Global Environmental Change 23 (2013) 14881500

Contents lists available at ScienceDirect

Global Environmental Change

journal homepage: www.elsevier.com/locate/gloenvcha

Who remembers a hot summer or a cold winter? The asymmetric

effect of beliefs about global warming on perceptions of local climate
conditions in the U.S.
Peter D. Howe a,b,*, Anthony Leiserowitz b
a
b

Department of Environment and Society, Quinney College of Natural Resources, Utah State University, 5215 Old Main Hill, Logan, UT 84322, USA
School of Forestry and Environmental Studies, Yale University, 195 Prospect Street, New Haven, CT 06405, USA

A R T I C L E I N F O

A B S T R A C T

Article history:
Received 2 April 2013
Received in revised form 19 September 2013
Accepted 25 September 2013

This paper explores the phenomenon of local climate perception and the extent to which public
perceptions match climate conditions as recorded in instrumental climate data. We further examine
whether perceptions of changes in local climates are inuenced by prior beliefs about global warming,
through the process of motivated reasoning. Using national survey data collected in the United States in
2011, we nd that subjective experiences of seasonal average temperature and precipitation during the
previous winter and summer were related to recorded conditions during each season. Beliefs about
global warming also had signicant effects on subjective experiences with above-normal temperatures,
particularly among those who believed that global warming is not happening. When asked about the
summer of 2010, those who believed that global warming is not happening were signicantly less likely
to report that they had experienced a warmer-than-normal summer, even when controlling for
demographics and local climate conditions. These results suggest that the subjective experience of local
climate change is dependent not only on external climate conditions, but also on individual beliefs, with
perceptions apparently biased by prior beliefs about global warming.
2013 Elsevier Ltd. All rights reserved.

Keywords:
Climate change
Global warming
Risk perception
Experience
Motivated reasoning
Spatial analysis
Survey research

1. Introduction
Thus far, global warming has manifested gradually over many
decades and at spatial scales well beyond the direct perceptual
capabilities of any individual human being (e.g. the global or
continental scale). Local weather conditions, on the other hand, are
a readily available source of information that, when aggregated
over time, may enable people to detect long-term climate trends at
the local scale (Howe et al., 2013; Orlove et al., 2010). The
translation of personal experience of changes in local weather
conditions to perceptions of climate variability and change is an
important component of individual and community adaptation
(Adger et al., 2007). Research on local climate knowledge has found
that people are able to detect and respond to changes in climate
(Strauss and Orlove, 2003), but the characteristics of local
manifestations of climate change that are perceived have been
hypothesized to be dependent on a variety of individual and
contextual factors. These factors include the importance of specic

* Corresponding author at: Department of Environment and Society, Quinney

College of Natural Resources, Utah State University, 5215 Old Main Hill, Logan, UT
84322, USA. Tel.: +1 4357979457.
E-mail addresses: peter.howe@usu.edu (P.D. Howe),
anthony.leiserowitz@yale.edu (A. Leiserowitz).
0959-3780/$ see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.gloenvcha.2013.09.014

climatic conditions to individual livelihoods (Meze-Hausken,

2004; Osbahr et al., 2011; Roncoli et al., 2002), the spatial scale
of changes (Howe et al., 2013; Ruddell et al., 2012), and the
reference periods over which individuals establish representations
of a normal climate (Hulme et al., 2009; Sanchez-Cortes and
Chavero, 2011). Perceptions of change in local climate, as with
other individual judgments, may be subject to systematic cognitive
biases that favor experiential over descriptive learning (Marx et al.,
2007). However, there has been little attention to the possibility of
biased perceptions of climate change at the local scale due to preexisting beliefs about climate change at the global scale. The
existence of strongly held beliefs about the direction of change in
the global climate may bias judgments about local climate in the
direction predicted by ones prior beliefs about the global climate.
Such biases in local climate perceptions, if present, may act as a
barrier to accurate detection of local climate change and an
impediment to effective climate change adaptation.
Recent research has shown that perceived personal experience
with global warming leads to heightened global warming risk
perceptions and greater certainty in the belief that global warming
is happening (Akerlof et al., 2013; Myers et al., 2012; Spence et al.,
2011). More specically, perceived experience appears to lead to
greater certainty that global warming is happening only among
those who have weakly held beliefs about global warming, while
motivated reasoning affects perceived personal experience among

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

those who have strongly held beliefs about global warming (Myers
et al., 2012). Motivated reasoning is the tendency to interpret
information to t pre-existing beliefs (Kunda, 1990). In this paper,
we extend previous ndings by exploring the effect of motivated
reasoning on perceptions of local seasonal climate while controlling for actual local climate conditions. Drawing from a nationally
representative survey of the U.S. population, we rst characterize
the relationship between instrumental climate data and perceptions of local seasonal climate. We subsequently examine the
relationship between sets of beliefs about global warming and
perceptions of local climate conditions.
The relationship between personal experience and beliefs about
global warming is of considerable interest as changes in local
weather and climate conditions continue to be consistent with
scientic projections of global warming. For example, between
January 2000 and September 2009 maximum temperature records
were broken more than twice as frequently as minimum temperature records in the contiguous U.S. (Meehl et al., 2009), and extreme
events such as the 2011 Texas heat wave and drought have become
much more likely (Peterson et al., 2012). While direct attribution of
any single weather event to long-term processes like global
warming is not possible, the accumulation of weather events that
fall outside the range of previous experience does provide evidence
that the climate is changing, since local extreme events become
more likely as the world warms (Hansen et al., 2012; Meehl and
Tebaldi, 2004; Rahmstorf and Coumou, 2011). But can individuals,
drawing upon their personal experience, accurately detect the
extent to which recent conditions have changed relative to the past?
It is therefore important to understand how people subjectively
experience their local climate, and what factors inuence their
judgments about whether local climates are changing. Previous
broad-scale survey research suggests that changes in local climate
conditions can inuence public perceptions of local warming trends
(Howe et al., 2013). While there is some evidence that recent
experience with short-term ambient temperatures may inuence
global warming beliefs (Akerlof et al., 2013; Borick and Rabe, 2010;
Egan and Mullin, 2012; Goebbert et al., 2012; Hamilton and
Stampone, 2013; Joireman et al., 2010; Li et al., 2011; Risen and
Critcher, 2011), there has been little attention to the possibility that
subjective experiences of local climate may also be inuenced by
pre-existing beliefs and attitudes about global warming, which
could affect the ability to recognize local climate change.
2. Background
Research in communities around the world has documented
many cases of people using personal experience to detect changes
in their local climate; such changes include altered plant and
animal phenology, new distributions of species, shorter or longer
growing seasons, and the changing frequency of extreme weather
events (Deressa et al., 2011; Orlove et al., 2000; Roncoli et al., 2002;
Smit et al., 1997; Thomas et al., 2007; Tschakert et al., 2010;
Weatherhead et al., 2010; West et al., 2008). Such research has
almost exclusively been carried out at the local scale using case
studies of one or a small number of communities. The communities
in these studies tend to be rural, with most residents engaged in
livelihoods based on agriculture, ranching, or shing. Because their
livelihoods are dependent on the weather, residents have a strong
incentive to pay attention to the variability of local weather and
climate, and indeed they tend to notice changes. The changes that
people notice tend to be closely related to the aspects of the
weather that have the most direct effect on the livelihoods in
which they engage, to such an extent that residents of the same
community may identify different changes depending on their
occupation (Hartter et al., 2012; Meze-Hausken, 2004; Osbahr
et al., 2011).

1489

Less clear is how individuals whose lives are not as directly

dependent on the weather might perceive local climate changes,
and if other individual factors might inuence their perception of
climate. Relatively few people in the U.S. are employed in a
profession that is as dependent on the vagaries of day-to-day
weather as, for example, a farmer who relies on rain-fed
agriculture (Meyer, 2000). Furthermore, the populations of
industrialized countries like the U.S. have adopted indoor climate
control technologies that insulate them in a thermal environment
disconnected from their local climate for much of the day
(Hitchings, 2011). These characteristics of the U.S. population
may be barriers to the public perceiving long-term changes in their
local climate. Despite ndings from national surveys in the U.S.
that majorities of the population believe that global warming is
affecting local weather and making extreme weather events worse,
it is not clear if the experience of such events is a causal factor in
belief change (Leiserowitz et al., 2013, 2012). Existing research is
mixed about the effect of short-term extreme events on beliefs and
behaviors related to climate change. For instance, there is
contradictory evidence that direct experience of ooding among
U.K. residents relates to concern about climate change (Spence
et al., 2011; Whitmarsh, 2008).
This analysis examines one facet of local climate perceptions:
judgments about whether seasonal temperature or precipitation
has differed from normal. Identifying abnormalities in seasonal
climate involves comparing current experience to memories of
past experience (Weber, 2010). If memories of past experience are
uninformative or inaccurate, it is possible that current conditions
may be falsely judged to be normal or abnormal, depending on the
direction in which memories of past experience have been
distorted (Rebetez, 1996). For example, an individual may assume
a July heat wave to be normal for that time of year if she
remembers previous years with hot summers, whether or not
those memories are accurate representations of her experience.
Thus, for personal experience with short-term climate conditions
to be an effective source of information about long-term local
climate change, individuals must be able to perform two
processes: perceive current weather conditions, and compare
their perceptions to their expectations of what is normal.
Expectations may be based on memories of personal experience or
descriptive information from external sources. If perceptions of
current conditions in relation to expectations are accurate, then
personal experience may be able to serve as a useful source of
information to motivate and guide adaptation to the specic
climatic changes happening in local places.
There are, however, multiple ways by which climate perceptions and expectations can be distorted. When asked to make a
rapid assessment of a complex phenomenon like the climate,
people may rely on intuitive processes, which may create
systematic biases in how people perceive seasonal climate.
Cognitive biases such as the tendency to overweight recent
experience in memory (known as the recency effect) and the
tendency to disregard the prior probabilities of events (known as
base rate neglect) may lead people to believe that the recent
weather that they have experienced is more representative of a
longer-term period than it really is. Base-rate neglect explains why
people may fail to account for prior probabilities when evaluating
conditional probabilities (Koehler, 1996), and may lead people to
assign an inordinate weight to unusual recent weather events
when evaluating the probability of long-term climate change.
Indeed, some existing evidence suggests that recency effects are
indeed present in farmer decision-making with respect to climate
information (Hansen et al., 2004).
Another potential source of error in judgments about seasonal
climate may result from motivated reasoning. Motivated reasoning
is the unconscious tendency to t information to conclusions that

1490

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

correspond with a preexisting belief or goal (Kunda, 1990).

Motivated reasoning may lead people to selectively misremember
their experiences in ways that reinforce their prior beliefs about
global warming. Evidence from a study of farmers in the
Midwestern U.S. supports this view, nding that farmers who
believed climate change was occurring were more likely to
accurately remember a local warming trend over the past ve
years (Weber, 1997). The increasing ideological polarization of
global warming in the U.S. may provide further additional
incentives for motivated reasoning in seasonal climate perceptions
(McCright and Dunlap, 2011). As global warming beliefs become
strongly associated with certain ideological orientations, ideologically motivated cognition gives people strong reasons to perceive
their experiences in ways that support their worldview (Kahan,
2013; Kahan et al., 2011).
Motivated reasoning may not inuence memories of all types of
weather uniformly, however; memories of local temperatures may
be more inuenced by global warming beliefs than memories of
local precipitation because of the intuitive association between
global warming and temperature (Leiserowitz, 2006, 2005). Global
climate models project a long-term rise in local average
temperatures across most of the world, resulting in a rise in
global mean temperature of 24 8C by 2100 relative to the late 20th
century (Meehl et al., 2007). Projected changes in precipitation,
however, are much more place-dependent and comparatively
uncertain (Christensen et al., 2007). These differences imply a more
intuitive association between the constructs of local temperatures
and global warming. The relationship between temperature and
global warming is also supported by ndings that ambient air
temperature signicantly affects global warming beliefs and
concern (Joireman et al., 2010; Li et al., 2011; Risen and Critcher,
2011) and by cognitive associations between global warming and
heat (Smith and Leiserowitz, 2012; Whitmarsh, 2009). Recent
research has also found that perceptions of local temperature show
a greater divergence from measurements of actual local weather
than do perceptions of hydrometeorological phenomena such as
ooding and drought (Goebbert et al., 2012). Motivated reasoning
may thus be more likely to bias perceived experience of local
temperature than that of local precipitation.
Our analysis begins with an exploration of the spatial pattern of
U.S. seasonal climate perceptions. We hypothesize that the spatial
distribution of perceptions of local climate conditionsthe
conditions to which individuals are exposed in the immediate
area of their householdwill broadly coincide with patterns of
recorded temperature and precipitation anomalies. Perceptions
should thus exhibit non-random spatial patterns, with perceptions
tending to be similar among individuals nearer to each other and
dissimilar among individuals at greater distances, since nearby
individuals would be more likely to share the same experiences. If
seasonal climate perceptions do exhibit a non-random spatial
distribution then we would further expect their distribution to
coincide with that of measured local climate anomalies during the
season of interest, under the assumption that people have
generally accurate perceptions of seasonal climate as recorded
in instrumental data.
Our second hypothesis relates to the role of motivated
reasoning in seasonal climate perceptions due to sets of beliefs
about global warming. Survey research in the U.S. has identied six
segments of the American public who have relatively homogenous
levels of concern, knowledge, and related beliefs about global
warming (Leiserowitz et al., 2009; Maibach et al., 2011). These
segments range along a spectrum from the Alarmed, who believe
climate change is occurring, perceive it to be a serious threat, and
support policies and personal behavior changes to mitigate the
threat, to the Dismissive, who believe climate change is not
occurring, do not perceive it to be a serious threat, and do not

support mitigation policies. Also along the spectrum are the

Concerned, who like the Alarmed believe climate change is
happening and a threat, but are not personally engaged with the
issue; the Cautious, who are uncertain whether climate change is
occurring or not; the Disengaged, who know little to nothing
about global warming; and the Doubtful, who tend to think
global warming is not happening, but if it is, it is not human caused,
and not a serious threat. We hypothesize that, if motivated
reasoning does inuence seasonal climate perceptions, then
respondents with different sets of beliefs about global warming
will exhibit different perceptions of local climate conditions, even
after controlling for the conditions experienced by each respondent. For instance, those who believe in the existence, certainty,
and threat of global warming (i.e. the Alarmed and the Concerned)
may tend to recall above-normal seasonal temperatures regardless
of actual conditions. Conversely, those who believe that global
warming is not happening, uncertain, or not a threat (i.e. the
Dismissive and the Doubtful) may be less likely to recall abovenormal seasonal temperatures. Motivated reasoning may also have
a stronger inuence on perceptions of climatic variables that are
intuitively associated with preexisting beliefs about global
warming, such as temperature, than on climatic variables that
are less associated with the construct of global warming, such as
precipitation.

3. Methods
This study is based on data from a nationally representative
survey of the United States in April and May 2011, conducted by
Knowledge Networks using a probability-based online panel.
Survey respondents were randomly sampled from a panel of over
50,000 members originally recruited using random-digit dialing
and address-based sampling. To ensure that the panel is nationally
representative, members without internet access receive a netbook and internet service. The survey was elded from April 22 to
May 11, 2011, with a total of 1,010 completed responses from
adults aged 18 or older. The survey had a recruitment rate of 15.7%
(contacted individuals who opt to join the panel), a prole rate of
60.9% (panel members who complete the prole survey), a
completion rate of 66.1% (invited panel members who completed
the survey), and a cumulative response rate of 6.3% (the product of
the recruitment, prole, and completion rates; see Callegaro and
DiSogra, 2008).
The geographic distribution of respondents roughly reected
that of the U.S. population. By state, the greatest number of
respondents lived in California (12%), followed by Texas (8%), New
York (8%), and Florida (7%). The global warming audience
segmentation was consistent with previous U.S. representative
surveys, with the majority of respondents classied as the
Concerned (27%) followed by the Cautious (25%), the Doubtful
(15%), the Alarmed (12%), the Disengaged (10%), and the
Dismissive (10%) (Leiserowitz et al., 2011).
Four survey items are the dependent variables in this study.
These items measure perceived departures from normal in average
temperature and precipitation during the preceding winter (winter
20102011) and the preceding summer (summer 2010). The
questions were presented as follows: (1) Has this winter in your
local area been warmer, colder, or no different than normal?, (2)
Has this winter in your local area brought more snow or rain, less
snow or rain, or was it no different than normal?, (3) Thinking
back to last summer, in your local area was it warmer, colder, or no
different than normal? and (4) Thinking back to last summer, in
your local area, did it bring more rain, less rain, or was it no
different than normal? The text of the survey did not specify a
normal period with respect to these items, since individual

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

denitions of normal are likely to be subjective and dependent

on individual experience.
The survey also contained a battery of items on global warming
beliefs, policy preferences, issue importance, and behavior that
was used to classify respondents into six like-minded global
warming audience segments using a 36-variable linear discriminant function (see Leiserowitz et al., 2009; Maibach et al., 2011).
The classication function was based on a latent class analysis of
nationally representative survey data collected in fall 2008.
Survey respondents were geocoded based on their street
addresses. To protect panel member condentiality geocoded
coordinates were randomly jittered within a radius of up to
0.15 km prior to analysis. Spatial climate data were obtained from
the PRISM climate mapping system, a high resolution monthly
dataset of the contiguous U.S. using a knowledge-based regression
model (Daly et al., 2002; PRISM, 2004). This analysis uses grids of
monthly mean temperature (derived as the average of the mean
monthly maximum and minimum temperature) and total precipitation at the 2.50 resolution (approximately 4 km between grid
points). For the purposes of this analysis, such high-resolution
gridded spatial climate data provide a more accurate spatial
representation of climate parameters than those available from
individual climate station records. These data represent a bestestimate of climate conditions in the immediate vicinity of
respondents households, since they account for ne-scale spatial
variations in climate such as rain shadows, temperature inversions,
and coastal effects that may not be apparent in station data or
coarsely gridded data. Since this analysis addresses perceptions of
the departure of local seasonal climate from long-term averages,
we derived anomalies for each month of the temperature and
precipitation datasets. Temperature anomalies were calculated as
the difference between the current months mean temperature and
the monthly mean temperature during the 19712000 period.
Precipitation anomalies, represented as a percentage of normal
precipitation, were calculated by dividing the current months
precipitation amount by the mean precipitation amount during the
period 19712000 and multiplying by 100. The choice of base
period for this analysis does not substantively affect the results,
since we examine relative anomalies between different sets of
respondents. Seasonal anomalies were derived by calculating a
three-month average based on winter (DecemberFebruary) and
summer (JuneAugust) climatological seasons.
3.1. Spatial analysis
If seasonal climate perceptions reect personal experience with
local climate conditions, then they should exhibit spatial patterns
signicantly different than what would be expected by random
chance. The spatial distribution of seasonal climate perceptions
would also be expected to coincide with the spatial distribution of
the climatic variable of interest during each respective season. To
examine the spatial distribution of the seasonal climate perceptions items, we mapped relative risk surfaces depicting the relative
probability density of each response. Rather than depicting the
probability density of each type of point alone, relative risk maps
represent the spatially varying probability of each possible
response conditional on the locations of all other responses. This
method accounts for the spatial inhomogeneity of the population
from which survey respondents are sampled. More specically, a
relative risk surface is derived from the ratio of bivariate kernel
density estimates for two sets of points (Hazelton and Davies,
2009). This ratio can then be used in hypothesis tests by calculating
a tolerance surface of p-values (Kelsall and Diggle, 1995).
In the current analysis, we created relative risk maps for each of
the possible responses to the seasonal climate perceptions items
using the spatstat (Baddeley and Turner, 2005) and sparr (Davies

1491

et al., 2011) R packages. Kernel densities were estimated by an

isotropic kernel smoother with optimal bandwidths selected by
cross-validation (Baddeley and Turner, 2005; Kelsall and Diggle,
1995). The optimal smoothing bandwidth for each map ranged
from 220 km to 517 km. Statistically signicant positive concentrations of responses were identied by calculating asymptotic pvalue surfaces based on a 50  50 grid (approximately 88 km
between grid points), with positive concentrations at the 95% and
99% condence level represented as contours superimposed on the
relative risk surface (Hazelton and Davies, 2009).
3.2. Bivariate and multivariate analyses
To examine the direct relationship between local conditions
and seasonal climate perceptions, we used the locations of
respondents to obtain monthly temperature and precipitation
data based on their household coordinates. We then summarized
and compared average climate conditions among respondents who
had similar seasonal climate perceptions using one-way analysis of
variance.
To investigate the effects of demographics, local temperature
and precipitation, and beliefs about global warming on seasonal
climate perceptions, we constructed two sets of four binary logistic
regression models. The dependent variable in the rst set of
models is a dichotomous variable indicating the perception of a
positive departure from normal seasonal temperature (warmer
than normal), as compared to a neutral or negative departure from
normal, during summer (models A1A2) and winter (models B1
B2). The dependent variable in the second set of models is a
dichotomous variable indicating the perception of a positive
departure from normal seasonal precipitation (more rain than
normal or more rain and snow than normal), as compared to a
neutral or negative departure, during summer (models C1C2) and
winter (models D1D2). In each set of models, model 1 includes
demographics and climate change beliefs as predictors, while
model 2 adds local temperature or precipitation anomalies.
Demographic variables include age (1829, 3044, 4559, 60+),
gender, educational attainment (less than high school, high school,
some college, bachelors degree or higher), race/ethnicity (White,
non-Hispanic; Black, non-Hispanic; other races, non-Hispanic; two
or more races, non-Hispanic; Hispanic), and political ideology
(conservative, moderate, liberal). Climate change beliefs are
represented by indicator variables classifying respondents into
one of the six audience segments (Alarmed, Concerned, Cautious,
Disengaged, Doubtful, Dismissive) with like-minded beliefs, risk
perceptions, and policy preferences about climate change (Maibach et al., 2011).
3.3. Seasonal context
The survey was elded in AprilMay 2011, following a period
with multiple record-breaking extreme weather events, including
tornadoes, oods, wildres, and drought in the U.S. (NCDC, 2011a).
The 751 tornadoes recorded in April 2011 set an all-time U.S.
record, while extensive rainfall and fast-melting snowpack caused
historic oods in the Ohio River valley and the lower Mississippi
River. March and April also brought abnormally dry weather in
Oklahoma, New Mexico, and Texas that caused extreme drought
conditions in large areas of those states (NCDC, 2011a).
For much of the U.S., the seasons examined in this study
contrast sharply with each other relative to historical averages. The
unusual weather of spring 2011 followed an unusually cold winter
of 20102011, both of which coincided with a relatively strong La
a that began in fall 2010. The December 2010 through February
Nin
2011 period was colder than average across most of the
Midwestern and eastern U.S. (Fig. 1a), particularly in the Southeast,

1492

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

Fig. 1. Mean temperature anomaly (panels A and B, top) and percent of normal precipitation (panels C & D, bottom) during climatological winter (DecemberFebruary) 2010
2011 and summer (JuneAugust) 2010. Data derived from monthly PRISM (2004) analysis.

where Florida experienced its tenth coldest winter in the 116-year

a
record (NCDC, 2011b). Winter 20102011, as in a typical La Nin
year, brought above-normal precipitation to the northern Midwest, the northeastern states, and the Ohio Valley, while the
Southern Plains and Southeastern states received below-normal
precipitation (Fig. 1c). By contrast, summer 2010 was the fourth
hottest summer on record in the U.S. as a whole, with the highest
temperature extremes occurring in the Mid-Atlantic and Southeast
(Fig. 1b). June through August 2010 was the hottest climatological
summer on record in Rhode Island, New Jersey, Delaware,
Maryland, Virginia, North Carolina, Tennessee, South Carolina,
Alabama, and Georgia. California and the Pacic Northwest were
not part of the pattern, experiencing normal or slightly below
normal temperatures. Precipitation during summer 2010 was
slightly below normal to slightly above normal across most of the
U.S. (Fig. 1d), with the northern Midwest ranking among the 10th
wettest summers on record (NCDC, 2010). There were also several
small regions with extremely below normal summer precipitation,
including southeastern Missouri and western Kentucky, the
California Central Valley, and southern Mid-Atlantic region.
The contrasting conditions between each season serve as a
natural quasi-experiment to examine perceptions of seasonal
climate. Summer 2010 was abnormally hot for muchbut not all
of the U.S., while the following winter was abnormally cold in a
similar swath of the country. It is also important to note that the
climatic extremes of spring 2011, when the survey was conducted,
may have inuenced survey responses about winter 20102011,
particularly regarding perceptions of seasonal precipitation in the
Ohio Valley and Northeast. These regions experienced persistent
above-normal rainfall from March through May 2011, which may
have inuenced respondents assessments of the overall average
precipitation for winter 20102011. This may particularly be the

case if an individuals understanding about what period constitutes the winter season does not correspond with the months of
climatological winter (DecemberFebruary). For example, astronomical winter in the northern hemisphere extends to the vernal
equinox on March 20, and snowfallassociated with winter
conditionsfrequently occurs after that date in the northern U.S.
Because of this disparity, we include temperature and precipitation
data for March and April 2011 in our analysis to control for the
effect of weather conditions immediately before and during the
survey.
4. Results and discussion
4.1. Descriptive survey results
Between 969 and 977 respondents in the contiguous U.S.
completed each of the four seasonal climate items (Table 1). The
majority of respondents reported that the previous winter had
been colder than normal (58%) and brought more rain and snow
than normal (59%). For the previous summer, the largest portion
(43%) of respondents reported that the season had been warmer
than normal, while nearly as many (42%) reported that the summer
had been no different from normal. A similar number (47%)
reported that precipitation during the previous summer had been
no different from normal.
Survey respondents experienced a wide range of local climate
conditions due to the spatial variability of climate in the U.S. On
average, local temperature anomalies among sampled respondents
differed more dramatically between the summer and winter
seasons than did local precipitation (Table 2). The average
respondent experienced above-normal temperatures and slightly
below normal precipitation in summer 2010, and below-normal

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

1493

Table 1
Seasonal climate perception items.
Temperature

Colder

No different than normal

Warmer

Winter 20102011a
Summer 2010b

603
136

61.8
14.0

259
420

26.6
43.2

113
417

11.6
42.9

Precipitation

Less snow or rain

n

Winter 20102011c
Summer 2010d

128
266

a
b
c
d

No different than normal

More snow or rain

13.2
27.2

222
484

22.9
49.5

619
227

63.9
23.2

Has this winter in your local area been warmer, colder, or no different than normal?
Thinking back to last summer, in your local area was it warmer, colder, or no different than normal?
Has this winter in your local area brought more snow or rain, less snow or rain, or was it no different than normal?
Thinking back to last summer, in your local area, did it bring more rain, less rain, or was it no different than normal?

temperatures and slightly below normal precipitation in winter

20102011.

were signicantly more likely to report more rain than normal,

coinciding with near record high precipitation in Iowa, Minnesota,
and Wisconsin (Fig. 2d).

4.2. Geographic patterns

The relative risk surfaces reveal substantively different spatial
patterns of responses across each of the four items. When asked
about the winter of 20102011 (Fig. 2a), respondents in the
northeastern states, extending from the Mid-Atlantic to New
England, were more likely to report that the previous winter was
colder than normal. This cluster coincides with colder than
normal conditions in the region during December 2010
February 2011. Respondents in the Northeastern U.S. also were
more likely to report more rain or snow than normal during the
previous winter, as did respondents in the upper Midwest and
the Pacic Northwest (Fig. 2c). A further cluster of respondents in
Colorado, New Mexico, and Texas reported less snow or rain than
normal during the winter. Although precipitation was indeed
below normal in the Southwest during December 2010 through
February 2011, these assessments may have also been magnied
by the drought and wildres in the region during March and
April 2011.
When asked to remember the summer of 2010, respondents in
the eastern U.S. were signicantly more likely to report warmer
than normal conditions, particularly along the Atlantic coast from
Rhode Island to North Carolina and inland to northern Georgia,
Alabama, and Tennessee, which coincides with the region affected
by that summers record heat wave (Fig. 2b). Respondents along
the Pacic coast, who were unaffected by the heat wave, were
signicantly more likely to report that the summer was colder than
normal. Respondents in a band from New Mexico and Colorado to
Arkansas and Mississippi were signicantly more likely than
others to report less rain than normal during summer 2010. Finally,
respondents in the Pacic Northwest and the northern Great Plains

4.3. Bivariate analysis of seasonal climate perceptions by local

temperature and precipitation anomalies
Differences in perceptions of seasonal climate conditions
corresponded with relative differences in measured climate
conditions, as would be expected if respondents had based their
perceptions on their local conditions. One-way analyses of
variance indicated signicant differences in three-month temperature anomalies among respondents to the items for winter
temperature, F(2,972) = 4.82, p = 0.01 (Fig. 3a). When asked about
temperatures during the most recent winter, individuals who
responded that the winter was warmer than normal were indeed
living in places where, during the months of climatological winter
(DecemberFebruary), the mean temperature anomaly was higher
than among those who responded that the winter was colder than
normal (Tukey HSD; mean difference: 0.26 8C, p = 0.01).
Summer temperatures also differed signicantly among
respondents with different perceptions of summer temperature,
F(2,970) = 33.98, p < 0.001 (Fig. 3b). The mean JuneAugust 2010
local temperature anomaly among those who responded that the
summer was warmer than normal was signicantly greater than
the anomaly among those who responded that the winter was
colder than normal (mean difference: 0.78 8C, p < 0.001). The mean
local temperature anomaly among those who responded that the
summer was no different than normal also differed signicantly
from that of those who responded that the summer was colder
than normal (mean difference: 0.69 8C, p < 0.001).
Likewise, there were signicant differences in relative December
2010February 2011 precipitation among respondents who perceived different amounts of winter precipitation, F(2,970) = 32.45,

Table 2
Summary statistics for temperature anomaly and percent of normal precipitation at survey respondent coordinates. Data derived from PRISM (2004), based on 19712000
normals.

Temperature anomaly (8C)

JuneAugust 2010
December 2010February 2011
MarchApril 2011
Percent of normal precipitation
JuneAugust 2010
December 2010February 2011
MarchApril 2011

Mean

Median

1.2
0.9
0.1

1.4
1.1
0.3

1.0
0.9
1.0

2.5
3.3
3.3

3.0
2.1
3.3

85.5
16.9
62.7

92.7
94.2
106.0

93.0
88.7
109.0

40.0
39.4
48.5

0.6
11.8
0.6

216.9
342.1
273.4

40.5
33.2
57.3

Std. dev.

Minimum

Maximum

Percent above
19712000 mean

1494

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

Fig. 2. Relative risk surfaces of the probability of responses to each item, conditional on the overall spatial distribution of responses. Each map depicts the ratio of two bivariate
kernel density estimates, the rst for case locations (respondents who responded as labeled), and the second for control locations (all remaining respondents). Darker shading
indicates a higher probability of responding as labeled. Optimal kernel smoothing bandwidths were selected by cross-validation: winter temperature, 276 km; winter
precipitation, 220 km; summer temperature, 517 km; summer precipitation, 294 km. Light blue contour lines indicate signicantly elevated probabilities at p < .05 (upper
tailed). Dark blue contour lines indicate signicantly elevated probabilities at p < .01 (upper tailed). Signicance contours based on asymptotic p-value surfaces (Hazelton and
Davies, 2009; Kelsall and Diggle, 1995). Dots indicate approximate household locations of labeled respondents. (For interpretation of the references to color in gure legend,
the reader is referred to the web version of the article.)

p < 0.001 (Fig. 3c). The mean December 2010February 2011

percent of normal precipitation among those who responded that
the winter had more rain and snow than normal was signicantly
greater than among those who responded that the winter was no
different than normal (mean difference: 18.4 percentage points,
p < 0.001), or had less snow or rain than normal (mean difference:
22.2 percentage points, p < 0.001). Signicant but smaller differences were present in mean JuneAugust 2010 precipitation for
those who perceived different amounts of summer precipitation,
F(2,974) = 3.54, p = 0.03 (Fig. 3d). The mean percent of normal
precipitation was signicantly greater among those who responded
that the summer had more rain than normal than among those who
responded that the summer was no different than normal (mean
difference: 8.5 percentage points, p = 0.02).

4.4. Multivariate analysis

4.4.1. Effect of local climate conditions on perceptions of seasonal
temperature and precipitation
Perceptions of a warmer-than-normal summer and winter were
related to local climatic conditions during each season (Table 3).
Despite the dramatic differences between winter 20102011 and
summer 2010, the effect of the local three-month temperature
anomaly on the perception that the season was warmer than normal
was positive and signicantly different from zero for both seasons.
Holding all other variables constant at their medians, a shift from the
5th percentile to the 95th percentile in the DecemberJanuary
temperature anomaly (from 2.2 8C to 0.5 8C) increased the
probability of recalling the winter as warmer than normal by

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

1495

Fig. 3. Time series plots showing mean climate conditions among groups of respondents to the seasonal climate perceptions items. Values are three-month moving averages.
Panels A (winter temperature) and B (summer temperature) report mean temperature anomaly (8C) by response; panel C (winter precipitation) and D (summer precipitation)
report mean percent of normal precipitation by response. Shaded gray box indicates the season referenced by each item. Vertical dotted line indicates the date of the survey.
Data derived from PRISM (2004).

1496

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

Table 3
Logistic regression models predicting perceived positive departure from normal in seasonal temperature, including climate change belief segments and local temperature and
precipitation anomalies.
Dependent variable

Winter, warmer

Model

A1

Summer, warmer
A2

B1

0.46 (0.45)

Demographics
Gender (female)a
Age (3044)b
Age (4559)b
Age (60+)b
Education (high school)c
Education (some college)c
Education (bachelors or higher)c
Black, non-Hispanicd
Other races, non-Hispanicd
Hispanicd
Two or more races, non-Hispanicd
Conservativee
Liberale

0.28
0.11
0.27
1.06
0.97
0.67
0.72
0.81
0.60
0.35
0.74
0.05
0.05

(0.21)
(0.29)
(0.29)
(0.33)**
(0.34)**
(0.33)*
(0.35)*
(0.30)**
(0.50)
(0.32)
(0.59)
(0.27)
(0.27)

0.28
0.14
0.37
1.14
0.99
0.78
0.78
0.82
0.57
0.04
0.56
0.08
0.03

(0.21)
(0.30)
(0.29)
(0.34)***
(0.33)**
(0.33)*
(0.36)*
(0.31)**
(0.51)
(0.35)
(0.61)
(0.27)
(0.27)

0.19
0.10
0.15
0.06
0.02
0.14
0.73
0.20
0.40
0.57
0.48
0.13
0.08

(0.14)
(0.22)
(0.21)
(0.21)
(0.25)
(0.25)
(0.27)**
(0.23)
(0.39)
(0.24)*
(0.49)
(0.17)
(0.18)

0.20
0.14
0.19
0.03
0.09
0.01
0.51
0.12
0.53
0.44
0.48
0.11
0.04

(0.14)
(0.22)
(0.21)
(0.21)
(0.25)
(0.26)
(0.27)
(0.23)
(0.41)
(0.25)
(0.49)
(0.17)
(0.18)

Climate change beliefs

Concernedf
Cautiousf
Disengagedf
Doubtfulf
Dismissivef

0.68
0.77
0.20
0.90
0.46

(0.33)*
(0.34)*
(0.40)
(0.40)*
(0.43)

0.57
0.70
0.14
0.86
0.39

(0.33)
(0.34)*
(0.40)
(0.41)*
(0.44)

0.29
0.49
0.74
1.20
1.65

(0.23)
(0.24)*
(0.29)*
(0.27)***
(0.32)***

0.25
0.46
0.72
1.16
1.67

(0.23)
(0.23)
(0.29)*
(0.27)***
(0.32)***

0.18
0.89
0.24
0.12

(0.14)
(0.26)***
(0.12)*
(0.29)

Local climate
Mean temperature anomaly, MA 2011 (8C)
Percent normal precipitation, MA 2011 (/100)
Mean temperature anomaly, DJF 201011 (8C)
Percent normal precipitation, DJF 201011 (/100)
Mean temperature anomaly, JJA 2010 (8C)
Percent normal precipitation, JJA 2010 (/100)
n
Deviance (2*log-likelihood)
AIC
Nagelkerke pseudo r2
Spatial autocorrelation of residualsg
Change in residual spatial autocorrelation

0.65 (0.61)

B2

Intercept

0.58 (0.35)

0.58 (0.43)

0.17 (0.10)
0.20 (0.17)

0.19 (0.09)*
0.25 (0.19)
975
650
688
0.10

975
633
679
0.13

973
1253
1291
0.10

973
1236
1276
0.12

0.05

0.03
37.4%

0.05

0.03
48.6%

Unstandardized regression coefcients (standard errors).

*
p < 0.05.
**
p < 0.01.
***
p < 0.001.
a
Reference category is male.
b
Reference category is 1829.
c
Reference category is less than high school.
d
Reference category is White, non-Hispanic.
e
Reference category is Moderate.
f
Reference category is Alarmed.
g
Univariate global Morans I. Distance weights based on 314 km threshold.

6 percentage points. The average temperature anomaly in the two

months prior to the survey, March and April 2011, did not have a
signicant effect on the perception that the winter had been warmer
than normal. Regarding summer temperature, a shift from the 5th
percentile to the 95th percentile in the JuneAugust temperature
anomaly (from 1.0 8C to 2.3 8C) increased the probability of
recalling the summer as warmer than normal by 15 percentage
points. The addition of local climatic variables improved the
predictive ability of the models of perceived summer and winter
temperature, as illustrated by the reduction in the AIC statistic and
residual spatial autocorrelation in each model.
Local climate conditions were also related to perceptions of
above-normal precipitation during the summer and winter
seasons (Table 4). The perception that the winter had more rain
and snow than normal was predicted by both temperature and
precipitation anomalies during DecemberFebruary, as well as by
precipitation during the two months immediately preceding the
survey. Holding all other variables constant, a shift from the 5th to
the 95th percentile in the local DecemberFebruary departure

from normal precipitation (from 46.6 to 176.1% of normal)

increased the probability of recalling the winter as having more
snow or rain than normal by 25 percentage points. Similarly, a shift
from the 5th to the 95th percentile in the MarchApril 2011
departure from normal precipitation (from 17.5 to 190.0% of
normal) increased the probability of recalling that the winter as
having more snow or rain than normal by 41 percentage points.
Regarding summer precipitation, a shift from the 5th to the 95th
percentile in the JuneAugust departure from normal precipitation
(from 7.8 to 152.8% of normal) increased the probability of
recalling that the summer had more rain than normal by 18
percentage points. As in the models predicting perceived seasonal
temperature, adding local climatic variables improved the predictive ability of the models of perceived winter and summer
precipitation.
4.4.2. Effect of global warming beliefs on perceptions of temperature
This section examines evidence for motivated reasoning in
seasonal climate perceptions, specically addressing the role of

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

1497

Table 4
Logistic regression models predicting perceived positive departure from normal in seasonal precipitation, including climate change belief segments and local temperature
and precipitation anomalies.
Dependent variable

Winter, more snow or rain

Model

C1

Intercept
Demographics
Gender (female)a
Age (3044)b
Age (4559)b
Age (60 + )b
Education (high school)c
Education (some college)c
Education (bachelors or higher)c
Black, non-Hispanicd
Other races, non-Hispanicd
Hispanicd
Two or more races, non-Hispanicd
Conservativee
Liberale
Climate change beliefs
Concernedf
Cautiousf
Disengagedf
Doubtfulf
Dismissivef
Local climate
Mean temperature anomaly, MA 2011 (8C)
Percent normal precipitation, MA 2011 (/100)
Mean temperature anomaly, DJF 201011 (8C)
Percent normal precipitation, DJF 201011 (/100)
Mean temperature anomaly, JJA 2010 (8C)
Percent normal precipitation, JJA 2010 (/100)
n
Deviance (2*log-likelihood)
AIC
Nagelkerke pseudo r2
Spatial autocorrelation of residualsg
Change in residual spatial autocorrelation

Summer, more rain

C2

0.60 (0.35)

D1

D2

2.06 (0.49)***

0.73 (0.38)

1.69 (0.52)**

0.09
0.29
0.47
0.50
0.11
0.14
0.20
0.38
0.63
0.77
0.88
0.04
0.03

(0.14)
(0.22)
(0.21)*
(0.21)*
(0.25)
(0.26)
(0.27)
(0.23)
(0.39)
(0.23)***
(0.44)*
(0.17)
(0.19)

0.11
0.47
0.70
0.69
0.23
0.12
0.14
0.33
1.05
0.55
0.72
0.00
0.19

(0.15)
(0.24)*
(0.23)**
(0.22)**
(0.28)
(0.28)
(0.29)
(0.25)
(0.42)*
(0.28)*
(0.49)
(0.19)
(0.20)

0.25
0.10
0.20
0.13
0.33
0.37
0.70
0.18
0.33
0.13
0.33
0.05
0.13

(0.16)
(0.25)
(0.23)
(0.24)
(0.27)
(0.27)
(0.29)*
(0.26)
(0.42)
(0.25)
(0.57)
(0.19)
(0.20)

0.23
0.15
0.25
0.10
0.30
0.33
0.70
0.31
0.42
0.37
0.26
0.05
0.09

(0.16)
(0.25)
(0.24)
(0.24)
(0.28)
(0.28)
(0.30)*
(0.26)
(0.44)
(0.27)
(0.58)
(0.20)
(0.21)

0.16
0.26
0.28
0.52
0.67

(0.25)
(0.25)
(0.31)
(0.27)
(0.30)*

0.03
0.43
0.32
0.70
0.82

(0.27)
(0.27)
(0.33)
(0.29)*
(0.33)*

0.32
0.25
0.85
0.21
0.40

(0.25)
(0.26)
(0.36)*
(0.29)
(0.34)

0.38
0.29
0.80
0.26
0.40

(0.26)
(0.27)
(0.36)*
(0.30)
(0.35)

0.04
1.45
0.20
1.08

(0.11)
(0.20)***
(0.10)*
(0.26)***

0.21 (0.11)
0.34 (0.19)

0.09 (0.10)
0.69 (0.21)**
969
1216
1254
0.07

969
1070
1116
0.25

977
1039
1077
0.03

977
1008
1054
0.08

0.28

0.16
44.2%

0.05

0.02
50.8%

Unstandardized regression coefcients (standard errors).

*
p < 0.05.
**
p < 0.01.
***
p < 0.001.
a
Reference category is male.
b
Reference category is 1829.
c
Reference category is less than high school.
d
Reference category is White, non-Hispanic.
e
Reference category is Moderate.
f
Reference category is Alarmed.
g
Univariate global Morans I. Distance weights based on 314 km threshold.

beliefs about global warming. The effect of global warming beliefs

on seasonal climate perceptions varied with the season and
climatic variable of interest, with the largest effects on perceptions
of the previous summers temperature. Even when controlling for
the local summer temperature anomaly, which was positive for
86% of the sample, respondents with different sets of global
warming beliefs had substantially different likelihoods of perceiving the summer as being warmer than normal. For example,
holding all other variables constant at their medians, respondents
in the Dismissive segment were 40% less likely than those in the
Alarmed segment, and 32% less likely than those in the Concerned
segment, to report experiencing a warmer-than-normal summer.
The standardized effects of belonging to the Doubtful and
Dismissive segments were larger than that of any included
demographic variable. The size of the effect also increased steadily
along the spectrum of belief certainty from those who believe
global warming is happening to those who do not believe it is
happening (Fig. 4). This pattern would be expected if individuals
who strongly believe that global warming is not happening are

more likely to bias their perceptions of their own experience

toward conditions that are inconsistent with global warming, e.g.
to report experiencing the summer as not being warmer than
normal despite having experienced above-normal temperatures.
If the effect of global warming beliefs on perceptions of seasonal
temperature is also present among those who strongly believe that
global warming is happening, then we would expect that these
individuals would tend to bias their perceptions of their own
experience toward conditions that are consistent with global
warming when faced with an experience that is supercially
inconsistent with global warming. In this case, those who strongly
believe that global warming is happening should report experiencing a warmer-than-normal winter despite extensive below-normal
temperatures. However, when controlling for the local temperature anomaly during winter 20102011, which was below normal
for 83% of the sample, global warming beliefs did not have a
consistent effect on the perception that the winter was warmer
than normal. During a season that was colder than normal for most
respondents, those in the Alarmed segment were no more likely

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

1498

Temperature perception ("warmer than normal")

Winter 2010-11

Summer 2010

Probability of response

80%

60%

40%

20%

ture, for both the winter and summer seasons. Holding winter
201011 precipitation and temperature anomalies constant at the
sample median, respondents in the Dismissive segment were 16%
less likely than those in the Alarmed segment to perceive the
winter as having more snow or rain than normal, while those in the
Doubtful segment were 13% less likely. Summer 2010 brought
above-normal precipitation to only 33% of the sample. Holding
precipitation and temperature anomalies constant at the sample
median, respondents in the Disengaged segment were 15% less
likely than those in the Alarmed segment to perceive the summer
as having more rain than normal. The differences between the
remaining segments were not signicant.
5. Discussion

0%
Alarmed

Concerned Cautious Disengaged Doubtful

Dismissive

Global warming belief segment

Fig. 4. Predicted probability of responding that winter 20102011 and summer
2010 were warmer than normal, by global warming belief segment and holding all
other variables constant at their sample medians. Based on model A2 (winter 2010
2011) and model B2 (summer 2010). Error bars represent 95% condence intervals.

than those in the Dismissive segment to perceive the winter as

warmer than normal, and only slightly more likely than those in
the Doubtful and Cautious segments to perceive the winter as
warmer than normal. This evidence points to an asymmetric effect
of global warming beliefs on perceptions of local climate
conditions, with those who do not believe global warming is
happening more likely to have biased perceptions of their local
climate when conditions are inconsistent with their beliefs.
To further explore the asymmetric effect of beliefs on
perceptions of local climate conditions, we ltered respondents
into two groups based on whether they were living in places with
above-normal or below-normal temperatures in summer 2010.
Within each group, we then compared the proportions of
respondents with different sets of global warming beliefs who
rated the season as warmer than normal. The proportion of
respondents who rated summer 2010 as warmer than normal
differed signicantly across global warming belief segments, x2(5,
N = 830) = 52.0, p < .000. Among respondents living in places with
above-normal summer temperatures, those in the Alarmed,
Concerned, Cautious, and Disengaged segments were signicantly
more likely than those in the Doubtful and Dismissive segments to
rate the summer as warmer than normal (p < .05). Among
Alarmed, Concerned, Cautious, and Disengaged respondents who
experienced a warmer-than-normal summer, 53% rated the
summer as warmer than normal, as opposed to 27% of Doubtful
and Dismissive respondents.
If those who believe climate change is happening and a threat
were to exhibit biased perceptions of seasonal climate conditions
when confronted with a colder-than-normal season, then we would
expect those who had experienced a colder-than-normal summer
during 2010 to be more likely to rate the season as being warmer
than normal. However, among those living in places with belownormal summer temperatures there were no signicant differences
across global warming belief segments in perceptions of summer
temperature, x2(5, N = 143) = 1.7, p = 0.89. These results provide
further support to the nding of an asymmetric effect of global
warming beliefs on perceptions of local climate conditions.
4.4.3. Effect of global warming beliefs on perceptions of precipitation
The effect of global warming beliefs on perceptions of local
precipitation was less extreme than for perceptions of tempera-

The results provide support for our rst hypothesis that the
spatial distribution of seasonal climate perceptions would coincide
with the spatial distribution of temperature and precipitation
anomalies. Spatial analysis of responses indicated that the
distribution of seasonal climate perceptions was non-random,
and analysis of local clusters found broad agreement between the
patterns of seasonal climate perceptions and patterns of local
climate anomalies. For instance, the extreme heat wave of summer
2010 in the eastern U.S. was visible in a map of the probability that
respondents would report that the summer was warmer than
normal, and the droughts of winter 20102011 in the Southwest
were visible in a local concentration of respondents who reported
that the winter was drier than normal. Furthermore, seasonal
climate conditions were signicantly different among groups with
different perceptions of seasonal temperature and precipitation.
However, the perception of a single climatic variable during a
specic season was not solely dependent on that variable,
particularly in regard to the perception of seasonal precipitation,
which was consistently associated with both local temperature
and precipitation. This nding reects the physical relationship
between temperature and the type and amount of local
precipitation. For instance, a colder-than-normal winter may lead
to more persistent and visible accumulations of snow.
The results also support our second hypothesis that global
warming beliefs may bias recollections of seasonal climate through
motivated reasoning. If global warming beliefs do affect seasonal
climate perceptions, we would expect perceptions to be biased
toward a state of the world that is consistent with beliefs about
global warming, e.g. toward seasons that are warmer than normal
for those who believe that global is happening, and seasons that are
not warmer than normal for those who doubt that global warming
is happening. Our results provide some evidence that global
warming beliefs bias perceptions of seasonal temperature, but the
effect is strongest among those who do not believe global warming
is happening. Controlling for local conditions, those who believed
in the existence and threat of global warming (the Alarmed and the
Concerned) were indeed signicantly more likely to recall the
summer of 2010 as being warmer than normal than those who
believed that global warming is not happening or is not a threat
(the Dismissive and the Doubtful). Conversely, for the winter of
20102011, a season that was colder than normal for most of the
population, those who believed in the existence and threat of
global warming were not signicantly more likely to recall the
winter as being warmer than normal than those who most strongly
believed that global warming is not happening and is not a threat
(the Dismissive), and only somewhat more likely than those who
believe global warming may be happening but is not a threat (the
Doubtful). These results therefore suggest that motivated reasoning may bias individual recollections of seasonal climate, but the
effect is asymmetric and more pronounced among those who most
strongly believe that global warming is not happening. The results

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

also suggest that the inuence of motivated reasoning may change

over time, with less of an effect on memories of more recent events
that are easier to recall, such as the conditions during the winter
that recently ended, than on memories of more distant events, such
as the conditions of the last years summer.
A number of recent studies have suggested that perceptions of
short-term weather patterns may shape climate change beliefs
(Borick and Rabe, 2010; Egan and Mullin, 2012; Goebbert et al.,
2012; Joireman et al., 2010; Li et al., 2011; Risen and Critcher,
2011), particularly among those who do not already hold strong
opinions about the issue (Hamilton and Stampone, 2013; Myers
et al., 2012). These ndings present challenges for climate change
communication, since they imply that attitudes among those who
have not yet made up their mind are sensitive to short-term
weather conditions that are only marginally informative about
long-term climate trends. Our results pose further challenges for
communication, since we show that those who strongly believe
that climate change is not happening are not likely to accurately
recall their experiences with local weather patterns when these
patterns are consistent with global climate change. In turn, they
are less likely to change their beliefs when confronted with realworld experiences consistent with the existence of climate change.
However, both this study and previous ndings imply that extreme
weather events can act as teachable moments for those who do not
have strongly held beliefs about climate change. We show that
excluding those who strongly believe climate change is not
happeningindividuals are likely to accurately recall their
experiences with seasonal climate conditions whether or not they
are consistent with long-term climate change. These results
suggest that many individual hold a mental model of climate
change that allows for attribution of extreme weather events to
long-term climate change, including both abnormally warm and
abnormally cold temperatures. This is in line with recent research
showing that people connect diverse extreme weather events to
climate change (Leiserowitz et al., 2013).
The ndings of this study are limited by the cross-sectional
nature of the survey data. Because data were collected at a single
point in time, we cannot denitively establish whether the
differing results between winter and summer seasons were due
to differences in how each season is perceived, or due to
respondents having more recently experienced the winter season.
In addition, we cannot rule out the possibility of reverse causality
in the model of global warming beliefs affecting seasonal climate
perceptions via motivated reasoning, since biased memories of
seasonal climate conditions may also have inuenced global
warming beliefs. Finally, although respondents were exposed to a
wide range of local climate conditions, exposure was not
distributed normally across the populationmost respondents
experienced a warmer-than-normal summer and a colder-thannormal winter. Further research conducted in different seasons,
and using longitudinal data, may provide additional evidence for
the effects of global warming beliefs on seasonal climate
perceptions identied in this study.
6. Conclusions
This study investigated how people perceive seasonal climate at
the local scale and how beliefs about global warming may
inuence subjective experiences of local climate conditions. To
address these questions, we compared judgments about two
seasons in the U.S. that had opposite extremes in temperature and
varying patterns of precipitation across much of the country.
Although previous research shows that populations with livelihoods directly reliant on local weather are able to perceive and
adapt to local climate variability and change, this study was one of
the rst to investigate whether the population within a large

1499

industrialized country such as the U.S. is able to detect changes in

their local climate. Our ndings indicate that perceptions of local
seasonal climate exhibit signicant non-random spatial patterns
that appear to derive from place-dependent experience with local
climate conditions. People generally are capable of characterizing
departures from normal in temperature and precipitation during
recent seasons, with several important caveats: as seasons recede
into the past and become harder to remember, the recollection of
whether they were warmer or colder than normal may become
more dependent on preexisting beliefs about global warming, with
perceptions tending to be biased in the direction consistent with
ones beliefs about global warming. Moreover, this effect appears
to be asymmetric, with larger biases evident among those who do
not believe global warming is happening.
While people are capable of recognizing and adapting to shortand long-term climate variability and change, these processes
depend, at least in part, on the changes in local climate conditions
being perceived by the affected individuals, either through
personal experience or description. As suggested by recent
research, climate change perceptions and beliefs arein part
dynamically constructed based on personal observations through
experience (Akerlof et al., 2013; Howe et al., 2013; Myers et al.,
2012). Even though personal experience with climate change may
elevate risk perceptions and belief certainty, the effect of personal
experience is dependent on the experience rst being recognized.
As we have shown, the recognition of local climate change can also
depend on ones prior beliefs about global warming.
Acknowledgements
This research was supported in part by the National Science
Foundation (Doctoral Dissertation Research Improvement Grant
1102785) and the Association of American Geographers. Survey
data collection was funded by the Surdna Foundation, the 11th
Hour Project, and the Grantham Foundation for the Protection of
the Environment. The authors acknowledge the contributions of
Brent Yarnal, Karl Zimmerer, Janet Swim, Frank Hardisty, Connie
Roser-Renouf, and Edward Maibach.

References
Adger, W.N., Agrawala, S., Mirza, M.M.Q., Conde, C., OBrien, K., Pulhin, J., Pulwarty,
R., Smit, B., Takahashi, K., 2007. Assessment of adaptation practices, options,
constraints and capacity. In: Parry, M.L., Canziani, O.F., Palutikof, J.P., van der
Linden, P.J., Hanson, C.E. (Eds.), Climate Change 2007: Impacts, Adaptation and
Vulnerability. Contributions of Working Group II to the Fourth Assessment
Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, pp. 717743.
Akerlof, K., Maibach, E.W., Fitzgerald, D., Cedeno, A.Y., Neuman, A., 2013. Do people
personally experience global warming, and if so how, and does it matter?
Global Environmental Change 21, 8191.
Baddeley, A., Turner, R., 2005. Spatstat: an R package for analyzing spatial point
patterns. Journal of Statistical Software 12, 142.
Borick, C.P., Rabe, B.G., 2010. A reason to believe: examining the factors that
determine individual views on global warming. Social Science Quarterly 91,
777800.
Callegaro, M., DiSogra, C., 2008. Computing response metrics for online panels.
Public Opinion Quarterly 72, 10081032.
Christensen, J.H., Hewitson, B., Busuioc, A., Chen, A., Gao, X., Held, I., Jones, R., Kolli,
a Rueda, V., Mearns, L., Menendez, C.G.,
R.K., Kwon, W.-T., Laprise, R., Magan
Raisanen, J., Rinke, A., Sarr, A., Whetton, P., 2007. Regional climate projections.
In: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor,
M., Miller, H.L. (Eds.), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge,
UK.
Daly, C., Gibson, W.P., Taylor, G.H., Johnson, G.L., Pasteris, P., 2002. A knowledgebased approach to the statistical mapping of climate. Climate Research 22, 99
113.
Davies, T.M., Hazelton, M.L., Marshall, J.C., 2011. Sparr: analyzing spatial relative
risk using xed and adaptive kernel density estimation in R. Journal of Statistical Software 39, 114.

1500

P.D. Howe, A. Leiserowitz / Global Environmental Change 23 (2013) 14881500

Deressa, T.T., Hassan, R.M., Ringler, C., 2011. Perception of and adaptation to climate
change by farmers in the Nile Basin of Ethiopia. Journal of Agricultural Science
149, 2331.
Egan, P.J., Mullin, M., 2012. Turning personal experience into political attitudes: the
effect of local weather on Americans perceptions about global warming.
Journal of Politics 74, 796809.
Goebbert, K., Jenkins-Smith, H.C., Klockow, K., Nowlin, M.C., Silva, C.L., 2012.
Weather, climate and worldviews: the sources and consequences of public
perceptions of changes in local weather patterns. Weather, Climate, and Society
4, 132144.
Hamilton, L.C., Stampone, M.D., 2013. Blowin in the wind: short-term weather and
belief in anthropogenic climate change. Weather Climate, and Society 5, 112
119.
Hansen, J., Marx, S., Weber, E.U., 2004. The Role of Climate Perceptions, Expectations, and Forecasts in Farmer Decision Making (IRI Technical Report No. 04-01)
IRI Technical Report. International Research Institute for Climate Prediction,
Palisades, New York.
Hansen, J., Sato, M., Ruedy, R., 2012. Perception of climate change. Proceedings of
the National Academy of Sciences of the United States of America 109, E2415
E2423.
Hartter, J., Stampone, M.D., Ryan, S.J., Kirner, K., Chapman, C.A., Goldman, A., 2012.
Patterns and perceptions of climate change in a biodiversity conservation
hotspot. PLoS ONE 7, e32408.
Hazelton, M.L., Davies, T.M., 2009. Inference based on kernel estimates of the
relative risk function in geographical epidemiology. Biometrical Journal 51,
98109.
Hitchings, R., 2011. Coping with the immediate experience of climate: regional
variations and indoor trajectories. Wiley Interdisciplinary Reviews: Climate
Change 2, 170184.
Howe, P.D., Markowitz, E.M., Lee, T.M., Ko, C.-Y., Leiserowitz, A., 2013. Global
perceptions of local temperature change. Nature Climate Change 3, 352356.
Hulme, M., Dessai, S., Lorenzoni, I., Nelson, D.R., 2009. Unstable climates: exploring
the statistical and social constructions of normal climate. Geoforum 40, 197
206.
Joireman, J., Barnes Truelove, H., Duell, B., 2010. Effect of outdoor temperature, heat
primes and anchoring on belief in global warming. Journal of Environmental
Psychology 30, 358367.
Kahan, D., 2013. Ideology, motivated reasoning, and cognitive reection. Judgment
and Decision Making 8, 407424.
Kahan, D.M., Jenkins-Smith, H., Braman, D., 2011. Cultural cognition of scientic
consensus. Journal of Risk Research 14, 147174.
Kelsall, J.E., Diggle, P.J., 1995. Non-parametric estimation of spatial variation in
relative risk. Statistics in Medicine 14, 23352342.
Koehler, J.J., 1996. The base rate fallacy reconsidered: descriptive, normative, and
methodological challenges. Behavioral and Brain Sciences 19, 117.
Kunda, Z., 1990. The case for motivated reasoning. Psychological Bulletin 108, 480.
Leiserowitz, A., Maibach, E., Roser-Renouf, C., 2009. Global Warmings Six Americas
2009: An Audience Segmentation Analysis. Yale University and George Mason
University. Yale Project on Climate Change Communication, New Haven, CT.
Leiserowitz, A., Maibach, E., Roser-Renouf, C., Feinberg, G., Howe, P., 2012. Extreme
Weather and Climate Change in the American Mind, Yale Project on Climate
Change Communication. Yale University and George Mason University, Yale
Project on Climate Change Communication, New Haven, CT.
Leiserowitz, A., Maibach, E., Roser-Renouf, C., Feinberg, G., Howe, P., 2013. Extreme
Weather and Climate Change in the American Mind: April 2013, Yale Project on
Climate Change Communication. Yale University and George Mason University,
Yale Project on Climate Change Communication, New Haven, CT.
Leiserowitz, A., Maibach, E., Roser-Renouf, C., Smith, N., 2011. Global Warmings Six
Americas in May 2011, Yale Project on Climate Change Communication. Yale
University and George Mason University, Yale Project on Climate Change
Communication, New Haven, CT.
Leiserowitz, A.A., 2005. American risk perceptions: is climate change dangerous?
Risk Analysis 25, 14331442.
Leiserowitz, A.A., 2006. Climate change risk perception and policy preferences: the
role of affect, imagery, and values. Climatic Change 77, 4572.
Li, Y., Johnson, E.J., Zaval, L., 2011. Local warming: daily variation in temperature
affects beliefs and concern about climate change. Psychological Science 22,
454459.
Maibach, E.W., Leiserowitz, A., Roser-Renouf, C., Mertz, C.K., 2011. Identifying likeminded audiences for global warming public engagement campaigns: an
audience segmentation analysis and tool development. PLoS ONE 6, e17571.
Marx, S.M., Weber, E.U., Orlove, B.S., Leiserowitz, A., Krantz, D.H., Roncoli, C.,
Phillips, J., 2007. Communication and mental processes: experiential and
analytic processing of uncertain climate information. Global Environmental
Change 17, 4758.
McCright, A.M., Dunlap, R.E., 2011. The politicization of climate change and polarization in the American publics views of global warming, 20012010. Sociological Quarterly 52, 155194.
Meehl, G.A., Stocker, T.F., Collins, W.D., Friedlingstein, P., Gaye, A.T., Gregory, J.M.,
Kitoh, A., Knutti, R., Murphy, J.M., Noda, A., Raper, S.C.B., Watterson, I.G., Weaver,
A.J., Zhao, Z.C., 2007. Global climate projections. In: Solomon, S., Qin, D.,
Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L.
(Eds.), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on
Climate Change. Cambridge University Press, Cambridge, UK.

Meehl, G.A., Tebaldi, C., 2004. More intense, more frequent, and longer lasting heat
waves in the 21st century. Science 305, 994997.
Meehl, G.A., Tebaldi, C., Walton, G., Easterling, D., McDaniel, L., 2009. Relative
increase of record high maximum temperatures compared to record low
minimum temperatures in the U.S. Geophysical Research Letters 36, 15.
Meyer, W.B., 2000. Americans and their Weather. Oxford University Press, New
York.
Meze-Hausken, E., 2004. Contrasting climate variability and meteorological
drought with perceived drought and climate change in northern Ethiopia.
Climate Research 27, 1931.
Myers, T.A., Maibach, E.W., Roser-Renouf, C., Akerlof, K., Leiserowitz, A.A., 2012. The
relationship between personal experience and belief in the reality of global
warming. Nature Climate Change 3, 343347.
NCDC, 2010. National Temperature and Precipitation Maps: June 2010August
2010. National Climatic Data Center, National Oceanic and Atmospheric Administration.
NCDC, 2011a. Spring 2011 U.S. Climate Extremes. National Climatic Data Center,
National Oceanic and Atmospheric Administration.
NCDC, 2011b. National Temperature and Precipitation Maps: December 2010
February 2011. National Climatic Data Center, National Oceanic and Atmospheric Administration.
Orlove, B., Roncoli, C., Kabugo, M., Majugu, A., 2010. Indigenous climate knowledge
in southern Uganda: the multiple components of a dynamic regional system.
Climatic Change 100, 243265.
Orlove, B.S., Chiang, J.C.H., Cane, M.A., 2000. Forecasting Andean rainfall and crop
o on Pleiades visibility. Nature 403, 6871.
yield from the inuence of El Nin
Osbahr, H., Dorward, P., Stern, R., Cooper, S., 2011. Supporting agricultural innovation in Uganda to respond to climate risk: linking climate change and variability
with farmer perceptions. Experimental Agriculture 47, 293316.
Peterson, T.C., Stott, P.A., Herring, S., 2012. Explaining extreme events of 2011 from a
climate perspective. Bulletin of the American Meteorological Society 93, 1041
1067.
PRISM, 2004. PRISM Climate Group, Oregon State University [WWW Document]..
http://prism.oregonstate.edu (accessed 05.01.12).
Rahmstorf, S., Coumou, D., 2011. Increase of extreme events in a warming world.
Proceedings of the National Academy of Sciences of the United States of America
108, 1790517909.
Rebetez, M., 1996. Public expectation as an element of human perception of climate
change. Climatic Change 32, 495509.
Risen, J.L., Critcher, C.R., 2011. Visceral t: while in a visceral state, associated states
of the world seem more likely. Journal of Personality and Social Psychology 100,
777793.
Roncoli, C., Ingram, K., Kirshen, P., 2002. Reading the rains: local knowledge
and rainfall forecasting in Burkina Faso. Society & Natural Resources 15,
409427.
Ruddell, D., Harlan, S.L., Grossman-Clarke, S., Chowell, G., 2012. Scales of perception: public awareness of regional and neighborhood climates. Climatic Change
111, 581607.
Sanchez-Cortes, M.S., Chavero, E.L., 2011. Indigenous perception of changes in
climate variability and its relationship with agriculture in a Zoque community
of Chiapas, Mexico. Climatic Change 107, 363389.
Smit, B., Blain, R., Keddie, P., 1997. Corn hybrid selection and climatic variability:
gambling with nature? Canadian Geographer/Le Geographe Canadien 41, 429
438.
Smith, N., Leiserowitz, A., 2012. The rise of global warming skepticism: exploring
affective image associations in the United States over time. Risk Analysis 32,
10211032.
Spence, A., Poortinga, W., Butler, C., Pidgeon, N.F., 2011. Perceptions of climate
change and willingness to save energy related to ood experience. Nature
Climate Change 1 1, 4649.
Strauss, S., Orlove, B.S., 2003. Weather, Climate, Culture. Berg, New York.
Thomas, D.S.G., Twyman, C., Osbahr, H., Hewitson, B., 2007. Adaptation to climate
change and variability: farmer responses to intra-seasonal precipitation trends
in South Africa. Climatic Change 83, 301322.
Tschakert, P., Sagoe, R., Ofori-Darko, G., Codjoe, S.N., 2010. Floods in the Sahel: an
analysis of anomalies, memory, and anticipatory learning. Climatic Change 103,
471502.
Weatherhead, E., Gearheard, S., Barry, R.G., 2010. Changes in weather persistence:
insight from Inuit knowledge. Global Environmental Change 20, 523528.
Weber, E.U., 1997. Perception and expectation of climate change: precondition for
economic and technological adaptation. In: Bazerman, M., Messick, D., Tenbrunsel, A., Wade-Benzoni, K. (Eds.), Psychological Perspectives to Environmental and Ethical Issues in Management. Jossey-Bass, San Francisco, CA, pp. 314
341.
Weber, E.U., 2010. What shapes perceptions of climate change? Wiley Interdisciplinary Reviews: Climate Change 1, 332342.
West, C.T., Roncoli, C., Ouattara, F., 2008. Local perceptions and regional climate
trends on the Central Plateau of Burkina Faso. Land Degradation & Development
19, 289304.
Whitmarsh, L., 2008. Are ood victims more concerned about climate change than
other people? The role of direct experience in risk perception and behavioural
response. Journal of Risk Research 11, 351374.
Whitmarsh, L., 2009. Whats in a name? Commonalities and differences in public
understanding of climate change and global warming. Public Understanding of Science 18, 401420.