Climatic Change DOI 10.1007/s10584-013-0708-5

Perceptions of climate change: Linking local and global perceptions through a cultural knowledge approach Beatrice Crona & Amber Wutich & Alexandra Brewis & Meredith Gartin

Received: 30 August 2011 / Accepted: 29 January 2013 # Springer Science+Business Media Dordrecht 2013

Abstract Understanding public perceptions of climate change is fundamental to both climate science and policy because it defines local and global socio-political contexts within which policy makers and scientists operate. To date, most studies addressing climate change perceptions have been place-based. While such research is informative, comparative studies across sites are important for building generalized theory around why and how people understand and interpret climate change and associated risks. This paper presents a cross-sectional study from six different country contexts to illustrate a novel comparative approach to unraveling the complexities of local vs global perceptions around climate change. We extract and compare ‘cultural knowledge’ regarding climate change using the theory of ‘culture as consensus’. To demonstrate the value of this approach, we examine cross-national data to see if people within specific and diverse places share ideas about global climate change. Findings show that although data was collected using ethnographically derived items collected through placebased methods we still find evidence of a shared cultural model of climate change which spans the diverse sites in the six countries. Moreover, there are specific signs of climate change which appear to be recognized cross-culturally. In addition, results show that being female and having a higher education are both likely to have a positive effect on global cultural competency of B. Crona (*) Stockholm Resilience Centre, Stockholm University, 106 91 Stockholm, Sweden e-mail: [email protected] A. Wutich : A. Brewis School of Human Evolution and Social Change, Arizona State University, Tempe, AZ 85287-2402, USA A. Wutich e-mail: [email protected] A. Brewis e-mail: [email protected] M. Gartin Urban Sustainability Research Coordinaton Network, Global Institute of Sustainability, Arizona State University, Tempe, AZ 85287-2402, USA e-mail: [email protected]

Climatic Change

individuals. We discuss these result in the context of literature on environmental perceptions and propose that people with higher education are more likely to share common perceptions about climate change across cultures and tentatively suggest that we appear to see the emergence of a ‘global’, cross-cultural mental model around climate change and its potential impacts which in itself is linked to higher education.

1 Introduction Over the last two decades, a growing body of scholarship has focused on understanding public perceptions of global warming and climate change, including how people recognize, understand, and respond to risk (Kempton 1991; O’Connor et al. 1999; Leiserowitz 2005, 2006; Brody et al. 2008; Agho et al. 2010). Documenting and explaining this aspect of climate change is fundamental because it defines the local and global socio-political contexts within which policy makers and scientists must operate and public perceptions fundamentally compel or constrain local and global political, economic, and social action in response to this issue (O’Connor et al. 1999; Leiserowitz 2005; Brody et al. 2008; Agho et al. 2010). To date, most studies addressing perceptions of climate change have been locally situated (e.g. Leiserowitz 2005; Brody et al. 2008; Battaglini et al. 2009; Byg and Salick 2009; Bunce et al. 2010; Speranza et al. 2010). This place-based research agenda is important for several reasons. First, climate change effects and susceptibilities to them will most likely be regionally and locally uneven (Walther et al. 2002). Second, a growing understanding that people are good natural observers of their local environmental has led to an appreciation for their knowledge, which is situated in cultural and ecological contexts, and how it can provide important models and unique understandings of climate change (Salick et al. 2009; Turner et al. 2009). Third, it recognizes that local perceptions reflect real-world and tangible concerns (Crate and Nuttall 2009). Collectively, these place-based studies clearly show that both local ecology and culture matter in how people conceptualize climate change. However, comparative studies across sites are important for building a more generalized theory around why and how people understand and interpret climate change, and for innovative hypothesis generation. Comparative studies also help link perceptions at local and more global scales, an important aspect given the pervasive character of climate change and the growing demand for mitigation at both global and local levels (Lorenzoni and Hulme 2009). Existing comparative studies are hampered by two key problems. First, most comparative studies of climate change perceptions have compared and contrasted sites with similar culture, language, and or socio-economic contexts, such as the US and Britain (Brechin 2003; Lorenzoni et al. 2006; Lorenzoni and Pidgeon 2006), or within Europe (Battaglini et al. 2009; Lorenzoni and Hulme 2009). Comparisons across different cultural and socio-economic contexts are few (Eurobarometer 2007; Brechin and Bhandari 2011). In particular, perspectives of indigenous people and citizens of developing countries are barely included, even though their vulnerabilities to climate change can be profound and their perceptions quite distinct (Salick et al. 2007). Second, most cross-national studies lack a well-grounded theory and design for how to capture local variability while also testing for various forms of universality in how people conceptualize climate change. Hence, most have relied heavily on summarizing and comparing findings of previous surveys (Bord et al. 1998; Lorenzoni and Pidgeon 2006; Brechin and Bhandari 2011), i.e. not conducting standardized data collection with locally-derived and/or locally-relevant indicators across diverse contexts. Here we present a cross-sectional study of six different contexts to illustrate a novel comparative approach to unraveling the complexities of local vs global perceptions around

Climatic Change

climate change. The approach focuses on extracting and comparing what cognitive anthropologists refer to as ‘cultural knowledge’ regarding climate change using the theory of ‘culture as consensus’ (Romney et al. 1986; Romney 1999). To demonstrate the value of this approach, we examine this cross-national dataset to see to what degree people within specific and diverse places share perceptions about global climate change. We expect that each site examined will exhibit its own ‘cultural model’ of climate change perceptions. For example, indigenous communities in coastal Fiji are more likely to be concerned or knowledgeable about issues related to sea-level rise, while drought may be more salient in the minds of respondents in the southwest US and Australia (Karoly et al. 2003). This proposition is supported by work showing that public risk valuations vary predictably with individual personal and historical experiences (Kempton 1991; Kraus et al. 1992; Savage 1993). Studies have also shown that gender and education tend to predict perceived climate risks (Savage 1993; O’Connor et al. 1999; Agho et al. 2010; Wolf and Moser 2011). We therefore also predict that these two variables will affect the cultural knowledge of respondents in consistently patterned ways across our studied populations.

2 Methodology This study makes use of cultural consensus analysis (CCA) to assess perceptions related to climate change in six countries. CCA is both a theory and a method (Romney et al. 1986). It conceptualizes culture as cognitive – i.e. culture is in the mind, not in artifacts or enacted behaviors. As a theory it is assumes that consensus is an indication of an ‘underlying truth’. Many epistemological systems rely on this connection between agreement and truth; for example the US court jury system, which does not regard a prosecutor’s claims as true unless a jury of 12 independent people agree (Borgatti and Halgin 2011). For a more comprehensive explanation of the formal model behind CCA please refer to Romney et al. (1986) and Weller (2007). D’Andrade (1987) suggests that the cultural beliefs of a group are beliefs held by a majority of culture members. The problem for anthropologists has traditionally been to know what that culture - held in the minds of the majority – is. Even if a culture exists, asking members of this ‘culture’ simple questions about cultural ‘facts’ will often result in variable and possibly conflicting answers. This is a manifestation of the difference in’cultural competency’ (Weller 2007) among respondents. Paraphrasing Borgatti and Halgin (2011:171), CCA as a methodology provides three things to address this issue. First, it provides a way to determine whether observed variability in beliefs is cultural, that is whether informants are sampled from different cultures with systematically different beliefs, or whether it simply reflects differences in individual familiarity with elements in their own culture. This is achieved by calculating the ratio between the 1st to 2nd eigenvector. It indicates whether there is only one set of answers present in the data. When responses are homogeneous across informants, a single pattern of responses will emerge and eigenvalue ratios will indicate a singular dimension present in the data. A general rule is that the ratio should be 3 to 1 or greater, in which case the consensus model may be used to represent the group’s responses with a single set of answers. Average competency scores should also be above .50 (Weller 2007). Second, for a group determined to belong to a single culture, CCA measures how much of the culture each individual knows – i.e. each individual’s ‘cultural competence’. Third, for each culture represented in a data set, CCA attempts to determine the culturally correct answer to every question that was put to respondents. This is done by creating an agreement matrix which is factored and the competence scores are used to weight the responses of each

Climatic Change

individual (by multiplication) and are then summed together. The competence scores are the first factor loadings, and the answers (a weighted, linear combination of responses) are found on the first set of factor scores. Fig. 1 graphically illustrates the culture-as-consensus approach as used here. The similarity between culture-as-consensus and work on public perceptions (e.g. Bord et al. 1998; O’Connor et al. 1999) is that both approaches collect individual level data and seek to identify predictors of individual variation in perceptions of climate change. In contrast to the public perceptions and standard survey approaches, however, CCA also explicitly characterizes the content of shared ‘cultural knowledge’ (perceptions) around climate change and assesses how much individuals conform to/deviate from this shared knowledge. The focus on shared knowledge also means that purposive locality-based sampling is a more appropriate means to capture ‘culture’ (Handwerker and Wozniak 1997) than randomized designs used in surveys. Unlike population-representative sampling results derived from this method are therefore not generalizable to that country; rather they represent the shared cultural model identified at a specific site (if one in fact exists). But by identifying cultural answers for distinct samples at the local level, CCA allows meaningful assessments of knowledge that is limited to local culturally distinct samples versus that which is shared across samples (i.e. globally) (Weller and Baer 2002). The method has been widely used (Weller and Baer 2001; Shafto and Coley 2003; Gartin et al. 2010) to understand ‘culture’ among groups of individuals. In our case ‘culture’ is akin to shared perceptions of climate change and does not necessarily reflect deeply held values but rather beliefs about causal relationships related to climate.

Fig. 1 Illustration of the culture-as-consensus approach, the analysis used and how it was applied in this study

Climatic Change

2.1 Research sites, sampling and data collection This research was conducted as part of the Global Ethnohydrology Study, a multiyear, multi-sited study examining cross-cultural knowledge of water and climate. Data were collected in six countries: Fiji, in one indigenous Fijian village on the island of Viti Levu; Ecuador indigenous Quichua communities along the Napo River (a rural agricultural zone) and in the small local center of Tena; New Zealand in rural Piopio (a rural town) and Wellington (a large city); Australia in Brisbane; UK, in central London; and US, in two neighborhoods in the city of Phoenix, Arizona. Sites represent a broad range of socio-cultural and environmental contexts, including two developing nations with lower urbanization (Fiji and Ecuador), sites with arid, inland environments (Phoenix US), coastal environments (Fiji and New Zealand), tropical rainforest environment (Ecuador), sub-tropical riverine environment (Australia), and temperate riverine environment (U.K.). In each site, we used non-probabilistic, purposive sampling designed to capture only local residents, as we reasoned they were most likely to share a local cultural model of climate change. The theoretical rationale is that if agreement exists in local cultural knowledge it should be represented in any public space sampling (Handwerker and Wozniak 1997). Within each site, effort was made to purposively sample respondents with as broad a range of demographic characteristics as possible (esp. with regard to gender, age, and race/ethnicity). Minimum sample size was set at 29, as this is the minimum needed to correctly classify 95 % of cultural items at a .99 confidence level with a conservative assumption of .50 cultural competency in the general population (Weller 2007). Data were collected with a total of 279 adults, spread across sites (Table 1). Respondents were recruited from public places in New Zealand, Australia, the U.K., and the U.S., door-to-door in Fiji and in Ecuador. The sample sizes and some sample-specific characteristics are given in Table 1. Data were collected through interviews by field researchers and were conducted in English in New Zealand, Australia, the U.K. and the U.S., Spanish and Quichua in Ecuador, and English and Fijian (with translators, as needed) in Fiji. While consequences of climate change are experienced at local levels the phenomenon itself is largely constructed by the scientific community. Asking local respondents how they deal with climate change can therefore be problematic. To address this, the survey was developed inductively, using freelisting techniques as a first step to elicit information about how and if respondents in each sites understood and conceptualized changing climate. The freelists were analyzed and the most commonly cited signs of climate change in each region identified. From these lists 36 (6 from Table 1 Demographics of the sites in each respective country Sites in…

Sample size (N) Female (%) Avg Age (yr) Avg Income (Self-rated Education= College scale, 1–10) or higher (%)

Australia

49

36.7

33.1

5.38

63.3

Ecuador

29

62.1

32.3

3.42

13.8

Fiji

61

55.7

34.7

4.26

14.8

New Zealand UK

63 43

41.3 41.9

37 33

5.66 5.41

52.4 69.8

US

33

63.6

52.4

6.17

78.8

Climatic Change

each area) signs were chosen, phrased as questions eliciting yes or no answers, and included in the final cultural consensus interview protocol (Table 2). This approach allowed us to examine whether respondents in each site recognize only local signs of climate change vs. recognizing signs induced from other research sites. Table 2 Cultural consensus questions, created based on freelists on local signs of climate change across sites. Right hand = “culturally correct” answers (and percentage answered correctly) from the global model. Culturally correct answers are determined by weighting most heavily the responses of people who performed best on the overall model (rather than deferring to a simple majority) Cultural consensus question

Answer

Have rain patterns changed in the last 20 years?

Y (84 %)

Are summers hotter now than they once were?

Y (73 %)

Are winters colder now than they once were?

Y (56 %)

Do asphalt/paved surfaces cause changes in local climate?

Y (68 %)

Are changes in monsoons a sign of climate change?

Y (76 %)

Can pollution cause global climate change? Are there fewer mosquitoes and other insects than there were 20 years ago?

Y (90 %) N (61 %)

Are changes in sandstorms a sign of climate change?

Y (63 %)

Is climate change causing it to rain more than it once did?

Y (56 %)

Do roads flood less heavily than they used to?

N (58 %)

Are allergies a bigger problem for people than they used to be?

Y (77 %)

Do farmers need less water for crops and animals than they used to?

N (73 %)

Is population growth a cause of climate change?

Y (70 %)

Is the sea level rising? Are droughts becoming less frequent?

Y (77 %) N (74 %)

Are more hurricanes or cyclones a sign of climate change?

Y (77 %)

Is ocean life dying off?

Y (77 %)

Is climate change causing people to suffer more sickness?

Y (71 %)

Is crop failure a sign of climate change?

Y (77 %)

Is it easier to make a living fishing as a result of climate change?

N (79 %)

Are fish migration patterns changing as a result of climate change?

Y (73 %)

Are glaciers melting as a result of climate change? Are stronger winds a sign of climate change?

Y (86 %) Y (76 %)

Are natural disasters increasing as a result of climate change?

Y (77 %)

Is snowfall decreasing as a result of climate change?

Y (61 %)

Are crop growing seasons changing?

Y (75 %)

Are fresh water sources being improved by climate change?

N (73 %)

Is the food supply made more secure by climate change?

N (74 %)

Is climate change making people healthier?

N (85 %)

Are diseases like malaria or dengue on the rise as a result of climate change? Is water pollution a sign of climate change?

Y (64 %) N (49 %)

Is rainfall decreasing as a result of climate change?

Y (60 %)

Are islands disappearing because of climate change?

Y (73 %)

Is there less fresh water available for people to use now than 20 years ago?

Y (70 %)

Is the weather more predictable now than it was 20 years ago?

N (60 %)

Are animal migration patterns changing?

Y (77 %)

Climatic Change

2.2 Analysis To examine the existence of a globally shared understanding about climate change, cultural consensus analysis of the responses to the questionnaire was conducted using the CCA module within the UCINet software. First raw response data was entered from the survey questions (0=No; 1=Yes) into SPSS and a variable created to sum the number of missing responses. Missing responses were replaced with a value of either 0 or 1 using a random integer-number generator, which is the conventional way to deal with missing data in this approach (see Weller 2007). Data was then imported into UCINet for cultural consensus analysis. This provided three outputs of relevance here (Weller 2007): (1) a measure of overall fit which indicates existence of a shared cultural model; (2) a culturally correct answer for each question based upon the frequency of shared responses to each question in the survey. Answers were determined by weighting most heavily the responses of people who performed best on the overall model (rather than deferring to a simple majority), (3) cultural competency scores which correspond to the number of correct answers of each respondent. Next we used the cultural competency scores from CCA as a dependent variable to examine to which degree gender (female=0, male=1) and level of education (college education or higher=1, high school or lower=0) could predict cultural competency across our global sample. We controlled for country level effects by running a 2-block (nested) regression, where countries were entered in the first block and gender and education entered in the second. This allowed us to tease out to what degree these two variables could improve the model above and beyond country level effects. Least squares multiple linear regression in SPSS (version 17.0) was used. Data was examined for multicollinearity and heteroskedasticity.

3 Results and discussion 3.1 Cultural consensus around climate change We began by examining the presence of locally distinct models of climate change in each of the six areas (by combining proximate sites). Using the criterion of the ratio of the 1st to 2nd eigenvalues and the average cultural competency scores (Table 3), we find clear evidence of locally shared models for the sites in the U.K., U.S., Australia, and New Zealand; however, the Fiji and Ecuador sites do not reach the conventional cutoff for a shared cultural model. This means that there is not a clear consensus of agreement among the people responding in the latter two samples. Next we examined whether there is also a “globally shared” cultural model of climate change perceptions. By “globally shared” we mean the existence of a core set of items in the survey that people generally agree on, regardless of where they live. Based on the results of the CCA (Table 3), we do find evidence supporting a globally shared cultural model of climate change across our entire sample (ratio of the 1st to 2nd factors=5.895). However, average competency score was .41, somewhat lower that the recommended level of .5. Looking at the distribution of average competencies across sites, all are at or near .5 with the exception of Ecuador and Fiji. Table 2 provides the list of items with the shared answer to each item. Agreement across the global model is reflected in high values (%). For example, the shared model includes shared recognition of climate change as having anthropogenic causes, creating extensive changes in the natural world, causing more frequent natural disasters, and resulting in deleterious human health effects.

Climatic Change

Table 3 Average factor loadings of the global cultural model, by country

Model demographic

Average competency score

“Global”

0.41

Country Australia

0.48

Ecuador

0.23

Fiji

0.23

New Zealand

0.50

UK

0.49

US

0.55

Examining our results in the context of existing literature on climate change perceptions (e.g. Crate and Nuttall 2009; Turner et al. 2009) we expected distinctive cultural models of climate change across our sample sites, with no clear global model of climate change. However, our results do not confirm this. Even though we do see evidence of cultural models in four of the studied sites, the CCA also reveals a shared cultural model of climate change which spans the diverse sites in the six countries. Furthermore, there are some specific signs of climate change which appear to be recognized cross-culturally. For example, we find a remarkably high level of agreement across sites regarding the causes, signs, and consequences of climate change. For instance, 90 % of respondents across all global sites agree that pollution causes climate change. The site with the highest agreement was in the U.K. (93 %), while the lowest agreement was in Ecuador (86 %). Regarding signs of climate change, 84 % of respondents globally agreed that rain patterns have changed in the last 25 years. The site with the highest agreement was in the U.S. (94 %), while the lowest agreement was again Ecuador (76 %). For the consequences of climate change, 85 % of respondents responded “no” to the question “Is climate change making people healthier?” The U.S. site had the highest percentage of respondents saying “no” (97 %), while the Fiji site had the lowest percentage saying “no” (70 %). It is interesting to note that even though the data presented here was collected using ethnographically derived items collected through place-based, open-ended interviews regarding local signs of climate change, we still find a shared understanding of signs related to climate change that span across geographic settings. Two examples highlight this. In our prior interviews in Fiji, we asked respondents to tell us some signs of climate change. Given Fijians’ dependence on marine resources, several people said that climate change was undermining fishing livelihoods and incomes. On the cultural consensus survey, we thus included the following reverse-worded question: “Is it easier making a living fishing as a result of climate change?” We expected that only Fijians would say “no” in large numbers. Globally, however, 79 % of respondents replied “no” to this question. In fact, the site where the most respondents gave the culturally correct answer was in the U.S. site (100 %)—a completely landlocked, desert locale with no marine economy whatsoever. Similarly, in our early interviews in New Zealand, people told us that melting glaciers are a sign of climate change. This is particularly salient in New Zealand, where glaciers are projected to shrink up to 72 % by 2100 (Radic and Hock 2011). Thus, we expected that glacier melt would be a much more salient indicator of climate change in the New Zealand site than in our other sites. While there was clear consensus in New Zealand that the correct response to the question “Are glaciers melting as a result of climate change?” is “yes” (87 %), this was only marginally higher than in the global model, where 86 % said “yes”. The site with the highest

Climatic Change

consensus that climate change is causing glaciers to melt was in Australia (96 %)—where there are no glaciers at all. Even for the four sites in which there were locally distinct cultural models surrounding climate change perceptions the substantive differences in perceptions across sites were quite modest. In fact, the culturally correct answers for local models only deviate from culturally correct answer in the global model for 5 out of 36 questions, mainly items having to do with changing patterns of rain and snow. In each case, we believe that local weather and environmental conditions explain the deviation of local sites from the global model. Beyond these relatively modest differences, there was complete agreement across the global and four local models regarding the culturally correct answers to the remaining 31 of our 36 place-based, locally-induced survey items. It is worth noting that all four countries have a strong Anglo-cultural influence which may very well account for the strong similarity in cultural models observed between these sites. It may also partly explain why the two study areas with no Anglo influence were significantly different, but it cannot explain why the cultural consensus within these sites was low. 3.2 Education and gender as important predictors of perceptions around climate change Results from the regression of global competency scores are shown in Table 4. Countries were entered as covariates in the first block, and gender and education in the second to test if these two demographic variables could predict global cultural consensus above and beyond country level effect. The addition of these two variables significantly improved the model (F change=8.344, p=0.000) (Table 4) and indicates that being female and having a higher education are both likely to have a positive effect on global cultural competency of individuals. The overall model also shows that countries do have an effect on individual competency scores (F change=14.579, p=0.000). In particular, we see that Fiji (−0.370) and Ecuador (−0.257) appear to be significantly different from the rest of the countries, as noted above. We cannot determine exactly why but note that respondents in these two country sites on average score less well on the global cultural consensus model and that it may be linked to the mutual absence of Anglo-cultural influence in these sites. With regard to gender and education, both of these variables have been shown to influence how people perceive climate change risk in previous studies (Savage 1993; O’Connor et al. 1999; Agho et al. 2010). If we begin with gender, our results to some degree resonate with some of the broad themes emerging in work on ecofeminism. Although a diverse field in itself, the different strands in ecofeminism can be said to share the belief that gender, class and race mediate the ways in which people live in and with local environments, and as such, being female distinguishes women’s experiences with their environment from those of men. Women are often tasked with providing for the family by extracting environmental services (Rocheleau et al. 1996; Terry 2009) and they are therefore likely to also be affected differently by factors altering the environment, such as climate change – which could arguably affect their perceptions of this issue, even across cultures (see Wolf and Moser 2011 for a review). The way in which gender affects humanenvironment perceptions have been one interest in the ecofeminism literature and studies have shown quite consistently that women react differently to environmental risk than men (Pilisuk and Acredolo 1988; Stallen and Thomas 1988). The explanations for this have been diverse, ranging from the separation of the world into human and non-human realms as an inherently masculine endeavor (Plumwood 1993), to educational aspects where women are suggested to perceive environmental change and risk differently than men based on lower education (Gutterling and Wiegman 1993), or because of a less prominent orientation toward

Climatic Change Table 4 Results for regression of global individual cultural competence scores Variable

Block 1

Block 2

Model 1

Model 2

Standardized Coefficients

Standardized Coefficients

Australia

-0.094

-0.046

Ecuador

-0.327**

-0.257**

Fiji UK

-0.482** -0.076

-0.370** -0.044

New Zealand

-0.061

-0.003

Gender

-0.122*

Education (college or higher)

0.201** N= 270

N= 270

r2=0.216

r2=0.263

Adj. r2=0.202

Adj. r2=0.244

F=14.579

F change=8.344, sig.=0.000

sig.=0.000 * and ** are significant at the 5% and 1% levels, respectively

economic importance and more focus on personal well-being (Stallen and Thomas 1988). Kahan et al. (2007) challenge all of these assumed relationships and suggest that risk sensitivity is primarily related to ‘cultural worldviews’ structured along two axis “hierarchy-egalitarianism” and “individualism-communitarianism”. Their work shows that gender only affects risk perception in conjunction with particular worldviews, and for environmental risk, hierarchical worldviews influence risk perception the most. This study did not address risk perceptions related to climate per se and therefore direct comparisons with above cited literature should be made with caution. However, our findings suggest being female and highly educated both have a positive effect on global cultural competency of individuals, i.e. their overall agreement with the globally identified cultural model. We can therefor rule out the idea that women differ from men based on lower education. However, beyond this we are not able to determine the reasons behind why women are more similar and ‘culturally competent’ in their perceptions of climate change in our global cultural model, but can only conclude that our findings are in line with previous work examining gender and human-environment relations. We also find that higher education appears to predict cultural competency. However, with regard to education and environmental perceptions the literature is less conclusive. In fact, work on environmental or technological hazards has shown quite contradictory trends. Some work shows that higher education is related to a reduced sense of risk (Pilisuk and Acredolo 1988). Others have seen the exact opposite trends (Midden 1986). In relation to climate change specifically, O’Connor et al. (1999) found that higher education was associated with less perceived risk as a result of climate change, while Agho et al. (2010) find that those with a university degree were more likely to think that global warming would increase. Although seemingly at odds, these results are not necessarily contradictory upon closer examination, as one could believe global warming could increase without necessarily associating any risk with it. This could be because higher education often correlates with higher income and lower involvement in rural economies or direct extraction of natural resources, and hence a lower sense that risks are a threat to oneself. In this paper we do not attempt to measure risks related to climate change specifically. Instead our results inform us that people with higher

Climatic Change

education are more likely to share common perceptions about climate change across cultures, despite the seeming effect of an Anglo-cultural influence on some of the sites. From this we could thus tentatively deduce that we appear to see the emergence of a ‘global’, cross-cultural mental model around climate change and its potential impacts which in itself is linked to higher education. There are several plausible explanations for this. One is that more highly educated individuals are more likely to consume similar (globally accessible) media which affects their way of thinking about climate change. Another possible mechanism behind our observed results is that more highly educated people are also more likely to have shared understanding of scientific knowledge (Kempton 1991; Etkin and Ho 2007). Finally, more highly educated people are less likely to work in professions defined by heavy natural resource extraction (such as agriculture or fisheries) and this commonality in their modes of employment, and reduced likelihood of exposure to fluctuations in the natural environment, may affect how they conceptualize climate change (O’Connor et al. 1999; Ungar 2000; Kahlor and Rosenthal 2009; Sundblad et al. 2009; Wolf and Moser 2011).

4 Conclusion CCA complements and enhances local place-based studies of perceptions of climate change, giving us the means to identify cross-cultural patterns in how people conceptualize climate change. This preliminary study illustrates how comparative studies provide a way of linking perceptions at both local and global scales which can be important given the pervasive character of climate change and the growing demand for mitigation at both global and local levels. CCA also seeks to explicitly characterize the content of shared cultural knowledge and to assess the degree to which individuals conform to or deviate from this, and as such can assist with building more generalized theory around why and how people understand and interpret climate change. Findings can thus also be important for generating further hypotheses. For example, the variation in ideas within – rather than across – sites presented here, compel us to search for theories that explain the lack of shared models observed in Ecuador and Fiji. The way climate change information is created and disseminated in Ecuador and Fiji may differ substantively from the other four sites where large-scale university and government research efforts exist to uncover and disseminate information about projected climate change impacts. It is therefore possible that where scientific, governmental and news reporting do not serve to homogenize information and knowledge around climate change, local perceptions may be more personal and idiosyncratic and thus less likely to form broad cultural consensus. This impacts on how we engage different segments of the general public in the climate change debate, or how we best frame the complexity of climate change for lay audiences (e.g. Budescu et al. 2009; Crate and Nuttall 2009). These questions deserve further examination but show how a cross-cultural study using CCA can provide hypotheses that can be tested in more depth at the local level. Finally, a few limitations of this study, and the use of CCA more generally, deserve a mention. The first relates to sampling. While a purposive sampling strategy is appropriate for CCA it does limit generalizability. Randomized samples drawn from a non-localized populations would assist in determine whether these results generalize to larger regions or countries. Second, site selection here was designed to include culturally, climatically, and ecologically diverse sites. While this is helpful for preliminary studies designed to detect the presence of ‘global’ consensus around climate change across diverse sites, further replication of each site ‘type’ is required to allow inference about differences between sites. Third, the highly inductive nature of our protocol design helped demonstrate the existence of a true

Climatic Change

‘global’ consensus. However, the inductive items do limit our ability to draw parallels between our findings and those of studies that use deductive, theoretically-derived concepts, scales, and survey items. A combined inductive-deductive design that uses inductive language and concepts in survey items, but link these more clearly to existing theoretical concepts from the literature, and compare these across sites is therefore desirable for future work. Finally, further studies using CCA to understand the interplay between local and global ‘cultural models’ may do well to incorporate aspects of the cultural cognition approach advocated by Kahan et al. (2007) to allow for testing of the degree to which alignment along these worldview axes influence cultural competencies. Acknowledgments Work was locally seeded by the National Science Foundation (NSF) Grant No. SES0345945 Decision Center for a Desert City (DCDC) and NSF grant number DEB-0423704 Central Arizona– Phoenix Long-Term Ecological Research,and ASU President’s Late Lessons from Early History initiative. ASU global health and anthropology undergraduate students assisted with field-based data collection. Input by Crona has been made possible by funding from The Swedish Research Council Formas, and by Mistra through a core grant to the Stockholm Resilience Centre. Any opinions, findings and conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the funding agencies.

References Agho K, Stevens G, Taylor M, Barr M, Raphael B (2010) Population risk perceptions of global warming in Australia. Environ Res 110:756–763 Battaglini A, Barbeau G, Bindi M, Badeck FW (2009) European winegrowers’s perceptions of climate change impact and options for adaptation. Reg Environ Chang 9:61–73 Bord RJ, Fisher A, O’Connor RE (1998) Public perceptionsof global warming: United States and international perspectives. Clim Res 11:75–84 Borgatti SP, Halgin DS (2011) Consensus analysis. In: Kronenfeld DB, Bennardo G, de Munck VC, Fischer MD (eds) A companion to cognitive anthropology. Wiley-Blackwell, Oxford, UK. doi:10.1002/ 9781444394931.ch10 Brechin SR (2003) Comparative public opinion and knowledge on global climatic change and the kyoto protocol: the U.S. versus the world? Int J Sociol Soc Policy 23(10):106–134 Brechin SR, Bhandari M (2011) Perceptions of climate change worldwide. Wiley Interdiscip Rev Clim Chang 2(6):871–885 Brody SD, Zahran S, Vedlitz A, Grover H (2008) Examining the relationship between physical vulnerability and public perceptions of global climate change in the United States. Environ Behav 40(1):72–95 Budescu DV, Broomwell S, Por H-H (2009) Improving communication of uncertainty in the reports of the Intergovernmental Panel on Climate Change. Psychol Sci 20(3):299–308 Bunce M, Rosendo S, Brown K (2010) Perceptions of climate change, multiple stressors and livelihoods on marginal African coasts. Environ Dev Sustain 12:407–440 Byg A, Salick J (2009) Local perspectives on a global phenomenon - climate change in Eastern Tibetan villages. Glob Environ Chang 19:156–166 Crate S, Nuttall M (eds) (2009) Anthropology and Climate Change: From Encounters to Actions. Left Coast Press, Walnut Creek D’Andrade RG (1987) Modal responses and cultural expertise. Am Behav Sci 31(2):194–202 Etkin D, Ho E (2007) Climate change: perceptions and discourses of risk. J Risk Res 10(5):623–641 Eurobarometer (2007) Attitudes of European citizens towards the environment. Directorate General Environment, European Commission, Brussels Gartin M, Crona BI, Wutich A, Westerhoff P (2010) Urban ecological knowledge? Urban perceptions of water quality in a desert city. Ecol Soc 15(4):art36 Gutterling JM, Wiegman O (1993) Gender-specific reactions to environmental hazards in the Netherlands. Sex Roles 28(7/8):433–447 Handwerker WP, Wozniak DF (1997) Sampling strategies for the collection of cultural data. Curr Anthropol 38:869–875 Kahan DM, Braman D, Gastil J, Slovic P, Mertz CK (2007) Culture and identity-protective cognition: explaining the white-male effect in risk perception. J Empir Leg Stud 4(3):465–505

Climatic Change Kahlor L, Rosenthal S (2009) If we seek, do we learn? Predicting knowledge of global warming. Sci Commun 30(3):380–414 Karoly D, Risbey J, Reynolds A (2003) Climate change - global warming contributes to Australia’s worst drought. World Wildelife Fund Australia, Sydney Kempton W (1991) Lay perspectives on global climate change. Glob Environ Chang 1(3):183–208 Kraus N, Mamlfors T, Slovic P (1992) Expert and lay judgements of chemical risks. Risk Anal 12(2):215–232 Leiserowitz AA (2005) American risk analysis: is clilmate change dangerous. Risk Anal 25(6):1433–1442 Leiserowitz AA (2006) Climate change risk perception and policy preferences: the role of affect, imagery and values. Climate Change 77:45–72 Lorenzoni I, Hulme M (2009) Believing is seeing: laypeople’s views of future socioeconomic and climate change in England and in Italy. Public Underst Sci 18:383–400 Lorenzoni I, Pidgeon NF (2006) Public views on climate change: European and USA perspectives. Climate Change 77:73–95 Lorenzoni I, Leiserowitz AA, De Franca Doria M, Poortinga W, Pidgeon NF (2006) Cross-national comparisons of image associations with “global warming” and “climate change” among laypeople in the United States of America and Great Britain. J Risk Res 9(3):265–281 Midden CJH (1986) Individual and largescale technology: A comparative attitude study to the production of electricity with coal and uranium. Leiden University, Netherlands O’Connor RE, Bord RJ, Fisher A (1999) Risk perceptions, general environmental beliefs, and willingness to address climate change. Risk Anal 19(3):461–471 Pilisuk M, Acredolo C (1988) Fear of technology hazards: one concern or many? Soc Behav 3:17–24 Plumwood V (1993) Feminism and the mastery of nature. Routledge, London Radic V, Hock R (2011) Regionally differentiated contribution of mountain glaciers and ice caps to future sealevel rise. Nat Geosci 4:91–94 Rocheleau DE, Wangari E, Thomas-Slayter BP (1996) Feminist political ecology. Cross-cutting themes, theoretical insights, policy implications. In: Rocheleau DE, Thomas-Slayter BP, Wangari E (eds) Feminist political ecology: global issues and local experiences. Routledge, London Romney AK (1999) Culture consensus as a statistical model. Curr Anthropol 40:S103–S115 Romney AK, Weller SC, Batchelder WH (1986) Culture as consensus: a theory of culture and informant accuracy. Am Anthropol 88:313–338 Salick J, Byg A, Posey D (2007) Indigenous people and climate change. Symposium Report (April 12–13). Environmental Change Institute, Oxford Salick J, Fangb Z, Byg A (2009) Eastern Himalayan alpine plant ecology, Tibetan ethnobotany, and climate change. Glob Environ Chang 19(2):147–155 Savage I (1993) Demographic influences on risk perceptions. Risk Anal 13(4):313–420 Shafto P, Coley J (2003) Development of categorization and reasoning in the natural world: novices to experts, naïve similarity to ecological knowledge. J Exp Psychol 29(4):641–649 Speranza CI, Kiteme B, Ambenje P, Wiesmann U, Makali S (2010) Indigenous knowledge related to climate variability and change: insights from droughts in semi-arid areas of former Makueni District, Kenya. Climate Change 100:295–315 Stallen PJ, Thomas A (1988) Public concern about industrial hazards. Risk Anal 8:237–245 Sundblad E, Biel A, Gärling T (2009) Knowledge and confidence in knowledge about climate change among experts, journalists, politicians and laypersons. Environ Behav 41(2):281–302 Terry G (ed) (2009) Climate change and gender justice. Practical Action Publishing and Oxfam, Oxford Turner WR, Oppenheimer M, Wilcove DS (2009) A force to fight global warming. Nature 462:278–279 Ungar S (2000) Knowledge, ignorance and the popular culture: climate change versus the ozone hole. Public Underst Sci 9(3):297–312 Walther G-R, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC, Fromentin J-M, Hoegh-Guldberg O, Bairlein F (2002) Ecological responses to recent climate change. Nature 416(6879):389–395 Weller SC (2007) Cultural consensus theory: applications and frequently asked questions. Field Methods 19(4):339–368 Weller SC, Baer RD (2001) Intra- and intercultural variation in the definition of five illnesses: AIDS, diabetes, the common cold, empacho, and mal de ojo. Cross-Cult Res 35(2):201–226 Weller SC, Baer RD (2002) Measuring within- and between-group agreement: identifying the proportion of shared and unique beliefs across samples. Field Methods 14:6–25 Wolf J, Moser SC (2011) Individual understandings, perceptions, and engagement with climate change: insights from in-depth studies across the world. Wiley Interdiscip Rev Clim Chang 2(1):547–569