Polarization within consensus? An audience segmentation model of politically dependent climate attitudes in Denmark

Effectively communicating climate science requires context-dependent understanding of the target audience and the underlying drivers of their attitudes. The current study explored the landscape of public opinions on climate change in Denmark using segmentation analysis, and investigated its political orientation predictors. Using latent class analysis, we identified the AAACID model; 6 distinct groups that represent the salient differences in Danish climate attitudes, namely the Alarmed, Alert, Ambivalent, Cautious, Indifferent and Doubtful . Climate concern and prioritizing climate mitigation agendas when voting were both negatively associated with Right-orientation, hierarchy, and individualism. A cross-cultural comparison, using the Global Warming ’ s Six Americas framework, further revealed substantial differences between Danish and American public attitudes on climate change. We suggest communication strategies suitable for each segment, and conclude that despite the consensus culture and general acceptance of climate science, political orientation still plays an important role in the remaining disagreement on climate issues in Denmark.


Introduction
The implementation of policies to address the extensive risks of climate change is dependent on a public that is supportive of climate change mitigation and adaptation.This requires public recognition of the severity and causes of the crisis.Understanding how to best align public opinion with the scientific consensus on climate change is a central task for social and communication sciences (Cook et al., 2016).Communicating climate science, however, is not a straightforward task (Markowitz & Guckian, 2018;Moser, 2010).Individual risk assessments, and decisions about appropriate action, are complicated by the inherent complexity of climate change as a phenomenon (Nisbet, 2009).Beyond issues of understanding, people also respond differently to climate messages depending on their existing knowledge and attitudes (Lord & Taylor, 2009), as well as broader worldviews, including political orientations and values (Kahan, 2015b).
In the past decade, significant advances have been made in understanding how to communicate climate science with the most impact (Goldberg, Gustafson, & van der Linden, 2020).Looking ahead, one substantial challenge that remains is understanding how well this research translates when applied to different local contexts across the globe.The current research addresses this challenge by exploring the landscape of public opinion on climate change in Denmark using segmentation analysis, and by identifying cultural, and political factors that underlie the commonalities and differences among segments of the Danish public.Hereby, we provide climate science communicators with a new topology to understand their target groups in this context.Additionally, we provide an intercultural comparison of Danish and American climate attitudes to understand the extent of their differences and similarities.

Biases in climate change attitudes
It is well-known that people are prone to appraise new information based on its compatibility with pre-existing beliefs, attitudes, and motivations (Druckman & McGrath, 2019;Lord, Ross, & Lepper, 1979).Some scholars argue that such results can be explained under an assumption that people interpret incoming information in ways that are in line with Bayesian updating (Tappin, Pennycook, & Rand, 2020).Others emphasize motivations, such as those stemming from concerns about identity-protection, in driving such processes (Kahan et al., 2007;Kunda, 1990;Kahan, 2015a;Hahn & Harris, 2014).In the climate domain, an illustrative study by Corner, Whitmarsh, and Xenias (2012) found that climate change skeptics tend to perceive messages as more convincing when they emphasize uncertainties about climate change rather than consensus (consistent with their current beliefs).In contrast, non-skeptics tend to find messages conveying certainty more convincing.Such findings imply that the effectiveness of climate science messages can vary depending on how they match audiences' pre-existing beliefs about climate change.In its most dire forms, these biases could cause information campaigns to trigger polarization despite providing accurate information (Hart & Nisbet, 2012), as groups differentially embrace and reject messages in ways that entrench preexisting positions.Public polarization on climate change has indeed been observed in multiple countries and particularly in the United States (Capstick, Whitmarsh, Poortinga, Pidgeon, & Upham, 2015).

Tailored messages through audience segmentation
In the context of potential problems of polarization, it is crucial that communication strategies are informed by a nuanced understanding of the audience's current beliefs and attitudes.Since these can differ substantially across individuals, approaching the population as a single, homogenous audience is too reductive.Tailoring messages to capture the interest and values of a specific audience, on the other hand, can motivate more thorough engagement with the material, and lead to stronger and more lasting attitude changes (Hine et al., 2014;Petty et al., 1981).In health communication, for example, tailoring messages to the information needs and interests of specific groups can be significantly more effective in bringing about behavioral change than un-targeted messages (Noar, Benac, & Harris, 2007).
Audience segmentation is a valuable tool that allows researchers and communicators to identify prevalent attitude and belief patterns within the population, and thereby tailor communications more precisely to clusters of like-minded individuals.Perhaps the most prominent use of audience segmentation in climate communication is the Global Warming's Six Americas (SA; Maibach, Leiserowitz, Roser-Renouf, & Mertz, 2011;Leiserowitz, Roser-Renouf, Marlon, & Maibach, 2021).The SA model identified 6 distinct subgroups within the American population, which were dubbed the Alarmed, Concerned, Cautious, Disengaged, Doubtful, and Dismissive.Each segment represents a distinct set of attitudes about climate change, ranging on a continuum from least to most concern about, belief in, and motivation to act on climate change.The SA methodology has been applied in other contexts like Germany (Metag, Füchslin, & Schäfer, 2017), India (Leiserowitz, Thaker, Feinberg, & Cooper, 2013), and Australia (Morrison, Parton & Hine, 2018), and revealed varying consistency with the original 6 categories.Such international comparisons highlight differences in how populations perceive the climate crisis.Other segmentation studies have developed models from entirely different frameworks.For example, a recent study using European Social Survey data identified only four classes as the best division to collectively describe 23 countries (Denmark not being one) across all of Europe (Kácha, Vintr, & Brick, 2022).

Political orientation and climate change attitudes
In addition to basing interventions on observed profiles of expressed attitudes, understanding the driving factors behind these profiles, like values, identities, and worldviews, might further facilitate attempts to create deeper and more lasting change (Hine et al., 2014).A substantial body of evidence has demonstrated that political identity often plays a key role in driving disagreement on climate change.Rejection of anthropogenic climate change is consistently explained by partisanship (Hornsey, Harris, Bain, & Fielding, 2016), ideology (Hamilton, Hartter, Lemcke-Stampone, Moore, & Safford, 2015), and worldviews (Hornsey, Harris, & Fielding, 2018).Specifically, right-wing ideology, conservatism and individualist worldviews are associated with higher rates of climate science rejection (for a comprehensive review of the role of worldviews in climate attitudes, see Hornsey, 2021).In fact, in meta-analyses individualist (as opposed to communitarian) and hierarchical (as opposed to egalitarian) ideologies have been found to be stronger predictors of climate change beliefs than education, scientific literacy, activist/green identity, and personal experience of extreme weather events (Hornsey et al., 2016).
Political outlook will inevitably color one's policy preferences.For example, some common climate mitigation initiatives (like government sanctioned restraints on industry to reduce pollution) tend to receive less support from people holding firm free-market beliefs (characteristic of the political right) than those who believe more strongly in state intervention (characteristic of the political left).The influence of political outlooks constitutes a particular issue of concern, however, when it extends beyond the acceptance or rejection of specific forms of policy and into the acceptance or rejection of the science on which policies should be based.This is because recognizing climate change as a scientifically supported reality precedes motivation to take action, even if action can itself take a variety of forms.Given observed links between political orientations and climate positions, it is essential for communicators to know whether and how their target group's attitudes relate to their political orientation.
The majority of research linking political orientations to climate positions originates from the United States of America (USA) and other English speaking nations.These national contexts have been suggested to have particular political cultures "that encourage citizens to appraise climate science through the lens of their conservative ideologies" (Hornsey et al., 2018, p. 619).However, the attitudinal profiles and political underpinnings of these remain underexplored in alternative contexts, like Northern Europe.Cross-cultural differences in both political culture and climate beliefs between USA and Northern European populations, renders straightforward translation between these contexts difficultpotentially limiting the applied value of tools like Six Americas.In this study, we focus on such a Northern European context, namely Denmark, which is interestingly different from the USA, both on cultural and public climate opinion dimensions.
Public opinion in Denmark is generally characterized by less political polarization than in the USA; Firstly, the Danish multi-partisan political structure differs dramatically from the bi-partisan US system, in that it typically promotes a "moderate rather than polarized debate" by cooperating on social issues across partisan lines (Eskjaer, 2017).Secondly, climate change denial in Denmark has been marginal (Andersen, 2019), partly because climate change reporting has been largely consistent with the scientific consensus (Eskjaer, 2017).Additionally, polls frequently note relatively high prioritization of climate politics in both left-and right wing voters (Altinget, 2019).Further questioning the translatability of extant research to Northern European contexts, English speaking populations, and the USA in particular, appear to be outliers when it comes to how political orientation and identity shapes climate attitudes.A meta-analysis showed that both left-right ideology and individualist worldviews were more strongly connected to climate change skepticism in the USA than in any of the other 24 nations included in the study (incl.Sweden & Germany) followed by the 5 other Western, English-speaking countries1 (Hornsey et al., 2018).Jointly, these cultural differences in political division and public opinion on climate change, make it problematic to generalize conclusions obtained in the USA about politicized and identity-driven climate polarization to a Danish context.At the same time, these topics are under-investigated in this area, and much research is still needed to understand how to best engage and inform the public, including how political identity factors influence our attitudes.

The current research
Our overarching aim is to explore the characteristic differences in Danish public opinions on climate change to ultimately improve targeted climate communication in Denmark.In addressing this aim, the current research extends previous work by: 1) exploring the similarities and differences between Danish and American public climate change opinions by comparing these groups in a common American segmentation framework, in order to better understand the translatability of American research to Denmark and investigate the need for a local segmentation model; 2) providing a new audience segmentation model for the Danish population, AAACID, based on items designed to better capture the attitudinal differences in this cultural context.This model offers a topology of distinct groups defined by their sharing similar understandings of climate change and therefore a meaningful way to define communication targets, and; 3) investigating the extent to which the observed differences in Danish climate attitudes can be explained by political orientation measures, such as Left-Right orientation and cultural worldviews.Cultural characteristics suggest that this link may be weaker in Denmark than in, for example, the USA.This link has been reported previously in Denmark and neighboring countries, and thus we still expected political orientation to predict climate change attitudes.

Methods and materials
The current study is based on survey data collected through an online platform in April 2021.All respondents gave written informed consent prior to participation.Ethical approval for the study was granted by the Research Ethics Committee at the Faculty of Humanities, University of Copenhagen, on the 13th of January 2021 in accordance with the relevant International and Danish ethical guidelines and regulations.Analytical methods used for audience segmentation, cross cultural comparison, and investigating the link between political orientation and climate attitudes will be detailed along with their respective results.

Respondents
A sample (N = 1769) of survey respondents was recruited through an online panel of the agency UserNeeds.Recruitment into the sample was specified to be demographically-representative based on quotas for age, gender, and region that were informed by national statistics.In the recruitment, we made no specifications as to the ethnicity or cultural background of the respondents.The sample therefore comprises Danish speaking inhabitants of Denmark.Respondents were paid through the agency for their participation.Of the total sample, 404 respondents were excluded for insufficient effort responding (Dunn, Heggestad, Shanock, & Theilgard, 2018) based on unreasonably fast response times (below 5 min; sample median = 14.1 min), failing or disregarding the test item, and straightlining tendencies.Of the remaining 1365 responses, ages ranged from 18 to 91 (mean = 51.3,SD = 17.3) with 51% female respondents.Two respondents reported their gender as other/prefer not to say.While there are no clear conventions as to minimum sample sizes in LCA, we chose a sample size of well above 1000 respondents because our model includes many indicators and to ensure sufficient power to distinguish small, but substantially relevant classes (Nylund-Gibson & Choi, 2018).

Climate items
To maximize the nuances in attitudes captured by the survey, we assembled a battery of items regarding climate change beliefs, concern, and engagement, as well as willingness to change and openness towards political initiatives.This comprised 43 items answered on a 5-point Likert scale (disagree-agree; full list available in Supplementary Materials/SM A).
In addition to the items commonly used in segmentation models regarding climate change beliefs and concern, we also assessed issues like perceived involvement among peers (Fielding & Hornsey, 2016;Goldberg, van der Linden, Leiserowitz, & Maibach, 2019), climate conversation frequency (Goldberg, van der Linden, Maibach, & Leiserowitz, 2019), shame related to own emissions (Rees, Klug, & Bamberg, 2015), willingness to act, technological optimism (Gardezi & Arbuckle, 2020;Lamb et al., 2020), and perceived public and scientific consensus (Goldberg, van der Linden, Leiserowitz, & Maibach, 2019) on climate change.These items were selected due to their previously reported empirical or theoretical relevance to climate attitudes, or because we expected the items to be relevant in discriminating attitude segments.This is particularly important since we assumed that simply belief items would not sufficiently capture attitudinal differences in this population.Items with clear ideological connotations (e.g."Danish companies should not be burdened with climate taxes"; see full list in SM A: (not) Included in LCA) were placed at the end of the questionnaire to ensure that the belief and concern questions were not biased by the explicit activation of political perspectives.

Political orientation
As a broad measure of political orientation, respondents were asked to indicate their position on a scale from the political left to the political right ("In politics, people often speak of Left and Right; where would you place yourself?"(1 = Left, 11 = Right).
We also assessed cultural worldviews as a further indicator of political orientations.This study used a version of the Cultural Cognition Worldview Scale (Kahan, 2012), which has been validated after being culturally adjusted and translated to Danish (Andersen & Hallsson, 2022;SM L).It contained 28 items on a 6-point Likert scale (disagree-agree) and assessed the two sub-scales hierarchy (13 items) and individualism (15 items).In previous research, individuals who are oriented more to the political right (left) have been found to score higher (lower) on preferences for hierarchy and individualism (Hornsey, 2021).

Six Americas super short survey (SASSY)
SASSY is a shortened version of the original 36 item Global Warming's Six Americas questionnaire, consisting of 4 items (see SM B) which could allocate respondents to the SA segments with a minimum accuracy of 70% (Chryst et al., 2018a).These items were identified through machine learning algorithms and 14 USA SA samples.Odds-ratios for these four items have then been derived through multinomial logistic regressions and are used to predict the likelihood of membership in each of the SA segments.We used a Danish translation of SASSY to enable a comparison between Danish and USA samples in the SA framework.That is, these items were not used to derive the new segmentation model presented here, but rather for comparative purposes with existing models.
Finally, demographic information was collected from all respondents.

Cross-cultural comparison of US and Danish climate attitudes
As a preliminary exploration of the differences and similarities between Danish and American public attitudes towards climate change, as well as the applicability of the SA model in Danish contexts, we fed the Danish sample data into the SA model.Using the online SASSY Group Scoring Tool, 2 respondents were allocated to the SA segments based on their answers to the SASSY questionnaire and compared to a US sample from December 2020 (Chryst et al., 2018a(Chryst et al., , 2018b)).

Results and discussion
As can be seen in Fig. 1, when we compared our Danish sample to American data there were substantial differences in the prevalence of specific segments across these populations.Two segments, the Disengaged and the Dismissive, each represented the attitudes of less than 1,5% of respondents from this sample, suggesting that these attitude profiles are rare in the Danish population.Interestingly, these are segments characterized by the highest levels of indifference and climate change denial, respectively.Moreover, 70% of the Danish sample fell into the Concerned and Cautious segments, with the disparity between the prevalence of Cautious people in Denmark versus the USA being particularly striking.
This comparison highlights two things.First, the Danish sample is concentrated primarily in the three segments who have the highest belief in and concerns about climate change as well as motivations to act as the USA sample is also.However, fewer Danish than American respondents fit in the outer, most extreme segments, indicating that there is less polarization on climate matters in Denmark, and that overall there is less denial.Secondly, two segments (the Disengaged and Dismissive) are minimally represented, while there is a concentration around two major segments (the Concerned and Cautious).In fact, Goldberg, Gustafson, Rosenthal, and Leiserowitz (2021) have recently suggested that even in the USA, a more fine-grained segmentation might be becoming necessary in response to the opinion shifts that have occurred since the tool was first developed.Nonetheless, this first set of findings suggests that, while the model has been appropriate for a USA context, it does not display the same ability to capture variance in Danish attitudes towards climate change.This highlights the need to construct a segmentation model from dimensions that can better discern relevant subgroups in this particular population.

Segmentation analysis
We used Latent class analysis (LCA) for the segmentation analysis.This is a type of finite mixture modeling, which identifies distinct subgroups in the population based on covariance in response patterns.In other words, LCA allowed us to search for the best possible division of our sample into groups, where respondents are highly similar to their own group while meaningfully different from others.Specifically, we used the R package poLCA (Linzer & Lewis, 2011) to conduct the LCA.For thorough description selection criteria and best practices in LCA, we refer to Weller, Bowen, and Faubert (2020).
29 items were included in the analysis (full list in SM A).We omitted climate attitude items with political/ideological facets (e.g."Danish companies should not be burdened with climate taxes.")from the LCA, in order to ensure that the model solution was not "contaminated" with ideological content, which could by default interact with political identity measures.A series of analyses of variance (ANOVA) showed that there were significant differences between segments p < .001for all variables included in the LCA (in all cases p < .001;SM K).This indicates that all included items played a distinguishing role in the analysis.

Model selection
A range of model solutions were compared in order to select the model with the best fit and interpretability.These models ranged from 2 to 8 classes and were each initiated with 500 random starts: a large number to avoid convergence at local rather than global solutions (McLachlan & Peel, 2000).This number was sufficient for all models to consistently find solutions with the highest log-likelihood (see SM C).Multiple fit criteria were then considered to evaluate the model which best fit the data (for more details, see SM D).
The fit criteria pointed towards multiple models with similar informativeness.However, the Bayesian Inference Criterion (BIC), which simulation studies show to be the most reliable information criterion (Nylund, Asparouhov, & Muthén, 2007), favored the 6 class solution (Table 1).Evaluating the statistical fit criteria in conjunction with interpretability (Weller et al., 2020), we selected the 6 class model as the optimal solution in part because it distinguishes a theoretically interesting "Ambivalent" segment, which was lost in more parsimonious solutions.Across all solutions, the small "Doubtful" segment was the most robust and persistent.

The six segments: AAACID
The resulting 6 segments were dubbed according to their respective characteristic attitude patterns: the Alarmed (15%), the Alert (21%), the Ambivalent (13%), the Cautious (27%), the Indifferent (17%), and the Doubtful (7%).The segments roughly range from the most to the least concerned, but represent a more nuanced division of respondents that is not captured by a single dimension (see Fig. 2).For example, where the Ambivalent are almost as concerned as the Alert, they are less open to climate policies than the Cautious.And where the Indifferent are more likely to believe in climate change than the Doubtful, they report having the weakest interest of all.A detailed description of each segment, along with their item response distributions and demographics, can be found in SM E-G.

Alarmed (15%).
The Alarmed is the most concerned, socially engaged, and pro-policy segment.Supported by a firm trust in climate scientists, they have full certainty of the reality and human causes of climate change.They anticipate severe consequences, also in Denmark, and think that climate policies are urgently needed.Mitigating action is extremely important to this group, which strongly supports CO 2 taxes and is the segment most willing to make compromises on other political areas to reduce emissions (e.g., welfare or economic growth).It is important for them to express the demand for climate action when voting.Alarmed people typically also perceive themselves as having strong attitudes about climate change (see Fig. 3), although they do not perceive their social circles as particularly engaged in the climate Fig. 1).Bar chart of percentages of our Danish and American sample (collected in December 2020) ascribed to each SA segment.Figure adapted from SASSY online tool output (Chryst et al., 2018a(Chryst et al., , 2018b)).

Table 1
Akaike (AIC) and Bayesian (BIC) information criteria for the four model solutions of most interest.Lower information criteria indicate better fit (Weller et al., 2020).debate.The Alarmed tend to be younger than average (5 years), and slightly more often women than men (57% female).

Alert (21%).
This highly concerned segment is in many ways similar to the Alarmed, only with slightly more moderate attitudes.They too perceive climate change as an extremely dangerous and urgent threat.They separate themselves from the alarmed particularly in perceiving themselves as having less strong attitudes about climate change.It is, however, still somewhat important for this group to demand ambitious climate policies when voting.The Alert do not differ much from the average age, but like the Alarmed, they are more often women than men (55,4% female) and more likely to live in the capital than the average.

Ambivalent (13%).
The Ambivalent are characterized by an unusual mismatch between risk perception and readiness for change.This group perceives climate change as an extremely serious threat, almost to the same extent as the Alert.But when it comes to policy support and readiness to, for example, pay more for high-emission goods, the Ambivalent are even less supportive than the Cautious, despite the Cautious being less concerned.They also report feeling less emission shame than both Cautious and Alert (see Fig. 4).
Interestingly, they are also much less likely to value climate agendas highly when voting than the cautious (see Fig. 5).This mismatch is bridged by perceiving the consequences as much more temporally distant than the Alert and Alarmed.Climate change is not socially salient in this group; the Ambivalent rarely discuss the issue, and perceive their social circles as disengaged from the topic.The Ambivalent are equally likely to be female or male, and slightly younger than the average respondent (around 4 years).

Cautious (27%).
The Cautious constitute the largest segment, characterized by moderate attitudes and beliefs about climate change.Overall, this group reports a relatively high awareness of the human caused nature of climate change.Contrasting this group with the Alarmed, however, the Cautious report much less certainty about this fact (see Fig. 8).At the same time, they perceive much less urgency around climate change (see Fig. 6).Nonetheless, this segment is overall in favor of taking measures against climate change, and are even somewhat willing to make compromises in their way of life, for example by paying a higher price for meat and fossil fuels.They generally believe that citizens should themselves do more to mitigate climate change, and that companies should be taxed for their emissions.Still, prioritizing climate agendas when voting remains inconsistent.Demographically, this group is very similar to the average.

Indifferent (17%).
The Indifferent stand out in their lack of engagement in climate matters.Most people in this group are aware that climate change is happening, but less than half acknowledge that human activity is the primary cause.This group also reports a much lower trust in climate scientists than, for example, the Cautious.For the Indifferent, climate change is a distant threat; they generally believe that potential harm to people is far off in the future, and are not expecting to be harmed by climate change themselves.
Of all segments, the Indifferent report having the very weakest attitudes towards climate change (see Fig. 3) as well as being the ones who most rarely discuss climate matters in their social circles (see Fig. 7).In fact, a t-test revealed that the Indifferent (M = 2.55, SD = 0.84) on average self-reported having weaker attitudes on climate change than the Doubtful (M = 3.04, SD = 0.63; t(117) = − 3.17, 95% CI [− 0.8, − 0.18], p = .002).Members of this segment are more frequently men (55%) than women, and are slightly older than the sample mean (approximately 2 years).

Doubtful (7%).
The Doubtful got their label from frequently rejecting the anthropogenic causes of climate change (see Fig. 8) and, to a much lesser extent, the reality of it.This stance is reflected in the     Doubtful's mistrust in climate scientists: 70% disagreed (somewhat or completely) that climate scientists are trustworthy.Interestingly, this segment perceives the absolute lowest amount of public and scientific consensus on the man-made nature of climate change.To the extent that the Doubtful believe in climate change, it is perceived as a distant issue both in time and geographically.Turning towards climate engagement, there is a strong sentiment among the Doubtful that the climate crisis is being exaggerated, and that ambitious action takes resources from more important and urgent issues.Furthermore, it is a pervasive notion that both personal and governmental action is futile as a means to reduce climate change.This is reflected in the group's pronounced opposition to carbon taxes on personal goods and industrial activity.More men (67%) than women are Doubtful, and this group tends to be older than the sample mean (about 7 years).

Varying perception of primary causes
We additionally observed differences in what the segments perceived as the primary cause of climate change.The Alarmed and Alert perceive both Western consumption and growth in the developing countries as primary causes of climate change, but they attributed slightly more responsibility to the West.The Indifferent and Doubtful, however, show the opposite pattern; Where the Doubtful are neutral about the role of the developing world, they specifically reject the influence of Western consumption (see SM H)].

Political orientation and climate change attitudes
In addition to characterizing Danish attitudes and opinions on climate change, a second aim was to investigate whether these attitudes were similarly explained by political orientations, as has been observed elsewhere in the world (though to varying degrees: Hornsey et al., 2018).To explore this, we first conducted a series of multiple linear regressions to predict the key items of climate change threat perception ("Global climate change is a serious threat") and voting priority ("In order to get my vote, a political party must have a clear and ambitious plan for tackling climate change") based on Left-Right identification, individualism, and hierarchy, controlling for demographics (education, household income, is respondent the only adult in household: yes/no, children: yes/no, age, and gender).
In these analyses, Left-Right identification, individualism, and hierarchy significantly predicted threat perception (F(9,1053) = 37.4, p < .001,adjusted R 2 = 0.236), and together these variables explained 19 percentage points more variance than a model containing only These relationships are also traceable in the political orientations of the segments.The Alarmed hold the strongest egalitarian and communitarian views (Fig. 9), whereas the Doubtful are the most hierarchical, individualistic and Right-oriented.The ambivalent shared the same levels of individualism as both the Indifferent and the Cautious, but are significantly less hierarchical than the Indifferent.All differences in individualism between segments were significant, with the exception of Ambivalent-Cautious and Ambivalent-Indifferent, as were all differences in hierarchy and Left-Right identification (Fig. 10) significant except Ambivalent -Cautious (results of t-tests can be found in SM I).

Usefulness of the segmentation model
The AAACID model contributes to a recent international wave of audience segmentation models, seeking to understand differences in local attitudes toward climate change.Such characteristics allow researchers and communicators to identify which receivers are most relevant to target, and in turn hint at how to best interact with this group.
International validation and comparisons of segmentation models is important because individuals are embedded in considerably different climate change contexts, both environmentally and culturally.The unevenly distributed impacts of climate change, and the means to cope with it, will inevitably bring about different perceptions of the climate crisis across nations.In Denmark, extreme weather events are still relatively infrequent compared to other parts of the world.Although less so than the USA, Denmark still has substantial per capita emissions.Yet compared to other nations, Denmark is in an advantageous position financially and socially to undertake local adaptive action.Such factors will in all likelihood influence how its citizens perceive climate change.As such, it is not surprising that the current model differs from segmentation models established elsewhere, like the Six Americas.
Other national segmentations have indicated and selected different numbers of segments (e.g. 5 Germanys; Metag et al., 2015), or different meanings of the segments due to the included variables.Similarly, when comparing the AAACID membership and SA membership of respondents in this study (see SM J), members of the least engaged and informed segment, the Indifferent, were more likely to be characterized as the SA Cautious than the SA Disengaged.As such, the meaning of indifference is relative to the culture within which one is situated.Other segments, such as the Alarmed or the Doubtful, were dubbed this way to reflect the large overlap in how respondents were classified in both models.Our crosscultural comparison, however, also revealed that the SA model describes segments (the Disengaged and the Dismissive) that are characterized by attitude constellations in the USA that are not evident in Denmark.As such, it appears that the model differences reported here are due to more than just design choices.

Communication strategies for the segments
The AAACID model identified 6 segments, namely the Alarmed, Alert, Ambivalent, Cautious, Indifferent, and Doubtful.From their characteristics and previous research on similar audiences, we can tentatively suggest appropriate communication for the individual segments.It is essential that the effectiveness of such strategies is tested empirically, and such studies will be important next steps for Danish climate communication research.To avoid accentuating differences, communicators tailoring messages to resonate with specific segments may want to prioritize bringing groups closer in their attitudes (Hine et al., 2014) and toward attitudes that are more conducive of individual, collective, and political action.This is particularly important in issues such as climate change, which necessitate coordinated collective action (Crompton, 2010).

Leveraging the engaged
Due to their already high risk perception and levels of policy support, the Alarmed and Alert should be ideal target groups for organizations aiming to foster collective action.A study by Doherty and Webler (2016) found evidence that targeting descriptive social norms and efficacy beliefs was an effective way of promoting public-sphere behavior in the SA Alarmed (e.g.voting, contacting elected officials, or protesting).Emphasizing the pro-climate actions that similar others are already taking part in, as well as promoting self-and collective efficacy beliefs, could therefore also be opportune strategies for Danish communicators to engage the Danish Alarmed and Alert due to their shared characteristics with the US Alarmed.

Social solutions to ambivalence
In contrast to the Indifferent and the Cautious, the Ambivalent are highly aware of the risks and urgency of climate change.Still, they show even weaker policy support and openness to change than the Cautious.As such, risk perception and policy support are not contingent in this group, so further emphasizing the consequences of climate change may be redundant here.The Ambivalent also report lower (both personal and governmental) response efficacy than the Cautious, Alert and Alarmed.In combination with their high risk perceptions, this could render the Ambivalent more avoidant rather than responsive to the issue (Rimal & Real, 2003).Efforts to strengthen efficacy beliefs, for example through success stories, or promoting vicarious learning (McLoughlin, 2021), could help the Ambivalent to align their actions and policy support with their high risk perceptions.
The Ambivalent also perceive their social circles as disengaged in the climate debate and rarely discuss the issue.Consequently, Ambivalents may not often be exposed to positive normative statements from their peers about which behaviors and attitudes are socially approved and disapproved.As people holding ambivalent attitudes are particularly likely to be persuaded by social consensus (Hodson, Maio, & Esses, 2001), promoting private-sphere conversations about climate change may be especially promising in this group.In addition to exposing people to normative information, conversations about climate change can also lead to learning important climate facts, and in turn to stronger beliefs and higher concern (Goldberg et al., 2019).As such, initiatives encouraging and equipping people to discuss climate facts with friends and family can contribute to a positive cycle of climate engagement and discussions.

Underscoring personal relevance for the less concerned
The Cautious constitute a significant group with 1 in 4 adults belonging to this segment.The Cautious are predominantly concerned about climate change and are positive toward climate action, yet they see the issue as somewhat distant.To emphasize personal relevance, organizations raising awareness about the proximity and urgency of climate change, may for example leverage trusted, local weather news sources to emphasize the link between climate change and local extreme weather events or long-term impacts (Halperin & Walton, 2018).This strategy may also be fruitful for the Indifferent.
Due to a lack of interest and engagement, particularly in the Indifferent, this group will likely be less willing to seek out information themselves, or spend time on learning complex facts.A more promising way to engage this segment could be to provide frequent, accessible, and entertaining messages (Flora et al., 2014) through familiar media and trusted sources.Once again, frequent, factual conversations about climate change with friends and family can be an important, highly trusted source of such information (Goldberg et al., 2019).Additionally, the expectation of public discussion draws people's attention to their own attitudes and encourages them to bolster their beliefs (Lambert, Cronen, Chasteen, & Lickel, 1996).

Political orientation
We found that individual climate attitudes and segment membership are tied to both cultural cognition worldviews and political left/right orientation.These findings align with evidence from several other countries (Hornsey et al., 2018), which show relationships of varying strength depending on the country, with the strongest relationships found in the USA.Thus, despite the Danish consensus culture, and the fact that the values underlying the Left-Right spectrum can vary substantially across different countries (Wojcik, Cislak, & Schmidt, 2021), political orientation still plays an important role in climate change attitudes.
Solution aversion has been highlighted as a potential mediator of the effect of political orientation on climate attitudes (Campbell & Kay, 2014).That is, conservative resistance to climate science may be less about perception of the science, and more a spill-over of the resistance to the most popular solutions, as these often go against typical conservative ideologies.Based on a relatively weak relationship between political orientation and attitudes in many non-English speaking countries, however, others argue that there is little "inherent to conservative ideologies that predisposes people to reject climate science" (Hornsey et al., 2018, p. 614).In our data too, it is not entirely clear whether climate attitudes and political orientation are inherently connected.The fact that the Cautious and Ambivalent segments, despite substantial differences in their attitude patterns, share highly similar political orientation indicates that other important factors differentiate climate attitudes, besides ideology and worldviews.
Another causal connection is argued to pass from political orientation through social group membership, where especially attitudes toward divisive issues are associated with certain groups (Druckman & McGrath, 2019).This can act as a motivation to seek interpretations in line with one's own ingroup or worldviews.A recent study found causal evidence that some Danish individuals engage in identity protective reasoning driven by cultural cognition worldviews, when processing climate change data (Hallsson & Andersen, 2022).While the current study investigated only correlational relationships, this suggests the possibility that the link observed here is at least partially caused or sustained by identity protective reasoning.
The exact implication of this relationship for communication is still a matter of discussion.There is some evidence that informative messages can have a backfiring effect on science deniers, causing the receivers to polarize even further in order to protect either identity or beliefs (Hart & Nisbet, 2012;Gromet, Kunreuther, & Larrick, 2013).However, this phenomenon is inconsistent (Druckman & McGrath, 2019).Perhaps more likely is the possibility that messages containing connotations of political orientations or worldviews simply have less impact on receivers (Hornsey, 2021).In either case, a growing body of research suggests that an effective way of bypassing the potential problem of resistance is to communicate climate science in a frame that affirms the values or identity of the target group (Wolsko, Ariceaga, & Seiden, 2016;Cohen & Sherman, 2014).This has for example been investigated with elite cues and republican in-group messengers (Goldberg, Wang, et al., 2021) and conservative moral frames (Wolsko et al., 2016).It should be noted, however, that identity affirmation and value alignment are not always proven effective (Severson & Coleman, 2015).Nonetheless, aligning messages with values more congruent with conservative Danish values may indeed be a promising way to gain support for climate change initiatives from a broader part of the population.The challenge will be how to best create appropriate frames that capture the values of each group while also promoting a common goal across political orientations, values and worldviews.

Status on polarization
We did not observe considerable signs of public polarization in beliefs about anthropogenic climate change in our data.The disagreement we do observe concerns levels of involvement and concern about the issue, willingness to act, and placement of responsibility.Outright denialism is marginally present only in the small Doubtful segment.Yet, in the place of outright denialism, more sophisticated, policy-oriented discourses of opposition may still develop.For example, Lamb et al. (2020) recently conceptualized four new overarching discursive strategies, namely 1) redirection of responsibility, 2) pushing for non-transformative solutions, 3) emphasizing the downsides of action, and 4) surrendering to climate change.While these discourses do not deny the existence of climate change and often contain partial truths, they threaten transition initiatives by justifying inaction or inadequate efforts.
Attitudes consistent with such discourses are traceable particularly in the Doubtful and Indifferent groups.For example, many Doubtfuls reject that Western consumption and lifestyles are primary causes of climate change, while also expressing indifference about the climate impact of the developing world (section 4.1.3),a pattern reminiscent of the 'redirect responsibility' delay discourse.When many Doubtful also report that climate change is not caused by human activity, this denial therefore seems to be particularly tied to the Western world and less so to the Developing countries.Furthermore, technological optimism is one of the discourse of 'pushing for non-transformative solutions'.Generally, the Alarmed and the Alert are the only segments opposing the idea that new technologies will be able to solve the climate crisis without individuals having to make substantial changes to their lifestyles.In the remaining segments there is a degree of technological optimism, which can be problematic in diverging attention from immediate consumer and voter action possibilities.
The current study suggests that a number of these discourses may be disorienting the Danish climate debate and discouraging more concerted action.In the process of engaging and informing the Danish public, outreach initiatives should therefore consider that acceptance of climate change is not necessarily sufficient in ensuring public engagement in, and demand for, climate action.Rather, it may be necessary to address and correct the misconceptions of climate delay discourses.It is possible that segments would respond differently to being exposed to such discourses on the basis of their varying levels of certainty, knowledge, and involvement.

Limitations and future directions
The current study is cross-sectional, and it is an important consideration that attitudes are expected to change over time.For example, data was collected in the spring of 2021 during the first year of the Covid-19 pandemic, and climate attitudes are known to vary in response to societal and economic changes (Andersen, 2019).Continuous data collection and analysis with the AAACID tool in the future would allow some examination of the stability of clusters and a picture of how segment sizes and membership change over time.It is furthermore of utmost importance that the effectiveness of tailored communication strategies is tested in the different segments in the future.Future research could advantageously go into greater depth of how attitudes about different aspects of the climate change are interrelated within different groups, thus improving our current understanding of the feasibility of interventions targeting different attitudes.Both quantitative and qualitative studies could look beyond political orientation and focus on different predictors of these attitude differences.Interviewing members of different segments could help further understand what sets these groups apart, for example, whether the Indifferent and Ambivalent experience socio-economic factors having influence on their active engagement in the issue.

Conclusion
Motivated by the sparsity of research on effective climate communication in Northern Europe, this study explored the landscape of climate attitudes within the Danish population through audience segmentation.Firstly, a cross-cultural comparison based on the Global Warming's Six Americas model revealed that public perceptions of climate change in Denmark differ substantially from those of US citizens.The Danish sample was less polarized and more concerned overall.Such cultural differences highlight the need for local climate communication research.The main contribution of this study was the identification of six distinct climate segments, each representing a distinct way in which members of the Danish public understand climate change.While we generally observe a strong belief in the reality of anthropogenic climate change and openness to climate action, there are considerable differences between segments, which hint at potentially effective strategies for communication and engagement.Finally, we observed that Danish public opinions on climate change are linked to political identity measures like identification with the political right/left and associated cultural worldviews.Specifically, typically conservative characteristics were associated with weaker beliefs in climate change and support for counter-measures.NNF17SA0031368, and by the University of Copenhagen.Bjørn Hallsson was funded by the Danish National Research Foundation under grant DNRF144.

Fig. 2 .
Fig. 2. The AAACID model with segment proportion and main characteristics.

Fig. 3 .
Fig. 3. Segment conditional responses to "I have strong opinions about climate change".

Fig. 4 .
Fig. 4. Segment conditional responses to "I often feel ashamed about my consumptions or lifestyle not being climate friendly."

Fig. 5 .
Fig. 5. Segment conditional responses to "In order to get my vote, a political party must have a clear and ambitious climate agenda".

Fig. 7 .
Fig. 7. Segment conditional responses to "I often discuss climate matters with family and friends".

Fig. 8 .
Fig. 8. Segment conditional responses to "Human activity is the primary cause of climate change".

Fig. 9 .
Fig. 9. Standardized Hierarchy and Individualism scores for each segment.Point size represents segment size.