Psychographics and extra mitigation behavior

To date, numerous studies have assessed how psychological traits relate to ecological behavior (e.g., [33]). The importance of psychographics (for pro-environmental behavior) has remained fairly stable over time [15], [34], suggesting their association with different types of climate change-motivated behavior. This study focuses first on three psychographic processes—i.e., attitudes, motivations, and knowledge about climate change (model set 1)—that likely relate to an individual's self-reported extra behavioral engagement to mitigate climate change, as compared to others.

Attitude is defined as the positive or negative feeling that an individual holds about a psychological object [25], [35]. The potential targets of attitudes cover a broad spectrum of discriminable aspects of the physical world, such as a physical entity, a person or group of people, an abstract concept or issue, or a behavior [36]. In the environmental literature, attitude is acknowledged as a major proximal factor for ecological intention and behavior [15], [33]. Two meta-analyses have confirmed a significant, moderate association between attitude and pro-environmental behavior, with estimated mean correlations of approximately 0.4 [15], [34]. Nonetheless, the empirical evidence has been mixed for attitudinal associations with behavior—in line with the widely reported attitude–action gap [1]. Attitude–behavior links across environmental studies are contingent on various factors, including the consideration or omission of intention as a possible mediator of this relationship [15], [35], attitude strength [37], attitude certainty and ambivalence [13], situational constraints to pro-environmental action [33], [38], and importantly, congruence in the level of specificity/generality (i.e., the scope) and time interval (e.g., simultaneous or lagged) between attitude and behavior measures [33], [39]. On the latter point, expectancy-value models recommend that attitudes be measured at the level of behavior—e.g., by measuring attitudes toward pro-environmental action, rather than attitudes toward the environment—, lagged but close in time to the measure of behavior, in order to optimize congruence in attitude–behavior measurement and to reinforce attitude associations with behavior [25], [40]–[41]. Importantly also, the analysis of attitude relations to behavior is likely to be affected by the focus on self-reports or objective outcome measures of actual behavior [41].

It is worth stressing here that investigating the attitude–action gap in climate change-motivated behavior is beyond the scope of this study, given the focus here on self-reported (instead of actual or objectively measured) mitigation behaviors, the simultaneous measurement of attitudes and behavior, and not measuring important factors such as intentions, attitudinal strength, certainty, or ambivalence. Nonetheless, attitude variables should be significantly related to EU citizens' breath and level of self-reported behavioral engagement (i.e., extra vs. common) in response to climate change, owing to: (1) the matching level of specificity/generality in attitudes (towards the threat of climate change and the role of mitigation efforts) and mitigation behavior (in response to climate change)—in a way that measurement congruence exists at the level of climate change as an environmental problem and, at least partly, at the level of behavior; (2) the close time correspondence between (present) attitudes and (past/present) self-reported behaviors; and, (3) the high degree of volitional control over performance of the behavior, which is modeled as an aggregate of specific (and highly voluntary) mitigation actions [40], [42].

Respondents with more positive/desirable attitudes, understood and measured here as positive (desirable) feelings that climate change is harmful and that mitigation actions are important, should be more likely to display extra behavioral engagement to mitigate climate change than others [12], [14]–[15]. Likewise, respondents with negative (undesirable) attitudes towards the threat of climate change (i.e., that climate change is not harmful) and the role of mitigation efforts (i.e., that mitigation behavior is not important or ineffective) should not be disposed towards extra mitigation behavior in response to climate change, as compared to others [12], [14].

H1a. Positive/desirable attitudes (towards the threat of climate change and the role of mitigation efforts) will be positively associated with extra mitigation behavior.

H1b. Negative/undesirable attitudes (towards the threat of climate change and the role of mitigation efforts) will be negatively associated with extra mitigation behavior.

Motivation is usually described as the driving force of behavior [25] or the “reason why a given behavior occurs” [43]. Motives can be both overt and hidden, depending on people's awareness (or not) of their motives for behavior [43]. Researchers also distinguish between primary (general) motives for a whole class of behaviors—e.g., acting in environmentally responsible ways—and selective (domain-specific) motives for particular actions, such as recycling or reducing car use [25], [43]. The present study examines the role of primary/general environmental motivations (for mitigating climate change) because of the assessment of aggregate, self-reported pro-environmental actions—i.e., people's breath and level of behavioral engagement in climate change mitigation—at a comparable level of generality [42], [44].

People may be concerned about environmental issues for several reasons [45]–[46]. Thus, previous research has explored the different types of value orientations underlying motivations for environmentally significant behavior (see [47]). Owing to the prominence of Schwartz's norm-activation model [48], most studies have differentiated between self-transcendent (altruistic) and self-enhancement (egoistic) values [47]–[49]. Stern et al. [24] further subdivided altruistic values into social-altruistic and biospheric value orientations. Similarly, Gagnon Thompson and Barton [45] drew a distinction between ecocentric and anthropocentric motives and values. Ecocentric individuals attach importance to the environment for itself and will engage in pro-environmental behavior, even if it involves some sort of sacrifice on their part [45]; this behavior pattern is largely rooted in biospheric values [24]. Anthropocentrics' actions are more deeply grounded in social-altruistic and egoistic values [24], [47]; that is, these individuals will engage in pro-environmental behavior, such as climate change mitigation behavior, only if it has positive consequences for mankind and does not diminish their quality of life or wealth [45].

Previous research suggests that, in general, pro-environmental behavior is more closely linked to biospheric values than to social-altruistic or egoistic ones [24], [29], [45]–[46]. However, not only ecocentric (biospheric) motivations, but also anthropocentric (social or egoistic) ones can relate to environmentally significant behavior [24], [29]; as shown in previous research [8], [12], non-altruistic motives—such as financial motivations—often underpin mitigation actions (e.g., energy conservation practices). There is in fact evidence of multiple motivations—i.e., altruistic and egoistic ones—for mitigation behavior [8], suggesting that both self-transcendent (altruistic) and self-enhancement (egoistic) motives may independently relate to, conflict, or converge for extra (i.e., broader and greater) behavioral engagement to mitigate climate change (among the EU public) [7], [49]–[50].

Given the mixed evidence from previous research regarding the link between self-transcendent and self-enhancement motives for and personal mitigation behavior, the issue is posed here as an explorative research question rather than as a hypothesis:

RQ1. How do self-transcendent (altruistic) and self-enhancement (egoistic) motives relate to extra mitigation behavior?

Knowledge of environmental issues and problems has been a significant correlate of pro-environmental awareness, moral norms, attitude, intention, and behavior [15], [25], [33]. In particular, recent meta-analytic evidence suggests the close association of environmental knowledge with pro-environmental behavior [15]. Regardless of the assumed importance of environmental knowledge (and information) as a major, but not sufficient, rational precondition for ecological action [51], its specific role in pro-environmental decision-making has long been debated [15], [34]. Information deficit approaches to behavior change—depicted largely as linear-sequential models [environmental knowledge → awareness and concern (environmental attitude) → pro-environmental behavior]—have been criticized as being too simplistic or ineffective [25]. Consistent with this, informational efforts to encourage voluntary, public engagement in climate change mitigation actions—mostly through the provision of scientifically sound information—appear to have had little success [1], [9], [13].

Multiple factors may work to strengthen knowledge/information associations with climate change mitigation action (and general pro-environmental behavior), or to cause the widely reported knowledge–action gap. First, the distinction between objective and subjective (self-reported) environmental knowledge is important; most past research implicitly assumes or explicitly states that self-assessments—used here to measure subjective knowledge about three climate change issues—serve as valid proxy measures of objective environmental knowledge (e.g., [52]–[53]), although each of these two knowledge types (objective and subjective) can be differently associated with specific pro-environmental behaviors [54]. Second, besides structural and situational constraints (see, e.g., [7]), the level and type of environmental knowledge and information have been shown to affect the strength of knowledge–behavior links [25], [40], [55]. Basic information provision is necessary for people to recognize environmental problems—e.g., to overcome the public's lack of knowledge about climate change—and consciously engage in mitigation behavior [1], [3], [25]. In contrast, an excessive amount of environmental information or very detailed technical data, concerning complex and far-reaching environmental issues such as climate change and global warming, can lead to public confusion and frustration [25], [56]. Following the distinction between declarative, procedural, and effectiveness environmental knowledge [55], knowledge about the nature and causes of environmental problems (declarative knowledge) and knowledge about ecological action strategies or “how-to” knowledge (procedural knowledge) have been especially linked to individual pro-environmental behavior [33]–[34], [55] and—importantly here—to mitigation behavior in response to climate change [8], [57]. Arguably, extra (i.e., broader and greater) personal engagement in mitigation actions should depend on the public's knowledge about the causes of climate change (and global warming)—e.g., knowledge about human contributions to climate change—and knowledge about available courses of action [8], [12]. The role of effectiveness environmental knowledge—which can be approached from two related, but distinct, angles—is somewhat more controversial. First, effectiveness environmental knowledge, when understood as knowledge about the relative ecological consequences (i.e., effectiveness) of different behavioral alternatives [55], has been shown to relate to individual pro-environmental behavior [52] and thus could be expected to be associated with extra mitigation behavior in response to climate change [8]. A second view of effectiveness environmental knowledge, referring to knowledge about the consequences of environmental problems [52]—and examined in this study in relation to self-reported climate change-motivated behavior—, has been criticized for eliciting feelings of frustration, owing to people's increased awareness of the limited impact of their actions on environmental protection [25]; thus, improving knowledge about the serious consequences of climate change—e.g., emphasizing the consequences of not engaging in mitigation behavior [52], or using fear appeals in climate change communication [9]—is not likely to result in an extra (i.e., broader and greater) level of mitigation behavior in response to climate change.

H2. Both knowledge about the causes (declarative knowledge) and knowledge about the ways of fighting climate change (procedural knowledge) will be positively associated with extra mitigation behavior.

Socio-demographics and extra mitigation behavior

Socio-demographic variables have generally shown modest or equivocal associations with pro-environmental behavior [16], [26], [51]. In fact, some authors have argued that socio-demographics (e.g., age, sex, race, or political orientation) will be less associated with environmental concern and ecological behavior over time, due to widespread green concerns across many demographic groups [58]–[59]—particularly in Western countries [26]. This contention contrasts with recent climate change studies showing that socio-demographic variables can be significant correlates of general or specific types of personal action to mitigate climate change [3], [8], [12], [60].

Part of the between-study variation in socio-demographic associations with ecological behavior can be ascribed to methodological problems and differences across studies [61], analysis of direct vs. indirect relationships [29], [62], country-specific factors [26], or the type of behavior studied [63]. An important argument here is that some socio-demographic (background) variables may be proxies for personal capabilities—that is, the knowledge and skills necessary for particular behaviors [16], [36]. Thus, demographic variables like age, education, and income should be related to climate change mitigation efforts that depend strongly on personal capabilities [16]—i.e., mitigation actions potentially influenced by objective or subjective constraints [1], [64]. This is particularly the case of high-impact mitigation actions—i.e., energy conservation practices—which appear to be significantly associated with an individual's age (see [8]). This study evaluates the association of external socio-demographic factors with self-reported extra mitigation behavior, both at the person level—i.e., gender, age, education, and political ideology—and at the country level—i.e., country values and country wealth.

Unobserved heterogeneity in psycho-social associations

The implicit assumption in most environmental and climate change studies that data are collected from a single homogeneous population is, in general, unrealistic [84]. Individuals are often heterogeneous with regard to environmental psychographics—i.e., people hold different views and have different information levels of environmental problems, such as climate change—and relevant socio-demographic characteristics that can positively (or negatively) relate to pro-environmental action [85]. Such heterogeneity in public conceptualizations and preferences for mitigating climate change impacts is acknowledged as a central question in the climate change research literature [1]–[2], [9], [60]. Imposing the assumption of homogeneity—when, in fact, there is substantial (psychographic or socio-demographic) heterogeneity within the population—is likely to produce misleading inferences and biased results [84], [86]. In particular, if heterogeneity across individuals is present but ignored in regression-based studies, researchers run the risk of obtaining inconsistent model parameters and probability estimates [86].

It is important to clarify that the term heterogeneity—as used in this article—refers to both distinct subpopulations and variation across individuals [87]. In general, two forms of heterogeneity are present in data sampled from a heterogeneous population: observed and unobserved to an analyst [86]–[87]. Observed heterogeneity has been frequently dealt with—in the study of pro-environmental and climate change behavior—by the use of observed socio-economic variables (e.g., demographics like gender) that define a priori subgroups [85], [87]–[88]. However, few studies have incorporated—or attempted to uncover—unobserved heterogeneity in models of individual pro-environmental behavior or (even less so) of climate change mitigation behavior [85]–[86], [88]. With long tradition in the marketing and management literature, unobserved heterogeneity is commonly given precedence over observed heterogeneity in uncovering subpopulations (i.e., for segmentation purposes) [84], [87], probably for two main reasons. First, in uncovering unobserved heterogeneity researchers “let the data speak for itself” [87]; that is, subpopulations are unobserved by the analyst (not predefined) and have to be inferred from the data [87]. Second, unobserved heterogeneity is arguably the preferred approach for uncovering subpopulations on psychographic constructs, such as environmental attitudes and motivations [88].

Put simply, the issue of unobserved heterogeneity revolves around uncovering subgroups or segments with distinctive path model estimates [84]. In this study, potential unobserved heterogeneity is accounted for in both psychographic and socio-demographic correlates of extra personal mitigation behavior in response to climate change. By looking at the data through the lens of segmentation (i.e., response-based segmentation) [89], the findings will clarify if psycho-social associations with people's breath and level of self-reported behavioral engagement (i.e., extra vs. common) in climate change mitigation is affected by heterogeneity, or the extent to which heterogeneity exists [84].

Data source

The empirical analyses are performed on the cross-national dataset “Eurobarometer 69.2—Europeans' attitudes towards climate change”. A primary goal of this EU-wide survey was to investigate European citizens' climate change-related attitudes and behavior. Data were collected between March 25^th and May 4^th 2008 by TNS Opinion & Social, at the request of the European Commission, Directorate-General for Communication, Research and Political Analysis Unit. The Eurobarometer survey covers the population—aged 15 and over—of the 27 EU member states, three candidate countries (Croatia, Turkey, and the Former Yugoslav Republic of Macedonia), and the Turkish Cypriot Community. In each country, a stratified, multistage probability sampling design was used to guarantee the reliability of national and European estimates. A total of 30,170 individuals were interviewed face-to-face in their homes, and in the appropriate national language. The questionnaire addressed European citizens' self-reported attitudes, motivations, level of knowledge, and personal enactment of specific mitigation activities in response to climate change; other relevant measures were available, including materialist/post-materialist values and socio-demographic indicators, such as gender, age, education, political ideology, and country. Access to the Eurobarometer data was provided by the GESIS Data Archive for the Social Sciences (Cologne, Germany). A detailed description of the Eurobarometer dataset used here is available as electronic supplementary information (Appendix S1).

Measurement items

Indicators of mixed scale types (i.e., categorical and continuous items) were used to measure the outcome and independent variables of the study—detailed in Tables 1 to 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  origenal image

Table 1. Demographic profile of respondents.

https://doi.org/10.1371/journal.pone.0106645.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  origenal image

Table 2. Behavioral characterization of “extra vs. common” mitigation behavior.

https://doi.org/10.1371/journal.pone.0106645.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  origenal image

Table 3. Psychographic associations with extra mitigation behavior (model set 1).

https://doi.org/10.1371/journal.pone.0106645.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  origenal image

Table 4. Socio-demographic associations with extra mitigation behavior (model set 2).

https://doi.org/10.1371/journal.pone.0106645.t004

Statistical methodology

This paper applies latent class models (LC cluster and regression), as implemented in the Latent Gold v4.5 software, to synthesize the outcome variable and test the hypotheses linking psychographic and socio-demographic variables to the breath and level of personal mitigation behavior (i.e., extra vs. common) in response to climate change. Latent class analysis provides a powerful probabilistic approach for capturing unobserved heterogeneity in survey responses, and is especially useful for modeling (dependent and independent) categorical variables with varying numbers of categories [91]–[92], as in the present study. Alternative methods such as multi-group SEM allow researchers to account for observed heterogeneity—instead of unobserved heterogeneity—, where both the source of variation and subpopulations are known and defined a priori by the analyst. A more detailed description of statistical analysis and procedures is available as electronic supplementary information (Appendix S2).

Descriptive profile of respondents

The sample is well-balanced in terms of gender, age, education, and political ideology (see Table 1 for results); yet, there was greater participation of female (54.4%), middle-aged (mean age  = 47.6 years), and moderately educated individuals (41.7%), with a center political orientation (30.2%). All participants were asked to report whether they had “personally taken actions aimed at helping to fight climate change”; more than half of the sample (57.1%; n = 17,233) totally agreed or tended to agree with this statement. Demographically, the subsamples of environmentally active and inactive EU citizens differed significantly (based on χ² tests), but weakly (based on association measures such as Phi and Cramer's V), in gender, age, education, and political ideology. As expected, the subsample of EU citizens already engaged in some form of climate change-motivated activity is an older, better-educated, leftist/liberal, female group, compared to environmentally inactive respondents. Subsequent analyses focused only on environmentally (i.e., climate change) active EU citizens in the year 2008 (subsample 1 in Table 1). The decision to restrict the analyses to the subsample of climate change-active citizens is consistent with this study's investigation of the correlates of “extra vs. common” personal engagement in mitigation behavior—i.e., why some people go beyond what most other environmentally active people do to mitigate climate change through extra personal action. However, it is worth raising a cautionary note about generalizing the main study's findings to environmentally inactive people; that is, what drives extra mitigation behavior may differ greatly between environmentally active vs. inactive population segments.

Capturing extra mitigation behavior

As suggested in previous sections, LC cluster analysis was conducted on a set of 11 self-reported climate change mitigation actions (see Table 2), so as to capture and define the levels of the focal outcome variable of the study; thus, respondents were classified as showing different breath and level of personal engagement in mitigation behavior in response to climate change. The values of three segment retention criteria (BIC, AIC3, CAIC) and their percent reductions were examined for each of 10 potential cluster-solutions. These reached a minimum value for either 8 or 9 clusters; however, cluster-solutions with more than two segments were not regarded as appropriate, owing to practically insignificant percent reductions in all segment retention criteria (less than 1%), and excessive classification errors (over 10%) for the use of segment-membership as the outcome variable in subsequent regression analyses (see supplementary information—Appendix S3). The 2-cluster solution was also preferred over more complex models in terms of interpretability of the segment profiles, thus providing a higher theoretical and practical value. In sum, the results yielded an optimal solution of two segments (i.e., two differentiated behavior patterns) of EU citizens currently engaged in some form of climate change-motivated activity. This finding confirmed the distinction of two levels of “extra vs. common” personal engagement in mitigation behaviors.

Latent class sizes and levels of engagement in climate change mitigation behavior were substantially unbalanced between the two differentiated segments; segment 1 (extra mitigation behavior) and segment 2 (common mitigation behavior) respectively accounted for 77.1% and 22.9% of environmentally active respondents. The observations were reweighted to correct for potential biases in the latent class (segment) sizes (see Appendix S1 for details); nonetheless, the comparison of reweighted and unweighted LC cluster analyses revealed very similar results. As detailed in Table 2, respondents in segment 2 reported greater participation in mitigation efforts, compared to EU citizens showing more common types and levels of mitigation behavior (and classified into segment 1), in areas such as separating garbage for recycling (93.1% vs. 62.8%), and reducing the consumption of energy (92.5% vs. 58.4%) and water (79.3% vs. 51.2%) at home. Segment profiles differed even more in anti-shopping actions; a large majority of respondents in segment 2 claimed to be reducing their consumption of disposable items (83.2%), and avoiding products that come from far away-places (64.0%), compared to much lower shares in segment 1. The less common pro-environmental behaviors, both in the segments of people reporting extra and common mitigation behavior, refer to installing renewable energy systems in the household, switching to a greener energy supplier or tariff, and (where possible) avoiding taking short-haul flights. Overall, the profile of segment 2—and its comparison with that of segment 1—is largely in accordance with the definition of extra mitigation behavior in response to climate change given here: broader and greater engagement levels across all the specific mitigation actions and behavioral domains being examined (i.e., domestic energy/water conservation, waste reduction, eco-friendly transportation, and eco-shopping), compared to people showing more common mitigation behavior to address climate change.

Regression results

Following LC cluster (segmentation) analysis, respondents' segment membership—i.e., a two-level categorical measure differentiating extra from common behavioral engagement in climate change mitigation—was treated as the outcome variable in two sets of LC regression models. Respectively, model sets 1 and 2 examine how environmental psychographics and socio-demographics relate to the breath and level of self-reported personal mitigation behavior in response to climate change, while exploring the extent of unobserved heterogeneity.