. Random selection of the individual participant at each site
is a key aspect of this research protocol, and it provides data that can be used
to calculate a littering rate for each site as well as the data needed to analyze
the personal and contextual-level predictors of littering.
The research team recorded each participant’s gender, approximate age,
whether the selected participant was alone or with one or more others, and noted
one of three possible disposal options: the participant did not have an item to
dispose, the participant had an item to dispose but left the site carrying the object,
or the participant disposed of the object. No other observations were made for
participants who had no object to dispose or who left the site with the object.
The research team made additional recordings for only those participants
who disposed of an object. They recorded whether the object was disposed of
properly or improperly. Proper disposal was operationalized as any disposal
that resulted in the object being placed in a receptacle, including ashtray,
trash can, or recycling bin. Items placed in the wrong receptacle (e.g., trash in
a recycling container or cigarette butts in a trash can) were not coded as litter.
For analytic purposes, these disposals were coded as “proper.” Also coded as
proper were pocketing the item or handing it to another person. Improper
disposals included disposals on the ground, planters, bushes or shrubbery, or
disposals on or around receptacles. The research team also recorded the type
of object disposed using a code sheet with 13 options, including an open-
ended option for “other.”
42 Environment and Behavior 45(1)
Of those who were observed to have littered, the researchers recorded the
person’s intent to litter using eight categories, drawing on prior work by
Williams et al. (1997): drop without intent, drop with intent, flick, shoot and
miss, inch away, wedge, sweep, or 90%. All but the first coded category were
classified as “with intent.” Drop with intent was a subjective classification
made by the coder and required one of two specific actions: the individual
visually inspected the item either at the point of disposal or immediately fol-
lowing, or there was an observable hand movement indicating an intentional
discard (e.g., flick, toss, fling, wedge, and sweep). The disposal strategies of
drop, flick, and shoot and miss involved the intentional placement of the item
in an improper location; sweep strategies involved brushing items from a flat
surface unto the ground; and “90%” codes included instances where the indi-
vidual collected other items for proper disposal but intentionally left one or
more objects behind. More detailed descriptions of these eight intentional
disposal strategies can be found in Williams et al.
Finally, the research team recorded littering participants’ distance (in feet)
from receptacles at the time they littered. Any discrepancies were resolved
through discussion. Observations within each site continued until 30 partici-
pants were observed making a disposal (proper or improper) or until the con-
clusion of an 8-hour observation period. A minimum of 4 hours of observations
were conducted at each site.
Results
Reliability Measures
During training, the research team conducted multiple sessions during which
they practiced coding both the settings and individuals. These training ses-
sions were conducted until the pairs of raters achieved a minimum of 80%
agreement for the categorical variables of the setting and r = .70 for continu-
ous variables of the setting. The reliabilities reported below are the average
percentage agreements and correlations across pairs of team members on the
final day of training:
•Did the person have an item for disposal in his or her hand (dichoto-
mous, 1 = yes or 0 = no)? Percentage of agreement, 93%.
•Did the person dispose of an item while within the observational bound-
ary (dichotomous, 1 = yes or 0 = no)? Percentage agreement, 95%.
•Did the disposal result in litter (dichotomous, 1 = yes or 0 = no)? Percent-
age of agreement, 97%. This variable served as the primary outcome.
Schultz et al. 43
•Type of item disposed? (categorical: 13 options plus “other”). Per-
centage of agreement excluding “other,” 97%.
•For individuals who were observed littering, was there clear intent
(dichotomous, 1 = yes or 0 = no)? Percentage of agreement, 98%.
•For individuals who were observed littering, what was the distance
to the closest receptacle at the point of littering, in feet (r = .94).
During actual data collection, additional reliabilities were computed using
data from 127 observed individuals at three sites obtained from three pairs of
the field team. The following percentage agreement was found: gender (95%
agreement), approximate age (r = .94), whether the individual was alone
(96% agreement), and time of day (100% agreement).
The data were analyzed as a multilevel model, which allowed for both
individual- and context-level predictors of littering behavior (Hox, 2002;
Raudenbush & Bryk, 2002). The summary below begins with basic descrip-
tive statistics from the observations and then proceeds to report the results
from the hierarchical linear model. Results from the multilevel models are
reported following convention, with β representing unstandardized Level-1
coefficients (in these analyses, person-level predictors) and γ representing
unstandardized Level-2 coefficients (in these analyses, context-level predic-
tors). Measures of variability are also reported, with σ representing variance
at Level 1 (person-level) and τ for variance at Level 2.
Descriptive Statistics
Across the 130 sites, 118 of them (91%) had at least one trash receptacle.
These included 64 sites with uncovered trash cans, 58 sites with lidded trash
cans, 43 sites with ash receptacles, 16 sites with recycling bins, 18 sites with
combined trash can/ash receptacles, and 12 sites with dumpsters. Many of
the sites had several types of receptacles, so the total exceeds 130. Of the 130
sites, only 2 had no visible litter within the observation boundary. By count,
the most frequently observed visible litter included cigarette butts and mis-
cellaneous paper. The number of sites with various types of litter is shown in
Table 1. “Other” items included diapers, dog waste, fishing gear, clothing,
and children’s toys. These findings indicate that although trash receptacles
are quite common in public spaces, ash receptacles, and (particularly) recy-
cling bins are less common.
Observations of littering in general were made at 86 sites across 10 states.
A total of 8,990 general observations were made; an additional sample of 767
smokers is reported separately below. The general observations were evenly
44 Environment and Behavior 45(1)
divided across rural (33%), suburban (34%), and urban areas (33%).
Observations were made throughout the day, with 27% made in the morning
before noon, 58% in the afternoons between noon and 4:00p.m., and 16% in
the evening after 4:00p.m. Of the observations, 56% of the observed targets
were male, and 44% were female. Observed ages ranged from 1 to 82 (M =
38, SD = 16), and 50% of the observed individuals were alone.
Of the 8,990 people who were observed, 2,472 left the site with no object
for disposal (28%), 4,534 left the site with an object (50%), and 1,962 dis-
posed of an object while on site (22%). Among these disposals, there were
342 instances of littering observed. That is, of all 8,990 individuals that were
observed moving through a diverse range of sites, 4% littered. In addition, of
all the disposal behaviors that were observed (N = 1,962), 342 (or 17%) were
improperly disposed by littering. The remaining proper disposals included
trash receptacle (60%), pocketing the item (9%), handing the item to another
person (6%), ashtray (6%), and recycling bin (1%).
Table 1. Percentage of Sites (out of 130) With that Litter Type Present and Percent
of Participants who Disposed of that Item Improperly
Litter type
% sites with this type of
litter present
% participants who
disposed of item
improperly
Cigarette butts 82 (fo = 106) 57 (n = 194)
Paper 67 (fo = 87) 7 (n = 20)
Food wrappers 45 (fo = 58) 14 (n = 14)
Confections 34 (fo = 44) 0 (n = 0)
Napkin/tissue 34 (fo = 44) 8 (n = 9)
Miscellaneous plastic 33 (fo = 43) 0 (n = 0)
Food remnants 24 (fo = 31) 20 (n = 16)
Beverage cup 16 (fo = 21) 3 (n = 5)
Beverage bottle: plastic 11 (fo = 14) 5 (n = 5)
Food containers 9 (fo = 12) 2 (n = 1)
Plastic bags 8 (fo = 11) 5 (n = 2)
Beverage can 6 (fo = 8) 12 (n = 8)
Beverage bottle: glass 5 (fo = 6) 0 (n = 0)
Yard waste 5 (fo = 6) 0 (n = 0)
Combination/mixed trash 0 (fo = 0) 4 (n = 12)
Other 27 (fo = 35) 37 (n = 46)
Unknown 0 (fo = 0) 8 (n = 10)
Schultz et al. 45
Of the 1,962 coded disposals, the most frequent were cigarettes (N = 340),
mixed trash (N = 337), and paper (N = 272). Table 1 also shows the types,
frequencies, and littering rates for the disposed objects. The table shows the
frequency of proper and improper disposals, along with the percentage of
each type of material that was littered (computed as improper / proper +
improper). The “other” category includes a number of low-frequency dispos-
als, including pet waste, candy and other confections, matches and cigarette
lighters, diapers, straws, chewing tobacco, and miscellaneous product pack-
aging like price tags, foil wrappers, and twist ties.
The 342 acts of littering were coded into discrete disposal strategies, along
with coded intent. The most frequent littering strategy was to drop with intent
(N = 183, 54%). That is, the person committed a clear and deliberate act of
littering. Other litter strategies included flick (N = 68, 20%) and drop without
intent (N = 42, 12%). The behaviors were also coded into littering strategies
found in prior research (Williams et al., 1997): inch away (N = 8), shoot and
miss (N = 8), wedge (N = 4), sweep (N = 3), and 90% (N = 2). When com-
bined, an estimated 81% of observed littering occurred with intent.
The observation team coded the distance (in feet) from the disposer to the
nearest receptacle (trash, recycling, or ashtray). Although there were several
instances of littering that occurred immediately adjacent to a receptacle, most
littering occurred at a considerable distance (mean distance to a receptacle at
time of littering was 29 feet).
Multilevel Modeling
Finally, a series of statistical analyses were conducted to examine the indi-
vidual and contextual variables that were predictive of littering. The analysis
was conducted using only data from observations where a disposal (either
proper or improper) occurred (N = 1962). Multilevel modeling is a statistical
technique that allows for “nested” data structures (in this case, individuals
nested within site). In addition, the multilevel approach does not require bal-
anced data (i.e., equal numbers of observations per site) and instead utilizes
all available information to estimate the underlying effects. The approach
makes it possible to simultaneously model individual-level and contextual-
level variables and to estimate the percentage of total variance in the out-
come measure that results from each (quantified as the IntraClass Correlation
Coefficient [ICC]). The analysis was conducted as a two-level model, with
person at Level 1 and context at Level 2. The analyses were conducted in
SPSS 19 using MIXED LINEAR. This analysis assumes a continuous and
normal dependent variable (which was violated). A parallel set of analyses
46 Environment and Behavior 45(1)
were also performed using the SPSS 19 GENLINMIXED procedure and
specifying a logistic link function and binomial distribution. For ease of
interpretation, we have presented the results in the original probability units
from the continuous SPSS MIXED multilevel model (0 = no littering and 1
= littering), rather than log-odds units. The conclusions are the same as those
obtained using a logistic link function.
The initial random effects model showed that the overall littering rate was
.17. That is, of all disposals, 17% were improper, t(74.86 = 9.87, p < .001).
Across the 1,962 disposals, σ = .12, Z = 30.66, p < .01, and the 86 locations, τ00
= .022, Z = 5.71, p < .01, there was considerable variability in the littering rate.
The ICC was .15. This statistic is directly interpretable, and it indicates that
15% of the variance in littering behavior resulted from site-level variables,
whereas 85% resulted from individual variability. This finding shows that on a
national level, the large majority (85%) of littering behavior results from indi-
vidual-level variables (e.g., age, gender, attitudes, and motivation). This is not
saying that physical context does not matter, and in fact, these results show that
15% of the variance in observed littering behavior was due to some aspect of
the context (e.g., existing litter, lack of convenient receptacles, etc.).
The second set of analyses focused on individual-level predictors of litter-
ing behavior: age, gender, time of day, and whether the individual was alone.
Results showed that age, β = −.001; df = 1943, t = −2.15, p < .05, and gender,
β = −.06; df = 1943, t = −3.42, p < .05, were the only significant predictors,
with older individuals littering less than younger and males littering more
than females. Time of day and being alone were not significantly predictive
of littering. For clarification, age was coded into demographic categories.
The highest rate of observed littering occurred for younger individuals, aged
18 to 29 years, for whom the littering rate was 26%. For adults 30 years and
older, the littering rate remained steady at approximately 15%. Children and
adolescents (younger than 18 years) had a littering rate of 13%. The gender
effect in the general littering observations showed that men (21%) littered
more than women (15%). No other individual-level variables were predictive
of littering. However, the variability in the Level-1 equation remained statis-
tically significant, indicating that other variables are required to fully explain
individual variability in littering. Combined, age and gender explained less
than 1% of the residual variance at Level 1.
Using the hierarchical structure of these data, the contextual predictors of
littering behavior were analyzed: site type (e.g., city center and fast food),
location type (rural, urban, and suburban), amount of existing litter present,
beautification efforts in the area (included ratings for cleanliness, landscap-
ing, and infrastructure), availability and number of receptacles (summed
Schultz et al. 47
number across trash, ash, and recycling), posted signage about littering, and
crowdedness of the location.
Given the relatively large number of predictor variables, particularly with
the dummy coded categorical variables of site type and location type, the
multilevel model was conducted through a building process as recommended
by Raudenbush and Bryk (2002). The analysis started by testing each dummy-
coded categorical variable, sequentially, and removing nonsignificant predic-
tors. The continuous predictors were then examined, again removing
nonsignificant predictors. The cumulative results from these analyses
revealed two uniquely and statistically significant predictors: availability of
disposal receptacles and amount of litter present. The first was the number of
disposal receptacles. As part of the site observations, the team counted the
number of receptacles (trash, recycling, cigarette, and dumpster), along with
the distance from the person at the time of disposal. The average was 5.8 bins
per location, with a range from 0 to 19. The analysis for presence of recep-
tacles revealed the expected finding that littering rates were higher when no
receptacle was present. But more relevant to the current research questions,
among sites with at least one receptacle, the statistical analysis showed that
locations with more receptacles had a lower littering rate, γ = −.01, df =
73.42; t = −2.52, p < .05. This statistical coefficient can be interpreted directly,
such that for every added trash receptacle, the littering rate decreased by 1%
(from the overall rate of 17%).
The second statistically significant predictor of littering behavior was the
presence of litter in the site. Locations with more litter were associated with
a higher littering rate. The statistical analyses showed that the presence of
existing litter (rated by the observers on a scale from 0-10) was predictive of
littering behavior, γ = .02, df = 82.01; t = 2.40, p = .018. This indicates that
for every unit increase in the amount of existing litter (from 0-10), the
observed littering rate increased by 2%. With both predictors in the model,
the variance at Level 2 remained statistically significant, indicating that more
variables are needed to fully explain the variability across site. Combined, the
two variables explained 9% of the Level-2 variance, but the variability in lit-
tering rates across the sites remained statistically significant, τ = .019; Z =
4.60, p < .001.
Supplemental analyses were also conducted using the Level-1 predictor of
distance to the nearest receptacle. This analysis was performed using the mul-
tilevel framework but only for observations at sites with at least one recep-
tacle (of any type). Results showed that distance to a receptacle at time of
disposal was strongly related to the likelihood of littering, β = .007; df =
1926, t = 17.95, p < .001. Distance to the receptacle explained 11% of the
48 Environment and Behavior 45(1)
residual variance at Level 1, σ = .108, Z = 30.55, p < .001. Note that distance
was coded in feet, so that for each added foot of distance from a receptacle at
the time of disposal, the probability of littering increased by .007. For clarifi-
cation, we calculated littering rates for disposals at seven different distances,
each of 10-foot increments. For disposals that occurred within 0 to 9 feet of a
receptacle, littering rates were 12%. At the largest distance (60 or more feet),
littering rates were 30% of disposals.
Observations of smokers. In addition to the large number of general litter-
ing observations, the field team also conducted a smaller number of obser-
vations of smokers. The separate focus on smokers was based on two
considerations. First, unlike the general littering observations of individu-
als moving through a public space, all smokers have something to litter—a
cigarette butt. Second, cigarette butts constitute the most frequently collected
litter worldwide.
The observations were made using the same protocol described above, with
a few modifications. First, only individuals who were (estimated) over the age
of 21 were utilized. This qualifier was imposed to provide consistency across
our observational protocol and to respond to the possibility of local restric-
tions on tobacco use for individuals younger than 21 years. Second, a measure
of existing litter was included at each site that focused on the number of ciga-
rette butts within the observational boundary. Third, the measure of existing
receptacles focused on only ashtrays (or trash/ash combinations). As with the
previous study, disposals of cigarette butts were coded as proper if they
reached any type of receptacle and not necessarily an ashtray.
In total, observational data were obtained from 767 smokers from 44 sites
(11 recreational, 12 bars/restaurants, 12 retail, 8 medical, and 1 city center).
There were 412 males and 344 females, ranging in age from 21 to 72 (M = 40,
SD = 13; 11 not coded). Of the 767 observed individuals, 206 (27%) left the
observation area still smoking, and the disposal behavior of 31 smokers could
not be clearly established. Of the remaining 530 smokers, 187 properly dis-
posed of the butt (35%) and 343 improperly disposed (65%). When the butt
was littered, drop with intent was the most frequently used strategy (35%),
followed by flick (27%), stomp (27%), and “other” (1%, including placing
the butt on or near a receptacle). Although there were several instances of
littering near a receptacle, most littering occurred at considerable distance
from a receptacle (average distance at time of littering was 31 feet).
The data analytic strategy followed the multilevel model approach used
above, in which the individual and contextual predictors of littering were
examined simultaneously. The reported results were calculated using the
MIXED command in SPSS version 19, and the results are reported in original
Schultz et al. 49
probability units. The analysis was performed on 530 cases (187 proper dis-
posals and 343 littered). The results from the multilevel model showed that
across the 530 individuals, σ = .135, Z = 15.70, p < .001, and the 44 locations,
τ00 = .081, Z = 4.73, p < .01, there was considerable variability in the littering
rate. The ICC was .38, indicating that 38% of the variance in cigarette litter-
ing resulted from contextual variables, whereas 62% resulted from individual
variability. This is a considerably higher clustering effect than that observed
for general littering behavior, and it suggests that cigarette butt disposal is
more affected by contextual-level variables than are general disposals (see
recommendations section below).
At the level of the individual (Level 1), only age emerged as a statistically
significant predictor, with older individuals littering less than younger, β =
−.004, df = 518.49; t = 2.93, p < .01. The highest littering rates occurred for
smokers in their 20s (66% littering rate) and 30s (72%), compared with
smokers in their 40s (58%), 50s (66%), and 60s (50%). Age explained less
than 1% of the Level-1 variance, and the variability across individuals
remained statistically significant, σ = .134, Z = 15.59, p < .001, suggesting
the need for additional predictors. Neither gender, time of day, nor being part
of a group were related to cigarette butt littering.
At the level of site (Level 2), analyses utilized the contextual predictors used
in the analyses of general litter (with minor modifications noted above). The
results showed three uniquely predictive variables: site type, existing litter, and
presence of ash receptacles. One of the strongest predictors of cigarette littering
was the number of ash receptacles, γ = −.09, df = 31.91, t = −2.13, p < .01). The
parameter estimate from the analysis is directly interpretable, and it indicates
that for every added ash receptacle, the littering rate for cigarette butts decreased
by 9% (from the initial base littering rate of 65%). The second significant pre-
dictor of cigarette litter was the amount of existing litter, γ = .05, df = 39.01,
t = 2.24, p = .03, with more littered environments attracting more cigarette butt
litter. Note that the existing litter is of any type and not just cigarette butts.
Results also showed an effect for site type, where retail locations were associ-
ated with the lowest rate of littering (58%), followed by city centers (58%).
Bars and restaurants were third (62%), whereas recreational (74%) and medi-
cal/hospital sites (75%) had the highest littering rates. Combined, these three
variables explained 24% of the variance at Level 2, although the variability in
littering rates across site remained statistically significant, τ = .061, Z = 3.21,
p < .01, suggesting the need for additional predictors.
Finally, supplemental analyses were conducted to examine distance to an
ash receptacle at time of disposal. Commensurate with the previous analyses,
this Level-1 predictor was examined within the multilevel framework but
50 Environment and Behavior 45(1)
only using data from sites with at least one ash receptacle. Results showed
that distance to the nearest receptacle was strongly predictive of littering, β =
.005, df = 292.79, t = 6.93, p < .001. Although a few instances of littering
were observed immediately adjacent to an ash receptacle, the average dis-
tance for litterers was 31 feet away.
Discussion and Conclusions
The results from these litter observations support a number of conclusions.
First, the overall littering rate was 17%. That is, of all the disposals observed
across the country, 17% were improper. In addition, of the individuals sam-
pled from 86 locations nationwide, 4% littered as they passed through the
site. For cigarette butts, the littering rate obtained from the focused observa-
tions was 65%. This is a strikingly high number, despite the strong norm
favoring proper disposal that has emerged over the past 40 years (see Bator,
Bryan, & Schultz, 2011).
Importantly, this littering rate was generated from a random sample of
individuals across a range of different locations and not just a few isolated
observations or of one type of location. In addition, the results showed that in
the majority of instances (81%), the littering occurred with intent.
The results from these analyses underscore findings from studies con-
ducted nearly 40 years ago (Burgess et al., 1971; Finnie, 1973; Geller et al.,
1977). Although much has been said about litter and littering over the years,
no study has afforded the opportunity to simultaneously test the degree to
which it is affected by personal and contextual variables. In fact, to our
knowledge, this is the first article to examine the same behavior across a large
number of contexts—a procedure which allows for a quantitative analysis of
“personal” and “environmental” influences on behavior. The results of the
current research indicate that 15% of general littering acts result from contex-
tual variables, and 85% result from personal qualities. This finding is particu-
larly instructive because it indicates that given the same infrastructure and
opportunities to properly dispose, individuals will vary tremendously. Note
that if the trend had been reversed, such that 85% of the variance was due to
the situation, it would indicate that while individuals vary across settings,
within a setting they act similarly (e.g., littering or not).
The results from the analyses of littering behavior identified only a couple
of significant predictors. Interestingly, gender was not a consistent predictor
of littering behavior. Gender was a significant predictor of littering in the
general observations, with males littering more than females. However, gen-
der was not a significant predictor of littering for cigarette butts. This second
Schultz et al. 51
finding runs contrary to prior data showing that men are more likely to litter
than women (Meeker, 1997; Torgler, García-Valiñas, & Macintyre, 2008) but
is consistent with other observational studies showing no gender effects
(Finnie, 1973; Geller et al., 1977; Williams et al., 1997).
At the individual level, the results did show a consistent and statistically
significant effect for age, with young adults (18-29) more likely to litter than
older adults. The negative relationship between age and littering has been
documented in several survey studies of littering behavior (Beck, 2007), with
researchers reporting that younger people tend to litter more often than those
who are older (e.g., Durdan, Reeder, Hecht, 1985; Finnie, 1973; Heberlein,
1971; Krauss, Freedman, & Whitcup, 1978). Krauss et al. (1978) also found
that younger participants were more likely to litter. They considered that nor-
mative control requires both internal controls and cognitive information, both
of which develop through the socialization process.
At the level of the location, presence and number of trash receptacles,
along with the amount of litter present were significant predictors of littering
behavior. These findings are consistent with previous studies (Cialdini et
al.1990; Meeker, 1997), although Roales-Nieto (1988) reported results show-
ing that adding more receptacles did not result in reductions in litter. This
latter finding suggests that a raw count of receptacles is probably an overly
simplistic consideration. Indeed, the current study shows that convenience
(i.e., distance to a receptacle) plays an important role. One well-placed recep-
tacle is likely to produce a larger reduction in littering than several inconve-
niently placed receptacles.
To this end, it is tempting to ask about the “optimal” spacing between
receptacles. Although these data do not speak directly to this issue, there is
evidence that the lowest littering rate occurs when receptacles are available
and close at hand. This effect was consistent for both general littering and
disposals of cigarette butts. Further inspection of the data showed that aggre-
gated observed general littering rates were low (and relatively flat at 12%) for
receptacles less than 20 feet away. The littering rates increased linearly
between 21 and 60 feet and then remained relatively flat at 30% for recepta-
cles 61 feet away and beyond. It is also important to point out that “optimal”
spacing will vary by location, and the key consideration is the distance to the
receptacle when the individual has an item for disposal. The current results
showed that the lowest rate of littering occurred when a receptacle was fewer
than 20 feet away. To deter littering, we encourage thoughtful placement of
receptacles so they are in the most easily accessible location depending on
where pedestrians are likely to be when they are in need of making a
disposal.
52 Environment and Behavior 45(1)
The observations of smokers revealed similar findings. First, with regard to
cigarette butt litter, results showed an average national littering rate of 65%.
This is substantially higher than that found for littering in general and corrobo-
rates the high number of cigarette butts collected in cleanups worldwide. As
with general litter, younger individuals were more likely to litter than older,
although the overall rate of improper cigarette disposal was above 50% for all
age groups. With regard to the multilevel analyses, results showed a clustering
effect of .38, indicating that a substantial amount of variability in littering
behavior results from contextual variables. Subsequent analyses revealed that
the lack of convenient ash receptacles, and sites with high levels of existing
litter (of any type, not just cigarette litter), were predictive of higher litter rates.
Although the littering rates reported in this article are based on a large,
national sample, it is important to acknowledge a few methodological limita-
tions. First, the sample cannot be considered representative of all individuals
in the United States. Although the reported results are based on random sam-
ples of individuals within each site and of randomly selected specific site loca-
tions across the country, the type of sites where observations were made was
not randomly determined. That is, eight specific types of sites were selected
for observations (e.g., retail and recreational) at the outset of the study, and
although diverse, these eight site types cannot be considered representative of
all physical environments across the country. Although drawing a random
sample of physical locations across the country is methodologically desirable,
it was not practically feasible in the current project.
Another limitation with the littering rate reported in this article is the
potential bias toward large or more readily observable items. The observed
littering rate of 17% is probably an underestimate for the true littering rate
because there were certainly littered objects that went undetected by the field
team. Small items, in particular, are likely to be underrepresented using the
observational protocol by virtue of the difficulty in seeing them from a dis-
tance. It is also possible that the mere presence of the research team in the
environments muted the littering rates, although the observational protocol
was designed to minimize this influence.
Just as the observational protocol is likely to underestimate the littering
rate, it is also likely to overestimate “intended” littering. In coding behaviors
in the field, proper disposals are relatively easy to determine, as are intended
littering behaviors. Although the observational team was meticulous in observ-
ing and coding behaviors, unintentional littering is inherently more difficult to
detect. As a result, the reported 81% of litter that is intentional should be inter-
preted with caution. This limitation is more applicable to the general littering
observations and less so for the focused smoker observations. Given that
Schultz et al. 53
intentional littering has been found to be more easily deterred than uninten-
tional littering (Sibley & Liu, 2003) and that we recorded substantially more
unintentional littering behavior, we encourage tests of litter prevention tech-
niques that promote awareness and individual-level motivation.
Finally, we offer a caution on our interpretation of the ICC. The ICC repre-
sents the degree to which the data “cluster”, and in our study, it quantifies the
proportion of variance in the dependent variable that is attributed to the site,
rather than to the individual. At one extreme, an ICC = 1.0 would indicate that
all of the variance in littering behavior was associated with site such that all
individuals observed within each site were the same (either littering or not) but
that littering occurred at some sites but not others. On the other extreme, with
an ICC = 0, all sites would show the same littering rate, but individuals would
vary within each site. At the site level, we have used contextual variables like
the availability of trash receptacles or the amount of existing litter as predic-
tors. However, it is important to point out that site-level variance could also be
due to shared regional or local norms associated with littering in different
types of locations or even shared norms about littering in general. To illustrate,
consider the case of cigarette butt disposal, for which we found an ICC of .38.
Although some of this clustering is certainly due to contextual variables like
the availability of an ashtray, some may also be due to the strong norm against
smoking and littering in some contexts. For instance, had we included ciga-
rette butt disposal by staff smoking in a designated area near an elementary
school, the littering rates would likely have been affected more by the norm of
social responsibility than by the availability of ashtrays.
Implications for Litter Prevention
The findings from this research point to several strategies for litter preven-
tion. These strategies include a combination of both structural and motiva-
tional activities. This section of the article provides a series of
recommendations for litter prevention that are consistent with the research
findings. Importantly, this is not an exhaustive list, and readers are encour-
aged to think creatively about ways to link the reported findings to litter
prevention. In addition, it seems likely that any single prevention activity
will yield only small results, and the most effective approach will utilize
multi-pronged strategies that target both structural and personal variables.
Beautification. The current results clearly show that litter begets littering.
This finding is not new, and indeed, it was noted in the early studies of litter
(e.g., Cialdini et al., 1990; Keizer et al., 2008). However, although most of
the early studies presented participants with an object to either litter or
54 Environment and Behavior 45(1)
dispose of properly, the current research was completely nonintrusive. We
observed genuine behavior in a variety of settings across the United States.
Given this methodology, we were able to learn that individuals use a variety
of cues from their surrounding environment to determine what is a common
and accepted behavior. The presence of litter communicates the norm for that
situation and the acceptability of littering. In addition, the existing litter will
require cleanup, so one more piece may seem inconsequential.
To this end, a key to the success of any litter prevention activity is to clean
up and remove existing litter. Reducing the amount of existing litter in a loca-
tion is a surefire way to reduce the rate of littering behavior (Casey & Lloyd,
1977; Huffman, Grossnickle, Cope, & Huffman, 1995). In addition, prior
studies have found that involving community residents in cleanup activities
can promote a long-term reduction in litter and increase an individual’s moti-
vation not to litter (Roales-Nieto, 1988).
Behavioral opportunity. Related to the recommendation for beautification
efforts (above), there is also consistent evidence for the importance of oppor-
tunity. That is, the context should provide a convenient and accessible means
for proper disposal of trash and recyclables. Although the current results
show the widespread availability of receptacles in public places, results also
revealed that distance to a trash can was a strong predictor of littering behav-
ior. Providing easily identifiable, accessible receptacles, with clear and rec-
ognizable messaging and prompts, can go a long way toward reducing
littering rates (De Kort, McCalley, & Midden, 2008; National Cooperative
Highway Research program, 2009).
The issue of behavioral opportunity is especially important for cigarette
butts. The reported observational data suggest that disposal of cigarette butts
is more strongly clustered within locations, yet less than half (47%) of the
locations in the sample provided an ash receptacle. Indeed, Liu and Sibley
(2004) reported a 64% drop in cigarette butt littering by adding ashtrays on a
university campus, although the change did not affect attitudes about litter.
Similar results were reported by Geller, Brasted, and Mann (1980) and by
Sibley and Liu (2003). Given the increase in legislation prohibiting indoor
smoking, an increasing number of smokers are moving outside to smoke.
However, the infrastructure for collecting ashes and lit cigarettes is woefully
behind these policies, and the reported data suggest that more efforts to afford
smokers an opportunity for proper disposal are needed.
Awareness and motivation campaigns. In addition to the recommendations
for beautification and infrastructure, there is an important role for litter pre-
vention strategies that target individual-level motivation. The statistical anal-
yses showed that 85% of the variance in general littering and 62% of the
variance in cigarette butt littering resulted from individual differences. These
Schultz et al. 55
include demographic, attitudinal, and motivational differences (among oth-
ers), and they speak to the importance of understanding the individual-level
motivations and barriers to littering (McKenzie-Mohr, 2002).
One way to promote individual-level motivations is through outreach and
media messages (Nolan, Schultz, & Knowles, 2009). Although prior research
has shown that such campaigns typically only produce small changes in
behavior (if any), there is reason to continue utilizing media messages, and
more importantly branding, in litter prevention efforts. Based on the reported
data, and background literature, messages should highlight the dramatic
decline in littering rates over the past 40 years, the generally infrequent over-
all littering rate, and the widespread disapproval for individuals who litter
(see also Cialdini, 2003).
In a related data set collected as part of the current study, surveys were con-
ducted with observed individuals randomly sampled across the country. These
surveys were conducted with both litterers and nonlitters, and the findings show
a near unanimous disapproval for littering (Bator et al., 2011). This finding,
coupled with other research on the role of injunctive and personal norms, sug-
gests that messages should emphasize that only a few deviant individuals litter
and that these individuals are disapproved of by the majority (see Cialdini, 2003;
Grasmick, Bursik, & Kinsey, 1991; Schultz, Tabanico, & Rendón, 2008).
In closing, the data reported in this article represent the largest single study of
littering behavior conducted to date. Data are reported from unobtrusive obser-
vations of nearly 10,000 randomly selected individuals across 130 diverse pub-
lic locations across the country. The results show that of all the disposals that
took place in these locations, 17% resulted in litter. For disposals of cigarette
butts, the littering rate was even higher, at on observed rate of 65%. Statistical
analyses are reported that utilize the multilevel framework and simultaneously
examine both the contextual and personal predictors of littering behavior. The
findings show that littering results from both personal and contextual factors and
that both are critical in understanding littering behavior. This perspective is con-
sistent with the traditional approach utilized by environmental psychologists
and can be particularly instructive in efforts to reduce littering rates.
Author’s Note
Supplemental Appendix of this article is available on the Environment and Behavior’s
Web site: http://eab.sagepub.com/.
Acknowledgment
The authors want to acknowledge the excellent work and commitment of their field
research team: Jenna Albert, Sara Aguilar, Michelle Cugini, Tracy Galea, Elizabeth
Morales, Belinda Rojas, and Michael Stringham. They also thank their collaborating
56 Environment and Behavior 45(1)
teams in New York and Vermont: Montgomery Bopp, Kara Carpenter, Ashley Doyle,
Cassie Fortney, Jamie Kuhn, Nicole LeFevre, Andrea Martino, Eva Richardson, and
Ashlee Rock. Finally, they want to thank the collaborating team in New Mexico:
Angela Bryan, Maddia Ikeda, Jenna Kicklighter, Stefan Klimaj, Eva Padilla, Jenna
Tonelli, and Kiani Wong. The authors want to express their appreciation for the sup-
port and encouragement of Susanne Woods and the staff at Keep America Beautiful.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.
Funding
The research reported in this article was conducted with funding from Keep America
Beautiful, using financial support from Philip Morris, An Altria Company.
References
Bator, R., Bryan, A., & Schultz, P. W. (2011). Who gives a hoot? Intercept surveys of
litterers and disposers. Environment and Behavior, 43, 295-315.
Beck R. W. (2007). Literature review—Litter: A review of litter studies, attitudes sur-
veys, and other litter related literature. Retrieved from http://www.kab.org/site/
DocServer/Litter_Literature_Review.pdf?docID=481
Brown, B., Perkins, D., & Brown, G. (2004). Crime, new housing, and housing inci-
vilities in a first-ring suburb: Multilevel relationships across time. Housing Policy
Debate, 15, 301-345.
Burgess, R. L., Clark, R. N., & Hendee, J. C. (1971). An experimental analysis of anti-
littering procedures. Journal of Applied Behavior Analysis, 4, 71-74.
Casey, L., & Lloyd, M. (1977). The cost effectiveness of litter removal procedures in
an amusement park. Environment and Behavior, 9, 535-546.
Cialdini, R. B. (2003). Crafting normative messages to project the environment. Cur-
rent Directions in Psychological Science, 12, 105-109.
Cialdini, R. B., Kallgren, C. A., & Reno, R. R. (1991). A focus theory of normative
conduct: A theoretical refinement and reevaluation of the role of norms in human
behavior. Advances in Experimental Social Psychology, 21, 201-234.
Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative
conduct: Recycling the concept of norms to reduce littering in public places. Jour-
nal of Personality and Social Psychology, 58, 1015-1026.
Cone, J. D., & Hayes, S. C. (1980). Environmental problems: Behavioral solutions.
Monterey, CA: Brooks/Cole.
De Kort, Y., McCalley, L., & Midden, C. (2008). Persuasive trash cans: Activation of
littering norms by design. Environment and Behavior, 40, 870-891.
Schultz et al. 57
Durdan, C. A., Reeder, G. D., & Hecht, P. R. (1985). Litter in a university cafeteria:
Demographic data and the use of prompts as an intervention strategy. Environ-
ment and Behavior, 16, 387-404.
Finnie, W. C. (1973). Field experiments in litter control. Environment and Behavior,
5, 123-144.
Geller, E. S. (1980). Applications of behavioral analysis for litter control. In D. Glen-
wick & L. Jason (Eds.), Behavioral community psychology: Progress and pros-
pects (pp. 254-283). New York, NY: Praeger.
Geller, E. S., Brasted, W., & Mann, M. (1980). Waste receptacle designs and interven-
tions for litter control. Journal of Environmental Systems, 9, 145-160.
Geller, E. S., Winett, R. A., & Everett, P. B. (1982). Preserving the environment: New
strategies for behavior change. Elmsford, NY: Pergamon.
Geller, E. S., Witmer, J. F., & Tuso, M. E. (1977). Environmental interventions for
litter control. Journal of Applied Psychology, 62, 344-351.
Grasmick, H., Bursik, R., & Kinsey, K. (1991). Shame and embarrassment as deter-
rents to noncompliance with the law: The case of an antilittering campaign. Envi-
ronment and Behavior, 23, 233-251.
Heberlein, T. (1971). Moral norms, threatened sanctions, and littering behavior
(Unpublished doctoral dissertation). University of Wisconsin, Madison.
Hox, J. (2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Erlbaum.
Huffman, K., Grossnickle, W., Cope, J., & Huffman, K. (1995). Litter reduction: A
review and integration of the literature. Environment and Behavior, 27, 153-183.
Keep America Beautiful. (2007). KAB’s seven primary sources of litter. Retrieved
from http://www.kab.org/site/PageServer?pagename=litter_sevensources
Keizer, K., Lindenberg, S., & Steg, L. (2008). The spreading of disorder. Science,
322, 1681-1685.
Krauss, R. M., Freedman, J. L., & Whitcup, M. (1978). Field and laboratory studies of
littering. Journal of Experimental Social Psychology, 14, 109-122.
Liu, J., & Sibley, C. (2004). Attitudes and behavior in social space: Public good inter-
ventions based on shared representations and environmental influences. Journal
of Environmental Psychology, 24, 373-384.
McKenzie-Mohr, D. (2002). The next revolution: Sustainability. In P. Schmuck & P.
W. Schultz (Eds.), Psychology of sustainable development (pp. 19-36). Norwell,
MA: Kluwer.
Meeker, F. L. (1997). A comparison of table-littering behavior in two settings: A case
for a contextual research strategy. Journal of Environmental Psychology, 17, 59-68.
MSW Consultants. (2009). National visible litter study. New Market, MD: Author.
Retrieved from www.kab.org/research09
National Association of Home Builders. (2009). House price estimator. Available
from www.nahb.org
58 Environment and Behavior 45(1)
National Cooperative Highway Research Program. (2009). Reducing litter on road-
sides. Washington, DC: Transportation Research Board of the National Academies.
Retrieved from http://onlinepubs.trb.org/onlinepubs/nchrp/nchrp_syn_394.pdf
Nolan, J., Schultz, P. W., & Knowles, E. (2009). Using public service announcements
to change behavior: No more money and oil down the drain. Journal of Applied
Social Psychology, 39, 1035-1056.
Ocean Conservancy. (2007). Start a sea change. Washington DC: Author. Retrieved from
http://www.oceanconservancy.org/site/DocServer/ICC_AR07.pdf?docID=3741
Pandey, J. (1990). The environment, culture, and behavior. In R. Brislin (Ed.), Applied
cross-cultural psychology (pp. 254-277). Thousand Oaks, CA: SAGE.
Public Opinion Surveys, Inc. (1968). Who litters and why. Princeton, NJ: Keep Amer-
ica Beautiful, Inc.
Raudenbush, S., & Bryk, A. (2002). Hierarchical linear models: Applications and
data analysis methods. (2nd ed.). Thousand Oaks, CA: SAGE.
Roales-Nieto, J. G. (1988). A behavioral community programme for litter control.
Journal of Community Psychology, 16, 107-118.
Schultz, P. W., Tabanico, J., & Rendón, T. (2008). Normative beliefs as agents of influ-
ence: Basic processes and real-world applications. In R. Prislin & W. Crano (Eds.),
Attitudes and attitude change (pp. 385-409). New York, NY: Psychology Press.
Sibley, C., & Liu, J. (2003). Differentiating between active and passive littering: A
two-stage process model of littering behavior in public spaces. Environment and
Behavior, 35, 415-433.
Skogan, W. (1990). Decline and disorder: Crime and the spiral of decay in American
neighborhoods. New York, NY: Free Press.
Torgler, B., García-Valiñas, M., & Macintyre, A. (2008). Justifiability of littering: An
empirical investigation (Working Paper No. 2008-8). Basil, Switzerland: Center
for Research in Economics, Management, and the Arts. Retrieved from http://
www.crema-research.ch/papers/2008-08.pdf
Williams, E., Curnow, R., & Streker, P. (1997). Understanding littering behaviour in
Australia: A report for the Industry Environment Council. Retrieved from http://
www.afgc.org.au/cmsDocuments/LBS%20I.pdf
Bios
P. Wesley Schultz is professor of psychology at California State University, San
Marcos. He earned his bachelor’s degree from the University of California, Irvine and
his doctoral degree from the Claremont Graduate University. His research interests are
in applied social psychology, particularly in the area of sustainable behavior. His cur-
rent work focuses on social norms and the importance of social norms in promoting
conservation. He has worked on projects for a variety of organizations, including the
U.S. Environmental Protection Agency, National Institute of General Medical Science,
National Institute of Justice, and the California Integrated Waste Management Board.
Schultz et al. 59
Renée J. Bator is professor and cochairperson in the psychology department at the
State University of New York, Plattsburgh. She completed her bachelor’s degree in
psychology at the University of California at Santa Cruz. She earned her MA and PhD
from the Social Psychology Program at Arizona State University. Her research inter-
ests focus on the application of persuasion theory to prosocial outcomes, especially
those with environmental or health benefits.
Lori Brown Large is project director at Action Research. She earned her BA and MA
degrees in sociology from California State University, Fullerton. She has extensive
experience in the area of applied social science with an emphasis in survey research
design and implementation. Her particular interests are in the area of waste reduction,
energy conservation, and pollution prevention. At Action Research she works with a
variety of organizations to develop, implement, and evaluate outreach campaigns
aimed at changing behavior.
Coral M. Bruni is a doctoral student at the Claremont Graduate University. She
completed her bachelor’s and master’s degrees in psychology at California State
University, San Marcos. Her research interests include connectedness with nature,
self/identity, and implicit social cognition. Her recent studies have focused on the role
of implicit connectedness with nature across a variety of ages and places.
Jennifer J. Tabanico is principal at Action Research. She earned her BA degree in
psychology and MA degree in experimental psychology both from California State
University, San Marcos. Her research interests are in applied social psychology with
particular emphasis on the applications of psychological theory to the development
of public policy and behavior change programs. In both academic and professional
appointments, she has completed numerous studies of environmental attitudes, haz-
ardous waste management, energy conservation, and pollution prevention. At Action
Research, she works with a variety of organizations to develop, implement, and
evaluate outreach campaigns aimed at changing behavior.
