Segments of Environmental Concern in Kuwait

Abstract

This discussion examines segmentation by environmental attitudes in Kuwait using a survey (n = 1121) of adult Kuwaiti citizens. Attitudes were measured with a shortened version of the Environmental Attitudes Inventory (EAI) scale as adapted in prior work about the UAE. Consistent with research suggesting that environmental attitudes frequently reflect local conditions, the dimensional structure in Kuwait was not very similar to that of the UAE case. The Kuwait data do, however, yield a dimensional structure which reflects the broader debate in the sustainability literature on the relative balance of environmental vs. socio-economic concerns. Cluster analysis on these dimensions yielded five distinct segments ranging from not very strong concern about either to strong concern about both, for focusing on one or the other. Policy implications of addressing different views about environmentalism are discussed.

1. Introduction

This paper examines the segmentation of attitudes about environmental issues in Kuwait. Environmental problems and attitudes about the environment evolve over time, various countries face different conditions, and understanding of how to deal with these issues can vary. It has been known for quite some time that specific details of pro-environment attitude and behavior can depend at least partly on country socio-economic conditions [1,2,3]. Even within a particular country, consumers are also likely to have a range of views, i.e., there will be multiple segments with differing attitudes and behaviors [4,5]. Discussing the often-lamented gap between green attitudes and green behavior, Sharma [6] (p. 1225) asserts that “segmentation may help in better understanding the varied concerns of the consumers”.

A reasonable approach for investigating a specific country context is to start with a comprehensive, well-tested scale which covers the range of attitudinal issues that need to be addressed but then to adapt to country-specific thinking. This research uses an adaptation of the Environmental Attitudes Inventory scale (EAI; [7]) to examine the underlying conceptual dimensions of attitudes in Kuwait and then explores segmentation in those attitudes to highlight the different ways consumers think about and behave toward the environment. Many of the somewhat rare green segmentation studies focus on socio-demographic variables, but this has limitations. “The literature on segmenting environmentally conscious consumers suggests that … attitudinal and behavioural variables best determine green or sustainable consumer segments” [4] (p. 2).

The version of the EAI used here was taken from research adapting the EAI scale to conditions in the United Arab Emirates (UAE) [8]. This somewhat-downsized adaptation still covers a fairly wide range of views about nature, technology, and societal response and touches on the broader issues of how environmentalism interacts with socio-economic issues. The UAE is usually considered to be roughly similar to Kuwait culturally and economically, but the Kuwait survey results did not show a very similar attitudinal structure. The Kuwait data, however, did fit well into a framework outlined some time ago for understanding the range of views regarding the relative weights between ecological vs. socio-economic orientation in the debate on sustainable development [9]. While not specifically citing this framework, Whyte and Lamberton [10] categorize current views of sustainability experts as falling somewhere along a continuum between having strong environmental concern, strong socio-economic focus, or a trade-off somewhere between. Kemper and Ballantine [11] do explicitly refer to Hopwood et al. [9] in organizing how the marketing field defines “sustainability marketing”.

Kuwait is a small city-state at the head of the Arabian Gulf, situated entirely in the desert climate characteristic of much of the Middle East, particularly the Arab countries of the Gulf [12]. The most problematic issues among the Gulf Cooperation Council states (from north to south: Kuwait, Saudi Arabia, Bahrain, Qatar, United Arab Emirates, and Oman) revolve around three key resources, often called the water–energy–food nexus [13,14]. Each element has a host of related environmental issues. There is no comprehensive GCC-wide plan regarding the environment, but the common conditions and common problems foster a fair amount of environmental cooperation, loosely oriented toward achieving the UN’s sustainable development goals (SDG) [15]. Kuwait, similar to most GCC states, has strong educational content about environmental issues [16]. Thus, in general, attitude segments would reflect thinking based on at least some knowledge of the issues. It is not likely that any substantial segment is simply completely ignorant about the Gulf’s environmental issues.

Therefore, the objective here is to look at segments of environmental attitudes in Kuwait. Marketers and policy makers require knowledge of the different segments to develop good strategies and policies. Much environmental attitudes research implicitly assumes homogenous thinking. This gives somewhat misleading information; there is no one-size-fits-all strategy or policy. Yilmazsoy et al. [5], for example, say marketing managers need to understand that even several very green segments may have different considerations. Golob and Kronegger [4] recommend distinct policy elements to address different environmental thinking across a range of segments. Further, thinking is unlikely to exactly match attitudes in other countries, even in the GCC. The segments should, however, align with the broad patterns of thought structures known from assessment of such attitudes across many contexts.

Specifically, we aim to assess how well segments roughly correspond to the discussion of the range of sustainability thinking in Hopwood et al. [9], which is the main framework Kemper and Ballantine [11] use in discussing sustainability marketing. Snowball sampling was used to ask Kuwaiti citizens about their attitudes toward environmentalism, using items from a similar survey in the UAE, another Gulf country with somewhat similar conditions [8]. Factor analysis in SPSS was used to identify seven dimensions of environmental thinking, and two stage cluster analysis identified five distinct segments.

2. Materials and Methods

The next sections briefly discuss three key issues to justify the need for examining segmentation of psychological attitudes toward environmentalism. The scale here is based on a short form of the EAI, but, as with most versions of any of the major scales, it has been adapted to the situation in the Gulf. Sometimes the lack of standardization is criticized, but many observers believe that environmental attitudes are strongly influenced by local conditions. Segmentation is needed because a gap between pro-environmental attitudes and pro-environmental behavior is often observed and lamented. Part of the reason this gap seems so prominent is that researchers neglect to assess segments. Different segments have stronger or weaker attitudes about the environment and what can or should be done. Weak attitudes generally show more prominent attitude–behavior gaps in any area; but behavior is more consistent with attitude among people who have stronger attitudes. Finally, we outline the framework used here for assessing attitudes, based on a schema categorizing views on sustainability according to the relative balance between environmental vs. socio-economic concerns.

2.1. The Environmental Attitudes Inventory (EAI) Scale

For several decades, from when environmental issues began to gain prominence as an important research topic, a number of measurement scales for pro-environment attitudes have been proposed. There is little consistency about whether they measure single environmental issues or a range or are fairly specific or quite general. The EAI [7] is the foundation used here. It contains a wide coverage of relevant issues, including themes of technology, societal response, and views of nature. Possibly the most prominent alternative is the older but fairly widely used new environmental paradigm (NEP) scale. It is still often employed as a useful tool for capturing psychological attitudes toward environmentalism (e.g., [17]). Bernstein and Szuster [18] say that NEP captures the three main contemporary issues about views of technology, views of societal response, and views of nature. They question, however, whether it may focus on a somewhat outdated view of environmentalism, not fully representing the range of current views though covering issues that remain current.

Bernstein and Szuster [18] note that the EAI scale [7] is probably the most recognized alternative to NEP, because it is one of the most carefully developed. The EAI has an extensive list of items, divided into twelve dimensions and tested in two major surveys (in New Zealand and Brazil) during development. The long version has 120 items, but Milfont and Duckitt [7] also offer a “shorter” 72-item version, similarly carefully well tested. The EAI has also been used extensively, although, similar to the NEP, often with adaptation to local conditions. One useful characteristic, however, is that short forms of the scale have been carefully tested (e.g., [19,20]), including in the Middle East (e.g., [8]).

Bernstein and Szuster [18] were concerned about the lack of standardization in NEP due to wide adaptation to local conditions, an issue already recognized some time ago (e.g., [21]). This lack of standardization is also apparent with the EAI, which has been adapted to a wide range of contexts. This lack of standardization probably should not be considered a weakness; rather, a well-tested comprehensive scale can provide a useful template, but it needs to fit local conditions. Researchers are usually well aware of the variation across countries noted in the introduction [1,2,3]. Most recognize that there is no one-size-fits-all scale which can give anything but a vague, very general overview of environmental attitudes. Some degree of environmental consciousness is evident almost everywhere, but “cultural factors shape beliefs about how nature works and how individuals interact with nature” [22] (p. 194).

2.2. Environmental Consciousness and the Attitude–Behavior Gap

It is essential to understand why research often shows an attitude–behavior gap, whatever the specific attitudes are. Simply stated, relationships in social science are subject to random error. A simple regression equation, for example, does not estimate only y_i = a + bx_i; where x is the attitude component, and y is the behavior. Rather, the model is actually y_i = a + bx_i + e_i; i.e., it includes random error. Of course, if x is only one of multiple influences on y, leaving those other influences out will inflate the error, which becomes not entirely random but must incorporate impacts on y not in the model. Too much error and the equation will be non-significant, seemingly indicating no relationship of attitude to behavior. Including those additional attitude components might work better, as in y_i = a + b₁x_1i + b₂x_2i + … b_nx_ni + e_i, which puts several aspects of attitude in as antecedents to behavior. But however carefully one accounts for all the variables that can have an impact on behavior, the e_i term remains: the gap can never be completely eliminated. “Reality is certainly not deterministic and individual behavior has a certain variance” [23].

The problem is even more difficult. Failure to recognize different thinking in consumer segments means that even with multiple attitude components, it is difficult to obtain accurate estimates. Some of the attitude influences will be relevant to one segment, but not another. Mehta and Chahal [24] are skeptical that merging various consumer segments in analysis of policy responses yields useful results. The different segments have different attitudes and will respond differently to policies. Even with the range of important attitude elements in the analysis, the weaker attitude component for some segments is mixed with the stronger attitude component of other segments. In other words, component by attitude component, the analysis uses mixed data from respondents who care about the issue (so that attitude is more likely to relate to behavior) and those who do not care very much.

Some time ago, Van Doorn et al. [25] demonstrated what they called a zone of indifference: a range where attitudes are somewhat weak. Using attitudes toward organic products, they showed that in the weak-attitude range, there is essentially no relationship between attitude and behavior. Once attitudes become stronger, there is a relationship. Kumar [26], for example, found such a non-linear impact on green buying behavior in India. Bechler et al. [27] show that this sort of non-linear relationship between attitude and behavior is actually fairly common. Figure 1 illustrates this concept. We did a simple simulation for mixing two samples, one with no relationship (zone of indifference in the figure), and the other where there is a relationship. As the sample proportion shifts toward more representation for weak-attitude people, the parameter estimating the relationship declines, until it eventually becomes non-significant.

Beyond the frequently observed divergence of environmental attitudes country by country, noted in the introduction, there is also divergence segment by attitude segment. Merging the segments into one analysis does not allow very precise estimation of how attitude relates to behavior, even if the results are significant. Even some of the relatively infrequent segmentation about environmental attitudes does not solve this problem. Green segmentation traditionally has been mostly socio-demographics: this is one reason for conflicting results. As Golob and Kronegger [4] observed, attitudes and behaviors are much better ways to understand consumers. As just summarized here, this includes not just an overall merger of attitudes across various segments but assessing things segment by segment, so that weak attitudes can be distinguished from strong across the various components.

2.3. Conceptualizing Concerns of Green-Oriented Consumers

Generally, views on sustainability fall along a continuum that balances between strong environmental concern, strong socio-economic focus, or some combination of both [10]. While the discussion here is primarily about environmental concerns, it is actually not possible to fully separate these issues. The United Nations explicitly argues that its sustainable development goals are strongly interconnected: “the 17 SDGs are integrated—that is, they recognize that action in one area will affect outcomes in others, and that development must balance social, economic and environmental sustainability” [28]. This is also the view in academic discussion about the three pillars of sustainability [29,30,31]. As discussed below, items in the EAI [7] do not ignore this but include a range of views about nature, technology, and societal response. Many consumers nowadays recognize the interconnections and consider them as a package, not one pillar at a time (e.g., [32]).

There is quite a range of positions in the debate on sustainability. Here, we adopt the framework of Hopwood et al. [9], who proposed a useful mapping of where various views fit regarding their emphasis on environmental vs. socio-economic concerns. Figure 2 illustrates their basic schema. A few of the debate participants they discussed are noted in the figure because they are well known and, thus, can help in visualizing positions in the schema. The World Bank and OCED are solidly status quo, recognizing that something needs to be done, but not so much that there is any need for substantial systemic change. Strong growth can simply be better managed, often with better technology, to avoid serious problems [9] (pp. 42–43).

The Brundtland Report, the foundation of much of the modern three-pillar sustainability discussion, is positioned on the boundary of status quo and reform in the schema [9] (pp. 43–45). More needs to be done, and some aspects of the current system need to be changed to be able to do it. The reformed system, however, will enable “a new era of economic growth: growth that is forceful and at the same time socially and environmentally sustainable” [33] (p. 7). Figure 2 also notes where Hopwood et al. [9] place “green consumers”: relatively low on socio-economic concern (similar to World Bank, OECD), but on the boundary of status quo and reform on environment (similar to Brundtland). Their discussion suggests this is where the government and businesses feel they should be so that they can drive demand for economic growth.

Mainstream environmental groups—at the time their schema was proposed—were often solidly reform and came in slightly above [34] in Figure 2. They had often evolved from earlier years when they were much more focused exclusively on environmental issues, as noted in the contrast with [35] in the figure. Schumacher [36], a prominent proponent of “Small is beautiful”, is placed on the boundary between reform and transformation, which believes the current system cannot entirely solve sustainability problems and needs some degree of far-reaching change [9] (p. 45). Transformation aims for completely rebuilding the system. Some forms of transformation are outside the range of what is sustainable. Deep ecology, for example, is far into the transformation stage on the environmental axis but well below the oval representing the sustainability debate, because it is not particularly concerned about socio-economic impacts of environmental actions at all. Similarly, some views, e.g., socialist cornucopians in Hopwood et al. [9] (p. 45) are very strongly concerned with full socio-economic equality but have virtually no concern for the environment.

Positions of participants in the debate, of course, may well have evolved since 9 Hopwood et al. [9]. They even note several changes over the years up to the time of their discussion, such as IUCN and cooperating organizations noted in Figure 2. Chavalittumrong and Speece [32], for example, found some consumers are no longer “green” but have gone beyond to consider all three pillars of sustainability. The volume of discussion on sustainability has also greatly expanded. Nevertheless, the range of views still fit well within the schema, and it is frequently used to examine current work, particularly in discussions of “sustainability marketing”. Kemper and Ballantine [11], for example, say that the framework “is the most frequently cited and comprehensive typology concentrating on a wider and more specific variety of environmental, social (inequality, inequity) and economic (economic models, responsibility) aspects of sustainability”.

The work of Kemper and colleagues (e.g., [37,38]) aims to rethink how marketing can more substantially contribute to sustainability. A few other marketing scholars are also challenging marketing’s traditional role in driving consumption. Sheth and Parvatiyar [39] argue that rather than consumer demand determining how marketing can best foster consumption, marketing needs to drive the nature of demand toward a strong sustainability orientation. White et al. [40] discuss how appeal to various psychological and social motivations can help shift consumers toward more sustainable behaviors. The range of views in such “sustainability marketing” discussion probably runs from reform to slight transformation, but more traditional views of marketing are probably still dominant. Kemper et al. [38], for example, found that just over one-third of their mostly Western sample of marketing academics (n = 437) did not think sustainability was a very important issue, and saw little need to change how marketing is taught. They categorize another quarter as “moderate status quo” and just over one-third as strong concern. However, both of these groups were similar in their only-slight agreement that marketing needs to change. Less than seven percent felt strongly that sustainability is a critical problem requiring major change in marketing.

However much (or little) marketing academics think the marketing curriculum should integrate sustainability issues, the issues cannot be completely ignored. Markets are gradually shifting toward more sustainability consciousness [6]. For example, “market orientation will automatically lead organizations toward environmental sustainability as customers become more aware of the overall need for environmental sustainability” [41] (p. 33). Mitchell et al. [42] discuss this in terms of full three-pillar sustainability. Most companies, as well as policy makers, will need to understand the growing sustainability orientation in markets, but simply merging across a range of consumer segments into a single “sustainability segment” is unlikely to yield useful insight (e.g., [24]). There are multiple versions of sustainability orientation, as the framework in Hopwood et al. [9] makes clear. Here, we show that consumer attitudes can also fit well into that framework (Figure 2).

3. Methodology

Measurement of environmental attitudes was indirectly derived from the Environmental Attitudes Inventory (EAI) developed by Milfont and Duckitt [7], which is recognized as one of the most carefully developed scales for the concept [18]. With either 120 or 72 items, however, it is too lengthy to be very effective in most survey situations. Further, the scale was developed with data from New Zealand and Brazil. It may also be problematic to expect it to fully fit all countries/cultures where environmental attitudes may be somewhat different or somewhat more/less developed than in their study areas. (For example, Kuwait is substantially different from New Zealand and Brazil on Hofstede’s cultural dimensions [43].) Several researchers have worked on abbreviated versions which represent the issues in the original EAI, including one with 37 items, validated in Australia [44], and one with 36 items, validated in Portugal [45]. These two versions are not exactly alike, and each has adapted the EAI slightly differently, although the general sense of the EAI is preserved. Again, neither of these countries is very similar to Kuwait culturally [43].

AlMenhali et al. [8] adapted a 35-item version to the United Arab Emirates and reduced the scale to an even shorter version with 21 items. Their work depends heavily on statistical fit using structural equation modeling (SEM), and they eliminated some of the conceptual dimensions of the EAI. While it fits their UAE data well, there may still be questions about how well it would work in other contexts. This research used their scale, partly because it is the shortest of several EAI adaptations, so that it is the easiest to administer without unduly overloading respondents. In addition, it has been adapted for use in an Arab Gulf country with quite similar conditions to Kuwait in many aspects, including the use of regional Gulf Arabic, so that there was not much need for major changes in item wording.

Bernstein and Szuster [18], however, note that it can be problematic to rely too heavily on a specific scale, because the nature of environmental problems and the public’s understanding of them evolves. They recommend case-by-case assessments, with scales broadly including items to cover three key issues (as does the EAI): views about nature, about technology, and about societal response. For our context, this is more of an issue in attitude structure than item wording, and it did have an impact. Kuwait broadly shares culture with other GCC states; traditionally it was part of “Gulf Culture” [46,47]. Despite strong similarity, however, there are a few key differences from the UAE, notably in terms of substantially greater political participation [48] and on Hofstede’s cultural dimension assessing nurturing/caring vs. competitive/achievement-oriented. Kuwait has a relatively low score, which indicates that “the dominant values in society are caring for others and quality of life” [43]. The UAE scores more-or-less in the middle of this scale, so comparatively, it is more competitive and achievement-oriented.

Sampling and Sample Characteristics

Student groups in several sections of an undergraduate marketing class collected data using the common questionnaire assigned by the instructor. For each section, several groups had to give a brief presentation of their results, after which there was an overall discussion of what was learned about consumer views and about marketing research for uncovering those views. A graded report analyzing consumer thinking about environmental issues was submitted end-of-term, so each group had some incentive to ensure data quality. Most respondents had provided contact information, and a sampling for follow-up validation indicated little problem with the manufactured data. The student reports, of course, were largely basic descriptive statistics, but the data here combine all of the group data sets.

The data were collected by snowball sampling. Van Meter [49] shows that carefully implemented snowball sampling can be quite representative. In this research, each student in the group provided a snowball seed, so there were a great many snowball connection trails not confined to any particular social network. Generally, researchers can be more confident of sample diversity when there are a great many seeds [50]. Meta-analysis of research that has specifically compared student-recruited samples to other methods has found little difference in demographics of the sample, or in parameter estimates [51].

Snowball sampling is often quite useful for adapting sampling to specific cultural conditions (e.g., [52]). Specific to this context, it is difficult in Kuwaiti culture to directly approach potential respondents without prior introductions, especially respondents of the opposite sex. Abalkhail and Allan [53] used snowball sampling in research on women and wasta (connections) in several Gulf states, including Kuwait. Al-Salem and Speece [54] report on a number of small surveys, mostly using snowball sampling, to piece together a more comprehensive view of attitudes toward women in leadership in Kuwait.

A total of 1121 questionnaires were collected overall. Preliminary analysis of the key environmental attitude questions identified 15 questionnaires on which 3 or more of the 23 items were missing, and they were eliminated. Scattered missing data not concentrated on a single questionnaire was not considered a problem, so a total of 1106 responses were kept for analysis here.

The sample was nearly equally divided between men (50.8%) and women (49.2%). Although Kuwait does have a strong gender imbalance because of migrant labor, among citizens there is very little imbalance. The relative affluence of Kuwaiti citizens is reflected in the monthly personal spending budget. Only one-third of the sample reported this was less than KWD 500 (about EUR 1500), and one-third reported KWD 1000 to 2000. Ten percent have monthly personal spending budgets above KWD 2000. Forty-six percent of the respondents were between 18 and 25 years old, i.e., partly students, partly early-career, and 30 percent of the respondents were 25–34, being somewhat more established. (The lowest monthly personal budget was strongly concentrated in the young age category.) Only a few, however, were older than 60, so nearly all the respondents are people who are active in Kuwait’s economy now, not retirees or children. Three-fourths have a university degree, and most of those who do not are currently university students. Overall, the respondent characteristics reflect Kuwait’s affluent society (among citizens).

4. Results

Responses on the EAI are shown in Appendix A, but the item means will be briefly noted dimension by dimension, along with discussion of the EFA results, which are shown in Appendix B. The exploratory approach here implies that there is no a priori theory about the dimensional structure of the scale, and thus, confirmatory factor analysis (CFA) is inappropriate. In fact, CFA on the Kuwait data using the dimensionality in AlMenhali et al. [8] showed very serious problems, consistent with the assertion above that specifics of environmental attitudes may differ across contexts. Exploratory factor analysis (EFA) was used to determine seven useful dimensions for the Kuwaiti data.

A few items, largely about what activities people liked to do, did not load well in any interpretable dimensional structure, because of weak factor loadings, poor communality, and difficult-to-interpret associations with the other items on the factor. More than one problem was required to kick an item out, following Hair et al. [55] who advise deleting items only for major problems, not minor ones. The seven identified dimensions accounted for 63 percent of variance, with minimal crossloading (Appendix B). As Hair et al. [55] recommend, solutions with six and eight factors were tested, but they were difficult to interpret and showed a larger number of problem items. A split sample (random half/half) tested that the dimensional structure was stable, and the seven dimensions were used.

The composite variables in further analysis were the EFA orthogonal factor scores, which exclude variance not associated with the dimension [55].

4.1. Mapping Attitude Dimensions to the Broader Sustainability Debate

Each of the seven EFA dimensions can be roughly placed on the framework used here; Figure 3 shows their positions. It should be kept in mind, however, that these are very approximate. They are useful for visualizing the range of issues consumers might think about, but they do not represent a specific mindset, because consumer attitudes are a combination of the various dimensions (as Section 4.2 discusses). Varying font colors are used to help keep track of the seven dimensions in Section 5,

Table A1. Summary statistics on the environmental attitude items.

Questionnaire Item	Mean	Std. Dev
f1: Unconcerned about environment: no information about views on socio-economic axis (q19, q21, and q22)
q22: It does not make me sad to see recyclable plastic go to waste and destroy the natural environment	1.98	1.189
q21: It does not make me sad to see natural environments destroyed	1.95	1.184
q19: I do not believe protecting the environment is an important issue	1.99	1.191
f2: Eco-centered/deep ecology: environment > socio-economic (q7, q8, and q9)
q9: We should protect the environment even if it means people’s welfare will suffer	3.15	1.165
q7: Conservation is important even if it lowers people’s standard of living	3.18	1.153
q8: We need to keep the Arabian Gulf clean to protect the environment and not as places for people to enjoy water sports	3.45	1.241
f3: Techno-centered approach (q10, q11r, and q12)
q10: Science and technology will eventually solve our problems with pollution and diminishing resources	3.33	1.146
q12: Modern science will solve our environmental problems	3.34	1.145
q11r: The belief that advances in science and technology can solve our environmental problems is completely wrong and misguided	3.22	1.184
f4: Humans must adapt to nature (q14, q17, and q20)
q20: Despite our special abilities, humans are still subject to the laws of nature	3.26	1.143
q17: Humans do not have the right to damage the environment just to get greater benefits and promote economic prosperity	3.48	1.207
q14: Human beings should not tamper with nature even when nature is uncomfortable and inconvenient for us	3.17	1.166
f5: Willing to engage (q4r and q6r)
q6r: I would not want to donate money to support an environmentalist cause	3.48	1.146
q4r: I would not get involved in an environmentalist organization	3.57	1.217
f6: Socialist cornucopia: social reform > environment (q15, q16, and q18)
q16: Protecting people’s jobs is more important than protecting the environment	2.70	1.103
q15: When nature is uncomfortable and inconvenient for humans, we have every right to change and remake it to suit ourselves	3.05	1.187
q18: The benefits of modern consumer products are more important than the pollution that results from their production and use	2.45	1.151
f7: Libertarian: not specified about environment, but anti-government intervention (q1 and q2)
q2: I am opposed to governments controlling and regulating the way raw materials are used to try and make them last longer	2.96	1.141
q1: Factories and industries should be able to use raw materials rather than recycled ones if this leads to lower prices and costs, even if it means the raw materials will eventually be used up	2.68	1.211

Scale: 1 = strongly disagree, 5 = strongly agree; items with ‘r’ indicate the scale was reversed for analysis.

and

Table A2. EFA dimensions in consumer thinking.

item	f1	f2	f3	f4	f5	f6	f7	Comm.
q9.22	0.874							0.788
q9.21	0.883							0.808
q9.19	0.737							0.672
q9.9		0.780						0.633
q9.7		0.724						0.551
q9.8		0.681						0.586
q9.10			0.761					0.663
q9.12			0.769					0.678
q9.11r			0.533					0.543
q9.20				0.758				0.618
q9.17				0.695				0.592
q9.14				0.526				0.462
q9.6r					0.771			0.653
q9.4r					0.764			0.635
q9.16						0.685		0.568
q9.15						0.684		0.612
q9.18	0.434					0.530		0.540
q9.2							0.836	0.723
q9.1							0.738	0.643
rotated % variance	13.020	10.708	8.456	7.874	7.801	7.666	7.456

Loadings < 0.4 are not shown.

and on the following figures.

• f1: Unconcerned about environment: no information about views on socio-economic axis (q19, q21, and q22)

These three questions explicitly state that it does not upset respondents to see several aspects of environmental degradation. Overall, the sample disagreed with these statements strongly; with means between 1.95 to 1.99 on the 1–5 scale where 1 = strongly disagree. (But note below that one segment agreed strongly.) These are somewhat vague about what exactly they feel about the environment, and they do not indicate anything about socio-economic issues, so it was judged that this dimension probably represents a range of status quo environment thinking. The slight crossloading from f6 suggests, however, that socio-economic concerns are prioritized over the environment, so it does not extend to the parts of status quo with somewhat more environmental concern.

• f2: Eco-centered/deep ecology: environment > socio-economic (q7, q8, and q9)

These questions state several versions of environmental protection being more important than people’s welfare, standard of living, and recreational use. There was mild agreement with the statements (3.15, 3.18) and somewhat stronger for the specific issue about recreational use of the Gulf for water sports (3.45). This suggests a dimension representing deep ecology, so the green oval in Figure 3 is in the reform–transformation area. The questions do not explicitly include how unimportant socio-economic issues are, so the green oval extends upward into parts of the sustainability debate, but the view is likely outside the scope of the debate according to Hopwood et al. [9].

• f3: Techno-centered approach (q10, q11r, and q12)

These three questions suggest that science and technology will be able to solve environmental problems. They gained moderate agreement overall (3.22 to 3.34). Only q10 slightly includes socio-economic issues, by including the problem of diminishing resources. This is the techno-centered approach in Hopwood et al. [9], mostly characteristic of slightly more concerned people who are mainly status quo and into reform. This approach rarely shows really high levels of concern for socio-economic issues.

• f4: Humans must adapt to nature (q14, q17, and q20)

These three questions are stronger statements of environmental concern that those in f1, in that they explicitly state humans should not do things that damage the environment even if it would bring benefits. Agreement overall ranged from somewhat mild (3.17) to moderately strong (3.48). This thinking is in the reform range, moving toward transformation regarding the environment.

• f5: Willing to engage (q4r and q6r)

These two questions state unwillingness to actually take action supporting environmentalism, but the scale is reversed in the analysis. Thus, they represent moderate willingness to get involved, but the questions do not indicate anything about socio-economic concern. This dimension may be in the reform range on environment if socio-economic concern is lower or could possibly be in the transformation range with higher socio-economic concern.

• f6: Socialist cornucopia: social reform > environment (q15, q16, and q18)

These questions represent the socialist cornucopia thinking in Hopwood et al. [9], but overall, there was moderate disagreement to neutrality in this sample (2.45 to 3.05). One segment did agree somewhat (see below). In Figure 3, this dimension is high on socio-economic concern and low on environmental concern, representing views outside the range of the sustainability debate, according to Hopwood et al. [9]. This dimension had a slight crossloading with f1, the nature of which suggests that there may be slight environmental concern. This would extend the red oval slightly into the debate boundaries.

• f7: Libertarian: not specified about environment, but anti-government intervention (q1 and q2)

These two questions are about opposition to government intervention into managing resources. Respondents actually slightly disagree with the opposition (2.68) or are neutral (2.96). One segment, however, agreed somewhat strongly (see below).

4.2. Identifying Attitude Segments

Factor scores of the seven dimensions were used to identify five segments using cluster analysis, following standard practice outlined in Hair et al. [55]. A two-stage approach was employed, starting with hierarchical cluster analysis using squared Euclidian distances and Ward’s method. This yielded the dendogram used to identify the most appropriate number of clusters. Visual inspection suggested a five-cluster solution, but four and six clusters were also checked to see if they provided more useful segments. There was an advantage in better discrimination moving to five clusters but little additional benefit in six, so the five clusters were used. From there, the procedure closely matches that of Wallace et al. [56], who examined segments of Facebook usage. K-means clustering used the means of each dimension in the five-cluster hierarchical clustering case as seed cluster centroids. Discriminant analysis using the seven factor scores for predicting cluster membership showed 95.3 percent of cases were correctly predicted, substantially better than with a single step based on hierarchical clustering alone.

Figure 4 illustrates the characteristics for each of the five segments in terms of their scores on the seven dimensions identified in the EFA, after which the segments are briefly discussed.

segment1: Factor1 is the only dimension that is strong for this segment, indicating that these people are not very concerned about the environment. Factor5 shows slight unwillingness to engage with environmental organizations or make donations, consistent with the lack of concern on factor1, but this is not very strong. Factor6 shows slight concern with socio-economic issues without much concern for the environment. These consumers seem more-or-less not to really be thinking about environmental issues at all.

segment2: Factor4 is quite strong here, indicating thinking that humans should adapt to nature. Factor2 is somewhat prominent also, suggesting that environmental protection has some priority over the social and economic concerns, while factor6 shows strong disagreement that social and economic considerations should override environmental ones. Factor5 is likewise somewhat prominent; these consumers are willing to donate to environmental causes and engage with environmental organizations. They disagree fairly strongly with the thinking that various environmental issues are not important enough to worry about (factor1). They mildly disagree that science and technology will be able to solve all environmental problems (factor3), as well as with libertarian thinking that the government should not regulate environment issues (factor 7). On the map of debate about sustainability in Hopwood et al. [9], this seems to be a strongly eco-centered approach.

segment3: These consumers also disagree fairly strongly that environmental issues are not important enough to worry about, but they have different thinking about what to do about it. They are fairly strongly libertarian, i.e., agreeing on factor7 that the government should not regulate environment issues. They seem to view these issues more as an individual responsibility and are quite willing to get involved (factor5). They are somewhat ambiguous about whether strong action on socio-economic issues (factor6) or environmental protection (factor2) should take precedence; both are strong, probably reflecting a feeling that both are important and need to be balanced. There is fairly strong disagreement that humans must adapt to nature at any cost (factor4), but little belief that science and technology will solve all problems (factor3).

segment4: These people mildly agree that they are not very concerned about the environment (factor1), but they also show mild willingness to engage with environmental organizations or donate to environmental causes (factor4). Likely this reflects a lack of interest broadly but some interest in a few specific issues. They definitely do not believe that environmental issues are more important than socio-economic welfare (factor2) or that humans must adapt to nature rather than attempt to control it (factor4), but neither do they believe that socio-economic considerations take precedence. This thinking seems to be more or less about balance, as in segment3, but with an orientation toward relying on the government to solve problems, rather than individual action. The segment disagrees with opposition to government regulation (factor7).

segment5: This segment also disagrees fairly strongly that environmental issues are not important enough to worry about (factor1), but they are quite unwilling to engage themselves (factor5). They somewhat disagree with thinking that the government should not regulate these issues (factor7), and they have the strongest faith in science and technology solving environmental problems (factor3). There is mild disagreement that environment should be a prioritized over socio-economic issues (factor2) and also mild agreement that jobs and living standards are more important (factoro6).

Table 1. Summary of segment descriptions.

	Description
segment1	Very unconcerned about environment. No other dimension goes beyond fairly weak attitudes, whether positive or negative. Strongly status quo in [9].
segment2	Strong thinking that humans must adapt to nature and strong thinking that environment takes priority over socio-economic concerns. This is deep ecology in [9].
segment3	Mostly fairly strong attitudes, positive for willing to engage, socio-economic, and libertarian; somewhat strong for eco-oriented. Strong disagreement about being unconcerned with environment. This borders on transformation in [9].
segment4	Strong to somewhat strong disagreement with most elements. Slightly willing to engage and disagree with a libertarian approach. Citizens and the government should both do something but nothing radical. Status quo bordering on reform in [9].
segment5	Fairly strong disagreement that they are not concerned with environment but not willing to get engaged. Strongest belief of any segment that technology can solve things. Status quo bordering on reform in [9].

summarizes these segment descriptions, and Figure 5 illustrates the placement of these five segments on the framework in Hopwood et al. [9]. As can be seen, the consumer segments range across the scope of the sustainability debate. Segment2 even seems to be an example which Hopwood et al. [9] say is not really sustainable because there is no balance of very strong environmental concern with socio-economic implications. A socialist cornucopia segment (very strong concern on the socio-economic dimension, but very little concern for the environment) is not represented here. This is probably simply an artifact of the scale used here. The EAI contains enough about socio-economic issues to assess trade-offs with environmental concerns, but it does not have sufficient detail to determine that the socio-economic aspects are far stronger than environmental concerns.

5. Discussion

The results of the large consumer survey have demonstrated several assertions made in the discussion here. First, even a well-tested scale, such as various modifications of the EAI, cannot be used uncritically. The individual items, taken from an adaptation to a Gulf environment (UAE), did make sense to Kuwaiti respondents. However, the Kuwait data did not fit well at all into the dimensionality found in the UAE. The Kuwaiti sample perceived a much different dimensionality. Some of the literature already suggests that attitudes are likely to differ depending on local culture and conditions across countries (e.g., [22]).

Second, much work on segmentation of environmental concerns tends to focus on demographic segmentation, but some observers say attitude segmentation is a more useful approach (e.g., [4]). Here, the environmental attitudes show very distinct dimensionality, which reflects the issues covered in the Hopwood et al. [9] framework summarizing the sustainability debate. Cluster analysis using the attitude dimensions uncovered in the Kuwaiti data give five distinct segments which correspond well with specific positions in that framework. The assertion that attitude segments will show a range of views is supported. There is no one-size-fits-all view about the environment, or more broadly, sustainability, among Kuwait consumers. Skepticism about merging various consumer segments [24] seems well justified. Here, an average across all five segments would not give useful information representing any segment for policy makers.

Third, the way Kuwaiti consumers think about the issues can easily fit into the broad discussion of sustainability organized in the Hopwood et al. [9] framework; i.e., these sorts of views are consistent with the range of academic and policy debate about the issues. Policy makers do not need to learn entirely new views; rather, they need to recognize that policy must address many of the multiple ways of looking at environmentalism/sustainability which have already been under discussion for some time.

The overall consumer attitudes are constructed from a mix of the various attitude dimensions. Some of the dimensions can be very weak in particular segments. Figure 4 shows that several segments barely consider some dimensions at all, but all segments have one or a few dimensions that are their key concerns. According to Van Doorn et al. [25], this suggests that those weak-attitude dimensions will play little role in how attitudes may translate into actual intentions or behavior toward the environment. The stronger attitudes might translate to these, with proper policy to encourage behaviors.

A specific policy implementation is likely to influence people in some segments but not others. In a few cases, the specific policy, attractive to some, may actually be disliked by another segment when their attitude about the issue is strongly different. The data here show substantial disagreement among the segments about what the most important issues are. In most segments, attitudes on some of the dimensions are quite weak but strong on other dimensions. This is no different than in the broader debate on sustainability. Kuwait maps well to the wide range of views about the environment and how it trades off with socio-economic issues.

Future research is needed to assess trade-offs consumers might be willing to make if policies must balance quite different views about what needs to be done regarding sustainability. For example, there are different priorities across segments between environmental and socio-economic concerns, as well as how big or small the government’s role should be. It is impossible to maximize what any segment wants without alienating some other segment. How can policy be structured to optimize sustainability impact?

Carefully crafting policy to account for a range of consumer thinking implies a methodology which facilitates this. Simply copying scales (such as the EAI, [7]) from prior work, no matter how good they are in their original context, does not account for well-documented variation in environmental attitudes depending on local conditions [1,2,3]. On the other hand, an entirely new scale does not allow much comparison. The EAI, used here, has a number of adaptations which allow better measurement of attitudes in local contexts (e.g., [19,20]). Ideally, scale adaptation should start from already-adapted work in a place with similar environmental issues and socio-cultural aspects (here, [8]). Kuwait is quite similar to the UAE in many respects. Nevertheless, there are also some subtle differences. Examining these carefully may require modifying item wording and, often, revising the dimensional structure of attitudes. With these adaptations, we can see that Kuwaiti consumers fit into the broader range of thinking about sustainability [9]. But there is a broad range, as is well known in sustainability discussions. Exactly where Kuwaiti segments fit may be slightly different from where segments are positioned in other contexts.

6. Conclusions

All of this suggests that policy to move Kuwaiti consumers toward stronger environmentalism must be nuanced. The AFED [16] points out, for example, that Kuwait has made a substantial effort at environmental education, and environmental consciousness has been present for some time. Most citizens are well aware of the issues but, nevertheless, often do not take much action [57]. Given that awareness is not much of a problem, a simple policy of more education will have little impact for most segments. In segment1, more effective education might help, but it would need to focus more on impacts. Segment1 people, already somewhat aware, need to clearly understand how lack of attention to the problems is detrimental to Kuwait’s long-term prospects and their own lives. Other segments mostly already agree something should be done, but there is no consensus on what exactly is important to do.

Another example shows similar issues. Recent research on water policy [57,58] uncovered strong views on the need for fairness and not harming the interests of lower-income citizens. Such thinking is reflected here in the socio-economic concerns that come up in thinking about environmental issues in some segments. For people in these segments, environmental policy must include mechanisms to improve conditions along the socio-economic axis in the framework or they might not be accepted. In general, the different views across segments suggest that multiple policy elements must be worked out which will appeal to the different segments, and they must be coordinated so that elements which some like do not alienate others. Very strong eco-centered policy, for example, would be attractive to segment5. But it would alienate several other segments, even if they care about the environment, if it fails to account for socio-economic aspects of sustainability.

None of the segments can be entirely ignored, because they are roughly similar in size, and even the smallest, segment2, is still substantial. Most already have attitudes that are conducive to contributing in some way to stronger environmental protection/sustainability. But there is little actual policy to harness the fairly strong awareness, as, for example, is apparent from the water policy example just noted in [57,58], which shows the willingness of citizens to cooperate, provided water policy is consistent with their beliefs. This study suggests that the same is true for much broader environment/sustainability issues.