What factors influence consumers to buy fake meat products? In this exploratory research the authors aim at identifying a variety of factors that explain why consumers buy fake meat products, which have been on the rise in the past years in Germany. The authors follow a split approach in form of one quantitaive (regression analysis) and one qualitative (means-end chain) study to gain insights.
Table of Content
List of Abbreviations
List of Figures
List of Tables
List of Appendices
1 Introduction
2 Research Design
2.1 Literature Review
2.2 Derivation of the Research Design
3 Expert Interviews
4 Study One
4.1 Hypotheses Derivation
4.2 Conceptual Model
4.3 Data Collection and Sample Description
4.4 Questionnaire Design
4.5 Data Preparation
4.5.1 Definition of Dependent and Independent Variables
4.5.2 Control Variables
4.5.2.1 Factor Analysis
4.5.2.2 Socio-demographics
4.6 Scenario Analysis
4.6.1 Research Design
4.6.2 Methodology
4.6.3 Results
4.6 Hierarchical Linear Regression Analysis
4.7.1 Research Design
4.7.2 Methodology
4.7.3 Results
5 Study Two
5.1 Means-end Chain Theory
5.1.1 Introduction
5.1.2 Assumptions of the Means-end Chain
5.1.3 Means-end Chain Design
5.1.4 Laddering
5.2 Means-end Chain Survey
5.2.1 Questionnaire Design
5.2.2 Means-end Chain Methodology
5.3 Derivation of Factors
5.4 Results
6 Implications
7 Limitations and Further Research
References
Appendices
List of Abbreviations
Abbildung in dieser Leseprobe nicht enthalten
List of Figures
Figure 1: Research Structure
Figure 2: Conceptual Model
Figure 3: Hierarchical Association Chain
Figure 4: HVM - all Participants
List of Tables
Table 1: Factor Analysis Results
Table 2: Coefficients of the Final Regression Model
Table 3: Comparison of Laddering Techniques
Table 4: Means-end Chain Factors
List of Appendices
Appendix 1 - Experts on type of producer
Appendix 2 - Key notes expert interviews
Appendix 3 - Sample description
Appendix 4 - Purchase intention
Appendix 5 - Consumption of meat substitutes by customer type
Appendix 6 - Outlier test (5/10 % confidence interval)
Appendix 7 - Correlation of variables
Appendix 8 - Questionnaire constructs, items and order of questions
Appendix 9 - Cronbach’s Alpha: purchase intention
Appendix 10 - Cronbach’s Alpha: independent variables
Appendix 11 - Cronbach’s Alpha: scenario purchase intention
Appendix 12 - Fit of data for factor analysis
Appendix 13 - Anti-image covariance matrix and anti-image correlation matrix
Appendix 14 - Communalities
Appendix 15 - Communalities_2
Appendix 16 - Cronbach’s Alpha - factors: variety seeking, meat reduction and curiosity
Appendix 17 - Criteria final factor solution
Appendix 18 - Argumentation factor self-optimization
Appendix 19 - Kaiser criterion, screeplot and Eigenvalue >1 criterion
Appendix 20 - Final factor solution with reliabilities
Appendix 21 - Test for normal distribution of DV
Appendix 22 - Purchase intention hybrid and pure
Appendix 23 - Mann-Whitney-U test H1
Appendix 24 - Purchase intention flexitarians
Appendix 25 - Purchase intention heavy meat consumers
Appendix 26 - Purchase intention meatless consumers
Appendix 27 - Purchase intention meat consumers
Appendix 28 - Purchase intention hybrid - meat and no meat consumers
Appendix 29 - Purchase intention pure - meat and no meat consumers
Appendix 30 - Customer type specific purchase intention - hybrid producer
Appendix 31 - Customer specific purchase intention - pure
Appendix 32 - Test for correctly specified model
Appendix 33 - Expected value of error term
Appendix 34 - Histogram of standardized residuals
Appendix 35 - Histogram of standardized residuals
Appendix 36 - K-S test for unstandardized residual
Appendix 37 - ANOVA table of the final regression model
Appendix 38 - Model summary of the final regression model
Appendix 39 - Descriptive statistics: nutrition involvement
Appendix 40 - Final model (household size = 1 included)
Appendix 41 - Summary implication matrix
Appendix 42 - HVMs meat consumers and no meat consumers
Appendix 43 - HVMs customer types
Appendix 44 - HVMs heavy and mediocre meat substitute consumers
Appendix 45 - Comparison of terminal values per consumer type
Appendix 46 - HVMs males and females
Appendix 47 - Syntax
1 Introduction
August 2013, London: two volunteers tried the possibly most expensive burger that was ever cooked. The production of the beef patty gobbled 300,000 EUR and required years of extensive research (Schadwinkel 2013). The burger was the result of an attempt to produce meat in the laboratory, avoiding any animal slaughter or waste of natural resources that are needed to produce meat. While this event may have taken place mainly for publicity reasons, it represents one of the sensational highlights of a constant development to replace animal based meat with meat-like substitutes.
While artificial meat from laboratories is yet far away from market maturity, there are several other, mostly plant, eggs or fungus based approaches to create a product, which is able to fulfill consumer’s demands with respect to meat consumption. The markets for these meat substitutes grew constantly in the last years and the world market is projected to grow with more than six percent compound annual growth rate until 2020, reaching a volume of more than five billion USD (Markets and Markets 2014).
From a marketing perspective one of the most recognized symptoms of this development is the evolution of one of Germany’s biggest meat and sausage producers, RÜGENWALDER MÜHLE. Originally producing traditional meat and sausages specialties, the company started to invest in the development of meat substitutes, but not without facing internal resistance of employees due to a perceived dilution of the company’s heritage (Kwas- niewski 2015). However, their product innovations were so successful, that RÜGENWAL- DER MÜHLE is supposed to reach the goal to generate thirty percent of its total revenue with meat substitutes in 2016 and thus four years earlier than originally targeted (Kwas- niewski 2015). This example illustrates the necessity for (marketing) management to un- derstand consumers in order to sense important developments in their behavior. These insights are the very foundation that a company strategy should be based on. With regard to the required investments and according risks, the importance of changing consumer behavior should not be underestimated.
While the impact of the development of the meat substitute market is visible on the aggre- gated as well as on the business level especially from a financial perspective, there are various factors that influence consumers in their decision to buy meat substitutes. Previous consumer research has identified various factors that play a role in this context, such as the impact of sensory characteristics (Elzerman, Van Boekel, and Luning 2013, p. 700), health considerations (Hoek et al. 2004, p. 265) or environmental concerns (Hoek et al. 281 2011, p. 663). Also, the impact of socio-demographics, such as gender (Van Doorn and Verhoef 2015, p. 444) or age (Schösler, Boer, and Boersema 2012, p. 46) play a role. However, there are still gaps that have not been addressed before, but bear the potential to deliver valuable insights for managerial implications. The aim of this work is therefore to identify these gaps in existing research and to empirically derive what factors influence consumers to buy meat substitutes.
In accordance with this research aim, Chapter Two will provide a brief literature review in order to reveal what has not been covered by research so far and to set the foci of this work. Subsequently, the research design will be derived based on the requirements that are inherent to tackle the emphases of this work. In order to deliver thorough results, this chapter will argue for a split up of the empirical analysis, conducting both, a quantitative and a qualitative study. Chapter Three will then explain the first step of the data gathering process, namely expert interviews, which serve as a foundation for both studies.
The quantitative study (Study One) will be presented in Chapter Four and consists of two parts. First, based on the findings in Chapter Two and Three hypotheses will be derived and the conceptual model will be introduced. Afterwards the data collection process will be depicted and a sample description will be provided, before the questionnaire design is discussed. Second, the data preparation processes, including factor analysis and the intro- duction of control variables are depicted. This is followed by data analysis, using scenario and regression analysis. Ultimately, the results will be presented and discussed.
The qualitative study (Study Two) will be subject of Chapter Five. In the beginning, a brief introduction to the means-end chain and its underlying assumptions, the design and different laddering techniques will be provided. Following this, the data gathering process will be illustrated by explaining the questionnaire design and the underlying methodology. Afterwards, the factors that were considered in the means-end chain will be derived. Finally, the results of the means-end chain analysis will be presented.
Chapter Six will align the results of Study One and Study Two and use these empirical results as a foundation to develop meaningful managerial implications. Finally, in Chapter Seven limitations of this research work will be discussed and questions for further research will be identified.
2 Research Design
2.1 Literature Review
The purpose of this chapter is to derive a general research design that will serve as the basic structure for the research. Therefore, an identification of gaps in the existing meat substitute research by means of a brief literature overview will be provided in this chapter. Subsequently, those gaps will be identified that lie in the focus of this research work. Next, the research approach that is most appropriate to deliver these missing insights about meat substitute purchase will be derived in Chapter 2.2.
Meat substitutes, also called meat alternatives, meat analogs or meat replacers (Hoek et al. 2011, p. 662) “are primarily vegetable based food products that contain proteins made from pulses (mainly soy), cereal protein, or fungi” (Hoek et al. 2011, p. 662). The market for these products experienced a massive uplift around the end of the 90s (Davies and Lightowler 1998, p. 90) and is growing ever since (Markets and Markets 2014). The main reason for this development is represented by the increasing diversification of the con- sumer segments (Sadler 2004, p. 251). While meat substitutes were initially consumed nearly exclusively by vegetarians, they reach a broader audience nowadays. More and more non-vegetarians are buying such products either for a more diverse diet or because of the pursuit to deliberately reduce meat consumption (Sadler 2004, p. 251). In this con- text, consumers that deliberately renounce meat for at least one day per week are defined as “flexitarians” (de Bakker and Dagevos 2012, p.882).
Meat substitute research and comparable organic food literature have identified several socio-demographics to have an impact on the purchase behavior towards meat substitutes. Studies show ambiguity with one identifying women as the main costumer group of meat substitutes (Hoek et al. 2011, p. 667) and others finding no significant difference in the gender distribution (Hoek et al. 2004, p. 268). Organic food literature supports the finding of women as the focus customers (Van Doorn and Verhoef 2015, p. 444). Changes in diet and a development to consume less meat and instead more meat substitutes are likely to be related to younger persons and thus younger persons might consume more meat sub- stitutes (Schösler, Boer, and Boersema 2012, p. 46). The level of education is assumed to be higher among those, who mostly consume meat substitutes (Hoek et al. 2011, p. 667) and as a peer subject, organic product consumption is also positively related to the level of education (Ngobo 2011, p. 95). Household size, which is the number of persons that jointly eat the same meals and thus purchase food for the entire household, might be smaller for meat substitute consumers (Hoek et al. 2004, p. 268). This effect is supported in the field of organic purchasing (Van Doorn and Verhoef 2015, p. 444; Ngobo 2011, p. 98). Organic food research implied a positive effect of high level occupation on purchas- ing, which could also be present for meat substitute purchasing (Ngobo 2011, p. 95). A high degree of urbanization is a characteristic of meat substitute consumers and needs to be considered (Hoek et al. 2004, p. 268). Finally, income has not been found to have a significant influence on meat substitution patterns (Schösler, Boer, and Boersema 2012, p. 46) but organic food research indicated a positive influence of a higher income on organic food purchasing (Verhoef 2005, p. 257; Ngobo 2011, p. 98) and could potentially have an influence on the purchase intention.
In summary, gender, age, education, household size, occupation, degree of urbanization, consumer type and income are socio-demographic variables explored by recent literature to have an impact on the purchase intention towards meat substitutes or comparable prod- ucts.
Existing literature already assumed and tested a great amount of factors that in addition to the previously discussed socio-demographics may influence consumption of meat substitutes. The aim of this review is the identification of loops in literature, meaning the finding of important factors not yet targeted by past studies. In order to reach this goal, it is indispensable to uncover the actual research status.
The two most important, and in the literature most discussed, purchase factors, which hold for all customer segments, are environmental issues and animal welfare (Hoek et al. 2011; Hoek et al. 2004; Rozin, Markwith, and Stoess 1997). In this context, the environmental issue describes higher pollutions caused by meat production in comparison to meat sub- stitutes as discovered by Zhu and Van Ierland (2004). Aspects of the personal health were observed to have only a significant influence on the purchase intention of vegetarian meat substitute consumers but not on the intention of other customer segments (Hoek et al. 2004, p. 68). A possible explanation for this finding could be the above mentioned diverse con- sumer structure of meat substitute buyers, which comes along with a variety of different purchase reasons per segment. Thus, a differentiation between vegetarians and other meat substitute consumers as proceeded in Hoek et al. (2004) is useful. The similarity of meat substitutes to real meat, especially the taste, optic and texture, is a product attribute for that such a differentiation is needed since it is only preferred by small amount meat substitute consumers and thus only leads to a higher purchase intention of those customers that do not buy these substitutes regularly (Hoek et al. 2011, p. 669). Furthermore, a qualitative study of Elzerman, Van Boekel, and Luning (2013) indicates that a convenient preparation process of the products is an important factor. Moreover, the familiarity with the product and the frequency of consumption has also a high influence on the liking of these and thus also on the purchase intention (Hoek et al. 2013, p. 257). A related factor is food neophobia, which describes the phenomenon of avoidance of novel food by some customers (Pliner and Hobden 1992). Research revealed, that a high food neophobia leads to a lower pur- chase intention of meat substitutes (Hoek et al. 2011, p. 668). The price/quality relation is another product attribute, which influences the purchase intention. Hoek et al. (2013), p. 258, state, that the price/quality relation is more important for meat consumers buying meat substitutes.
In summary, some research has already been conducted in the field of meat substitute purchase. The influence factors that were examined so far can be roughly categorized. One category comprises factors that can be described as directly product related, such as meat similarity, price, convenience or quality perception. Another category includes con- sequences that might be evoked by meat substitute consumption, such as health protection through meat reduction, environmental protection or animal welfare. A third category contains socio-demographics.
Nevertheless, there are several factors of meat substitute purchase that have not been ex- plored by previous research. One possibly influential factor is the consumer’s knowledge about the producer, which could be either classified as a pure meat substitute producer or a hybrid producer, which also manufactures meat products. These insights would provide valuable orientation for different types of producers and help them to communicate effec- tively, how they produce. In addition, the impact of consumers’ nutrition involvement remains unclear. For producers, this impact is important to know in order to determine the importance of ingredient choice and consequently how products should be labelled. From a societal perspective, it has also not been investigated how meat substitutes and their con- sumers are being viewed and what impact these prevalent opinions might have on meat substitute consumers. In this context, no research has been conducted focussing on the possibility to sharpen the personal image by consuming meat substitutes. Moreover, as pointed out before, the consumer segments of this product category are rather diverse, but it is unclear, which segment represents the main consumer group.
Additionally, no research has been conducted focussing on the intrinsic value structure that lies at the core of the decision to consume meat substitutes. Research of these value structures has been conducted in related areas, such as organic food, and yielded valuable insights (Grunert and Grunert 1995). These can be seen as a foundation of understanding consumers and how different consumer groups may be successfully addressed by market- ing activities. Tying on to this gap, identifying how these intrinsic and hence more general values connect to specific product attributes delivers valuable insights about consumers of meat substitutes from a marketing perspective. The derived value may be even amplified considering the diversity in the consumer structure. The identification of these attribute- value chains constitute an important objective in order to further close research gaps.
2.2 Derivation of the Research Design
From a research perspective, the identified gaps of interest may be divided into two different types. First, there are single factors that are suspected to have an influence on the consumption of meat substitutes and have not been tested yet. The research should therefore aim at delivering results about their general impact, as well as about their strength of impact in comparison to other factors. Second, the research should be able to connect different levels of factors, such that the structure of intrinsic values, which lie at the core of the consumer’s decision making process, can be derived.
In order to reach both goals sufficiently, a Convergent Parallel Mixed Methods Design approach as proposed by Creswell (2014), will be applied. The idea behind this approach is, that different methods provide different types of information, which in the end should confirm or disconfirm each other (Creswell 2014, p. 269). While originally the confirmation purpose was emphasized, this approach will be mainly used to generate complementing results (Campbell and Fiske 1959, pp. 100-105, Jick 1979, p. 610).
In line with these theoretical suggestions, this research work will be divided into two stud- ies. As a basis for both studies, expert interviews are conducted upfront. Study One then focuses on generating quantitative results by means of developing a conceptual model, formulating hypotheses, gathering quantitative consumer data and regression/ scenario analysis. Study Two focuses on generating qualitative results by means of identifying rel- evant product and consumer related factors that might influence meat substitute consump- tion and their subsequent interconnection in a means-end chain analysis (MEC). The re- sults of both studies are ultimately used jointly to provide a differentiated answer to the research question. This structure is depicted in figure 1.
Summing up, this chapter provided a brief literature review and research gaps were identified. Afterwards, the Convergent Parallel Mixed Methods Design was introduced and it was pointed out, that based on this theoretical framework two studies, one qualitative, one quantitative, are most appropriate to address the aim of this work.
Figure 1: Research Structure
Abbildung in dieser Leseprobe nicht enthalten
Source: Authors’ own illustration.
3 Expert Interviews
The needs for conducting expert interviews are eclectic and can be classified into gaining explorative insights (Malhotra and Birks 2007, p. 212), supporting the hypotheses formulation, increasing reliability of factors not indicated by literature, specifying and exercising the MEC and giving estimations of the future of meat substitutes. Continuing, expert interviews enable to specify the literature review and tailor the implications gained to the needs of the research topic meat substitutes.
The expert interviews are downstream of the literature review, which leads to the advantage, that possible factors or hypotheses that have not been mentioned by existing literature but have evoked the researcher´s interest can be assessed and supported by experts. This also supports the selection of the most relevant hypotheses and factors.
All experts were asked to consecutively indicate meat substitute specific product attributes, psychological consequences and terminal value insights following the MEC approach. Since those categories are specified in advance for computerized presentation, a violation of the criterion that the raw data should be a result of the respondent’s cognitive structure and linkages and not those of the authors is possible (Grunert and Grunert 1995, p. 212). This can be only avoided if the answers that are predetermined by the researchers are highly exhaustive. It follows, that the experts next to the literature review enable to allevi- ate this bias of the computer based MEC design on the participants by increasing the de- gree of exhaustiveness of the given answers. In addition to this, computer based MEC, if already set up, cannot be influenced by the researcher anymore and must be feasible from the beginning. Having this in mind, expert interviews further support the practicality to exercise a MEC laddering for meat substitutes. Last, the interviews enable to derive esti- mations of future trends that will increase the relevance of the managerial implications.
Seven different experts were selected with the means of capturing the whole bandwidth of opinions from meat producer to hybrid producer and pure meat substitute producer, nutrition and marketing specialists and food technologists. This selection is particularly chosen to have contrary arguments and mirror the different involved stakeholders. Since the kind of the producer will be part of the analysis, the opinions of all different producer types provide major differentiation possibilities.
The primary data generation was conducted via telephone to account for obstacles concerning distance, time and convenience for the respondents. In detail, semi-structured interviews with the same questions for all experts consisting of broad open questions and structured questions following the laddering approach (Malhotra and Birks 2007, p. 212) were asked in the same direction and wording for all conducted interviews. All interviews have been recorded with consent of the interviewees.
Analyzing the responses of the experts, similarities in the answers concerning the three stages of the MEC were verified and aligned with the results derived by literature review. Continuing, these responses were either used to support a finding or to create a new one. Answers to the hypotheses specific questions were furthermore used to support both the relevance of the hypotheses and the direction of the effect. Answers to the open stated questions gained insights into general meat substitute market developments and future trends.
4 Study One
In this chapter, the first study within the Convergent Parallel Mixed Methods Design is presented. After deriving hypotheses and setting up the conceptual model (Chapter 4.1 and 4.2), the data collection and sample description (Chapter 4.3) as well as the question- naire design (Chapter 4.4) are outlined. This is followed by data preparation (Chapter 4.5) before the analyses and the according results (Chapter 4.6 and 4.7) are presented.
4.1 Hypotheses Derivation
Screening the company landscape of meat substitute producers, two different types of pro- ducers can be observed: hybrid and pure ones. Hybrid producers were originally pure meat product producers and added a line of meat substitute products. Hybrid producers in Ger- many are for instance RÜGENWALDER MÜHLE or HERTA. Both brands use their estab- lished brand name to extend it to meat substitutes. In this setting, brand attribute associa- tions are carried over (Aaker and Keller 1990, p. 36). This could be either the association of a professional and renowned meat processor, which is best at producing a meat substi- tute product that is closest to the original meat product or the association of animal slaugh- tering and low environmental friendliness. Another type of hybrid producers are the ones that established a subsidiary for the meat substitute products. An example is LUTZ FLEISCHWAREN GMBH, which sells their meat substitutes through their subsidiary ART- LAND CONVENIENCE GMBH branding their products VEGETARIA (vegetaria-food.de 2016).
The other category is the pure producer. They do not have their origin in the meat pro- cessing business and were founded following the trend towards meat-less products. The question of interest is, whether consumers care about the type of meat substitute producer. Since vegetarians and vegans score high on moral motivations like animal protection and environmental issues (Beardsworth and Keil 1992, p. 269), it is assumed, that their pur- chase intention towards meat substitutes is lower, knowing that the product was produced by hybrid producer compared with a pure one. Otherwise, they would cross-subsidize the meat processing industry, which is not perceived as animal and environment friendly (Zhu and Van Ierland 2005, p. 738).
Since the purchase of meat substitutes is usually linked to a reduction of meat, ecological arguments and the moral dimensions of animal treatment are main reasons for meat re- nouncement (Rozin, Markwith, and Stoess 1997). Both are not attributed to a hybrid pro- ducer and this supports the idea of a lower purchase intention. For other consumer groups that do not renounce meat products or want to reduce their meat consumption, the pre- dicted behavior is less intuitive and thus of high interest. Expert interviews revealed a strong interest in this topic and all experts see a link between the type of producer and the purchase intention. The majority assumes the direction of effect to be negative for hybrid producers but also the opinion was stated that for meat consumers, a hybrid producer is most trusted and credible to manufacture meat substitutes (see Appendix 1). This leads to the following hypothesis:
H1: Consumer ’ s knowledge about the producer being hybrid decreases the pur- chase intention towards meat substitutes
Tying on to the argumentation above, especially the consumers, who renounce meat prod- ucts, are assumed to react stronger on the negative effect of a hybrid producer on the pur- chase intention. Compared to the other groups, flexitarians and heavy meat consumers (meat abandonment less than once per week), vegetarians and vegans have the highest moral motivation (Lindeman and Sirelius 2001, p. 178). Additionally, vegetarians attrib- ute higher importance on product information like the producer compared to meat con- sumers (Hoek et al. 2004, p. 268). This is also supported by several experts, who suggest that this negative effect is mostly present for vegetarians (see Appendix 1). Hence, the following hypothesis is stated1:
H2: No meat consumption enhances the negative effect of a hybrid producer on the purchase intention towards meat substitutes
Literature has shown that a product being perceived as unique increases the purchase in- tention of consumers (Van Kleef, Van Trijp, and Luning 2005, p. 303). To the best of the author’s knowledge, it is yet unclear, how the consumer’s individual need for uniqueness influences their purchase intention towards meat substitutes. Consumer´s need for unique- ness was introduced by Tian, Bearden, and Hunter (2001) and describes their behavior to avoid products that are popular and bought by the majority. Following this, consumers with a high need for uniqueness aim to express their image through buying unique and rare products. Two experts indicated that meat substitutes are products for consumers who want to express alternative lifestyles and want to be hip through contrast (see Appendix
2). Supporting this, it is implied that individuals believe food to symbolize who they are (Weaver and Lusk 2014, p. 10). Knowing that meat substitute products are on the rise and gaining popularity, the authors still regard it as a rare and unique product that attracts attention of others and thus the following hypothesis is stated:
H3: The higher the need for uniqueness, the higher the purchase intention to- wards meat substitutes
Meat substitutes are foods and it is inevitable to assess their nutritional value for the con- sumers. A producer of meat substitutes and meat products has stated that meat substitute products highly consist of artificial flavors and flavor enhancers (see Appendix 2). Addi- tionally, he stated that it would not be possible to sell these products following the German standards of meat products. Adding to this, two ecotrophologists criticize the high degree of processing as well as the usage of artificial coloring and preservatives (see Appendix
2). It is proven, that vegetarians have a high involvement with nutrition (Janda and Troc-
chia 2001, p. 1237) and thus should care about the above mentioned problems with meat substitutes. In addition, meat substitute consumers are suspicious about the nutritional value (Elzerman, Van Boekel, and Luning 2013, p. 705). Assessing the consumers in general, nutrition is among the top three food choice attributes (Walker and Olson 1991, p. 16). It is concluded that those consumers, who are very involved in their nutrition, are likely to avoid meat substitutes and thus following hypothesis is stated:
H4: The higher the nutrition involvement, the lower the purchase intention to- wards meat substitutes
All in all, four different hypotheses will be tested within this study to evaluate the impact of the mentioned factors on the purchase intention towards meat substitutes.
4.2 Conceptual Model
In this study, the influence of the origin of the producer on the purchase intention towards meat substitutes will be tested. Adding to this, the impact of no meat consumption is of special interest and assumed to be a moderator. Next, the influence of the constructs need for uniqueness and nutrition involvement on the purchase intention towards meat substi- tutes will be further tested. This leads to the overall conceptual model of Study One as depicted in figure 2.
Figure 2: Conceptual Model
Abbildung in dieser Leseprobe nicht enthalten
Source: Authors‘ own illustra- tion.
4.3 Data Collection and Sample Description
Before the actual data was collected, a pretest with 16 participants was conducted in order to test the questionnaire for ambiguities of any kind. Only small obstacles were detected leading to a minor change in the wording of the scenarios and the adding of “usually” in the question “for how many household members do you usually go shopping?”.
Data was gathered within a survey period of one week using the online tool Qualtrics. In order to statistically prove H2, it was necessary to attain a sufficient number of no meat consumers. Therefore, the survey was spread not only in student fora, such as the Face- book group “Münster Marketing Major 17”, but was also supported by websites with a focus on no meat consumers, such as “www.vegan-freeletics.com” and fora such as “PETA ZWEI”.
Through this strategy, 806 participants were gathered, of which 212 were deleted because of unfinished questionnaires resulting in a valid sample of 594 participants. While a de- tailed sample description can be seen in Appendix 3, the key characteristics of the sample will be disclosed next.
The mean age of the sample is 27.88 years and more than two-thirds of the respondents are female. Most of the participants live in single households and towns with 100,000 to 500,000 inhabitants and indicated at least Abitur as highest educational degree. The ma- jority of the sample is still studying and has an income between 500 and 1,000 EUR. The distribution of the four different consumer types is nearly equal and 84 % of the respond- ents already have purchased meat substitutes. Only 2.7 % of the participants indicated no purchase intention at all (purchase intention = 1, see Appendix 4). Concerning the con- sumption of meat substitutes, the consumer types can be ranked with decreasing frequency: vegetarians, vegans, flexitarians and heavy meat consumers (see Appendix 5). Due to the measurement of the variables of interest using Likert scales, having both a floor and ceil- ing (1 and 7), it was decided not to discard individuals, who responded at either the low or high end of the limited scale. Furthermore, the sufficient sample size ensures, that pos- sible outliers do not affect the validity of the results (see Appendix 6). On the other hand, the metric variable age, which was measured on an infinite scale, was tested for outliers as well. One participant, who indicated an age of 9,999, was assigned the mean age of
27.88. The correlations of the variables of interests with the most important socio demographics can be seen in Appendix 7.
4.4 Questionnaire Design
The questionnaire for Study One consisted of one three-item-construct for the dependent variable, the independent variables (need for uniqueness and nutrition involvement), the two scenarios as well as the control variable product involvement due to its high im- portance. Furthermore, 24 additional single item control variables were implemented in order to test for a wide range of possible influences. To reduce the complexity, these con- trol variables were grouped within a factor analysis, which will be described in chapter
4.5.2.1. All used constructs as well as their sources can be found in Appendix 8.
Additionally, the questionnaire was expanded by eight socio-demographical questions. Each participant was asked for their gender, age, household size, education, degree of urbanization, occupation, income and consumer type. Consumption related characteristics, such as meat consumption frequency and purchase frequency for meat substitutes and the question for the duration of the meatless diet (respectively the question for the conscious meat abandonment), which was only displayed for the consumer types vegetarian and vegan (meat consumers), were asked additionally.
4.5 Data Preparation
4.5.1 Definition of Dependent and Independent Variables
In order to analyze the purchase intention for meat substitutes, the construct of the dependent variable (purchase intention), containing three items, was recoded into a single variable by calculating the mean value of the three question items. Afterwards, a reliability test was executed, yielding a Cronbach’s alpha of .862 (see Appendix 9). The same procedure was implemented for the two independent variables of interest (need for uniqueness and nutrition involvement). Both are reliable having a Cronbach’s alpha higher than .6 (see Appendix 10). Furthermore, the purchase intention of the two scenario constructs was reconstructed in the variables s1 purchase intention and s2 purchase intention, both also with a Cronbach’s alpha higher than .6 (see Appendix 11).
Additionally to the recoding, new variables were built in order to enrich the analysis. A variable containing the type of producer called scenario with value 1 for a pure producer and 2 for a hybrid producer and the variable scenario pi, which merges s1 purchase inten- tion and s2 purchase intention, were created in order to conduct the scenario analysis. Moreover, the new variables customer type (1 = heavy meat consumers, 2 = flexitarian, 3 = vegetarian, 4 = vegan) and customer type 2 (1 = meat consumer and 2 = no meat con- sumer) were created to enable data analysis at a greater level of detail.
4.5.2 Control Variables
To consider the influence of other variables and isolate the influence of nutrition involvement and need for uniqueness on the purchase intention towards meat substitutes, the influence of other variables has to be controlled for. Next to socio-demographic control variables, all 24 single item control variables and the product involvement included in the survey were suspected to have an influence on the purchase intention.
4.5.2.1 Factor Analysis
To summarize and reduce the number of the 24 single item control variables to a more manageable level and also to reduce the number of variables in a regression, an explorative factor analysis is conducted. Hereby, the principal components analysis method was used. Factor analysis detects correlations among variables and replaces them with a new set of uncorrelated variables (factors) (Malhotra and Birks 2007, p. 647). All variables were measured on a seven point Likert scale and thus, no standardization of the variables is necessary. Since factor analysis aims to group highly correlating variables into a factor, a prerequisite is a significant correlation between the observed variables. The Bartlett’s test of sphericity is significant (p-value < .001), which indicates a prevalent correlation be- tween the variables and furthermore, the Kaiser-Meyer-Olkin (KMO) measure of sam- pling adequacy has a sufficient value of .768 (see Appendix 12) also indicating that the data is suited for factor analysis (Malhotra and Birks 2007, p. 647). Supporting this, only 5.68 % of the absolute values of the elements in the anti-image covariance matrix beneath the diagonal are smaller than .09 and all variables have a measure of sampling adequacy larger than .5 (see Appendix 13). Running the factor analysis, variables with a low extraction will be excluded since they have only little shared variance and do not fit to other variables well. Following this, the variable value for money is excluded (extraction of .377) (see Appendix 14). After this deletion convenience still has a low shared variance (.436) and is thus removed from the factor analysis as well (see appendix 15). After this exclusion, Bartlett’s test of sphericity and KMO are still sufficient.
Following the exclusion of two variables from the factor analysis, 22 variables entered the factor analysis again and the Kaiser (latent root) criterion was used to determine the num- ber of factors. To check for reliability, Cronbach’s alpha was calculated for all factor so- lutions and should be higher than .6 (Malhotra and Birks 2007, p. 647). The factor com- bining the variables curiosity, meat reduction and variety seeking only scored .442 and was split up (see Appendix 16). After this change, all variables were still assigned to the same factor as for the first solution and only the importance of factor four and five have changed (former factor number five was product origin). The final factor solution now only consists of 19 variables with an extraction higher than .46 and significant Bartlett’s test of sphericity (p-value < .001) and middling KMO of .75 (see Appendix 17).
The variable with the lowest extraction (self-discipline) was retained because a deletion would decrease the reliability of the whole factor from .551 to .519 (see Appendix 18). Kaiser criterion, the eigenvalue larger than one and the scree plot all indicate a solution of six factors (see Appendix 19).
In general, the factor analysis reduced the number of control variables from 24 to eleven. Variables were allocated to a factor following the rotated component matrix. The previ- ously derived factors are sorted in decreasing order by their Eigenvalues. Two factors still resulted in a Cronbach’s alpha lower than .6 but evaluating the variables that were as- signed to the factors self-optimization and sensory characteristics, they all point mainly in the same direction and can be argued for following their meaning for the reality. The factor self-optimization combines items that display a relentless pursuit of quality, health and indulgence through consumption of food. The other critical factor, sensory characteristics combines all sensory variables of meat substitutes like the taste of its own, the consistency and the appearance. The final factor scores are depicted in table 1 and were calculated as the mean value of the group variables of the factor.
Table 1: Factor Analysis Results
Abbildung in dieser Leseprobe nicht enthalten
Source: Authors’ own illustration (see Appendix 20).
4.5.2.2 Socio-demographics
The ordinal socio-demographic variables income and degree of urbanization were treated as quasi-metric since the informational value of a higher or lower level of these variables is sufficient for this study. In contrast, the ordinal variable household size was dummy coded with the base category household size 1 to explore differences between the different household sizes in detail.
The nominal socio-demographical variables education, gender and occupation were dummy coded with the base categories Hauptschulabschluss, male, and pupil.
4.6 Scenario Analysis
4.6.1 Research Design
To test if the origin of the producer has an influence on the purchase intention, two ran- domly assigned scenarios were implemented in the questionnaire. A participant was either assigned to scenario one or scenario two. The first scenario is shaped for a hybrid producer and the second one for a pure producer. In both scenarios, the consumers were encouraged to think about being in the supermarket and wanting to buy meat substitutes. In the first scenario, they are asked to imagine finding a product in the supermarket from a producer, of which the consumers know that it is also producing conventional meat products. In the second one, the consumer is asked to imagine finding a product in the supermarket from a producer, of which they know that it only produces meat substitutes and no conventional meat products.
4.6.2 Methodology
To assess and statistically prove that the purchase intention of the consumers towards the meat substitute products of a hybrid producer is significantly lower compared with a pure producer, the mean values of the purchase intentions of the two scenarios need to be sig- nificantly different from each other. Since the dependent variable purchase intention is not normally distributed and the logistic transformation did not lead to a normal distribution either (see Appendix 21), non-parametric tests are suited best to test for differences in the group means (Zimmerman 1987, p. 171). The Mann-Whitney U test can be seen as the counterpart of the t-test and tests for significant differences in the mean of two groups of an independent sample (Malhotra and Birks 2007, p. 526). The Kruskal-Wallis test builds the non-parametric counterpart of the ANOVA (Malhotra and Birks 2007, p. 565) and is used to assess, whether the purchase intention towards meat substitutes in the two scenar- ios is different across more than two compared groups (heavy-meat consumers, flexitari- ans, vegetarians and vegans). Generally, in this study parameters are evaluated based on a significance level smaller than five percent.
4.6.3 Results
The random assignment of the scenarios leads to rather equal group sizes of 299 respond- ents for scenario one and 295 for scenario two. The average purchase intention for prod- ucts from the hybrid producer is lower compared to the pure producer (3.67 vs. 5.04) (see Appendix 22). Testing, whether this difference is statistically significant, the Mann-Whit- ney U test indicates a significant difference in the mean values of the purchase intention (p-value < .001) and thus, H1 can be supported: A hybrid producer decreases the purchase intention towards meat substitutes (see Appendix 23). This means that disregarding the type of consumer, a hybrid producer scores a lower purchase intention compared with a pure producer.
To gain more detailed insights, flexitarians and heavy meat consumers are further ana- lyzed. Flexitarians also care about the origin of the producer and score a significantly (p- value = .002) higher purchase intention for the meat substitutes of the pure producer (see Appendix 24). In contrast, if only heavy meat consumers are considered they have no statistically significant (p-value = .052) different purchase intention concerning the two types of producer and do not care about the origin of the producer (see Appendix 25).
The second hypothesis can be rejected. When analyzing no meat consumers (vegetarians and vegans) and meat consumers (flexitarians and heavy meat consumers), both groups show significant differences in the purchase intention concerning the type of producer. For both types, the purchase intention is higher towards meat substitutes of a pure producer, for meatless consumers the difference in the mean values is 2.22 and for meat consumers .61 (see Appendix 26 and 27). But the mean purchase intentions for the hybrid producer of no meat consumer are not significantly (p-value =.420) lower than the purchase intention of meat consumers (see Appendix 28). Thus, no moderation effect of no meat consumption could be detected and therefore H2 is rejected.
It is additionally revealed that the mean purchase intention of vegetarians and vegans to- wards meat substitutes of a pure producer is significantly higher than the purchase inten- tion of meat consumers (5.82 vs. 4.35), which means that no meat consumption moderates the positive effect of a pure producer on the purchase intention (see Appendix 29).
Analyzing the purchase intention towards meat substitutes of a hybrid producer yielded that neither between all four different categories of consumers (p-value = .216) nor be- tween vegans and all other consumers (p-value = .057) statistically significant differences in the mean values exist (see Appendix 30). In combination with the findings from H1, this means that flexitarians, vegetarians and vegans care about the origin of the producer, but no group significantly stronger.
Considering the second scenario, the influence of a pure producer on the purchase inten- tion, significant (p-value < .001) differences across all four groups exist: Following the Kruskal-Wallis test, all four groups differ significantly in their purchase intention towards meat substitutes, when assigned to the pure producer scenario (see Appendix 31). The purchase intention is highest among vegetarians, closely followed by vegans and flexitarians. Heavy meat consumers show the lowest purchase intention.
Compared to the results for the hybrid producer in scenario one, mainly vegetarians and vegans drive the higher purchase intention towards meat substitutes from a pure producer. Thus, a pure producer generally leads to a higher purchase intention and this effect is mainly driven by no meat consumption. These findings are in line with expert opinions and help to close the gaps in literature.
4.7 Hierarchical Linear Regression Analysis
4.7.1 Research Design
As described in chapter 4.1, the goal of this study is to uncover the effect of the need for uniqueness and the nutrition involvement on the purchase intention of meat substitutes. Thus, need for uniqueness and nutrition involvement serve as IVs. Furthermore, the re- gression function was extended by social demographic data (gender, education, age, oc- cupation, income, household size and degree of urbanization), the eleven factorized con- trol variables from chapter 4.5.2.1 and the product involvement because of their potential influence on the purchase intention. As described in 4.5.2.2, the four variables education, gender, occupation and household size are split into 24 sub-variables. Therefore, the final regression model consists of the constant term and 41 variables. The high number of par- ticipants prevents the regression model from overfitting because the ratio variables per participant (>10) is satisfying.
4.7.2 Methodology
In order to measure the influence of the need for uniqueness and the nutrition involvement as well as their effective direction on the purchase intention towards meat substitutes, a hierarchical ordinary least squares regression (in the following called OLS regression) was conducted. In this context, the OLS regression is able to measure the relationship of a metric dependent variable (DV) with independent variables (IVs) as well as control vari- ables (Malhotra and Birks 2007, p. 581) by constructing the best fitting line (so called regression line) between them that minimizes the sum of squared errors (Malhotra and Birks 2007, p. 584). In this procedure, it is necessary to estimate the value of beta param- eters for the constant term (β0), the IVs and the possible control-variables (ßn) within the regression equation as well as to calculate their p-values as an indicator of significance (Malhotra and Birks 2007, p. 584). The beta factor displays the change of the DV if the specific IV is changed by one unit and its sign gives insights of its effective direction (Malhotra and Birks 2007, p. 584). The strength of the overall regression model can be evaluated by the coefficient of determination R² as well as its equivalent adj. R², which is the R² adjusted for the number of IVs and control variables (Malhotra and Birks 2007, p. 591).
The regression model was implemented in three blocks within SPSS. The first block only consists of the two IVs in order to be able to measure their influence on the purchase intention by the assessment of R². The second block adds the factorized control variables and the product involvement. Last, the third block complements the socio-demographical variables and thus builds the final model.
In order to support the validity of the regression model, the underlying assumptions have to be tested. First of all, the correlation matrix was inspected (see Appendix 7). Visual inspection of the scatterplot of standardized predicted value and the studentized residual as well as the relationship between the DV and the IVs facilitate the correct specification of the model (see Appendix 32 (1-2.41)). Besides the expected error term of zero (see Appendix 33), errors are also normally distributed (see Appendices 34 and 35 for a visual inspection and 36 for the Kolmogorow-Smirnow-test). Endogeneity is not present by log- ical argumentation, because the purchase intention does not influence any IV. Multicol- linearity in this sample is low, indicated by the VIF values lower than three for metric variables, which is a result of the prior factor analysis (see table 2). The present study does not consist of any time related data and the variance of the error term is constant (see Appendix 32).
4.7.3 Results
The final model as described in the previous chapter is significant with an empirical F- value of 26,791 (see Appendix 37) and is able to explain 66.6 % (R² = .666; adj. R² = .641) of the variance of the DV (see Appendix 38). As can be seen in model one of Appendix
38 the explanatory power of a regression model consisting of only nutrition involvement and need for uniqueness is rather small explaining about three percent of the variance of the purchase intention (R² = .032; adj. R² = .029) (see Appendix 38), which supports the general consensus of this paper and existing literature, which describes the purchase in- tention of meat substitutes as being driven by many different factors.
Table 2 shows the significances of the coefficients of the final regression model as well as their standardized beta values as a signal of effect size and enables a comparison between the significant variables. Both, the overall results and especially the two independent var- iables of interest (nutrition involvement and need for uniqueness) will be described in the following.
The IV need for uniqueness is significant (p-value = .023) but has a negative coefficient of -.096 meaning the higher the need for uniqueness the lower the purchase intention to- wards meat substitutes. This finding contradicts hypothesis three of this paper since the effective direction is negative, thus H3 is rejected. An explanation of the observed results might be the high growth rate of meat substitutes as described in chapter 2.1. Since meat substitutes are an established product segment nowadays they might not attract people with high needs for uniqueness anymore. Those people are likely to change their con- sumption behavior when they learn that their preferences are shared with a majority (Ber- ger and Heath 2007, p. 128). Another possible explanation arises when considering that meat substitutes try to substitute (or often even imitate) meat. Therefore, it could be as- sumed that especially people, who try to stay connected with their meat eating social en- vironment, consume meat substitutes in order to not diverge. In other words, they consume the products due to their high need for belonging (in opposite to a high need for uniqueness) as described by Baumeister and Leary (1995).
The independent variable nutrition involvement is significant (p-value = .017) as well but with a positive beta coefficient of .120, meaning that the purchase intention for meat sub- stitutes increases with an increasing nutrition involvement. This finding rejects H4 since the effective direction is not negative as expected. A possible explanation for this result could be the wish to avoid specific ingredients. Specifically, the high average nutrition involvement of meat substitute consumers of 5.48 (see Appendix 39) could be explained by consumers scanning nutrition information in order to test if the product contains any undesired ingredients such as an excessive proportion of soy. Continuing, brands label their products and partly mislead the consumers by displaying only selected positive in- formation about ingredients.
Furthermore, some control variables were found to be significant and will be discussed in the following. The biospheric preservation was found to have a significant (p-value = .036) 28 20 positive influence on the DV, which is intuitive since an increase of such utilitarian values (especially of animal protection) should increase the purchase intention of meat substitutes. The product attribute meat similarity was found to have a significantly negative influence on the purchase intention, which can be explained by findings of Rozin, Markwith, and Stoess (1997), p. 68, showing that vegetarian consumers often dissociate themselves from any kind of meat and often enhance this to a pure reluctance against sensory characteristics of meat.
On the other hand, the findings of this study reveal a significant (p-value = .009) positive effect of sensory characteristics of meat substitutes on the DV. This finding is reasonable since meat substitutes should have an appealing taste, texture and optical appearance in order to be attractive to potential buyers. In addition, this study reveals that an increasing self-optimization significantly (p-value = .007) decreases the purchase intention for meat substitutes. Moreover, this study indicates that the purchase intention of potential meat substitute consumers increases significantly (p-value = .032) with an increasing curiosity. This finding is in line with former findings indicating that meat substitute consumers tend to have a lower food neophobia (Hoek et al. 2011, p. 668). P roduct involvement has the highest significant positive effect on the purchase intention (p-value < .001; standardized coefficient = .538), which is very intuitive. Another instinctive indication is the significant (p-value < .001) positive influence of the need for meat reduction on the DV since meat substitutes are perceived as a means to meat reduction.
Of the socio-demographic control variables, table 2 reveals that the household size 2 (p- value = .007), the consumer type vegetarians (p-value < .001) and vegans (p-value < .001) as well as the degree of urbanization (p-value = .040) were found to have a significant positive effect on the purchase intention of meat substitutes. While the findings concern- ing the consumer type are intuitive, the effect of the other two variables requires further interpretation. The fact that only the small household size of two was found to have a significant positive influence leads to the assumption that small household sizes could have a significant positive relation with the purchase intention towards meat substitutes. Therefore, a second regression analysis with a different base category coding proved that a size of one is also significant (see Appendix 40). This might be explained by the ability of small household sizes to experiment with regard to food and that less food related re- strictions have to be considered. The discovery of a rising purchase intention of potential
Abbildung in dieser Leseprobe nicht enthalten
Source: Authors‘ own illustration.
This and the finding regarding the household size is in line with the findings of Hoek et al. 2004, p. 268 as described in chapter 2.1.
5 Study Two
5.1 Means-end Chain Theory
5.1.1 Introduction
This section presents Study Two within the Convergent Parallel Mixed Methods Design. Chapter 5.1 introduces the means-end chain as a qualitative analytical tool from a theoret- ical perspective. Subsequently, in Chapter 5.2 the questionnaire design and the methodol- ogy of analysis are explained. The factors that are specific to the underlying topic of re- search are presented in Chapter 5.3. Last, Chapter 5.4 highlights the most important results.
The means-end chain theory, introduced by Gutman in 1982, provides an approach for linking consumers’ product values to their underlying terminal values. The theory sug- gests, that product attributes are linked to psycho-social consequences produced through consumption, and personal values, which determine consumers’ decision-making process (Gutman 1984, p. 25). Thus, MECs connect product knowledge to self-knowledge in or- der to analyze and interpret how consumers perceive products in relation to themselves (Walker and Olson 1991, p. 112). In one of the most commonly used examples of MEC analysis, offered by Reynolds and Gutman (1988), p. 796, consumers were asked for their beverage choices. Thereby, a female consumer stated, that she buys wine coolers when she stays at the bar with her coworkers, because she likes the bottle shape and the fancy label (product attributes). She explained, these product attributes let her look more femi- nine compared to drinking beer (psychological consequence), which in turn gives her the feeling of belonging to the group (terminal value). This chain of reasoning illustrates the idea of the MEC and serves as a structure for this study.
5.1.2 Assumptions of the Means-end Chain
The chain of reasoning underlies different assumptions. The first two assumptions claim, that (1) the desirable end-states of existence, here named terminal values, play a dominant role in the decision-making process of consumers and (2) that they succeed in grouping the enormous diversity of products that potentially satisfy their values into sets or classes in order to reduce the complexity of choice (Gutman 1982, p. 60). The last two, more general assumptions state, that (3) all consumer actions have consequences and (4) that all consumers learn to associate particular consequences with particular actions they may take (Gutman 1982, p. 70). Moreover, in order to use the MEC concept it is necessary, that consumers are highly involved in the product, to ensure a linkage between psychological consequences and terminal values (Gutman 1982, p. 66).
Hereby, the question arises, if meat substitutes are high involvement products. Hoek et al. (2011), p. 372, argues, that meat substitutes are food and therefore generally classified as 2823
low involvement products. In contrast, Verbeke and Vackier (2004), p. 166, already ar- gued, that this generalization shows a lack of appreciation due to increasing interest of food consumers in animal treatment, product origin and health issues. They found that especially “concerned meat consumers” have a high product involvement when testing effects of consumer involvement in fresh meat (Verbeke and Vackier 2004, p. 166). Like- wise, motivations for vegetarian consumption indicate high involvement in various prod- uct aspects, such as concerns with nutrition, fitness or environment (Janda and Trocchia 2001, p. 1225). During the interviews of this study, the experts were asked about the per- ception of meat substitutes in society. They answered, that it is a “divisive topic” forming “strong opinions”, which can result in a positive but also negative image of meat substi- tutes (see Appendix 2). This indicates, that meat substitutes are rather high than low in- volvement products, which is therefore assumed in the context of this study. Thus, all assumptions are seen as fulfilled.
5.1.3 Means-end Chain Design
One critical issue of MEC analysis is how to make consumers’ decision making processes tangible. In line with the wine cooler example, most literature uses three different levels, in which consumers’ answers are classified: (1) product attributes, (2) consequences of consumption, as well as (3) terminal values, which are worth to be achieved by the con- sumer (Jung and Kang 2010, p. 220; Paul et al. 2009, p. 217; Reynolds and Gutman 1988). Some literature uses more than three levels for a more detailed distinction between con- crete and abstract product attributes, functional and psychological consequences and in- strumental and terminal values (Veludo-de-Oliveira, Ikeda, and Campomar 2006b, p. 629). In any case, a means-end chain concept has a hierarchical structure, guided by the level of abstraction (Gutman 1982, p. 71). However, it is hard for consumers to activate their self- and product knowledge at any given time, wherefore results stay more con- sistent using less differentiations (Walker and Olson 1991, p. 113). Furthermore, Schwartz (1992), p. 49, emphasized the existing lack of empiricism in the distinction between in- strumental and terminal values, the former status quo defined by Rokeach (1973). This study is adapted to the Schwartz Value Survey, wherefore no distinction between instru- mental and terminal values is made. As a result, the hierarchical association chain as
Figure 3: Hierarchical Association Chain
Abbildung in dieser Leseprobe nicht enthalten
Source: Authors‘own illustration based on Gutman (1982).
shown in figure 3 is used in this study.
5.1.4 Laddering
Within the means-end chain methodology, laddering is the most commonly used method to connect product attributes with the terminal values, which are the underlying reasons for the purchase of products on an abstract level (Gutman 1982, p. 70). Through so called “ladders” the researcher tries to get a deeper insight of consumers’ intrinsic motivation for the purchase.
Laddering can be divided into soft and hard laddering, of which former is mostly realized by one-on-one in-depth interviews and the latter by the use of a questionnaire with predefined answering options or a free association paper-and-pencil experiment (Russell et al. 2004a, p. 569). A comparison of both techniques is depicted in table 3.
Table 3: Comparison of Laddering Techniques
Abbildung in dieser Leseprobe nicht enthalten
For this study it was decided to use a computerized presentation, not only because of the time and cost advantages mentioned in table 3, but also because of the user friendliness of the method. Moreover, one of the main purposes of this study is to give managerial impli- cations. As previous studies revealed, hard laddering is more suitable for this purpose (Russell et al. 2004a, p. 582). Additionally, evidence shows that the overall differences within the hard laddering techniques paper-and-pencil and computerized presentation are comparably small and, depending on the level of the cut-off value, up to non-existing (Russell et al. 2004b, p. 290). This will be described in chapter 5.2.2 in more detail. Last, computer based MECs enhance the chance to create a sufficient sample sizes.
5.2 Means-end Chain Survey
5.2.1 Questionnaire Design
The replication of the means-end chain and the laddering process within the online ques- tionnaire was realized based on the theoretical background as described. For the means- end chain analysis the pre-test revealed no obstacles. The survey started with the means- end chain in order to gather answers that are unbiased by the scenario or the conventional questionnaire part. The stages and their answering options can be seen in Appendix 8.
It was decided to allow respondents to choose more than one of their most important prod- uct attributes (consequences) in stage one (stage two) and afterwards in a subsequent set, indicate the most important in order to build a direct ladder (Russell et al. 2004a, p. 572). This approach supports the selection by decreasing complexity. In case a consumer chooses only one product attribute (consequence), he or she was directly guided to stage two (stage three). Furthermore, an open text field was implemented in stage one and two in order to give respondents the opportunity to list attributes or consequences that are po- tentially missing.
5.2.2 Means-end Chain Methodology
The most frequently used analysis technique of the laddering data, which leads to useful managerial implications, is based on two steps. First, a summary implication matrix (SIM) has to be created based on laddering data. Based on this matrix a hierarchical value map (HVM) is drawn in a second step in order to visualize the means end chain results (Reynolds and Gutman 1988). Both steps will be discussed in detail.
In comparison with Reynolds and Gutman (1988), the SIM in this study was reduced to display only the product attributes and consequences in the rows whereas consequences and terminal values are in the columns (see Appendix 41). This procedure was imple- mented, because the computerized laddering process in this study lacks interconnections within different levels. The cell entries in the SIM are the frequencies of the chosen path- ways of the means-end chain, while they can be distinguished between direct (stage one to stage two and stage two to stage three (lower block) in Appendix 41) and indirect (stage two to stage three (upper block) in Appendix 41) connections. A direct connection measures the number of times participants chose a particular consequence after a specific product attribute or a particular terminal value after a specific consequence, for example the number of times respondents selected organic ingredients as a product attribute and health as a consequence of choosing organic. Indirect connections on the other hand measure the frequency of a chosen terminal value after a particular product attribute was selected (Reynolds and Olson 2001, p. 129). In this study, the analytical focus lies on the direct connections, because it is possible, that certain consequences may lead to specific terminal values, but show no distinctive connection to any product attribute. Vice versa, a certain product attribute may connect to a particular consequence, but no clear connections to any terminal value follow.
In order to depict the cognitive structure of the participants for the studied product, a HVM is drawn. This step is especially useful for getting deeper insights for the planning of the marketing-mix elements (Botschen and Hemetsberger 1998, p. 153). This goal can only be reached if a clear structure lies behind the HVM. In line with Russell et al. (2004a), only the top five connections are displayed in the HVM. Therefore, the SIM is used to indicate the top five ladders of each stage, first from product attributes to psychological consequences and second from psychological consequences to terminal values. This ap- proach automatically indicates the top attributes, consequences and values. As the map can be compiled for different samples with certain characteristics, such as gender or con- sumption behavior, the top five connections may change and consequently also the top attributes, consequences and values change. The advantage of this consistent approach is the reduction and thus the focus on the most important structure elements, which enhances the extraction of meaningful insights (Reynolds and Gutman 1988, p. 20). An alternative approach is the determination of a fixed cut-off value for single connections. This ap- proach on the other side can lead to an overrepresentation of minorities, especially in com- puterized laddering studies, and can thus impede the extraction of meaningful insights (Reynolds and Gutman 1988, p. 20).
It is important that the researcher avoids crossing lines of connections, which interferes the interpretability of the map and the coherencies between particular elements (Reynolds and Gutman 1988, p. 21). Within the HVM the most important attributes, consequences and terminal values can be represented through a variation of their box size, increasing with the number of frequencies of the specific element. In addition, the HVM can indicate the most important ladders by a variation of the thickness of the lines between the particular elements. Analogous to the box sizes, a thicker line states a stronger connection between two elements (Voss, Gruber, and Szmigin 2007, p. 954).
5.3 Derivation of Factors
In order to create an accurate set of factors in the different stages, literature on meat sub- stitutes, organic/fresh meat and vegetarianism was scanned. Moreover, within the inter- views conducted for this study, the experts were asked about possible factors for each stage. The final set of attributes for product characteristics and psychological conse- quences, based on the authors’ and experts’ judgments as well as literature are outlined in table 4. In total, 17 product attributes and 16 psychological consequences were defined. In order to determine a set of terminal values, all 53 values from the Schwartz List of Values (Schwartz 1992) were assessed due to feasibility reasons 15 values that represent all ten value domains were adjusted to the context of meat substitutes.
Table 4: Means-end Chain Factors; Source: Authors’ own illustration.
Abbildung in dieser Leseprobe nicht enthalten
[...]
1 Note that the group “no meat consumers” comprises all vegetarians and vegans.
-
-
-
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X.