Printer Friendly

Communicating climate change: spatial analog versus color-banded isoline maps with and without accompanying text.

Effective climate change outreach requires a close fit between climate information, its mode of presentation, and its intended audience. This fit is in need of adjustment: among both climate scientists and the public, dissatisfaction with the current state of climate change communication is widespread. According to a fall 2008 survey, 82% of American adults think that they do not have enough information to "form a firm opinion" about global warming (Maibach, Roser-Renouf, and Leiserowitz 2009). Scientists similarly recognize the need to improve communication of the causes and consequences of climate change, with practitioners from environmental sciences (Leiserowitz 2006; Lorenzoni, Nicholson-Cole, and Whitmarsh 2007), geography (Slocum 2004; Moser 2007; Hulme 2008), landscape architecture (Sheppard 2005; Brody et al. 2008), psychology (Gifford 2008), and resilience studies (Tschakert and Dietrich 2010) all calling for better tools and methods for communicating this information to the public. This article considers some explanations for these difficulties communicating climate change information, tests whether spatial analogs can help overcome these difficulties, and assesses the results. It begins with a brief literature review of climate change perceptions and communication.

Climate change perceptions and communication

Barriers to understanding and engagement

Climate scientists' calls for improving climate change communication generally reflect two closely related goals: (1) improving public awareness and understanding of the science of climate change, including causes and impacts; and (2) increasing personal engagement with climate change and its possible impacts (Lorenzoni, Nicholson-Cole, and Whitmarsh 2007). The first goal hews closely to the "deficit model" of scientific communication, which attributes people's skepticism about science to gaps or distortions in their understanding of the relevant scientific facts (Sturgis and Allum 2004). Meanwhile, the second goal suggests a need to move beyond understanding to engagement. Following Lorenzoni, Nicholson-Cole, and Whitmarsh (2007), engagement is defined as one's personal state of cognitive, affective, and behavioral connection with climate change. Communicating to increase engagement thus requires consideration of how scientific information is contextualized by personal experience and societal values in ways that help shape people's thoughts, feelings, and actions (Lorenzoni, Nicholson-Cole, and Whitmarsh 2007). To achieve behavioral engagement, communicators need to go beyond understanding and concern by taking steps to develop the public's motivation and capacity to adapt to and mitigate climate change; as noted in Grothmann and Patt (2005), these efforts must address both perceived self-efficacy and adaptation costs.

Several studies have attempted to explain why many people have difficulty understanding and personally engaging with climate change. Some explanations suggest that scientists have failed to tailor climate change information to their audiences' characteristic modes of thought. For example, the dual process theory posits that people use two modes of thought to understand risks and other stimuli: analytic processing and experiential processing (Evans 2003; Slovic et al. 2004). Analytic processing is more deliberate, consciously employs formal tools such as statistical analysis and algorithms, and is the preferred mode of scientific thought; experiential processing is more immediate, is grounded in imagery and affect, and is often used in intuitive or "spur-of-the-moment" decision making (Slovic et al. 2004). Several authors have therefore suggested that the communication of climate change information could be improved by translating scientists' analytic conclusions into experiential modes, which are potentially more concrete, immediate, and memorable, and thus more likely to prompt behavioral change (Sheppard 2005; Leiserowitz 2006; Marx et al. 2007; CRED 2009).

Other explanations for the public's failure to understand and engage with climate change have identified cognitive barriers that can blunt perceptions of climate change and its projected impacts. One frequently mentioned barrier is discounting. Many people believe that any negative impacts of climate change will primarily be felt in the distant future, by people who live far away or by non-human nature; they therefore tend to downplay the personal importance of climate change (Nicholson-Cole 2005; Leiserowitz 2007; Lorenzoni, Nicholson-Cole, and Whitmarsh 2007; APA 2009). To overcome this tendency, several authors have recommended a shift from a global climate change discourse toward a local discourse that shows how climate is embedded in communities and how changing climates will affect people's daily lives; in short, how climate change is locally relevant and personally meaningful (see work by perceptions specialists such as Slocum (2004), Leiserowitz (2007), and Hulme (2008), as well as work by geographers and other global change scientists on the regional impacts of climate change, such as the Global Change in Local Places project, Association of American Geographers (2003)).

Techniques for overcoming barriers to climate change communication

Visualization and mapping

Visualization and mapping may help improve climate change communication. As a possible means of making climate change more concrete and locally relevant, visualizations and maps may activate experiential processing, overcome discounting, and thereby improve climate change understanding and engagement. Experiential processing "encodes reality in images, metaphors, and narratives to which affective feelings have become attached" (Slovic et al. 2004). Graphical representations of the impacts of climate change on an environment that is well known to the viewer might therefore be expected to activate experiential processing and generate a strong affective response. For example, 2-D or 3-D visualizations can be used to show the projected impacts of climate change on local landscapes, driving home the personal relevance of climate change more "quickly and powerfully" than would be possible using text alone (Nicholson-Cole 2005; Sheppard 2005). According to cognitive science, static graphics can "facilitate comprehension, learning, memorization, problem solving, and communication, including inference of dynamic processes" (Fabrikant, Hespanha, and Hegarty 2010). Moreover, dual coding theories from psychology suggest that information that is simultaneously presented both textually and graphically (whether through pictures, maps, or graphs) is more likely to be understood and remembered than information that is presented using text or graphics alone, provided that certain design principles are met (Schnotz 2002). Indeed, graphics may prove particularly effective in increasing comprehension of complex subjects about which the audience has little prior knowledge, e.g., climate change (Schnotz 2002).

As a popular and accessible graphical form, maps may hold great potential for climate change communication, particularly to the extent that their depictions of climate change are iconic rather than abstract, and local to regional in scale (Nicholson-Cole 2005; Sheppard 2005). Maps are a popular means of depicting climate change and its impacts, as evidenced by their frequent inclusion in Intergovernmental Panel on Climate Change reports (e.g., Alley et al. 2007) and mass media publications (e.g., Dow and Downing 2006). The popularity of maps may be due in part to their generally high accessibility; unlike graphs and other "logical-pictures" that require many viewers to develop new cognitive schema, maps are often familiar and comprehensible to both novices and experts (Schnotz 2002, 114).

However, this may not be true of all climate change maps, which can be fairly abstract. First order impacts of climate change, such as changes in temperature or precipitation, are often mapped using bands of color between isolines that depict areas of equal change in temperature or precipitation (isallotherm and isallohyet mapping, respectively). To understand such maps, viewers must understand the abstract idea that isolines are lines of equal value (MacEachren 2004). Moreover, although the public is generally familiar with isoline weather maps from print, television, and online sources (Fabrikant, Hespanha, and Hegarty 2010), isoline maps of climate change differ from these weather maps. Unlike weather maps, which generally show daily extremes, these climate change maps often show differences in long-term averages. Applying the same schema used for viewing maps of meteorological extremes to maps of changes in multi-decadal averages could lead to discounting of climate change impacts, since small changes in averages can mask large changes in the severity and frequency of extreme events. Moreover, unlike weather maps, isoline maps of climate change require viewers to consider abstractions such as long-term averages that are not perceived directly and may therefore be difficult to engage with and understand. Indeed, research has shown that viewers often have difficulty understanding maps of statistical climate forecasts, such as probabilities of above- or below-normal precipitation (Ishikawa et al. 2005). These difficulties may also extend to maps of other statistical abstractions, such as changes in climatological averages.

While isoline maps of climate change present abstract ideas that may confuse viewers, filled variants of these maps use color in a way that may enhance engagement particularly when depicting temperature change. Filled-isoline maps of temperature-change projections typically use red, orange, and purple hues to represent areas with warmer projected temperatures; owing to the large temperature increases predicted to occur by the end of this century under most climate models and scenarios, these maps often acquire a "ball of fire" appearance. For audiences in the United States, this extensive use of reddish hues can carry connotations of activity, importance, and even danger (Madden, Hewett, and Roth 2000; Propen 2007; Tufte 1983). Thus, audiences may find filled-isoline maps of temperature change that feature many reddish hues more attention grabbing and engaging than other, less-colorful presentations of temperature change.

Temporal and spatial analogs

Similar to maps and visualizations, analogs have been suggested as a possible means of overcoming some of the barriers to effective climate change communication. Analogs for climate change are refined analogies for projected--but ultimately unknown--impacts and adaptations to climate change (Jamieson 1988, 81). Analogs may be temporal/historical (e.g., the Medieval Warm Period (Bolin et al. 1986) or 2003 European heat wave (Beniston and Diaz 2004) as analogs for European climate change) or spatial/regional (e.g., the climate of present-day central Spain as an analog for the climate of Paris in 50 years (Kopf, Ha-Duong, and Hallegatte 2008)) (Knight and Staneva 2004, 33). Analogs are grounded in real places and times, and therefore can provide a verisimilitude and narrative depth impossible to achieve using output from general circulation models (Glantz 1988; Jamieson 1988, 82). Several authors have argued that the depth of detail provided by analogs can help generate "moving stories" that aid communication and motivate action by policymakers, and especially by the public (Jamieson 1988; Kopf, Ha-Duong, and Hallegatte 2008). Accordingly, analogs may play a key role in translating information that appeals to analytical processing (such as changes in the long-term average temperatures) into information that appeals to experiential processing (such as accounts of direct or vicarious experiences with existing or past climates; Marx et al. 2007). Analogs may also help combat discounting, particularly in cases where the audience has direct experience of the region and time frame being used as an analog for future climate change.

However, analogs may not always address the problem of discounting. For example, the Medieval Warm Period would likely be seen as so temporally and culturally distant from modem Europe that its use as an analog could accentuate discounting effects. Analogical reasoning about climate change also faces other challenges. Jamieson (1988, 87-88) notes that analogies can become strained if climate researchers ignore differences and emphasize the points of similarity. Moreover, as Williams, Jackson, and Kutzbach (2007) point out, climate change may create "novel" climates for which there are no present-day analogs.

Purpose and research questions

The potential benefits of both graphical depictions and analogs for the communication of climate change may be combined in one method: spatial-analog mapping. Kopf, Ha-Duong, and Hallegatte (2008, 914) claim that their maps of spatial analogs for climate change in 12 European cities have significant communication value in light of their use "in teaching and in European popular science and mass media." However, the effectiveness of spatial-analog mapping in promoting understanding of and engagement with climate change information has yet to be compared with the effectiveness of other means of presenting the same climate change information, including filled-isoline mapping and plain text.

The purpose of this article is, therefore, to assess whether maps of spatial analogs for climate change are an effective means of overcoming the difficulties in communicating climate change information described above. To simplify the research design, the article considers only one aspect of climate change--projected increases and decreases in temperature--and one spatial analog. Accordingly, preliminary answers to the following questions are developed:

* Are climate change data easier to understand and more engaging if they are presented using spatial-analog maps as opposed to filled-isoline maps?

* Are these data more or less understandable or engaging when presented using only text, only mapping, or both maps and text?

Based on the review of the literature, it was expected that spatial-analog presentations would be easier to understand and more engaging than temperature-change presentations. It was also expected that presentations with maps would be easier to understand and more engaging than text-only presentations, and that the effects of dual coding would make presentations that combine maps and text more understandable and engaging than presentations of maps or text alone. These research questions compare only a handful of the many different ways of communicating climate change information. However, since spatial analogs are usually communicated using maps or text, and filled-isoline maps are a popular choice for mapping climate change, the research design covers many of the likely alternatives to spatial-analog mapping. Although the survey used to test the research questions focuses on climate change in only one place, the results suggest that we may need to reexamine widely held assumptions about the effectiveness of spatial analogs for communicating climate change.

Methods

To address the research questions, the author surveyed a convenience sample of residents from Pennsylvania's Centre Region (Figure 1). This convenience sample was drawn from members of a diverse group of environmentally minded Centre Region organizations. In combination with this sampling frame, the convenience sampling approach ensured that respondents were already at least somewhat interested in climate change, and therefore good candidates for testing whether maps and spatial analogs can push audiences beyond understanding and awareness and toward engagement. The survey depicted one climate model's projection of temperature change in the Centre Region in one of six different ways, one for each of six different survey forms. As shown in Table 1, each of the survey forms: (1) described temperature change in the Centre Region in terms of either a simple change in temperature or a spatial analog representative of this temperature change; and (2) presented this information using text, maps, or both text and maps. Respondents were randomly assigned to one of the six survey forms; random assignment was used to permit assessment of the effects of each of the six survey forms on respondents' opinions about temperature change in the Centre Region. The survey was designed to test the hypothesis that there is a relationship between the survey form to which a respondent was assigned and that respondent's assessed level of understanding and engagement.

Survey design and questions

Table 2 provides an overview of the sections of the questionnaire. Sections are listed in order and include a description of the content, format, and number of questions asked in that section. The table also indicates whether a section was administered to respondents before or after the experimental treatment was applied, i.e., before or after the presentation of one of the six scientific projections of temperature change in the Centre Region. Respondents were required to complete the sections in the order presented, and could not return to earlier sections to change their answers.

Sections 1 and 2

After reading an introduction to the survey and being assigned randomly to one of the six survey forms in Section 1, respondents completed Section 2, which included questions about their beliefs and feelings regarding climate change. Respondents used a Likert scale to indicate their level of agreement or disagreement with five statements about climate change. Statements evaluated by respondents included: "The climate appears to be changing more rapidly now than it was 100 years ago"; "Concern about climate change is exaggerated"; and "Climate change is a serious problem that requires urgent action." These questions established a baseline measure of respondents' beliefs and feelings about climate change.

Section 3

In Section 3, respondents completed two or three questions about how they expect average temperatures in the Centre Region to change over approximately the next 50 years. Respondents were asked these questions before viewing a scientific projection of temperature change so that their "pre-treatment" responses could be compared with their "post-treatment" responses to the same questions. Some of these questions were presented in different ways for different treatment groups to balance the use of graphics and interactivity across groups.

Section 3a asked all respondents to predict the sign and magnitude of temperature change in the Centre Region, first on a subjective scale and then in terms of a change in degrees Fahrenheit. Before asking any questions about temperature change in the Centre Region, respondents were shown a map of the Centre Region to familiarize them with its location (Figure 1). The first question in this section asked respondents to complete the following sentence: "I believe that average Centre Region temperatures in 2041-2070 will be--than average Centre Region temperatures in 1971-2000." To ensure comparability across treatment groups, all respondents were asked this question in the same way. Respondents answered on a seven-point Likert scale ranging from "much colder" to "much warmer." Because of the subjectivity inherent in using this scale to measure expected temperature change, respondents were also asked to match their subjective prediction with an expected change in temperature in degrees Fahrenheit.

Section 3b was only presented to respondents in the three spatial analog survey forms. It asked these respondents to select a spatial analog location that matched their predicted change in Centre Region temperatures. These pre-treatment spatial analog predictions were recorded for comparison with post-treatment spatial analog predictions that respondents made after viewing Section 4.

Section 4

Section 4 presented the experimental stimuli. After being asked to make predictions, respondents were shown maps and/or text presenting a scientific projection of temperature change in the Centre Region in terms of either a simple change in temperature or a spatial analog that is representative of this change in temperature. As shown in Table 1, this section of the questionnaire depicted temperature change in six different ways, one for each survey form. Two of these six ways of presenting temperature change are shown in Figures 2 and 3, which present map-and-text versions of the change-in-temperature and spatial-analog survey forms, respectively. The change-in-temperature map was designed to resemble the filled isoline maps used in weather reports, and was produced using NASA's Panoply software (see Schmunk 2010); the spatial-analog map was modeled on the Union of Concerned Scientists' map of "Migrating Climates," as depicted in their report, Climate Change in Pennsylvania (Figure 4). All six survey forms were produced using a projected temperature increase from 1971-2000 to 2041-2070 of 2.8[section]C (presented to respondents as "about 5[section]F"). This projection was based on output for North America from the Canadian Regional Climate Model-Third Generation Coupled Global Climate Model (CRCM-CGCM3) regional climate model, run using the A2 emissions scenario by the North American Regional Climate Change Assessment Program (NARCCAP) (Mearns et al. 2009). Ashville, North Carolina, was used as the analog for future Centre Region temperatures

because it was the most southerly spatial analog compatible with this projected temperature increase. The climate model and spatial-analog location used in the survey were chosen to maximize the apparent change in Centre Region temperatures. By displaying only the most dramatic changes, it was anticipated that any differences in the ability of the different survey forms to generate engagement would be magnified, and therefore easier to detect.

Section 5

Immediately following this depiction of Centre Region temperature change, Section 5 asked respondents to revisit their earlier predictions about temperature change (some respondents also revisited their spatial analog predictions in this section). Respondents from each treatment group updated their answers to the same set of questions that they answered in Section 3 of the survey. The spatial analog and objective temperature-change questions were repeated to assess how well respondents understood the depiction of Centre Region temperature change presented in Section 4. The subjective temperature-change question was repeated to gauge respondents' affective engagement with this depiction. Each question was altered slightly to remind respondents that they should answer in light of the temperature projection presented in Section 4. Because all survey forms presented a projected increase in Centre Region temperatures of about 5[section]F (either directly or in the form of a spatial-analog projection), respondents who correctly understood the projection--and did not strongly disagree with it--should have selected "4 to 6[section]F warmer" as their post-treatment prediction. The questions in Section 5 thus measured not only respondents' understanding, but also to some extent how engaging and persuasive they found the projection. However, both the sampling frame of environmentally minded organizations and the instructions to answer in light of the scientific projection were designed to limit the number of respondents who would completely disregard the projection when answering.

Section 6

Section 6 then asked respondents a series of questions designed to assess their affective and behavioral engagement with the depiction of temperature change in the Centre Region first presented in Section 4 of the questionnaire. To gauge affective engagement, respondents were asked the following question: "Over the next 60 years, how do you expect the above temperature prediction to affect your: well-being; health; finances; energy costs." For each of the areas affected, respondents answered on a Likert scale ranging from "very harmful" to "very beneficial." To gauge behavioral engagement, respondents were asked: "How large an impact do you expect the above temperature prediction to have on your decision making for each of the following activities over the next 60 years: installing or upgrading air conditioning systems; installing or upgrading heating systems; gardening; purchasing clothing; purchasing winter sporting equipment; purchasing summer sporting equipment; purchasing a car; purchasing a home." For each activity, respondents answered on a Likert scale ranging from "No impact" to "Very large impact." It was hypothesized that respondents' answers to these questions would be affected by both the treatment group to which they were assigned and their beliefs and feelings about climate change, as assessed in Section 2.

Section 7

Finally, Section 7 asked a series of demographic questions. Questions focused on measures of place-attachment, such as places and lengths of residence; in the interest of brevity, demographic questions asked for age and gender but did not include queries about educational attainment or other demographic factors. Answers were used as controls in regression models. Referral information was also requested. By asking respondents to describe how they learned about the survey (including the name of the

organization through which they received their invitation), the author was able to identify--and eliminate from analysis--25 respondents who stated that they had learned about the survey through an unsanctioned channel (e.g., the Internet). These responses were eliminated because they lacked a clear connection to the Centre Region and did not fit the chosen sampling flames (members of environmentally minded Centre Region organizations).

Survey implementation

Multiple sample flames were developed through contacts with several citizen organizations in the Centre Region. Because the survey was designed to test how different ways of presenting local climate change information affect understanding and engagement, the objective was to reach respondents that generally would not reject new information about climate change out of hand, but were instead interested in climate change in their community and open to learning more about it. To this end, the organizations that were approached met three criteria: (1) they were headquartered or had a chapter in the Centre Region, (2) their leadership had shown interest in climate change, and (3) their membership did not join primarily due to concern about climate change. These criteria helped locate organizations that were interested in supporting the work, but had members whose views on climate change were potentially quite diverse. For a list of the 11 partner organizations that agreed to participate, along with each organization's number of respondents, see Table 3.

An online questionnaire was used to survey members of partner organizations. Based on an estimated 2008 population for the Centre Region of 86,106 (Centre County Planning and Community Development Office 2008), a 95% confidence interval, a sampling error of plus-or-minus 5%, and a worst-case expected variation in answers (50-50 split), a target sample size of at least 382 individuals was established (Dillman 2007). The survey was administered using LimeSurvey (2011)--an online survey tool--and was open for 2 weeks, beginning on 2 December 2010. To inform their members of the survey, partner organizations sent pre-notification, invitation, and reminder emails.

Preparatory data analysis

Of the 565 responses to the survey, 444 valid responses were retained for analysis, thus exceeding the required minimum sample size. Other responses were rejected as incomplete or uninvited; a single corrupt response (answers recorded in wrong database fields) was also eliminated. Random assignment of the respondents to survey forms functioned correctly, with the percentage of respondents assigned to each of the six survey forms being within 3.5% of the expected 16.6% value. Random assignment to the six survey forms allowed analysis of the effect of form assignment independent of demographic or other factors.

All survey data were imported to PASW Statistics for Windows, Version 18.0 (SPSS Inc. 2009) for analysis. Aggregate scores were developed for each of the three survey sections that used multiple questions to measure a single sample characteristic: the five pre-treatment Likert scale questions assessing beliefs and feelings about climate change, the four post-treatment Likert scale questions assessing benefit/harm to lifestyle and finances, and the eight post-treatment Likert scale questions assessing perceived impacts on purchasing decisions and outdoor activities. These aggregate scores were developed so that statistical tests could assess the relationship between these characteristics and other characteristics measured by the survey.

Before aggregating scores for these three survey sections, answers were recoded so that the direction of the ordinal relationship was preserved across questions. Following recoding, a principal components analysis confirmed that the questions in each of the three survey sections did in fact measure one and only one characteristic of the sample. Rather than use a traditional principal components analysis (PCA), a categorical principal components analysis (CATPCA) was performed, which is similar to traditional PCA but better suited for use with categorical or ordinal variables. For each of the three sections, the component loadings identified by CATPCA were consistent with the assumption that each section measured a different sample characteristic. The value of Cronbach's alpha for each section was greater than 0.85; this was higher than the 0.7 level generally required to establish internal consistency (Nunnally 1978).

Based on the results from this CATPCA and Cronbach's alpha analysis, three aggregate scores were created for each respondent, one for each of the three sections. To aid comparison across treatment groups, each of the respondents' three aggregate scores were then assigned a quartile ranking. In the absence of measures of the mean and standard deviation (which are not appropriate for summed ordinal data), quartile rankings enable comparison of respondents' aggregate scores across both the three survey sections and treatment groups.

To capture how respondents' subjective temperature predictions changed from pre- to post-treatment, a change score was derived for this variable. Because of the ordinal nature of the subjective predictions, the change score records the direction of the change (decrease, no change, or increase), but not its magnitude. For example, a respondent whose response changed from "somewhat warmer" to "much warmer" would be coded as "increase," whereas a respondent whose response changed from "warmer" to "somewhat colder" would be coded as "decrease."

Results

Respondents' subjective and objective temperature-change predictions increased from pre- to post-treatment (Figure 5). However, this increase was not uniform across survey forms. Respondents using spatial-analog or text-only forms appeared to have lower post-treatment subjective temperature predictions than did those using change-in-temperature forms or forms with maps (Figure 6). Post treatment, respondents using change-in-temperature forms were much more likely than respondents using spatial-analog forms to predict that temperatures would be "4 to 6 degrees F warmer" (Figure 7).

Among respondents using spatial-analog forms, those who predicted higher temperature changes tended to predict more southerly analog locations. Respondents' pre-treatment predictions of change in temperature were positively correlated with the southerly component of the distance between the Centre Region and their pre-treatment analog predictions (Spearman's rho = 0.660, two-tailed p < 0.001, N = 130). Similarly, respondents' post-treatment predictions of change in temperature were also positively correlated with the southerly component of the distance between the Centre Region and their post-treatment analog predictions; however, the relationship was weaker (Spearman's rho = 0.379, two-tailed p < 0.001, N = 125). The southerly component of the shift in the location of respondents' spatial-analog predictions from pre- to post-treatment was also positively correlated with the corresponding change in respondents' objective temperature predictions (Spearman's rho = 0.408, two-tailed p < 0.001, N = 123).

High aggregate scores for pre-treatment beliefs and feelings about climate change suggest that most respondents were already very concerned about climate change before taking the survey (Figure 8A). Despite these high levels of concern, most respondents believed that the projected temperature change would be only somewhat beneficial or harmful to their lifestyle and finances (Figure 8B), and would not influence their future purchasing decisions and outdoor activities (Figure 8C). Respondents using temperature-change forms expected more harmful effects and somewhat greater impacts on purchasing decisions and outdoor activities than did respondents using spatial-analog forms; effect and impact scores appeared to have only a weak relationship with assignment to text-only, map-only, or map-and-text forms (Figure 9).

Explanatory models

Based on the relationships suggested by these results, the author developed regression models to explain how assignment to different survey forms affected three dependent variables: the direction of change in subjective temperature-change predictions from pre- to post-treatment; the post-treatment measure of how respondents expected the scientific temperature projection to affect their lifestyle and finances; and the post-treatment measure of how respondents expected this projection to influence their purchasing decisions and outdoor activities. These dependent variables were selected to assess respondents' understanding of--and engagement with--the survey's six different ways of presenting temperature change. Two survey form variables were used as independent variables in the analysis: one that compares spatial-analog forms and change-in-temperature forms, and one that compares forms containing maps with text-only forms. Chi-square tests of independence (not shown) supported grouping map-only and map-and-text survey forms together for comparison with text-only forms.

Logistic regression techniques were used to build the three explanatory models. Several control variables were considered for inclusion in each model. Quartile scores for pre-treatment concern about climate change were considered because several studies found that one's existing beliefs about climate change--e.g., whether one is an alarmist or a denier--strongly color how one interprets new information about climate change (Dunwoody 2007) and are also predictive of one's beliefs about the likely severity of local impacts of climate change (Leiserowitz 2007). Several demographic factors that have been found to correlate with concern about climate change were also considered, including age and gender (more concern among youth and women; Maibach, Roser-Renouf, and Leiserowitz (2009)) and measures of distance and place-attachment (discounting of perceptually distant climate change impacts; Nicholson-Cole (2005), Leiserowitz (2007), Lorenzoni, Nicholson-Cole, and Whitmarsh (2007), APA (2009)). Finally, chi-square tests of independence were used to test whether a relationship exists between the three dependent variables and each of the possible control variables. To avoid reducing statistical power, only control variables that show a significant relationship with a dependent variable are included in the regression model for that variable.

Model of direction of change in subjective temperature prediction

The results of a multinomial logistic regression for the direction of change (decrease or increase) in the subjective temperature prediction from pre- to post-treatment are shown in Table 4. The final model was a significantly better fit than the intercept-only model. The parameter estimates include beta coefficients for each non-reference category of each independent variable. The beta coefficient is the natural log of the odds ratio; the odds ratio is the odds of being in the category of interest divided by the odds of being in the reference category for the independent variable. Among respondents who decreased their predicted change in temperature, the beta coefficients were found to be significant (p < 0.05) for only the variable comparing spatial-analog and temperature-change survey forms. Among respondents who increased their predicted change in temperature, the beta coefficients were found to be significant (p < 0.05) for the variable comparing spatial-analog and temperature-change survey forms and for the variable comparing survey forms with maps and text-only survey forms.

Exponentiation of the beta coefficients was used to derive odds ratios. These odds ratios describe the strength and direction of the relationships between the independent and dependent variables. Among respondents who decreased their temperature prediction, odds of being in a spatial-analog form were 2.455 times the odds of being in a change-in-temperature form. Among respondents who increased their temperature prediction, odds of being in a temperature-change form were 1.795 times the odds of being in a spatial-analog form, and odds of being in a form with maps were 1.787 times the odds of being in a text-only form.

Model of effects of projected temperature change on lifestyle and finances

The ordinal logistic regression presented in Table 5 shows how survey-form assignment affects respondents' quartile scores for items assessing the effects of projected temperature change on lifestyle and finances (the dependent variable for this model). The final model was a significantly better fit than the intercept-only model. The beta coefficients were found to be significant (p < 0.05) for the variable comparing spatial-analog and temperature-change survey forms. The beta coefficient associated with the variable comparing forms with maps and text-only forms was found to be just beyond statistical significance (p = 0.054).

Odds ratios were derived by exponentiation of the additive inverse of the significant coefficients. These odds ratios are the odds that one level of an independent variable has a score on the dependent variable that is less than or equal to a given value, divided by the odds that the reference level of the independent variable has a score on the dependent variable that is less than or equal to this same value (Kleinbaum 2010, 470). For any given quartile score for effects on lifestyle and finances, the odds that respondents that used spatial-analog forms would be in that quartile or lower were 1.916 times the odds for respondents that used temperature-change forms. The odds for an equal or lower score for respondents that used text-only forms were 1.446 times the odds for respondents that used forms with maps.

Model of impacts of projected temperature change on purchasing decisions and outdoor activities

The ordinal logistic regression in Table 6 shows how form assignment affected respondents' quartile scores for items assessing the impact of projected temperature change on purchasing decisions and outdoor activities (the dependent variable for this model). The final model was a significantly better fit than the intercept-only model. The beta coefficient associated with the variable comparing spatial-analog forms and temperature-change forms was found to be just beyond statistical significance (p = 0.052). The beta coefficient associated with the variable comparing text-only forms and forms with maps was not found to be significant (p = 0.295). For any given quartile score for impacts on purchasing decisions and outdoor activities, the odds of respondents with spatial-analog forms being in that quartile or lower were 1.443 times the odds for respondents with temperature-change forms.

Discussion

The implications of these results for the communication of regional climate change projections to the public are considered below. In line with the two barriers to effective climate change communication described in Lorenzoni, Nicholson-Cole, and Whitmarsh (2007)--lack of understanding and lack of engagement--this discussion compares scores for understanding and engagement across the six treatment groups.

Comparison of change-in-temperature and spatial-analog forms

Although respondents using spatial-analog forms appeared to understand the spatial-analog concept, they were less likely than respondents using change-in-temperature forms to understand that the projection presented represented a "4-6[degrees]F" increase in Centre Region temperatures. Respondents using spatial-analog forms also scored lower than respondents using change-in-temperature forms on measures of affective engagement.

Understanding of spatial-analog concept

A key facet of the spatial-analog concept is that warmer climates are generally found at lower latitudes--in the northern hemisphere, farther south (Kopf, Ha-Duong, and Hallegatte 2008, 13). The results showed that predictions of a warmer climate were generally associated with more southerly spatial analog predictions. Based on the moderately strong correlations between respondents' objective temperature-change predictions and the southerly component of the distance from the Centre Region to their predicted analog locations, it is likely that most respondents did understand this facet of the spatial-analog concept.

Understanding of temperature projection

Respondents using change-in-temperature survey forms appeared to have a more precise understanding of the temperature projection than did respondents using spatial-analog survey forms. The most frequently selected post-treatment temperature-change prediction for both temperature-change and spatial-analog survey forms was "4-6[degrees]F warmer." Because this matches the projected temperature increase of about 5[degrees]F that was used to construct all presentations of temperature change used in the survey, it suggests that both groups of survey forms understood the presented change in temperature fairly well. However, the distribution of post-treatment temperature-change predictions is different for each group of survey forms: whereas the predictions of change-in-temperature respondents are tightly clustered in the "445[degrees]F warmer" category, predictions of spatial-analog respondents are almost evenly split between the "2-4[degrees]F warmer" and "4-6[degrees]F warmer" categories. Moreover, while almost 20% of spatial-analog respondents selected the "more than 6[degrees]F warmer" category, less than 5% of temperature-change respondents selected this category.

These results suggest that respondents that used spatial-analog forms understood that the analog presented represented temperatures that would probably be about 2 to 6[degrees]F higher than present-day Centre Region temperatures, but were less certain than respondents that used temperature-change forms that this increase in temperature would fall into the "4-6[degrees]F warmer" category. Because change-in-temperature forms presented the projected temperature increase of 5[degrees]F directly, it is not surprising that respondents using these survey forms were more likely to match this projection in their post-treatment predictions than were respondents using spatial-analog forms, who saw this projection only after it had been translated into an analog location.

Engagement with temperature projection

Respondents using spatial-analog forms exhibited lower levels of affective engagement with the projection than did respondents using temperature-change forms. Compared to respondents using temperature-change forms, respondents using spatial-analog forms were less likely to predict, post treatment, that the climate in the Centre Region would become "much warmer" and slightly more likely to predict that the climate would become "somewhat warmer" or "warmer" (Figure 6). This suggests that respondents using temperature-change forms expected a change in temperature that was subjectively more intense than that expected by respondents using spatial-analog forms. The expectation of a "much warmer" climate might be expected to generate a stronger affective response--and a corresponding change in the way respondents think about climate change--than the expectation of a "somewhat warmer" or "warmer" climate.

Similar conclusions can be drawn from the results of the logistic regression of the direction of the shift in these subjective temperature-change predictions from pre to post treatment (Table 4). Using a spatial-analog form more than doubled the odds that respondents would shift their expectations toward a lower temperature change, while being in a change-in-temperature form nearly doubled the odds that respondents would shift their expectations toward a higher temperature change. If increases in subjective temperature predictions are taken as a proxy for increased affective engagement, then assignment to a temperature-change form increased the odds that a respondent would engage with the temperature projection, whereas assignment to a spatial-analog form decreased these odds.

Respondents using spatial-analog forms were also less likely than respondents using temperature-change forms to expect that the temperature projection they viewed would have harmful effects on their lifestyle and finances. Both descriptive statistics and ordinal logistic regression supported this conclusion: compared to respondents using temperature-change forms, those using spatial-analog forms were about 10% less likely to score in the fourth quartile for these harmful effects (Figure 9) and had about twice the odds of an equal or lower quartile score (Table 5). By lowering the expected harm caused by temperature change, assignment to spatial-analog forms may not only decrease the intensity of concern about this change in temperature, but may also thereby decrease affective engagement.

Spatial-analog respondents' relatively low quartile scores for impacts on purchasing decisions and outdoor activities suggest that these respondents may also have lower levels of behavioral engagement. As shown in Figure 9, respondents using spatial-analog forms appear to have been slightly more likely than those in change-in-temperature forms to score in the first or second quartile, which are associated with less-disruptive behavioral impacts. Logistic regression supports this conclusion: respondents using spatial-analog forms were about one and a half times as likely as those using temperature-change forms to be assigned to an equal or lower quartile score, although this relationship was just beyond statistical significance at the 95% confidence level (Table 6). This suggests that assignment to spatial-analog survey forms may have had a small negative effect on behavioral engagement.

Reasons for differences in engagement

Several factors may have led respondents using change-in-temperature forms to engage more strongly with the projection than did respondents using spatial-analog forms. Respondents who dislike snowy winters may have found a shift to a North Carolina-like climate somewhat attractive, speculating that it would bring warmer weather and less snow. However, respondents using change-in-temperature forms predicted only slightly warmer subjective and objective temperatures than did respondents using spatial-analog forms, suggesting that an aversion to cold weather and an affinity for the North Carolina climate probably does not fully explain the different results for these two groups of survey forms.

While authors have suggested that spatial analogs' verisimilitude and narrative depth can help improve public engagement with climate projections, in this case the "moving story" told by the analog might have worked against engagement (Kopf, Ha-Duong, and Hallegatte 2008; Glantz 1988; Jamieson 1988, 82). Asheville, North Carolina, is a popular travel destination, noted for its mountain scenery, fine food and drink, and retirement amenities (Buncombe County Tourism Development Authority 2011); enthusiasm for the Asheville lifestyle may have tempered some respondents' concern about rising temperatures in the Centre Region. Although respondents were asked to consider only the effects of a change to Asheville-like temperatures, the penumbra of positive feelings and associations that may have surrounded Asheville for some respondents could have led them to predict effects on their lifestyle and finances that were more beneficial and less harmful than they might have been otherwise. This could be an example of anchoring, a process whereby exposure to one stimulus affects how one subsequently perceives and responds to another (Wilson 1996; Nicholls 1999). Spatial analog respondents may also have been less likely to expect harmful effects on their lifestyle and finances because the Asheville analog called to mind an intact social-ecological system that is already well adapted to warmer weather. In contrast, temperature-change respondents may have been more likely to consider the social and ecological disruptions that would likely be caused by a rapid shift to a warmer climate, and may therefore have found the projection more threatening.

The design of the change-in-temperature maps may also have contributed to the higher engagement scores for change-in-temperature forms. It may be that respondents' familiarity with filled-isoline weather maps from print, television, and online sources (Fabrikant, Hespanha, and Hegarty 2010) enabled them to interpret and engage with the temperature-change map more easily than the spatial-analog map, which presented temperature change in a way that was likely new to most respondents. The use of color on these maps may also have been a factor. While the spatial-analog map featured primarily white, gray, and blue (with a small red "X"), the temperature-change map featured large swaths of orange and red. Cross-cultural studies have shown that red is generally assigned a higher level of importance than other colors, and is considered "active," "hot," and "vibrant," whereas blue and green are considered "peaceful," "gentle," and "calming"; within the United States, red carries the additional association of symbolizing "warning" (Tufte 1983; Madden, Hewett, and Roth 2000; Propen 2007, 243, 246). These associations suggest that the extensive use of red on the temperature-change map may have conveyed a strong sense of danger and importance, and therefore made it more attention-grabbing than the spatial-analog map.

The temperature-change map also provided respondents with temperature projections for a broader geographic area than did the versions of the projection shown in other forms. Although respondents were asked to answer based solely on the projected temperature change for the Centre Region, respondents who saw the temperature-change map may have also taken into consideration the projected temperature change of 4 to 6[degrees]F that it showed for much of the eastern United States and Canada; this information was presented to respondents using the two change-in-temperature survey forms that included maps, but was not available to respondents using other forms.

However, closer analysis of the results suggests that these differences in map color and coverage cannot fully explain the differences in engagement observed between temperature-change and spatial-analog forms. Additional logistic regression analysis (not shown) was conducted to compare the responses for text-only Versions of the tem perature-change and spataal-analog survey forms. This analysis revealed that, among text-only respondents, those in spatial-analog forms were still about 1.9 times more likely than those in temperature-change forms to have a lower score on items assessing effects on lifestyle and finances (p = 0.039). Similarly, those in the text-only version of spatial-analog forms were about 2.8 times more likely than those in the text-only version of temperature-change forms to have a lower score on items assessing impacts on purchasing decisions and outdoor activities (p = 0.001). However, among text-only respondents, assignment to spatial-analog or temperature-change forms did not have a significant effect on whether respondents increased or decreased their temperature-change predictions (p = 0.857). With the exception of this last metric, these odds ratios are similar to those observed for all spatial-analog and temperature-change respondents, suggesting that differences other than map coverage and color contributed to the lower levels of engagement observed among spatial-analog respondents.

Comparison of text-only, map-only, and map-and-text forms

Respondents using text-only, map-only, or map-and-text forms appeared to understand the temperature projection equally well, with more than 40% of respondents from each of the three groups of survey forms choosing the "4[degrees]F to 6[degrees]F warmer" category. Respondents assigned to forms that featured maps demonstrated higher levels of cognitive and affective engagement than did respondents assigned to forms that featured only text. Respondents using forms with maps not only expected a subjectively more intense change in temperature than did respondents using text-only forms (post treatment, respondents using maps chose the "much warmer" category approximately twice as often as text-only respondents), but also were more likely to increase their subjective temperature prediction from pre to post treatment. Respondents that used forms with maps might therefore be expected to think more deeply and feel more strongly about temperature change in the Centre Region than would respondents that used text-only forms. However, the inclusion of maps did not appear to affect behavioral engagement.

The finding that forms with maps were somewhat more engaging than text-only forms is not surprising, as it is in line with existing literature that supports the ability of graphics in general--and maps in particular--to improve understanding and engagement. As noted by Fabrikant, Hespanha, and Hegarty (2010), cognitive science has found that static graphics can improve understanding and learning. By encoding abstract temperature projections in graphics, the survey forms that featured maps may have been more likely to activate experiential processing and generate a strong affective response (Slovic et al. 2004). The finding is also in line with Nicholson-Cole (2005) and Sheppard (2005), who claim that 2-D visualizations such as maps express the personal relevance of climate change more "quickly and powerfully" than would be possible using text alone.

Conclusions

In answer to the first research question, it was found that--contrary to expectations--temperature-change data were more understandable and engaging when expressed directly as a change in temperature than they were when expressed indirectly using a spatial analog. Respondents appeared to understand the spatial-analog concept, but were somewhat uncertain about the amount of temperature change represented by the analog. For engagement, the lower "effect" and "impact" scores for spatial-analog forms suggest that respondents found these forms less engaging than temperature-change forms.

In answer to the second research question, it was found that--confirming expectations--temperature-change data were somewhat more engaging when presented using a map or a map and text than they were when presented using only text. Forms with maps tended to have higher subjective temperature-change predictions and "effect". scores than text-only forms, but these differences were generally smaller and less significant than differences due to assignment to spatial-analog or temperature-change forms. Results for map-only forms did not differ significantly from results for map-and-text forms.

In light of these findings, climate change communicators who wish to improve understanding and engagement should strongly consider using maps where possible, but should exercise caution before using spatial-analog approaches. Maps and other graphical elements can be a quick and powerful way of improving comprehension of--and generating a strong affective response to--the message (Slovic et al. 2004; Nicholson-Cole 2005; Sheppard 2005). The use of appropriate maps may therefore help address misunderstandings about climate change, while also making climate change more personally relevant and meaningful. However, before using a spatial analog, communicators should consider the full range of cultural and emotional associations that the analog might evoke. If there are more than two viable analog locations, it might be worthwhile to conduct a pilot study to assess the full range of climatic, cultural, and emotional associations that that each analog carries for the audience. Another option might be to use a regional--rather than city-level--spatial analog; this could reduce the likelihood that the audience will key on the strong cultural associations that some cities carry.

Moreover, if communicators wish to lead the public to consider behavioral changes, they may need to move beyond the techniques explored in this article. Across all survey forms, respondents demonstrated a high level of concern about climate change but low levels of behavioral engagement; this is consistent with an earlier survey of residents of central Pennsylvania, which found ambivalence toward efforts to reduce greenhouse gas emissions in general, and low levels of support for voluntary efforts that are difficult, costly, or require changes in lifestyle (O'Connor et al. 2002, 11).

These lessons for climate change communicators are preliminary, and should be viewed in light of the limitations of the research design. Of the several limitations to the research, the most important is probably its use of convenience sampling rather than random sampling. Although the survey reached members of a diverse group of Centre Region organizations, most of these organizations already had some interest in environmental issues. Thus, the results may not hold for less environmentally minded samples, and perhaps cannot be generalized to the full population of adult Centre Region residents or to residents of places outside of the Centre Region. In addition, the survey tested only one spatial analog, which was generated based on output from one climate model run, which was in turn based on one emissions scenario. Thus, it is possible that the results could have been an artifact of the particular spatial analog used, and not representative of the spatial-analog method in general. Finally, the online survey method prevented control of the environment in which the respondents took the questionnaire. Attempts to screen out uninvited participants may not always have been successful, and a few respondents may have collaborated with friends and family when completing the questionnaire. Future research could address many of these limitations.

Despite its limitations, the research is a valuable first attempt at exploring an important and little researched question. Barriers to effective climate change communication threaten efforts to tackle a problem that, if left un-addressed, could seriously threaten the health of both human and natural systems (Pachauri and Reisinger 2007). Given the importance of removing these barriers, and in light of the finding that spatial analogs may not always be the best choice for improving understanding and engagement, empirical assessments of other widely used but little-tested methods of communicating climate change may be worthwhile. Using surveys and other metrics, communication can be fine-tuned to improve understanding of the science, increase public engagement, and possibly drive mitigative and adaptive action. This research has shown that the use of maps may be one way to improve understanding and engagement; additional research is however needed to identify other ways.

http://dx.doi.org/10.1080/15230406.2013.826479

References

Alley, R., T. Berntsen, N. L. Bindoff, Z. Chen, A. Chidthaisong, P. Friedlingstein, J. Gregory, et al. 2007. Climate Change 2007: The Physical Science Basis-Summary for Policymakers. Geneva: IPCC Secretariat.

American Psychological Association's Task Force on the Interface between Psychology and Global Climate Change (APA). 2009. Psychology and Global Climate Change: Addressing a Multi-faceted Phenomenon and Set of Challenges. Accessed July 29, 2013. http://www.apa.org/ science/about/publications/climate-change-booklet.pdf

Association of American Geographers GCLP Research Team. 2003. Global Change and Local Places: Estimating, Understanding, and Reducing Greenhouse Gases. Cambridge: Cambridge University Press.

Beniston, M., and H. F. Diaz. 2004. "The 2003 Heat Wave as an Example of Summers in a Greenhouse Climate? Observations and Climate Model Simulations for Basel, Switzerland." Global and Planetary Change 44 (1): 73-81.

Bolin, B., B. R. Doos, J. Jager, and R. A. Warrick. 1986. Scope 29--The Greenhouse Effect, Climatic Change, and Ecosystems. Chichester: John Wiley. Accessed January 30, 2013. http:// www.scopenvironment.org/downloadpubs/scope29/contents.html

Brody, S. D., S. Zahran, A. Vedlitz, and H. Grover. 2008. "Examining the Relationship between Physical Vulnerability and Public Perceptions of Global Climate Change in the United States." Environment and Behavior 40 (1): 72.

Buncombe County Tourism Development Authority. 2011. Asheville Accolades and Media Praise|Asheville, NC's Official Tourism Web Site. Accessed April 26, 2011. http:// www.exploreasheville.com/press-room/asheville-quality-of-life-awards/index.aspx

Centre County Planning and Community Development Office. 2008. Annual Estimates of the Population for Counties of Pennsylvania: April 1, 2000 to July 1, 2008. Accessed April 11, 2011. http://www.co.centre.pa.us/planning/2008_ Population_Estimates.pdf

Center for Research on Environmental Decisions (CRED). 2009. The Psychology of Climate Change Communication: A Guide for Scientists, Journalists, Educators, Political Aides, and the Interested Public. New York: Earth Institute, Columbia University. Accessed July 29, 2013. http://www.cred.columbia.edu/guide/pdfs/CREDguide_full-res.pdf

Dillman, D. A. 2007. Mail and Internet Surveys: The Tailored Design Method. Hoboken, NJ: John Wiley & Sons.

Dow, K., and T. E. Downing. 2006. The Atlas of Climate Change: Mapping the World's Greatest Challenge. Berkeley: University of California Press.

Dunwoody, S. 2007. "The Challenge of Trying to make a Difference Using Media Messages." In Creating a Climate for Change: Communicating Climate Change and Facilitating Social Change, edited by S. C. Moser and L. Dilling, 89-104. Cambridge: Cambridge University Press.

Evans, J. S. B. 2003. "In Two Minds: Dual-Process Accounts of Reasoning." Trends in Cognitive Sciences 7 (10): 454-459.

Fabrikant, S. I., S. R. Hespanha, and M. Hegarty. 2010. "Cognitively Inspired and Perceptually Salient Graphic Displays for Efficient Spatial Inference Making." Annals of the Association of American Geographers 100 (1): 13-29.

Gifford, R. 2008. "Psychology's Essential Role in Alleviating the Impacts of Climate Change." Canadian Psychology/ Psychologic Canadienne 49 (4): 273-280.

Glantz, M. H. 1988. Societal Responses to Regional Climatic Change: Forecasting by Analogy. Boulder, CO: Westview Press.

Grothmann, T., and A. Patt. 2005. "Adaptive Capacity and Human Cognition: The Process of Individual Adaptation to Climate Change." Global Environmental Change Part A 15 (3): 199-213.

Hulme, M. 2008. "Geographical Work at the Boundaries of Climate Change." Transactions of the Institute of British Geographers 33 (1): 5.

Ishikawa, T., A. G. Barnston, K. A. Kastens, P. Louchouam, and C. F. Ropelewski. 2005. "Climate Forecast Maps as a Communication and Decision-Support Tool: An Empirical Test with Prospective Policy Makers." Cartography and Geographic Information Science 32 (1): 3-17.

Jamieson, D. 1988. "Grappling for a Glimpse of the Future." In Societal Responses to Regional Climatic Change: Forecasting by Analogy, edited by M. H. Glantz, 73-93. Boulder, CO: Westview Press.

Kleinbaum, D. G. 2010. Logistic Regression [Electronic Resource]: A Self-Learning Text. New York: Springer Science+Business Media, LLC.

Knight, C. G., and M. P. Staneva. 2004. "Analogs of Climate Change." In Drought in Bulgaria: A Contemporary Analog for Climate Change, edited by C. G. Knight, I. Raev, and M. P. Staveva, 29-38. Aldershot: Ashgate Publishing.

Kopf, S., M. Ha-Duong, and S. Hallegatte. 2008. "Using Maps of City Analogues to Display and Interpret Climate Change Scenarios and Their Uncertainty." Natural Hazards and Earth System Sciences 8: 905-918.

Leiserowitz, A. 2006. "Climate Change Risk Perception and Policy Preferences: The Role of Affect, Imagery, and Values." Climatic Change 77 (1): 45-72.

Leiserowitz, A. 2007. "Communicating the Risks of Global Warming: American Risk Perceptions, Affective Images, and Interpretive Communities." In Creating a Climate for Change: Communicating Climate Change and Facilitating Social Change, edited by S. C. Moser and L. Dilling, 44-63. Cambridge: Cambridge University Press.

LimeSurvey. 2011. LimeSurvey--The Free & Open Source Survey Software Tool! Accessed April 11, 2011. http://www.limesurvey.org/

Lorenzoni, I., S. Nicholson-Cole, and L. Whitmarsh. 2007. "Barriers Perceived to Engaging with Climate Change Among the UK Public and their Policy Implications." Global Environmental Change 17 (3-4): 445-459.

MacEachren, A. M. 2004. How Maps Work: Representation, Visualization, and Design. New York: The Guilford Press. Madden, T. J., K. Hewett, and M. S. Roth. 2000. "Managing Images in Different Cultures: A Cross-National Study of Color Meanings and Preferences." Journal of International Marketing 8 (4): 90-107.

Maibach, E., C. Roser-Renouf, and A. Leiserowitz. 2009. Global Warming's Six Americas 2009: An Audience Segmentation Analysis. New Haven, CT: Yale Project on Climate Change, School of Forestry and Environmental Studies, Yale University.

Marx, S. M., E. U. Weber, B. S. Orlove, A. Leiserowitz, D. H. Krantz, C. Roncoli, and J. Phillips. 2007. "Communication and Mental Processes: Experiential and Analytic Processing of Uncertain Climate Information." Global Environmental Change 17 (1): 47-58.

Mearns, L. O., W. Gutowski, R. Jones, R. Leung, S. McGinnis, A. Nunes, and Y. Qian. 2009. "A Regional Climate Change Assessment Program for North America." Eos Transactions, American Geophysical Union 90 (36): 311.

Moser, S. C. 2007. "More Bad News: The Risk of Neglecting Emotional Responses to Climate Change Information." In Creating a Climate for Change: Communicating Climate Change and Facilitating Social Change, edited by S. C. Moser and L. Dilling, 64-80. Cambridge: Cambridge University Press.

Nicholls, N. 1999. "Cognitive Illusions, Heuristics, and Climate Prediction." Bulletin of the American Meteorological Society 80 (7): 1385-1397.

Nicholson-Cole, S. A. 2005. "Representing Climate Change Futures: A Critique on the Use of Images for Visual Communication." Computers, Environment and Urban Systems 29 (3): 255-273.

Nunnally, J. C. 1978. Psychometric Theory. New York: McGraw-Hill.

O'Connor, R. E., R. J. Bord, B. Yamal, and N. Wiefek. 2002. "Who Wants to Reduce Greenhouse Gas Emissions?" Social Science Quarterly 83 (1): 1-17.

Pachauri, R. K., and A. Reisinger. 2007. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva: IPCC.

Propen, A. 2007. "Visual Commnnication and the Map: How Maps as Visual Objects Convey Meaning in Specific Contexts." Technical Communication Quarterly 16 (2): 233-254.

Schmunk, R. B. 2010. Panoply netCDF, HDF and GRIB Data Viewer (Version 2.9.4) (Software). Accessed July 29, 2013. Available from http://www.giss.nasa.gov/tools/ panoply/

Schnotz, W. 2002. "Commentary: Towards an Integrated View of Learning from Text and Visual Displays." Educational Psychology Review 14 (1): 101-120.

Sheppard, S. R. 2005. "Landscape Visualisation and Climate Change: The Potential for Influencing Perceptions and Behaviour." Environmental Science and Policy 8 (6): 637-654.

Slocum, R. 2004. "Polar Bears and Energy-Efficient Lightbulbs: Strategies to Bring Climate Change Home." Environment and Planning D 22: 413-438.

Slovic, P., M. L. Finucane, E. Peters, and D. G. MacGregor. 2004. "Risk as Analysis and Risk as Feelings: Some Thoughts about Affect, Reason, Risk, and Rationality." Risk Analysis 24 (2): 311-322.

SPSS Inc. 2009. PASW Statistics for Windows, Version 18.0. Chicago, IL: SPSS.

Sturgis, P., and N. Allum. 2004. "Science in Society: Re-evaluating the Deficit Model of Public Attitudes." Public Understanding of Science 13 (1): 55.

Tschakert, P., and K. A. Dietrich. 2010. "Anticipatory Learning for Climate Change Adaptation and Resilience." Ecology and Society 15 (2): 11.

Tufte, E. R. 1983. The Visual Display of Quantitative Information. Cheshire, CT: Graphics Press.

Union of Concerned Scientists. 2008. Climate Change in Pennsylvania: Impacts and Solutions for the Keystone State. Cambridge, MA: Union of Concerned Scientists.

Williams, J. W., S. T. Jackson, and J. E. Kutzbach. 2007. "Projected Distributions of Novel and Disappearing Climates by 2100 AD." Proceedings of the National Academy of Sciences 104 (14): 5738.

Wilson, T. D. 1996. "A New Look at Anchoring Effects: Basic Anchoring and Its Antecedents. Journal of Experimental Psychology." General 125 (4): 387.

David Pahl Retchless*

Department of Geography, The Pennsylvania State University, 302 Walker Building, University Park, PA 16802, USA

(Received 18 July 2012; accepted 23 April 2013)

* Email: dpr173@psu.edu

Table 1. Six treatment groups, each assigned to a different version of
the survey.

          Change in temperature          Spatial analog

Text      Text describing change in      Text describing spatial
          temperature (Form 1)           analog (Form 4)

Map       Map of change in temperature   Map of spatial analog (Form
          (Form 2)                       5)

Text      Text and map describing        Text and map describing
and map   change in temperature (Form    spatial analog (Form 6)
          3)

Table 2. Questionnaire sections, in order of appearance. Section 4
represents the experimental stimuli.

Sections                 Purpose                Number and type of
                                                    questions

Section 1       Introduction and random     NA
  ([dagger])      assignment

Section 2       Baseline for climate        5 Likert scale questions
  ([dagger])      change beliefs

Section 3a      Prediction of temperature   1 Likert scale question
  ([dagger])      change                      and 1 temperature-change
                                              scale question

Section 3b      Prediction of spatial       1 analog selection
  ([dagger])      analog                      question

Section 4       Presentation of             NA
                  temperature change in 1
                  of 6 ways

Section 5a      Prediction of temperature   1 Likert scale question
  ([double        change (understanding       and 1 temperature-change
  dagger])        and engagement)             scale question

Section 5b *    Prediction of spatial       1 analog selection
  ([double        analog (understanding)      question
  dagger])

Section 6       Effects on lifestyle and    12 Likert scale questions
  ([double        finances, purchasing
  dagger])        decisions, and outdoor
                 activities (engagement)

Section 7       Demographics                9-13 multiple choice and
  ([double                                    numerical input
  dagger])                                    questions

Note: * Spatial-analog survey forms only; ([dagger]) Pre-treaternent;
([double dagger]) Post-treatment.

Table 3. Response rate by organization, valid responses only.

Partner organization                           Respondents   Response
                                                    *        rate (%)

C1earWater Conservancy                            185          16.6
Penn State Eco-Action Club                         26           4.9
Unitarian Universalist Church                      80          15.3
State College Bird Club                            35          12.5
Trout Unlimited--Spring Creek Chapter              33          18.9
University Baptist and Brethren Church             20          13.0
Sierra Club, Moshannon Group                       16          13.9
League of Women Voters                              8          13.8
UNA-USA Centre County                               7          12.7
Good Shepherd Catholic Church (Social              11          26.8
  Justice Group)
Penn State Center for Sustainability                4          25.0
  (Staff Only)
Unknown **                                         36           NA
Total                                             444          14.4

Notes: * Some respondents indicated that they had been invited to
participate by multiple organizations; these respondents are counted
once for each organization, but are not counted multiple times in the
overall total.

** Includes respondents who did not indicate a referral source, or
described their referral source in general terms (e.g., "email," "my
church," or "two Centre-Region organizations"); respondents who
explicitly indicated a non-sanctioned referral source (e.g., "a
website" or "the Intemet") were removed from the results prior to
analysis.

Table 4. Multinomial logistic regression of direction of change in
subjective temperature prediction from pre-treatment to post-
treatment.

Direction of change in
subjective temperature        Survey                     SE
prediction (DV) (a)          form (IV)      [beta]     [beta]

Decrease in predicted       Analog (b)     0.898 *      0.439
  change in temperature      Maps (c)      0.090        0.420

Increase in predicted       Analog (b)    -0.586 **     0.209
  change in temperature      Maps (c)      0.581 *      0.225

                                            95% CI
                                        [e.sup.[beta]]
Direction of change in
subjective temperature      [e.sup.    Lower    Upper
prediction (DV) (a)         [beta]]    bound    bound

Decrease in predicted        2.455     1.039    5.805
  change in temperature      1.094     0.481    2.491

Increase in predicted        0.557     0.370    0.838
  change in temperature      1.787     1.150    2.777

Notes: (a) The reference category is: no change in predicted change in
temperature.

(b) The reference category is: change in temperature survey forms.

(c) The reference category is: text-only survey forms.

Control is gender (omitted from table).

[e.sup.[beta]] = exponentiated [beta]; Model fitting (chi-square):
[chi square] = 36.103 (df = 8, p < 0.001).

* p<0.05; ** p<0.01.

Table 5. Ordinal logistic regression of quartiles for effects on
lifestyle and finances.

                                                         95% CI
                                                     [e.sup.[beta]]

Survey form                             [e.sup.     Lower     Upper
(IV)             [beta]     SE [beta]   -[beta]     bound     bound

Analog (a)     -0.650 ***     0.180      1.916      1.346     2.726
Maps (b)        0.369         0.192      0.691      0.475     1.007

Notes: (a) The reference category is: change in temperature survey
forms.

(b) The reference category is: text-only survey forms.

Control is quartile score for pre-treatment concern about climate
change (omitted from table).

[e.sup.-[beta]] = exponentiated opposite of [beta]. Model fitting
(chi-square): [chi square] = 90.661 (df = 5, p < 0.001).

*** p < 0.001.

Table 6. Ordinal logistic regression of quartiles for impact on
purchasing decisions and outdoor activities.

                                               95% CI [e.sup.-[beta]]
Survey
form                                [e.sup.-    Lower     Upper
(IV)          [beta]    SE [beta]   [beta]]     bound     bound

Analog (a)    -0.367      0.189      1.443      0.997     2.091
Maps (b)       0.209      0.200      0.811      0.166     3.977

Notes: (a) The reference category is: change in temperature survey
forms.

(b) The reference category is: text-only survey forms.

Controls are quartile score for pre-treatment concern about climate
change, distance of respondent's zip code from Centre Region, and
plans to move away from Centre Region (omitted from table).

[e.sup.-[beta]] = exponentiated opposite of [beta]. Model fitting
(chi-square): [chi square] = 83.647 (df = 7, p < 0.001).
COPYRIGHT 2014 Taylor & Francis Group LLC
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Retchless, David Pahl
Publication:Cartography and Geographic Information Science
Article Type:Report
Date:Jan 1, 2014
Words:11040
Previous Article:Study of the attentive behavior of novice and expert map users using eye tracking.
Next Article:Finding least-cost paths across a continuous raster surface with discrete vector networks.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |