Promoting purchases of sustainable groceries: An intervention study

PROMOTING PURCHASES OF SUSTAINABLE GROCERIES Promoting purchases of sustainable groceries: An intervention study Daniel Hanssa and Gisela Böhma The final publication is available at Elsevier via http://dx.doi.org/10.1016/j.jenvp.2012.10.002 Author Note a Faculty of Psychology, University of Bergen, Christies gate 12, 5020 Bergen, Norway. E-Mail Daniel Hanss: daniel.hanss@isp.uib.no; E-Mail Gisela Böhm: gisela.boehm@psysp.uib.no Correspondence concerning this article should be addressed to Daniel Hanss, Tel: +47 555 888 99. 1 PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 1 Abstract Previous research suggests that consumers can be encouraged to purchase environmentally friendly groceries by means of persuasive messages. The present intervention study investigated whether providing information about how consumers can help mitigate environmental and social problems through everyday purchasing decisions can (a) strengthen consumers’ intentions to purchase sustainable groceries (e.g., ecological and fair trade foods), (b) promote actual purchases of sustainable groceries, and (c) strengthen selfefficacy beliefs in the domain of sustainable development (the assumed determinant of sustainable consumption). The results suggest that the intervention strengthened consumers’ intentions to purchase domestic, seasonal, and certified ecological products. In addition, the intervention promoted the actual purchasing of certified ecological and fair trade products. Effects of the intervention on self-efficacy beliefs were, however, nonsignificant. We discuss whether increased knowledge and consciousness might have accounted for some of the observed effects on purchasing intentions and actual purchases of sustainable groceries. Keywords: behavioral change, consumption, sustainable development, self-efficacy, intervention, information. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 2 Promoting purchases of sustainable groceries: An intervention study 1. Introduction When declaring the Millennium Development Goals in 2010, the UN underlined that sustainable consumption is an important pillar of environmental sustainability and of combating poverty (Marrakech Process Secretariat: UNDESA and UNEP, 2010). Because consumption activities of private households contribute substantially to environmental problems such as climate change and acidification of soil and water (for an overview see Ölander & Thøgersen, 1995), any serious progress towards sustainability will require that private people change their consumption habits. One way in which private consumers can contribute to sustainable consumption is by choosing environmentally friendly and fair trade products instead of conventional products when shopping for groceries (Doran, 2009; Tanner & Wölfing Kast, 2003; Tobler, Visschers, & Siegrist, 2011). For example, ecological foods help reduce the environmental impact associated with conventional farming (cf. Jungbluth, Tietje, & Scholz, 2000), and fair trade products support vulnerable small-scale producers by guaranteeing above-average market prices (cf. Doran, 2009). Currently, the United Nations Commission on Sustainable Development is preparing a ‘10-year framework of programmes’ on sustainable consumption and production. One of the aims is to support regional and national initiatives to promote sustainable consumption. An example for an ambitious national initiative is Norway’s sustainable development strategy that incorporates sustainable consumption with an emphasis on reducing any negative impact on the natural environment (Ministry of Finance, 2011). To that end, the Norwegian government has decided that by the year 2020, 15% of Norwegian food production and consumption should be ecological (Bye, Aarstad, Løvberget, Berge, & Hoem, 2010). This target appears quite ambitious given that, at present, Norway is among the countries with the lowest demand for ecological foods (i.e., about 1% of overall food sales) in Western Europe PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 3 (Framtiden i våre hender, FIVH, 2011). What is more, in Norway, the supply of many ecological products (e.g., ecological milk, eggs, and meat) exceeds their demand so that large shares of these ecological products are sold as conventional products (FIVH, 2011). Hence, the crucial question is: How can we encourage consumers to purchase sustainable groceries? The term sustainable groceries here refers to grocery products that are in line with the concept of sustainable consumption; in other words, products that help reduce the environmental and social impacts of conventional production and consumption (cf. Belz and Bilharz, 2005). According to this usage of the term, certified ecological and fair trade groceries can be regarded as sustainable groceries. Intervention studies that have addressed the question of how consumers can be encouraged to purchase sustainable groceries are scarce (cf. Lehman & Geller, 2004). Notable exceptions are the studies by Geller, Farris, and Post (1973), De Young et al. (1993), Bamberg (2002), and Biel, Dahlstrand, and Grankvist (2005). Bamberg showed that consumers can be encouraged to purchase ecological foods by fostering implementation intentions and providing monetary incentives. In Biel et al.’s study, participants shopped in a computer-simulated grocery store. After environmental values had been made salient by help of a poster, the participants purchased more ecological products. In the studies by Geller et al. (1973) and De Young et al. (1993), purchases of sustainable products (e.g., products with less or returnable packaging and products with fewer harmful chemicals) increased when environmental and economic arguments for buying these products were provided. In Geller et al.’s study, participants were additionally provided with information about how the targeted sustainable behavior contributes to environmental preservation, and in De Young et al.’s study, with information about how to perform the targeted behavior. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 4 A shortcoming of the studies by Geller et al. (1973) and De Young et al. (1993) is that the interventions were evaluated with regard to changes in the targeted consumption behaviors only but not with regard to changes in the determinants of these behaviors. According to Abrahamse, Steg, Vlek, and Rothengatter (2005, 2007), it is important to also investigate changes in behavior determinants in order to understand how exactly interventions affect the target behavior and how interventions can be further improved. A large body of psychological research has investigated which individual characteristics are correlated with preferences for sustainable products and services. Among these variables are attitudes (e.g., Schwepker & Cornwell, 1991), values (e.g., Collins, Steg, & Koning, 2007), and future orientation (e.g., Urien & Kilbourne, 2011). Efficacy beliefs, in particular, have been shown to be an important correlate of sustainable consumer behavior (e.g., Hanss & Böhm, 2010; Roberts, 1996). In recent versions of the theory of planned behavior (e.g., Fishbein & Cappella, 2006), the assumption is that self-efficacy beliefs determine behavioral intentions, and that behavioral intentions are the main determinants of actual behavior. One of the challenges of sustainable consumption is that many consumers need to contribute in order to make a significant difference. Because collective efforts are needed, individuals may believe that there is little they can personally do to promote sustainable consumption. Another challenge is that many of the benefits of sustainable consumption, such as a healthy natural environment, accrue to society whereas many of the costs, such as higher purchase prices of ecological products, must be borne by individuals. If consumers do not believe that their sustainable behaviors will make a difference, that is, if consumers have low self-efficacy beliefs, they may not be willing to accept the costs of sustainable consumption. This is a possible explanation for why efficacy beliefs can be expected to be an important predictor of sustainable consumption behaviors. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 5 The purpose of the present study was to investigate the effects of an informational intervention strategy on purchasing intentions, actual purchases of sustainable groceries1, and consumers’ self-efficacy beliefs with regard to contributing to sustainable development. Before turning to the study, we will outline key issues for interventions that are designed to promote sustainable consumption and elaborate on the assumed relation between self-efficacy beliefs and decisions to buy sustainable products as well as on strategies for strengthening self-efficacy beliefs. 2. Key issues for interventions designed to promote sustainable consumption Steg and Vlek (2009) review intervention studies in the domain of environmental psychology and provide a list of issues that should be considered in the promotion of environmentally friendly behavior. We assume that these issues are also crucial for the success of interventions that are designed to promote sustainable consumption (environmental friendliness being an important aspect of sustainable consumption). The four main issues are (cf. Steg & Vlek, 2009, Table 1): (a) selecting the behaviors to be changed, (b) identifying the determinants of these behaviors, (c) choosing an appropriate intervention strategy for changing the behaviors, and (d) evaluating the effectiveness of the intervention. With regard to the selection of the behaviors to be changed, Steg and Vlek (2009) stress that one should choose behaviors that have significant impacts (in our case: on sustainable development) and for which the desired changes are feasible. The intervention applied in the present study was targeted at purchases of sustainable groceries, such as ecological and fair trade foods and cosmetics. Promoting the purchase of sustainable groceries seems feasible in Norway because sustainable alternatives are available for many grocery products on the Norwegian market; for some products, the supply of sustainable alternatives largely exceeds their demand (see Section 1). PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 6 Concerning the determinants of sustainable consumption, we investigated sustainable development self-efficacy, that is, the degree to which individuals believe that they can personally contribute to sustainable development. In a previous cross-sectional study, we documented that sustainable development self-efficacy predicted purchases of sustainable groceries (Hanss & Böhm, 2010). Whether sustainable development self-efficacy is causally related to the purchasing of sustainable groceries has not been investigated before. This was one of the aims of the present study. With regard to intervention strategies, Steg and Vlek (2009) distinguish between structural and informational strategies. Structural strategies target changes in the decision context. Informational strategies aim to strengthen individual determinants (e.g., self-efficacy beliefs) of the behavior to be promoted. In this study, we used an informational strategy aimed at promoting sustainable consumption through strengthening sustainable development self-efficacy. Previous research has demonstrated that both environmental behaviors in general and sustainable consumption in particular can be promoted by means of informational strategies (e.g., De Young et al., 1993; Geller et al., 1973; combination of information and other intervention strategies: e.g., Abrahamse, et al., 2007; Staats, Van Leeuwen, & Wit, 2000). Our study complements these previous ones in a number of ways. First, most informational strategies used in the domain of grocery shopping focused on information about environmental effects, for example how to reduce waste. The effects of information about social impacts, for example on the working conditions of small-scale producers, received little attention. Second, some of the used measurements of consumption decisions focused on specific product categories, for example soft drinks in Geller et al.’s (1973) study. Third, there is a lack of research that assessed actual purchasing behavior over a longer period of time. Examining intervention effects over time is important if the aim is to understand whether consumers fall back to old habits after the intervention and hence whether the intervention PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 7 leads to long-lasting changes of consumption patterns. This study accounted for these limitations by including information about environmental and social effects of grocery consumption in the intervention and by assessing purchasing decisions concerning various product categories over a period of several weeks. As for the evaluation of the intervention’s effectiveness, we investigated changes in sustainable consumption (i.e., self-reported purchasing intentions and actual purchases) and in self-efficacy beliefs. 3. Self-efficacy beliefs and sustainable consumption Albert Bandura (e.g., 1977, 1997, 2006) coined the term self-efficacy to refer to the degree to which an individual feels capable of performing the behaviors that are required to bring about certain desired outcomes. The outcomes that are of central interest in connection with sustainable consumption are contributions to three dimensions of sustainable development (cf. UNWCED, 1987): environmental preservation, a socially fair distribution of resources, and economic welfare. Various cross-sectional studies have investigated self-efficacy beliefs with regard to environmental outcomes (e.g., climate change; Kellstedt, Zahran, & Vedlitz, 2008). Another line of research has investigated self-efficacy specifically in connection with consumer behavior. For example, Ellen, Wiener, and Cobb-Walgren (1991) use the term perceived consumer effectiveness (PCE) to refer to the degree to which consumers believe that they can contribute to environmental preservation through their consumption behaviors. PCE can thus be conceived of as self-efficacy with regard to the behavioral domain consumption and the outcome domain environmental preservation. Several studies have found that consumers who score high on PCE have stronger preferences for sustainable products than consumers who score low on PCE (e.g., Berger & Corbin, 1992; Vermeir & Verbeke, 2006). In studies by Roberts and colleagues, PCE turned out to be a more important correlate of environmentally PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 8 friendly consumer behavior than other individual characteristics, such as altruism (Straughan & Roberts, 1999) and environmental concern (Roberts, 1996). And yet other studies have investigated the relation between locus of control and sustainable consumption. Rotter (1966) distinguishes between internal and external locus of control, where the former represents a tendency to believe that events or outcomes of situations are a result of one’s own actions, and the latter represents a tendency to believe that events or outcomes of situations are controlled by external factors and are independent of one’s own actions. A common finding in this line of research is that people with an internal locus of control are more likely to engage in sustainable consumption behaviors than people with an external locus of control (e.g., Allen & Ferrand, 1999; Cleveland, Kalamas, & Laroche, 2005, 2012; Schwepker & Cornwell, 1991). Allen and Ferrand (1999) as well as Cleveland et al. (2005, 2012) measured locus of control in connection to environmental preservation; the former used the Environmental Action Internal Control Index (SmithSebasto, 1992), whereas the latter used different, self-developed instruments to measure environmental locus of control (ELOC). PCE and ELOC are conceptually similar to self-efficacy beliefs concerning environmental preservation, which is the degree to which one believes that one can personally help preserve the environment. Environmental preservation constitutes one dimension of sustainable development. Hanss and Böhm (2010) introduced the concept of sustainable development self-efficacy (SDSE); this concept covers (a) three dimensions of sustainable development (i.e., environmental preservation, socially fair distribution of resources, and economic welfare) as outcomes and (b) consumers’ perceived influence on these outcomes through everyday behaviors in general and consumption in particular. SDSE is thus a broader concept than PCE and ELOC. In a cross-sectional study among Norwegian consumers (N = 402) Hanss and Böhm (2010) explored the dimensionality of SDSE and the relation between PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 9 SDSE and sustainable consumption. It was found that consumers’ perceived direct influence on sustainable development and their perceived indirect influence (i.e., through encouraging other consumers to contribute to sustainable development) made up separate facets of SDSE. Self-efficacy regarding both one’s direct and indirect influences on sustainable development contributed to predicting purchases of sustainable groceries. These findings are in line with a study by Cleveland et al. (2012) demonstrating that people’s perceived direct and indirect impact on the environment are represented by separate facets of internal ELOC. A compound measure of internal ELOC that comprised both people’s perceived direct and indirect impact predicted various sustainable consumption behaviors. In sum, these studies suggest that self-efficacy beliefs are related to sustainable consumption. It was one aim of this study to investigate whether strengthening self-efficacy beliefs could promote purchases of sustainable groceries. We adopt the more encompassing concept of SDSE rather than restricting ourselves to the narrower concepts of PCE or ELOC. 4. Strengthening sustainable development self-efficacy beliefs A key issue of SDSE is whether consumers believe that their own actions make a difference for sustainable development. We assume that doubts in the efficacy of one’s own consumer behaviors might stem from two sources: (a) from a lack of knowledge about the environmental and about the social effects of grocery production and consumption, or (b) from the collective nature of sustainable consumption -- the fact that it requires the joint efforts of many consumers to make a significant contribution to sustainable development so that one’s individual actions have little effect in isolation (cf. Hanss & Böhm, 2012a). Therefore, we provided consumers with information that addressed these two potential diminishers of self-efficacy. More specifically, we presented examples of how people can reduce their negative environmental and social impact by buying sustainable instead of conventional groceries. In addition, we illustrated the cumulative effects that can emerge if a PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 10 consumer’s purchases of sustainable groceries encourage other consumers to do the same. One possibility of asserting influence on other consumers is through increasing other’s expectations of collective success. If other people notice one’s purchases of sustainable products they may become more optimistic that sustainable consumption can be attained, given the fact that joint efforts are necessary to make a difference. Consequently, others may be more willing to contribute their share by purchasing sustainable products. Another possibility of influencing other consumers is through social norms. One’s purchases of sustainable groceries can increase social norms for sustainable consumption and compliant behavior by others. In line with these assumptions, Von Borgstede, Dahlstrand, and Biel (1999) found that people were more willing to purchase sustainable products if they believed that others buy these products too. Goldstein, Cialdini, and Griskevicius (2008) provide empirical support for the influence of social norms on sustainable behaviors. These forms of indirect, socially mediated influence on sustainable development are particularly likely if our product purchases are visible to other people who surround us in everyday life. One part of the intervention was therefore to present examples of everyday situations in which choice of sustainable grocery products is socially visible, for example, when we eat lunch with colleagues, and to explain that in these situations other people may get encouraged to contribute to sustainable consumption. We assumed that by providing this information, we would be able to strengthen consumers’ SDSE. The use of persuasive information has been shown to strengthen efficacy beliefs in other domains, such as anti-discrimination (cf. Stewart, Latu, Branscombe, & Denney, 2010). 5. Hypotheses The following hypotheses were tested: PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 11 Hypothesis 1 (purchases of sustainable groceries): As a result of the intervention, the number of sustainable purchases will increase in the intervention group but not in the control group. Hypothesis 2 (purchasing intentions): As a result of the intervention, participants’ intentions to purchase sustainable groceries (PI) will increase in the intervention group but not in the control group. Hypothesis 3 (sustainable development self-efficacy): As a result of the intervention, participants’ sustainable development self-efficacy (SDSE) will increase in the intervention group but not in the control group. 6. Method The intervention study had a randomized pretest-posttest design with a six-month follow-up and was conducted from November 2009 to June 2010 over the internet. 6.1 Participants One hundred forty-five residents of the Bergen community (Norway) participated in the study. With approximately 260,000 inhabitants, Bergen is the second largest city in Norway. The participants were between 18 and 70 years of age (M = 38, SD = 14), and 93 participants (64%) were female. The median household income (yearly, after tax) category was NOK 350,000 to 449,000 (i.e., approx. $61,000 to $78,000 US) (17%), and most participants were married or had a life partner (63%) and held a college degree (64%). Compared to the population of the Bergen community,2 people between 20 and 66 years were overrepresented in our sample (63% in Bergen vs. 92% in our sample) as were women (50% female in Bergen), people with a college degree (35% in Bergen), and married people (33% in Bergen). Households with small incomes (i.e., less than 150,000 NOK / approx. $26,000 US) were slightly overrepresented, and households with high incomes (i.e., more than 550,000 NOK / approx. $96,000 US) were slightly underrepresented compared to the greater Bergen PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 12 area. The participants were randomly assigned to one of two groups: an intervention group (n = 73, M = 39 years of age, SD = 13; n = 48 female) and a control group (n = 72; M = 38 years of age, SD = 15; n = 45 female). We recruited the participants through newspaper announcements and mailbox postings (i.e., 2,000 flyers distributed to randomly selected households in different parts of the city). The newspaper announcements and flyers informed potential participants that the aim of the study was to learn about people’s opinions and preferences related to everyday consumption. No reference to the topics sustainability or sustainable consumption was made during the recruitment. All participants were offered a budget for purchasing groceries via the internet platform of the study. The groceries were delivered to the participants’ homes after the study by the authors. Together with the groceries, each participant received any unspent money that remained in the budget. The groceries and the budget remainder served as incentives for participating in the study. Moreover, cinema vouchers were raffled among those who completed the study. 6.2 Measures The study comprised several measures that are not related to the focus of this paper. Here, we report only those measures that are of interest with regard to our research questions. 6.2.1 Purchases of sustainable groceries We provided the participants with a budget of 500 NOK (i.e., approx. $90 US) to purchase groceries via the internet platform of the study. In each of four parts of the study (two before and two after the intervention), participants were presented with two pairs of grocery products. Each pair consisted of a sustainable product (i.e., certified as ecological and/or fair trade) and a conventional product of the same quantity and product category. A wide variety of product categories was used (e.g., orange juice, tea, strawberry jam, spaghetti, olives). Participants had to choose between the sustainable and the conventional product of PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 13 each pair. For example, the two pairs presented to a participant on one occasion might have consisted of two brands of orange juice (one ecological, the other conventional) and two brands of spaghetti (one ecological, one conventional); the participant would then have been required to choose one of the two brands of orange juice and one of the two brands of spaghetti. The number of sustainable products chosen per study part (minimum zero, maximum two) was recorded for each participant and used in the data analyses. The two product pairs were presented at the beginning of each study part. The order of the product categories was randomized across the study parts with different random orders for different participants. Consequently, in each study part, different participants were presented with different product categories, but at the end of the study, all participants had chosen from all product categories. The products were represented by a picture of the packaging, brand and product name, quantity, and purchase price. Sustainable products were described as “økologisk” (i.e., ecological) or “fair trade,” and the respective eco and fair trade labels were displayed on the packaging of the products. A previous study had revealed that the displayed labels were well known and considered indicative of sustainable groceries by Norwegian consumers (cf. Hanss & Böhm, 2012b). Purchase prices matched the average prices of the products in local supermarkets and were lower for the conventional products than for the ecological and fair trade product alternatives. Figure 1 displays an example of how the product pairs were presented. The maximum sum that the participants could spend for the products was smaller than the budget they were provided with. In other words, the budget was sufficient so that they could choose the sustainable alternative for all choices and still have some remaining budget left at the end of the study. At the beginning of the study, we informed the participants that they would receive the remainder of the budget together with the purchased groceries. In addition, prior to each product purchase, we reminded the participants that they had enough PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 14 money left to choose the product that appealed most to them. By means of providing participants with the budget remainder, we intended to implement a conflict between saving money and paying extra for certified ecological and fair trade products. 6.2.2 Intentions to purchase sustainable groceries Purchasing intentions (PI) were measured by means of a questionnaire that was composed of 18 items. The questionnaire was an adapted version of a questionnaire that we used in previous research to measure the habitual purchasing of sustainable groceries3 (see Hanss & Böhm, 2010). The items consisted of statements about the purchasing of groceries (e.g., food and cosmetics) with various sustainable attributes, such as fair trade or environmentally friendly production. The choice of product attributes that we included in the PI questionnaire was based on a study in which we explored which product attributes Norwegian consumers consider important for sustainable groceries (Hanss & Böhm, 2012b). Example items are: “When I buy food and have the choice, I will buy products that guarantee fair payment to the producers” and “When I buy fruits and vegetables and have the choice between ecological and conventional products, I will buy ecological products.” The items also included statements about the purchasing of groceries with eco and fair trade labels; for example, “When I buy cosmetics and have the choice, I will buy products with the Nordic Swan label.” Participants indicated how likely they were to engage in the described purchasing behavior on an 11-point scale that ranged from 0 (no, definitely not) to 10 (yes, definitely). 6.2.3 Sustainable development self-efficacy SDSE was assessed by a questionnaire that was composed of 24 items. Twenty items were adopted from an SDSE questionnaire that we presented in Hanss and Böhm (2010). These items measure various aspects of SDSE: (a) SDSE with regard to the direct impact that one’s actions (i.e., actions in general and buying behavior) have on sustainable development; PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 15 (b) SDSE with regard to one’s indirect impact on sustainable development, that is, through encouraging other people to promote sustainable development; (c) self-efficacy with regard to promoting three dimensions of sustainable development: environmental preservation, a socially fair distribution of resources, and economic welfare; and (d) self-efficacy with regard to mitigating climate change (adopted from Kellstedt et al., 2008). Four additional questionnaire items measured SDSE specifically with regard to contributing to sustainable development by purchasing sustainable products; for example, “By purchasing sustainable (e.g., ecological and fair trade) products, I can encourage more sustainable agricultural practices.” Participants indicated how much they agreed with each of the 24 statements on a fivepoint scale that ranged from 1 (strongly disagree) to 5 (strongly agree). 6.3 Procedure and materials Participants registered for the study on a website that could be accessed via an internet link that was provided in the newspaper advertisements and mailbox postings. The website was developed in cooperation with a web designer and served as the platform for the participant registration, the product purchases, the intervention, and the questionnaires for measuring purchasing intentions and self-efficacy beliefs. In the registration section of the website, people were informed about the duration and aims of the study (i.e., “learning about people’s opinions and preferences related to everyday consumption”). In addition, it was announced that participants would be offered a compensation “package” worth 500 NOK (i.e., approx. $90 US) for completing the study and that this package would be composed of monetary compensation and products that they would choose during the course of the study. To further increase the incentives for completing the study, we announced that a cinema voucher worth 1,000 NOK would be raffled among those participants who completed all parts PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 16 of the study. Another 10 cinema vouchers (each voucher worth one movie visit) were raffled among those who completed the follow-up. Those who decided to participate could enter their personal information (i.e., age, gender, yearly household income, and e-mail address) into a form provided in the registration section of the website. After the registration period, we deactivated the registration form and sent invitation e-mails to those who had signed up for the study. The invitation emails contained a personalized link to the first part of study and a deadline for participation. The study was comprised of nine parts -- a timeline of the study including the instruments and materials used in each study part are provided in Table 1: Parts 1 (1st week) and 2 (2nd week) served as the baseline measurement (pre-test). In both parts, participants were first asked to purchase their preferred products from the two pairs of groceries that we presented on the webpage.4 Then they filled out the questionnaires to measure purchasing intentions and self-efficacy beliefs. In Parts 3, 4, 5, and 6 (3rd to 6th week), participants took part either in a four-step intervention or in a four-step control condition (the two conditions are described in Sections 6.3.1 and 6.3.2). Product purchases, PI, and SDSE were not measured in Parts 3 to 6. Parts 7 (7th week) and 8 (8th week) resembled Parts 1 and 2 and served as the post-test measure of product purchases, PI, and SDSE. After completion of Part 8, we delivered the purchased products and the budget remainder to participants’ homes. Six months later, we contacted the participants again and asked them to take part in a follow-up (i.e., Part 9) in which we measured PI and SDSE. The follow-up allowed us to investigate whether the effects of the intervention had diminished over time. Our dependent variables were measured in Parts 1, 2, 7, and 8. In each of these four parts, participants first chose their preferred products from two pairs of groceries, and then they filled out questionnaires to measure PI and SDSE. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 17 For each part of the study, separate invitation emails were sent to those participants who had completed the preceding study part. Each invitation e-mail contained a link that participants used to log in to the internet platform of the study (i.e., participants who completed all study parts logged into the platform once a week for eight weeks -- Parts 1 to 8 -- and once more for Part 9, which was the follow-up that took place six months after Part 8). Those participants who had not completed a study part within four days received a reminder e-mail. 6.3.1 Intervention condition The intervention was comprised of four steps. In the first step (Study Part 3), we informed participants about (a) major environmental problems (e.g., pollution, deforestation, climate change) and their consequences for humans and other species, and (b) socio-economic problems (e.g., poverty and related health issues). We gave examples of these problems and their consequences in Norway and in other (mostly developing) countries. The aim of this first intervention step was to increase consumers’ awareness of the presented environmental and social problems. The aim of the second intervention step (Study Part 4) was to show that human activities are among the major causes of the problems that we presented in the first intervention step. The following activities were addressed: overconsumption of natural resources in general, energy use and greenhouse gas emissions, and littering and emissions of toxic substances. With regard to the agents involved in these activities, we covered private households (e.g., private transportation) and industrial and agricultural enterprises (e.g., production and transportation of products). Concerning poverty, we pointed out that the unfair payment of local producers in developing countries partly accounts for poor working and living conditions in these countries, and that small profit margins might hinder producers in developing countries from adopting environmentally friendly practices. We assumed that by PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 18 highlighting the fact that human activities are among the major causes of these problems, consumers would gain greater confidence in humanity’s ability to mitigate the problems. The first and second intervention steps did not directly target consumers’ sustainable development self-efficacy beliefs. The information provided in these intervention steps was meant to build a foundation for the subsequent two intervention steps in which we provided information that specifically targeted SDSE. The aim of the third intervention step (Study Part 5) was to strengthen consumers’ self-efficacy beliefs with regard to directly contributing to sustainable development (i.e., through their own actions). We informed the participants about sustainable consumption alternatives that would help mitigate the problems described in the first part of the intervention. Examples of consumption alternatives that were covered are: commuting by public transportation instead of by private car or purchasing locally produced groceries in order to reduce one’s carbon footprint, buying ecological instead of conventionally grown fruits and vegetables to reduce one’s impact on the environment, buying fair trade instead of conventionally traded products to help improve the working and living conditions of local producers. In order to familiarize consumers with labels that indicate sustainable products, we depicted examples of eco and fair trade labels (e.g., the Norwegian Debio økologisk label and the Fairtrade Max Havelaar label) together with the consumption alternatives. As for Fairtrade Max Havelaar, we further explained what the organization stands for and gave two examples of farmers in developing countries whose working and living conditions had been improved through fair trade. In order to illustrate opportunities for reducing one’s carbon footprint, we introduced two fictitious Norwegian consumers whose consumption habits differed with regard to how much they contributed to CO2 emissions. We contrasted the specific consumption habits of the two consumers (e.g., “never buys ecological or seasonal food” vs. “sometimes buys ecological or seasonal food” or “does not recycle glass and paper” vs. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 19 “recycles most glass and paper”) and compared their carbon footprints. At the end of the third intervention step, we emphasized that one’s consumption activities are a means of “voting” for more or less sustainable ways of industrial and agricultural production, product transportation, and trade. The aim of the fourth intervention step (Study Part 6) was to strengthen consumers’ self-efficacy beliefs with regard to indirectly contributing to sustainable development (i.e., through encouraging other people to consume sustainably). We provided examples of situations in which one’s own sustainable behaviors might be visible to other people (e.g., one’s purchases of ecological and fair trade foods might be visible to colleagues during lunch breaks at work -- particularly if the sustainable foods are indicated by eco and fair trade labels). We explained that if other people take notice of one’s sustainable behaviors, they may become encouraged to join in because they may realize that they would not be the only ones contributing to sustainable development. In addition, we illustrated “domino effects” (e.g., cumulative reductions of CO2 emissions) that would occur if one encouraged people in one’s environment to consume sustainably, who in turn encouraged some of their friends, colleagues, or relatives to consume sustainably (and so on). At those points in the intervention where we used statistics or pictures to illustrate the presented topics, we provided their sources from the literature or the internet. At the end of each part of the intervention, the participants were asked questions concerning the contents of the intervention. 6.3.2 Control condition Participants who took part in the control condition were given several tasks that dealt with aesthetic judgments that were unrelated to sustainable consumption: In Study Part 3, we asked the participants to match colors to car types (e.g., sports car or sedan); in Part 4, we PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 20 asked the participants to rate the same colors and car types according to how much they liked them; and in Parts 5 and 6, the colors and car types were rated on semantic differential scales. 7. Data handling and preliminary analyses Data were analyzed with the statistical package SPSS 19. We replaced missing data by means of multiple imputation (Rubin, 1996) before we analyzed the effects of the intervention. A more detailed description of the missing data and of the multiple imputation procedure is provided in Section 7.3. Graham (2009) recommends computing index scores of the variables with missing data before running the multiple imputation in cases in which the number of variables entered into the imputation exceeds 100. This was the case in our study. Two separate principal component analyses (orthogonal varimax rotation, eigenvalues greater than one, listwise deletion) were conducted to explore the structure of SDSE and PI. Only the items that measured SDSE and PI in Study Part 1 were entered into the principal component analyses. The data were suited for principal component analysis as indicated by the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy (SDSE: .86; PI: .89) and Barlett’s test of sphericity [SDSE: approximate χ2 (276) = 3541.43, p < .001; PI: approximate χ2 (153) = 1368.59, p < .001]. Based upon the loadings of the items on the extracted components, we computed index variables for SDSE and PI. 7.1 Computing index scores for SDSE For SDSE (n = 145 in Study Part 1), we retained five components. Together, these components explained 77.35% of the total variance in SDSE. The first component, SDSE Others, explained 18.02% of the variance (after rotation) and represented self-efficacy beliefs with regard to motivating others to promote sustainable development. The second component, SDSE Domestic, explained 16.24% of the variance and represented self-efficacy beliefs with regard to contributing to sustainable development locally (i.e., in Norway). The third component, SDSE Global, explained 14.92% of the variance and represented self-efficacy PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 21 beliefs with regard to contributing to sustainable development around the world. The fourth component, SDSE Sustainable Products, explained 14.13% of the variance and represented self-efficacy beliefs with regard to contributing to sustainable development by purchasing sustainable products. The fifth component, SDSE Environment-Climate, explained 14.05% of the variance and represented self-efficacy beliefs with regard to contributing to environmental preservation and mitigating climate change. These components resemble the SDSE facets found by Hanss and Böhm (2010), particularly the distinction in self-efficacy regarding one’s indirect impact (by motivating others) and self-efficacy regarding one’s direct impact on sustainable development. For each of the components, we computed an index variable by averaging participants’ scores on those items with loadings greater than .5 on the respective component. Cronbach’s alpha values were: α = .94 (SDSE Others, six items), α = .89 (SDSE Domestic, five items), α = .93 (SDSE Global, four items), α = .91 (SDSE Sustainable Products, four items), and α = .82 (SDSE Environment-Climate, five items). Next, we computed SDSE index variables for Study Parts 2, 7, 8, and for the follow-up. In order to facilitate a comparison of the index variables across study parts, we combined the SDSE items in the same way that we did for the index variables from Part 1. 7.2 Computing index scores for PI For PI (n = 119 in Study Part 1), we retained three components. Together, these three components explained 63.75% of the total variance in PI. The first component, PI ResourceSaving-Ethical, explained 30.12% of the variance (after rotation) and represented intentions to purchase groceries with various qualities, such as little and recyclable packaging, little energy use for production and product shipping, humane animal treatment, and fair payment of producers. The second component, PI Ecological Foods, explained 18.66% of the variance and represented intentions to purchase certified ecological foods and intentions to purchase PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 22 products directly from farmers. The third component, PI Domestic-Seasonal, explained 14.97% of the variance and represented intentions to purchase domestic products and seasonal fruits and vegetables. We computed index variables for the PI components for each study part (i.e., for Parts 1, 2, 7, 8, and the follow-up) following the same procedure that was used for the SDSE index variables. For the first part of the study, Cronbach’s alpha values were: α = .91 (PI ResourceSaving-Ethical, 10 items), α = .88 (PI Ecological Foods, five items), and α = .75 (PI domestic-seasonal, three items). 7.3 Handling of missing data Of the N = 145 participants (n = 73 intervention, n = 72 control), 143 completed Part 1 (n = 71 intervention, n = 72 control), 132 completed Part 2 (n = 67 intervention, n = 65 control), 111 completed Part 7 (n = 54 intervention, n = 57 control), 104 completed Part 8 (n = 50 intervention, n = 54 control), and 61 (n = 27 intervention, n = 34 control) completed the follow-up. Differences between the intervention and control groups in the numbers of participants who completed versus did not complete the individual survey parts were all nonsignificant. Participants who did not complete a study part were not invited to participate in subsequent study parts. Whereas the proportion of participants who dropped out during Parts 1 to 8 was relatively small, many dropped out after Part 8. Six months of time separated Part 8 and Part 9, which is a fairly long time period. We assume that this is the reason for the high drop-out rate after Part 8. In order to investigate whether participants who completed all study parts differed from those who dropped out during Parts 1 to 8 and from those who dropped out after Part 8, we compared the three groups with regard to age, gender, SDSE, PI, and general environmental attitudes. We used data from Study Part 1 for this comparison. The only PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 23 significant difference was for age, F(2, 142) = 6.38, p = .002. Bonferroni post hoc tests revealed that those who completed all parts of the study were significantly older (M = 43 years) than both those who dropped out during Parts 1 to 8 (M = 34 years), p = .004, and those who dropped out after Part 8 (M = 36 years), p = .027. Age was positively, albeit weakly, related to some of the dependent variables: SDSE Global (Part 1: r = .20, p = .017), PI Resource-Saving-Ethical (Part 1: r = .19, p = .022; Part 7: r = .18, p = .038), PI Domestic-Seasonal (Part 1: r = .24, p = .004; Part 2: r = .22, p = .009; Part 7: r = .25, p = .004; Part 8: r = .23, p = .011), and the number of sustainable products bought by the participants (Part 2: r = .21, p = .014; Part 7: r = .19, p = .032). However, it should be noted that the intervention and control groups did not differ with regard to age (i.e., in none of the study parts), and, thus, differences between the two groups on the dependent variables cannot be attributed to age differences. As a result of participants dropping out of the study and the possibility of skipping single PI items to indicate that one generally never buys the product type covered by the item (e.g., if a participant wanted to indicate that he/she never buys cosmetics), 83 data points (i.e., 1%) were missing in study Part 1, 512 (8%) in Part 2, 1484 (24%) in Part 7, 1808 (29%) in Part 8, and 3511 (58%) in the follow-up. The number of possible data points was 6235 for Parts 1, 2, 7, and 8 and 6090 for the follow-up. We replaced missing data by means of multiple imputation (except for missing answers to questions asked during the intervention; see Section 8.4).5 More specifically, we used the automatic imputation method available in SPSS 19. Multiple imputation has been shown to perform well with large numbers of missing values (even as much as 50%) in the dependent variables and with sample sizes comparable to the one of the current study (cf. Graham, 2009). PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 24 Variables that were entered into the imputation model were the product choice variables, the SDSE and PI index variables, and several auxiliary variables6: age and gender of the participants, variables measuring participants’ attitudes toward ecological and fair trade groceries, attitudes toward regional and healthy groceries, general environmental attitudes, perceived social norms of purchasing sustainable groceries, and moral judgments related to purchasing sustainable groceries. Five imputed data sets were generated -- according to Rubin (1996), three to five imputed data sets are sufficient in many cases. The five imputed data sets were generally similar. We imputed the data separately for the intervention and control groups and combined the complete data sets afterwards for subsequent analyses. Imputed values that fell outside of the possible range of scores (e.g., imputed values for purchasing intentions that were either negative or greater than 10) were rounded to the next possible value (cf. Enders, 2010). 8. Results Before the intervention, that is, in Parts 1 and 2 of the study, the differences between the intervention and control groups with regard to product purchases, SDSE, and PI were small (cf. Figures 2 to 4). In Part 1, two group differences were significant (both on SDSE indices, see Figure 4): SDSE Global [t(143) = -2.07, p = .040] and SDSE EnvironmentClimate [t(143) = -2.04, p = .044]. In addition, the mean difference for SDSE Others was marginally significant: t(143) = -1.97, p = .051. None of the group differences in Part 2 was significant. A straightforward approach for investigating whether the intervention had effects on the dependent variables would be to conduct a series of two-way analyses of variance (ANOVAs) with participant group (i.e., control vs. intervention) as a between-subjects factor and time (i.e., the measurement points) as a within-subjects factor. So far, however, there are no standard rules for pooling ANOVA results -- particularly the F statistics -- in multiple PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 25 imputation (Van Ginkel, 2010, p. 5). Thus, we formulated the ANOVAs in terms of a general linear model. In each of the models, the dependent variable was either purchases of sustainable products, one of the PI indices, or one of the SDSE indices. Participant group (intervention vs. control), time (measurement points), the group x time interaction, and participant ID were entered as predictors. Our main focus was on changes in the dependent variables from Study Part 2 (immediately before the intervention) to Part 7 (immediately after the intervention). We proceeded as follows: If a group x time interaction was significant, we further explored differences between the intervention and control groups by means of between-group post-hoc tests. If a group x time interaction was nonsignificant but the effect of time was significant, we further explored changes across time by means of within-group post-hoc tests. We used the imputed data sets in all analyses that are reported in sections 8.1 to 8.3. We report the pooled results, that is, parameter estimates combined across the five imputations. 8.1 Effects on purchases of sustainable groceries Figure 2 displays the mean numbers of sustainable groceries bought by the participants in each study part. According to Hypothesis 1, we assumed that after the intervention, the mean number of sustainable groceries purchased by the participants would increase in the intervention group but not in the control group. Indeed, one week after the intervention (i.e., in Part 7), the average number of sustainable groceries purchased by participants in the intervention group had increased by 22.7% and remained almost unchanged in Study Part 8. By contrast, the participants in the control group purchased only slightly more (i.e., 3.5%) sustainable groceries in Part 7 than in Part 2. To test whether these changes were significant, we formulated a general linear model with the number of sustainable groceries purchased by the participants as the dependent PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 26 variable. The group x time interaction turned out to be nonsignificant (cf. Table 2), indicating that the increase in purchases of sustainable groceries over the course of the study did not differ between the intervention and control groups. However, there was a significant effect of time (p < .001). We further explored how the number of sustainable purchases changed from one study part to the next by means of t tests. These t tests were conducted separately for the control and intervention groups. Significance levels for the six tests were adjusted to maintain the familywise error rate (Bonferroni correction, p = .0083). We first investigated whether participants in the intervention group and in the control group changed their purchasing behavior from Study Part 1 to Part 2. Two t tests were conducted, one for the intervention and one for the control group, in which we compared the mean number of sustainable groceries purchased in Part 1 with the mean number of sustainable groceries purchased in Part 2. The results revealed that the increase in sustainable purchases was significant for both groups, t(1202) = -3.84, p < .001 (intervention), t(11519) = -3.29, p = .001 (control),. Then, we explored the increase in sustainable purchases between Part 2 and Part 7. This difference was significant only for the intervention group, t(239) = -2.67, p = .0082 (intervention), t(300) = -0.46, p = .647 (control). These results indicate that participants in the intervention group purchased more sustainable groceries one week after the intervention than one week before the intervention, whereas participants in the control group did not change their purchasing behavior in the same time period. Changes from Part 7 to Part 8 were nonsignificant for both participant groups, t(19) = -0.21, p = .983 (intervention), t(98) = -1.36, p = .178 (control). In conclusion, there was a general increase in the number of sustainable purchases across time. The change from Part 2 to Part 7 (i.e., from before to after the intervention) was different for the intervention and control groups in that the intervention group increased PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 27 significantly, whereas the control group did not. This result is in accordance with Hypothesis 1, which predicted that the intervention would increase sustainable purchases. 8.2 Effects on intentions to purchase sustainable groceries The mean ratings of intentions to purchase sustainable groceries are depicted in Figure 3, separately for the three PI index variables. Three regression models were formulated (cf. Table 2); in each of the models, one of the PI components was entered as the dependent variable. We hypothesized that as a result of the intervention, participants’ intentions to purchase sustainable groceries would increase in the intervention group but not in the control group (Hypothesis 2). Support for our hypothesis came from the analyses of PI DomesticSeasonal: The interaction effect of group x time was significant (p = .039; cf. Table 2). Pairwise comparisons revealed that participants who had been exposed to the intervention reported stronger intentions to purchase domestic and seasonal groceries in Study Part 8 and in the follow-up than participants in the control group, t(38) = -2.43, p = .020, and t(35) = 2.40, p = .022, respectively. For Parts 1, 2, and 7 (cf. Table 1), differences between the control and intervention groups were nonsignificant: t(2575) = -0.46, p = .642; t(45) = 0.14, p = .887; and t(96) = -1.63, p = .107, respectively. Thus, the effect of the intervention on PI DomesticSeasonal was strong enough to reach significance only two weeks after the intervention and lasted until the follow-up. Support for Hypothesis 2 was less clear with regard to the index variables PI Resource-Saving-Ethical and PI Ecological Foods. The effect of the group x time interaction was nonsignificant for both of these PI indices (cf. Table 2). The effect of time was marginally significant for PI Ecological Foods (p = .071). When exploring changes in PI Ecological Foods over time, we found a significant increase from Part 2 to Part 7 in the intervention group, t(32) = -2.78, p = .009, but not in the control group, t(45) = -0.94, p = PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 28 .352. Changes in PI Ecological Foods from Part 1 to 2, Part 7 to 8, and Part 8 to the follow-up were nonsignificant in both participant groups (highest t value 1.20; lowest p value .230). In sum, these findings suggest that the intervention had effects on intentions to purchase domestic and seasonal products and -- somewhat less clear -- on intentions to purchase ecological foods. The intervention did not affect the facet representing resourcesaving and ethical consumption. Therefore, we can conclude that Hypothesis 2 was supported by differences between the intervention and control groups in two of the three PI facets. With regard to exploring the relation between purchase intentions and behavior, Table 3 shows the associations between purchasing intentions and actual purchases of sustainable products pooled across the intervention and control groups. All but one of these correlations were significant. Hence, the overall picture that emerges from these data is that the PI indices are positively associated with the number of sustainable products that participants bought (i.e., the stronger a participant’s purchasing intentions, the more sustainable products he/she bought). The absolute size of these correlations is moderate, but roughly in line with the literature on the relation between intentions and environmental behavior (cf. Bamberg & Möser, 2007). In the next section, we address the question of whether the intervention affected participants’ self-efficacy beliefs. 8.3 Effects on sustainable development self-efficacy beliefs We hypothesized that as a result of the intervention, participants’ sustainable development self-efficacy beliefs would be strengthened in the intervention group but not in the control group (Hypothesis 3). Figure 4 indicates that the intervention group and the control group hardly differed from each other with regard to changes in self-efficacy beliefs. For two of the SDSE indices, SDSE Domestic and SDSE Global, there was a (marginally) significant effect of time (p = .047 and p = .089, respectively) (cf. Table 2). T tests revealed PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 29 that SDSE Domestic increased marginally from Part 7 to Part 8 in the intervention group, t(129) = -1.94, p = .054, but not in the control group, t(25) = -1.24, p = .226. None of the other comparisons was significant (highest t value 0.73; lowest p value .476). Changes in SDSE Global were nonsignificant in both participant groups (highest t value 1.19; lowest p value .237). Taken together, these results indicate that self-efficacy beliefs were not affected by the intervention. 8.4 How well was the provided information conveyed? The finding that self-efficacy was not affected by the intervention raises the question of how well the provided information was conveyed to the participants. We addressed this question by means of exploring participants’ answers to the questions that were asked at the end of each intervention step. Note that these questions were asked only to participants in the intervention group; thus, we cannot compare these responses to the control group, but we believe that they are informative nevertheless. The first intervention step informed participants about environmental and socioeconomic problems. At the end of the intervention step, we asked participants to indicate for each of five major problems (i.e., poverty, air-pollution, water-pollution, deforestation, and climate change) how well-informed they felt about the problem on a 5-point rating scale ranging from 1 (very poorly informed) to 5 (very well-informed). The mean ratings suggest that the participants felt quite well-informed about climate change, poverty, and air-pollution (M = 3.97, SD = 0.71; M = 3.92, SD = 0.65; M = 3.83, SD = 0.63, respectively). Participants felt somewhat less informed about deforestation and water-pollution (M = 3.62, SD = 0.80; M = 3.57, SD = 0.75, respectively), lowest t value 1.91; highest p value .061. The second intervention step informed participants that human activities are among the major causes of the environmental and socio-economic problems addressed in the first intervention step; the intention was to create confidence in humanity’s ability to mitigate the problems. At PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 30 the end of the intervention step, we presented participants with 15 statements about the covered topics and asked participants to indicate the degree to which they agreed with the statements on a five-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). Participants agreed with statements that claimed that consumption-oriented lifestyles in industrialized countries contribute to environmental and socio-economic problems (average ratings from M = 4.08, SD = 0.96 to M = 4.43, SD = 0.64). Moreover, agreement with statements claiming that industrialized countries have the means to mitigate the problems was high (M = 4.08, SD = 0.90 to M = 4.29, SD = 0.81), whereas participants tended to disagree with the statement that developing countries have the means to solve the problems themselves (M = 2.16, SD = 1.11) and with the statement that countries affected by environmental problems should solve the problems themselves (M = 2.41, SD = 1.06), lowest t value 9.06; all p values < .001. There was broad agreement that industrialized countries should take a leading role in mitigating environmental problems and in fostering sustainable development (M = 4.56, SD = 0.59; M = 4.41, SD = 0.69), and Norway was believed to have the means to contribute to solving environmental problems (M = 4.41, SD = 0.66). Concerning the ability of Norwegian consumers to make a difference, agreement with two statements about their ability to help mitigate environmental problems was high (M = 4.17, SD = 0.77 and M = 4.16, SD = 0.88). However, participants agreed somewhat less with the statement that Norwegian consumers could help improve working and living conditions in developing countries (M = 3.90, SD = 0.96), t values 3.00 and 3.41; p values .004 and .001. Agreement that Norwegian consumers can foster sustainable development by choosing sustainable products was moderate (M = 3.98, SD = 0.99) and did not differ significantly from the other three statements about Norwegian consumers’ ability to make a difference. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 31 The third intervention step informed participants about sustainable consumption alternatives and provided examples of everyday consumption decisions that help reduce a person’s contribution to environmental and socio-economic problems. The aim was to create confidence in participants’ perceived ability to personally make a difference. At the end of the intervention step, we asked participants whether there was something that they could do in the course of the current week to contribute to sustainable development (open answer question). Participants named a wide range of behaviors; most frequently mentioned were recycling and composting waste (24 times), changing one’s diet -- for example, eating less meat (19 times), buying fair trade (17 times) and ecological (16 times) products -- and taking public transportation (15 times). Examples of behaviors that were mentioned several times were: avoiding packaging (11 times), saving energy at home (10 times), buying locally produced goods (9 times), buying less and making more conscious purchase decisions (each 7 times). Behaviors that were mentioned by only a few participants were, for example, reusing and repairing clothing and equipment, buying used products (each mentioned 3 times), buying seasonal produce (2 times), and spreading the information provided by the intervention (1 time). In addition, we presented participants with a list of 10 behaviors that were covered in the third intervention step and asked them to indicate for each of the behaviors how much they thought they could contribute to sustainable development by carrying out the behavior (five-point rating scale: 1 = very small contribution to 5 = very large contribution). Participants thought that they could make a large contribution by recycling glass and paper, avoiding products with unnecessary packaging, and reducing household waste (average ratings from M = 4.00, SD = 0.89 to M = 3.79, SD = 0.99). The perceived ability to contribute to sustainable development was somewhat lower for buying fair trade products, taking public transportation instead of a private car, buying seasonal and local products, eating less meat, PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 32 and buying ecological instead of conventional products (M = 3.54, SD = 1.10 to M = 3.26, SD = 1.11), lowest t value 1.94; highest p value .057. The purchasing of recycled goods (M = 3.70, SD = 0.96) was perceived to be more effective than eating meat and buying ecological and local products, lowest t value 2.46; highest p value .017, but less effective than recycling glass and paper, t(60) = 2.87; p = .006. The fourth intervention step illustrated how one’s own sustainable behaviors could encourage other people to join in (e.g., in situations in which one’s own sustainable behaviors are visible to one’s neighbors, colleagues, friends, or relatives). At the end of the fourth intervention step, participants were asked to indicate for each of five behaviors how likely they thought they would be to encourage other people to join in by means of carrying out the behavior. The behaviors (commuting by public transportation; having a message -- for example, at one’s mailbox -- to signal that one does not want to receive unaddressed advertisements; buying ecological products; buying fair trade products; recycling glass and paper) were rated on an 11-point scale ranging from 0 (definitely no) to 10 (definitely yes). Participants thought that they would be more likely to encourage others to join in if they recycled and had a no-advertisement message (M = 7.36, SD = 2.72 and M = 7.14, SD = 2.86, respectively) than if they carried out the other three behaviors (M = 6.09, SD = 2.45 to M = 5.90, SD = 3.12), lowest t value 3.12; highest p value .003. Furthermore, we asked participants if they could come up with everyday situations -- other than those covered by the intervention -- in which sustainable behaviors are visible to others (open answer question). Most participants replied by giving examples of behaviors that are visible to others (some of these behaviors were covered by the intervention -- others were new), such as using environmentally friendly reusable shopping bags, composting, and driving a fuel efficient car. In sum, these findings suggest that the information provided in the intervention reached the participants; most importantly, participants thought that they could contribute to PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 33 sustainable development directly (i.e., through their own behaviors, such as recycling and avoiding products with unnecessary packaging) and indirectly (e.g., recycling will encourage others to join in). Moreover, the intervention inspired participants to come up with ways of contributing to sustainable development that were not covered by the intervention. Hence, we conclude that the informative goals of the intervention were generally achieved. 9. Discussion In the presented study, we tested an informational intervention strategy that aimed to promote intentions to purchase sustainable groceries, actual purchases of sustainable groceries, and sustainable development self-efficacy beliefs. We found that participants in the intervention group purchased more sustainable groceries one week after the intervention than one week before the intervention. Participants in the control group did not significantly change their purchasing behavior during this time period. Furthermore, the intervention had effects on purchasing intentions: two weeks after the intervention and in the follow-up, intentions to purchase domestic and seasonal products were higher in the intervention group than in the control group. Intentions to purchase ecological foods were higher one week after the intervention than one week before the intervention among participants in the intervention group; this difference was not found in the control group. In conclusion, our findings suggest that sustainable consumption can be promoted by means of providing information about how individuals can reduce their environmental and social impact through everyday purchasing decisions and about how individual decisions to consume sustainably may encourage other people to join in. That sustainable consumption can be promoted by informational strategies is in line with previous studies (e.g., De Young et al., 1993; Geller et al., 1973). Perhaps even more importantly, our findings indicate that the effects of the intervention on purchasing intentions and actual purchases of sustainable groceries did not PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 34 diminish over time (for intentions up to the six-month follow-up). However, it should be noted that a self-selection bias might have partly accounted for this finding: It is possible that participants in favor of sustainable consumption were more likely to sign up for the follow-up than those with little interest in sustainable consumption. One hint that the composition of the sample might have changed over time is the finding that participants who completed all study parts including the follow-up were somewhat older than participants who dropped out during the course of the study. Age was positively, albeit weakly, related to two of the PI index variables and to the purchasing of sustainable products. Effects of the intervention on intentions to purchase domestic and seasonal products (PI Domestic-Seasonal) were among the strongest effects. Previous research suggests that Norwegian consumers tend to prefer foods with a Norwegian origin over imported foods and to consider domestic foods safer than imported foods (Berg, Kjærnes, Ganskau, Minina, Voltchkova, Halkier, & Holm, 2005). Berg et al. (2005) also found that trust in food control authorities is higher in Norway than in its neighboring countries, Denmark and Russia -- high trust in food control authorities might be a possible reason for why Norwegians consider domestic foods to be safe. These findings suggest that purchases of domestic and locally produced groceries constitute a domain of sustainable consumption that is particularly easy to promote in Norway. One facet of purchasing intentions was not affected by the intervention: the facet PI Resource-Saving-Ethical that represented various product qualities, such as little and recyclable packaging, fair trade, packaging that uses little energy for production, and humane animal treatment. A possible explanation for this finding is that the facet comprises purchases of products with sustainable qualities that were not covered sufficiently by the intervention. For example, the intervention did not cover humane animal treatment and the foregoing of PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 35 animal testing; both attributes were found to be of medium importance for sustainable foods and cosmetics by Norwegian consumers (Hanss & Böhm, 2012b). Contrary to our assumptions, sustainable development self-efficacy beliefs were not affected by the intervention. Terragni and Kjærnes (2005) state that Norwegians tend to hold positive attitudes and trust in public bodies and consumer organizations. This claim is supported by the findings of Berg et al. (2005). Perhaps self-efficacy beliefs were hardly affected by the intervention in the present study because Norwegian consumers attribute responsibility for sustainable development to “powerful others” such as politicians, but not to themselves. Vermeir and Verbeke (2006) tried to strengthen consumers’ perceived efficacy by means of giving an example of how pressure exerted by consumers can help improve the working and living conditions of banana farmers in Latin America. An additional statement underlined that individual consumers can make the world a better place by counteracting unsustainable practices. Contrary to the authors’ expectations, but similar to our results, the messages did not strengthen consumers’ perceived ability to make a difference. The information provided by our intervention was more comprehensive than the information that Vermeir and Verbeke (2006) provided in their study, and participants’ answers to the questions that we asked during the intervention indicated that the presented information was well-conveyed to participants. When asked during the intervention, participants believed that they could contribute to sustainable development directly and indirectly through various consumption-related activities. And yet, self-efficacy beliefs were not affected by our intervention. Thus, our findings suggest that providing information about how consumers can directly and indirectly contribute to sustainable development does not suffice for strengthening self-efficacy beliefs. Strengthening SDSE might require providing people with the opportunity to actually PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 36 experience that their behaviors make a difference (i.e., providing mastery experiences; cf. Bandura & Adams, 1977). Thus, self-efficacy beliefs might be more affected if a feedback component is added; for example, by informing consumers about how their everyday consumption decisions can reduce their ecological footprint. One domain where feedback strategies have been used is energy consumption. The feedback measures used there include the use of monitors that display daily energy consumption and monetary savings over time (for an overview, see Abrahamse, et al., 2005). With regard to grocery consumption, however, providing feedback about environmental and social impacts may be more complicated because these impacts depend on a multitude of product features (e.g., production, distribution, and trade) and consumption-related factors (e.g., shopping and disposal of the product). Hence, feedback systems would have to integrate product life-cycle assessment data with people’s consumption habits. We can conclude that the intervention affected purchasing intentions and actual purchasing behavior in the expected direction, but these effects were not due to changes in self-efficacy beliefs. If not self-efficacy beliefs, what other determinants of sustainable consumption might have been affected? Torjusen, Lieblein, Wandel, and Francis (2001) refer to several studies among citizens of Nordic countries in which it was found that people’s willingness to pay for environmentally friendly products depended on whether or not people believed that the products were linked to environmental problems. These findings suggest that knowledge about how the production, trade, and consumption of groceries are related to environmental and social problems might be a decisive factor in decisions between conventional and sustainable groceries. The information that we provided in Steps 2 and 3 of the intervention touched upon the environmental and social impacts of consumption activities and might thus have increased participants’ knowledge in this domain. Unfortunately, we did not measure changes in consumers’ knowledge about the environmental and social impacts of PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 37 grocery consumption. We can, however, refer to feedback that participants provided concerning the study (i.e., after Study Part 8): Ten out of 25 intervention group participants who commented on the study indicated that they had learned something new. For example, one participant (female, 29 years) wrote (translated from Norwegian): “The study was interesting and I learned a lot. I altered my choices during the study and I became more conscious about my purchasing and recycling habits.” The example also suggests that participants became more conscious about their purchasing decisions. Two other comments by intervention group participants pointed in this direction; for example, “I liked this study -it made me think more about the environment and my own actions” (female, 26 years). An interesting finding with regard to purchases of sustainable groceries was that participants in both conditions were more likely to choose sustainable products at the end of the study than in the first part of the study (the effect was stronger in the intervention group, though, cf. Figure 2). From participants’ feedback, it became apparent that some of the control group participants also became more conscious about their purchasing choices during the course of the study (eight out of 20 comments made by control group participants pointed in this direction). For example, one control group participant (female, 58 years) wrote: “[The study] made me think about my own lifestyle and my relationship to the resources on this planet.” Thus, it seems possible that merely responding to the questionnaires and being regularly presented with sustainable products increased participants’ awareness of sustainable development-related issues and strengthened their preferences for sustainable products. We see five limitations of the current study that should be considered in future research: First, participants did not use their own money for the product purchases. This is a limitation insofar as people may be less reluctant to pay extra for ecological or fair-trade products if they use money that was given to them instead of their own money. As a consequence, preferences for sustainable products may have been stronger in our study than PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 38 what could be expected in “real life.” However, this effect should have been equally strong in the control and intervention conditions. Second, the relatively large proportion of participants who dropped out of the study after Part 8 limits the extent to which valid conclusions can be drawn for the follow-up. Thus, future studies should try to reduce drop-out rates at the followup, possibly by providing larger incentives. A shorter time period for the follow-up would probably have led to a lower drop-out rate, but would also have rendered the results less informative. Third, our study would have benefited if we had measured participants’ knowledge of the topics that we covered in the intervention. A pre-/post-intervention measure of knowledge in both participant groups would have allowed us to investigate whether the information that we provided increased participants’ knowledge about sustainable consumption and whether learning about sustainable consumption issues can explain the intervention effects on PI and product purchases. Research by Frick, Kaiser, and Wilson (2004) suggests that knowledge about behavior alternatives and their effectiveness is important for promoting conservation behavior. Fourth, the present study does not allow any conclusions to be drawn with regard to how effective the individual intervention steps were. Therefore, if one aims to isolate the effects of the different intervention steps, they should be tested separately. This will reveal whether the single intervention steps also work independently or only in conjunction with each other. Fifth, it cannot be ruled out that some of the obtained effects on self-reported purchasing intentions and actual purchases are due to socially desirable response patterns. Particularly the participants in the intervention group may have felt a pressure to comply with the agenda of the researchers or with more general social norms to act in socially and environmentally responsible ways. We believe that, despite these limitations, the present study has helped improve our understanding of how sustainable consumption can be promoted. What are the main conclusions for practitioners? First, the market share of sustainable products can be increased PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 39 by informational campaigns. Second, as our results showed, informational campaigns can be effective for changing behavior, also without affecting self-efficacy beliefs. It is often argued that structural strategies, for example subsidies or green taxes, are more efficient for promoting sustainable behaviors than informational strategies (e.g., Steg & Vlek, 2009). However, informational strategies can help increase public support for policy measures and can therefore serve as a supplement to structural strategies. In this regard, future research should further investigate by means of longitudinal designs how persistent behavior changes, induced by informational campaigns, are and whether the new behaviors translate into other forms of engagement, for example, membership in environmental organizations or parties and opinion leadership. The information used in the intervention was quite extensive and was therefore provided in several steps, over the course of four weeks. “Real life” campaigns of similar length in which consumers do not receive a monetary incentive to participate may find it useful to embed the information in a format that attracts, involves, and entertains consumers, such as an educational game. A digital format may be particularly well suited to accomplish this; for example, a multiplayer online game that provides information about sustainable consumption alternatives and simulates a player’s indirect impact on sustainable development through encouraging other players to join in. In addition, the consumption decisions of the individual players could be made public in the player community. This may increase both social norms to consume sustainably and compliant behavior. To facilitate regular participation in the game, it may be useful to make the game accessible from mobile devices, such as smart phones or tablet computers. As for the target group of informational campaigns, a promising strategy may be to address primarily those consumers who already have an interest in sustainability topics and who may later serve as opinion leaders. This target group should be comparatively easy to PROMOTING PURCHASES OF SUSTAINABLE GROCERIES motivate even if no monetary incentives are provided. 40 PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 41 References Abrahamse, W., Steg., L., Vlek, C., & Rothengatter, T. (2007). The effect of tailored information, goal setting, and tailored feedback on household energy use, energyrelated behaviors, and behavioral antecedents. Journal of Environmental Psychology, 27, 265-276, doi: 10.1016/j.jenvp.2007.08.002. Abrahamse, W., Steg, L., Vlek, C., & Rothengatter, T. (2005). A review of intervention studies aimed at household energy conservation. Journal of Environmental Psychology, 25, 273-291, doi:10.1016/j.jenvp.2005.08.002. Allen, J. B., & Ferrand, J. L. (1999). Environmental locus of control, sympathy, and proenvironmental behavior: A test of Geller’s actively caring hypothesis. Environment and Behavior, 31, 338-353, doi:10.1177/00139169921972137. Bamberg, S. (2002). Effects of implementation intentions on the actual performance of new environmentally friendly behaviours – results of two field experiments. Journal of Environmental Psychology, 22, 399-411, doi:10.1006/jevp.2002.0278. Bamberg, S., & Möser, G. (2007). Twenty years after Hines, Hungerford, and Tomera: A new meta-analysis of psycho-social determinants of pro-environmental behaviour. Journal of Environmental Psychology, 27, 14-25, doi:10.1016/j.jenvp.2006.12.002. Bandura, A., & Adams, N. E. (1977). Analysis of self-efficacy theory of behavioral change. Cognitive Therapy and Research, 1, 287-310, doi:10.1007/BF01663995. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman. Bandura, A. (2006). Guide for constructing self-efficacy scales. In F. Pajares and T. Urdan (Eds.), Adolescence and education: Vol. 5. Self-efficacy beliefs of adolescents (pp. 307-337). Greenwich, CT: Information Age Publishing. Belz, F.M., & Bilharz, M. (2005). Nachhaltiger Konsum: Zentrale Herausforderungen für moderne Verbraucherpolitik (Consumer Science Diskussionsbeitrag 1). Retrieved PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 42 (22/12/11) from University of St.Gallen website: http://www.alexandria.unisg.ch/EXPORT/DL/41721.pdf Berg, L., Kjaernes, U., Ganskau, E., Minina, V., Voltchkova, L., Halkier, B., & Holm, L. (2005). Trust in food safety in Russia, Denmark and Norway. European Societies, 7, 103-129, doi:10.1080/1461669042000327045. Berger, I. E., & Corbin, R. M. (1992). Perceived consumer effectiveness and faith in others as moderators of environmentally responsible behaviors. Journal of Public Policy and Marketing, 11, 79-89. Biel, A., Dahlstrand, U., & Grankvist, G. (2005). Habitual and value-guided purchase behavior. Ambio, 34, 360-365. Bye, A. S., Aarstad, P. A., Løvberget, A. I., Berge, G., & Hoem, B. (2010). Jordbruk og miljø: Tilstand og utvikling 2010 (Report 48/2010). Retrieved from Statistics Norway website: http://www.ssb.no/emner/01/04/rapp_jordbruk/rapp_201048/rapp_201048.pdf Cleveland, M., Kalamas, M., & Laroche, M. (2005). Shades of green: Linking environmental locus of control and pro-environmental behaviors. The Journal of Consumer Marketing, 22, 198-212, doi:10.1108/07363760510605317. Cleveland, M., Kalamas, M., & Laroche, M. (2012). “It’s not easy being green”: Exploring green creeds, green deeds, and internal environmental locus of control. Psychology & Marketing, 29, 293-305, doi:10.1002/mar.20522. Collins, C. M., Steg, L., & Koning, M. A. S. (2007). Customers’ values, beliefs on sustainable corporate performance, and buying behavior. Psychology & Marketing, 24, 555-577, doi:10.1002/mar.20173. De Young, R., Duncan, A., Frank, J., Gill, N., Rothman, S., Shenot, J., Shotkin, A., & Zweizig, M. (1993). Promoting source reduction behavior: The role of motivational PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 43 information. Environment and Behavior, 25, 70-85, doi:10.1177/0013916593251003. Doran, C. J. (2009). The role of personal values in fair trade consumption. Journal of Business Ethics, 84, 549-563, doi:10.1007/s10551-008-9724-1. Enders, C. K. (2010). Applied missing data analysis. New York: Guilford Press. Ellen, P. S., Wiener, J. L. & Cobb-Walgren, C. (1991). The role of perceived consumer effectiveness in motivating environmentally conscious behaviors. Journal of Public Policy and Marketing, 10, 102-117. Fishbein, M., & Capella, J. N. (2006). The role of theory in developing effective health communications. Journal of Communication, 56, 1-17, doi:10.1111/j.14602466.2006.00280.x. FIVH. (2011). Norge på økomatbunnen (Arbeidsnotat 2/2011). Retrieved (22/12/11) from http://www.framtiden.no/view-document/449-norge-pa-okomatbunnen.html Frick, J., Kaiser, F. G., & Wilson, M. (2004). Environmental knowledge and conservation behavior: Exploring prevalence and structure in a representative sample. Personality and Individual Differences, 37, 1597-1613, doi:10.1016/j.paid.2004.02.015. Geller, E. S., Farris, J. C., & Post, D. S. (1973). Prompting a consumer behavior for pollution control. Journal of Applied Behavior Analysis, 6, 367-376, doi:10.1901/jaba.1973.6367. Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint: Using social norms to motivate environmental conservation in hotels. Journal of Consumer Research, 35, 472-482, doi:10.1086/586910. Graham, J. W. (2009). Missing data analysis: Making it work in the real world. Annual Review of Psychology, 60, 549-576, doi:10.1146/annurev.psych.58.110405.085530. Hanss, D., & Böhm, G. (2010). Can I make a difference? The role of general and domainspecific self-efficacy in sustainable consumption decisions. Umweltpsychologie, 14, PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 44 46-74. Hanss, D., & Böhm, G. (2012a). A social dilemma perspective on sustainable consumption: Social values, time perspective, and self-efficacy beliefs predict purchases of environmentally friendly and fair trade groceries. Manuscript submitted for publication. Hanss, D., & Böhm, G. (2012b). Sustainability seen from the perspective of consumers. International Journal of Consumer Studies, 36, 678-687, doi: 10.1111/j.14706431.2011.01045.x. Jungbluth, N., Tietje, O., & Scholz, R. W. (2000). Food purchases: Impacts from the consumers‘ point of view investigated with a modular LCA. The International Journal of Life Cycle Assessment, 5, 134-142, doi:10.1007/BF02978609. Kellstedt, P. M., Zahran, S., & Vedlitz, A. (2008). Personal efficacy, the information environment, and attitudes toward global warming and climate change in the United States. Risk Analysis, 28, 113-126, doi:10.1111/j.1539-6924.2008.01010.x. Lehman, P. K., & Geller, E. S. (2004). Behavior analysis and environmental protection: Accomplishments and potential for more. Behavior and Social Issues, 13, 13-32. Marrakech Process Secretariat: UNDESA and UNEP. (2010). Proposed input to CSD 18 and 19 on a 10 year framework of programmes on sustainable consumption and production (10YFP ON SCP). Retrieved (22/12/11) from: http://esa.un.org/marrakechprocess/pdf/Draft3_10yfp_RevisedUNEP-UNDESA.pdf Ministry of Finance. (2011). Nasjonalbudsjettet 2011 (Kapitel 7: bærekraftig utvikling). Retrieved from http://www.regjeringen.no/nb/dep/fin/dok/regpubl/stmeld/20102011/meld-st-1-20102011/7.html?id=616542 Ölander F., & Thøgersen, J. (1995). Understanding of consumer behaviour as a prerequisite for environmental protection. Journal of Consumer Policy, 18, 345-385, PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 45 doi:10.1007/BF01024160 Roberts, J. A. (1996). Green consumers in the 1990s: Profile and implications for advertising. Journal of Business Research, 36, 217-231, doi:10.1016/0148-2963(95)00150-6 Rotter, J. B. (1966). Generalized expectancies for internal versus external control of reinforcement. Psychological Monographs, 80(1,Whole No. 609). Rubin, D. B. (1996). Multiple imputation after 18+ years. Journal of the American Statistical Association, 91, 473-489. Schwepker, C. H., & Cornwell, T. B. (1991). An examination of ecologically concerned consumers and their intention to purchase ecologically packaged products. Journal of Public Policy & Marketing, 10, 77-101. Smith-Sebasto, N. J. (1992). Design, development, and validation of an instrument to assess the relationship between locus-of-control of reinforcement and environmentally responsible behavior in university undergraduate students (Doctoral dissertation). Retrieved from http://etd.ohiolink.edu/sendpdf.cgi/SmithSebasto%20Nicholas%20J.pdf?osu1219953285 Staats, H., Van Leeuwen, E., & Wit, A. (2000). A longitudinal study of informational interventions to save energy in an office building. Journal of Applied Behavior Analysis, 33, 101-104. Steg, L., & Vlek, C. (2009). Encouraging pro-environmental behaviour: An integrative review and research agenda. Journal of Environmental Psychology, 29, 309-317, doi:10.1016/j.jenvp.2008.10.004. Stewart, T. L., Latu, I. M., Branscombe, N. R., & Denney, H. T. (2010). Yes we can! : Prejudice reduction through seeing (inequality) and believing (in social change). Psychological Science, 21, 1557-1562, doi:10.1177/0956797610385354. Straughan, R. D., & Roberts, J. A. (1999). Environmental segmentation alternatives: A look at PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 46 green consumer behavior in the new millennium. Journal of Consumer Marketing, 16, 558-575, doi:10.1108/07363769910297506. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics. Boston: Allyn and Bacon. Tanner, C., & Wölfing Kast, S. (2003). Promoting sustainable consumption: Determinants of green purchases by Swiss consumers. Psychology & Marketing, 20, 883-902, doi:10.1002lmar.10101. Terragni, L., & Kjaernes, U. (2005). Ethical consumption in Norway: Why is it so low? Retrieved (22/12/11) from the National Institute for Consumer Research (SIFO) website: http://www.sifo.no/page/Forskning//10060/59109.html Tobler, C., Visschers, V. H. M., & Siegrist, M. (2011). Organic tomatoes versus canned beans: How do consumers assess the environmental friendliness of vegetables? Environment and Behavior, 43, 591-611, doi:10.1177/0013916510372865. Torjusen, H., Lieblein, G., Wandel, M., & Francis, C.A. (2001). Food system orientation and quality perception among consumers and producers of organic food in Hedmark County, Norway. Food Quality and Preference, 12, 207–216, doi:10.1016/S09503293(00)00047-1. UNWCED. (1987). Our common future: Report of the World Commission on Environment and Development. Retrieved (22/12/11) from UN Documents website: http://www.undocuments.net/ocf-02.htm#I Urien, B., & Kilbourne, W. (2011). Generativity and self-enhancement values in eco-friendly behavioral intentions and environmentally responsible consumption behavior. Psychology & Marketing, 28, 69-90, doi:10.1002/mar.20381. Van Ginkel, J. R. (2010). SPSS syntax for applying rules for combining multivariate estimates in multiple imputation. Retrieved (22/12/11) from Leiden University PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 47 webpage: http://www.socialsciences.leiden.edu/educationandchildstudies/childandfamilystudies/ organisation/staffcfs/van-ginkel.html Vermeir, I., & Verbeke, W. (2006). Sustainable food consumption: Exploring the consumer “attitude-behavioral intention” gap. Journal of Agricultural and Environmental Ethics, 19, 169-194, doi: 10.1007/s10806-005-5485-3. Von Borgstede, C., Dahlstrand, U., & Biel, A. (1999). From ought to is: Moral norms in large-scale social dilemmas (Göteborg Psychological Reports, 29, No. 5). Retrieved (28/08/12) from University of Gothenburg website: http://www.science.gu.se/digitalAssets/1286/1286092_gpr99_nr5.pdf PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 48 Footnotes 1 When we say groceries, we refer to products that are available in grocery stores, such as foods or cosmetics (e.g., soap, body lotion, and toothpaste). 2 Comparison data from Statistics Norway, available online at: http://www.ssb.no/ 3 The original questionnaire (Hanss & Böhm, 2010) measured purchase habits rather than purchase intentions. To measure purchase habits, the original questionnaire asked how one usually decides. In this study, we asked instead how one will decide in the future, in order to tap intentions. In addition, we improved the wording of some items. For example, one of the purchase habit items used by Hanss and Böhm (2010) was: “When shopping produce, I decide for ecological products.” In the present study, we changed this to “When I buy fruits and vegetables and have the choice between ecological and conventional products, I will buy ecological products.” 4 There were no differences between the intervention and control conditions with regard to the types of product categories that were presented during the course of the study. The sequence of product presentation was randomized and thus differed from participant to participant within and between the two participant groups. 5 The basic idea of multiple imputation is to produce several complete data sets by generating plausible values for missing values (Rubin, 1996). These plausible values are estimated from the existing data, for example, through linear or logistic regression. Once the complete data sets have been generated, analyses are conducted for each of the data sets separately. Next, the results are pooled (i.e., combined into a single set of results), commonly according to the rules provided by Rubin (e.g., 1996). Multiple imputation has several advantages: For example, it does not require the data to be missing completely at random (MCAR) -- according to Tabachnick and Fidell (2007), multiple imputation may even be used if data are missing not at random (MNAR) -- and it PROMOTING PURCHASES OF SUSTAINABLE GROCERIES 49 retains sampling variability, which may not be the case when missing data are imputed with some of the “older methods,” such as mean substitution (cf. Tabachnick & Fidell, 2007). 6 Auxiliary variables are variables that are correlated with the main variables of interest in the imputation model (cf. Graham, 2009; here: variables that are correlated with the product purchase variables, the SDSE, and the PI index variables). According to Graham (2009), the inclusion of auxiliary variables has two major advantages: “It can reduce estimation bias due to MNAR missingness, and it can partially restore lost power due to missingness” (p. 560). The variables were entered into the imputation model in the following order: (1) age, (2) gender, (3) attitudes toward ecological and fair trade groceries, (4) attitudes toward regional and healthy groceries, (5) norms, (6) moral judgments, (7) general environmental attitudes, (8) SDSE (others, domestic, global, sustainable products, environment-climate), (9) product choice, (10) PI (resource-saving-ethical, ecological foods, domestic-seasonal). For each variable, one needs to define its role in the imputation model. In SPSS 19, three options are available: Imputation only, that is, missing values on the variable are imputed by help of the available data. Prediction only, that is, original data of a variable serve as predictors for the imputation of missing values on other variables. Imputation and prediction, that is, missing values on the variable are imputed and then original and imputed data serve as predictors for the imputation of missing values on other variables. In the present study, age and gender served solely as predictors for variables that were entered subsequently into the model (i.e., missing values for age and gender were not imputed). For all other variables, missing values were imputed with the help of the preceding variables in the model; then, original and imputed data were used for imputing data on subsequent variables. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES Table 1 Overview of the procedure, measurements, materials, and number of participants per study parta Study part Part 1 (Pre-test 1) Week 1 Part 2 (Pre-test 2) Week 2 Part 3 Week 3 Part 4 Week 4 Part 5 Week 5 Part 6 Week 6 Part 7 (Post-test 1) Week 7 Part 8 (Post-test 2) Week 8 Part 9 (Follow-up) Six months after Part 8 Intervention (initially n = 73) Control (initially n = 72) Product purchases Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 71 Product purchases Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 72 Product purchases Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 67 Product purchases Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 65 Intervention step 1: awareness of environmental and socio-economic problems Matching colors to car types Intervention step 2: human actions as main causes for problems Rating the overall appeal of colors and car types Intervention step 3: strengthening self-efficacy with regard to directly contributing to a sustainable development Intervention step 4: strengthening self-efficacy with regard to indirectly contributing to a sustainable development Product purchases Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 54 Rating colors on semantic differential scales Rating car types on semantic differential scales Product purchases Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 57 Product purchases Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 50 Product purchases Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 54 Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 27 Purchasing intentions (PI) Self-efficacy beliefs (SDSE) Completed by n = 34 Note. aNumbers of participants who completed the mandatory questionnaires (product purchases, PI, and SDSE) in study Parts 1, 2, 7, 8, and 9. In study Parts 3 to 6, some of the questionnaires were optional; no participant numbers are reported for these study parts. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES Table 2 Effects of group, time, and group x time on purchases of sustainable products, PI, and SDSE Independent variables Group Dependent variable B Time Group x Time B t p B t p -.305 -1.57 .116 0.16 3.69 .000 0.70 1.15 .251 -.470 -.987 .328 .123 1.65 .101 .005 .048 .962 -1.51 -2.99 .005 .135 1.81 .071 .062 .490 .628 -.771 -1.88 .061 .056 .826 .409 .204 2.07 .039 .160 .399 .012 .370 .712 -.006 -.118 .907 -.034 -.177 .860 .067 2.01 .047 .008 .171 .864 -.057 -.271 .787 .060 1.71 .089 -.022 -.450 .653 -.042 -.237 .812 .024 .704 .484 .003 .063 .950 .016 -.009 -.308 .758 -.002 -.035 .972 t p Purchases of sustainable products PIa Resource-SavingEthical PIa Ecological Foods PIa Domestic-Seasonal SDSEb Others SDSEb Domestic SDSEb Global .845 SDSEb Sustainable Products SDSEb EnvironmentClimate .095 .924 Note. aPI = purchasing intentions; bSDSE = sustainable development self-efficacy. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES Table 3 Associations (Pearson correlations) between purchasing intentions and number of sustainable purchases per study part Study Part PIa Resource- PIa Ecological PIa Domestic- Saving-Ethical Foods Seasonal r p r p r p Part 1 No Sust.b .350 .000 .418 .000 .201 .032 Part 2 No Sust.b .352 .014 .326 .018 .179 .204 Part 7 No Sust.b .285 .003 .314 .005 .302 .000 Part 8 No Sust.b .188 .045 .243 .021 .249 .006 Note. aPI = purchasing intentions; bnumber of sustainable products purchased. The displayed correlations are pooled results (i.e., parameter estimates combined across the five imputations) for intervention and control group combined. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES Figure 1. Example product presentation: conventional (left) and sustainable (right) tea. Product description in Norwegian: “økologisk” means ecological; “teposer” means teabag; “kroner” refers to the local currency; “velg et produkt” means choose a product. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES Purchases of sustainable products Figure 2. Mean number of purchased sustainable products per study part: means for the intervention group and control group across time. In each study part, the participants had to purchase two products. The maximum number of sustainable products to be purchased per study part was two; the minimum number was zero. PROMOTING PURCHASES OF SUSTAINABLE GROCERIES PI Resource-Saving-Ethical PI Ecological Foods PI Domestic-Seasonal PROMOTING PURCHASES OF SUSTAINABLE GROCERIES Figure 3. Mean scores of purchasing-intention indices per study part: means for the intervention group and control group across time. Purchasing intentions were measured on 11-point rating scales that ranged from 0 (no, definitely not) to 10 (yes, definitely). PROMOTING PURCHASES OF SUSTAINABLE GROCERIES SDSE Others SDSE Domestic SDSE Global PROMOTING PURCHASES OF SUSTAINABLE GROCERIES SDSE Sustainable Products SDSE Environment-Climate Figure 4. Mean scores of sustainable development self-efficacy (SDSE) indices per study part: means for the intervention group and control group across time. SDSE was measured on 5-point rating scales that ranged from 1 (strongly disagree) to 5 (strongly agree).