Running head: GREEN PACKAGING CUES AND PACKAGE PROCESSING 1
University of Amsterdam Graduate School of Communication Research Master Communication Science
Master’s Thesis
Green Packaging Cues and their Influence on Package Processing 29 June 2018
Carolin Ischen 11354119
carolin.ischen@student.uva.nl c.ischen93@gmail.com
Abstract
Even though environmentally friendly or green products are valued and perceived as positive, they often do not attain consumers’ relatively limited consideration set for a purchase
decision. Based on the assumption that consumers’ attention and comprehension or
elaboration processes need to be activated first for green products to enter this consideration set, this thesis investigated whether green packaging cues, i.e. green package material and organic labelling, positively influence attention and comprehension processes and
furthermore, whether consumer-characteristics moderate the effects. Three experimental studies were conducted, both in a laboratory setting (n = 42) and partly replicated with a larger online-sample (n = 109). Mouse tracking, an implicit measurement method, was used to gauge lower-level processing (attention) and higher-level processing (comprehension and elaboration), heat mapping was used for lower-level processing, and these were
complemented by a questionnaire as an explicit assessment for higher-level processing. Results showed that green packaging cues influenced lower-level attention processing, meaning that packages carrying green cues positively influenced respondents’ perceived salience. A mediating effect of lower-level on the higher-level processing was not confirmed, neither were the moderating effects of the consumer-characteristics: green consumerism and environmental knowledge. However, a direct effect of green cues on higher-level processes (and several moderating effects of green consumerism and environmental knowledge on the direct effect) led to the conclusion that green cues are able to influence lower- and higher-level processing independently. This implicates that it is possible for corporations as well as policymakers to promote the environmental friendliness of a product and its packaging by implementing green packaging cues.
Keywords: green product packaging, package processing, attention and comprehension, mouse tracking, heat map
GREEN PACKAGING CUES AND PACKAGE PROCESSING 3
Packaging Cues and Influence on Processing
Although consumers have become increasingly aware of their environmental impact, our current excessive consumption patterns contribute to a variety of ecological challenges, including climate change, water and resource scarcity, and consumer health issues (Cho, 2015; Stolz, Molina, Ramírez, & Mohr, 2013). Packaging waste in particular represents one of the major environmental concerns society faces today, being directly observable through litter dumps and landfills, and involving nearly all consumers (Biod, Probert, & Jones, 1994; Van Birgelen, Semeijn, & Keicher, 2009). Not only can the purchasing of green packages help to decrease these negative outcomes, but eco-labelling can also be seen as a way “to encourage consumers to adopt more sustainable consumption patterns through the purchase of products that are more resource and energy efficient” (Horne, 2009, p. 175).
Green products can often be found in today’s supermarkets. Research concludes that consumers value environmentally friendly or green products (Rokka & Uusitalo, 2008). However, feeling positive towards green products often does not directly lead to more sustainable purchasing behaviors (e.g. Alwitt & Pitts, 1996; Bech-Larsen, 1996; Moisander, 2007; Thøgersen, 1999, 2004; Thøgersen & �lander, 2003; Uusitalo, 1989, 1990). Most studies in the field of green packaging focus on the relative importance of green attributes in comparison to other attributes for making a purchasing decision (Koutsimanis, Getter, Behe, Harte, & Almenar, 2012; Rokka & Uusitalo, 2008; Van Birgelen et al., 2009). However, even if the evaluation of green packaging is generally positive, this does not directly translate into considering these products for purchase.
Attention and comprehension processes must be acknowledged as a prerequisite for attitude formations and purchase intentions, and are therefore important to study (Edell & Staelin, 1983). The purchase of everyday products is often characterized by habits in which consumers choose between products in an evoked set, defined as a set of alternatives available at the place of purchase carrying relevance for the consumer (Bech-Larsen, 1996; Howard &
Sheth, 1988). Previous research finds that products within a set of alternatives are more likely to be chosen if they receive more attention (Krajbich, Armel, & Rangel, 2010; Van der Laan, Hooge, De Ridder, Viergever, & Smeets, 2015; Van der Laan, Papies, Hooge & Smeets, 2017). Different products are constantly competing for our attention, which is a scarce resource and highly selective, thus making it crucial to determine which packaging characteristics attract attention (Smit, Boerman & van Meurs, 2015).
In this regard, it is worthwhile to consider product packaging not only as a storage of everyday products, but as a communication method of greenness (Bech-Larsen, 1996). As food is usually packaged, it is possible to infer product characteristics from the package (Van der Laan, De Ridder, Viergever, & Smeets, 2012). Thus, to get into the consideration set and be able to influence consumers’ attitudes and behavior, green products have to firstly, attract attention and secondly, be understood as green. The overall research question is therefore: Do green packaging cues positively influence attention and comprehension processes?
In this study, a four-level model of audience involvement in advertising processing, introduced by Greenwald and Leavitt (1984), is used. It will be argued that green cues positively influence lower-level processing pre-attention and focal attention and higher-level processing comprehension and elaboration. Furthermore, the consumer characteristics of green consumerism and environmental knowledge are accounted for, based on the distinction between stimulus-driven processes triggered by the salience of information in the visual field and goal-directed processes based on consumer-characteristics (Orquin & Mueller Loose, 2013).
This thesis will contribute to the literature threefold. Firstly, attention and
comprehension as dependent variables enriches our understanding of the effects of packaging cues in general and green cues in particular. Secondly, novel methodological approaches will be used by comparing the unique measurements of mouse tracking and heat mapping.
GREEN PACKAGING CUES AND PACKAGE PROCESSING 5 unsustainable consumer behavior if it is proven that product packaging is able to draw
attention and communicate environmental friendliness.
Theoretical Background Green Packaging Cues
In a shopping situation, consumers make use of intrinsic (e.g. the product itself) or extrinsic (e.g. price, brand) cues to assess a product (Olson, 1972). Particularly in situations in which intrinsic cues, such as the product’s taste, are not available, extrinsic cues play an important role (Behe et al., 2014). Two of these cues are a green product packaging material (i.e.
looking recyclable or biodegradable) and product labels (e.g. the EU organic label) which can be used to promote the greenness of a product (Thøgersen, Haugaard, & Olesen, 2010).
In regards to the influence of green cues on attention, previous research is relatively scarce, but related research in the field of organic products and packaging, in general, gives insights in how these cues are perceived by consumers (for an overview, see Sleenbecker & Hamm, 2013). Several focus groups were conducted, which found that consumers would like to distinguish between organic and conventional packaging in regards to milk packaging (Hill & Lynchhaun, 2002) and that some consumers prefer glass as a package material, whereas others prefer plastic (Stolz, Bodini, Stolze, Hamm, & Richter, 2009).
While previous studies have shown that several product attributes (for example: price, convenience of use, or container size) are important for consumers when making a product decision, (Koutsimanis et al., 2012; Rokka & Uusitalo, 2008; Van Birgelen et al., 2009), Rokka and Uusitalo (2008) studied the relative importance of green packaging specifically when compared to other relevant product attributes. A conjoint analysis concluded that environmentally friendly packaging was the most important factor in the choice among functional drink products.1 Other studies provide evidence that consumers are willing to trade
1 Conjoint analysis is a survey-based method to assess the relative value of certain product features (Rokka and
off several product attributes in favor of environmentally friendly packaging (Van Birgelen et al., 2009) and that containers made from bio-based materials were highly appealing to
consumers (Koutsimanis et al., 2012).
Furthermore, organic labels as green cues can be used to promote sustainability and to reduce information search costs (Thøgersen et al., 2010). Although there are some
contradictory findings (e.g. D’Souza, Taghian, Lamb, & Peretiatko, 2007), most studies found a positive effect of green labels on product preference, i.e. perceived quality, the belief that the product offers good taste or the overall satisfaction with the product (e.g. D’Souza, Taghian & Lamb, 2006; Larceneux, Benoit-Moreau, & Renaudin, 2012; Thøgersen et al., 2010).
Concluding, green advertising in general and green cues in particular are important to consumers, but the question of which of these cues attracts attention such that they will be in the consideration set remains unanswered. To actually trade off other attributes in favor of green packaging, consumers must attend to the package first, and comprehend its message, to include the aspect of a package and its content being environmentally friendly, in their evaluation process. It is known from related research on packaging in general that attributes such as images, colors, or graphic form are proven to influence consumer choice (Ampuero & Vila, 2006; Underwood, 2003). Eye-tracking research conducted with different jam jars shows that the most looked-at areas are labels, then the information, and finally the border or outline of the product (Piqueras-Fiszman, Velasco, Salgado-Mentejo, & Spence, 2013). Further, research showed that consumers scan a product package from the largest to the smallest element and in alignment with reading patterns (Rebollar, Lidón & Puebla, 2015). Following the argument that certain cues indeed influence attention, this research specifically focuses on two green cues, green packaging materials and organic labels, to examine whether the attention and comprehension processes to those cues resemble the previous findings.
GREEN PACKAGING CUES AND PACKAGE PROCESSING 7
Levels of Processing
Attention and comprehension processes are considered to be “varying along a continuum from the automatic, lower effort processes of preconscious attention to the more controlled, higher effort comprehension processes involved in elaboration” (Celsi & Olson, 1988, p. 213-214). Greenwald and Leavitt (1984) introduced a four-level model of audience involvement in advertising processing, which can be applied to the processing of product packaging and its cues. At the pre-attention level, the consumer scans and filters the information on the package, while on the second level only a limited amount of package stimuli receive focal attention, influencing the selectivity in perception, which then plays an active role in constructing a decision (Orquin & Mueller Loose, 2013). This leads to the third level, comprehension, where the package is analyzed more closely and meaning is assigned, also called proposition formation (Finn, 1988). This implies that a higher involvement leads to propositional traces left in the memory. On the fourth level, elaboration, the information is embedded in existing knowledge structures (Greenwald & Leavitt, 1984). Elaboration takes place when a cognitive response is given to the stimuli and personal connections and imagery are generated, which requires full cognitive capacity and involvement, as details are
integrated into existing knowledge (Finn, 1988).
Processing is often assumed to have a hierarchical structure such that it must first occur on one level in order to evoke processing on the next higher level, including more capacity devoted to the task (Finn, 1988; Smit, Neijens & Heath, 2013). In contrast, the divergent processing model assumes that once a package has received baseline attention, it depends on an individual’s goal if the comprehension or the elaboration stage takes place (Finn, 1988; Mitchell, 1983). According to these scholars, it is possible to go directly to the elaboration stage after paying focal attention; as such, the comprehension level will be excluded.
Although there are some inconsistencies in how the sequence of stages is constructed, considering that the transition from one stage to another is not clear-cut, the dominant view in most scholars is that attention must be drawn first for subsequent processing to follow (Finn, 1988; Mitchell, 1983). Thus, two processing levels are distinguished for this research: lower-level processing, including pre-attention and focal attention, and higher-lower-level processing, including comprehension and elaboration.
Stimulus-driven and Goal-directed Processing
A distinction needs to be made between two types of processes, characterized as goal-directed and stimulus-driven (e.g. Chandon, Hutchinson, Bradlow, & Young, 2009).
Stimulus-driven, or bottom-up processing, is triggered by the salience of information in the visual field. This rapid and automatic type of processing is dependent on the stimulus and its intrinsic properties (Smit et al., 2013). In other words, consumers are likely to focus on attributes that stand out, either because of their salience or because of their unexpectedness (Meißner, Musalem & Huber, 2016). Previous research in persuasive, and more specifically marketing communication, has focused mostly on print advertisements, arriving at diverse determining factors for focal attention, most notably size and color (Pieters & Wedel, 2004; Pieters, Wedel, & Batra, 2010; Smit, et al., 2015; Zhang, Wedel, & Pieters, 2009), which is consistent with previous findings on product packaging (Ampuero & Vila, 2006; Underwood, 2003). Considering stimulus-driven processing, green cues can be seen as stimuli relevant for increasing the perceived salience of the package, as the variety of green packages in a
supermarket might still be lower than the variety of regular packages.
In contrast, goal-directed, or top-down processing, is based on
consumer-characteristics instead of stimulus-consumer-characteristics. More specifically, it is a volitional, focal process depending on personal characteristics and motivation (Smit et al., 2013). The motivational mindset for this type of processing is created when the cues correspond with consumers’ decision goals and enduring values (Meißner et al., 2016). A goal can, for
GREEN PACKAGING CUES AND PACKAGE PROCESSING 9 instance, be learning about a specific product type, but also can be showing an interest in sustainable behavior or the reduction of packaging waste.
Goal-directed and stimulus-driven processes are intertwined in the sense that the stimulus-driven process is more likely to be activated when a task-relevant attribute is
presented instead of one that is purely more salient (Eysenck & Keane, 2013). In other words, if a goal is present, the responsiveness of stimulus-relevant cues is higher, resulting in more processing.
Integrating Levels and Types of Processing
Previous research suggests that packaging attributes as stimulus-driven cues are more important as attractors of attention for lower-level processing (i.e. pre-attention and focal-attention), while consumer-specific characteristics as a goal-directed process are more important as retainers of attention for higher-level processing (i.e. comprehension and elaboration; Smit et al., 2013). Applying these considerations to the two green cues being studied, the following hypotheses regarding lower-level processing, as depicted in the conceptual model in Figure 1, are proposed:
Hypothesis 1: A green product packaging material leads to more lower-level processing than a non-green product packaging material.
Hypothesis 2: A product packaging with an organic label leads to more lower-level processing than a product packaging without an organic label.
Based on the sequential model, attention is necessary for higher-level processing (comprehension and elaboration) to emerge. Consequently, the influence of green cues on higher-level processing is assumed to be mediated by lower-level processing.
Hypotheses 3: Green cues (i.e. green product packaging material and an organic label) positively influence lower-level processing, which then influences higher-level processing.
Green Consumerism. A closer look at higher-level processing identifies motivation
as the decisive factor in whether or not consumers are directed towards a goal (Eysenck & Keane 2013). A green consumer, i.e. an “individual that wants and knows how to satisfy his or her needs in the everyday life causing as little as possible impact on the environment” (Sharma & Joshi, 2017, p. 208) is expected to have a higher motivation to comprehend and elaborate on green packaging cues, as these correspond with the consumer’s goals and values to behave in an environmentally friendly way. Even if it is assumed that a green intention does not necessarily transform into an actual green purchase, it will still positively influence the motivation to process green cues (Ajzen, 1991; Sharma & Joshi, 2017). Hence, it is suggested that green packaging materials and organic labels are more relevant for green consumers and function as an intrinsic source of personal relevance, resulting in a higher motivation to process green cues (Celsi & Olson, 1988). As green cues are able to
communicate environmental friendliness, they especially attract consumers who prioritize environmental protection (Thøgersen, 2000).
Previous research has found that consumers are indeed influenced by eco-labels depending on their environmental interest, i.e. consumers with a strong interest are affected by labels in building a product preference (Grankvist, Dahlstrand & Biel, 2004).
Correspondingly, green consumerism is expected to moderate higher-level processing (comprehension and elaboration).
Environmental Knowledge. Taking into account that the evaluation of green labels is
subjective, especially in situations where consumers do not have a thorough knowledge of the label and its meaning (Pedersen & Neergaard, 2006), research concludes that consumers must know, identify, and understand a label to use it in a decision making process (Thøgersen, 2000). Previous research found that knowledge of green packaging materials and eco-labels is still scattered (Scott & Vigar-Ellis 2014; Thøgersen, 2000). However, research also showed that consumers who are both familiar with eco-labels in general and confident about their
GREEN PACKAGING CUES AND PACKAGE PROCESSING 11 knowledge in this domain, are more likely to notice and understand new eco-labels
(Thøgersen, 2000).
The differences among consumers regarding environmental knowledge is
hypothesized to have an effect on their motivation to process, such that consumers with a higher knowledge are more likely to comprehend and elaborate on green products. Previous research showed that subjective knowledge, i.e. the subjective interpretation of knowledge about green products, is associated with organic vegetable consumption, while objective knowledge, i.e. being able to retrieve accurate information about green products from long term memory, is indirectly associated through increasing subjective knowledge (Pieniak, Aertsens, & Verbeke, 2010). Furthermore, Celsi and Olson (1998) concluded that domain knowledge influences the meaning produced by comprehension processes, since knowledge is related to long-term memory which needs to be activated for higher levels of processing. Thus, subjective and objective knowledge are both hypothesized to be important for higher-level processing. This leads to the following hypothesis regarding green consumerism and environmental knowledge:
Hypothesis 4: The positive impact of lower-level processing on higher-level processing is moderated by (1) green consumerism, in the sense that the effect of lower- on higher-level processing is stronger for green consumers and (2) environmental knowledge, in the sense that the effect of lower- on higher-level processing is stronger for consumers with more environmental knowledge.
Figure 1. Conceptual Model
Implicit and Explicit Measurements of Attention
Especially lower-level attention processes are difficult to measure. As often proposed, it is difficult to assess a person’s attention explicitly with survey questions, as these might be biased in a way that participants are reminded of the exposure or explicitly asked to recollect information from the exposed material (Vandeberg, Murre, Voorveld, & Smit, 2015). To implicitly measure attention, a common method is eye tracking, in which the gaze duration, fixation time etc. are measured (e.g. Bialkova & van Trijp, 2011). However, the assessment by means of an eye-tracker is not only costly, but is not suitable for certain research questions (Freeman & Ambady, 2010). Therefore, two novel implicit measurements of lower-level processing were implemented, namely heat map analysis and mouse tracking.
Overview of Studies
Three studies were conducted to test the hypotheses and compare the different measurement methods. Firstly, it was tested in a pre-test whether the developed stimulus material was perceived as intended. Secondly, a laboratory experiment was conducted in which the mouse
GREEN PACKAGING CUES AND PACKAGE PROCESSING 13 tracker was used to measure lower- as well as higher-level processing. Furthermore, heat mapping was used to assess lower-level processing and an explicit questionnaire was used to measure higher-level processing. Finally, the heat mapping and questionnaire measurements were replicated with a larger, non-student sample.
Pre-test
Before assessing the different relations between the processing levels and the moderators, it is necessary to pre-test whether the green packages used as a stimulus materials (manipulated with the image manipulation program GIMP, version 2.10.2, see Appendix A) are actually perceived as green.
A 2 (green label vs. no green label) x 3 (cardboard vs. white vs. brown package) x 2 (milk vs. chocolate) partially overlapping within- and between-subjects factorial design was conducted. Both, milk and chocolate bars were chosen as product types. Greenness of the package was manipulated on three levels. In the green condition, the package had a cardboard design to give it the appeal of a biodegradable material. In the non-green condition, white and brown (same color as the cardboard condition) packages were created. The different colors were chosen to control for the color of the package in contrast to the (white) background. The environmental label was manipulated on two levels. Half of the packages included the green EU organic label positioned on the front-package bottom left, while the other half of the packages did not include a label. The EU organic label was chosen because of its cross-national use.
A student sample of 43 respondents participated voluntarily in this study. As participants were randomly exposed to six out of twelve stimuli, the sample included both paired and independent observations. Thus, partially overlapping samples t-tests were conducted to test mean differences (Derrick, Toher, & White, 2017). The results revealed differences in the perception of greenness between the different packages. The product in the cardboard package with a label as well as the cardboard package without a label were
perceived as being highest in greenness compared to most of the other packages. In other words, a label as well as the package material significantly influenced the perception of product greenness, in the sense that the product in the package with a label was perceived as greener than the product in the package without a label and the same is seen for the product in the cardboard package in comparison with the product in the white or brown package. Based on the perception of realism, it was decided to use the milk packages as stimuli for the subsequent studies, as the different packages are perceived as equally high in realism. Furthermore, the chocolate packages were included in the second study to function as a “filler” to make the exposure to the packages less repetitive. A full description of the methods and results of the pre-test can be found in Appendix B.
Study 1 Method
Participants. An experiment with a 2 (green label vs. no green label) x 3 (cardboard
vs. white vs. brown package) within-subjects design was conducted. A sample of 42 Communication Science Bachelor students participated in the laboratory experiment at the University of Amsterdam in return for research credits. Participants were between 18 and 27 years old (M = 20.61, SD = 1.62) and 71.4% were female (28.6% male). None of the
participants were colorblind. Six participants reported eating a vegetarian diet (14.3%) and three reported a vegan diet (7.1%). The sample is overall international with 73.8% of the participants had lived in the Netherlands for five years or less at the time of the study. Among the respondents, 83.3% were familiar with the EU organic label.
Procedure. The study was approved by the ethical committee of the University of
Amsterdam (2018-PC-9053). After giving informed consent, participants were introduced to the mouse tracking exercise (see Appendix C). Firstly, the mouse tracker was used to measure lower-level processing. Participants were exposed to a computer screen with a “START” button on the bottom, a “LEFT” button on the left side and a “RIGHT” button on the right
GREEN PACKAGING CUES AND PACKAGE PROCESSING 15 side. When they clicked the “START” button, the pictures of two products appeared on the screen. Participants were instructed to move the mouse towards the corner corresponding to the image they noticed first. The order of the images was randomized and every combination of packages was displayed twice to randomize the position of the packages; this resulted in a total number of 30 milk package combinations.
The second part of the mouse tracking exercise was designed to measure higher-level processing. Respondents were exposed to a similar start screen with a “START” button on the bottom, a “NOT GREEN” button on the left side and a “GREEN” button on the right side. When they clicked on the “START” button, one image of one of the packages appeared on the screen and the participants were instructed to move their mouse towards the corner corresponding with their perception of whether the product is green or not green. Again, the order of the packages was randomized; this resulted in six milk packages in total.
After finishing this exercise, respondents were asked to complete a heat map exercise. They were exposed to two of the milk package images on the computer screen and were instructed to click with the mouse on the first three areas they noticed. The order of the packages and the position on the screen (left, right) was randomized. Afterwards, respondents were asked to fill out a questionnaire including explicit measures of higher-level processing and were thanked for their participation (the entire questionnaire can be found in Appendix D).
Manipulation. The different stimuli were manipulated in the same manner as in the
pre-test. This resulted in six different packages, i.e. cardboard package with label, cardboard package without label, brown package with label, brown package without label, white package with label and white package without label.
Measurements.
Lower-level processing. Lower-level processing was operationalized as perceived
mapping. For the mouse tracker, which of the buttons the participants actually clicked on (left or right) was used to measure which of the two packages was more salient to them. For the heat mapping, it was measured by which of the two packages the participants clicked on first.
Furthermore, the mouse tracker includes several complexity measures, based on the trajectories from the “START” button to the chosen corner. The first measurement used is the maximum deviation (MD), defined as “the largest perpendicular deviation between the actual trajectory and its idealized trajectory out of all time-steps“ (Freeman, 2018, p. 7). A higher MD thus indicates a trajectory more deviated toward the unselected alternative which means that respondents are more uncertain about their answer. The second measurement used is the calculated number of x- and y-flips, indicating the reversals of direction along the x-axis and y-axis, respectively. “This captures the fluctuations in the hand’s vacillation between response alternatives along the x-axis and y-axis” (Freeman, 2018, p. 8) and can be seen as a measurement of how complex the trajectories are, implying respondents’ uncertainty.2
Higher-level processing. Higher-level processing, comprehension and elaboration,
was measured in the second part of the mouse tracking exercise. Higher-level processing was operationalized as perceived greenness, because the perception of a package being green can be seen as the comprehension of, and elaboration on, its greenness. Participants indicated which of the packages they perceived as green or not green. By means of the mouse tracker, it was measured on which of the buttons (green, not green) the participants clicked, and MD and x- and y-flips were recorded. Perception of greenness as a measurement of higher-level
processing was further assessed in the questionnaire for each milk package. Participants
2 To be able to use these measures, the data was prepared using the Mouse Tracking Analyzer Program
(Freeman, 2018). All recorded trajectories were rescaled into a standard mouse tracker coordinate space with the “START” button on the bottom center with the coordinates “0, 0”, the top-left corner of the screen corresponds to “-1, 1.5” and the bottom-right corner corresponds to “1, 0, representing a 2 × 1.5 rectangle, which retains the aspect ratio of most computer screens (Freeman, 2018, p. 2). Seeing as the trajectories vary in duration, each trajectory was time-normalized into 101-time steps using linear interpolation.
GREEN PACKAGING CUES AND PACKAGE PROCESSING 17 evaluated the greenness on a 7-point Likert scale ranging from 1 (not green at all) to 7 (very green).
Moderators.
Green consumerism was measured with a scale based on Matthes and Wonneberger (2014), who included three underlying sub-concepts of green consumerism: environmental concern (e.g. “I am concerned about the environment”; Schuhwerk & Lefkoff-Hagius, 1995), attitudes toward green products (e.g. “I like green products”; Chang, 2011) and green
purchase behavior (e.g. “I make a special effort to buy products in biodegradable packages”, Kim & Choi, 2005; Shrum, McCarty, & Lowrey, 1995). The use of all items replicates Matthes and Wonneberger’s (2014) scale, resulting in a nine-item scale. 3
Subjective knowledge was adapted from Ellen (1994), and Pieniak et al. (2010), e.g. “I am very confident that I buy products in packages which are environmentally safe”. This resulted in a five-item scale.4
Objective knowledge was measured with an item battery tailored to the EU green label using true and false statements (e.g. “The label indicates that the production follows
regulations for the organic farming sector established by the European Union”, true), after being asked whether they were familiar with the label.
3 As a factors analysis including the initial eleven items yield a two factor solution with a correlation of .36., the
items “green products are good for the environment” and “I feel proud when I buy green products” were excluded. The new item battery is internally consistent with a Chronbach’s alpha of .85 (M = 5.08, SD = 0.88). The nine questions related to green consumerism were factor analyzed using principal component analysis with Varimax rotation. The analysis yielded two factors with eigenvalues greater than one explaining a total of 61.75% of the variance and with a correlation of .64. Bartlett’s test of sphericity, which tests the overall significance of all the correlations within the correlation matrix, was significant (χ2 (36)= 159.86, p < .001).
Thus, it was appropriate to conduct a factor analysis. A Kaiser-Meyer-Olkin measure of sampling adequacy indicated that it was acceptable to proceed with the analysis, since the strength of the relationships among variables was high (KMO = .78).
4 As both reversed items “Trying to figure out the best packaging in terms of the effects on the environment is
very confusing (REVERSED)” and “Every time I turn around, the ‘experts’ are changing their minds about what are the ‘best’ products or packages to buy (REVERSED),” load on a second factor, the decision was made to exclude them. The new item battery with five items is internally consistent with a Chronbach’s alpha of .80 (M = 3.36, SD = 0.86). The five questions related to subjective knowledge were factor analyzed using principal component analysis with Varimax rotation. The analysis yielded one factor with eigenvalues greater than one explaining a total of 57.02% of the variance. Bartlett’s test of sphericity was significant (χ2 (10)= 77.09, p <
.001). Thus, it was appropriate to conduct a factor analysis. A Kaiser-Meyer-Olkin measure of sampling adequacy indicated that it was acceptable to proceed with the analysis, since the strength of the relationships among variables was high (KMO = .79).
Results
Hypothesis 1: Packaging material and lower-level processing.
Mouse tracking: Packaging material. To answer the first hypothesis of whether a
green product package material leads to more lower-level processing than a non-green product package material, only combinations including a green package material (cardboard look) vs. a non-green package material (brown or white look) were considered. The binary variables (0 = non-green package more salient, 1 = green package more salient) were summed up, to create an index score ranging from 0 to 8, where 8 indicates that participants
consistently showed that the green package was most salient. Since the positioning of the packages on the screen (left, right) was not randomized, but participants were exposed to both combinations, the scale ranges from 0 to 85. A one-sample t-test was conducted to compare the sample mean with the estimated mean (in other words, if both would be similarly attention grabbing, the sample mean would equal 4.00; deviations from the mean thus indicate that a package is more or less salient). The one-sample t-test revealed that packages with a
cardboard look (M = 5.45, SD = 2.43) were perceived as significantly more attention grabbing than the estimated mean of 4.00; t(41) = 3.87, p < .001.
Heat map: Packaging material. The first hypothesis was also tested with the heat
mapping data, in which the same package combinations were used. The binary variables (0 = non-green package more salient, 1 = green package more salient) were summed up, to create an index score ranging from 0 to 4, where 4 indicated that participants consistently showed that the green package was most salient. As the position was randomized and respondents saw either the cardboard package on the left or on the right side, the scale ranges from 0-46. A
5 I.e. white with label vs. cardboard with label; cardboard with label vs. white with label; brown with label vs.
cardboard with label; cardboard with label vs. brown with label; white without label vs. cardboard without label; cardboard without label vs. white without label; brown without label vs. cardboard without label, cardboard without label vs. brown without label
6 I.e. white with label vs. cardboard with label; brown with label vs. cardboard with label; white without label vs.
GREEN PACKAGING CUES AND PACKAGE PROCESSING 19 one-sample t-test was conducted to compare the sample mean with the estimated mean equals 2.00. The one-sample t-test revealed that the package with the cardboard look (M = 2.76, SD = 1.23) was perceived as significantly more salient than the estimated mean of 2.00; t(41) = 4.03, p < .001. Thus, in line with the conclusions based on the mouse tracking data,
hypothesis 1 is supported, a green product package material leads to more lower-level attention than a non-green product package material.
Hypothesis 2: Labels and lower level processing.
Mouse tracking: Labels. To test the second hypothesis of whether a product package
with an organic label leads to more lower-level processing than a product package without an organic label, only combinations where one package featured a label and one package did not feature a label were considered. The binary variables (0 = package without a label more salient, 1 = package with a label more salient) were summed up resulting in a scale ranging from 0-67. A one-sample t-test was conducted to compare the sample mean with the estimated mean equals 3.00, revealing that the packages with a label (M = 4.38, SD = 1.58) were
perceived as significantly more salient than the estimated mean of 3.00, t(41) = 5.68, p < .001.
Heat map: Labels. For the heat mapping data, the binary variables (0 = package
without a label more salient, 1 = package with a label more salient) were summed up resulting in a scale ranging from 0-38. A one-sample t-test was conducted to compare the sample mean with the estimated mean equals 1.50, revealing that a package with a label (M = 2.48, SD = 0.67) was perceived as significantly more salient than the estimated mean of 1.50, t(41) = 9.42, p < .001. Hypothesis 2 is supported with both the mouse tracking and the heat mapping
7 I.e. white with label vs. white without label; white without label vs. white with label; cardboard with label vs.
cardboard without label; cardboard without label vs. cardboard with label; brown with label vs. brown without label; brown without label vs. brown with label
8 I.e. white with label vs. white without label; cardboard with label vs. cardboard without label; brown with label
data, a product package with an organic label leads to more lower-level attention than a product packaging without an organic label.9
Hypothesis 3: Mediation.
Mouse tracking: Lower- and higher-level processing. To test the third hypothesis of
whether lower-level processing leads to higher-level processing, new variables were computed with the mouse tracking data. Lower-level processing was assessed by showing respondents a combination of two packages, while higher-level processing was assessed by showing them the packages individually. Thus, the dependent variable perceived greenness of the package needed to be computed per package for the package material conditions. This led to scale from 0 to 2, with 2 meaning that both packages (with and without label) were
perceived as green10. Additionally, a per label condition (label, no label) measurement with a scale from 0 to 3 was implemented, with 3 meaning that all three packages with (without) a label were perceived as green was created11.
The three single linear regressions with perceived greenness of the package material conditions (higher-level) as the dependent variables and perceived salience (lower-level) as the independent variable revealed no significant regression equation (cardboard package, F(1,40) = 0.56, p = .459; brown package, F(1,40) = 0.57, p = .453; white package, F(1,40) = 0.33, p = .567; see Table 1). The same holds for the two single linear regressions with the label conditions as the dependent variables and perceived salience as the independent variable (label, F(1,40) = 2.59, p = .116; no label, F(1,40) = 2.00, p = .165, see Table 2).
9 To compare the two measurement methods, two paired samples t-tests were conducted, after the mouse
tracking data was recoded to a scale from 0-4 for the packaging conditions and 0-3 for the label conditions. A paired samples t-test of the salience of the package conditions between the heat map and the mouse tracking data reveals no significant mean difference between the heat map and the mouse tracking data. However, a significant mean difference can be found for the label conditions. On average, the package with a label was perceived as more salient with the mouse tracking measurement (M = 2.19, SD = 0.79) than with the heat map measurement (M = 2.48 SD = 0.67, t(41) = 2.50, p = .017).
10 I.e. sum of white with label and white without label; cardboard with label and cardboard without label and
brown with label and brown without label respectively.
11 I.e. sum of white with label, carboard with label and brown with label; and sum of white without label,
Table 2
Single Linear Regression Labels
Perceived greenness
Green label Green label No green label No green label Mouse tracking Heat map Mouse tracking Heat map
b (SE) b* b (SE) b* b (SE) b* b (SE) b* Constant 2.06 (0.32)*** 5.51 (0.52)*** 1.80 (0.35)** 3.12 (.58)*** Lower-level 0.11 (0.07) 0.25 -0.06 (0.20) - 0.05 0.11 (.08) -0.22 0.03 (0.23) -0.02 R2 .06 .00 .07 .00 F 2.59 0.09 0.91 0.23 Note. * p <.05. ** p <.01. *** p <.001., + p < .10, n = 42
Heat map: Lower- and higher-level processing. The same analysis was done
for the explicit assessment of perceived greenness (higher-level) on a 7-point scale. The dependent variable, perceived greenness of the package, needed to be computed per package (cardboard, white, and brown) with a scale from 2-14, with 14 meaning that both cardboard packages (with and without label) were perceived as very green12, and per label condition (label, no label) with a scale from 3-21, with 21 meaning that all three packages with (without) a label were perceived as very green13.
The three single linear regressions with perceived greenness of the package material conditions (higher-level) as the dependent variables and perceived salience (lower-level) as the independent variable revealed no significant regression equation for the cardboard
package (F(1,40) = 0.35, p = .558) or the brown package (F(1,40) = 1.06, p = .310, see Table 1). For the white package, the regression equation is marginally significant (F(1,40) = 4.00, p = .052). The two single linear regressions with the label conditions as the dependent variables and perceived salience as the independent variable revealed no significant regression equation
12 I.e. sum of white with label and white without label; cardboard with label and cardboard without label and
brown with label and brown without label respectively.
13 I.e. sum of white with label, carboard with label and brown with label; and sum of white without label,
GREEN PACKAGING CUES AND PACKAGE PROCESSING 23 (label, F(1,40) = .09, p = .763; no label, F(1,40) = .02, p = .880, see Table 2). Thus, the
results cannot support Hypothesis 3.
Hypothesis 4: Moderation.
Mouse tracking. To test the fourth hypothesis of whether the positive impact of
lower-level processing on higher-lower-level processing is moderated by (1) green consumerism and (2) environmental knowledge, three multiple linear regressions for the packaging material
conditions were conducted. Including perceived greenness of the package material conditions (higher-level) as the dependent variables, perceived salience (lower-level) as the independent variable, and the moderators (green consumerism, subjective and objective knowledge), the multiple linear regressions revealed no significant regression equations (cardboard package, F(7,34) = 1.16, p = .352; brown package, F(7,34) = .73, p = .647; white package, F(7,34) = .83, p = .572; see Table 3).14 The same holds for the multiple linear regression including the label-conditions as the dependent variables (label, F(7,34) = .37, p = .913; no label, F(7,34) = 1.81, p = .116; see Table 4).
Heat map and questionnaire. In the same manner, three multiple linear regressions
for the packaging material conditions were conducted with the heat map and questionnaire data. Including perceived greenness of the package material conditions (higher-level) as the dependent variables, perceived salience (lower-level) as the independent variable, and the moderators, the multiple linear regressions revealed no significant regression equations (cardboard package, F(7,34) = .62, p = .739; brown package, F(7,34) = 1.46, p = .213; white package, F(7,34) = 1.61, p = .167). The same holds for the multiple linear regression that included the label-conditions as the dependent variables (label, F(7,34) = .40, p = .894; no label, F(7,34) = 1.67, p = .150). Thus, the results cannot support Hypothesis 4.
14 Subjective knowledge and green consumerism are significantly correlated (r = .48). To account for
multicollinearity, separate multiple regressions including only one moderator respectively (e.g. cardboard x subjective knowledge) were conducted for all tests reported under the headline Hypothesis 4: Moderation., but these did not yield significant results either.
25 Table 4
Multiple Linear Regression Labels
Perceived greenness
Green label Green label No green label No green label Mouse tracking Heat map Mouse tracking Heat map
b (SE) b* b (SE) b* b (SE) b* b (SE) b* Constant 2.63 (2.69) 4.52 (4.65) 5.13 (2.59) 4.76 (4.61) LL 0.03 (0.60) 0.07 0.29 (1.75) 0.22 -1.23 (0.58)* -2.53 -0.10 (1.74) -0.07 GK -0.05 (0.57) -0.06 0.88 (0.97) 0.89 -1.28 (0.55)* -1.46 -1.30 (0.96) -1.19 LL*GK 0.01 (0.12) 0.13 -0.33 (0.36) -1.66 0.28 (0.11)* 3.69 0.28 (0.35) 1.31 SK -0.14 (0.65) -0.17 -1.16 (0.84) -1.15 0.78 (0.62) 0.87 1.63 (0.83)+ 1.47 LL*SK 0.02 (0.13) 0.16 0.41 (0.31) 1.64 -0.11 (0.13) -1.08 -0.46 (0.31) -1.64 OK 0.02 (0.39) 0.02 0.07 (0.63) 0.08 0.17 (0.38) 0.22 -0.13 (0.62) -0.13 LL*OK 0.01 (0.08) 0.05 0.01 (0.24) 0.05 0.00 (0.08) 0.01 0.03 (0.24) 0.08 R2 .07 .08 .27 .26 F 0.37 0.40 1.81 1.67
Note. * p <.05. ** p <.01. *** p <.001., + p < .10, n = 42, LL = Lower-level processing, GK = Green consumerism, SK = Subjective Knowledge, OK = Objective Knowledge
Direct effects on higher-level processing.
Mouse tracking. The direct effects of the different package materials on perceived
greenness were tested using a one-way repeated measures ANOVA. This yielded a significant effect of the packaging material conditions on perceived greenness (Wilk’s Lambda = .65, F(2,40) = 10.72, p <.001). Three paired samples t-tests were used to make post-hoc comparisons between conditions. A first paired samples t-test indicated that there was a significant difference in the perception of greenness of the cardboard package (M = 1.45, SD = 0.63, scale 0-2) and the white package (M = 0.81, SD = 0.10, t(41) = 4.60, p < .001) and the cardboard package and the brown package (M = 1.02, SD = 0.47, t(41) = 3.77, p < .001). Thus, a green packaging material was perceived as significantly greener than a non-green
packaging material. Lastly, the brown package was perceived as significantly greener than the white package (t(41) = 2.04, p = .048).
To test whether the effect of a green packaging material on higher-level attention is moderated by green consumerism and environmental knowledge, a general linear model with the condition as a within- and the moderators as between-subjects factors was tested. When including green consumerism and environmental knowledge as moderators, the main effect of package material condition on perceived greenness becomes insignificant (Wilk’s Lambda = .98, F(2,37) = 0.40, p =.673). All interaction effects are insignificant (label x green
consumerism, Wilk’s Lambda = .93, F(2,37) = 1.40, p =.260; label x subjective knowledge, Wilk’s Lambda = .91, F(2,37) = 1.90, p =.164; label x objective knowledge, Wilk’s Lambda = .99, F(2,37) = 0.18, p =.835).
Secondly, a direct effect of use of label on perceived greenness was tested. A paired samples t-test was conducted to compare the two conditions. The results show a significant mean difference in perception of greenness of the packages with a label (M = 2.55, SD = 0.71) and without (M = 0.74, SD = 0.77, t(41) = 11.26, p < .001). Thus, a product packaging with an organic label leads to more higher-level attention than a product packaging without an organic label.
To test, whether the effect of an organic label on higher-level attention is moderated by green consumerism and environmental knowledge, a general linear model with the condition as within- and the moderators as between-subjects factors was tested. The main effect of label condition on perceived greenness becomes insignificant (Wilk’s Lambda = .94, F(1,38) = 2.55, p =.188). All interaction effects are insignificant (label x green consumerism, Wilk’s Lambda = .99, F(1,38) = 0.28, p =.600; label x subjective knowledge, Wilk’s Lambda = 1.00, F(1,38) = 0.14, p =.715); label x objective knowledge, Wilk’s Lambda = 1.00, F(1,38) = 0.41, p =.524).
GREEN PACKAGING CUES AND PACKAGE PROCESSING 27
Questionnaire. The direct effects of the different package materials on perceived
greenness were also tested with the questionnaire data. A one-way repeated measures ANOVA yielded a significant effect of green packaging material condition on perceived greenness (Wilk’s Lambda = .48, F(2,40) = 21.48, p <.001). Two paired samples t-tests indicated that there was a significant difference in the perception of greenness of cardboard packaging (M = 5.05, SD = 1.06, scale 1-7) and white packaging (M = 3.48, SD = 1.01, t(41) = 6.63, p < .001) and cardboard packaging and brown packaging (M = 4.06, SD = 0.86, t(41) = 5.24, p <.001). Thus, a green packaging material was perceived as significantly greener than a non-green packaging material, which is in line with the mouse tracking analyses. Lastly, the brown package was perceived as significantly greener than the white package (t(41) = 4.28, p < .001).
To test, whether the effect of a green packaging material on higher-level attention is moderated by green consumerism and environmental knowledge, a general linear model with the condition as a within- and the moderators as between-subjects factors was tested. The main effect of packaging condition on perceived greenness becomes insignificant (Wilk’s Lambda = .93, F(2,37) = 1.45, p =.248). The interaction effects of label condition and green consumerism (Wilk’s Lambda = .84, F(2,37) = 3.59, p =.038) and subjective knowledge (Wilk’s Lambda = .86, F(2,37) = 3.12, p =.056) are marginally significant. The interaction effect of label condition and objective knowledge is not significant (Wilk’s Lambda = 1.00, F(2,37) = 0.07, p =.929).15
15 To examine the main effect of green consumerism on environmental knowledge in more detail, three multiple
linear regressions with green consumerism and subjective as well as objective environmental knowledge as the independent and perceived greenness of the packages in the different packaging material conditions were conducted. The multiple regression with the perception of greenness of the cardboard package reveals no significant regression equation (F(3,38) = 0.92, p = .439).Using the brown package as the dependent variable, the regression reveals a significant regression equation (F(3,38) = 3.43, p = .026) with an R2 of .21. Both, green
consumerism (b* = -.51, t = -3.09, p = .004, 95% CI [-0.83, -0.17]) and subjective knowledge (b* = 0.37, t = 2.22, p =.033, 95% CI [0.32, 0.70]) were significant predictors of perceived greenness for the brown package, while objective knowledge was not a significant predictor. Examining the perceived greenness of the white package, the regression equation is marginally significant (F(3,38) = 2.548, p = .070) with an R2 of .17. Green
Secondly, a direct effect of an organic label on perceived greenness was tested. A paired samples t-test was conducted to compare the two conditions, showing a significant mean difference in perception of greenness of the packages with a label (M = 5.36, SD = 0.87) and the packages without a label (M = 3.03, SD = 0.96, t(41) = 12.09, p < .001). Thus, a product packaging with an organic label leads to more higher-level attention than a product packaging without an organic label.
To test, whether the effect of an organic label on higher-level attention is moderated by green consumerism and environmental knowledge, a general linear model with the label condition as within- and the moderators as between-subjects factors was tested. The main effect of label condition on perceived greenness becomes insignificant (Wilk’s Lambda = .99, F(1,38) = 0.56, p =.485). The interaction effect of label condition and green consumerism on perceived greenness is significant (Wilk’s Lambda = .88, F(1,38) = 5.13, p =.029) and the interaction effect of label condition and subjective knowledge on perceived greenness is marginally significant (Wilk’s Lambda = .91, F(1,38) = 3.85, p =.057). The interaction effect of label condition and objective knowledge is not significant (Wilk’s Lambda = .97, F(1,38) = 1.17, p =.287).16
Implicit measurements.
In a next step, the MD and the amount of x- and y-flips measured by the mouse tracker were taken into account in order to measure the underlying processes implicitly. The higher the MD and number of x-, and y-flips, the more uncertain respondents were in their decision. Independent samples t-test were conducted for every combination, resulting in 30 tests for
greenness for the white package, while subjective and objective knowledge were not significantly associated with the perception of greenness.
16 Two multiple linear regressions with green consumerism and subjective and objective knowledge as the
independent and perceived greenness of the packages with a label and without a label respectively were
conducted. The regression with the package without the logo revealed a significant regression equation (F(3,38) = 3.29, p = .031) with an R2 of .21. Both, green consumerism (b* = .50, t = 3.00, p = .005, 95% CI [0.91,
-0.18]) and subjective knowledge (b* = 0.38., t = 2.27, p = .029, 95% CI [0.05, 0.80]) were significant predictors of perceived greenness of the package without a label, while objective knowledge was not a significant predictor. In contrast, the multiple regression with the perception of greenness of the package with a label reveals no significant regression equation (F(3,38) = 0.25, p =. 859).
GREEN PACKAGING CUES AND PACKAGE PROCESSING 29 MD, x-flips and y-flips respectively.17 The independent samples t-tests revealed no significant mean difference in MD and the amount of x-, and y-flips for the participants perceiving the package in a green package material as green in comparison to the ones perceiving it as not green. The same holds for the label conditions.
Study 2 Method
Sample. The second part of Study 1, namely the heat map exercise and the
questionnaire, was replicated and distributed online. All manipulations and measurements were identical to those used in Study 1.1819
A convenience sample of 110 participants took part in the online survey. One person was excluded because of colorblindness, resulting in a final sample size of 109. Of these participants, 23.9% reported to eat a vegan or vegetarian diet. The mean age was 27.36 (SD = 6.39) years and 71.6% of the sample was female (26.6% male, 1.8% other). In terms of education, 8.3% were high school graduates or equivalent, 47.7% held a Bachelor degree or equivalent, and 44.0% reported holding a degree higher than the Bachelor level. At the time of the survey, 38.5% of the respondents lived in the Netherlands and 34.8% lived in Germany, with these two countries making up the majority of the sample. Of the respondents, 83.5%
17 Due to the scope of this thesis, the t-tests are not reported here. For more detailed information, contact the
author.
18 The scale for green consumerism is internally consistent (Chronbach’s alpha = .91, M = 5.52, SD = 0.97). The
nine questions related to green consumerism (used in Study 1) were factor analyzed using principal component analysis with Varimax rotation. The analysis yielded one factor with eigenvalues greater than one explaining a total of 61.72% of the variance. Bartlett’s test of sphericity, which tests the overall significance of all the
correlations within the correlation matrix, was significant (χ2 (36)= 615.64, p < .001). Thus, it was appropriate to
conduct a factor analysis. A Kaiser-Meyer-Olkin measure of sampling adequacy indicated that it was acceptable to proceed with the analysis, since the strength of the relationships among variables was high (KMO = .91).
19 The scale for environmental knowledge is internally consistent (Chronbach’s alpha = .86, M = 3.92, SD =
1.15). A factor analysis including the seven items measured resulted in a two-factor solution. As the correlation between the two components is negative with -.16, it was decided to exclude the reversed items. The five remaining items related to subjective knowledge were factor analyzed using principal component analysis with Varimax rotation. The analysis yielded one factor with eigenvalues greater than one explaining a total of 65.16% of the variance. Bartlett’s test of sphericity, which tests the overall significance of all the correlations within the correlation matrix, was significant (χ2 (10)= 283.15, p < .001). Thus, it was appropriate to conduct a factor
analysis. A Kaiser-Meyer-Olkin measure of sampling adequacy indicated that it was acceptable to proceed with the analysis, since the strength of the relationships among variables was high (KMO = .79).
indicated to at least seldomly purchase milk. Finally, 63.9% indicated to be familiar with the EU organic label.
Results
Hypothesis 1: Package material and lower-level processing. To answer the first
hypothesis of whether a green product package material leads to more lower-level processing than a non-green product package material, only combinations including a green package material (cardboard) vs. a non-green package (brown or white) were considered. The binary variables (0 = non-green package more salient, 1 = green package more salient) were summed up, to create an index score ranging from 0-4, where 4 means that participants consistently indicated that the green package was most salient20. A one-sample t-test was conducted to compare the sample mean with the estimated mean (in other words, if both would be similarly attention grabbing, the mean would be 2.00; deviations from the mean thus indicate that a package is more or less salient). The one-sample t-test revealed that the package with the cardboard look (M = 2.60, SD = 1.16) was perceived as significantly more salient than the estimated mean of 2.00; t(108) = 5.35, p < .001. Thus, hypothesis 1 is supported.
Hypothesis 2: Labels and lower level processing. To test the second hypothesis of
whether a product package with an organic label leads to more lower-level processing than a product package without an organic label, only combinations where one package featured a label and one package did not feature a label were considered. The binary variables (0 = package without a label more salient, 1 = package with a label more salient) were summed up resulting in a scale ranging from 0-321. A one-sample t-test was conducted to compare the sample mean with the estimated mean equals 1.50. The one-sample t-test revealed that the
20 I.e. white with label vs. cardboard with label; brown with label vs. cardboard with label; white without label
vs. cardboard without label; brown without label vs. cardboard without label
21 I.e. white with label vs. white without label; cardboard with label vs. cardboard without label; brown with
GREEN PACKAGING CUES AND PACKAGE PROCESSING 31 package with a label (M = 2.13, SD = 0.97) was perceived as significantly more salient than the estimated mean of 1.50; t(108) = 6.75, p < .001. Hypothesis 2 is thus supported.
Hypothesis 3: Mediation. To test the third hypothesis of whether lower-level
processing leads to higher-level processing, the dependent variable perceived greenness of the package (on a 7-point scale) needed to be computed per package (cardboard, white and
brown) with a scale from 2-14 with 14 meaning that both cardboard packages (with and without label) were perceived as very green22 and per label condition (label, no label) with a scale from 3-21 with 21 meaning that all three packages with (without) a label were perceived as very green23. The three single linear regressions with perceived greenness of the package material conditions (higher-level) as the dependent variable and perceived salience (lower-level) as the independent variable revealed a marginally significant regression equations for the cardboard material condition, F(1,107) = 3.72, p = .057, but none for the brown (F(1,107) = 0.21, p = .651), or white (F(1,107) = 0.29, p = .594, see Table 5) packaging material
conditions.
The two single linear regressions with the label conditions as the dependent variables and perceived salience as the independent variable revealed no significant regression equation for the no label-condition, F(1,107) = .59, p = .443), but a significant regression equation for the label-condition, F(1,107) = 8.94, p = .003). The regression model can therefore be used to predict perceived greenness, but the strength of the prediction is low, as 8% of the variation in perceived greenness can be predicted on the basis of perceived salience (R2 = .08). Perceived salience, b* = 0.28, t = 2.99, p < .001, 95% CI [0.13, 0.66], has a significant, but weak association with perceived greenness. Thus, the overall results cannot support Hypothesis 3.
22 I.e. sum of white with label and white without label; cardboard with label and cardboard without label and
brown with label and brown without label respectively.
23 I.e. sum of white with label, carboard with label and brown with label; and sum of white without label,
Hypothesis 4: Moderation. To test the fourth hypothesis of whether the positive
impact of lower-level processing on higher-level processing is moderated by (1) green consumerism and (2) environmental knowledge, three multiple linear regressions for the packaging material conditions were conducted with the heat map and questionnaire data. Including perceived greenness of the package material conditions (higher-level) as the dependent variables, perceived salience (lower-level) as the independent variable, and the moderators, the multiple linear regressions revealed no significant regression equations (cardboard package, F(7,101) = 1.51, p = .173; brown package, F(7,101) = 0.71, p = .665; white package, F(7,101) = .28, p = .961; see Table 6). The same holds for the multiple linear regression including the label-conditions as the dependent variables (label, F(7,101) = 2.16, p = .044; no label, F(7,101) = 0.36, p = .923). Thus, the results cannot support Hypothesis 4.
Direct effects on higher-level processing. The direct effects of different packaging
on perceived greenness were tested with a one-way repeated measures ANOVA. This yielded a significant effect of the cardboard condition on perceived greenness (Wilk’s Lambda = .55, F(2,107) = 43.05, p <.001). Three paired samples t-tests were used to make post-hoc
comparisons between conditions. The first indicated that there was a significant difference in the perception of greenness of the cardboard package (M = 4.56, SD = 1.45) and the white package (M = 3.42, SD = 1.22, t(108) = 8.79, p < .001), and the cardboard package and the brown package (M = 3.61, SD = 1.21, t(108) = 8.71, p <.001). Thus, a green packaging material was perceived as significantly greener than a non-green packaging material. Lastly, the brown package was perceived as significantly greener than the white package (t(108) = 2.17 p = .032).
35 To test whether the effect of a green packaging material on higher-level attention is moderated by green consumerism and environmental knowledge, a general linear model with the condition as a within- and the moderators as between-subjects factors was tested. When including the moderators, the main effect of packaging material condition on perceived greenness remains significant (Wilk’s Lambda = .90, F(2,104) = 5.45, p =.006). The interaction effect with green consumerism is marginally significant (Wilk’s Lambda = .96, F(2,104) = 2.39, p =.097).24 The interaction effects with environmental knowledge are not significant (packaging condition x subjective knowledge, Wilk’s Lambda = 1.00, F(2,104) = 0.26, p =.773; packaging condition x objective knowledge, Wilk’s Lambda = 0.98, F(2,104) = 0.83, p =.437).
Secondly, a direct effect of use of label on perceived greenness was tested. A paired samples t-test was conducted to compare the two conditions. The results showed a significant mean difference in perception of greenness of the packages with a label (M = 4.73, SD = 1.40) and the packages without a label (M = 2.99, SD = 1.13, t(108) = 15.11, p < .001). Thus, a product packaging with an organic label leads to more higher-level attention than a product packaging without an organic label.
To test, whether the effect of an organic label on higher-level attention is moderated by green consumerism and environmental knowledge, a general linear model with the condition as a within- and the moderators as between-subjects factors was tested. When the moderators are included, the main effect of label condition on perceived greenness remains significant (Wilk’s Lambda = .87, F(1,105) = 15.09, p <.001). The interaction effects are insignificant (label condition x green consumerism, Wilk’s Lambda = .98, F(1,105) = 2.04, p
24 To examine the main effect of green consumerism in more detail, three single linear regressions with green
consumerism as the independent and perceived greenness of the packages in the different packaging conditions were conducted. The single linear regression with the perception of greenness of the brown package (F(1,107) = 2.65, p = .106) and the white package (F(1,107) = 0.21, p = .886) reveals no significant regression equation. Using the cardboard package as the dependent variable, the regression reveals a marginally significant regression equation (F(1,107) = 3.26, p = .074) with an R2 of .03. Green consumerism (b* = -.17, t = -1.81, p < .074, 95%
CI [-0.54, 0.03]) is a marginally significant but weak predictor of perceived greenness of the package with the cardboard package.
= .156; label condition x subjective knowledge, Wilk’s Lambda = 1.00, F(1,105) = 0.13, p = .719; label condition x objective knowledge, Wilk’s Lambda = .99, F(1,105) = 1.43, p = .234) Thus, the effect of the environmental label on higher-level attention is not moderated by green consumerism and environmental knowledge.
Discussion and Conclusion
The overall research question for this thesis was: Do green packaging cues positively
influence attention and comprehension processes? It was argued that green products need to draw attention and to be understood as green to achieve importance in consumers’
consideration set when making a purchase decision. This is particularly important, since previous research found that green products are valued and perceived as positive by
consumers and might thus lead to a more sustainable purchasing behavior (Rokka & Uusitalo, 2008). Thus, the purpose of this thesis was to investigate the impact of green cues, i.e. green packaging materials and organic labelling, on lower-level (attention) and higher-level (comprehension and elaboration) processing as introduced in a four-level processing model by Greenwald and Leavitt (1984).
Findings
Lower Level Processing. By means of different experimental studies, it was shown
that green cues indeed trigger lower-level attention processes, such that green packages (i.e. cardboard packages and packages featuring a label) are more salient and are looked at first. This is in line with previous research, showing that certain packaging cues like color, label, or graphic form are able to increase focal attention to a package (Ampuero & Vila, 2006;
Underwood, 2003; Piqueras-Fiszman et al., 2013). This thesis proved that green cues as stimulus-characteristics are influencing lower-level processes of attention, possibly because of their salience and unexpectedness (Meißner et al., 2016). Hence, green packages receive the necessary attention to have the potential to win the competition against other packaging attributes and make it into a consumers’ consideration set in a purchase situation.
