The Impact of Experience Attributes and Search Attributes in a O li e Custo er Re ie o the Reader’s Food Prefere ces
MASTER THESIS
Master of Science in Marketing, specialization Marketing Management
University of Groningen, Faculty of Economics and Business, the Netherlands
Author: Philipp Bergmann
1First supervisor: Dr. J.A. Voerman
2Second supervisor: Dr. D. Trampe
3Word count: 21.614
Date: January 12
th2015
1
Philipp Bergmann is a Marketing MSc student at the Faculty of Economics and Business, University of Groningen, The Netherlands. The research forms part of the Master of Science in Marketing, specialization Marketing Management, study program. Address for correspondence: Philipp Bergmann, Klarastrasse 4, 80636 Munich, Tel.:
+49 89 158 90 596; E-mail: Philipp-bergmann@mail.de; Student number: s2639084
2
Dr. Liane Voerman is Assistant Professor in the fields of Business and Marketing, Faculty of Economics and Business, University of Groningen, The Netherlands
3
Dr. Debra Trampe is Assistant Professor in the field of Marketing, Faculty of Economics and Business, University
of Groningen, The Netherlands
To precious hours of my life .
Which silently passed away while I was writing on this paper.
Loosing you was a frustrating, tiring and exhausting experience.
But I know I will see you again.
TABLE OF CONTENTS
ABSTRACT... I ACKNOWLEDGEMENT ………. I 1 INTRODUCTION ………
1.1 E perie e a d sear h attri utes a d their role i isleadi g advertisi g 1
1.2 Online customer reviews: a vital source of information 2
1.3 Online customer reviews and their role in experience and search products 3
1.4 Online customer reviews and their role in food purchase decision 5
1.5 Research question 7
1.6 Relevance and uniqueness of the thesis 7
1.6.1 The online food market 7
1.6.2 The age of the customer as an influencing factor 8
2 THEORY ……… 2.1 Experience and search attributes as the basis for opinion shaping 10
2.2 Food products 12
2.3 The type of product and the need for justification 13
2.3.1 Hedonic versus utilitarian food products 13
2.3.2 Hedonic food products and their need for justification 13
2.4 Age leading to emotional capacity and emotional skepticism 14
2.4.1 Age leading to emotional capacity 15
2.4.2 Age leading to emotional skepticism 16
2.4.3 Emotional capacity versus emotional skepticism 17
2.5 Overview 17
2.6 Conceptual model 19
3 METHODOLOGY ………... 3.1 Research design 20
3.2 Participants and descriptives 20
3.3 Procedure 21
3.4 Detailed description of all variables 24
3.4.1 Dependent variables 26
3.4.2 Experimental variable / manipulation check 29
3.4.3 Moderating variables 29
3.5 Results of the manipulation check 30
3.6 Preparation of data 31
3.6.1 Dependent variables 31
3.6.2 Calculating the interaction variables 37
3.7 Plan of analysis 37
3.7.1 Initial analysis 37
3.7.2 Main analysis 37
4 RESULTS………... 40
4.1 Initial analysis 40
4.1.1 Descriptives 40
4.1.2 Visualization 41
4.2 Main analysis 45
5 DISCUSSION………... 50
5.1 Summary and discussion of the findings 50
5.1.1 First hypothesis 52
5.1.2 Second hypothesis 54
5.1.3 Third and fourth hypothesis 56
5.2 Limitations and recommendations for future research 57
5.3 Managerial implications 58
REFERENCES ……… II
APPENDIX ………. XIV
I ABSTRACT
Before making a final purchase decision, customers want to be sure that a produ ct’s features are i deed as good as advertised by the producer. In an online setting, many people therefore use the recommendations of other customers in the form of so- called o li e custo er re ie s” OCRs . This research investigates the impact of OCRs o the reader’s food preferences, drawing on the paradigm of search and experience attributes. It argues that the occurrence of experience attributes in an OCR will lead to a higher explicit liking, implicit wanting and purchase intention. It furthermore claims that this effect is influenced by the type of food product, where hedonic food products additionally benefit from the occurrence of search attributes. The author furthermore suggests that this relationship is moderated by the age of the reader: as people mature, they develop higher levels of emotional skepticism and emotional capacity, which will ultimately positively as well as negatively impact the influence of experience and search attributes i OCRs o the reader’s food preference. The findings are as follows: A manipulation of OCRs towards experience attributes has a sig ifica t positi e i pact o the reader’s purchase intention.
Whereas the type of food product has no impact, the level of emotional skepticism and therefore the age of the reader does: People who are older show a higher emotional skepticism, which ultimately hampers their purchase intention.
ACKNOWLEDGEMENT
I would like to thank my supervisor Liane Voerman for her dedication in supporting me with my research.
Her availa ilit at a ti e , a i a le support a d vital ideas at a stage ere of ig help for
progress.
1 1 INTRODUCTION
1.1 Experience and search attributes a d their role i isleadi g advertisi g
On average, an adult is exposed to roughly 600 ads per day. This means that 600 times per day companies want to convince their potential future customers of the advantages of their products (American Mathematical Society 2007). In doing so, firms can base the formulation of their advertising claims on three different attributes: search attributes, experience attributes and credence attributes.
Search attributes comprise all those features of a product that can be verified and determined by the customers prior to the purchase (Klein 1998), whereas experience attributes can only be verified after the purchase (Klein 1998). Credence attributes can never be verified, neither before the purchase, nor during the purchase nor after the purchase. (Darby and Karni 1973)
This shows that for some product attributes, consumers lack vital information about their true nature, which creates a situation of incomplete information. This situation might pay for the producers:
They might lie about the true features of their product, exaggerate their advertising claims (Darby 1973) and thus create isleadi g ad e tisi g” Nelson 1974).
How can customers therefore verify that the company is not fooling them and that the facts promised are true? The price advertised is quickly verified: a visit to the store will show whether e.g. a chewing gum is priced as advertised and whether this is a good deal or not. Zeithaml (1981) and Nelson (1974) thus define the price as a search attribute, because it can be verified rather easily prior to purchase and at low time and money cost.
This verification is on the other hand more difficult for other attributes of the chewing gum such as the taste. The customer will not be able to determine the taste by simply looking at the information provided. He or she could search for information about the taste, e.g. in magazines, but this search would most likely become rather expensive time and money wise. Less expensive and time consuming is a direct sampling of the product (Nelson 1970). The taste of a chewing gum is best evaluated when purchased and tried. Consequently, taste, quality, flavor and color of a chewing gum are determined as experience attributes (Zeithaml 1981; Nelson 1970) because they are based on a subjective perception, cannot be verified easily prior to purchase and thus need to be experienced ”.
But what about the claimed fact that the chewing gum is capable of reducing the danger of tooth decay
or plaque? Even after the purchase of the product, the customer will not be able to fully relate his or her
assumed good mouth health to the chewing gum. He or she might assume that eating one chewing gum
2
a day helps to reduce the dangers of caries and plaque, but only a laboratory test at the dentist might prove right or wrong. Consequently, the caries and plaque reduction claim of the producer is defined as a credence attribute (Darby and Karni 1973).
This example shows that products can have search attributes, experience attributes or credence attributes, a mixture or even all of them at the same time (Srinivasan and Till 2002). To avoid situations where for every single product a purchase (to verify experience attributes) or a walk to the store (to verify search attributes) is needed, customers continuously seek for information about the product in order to ease their purchase decision through other ways (Nelson 1970). Nelson (1970) suggests the possibility of guided sa pli g”. He poi ts out that usto e s should i deed sa ple p odu ts, ut i a o e effi ie t a , hi h ea s ot at a do ”, ut athe ased o i fo atio . This i fo atio ” ould for instance be provided in the form of word-of-mouth (WOM), which is described as a i fo al ad i e passed et ee o su e s” (East, Hammond and Lomax 2008). WOM can be derived from consumer magazines or friends and relatives providing suggestions which product to purchase and which to avoid (Nelson 1970). This is highly effective since WOM lacks commercial bias (East, Hammond and Lomax 2008). Consequently, the expected utility derived from a sample suggested by friends is greater than the expected utility of a random sample (Nelson 1970).
In this way, consumers can use the opinions of others to determine whether the performance of the product is reliable (Childers and Rao 1992), whether a product is sold at a reasonable price or whether its usage is socially accepted (Cialdini and Goldstein 2004). This social influence is consequently highlighted by literature as a major influencing factor in the consumers consumption decisions. WOM as expressed through other consumers is deemed to be the most effective marketing tool (e.g. Arndt 1967;
Trusov, Bucklin and Pauwels 2009).
1.2 Online customer reviews: a vital source of information
WOM and social influence are not limited to the offline world, but can also be found in the online
world. One way of gaining insights into the experience of othe s as a asis fo guided sa pli g” a e
independent online customer reviews, labelled as OCRs in the following . OC‘s a e a positi e o
negative statement made by potential, actual, or former customers about a product or company, which
is made available to a ultitude of people a d i stitutio s ia the I te et” He ig-Thurau et al. 2004,
p.39). Klein (1998) highlights that this way of information transfer makes it possible to transform
experience attributes into search attributes, which then can be evaluated more easily by the customer.
3
The latter is thus able to experience the product similar to a true simulated product experience, which Klein (1998) alls i tual e pe ie e”.
OCRs as a tool for expressing electronic WOM is a highly powerful way for opinion shaping. Every fou th usto e fi st s uti izes othe usto e s opi io s a out a product before purchasing it (comScore 2010; Wagner and Wiehenbrauk 2014). Freedman (2011) even claims that more than 90 % of American Internet users read OCRs. Customers know that reviewers communicate their opinion voluntarily and thus deem OCRs as an authentic and objective sharing of opinions, contrary to classic advertising which is quickly debunked by customers as a source of deliberate influence (Decker and Trusov 2010; Moore 2012). Zhu and Zhang (2010) consequently see OCRs as a reliable representation of the overall WOM. (Zhu and Zhang, 2010).
Much research has outlined the influence of OCRs on consumers purchase decisions: While Sridhar and Srinivasan (2012), East, Hammond and Lomax (2008) and Floyd et al. (2014) prove that the online opinion of others indeed has a sig ifi a t i flue e o a usto e s pu hase de isions, other studies go more into the details of OCRs: Schlosser (2011) shows that the extent of ratings-text congruence positively influences the impact on the readers purchase behavior, while Sen and Lerman (2007) highlight valence as an influencing factor. Forman, Ghose and Wiesenfeld (2008) furthermore talk about the importance of the source of review and Jimenez and Mendoza (2013) prove the level of detail of the OCR to be an important influencing factor.
1.3 Online customer reviews and their role in experience and search products
Some authors also included a differentiation between search products and experience products by providing evidence that readers react differently to OCRs about search products than to OCRs about experience products.
Huang, Lurie and Mitra (2009) for instance state that the presence of OCRs has a greater effect on experience goods than on search goods. Mudambi (2010) is more precise and shows that experience products benefit from moderate reviews whereas search products from extensive reviews. Jimenez and Mendoza (2013) furthermore state that experience products as such involve higher purchase uncertainty and thus require more pre-purchase information, for instance through OCRs, than search products.
However the current research on search products and experience products only looks at the
product as a whole, defining it as eithe e ti el sea h” o e ti el e pe ie e”, lea i g o spa e fo
4
anything in between. Consequently, this only allows for two kinds of OCRs: OCRs about experience products and OCRs about search products. This narrow categorization leaves a question mark on whether a more fine-grained categorization of products might lead to different results. After all, research agrees on a categorization of products according to a continuum scale, where a search product always also contains experience attributes and vice versa (Huang, Lurie and Mitra 2009; Klein 1998; Mudambi 2010;
Nelson 1981). Furthermore, the findings of Mudambi (2010) and Jimenez and Mendoza (2013) apply to high price products that are intangible or normally not purchased every day. This limited scope of products leaves a question mark on whether and to what extent the mentioned findings can be transferred to other, more common product categories.
This research tries to find an answer to the mentioned two questions marks. It suggests to allow for a more fine-grained definition of products and thus of OCRs by making use of the possibilities that lie within the categorization of products according to the mentioned continuum scale. This allows to define products not only as either experience or search, but rather as a mixture of both (Huang, Lurie and Mitra 2009; Klein 1998; Mudambi 2010; Nelson 1981). Consequently, in this updated approach, there are not only two kinds of OCRs, but many more. Considering the example of a chocolate bar, an OCR could first exclusively address the experience attributes (taste), second exclusively the search attributes (price) or third even both. The last case allows for seemingly endless further differentiations: OCRs could be lassified e.g. as % sea h, % e pe ie e”, o % sea h, % e pe ie e”, depe di g o hi h attributes they specifically address. Due to this endless scope of new possibilities, these new OCRs will be efe ed to as fle i le OC‘s” i this pape . OC‘s that a e ased o the assu ption that products are eithe sea h o e pe ie e, ut othi g i et ee di hoto ous s ale , a e efe ed to as i fle i le OC‘s” i this pape .
To put it i a utshell: this pape fo uses o fle i le OC‘s” a d thus the fa t that experience attributes, search attributes or even both can be addressed within one and the same OCR.
The differentiation developed by Nelson (1974) originally contained only experience attributes and search attributes, while credence attributes were added later by Darby and Kerni (1973). Since then, research however has focused on the first two attributes, while credence attributes only played a side role that is mostly analyzed separately (Wright and Lynch 1995; Klein 1998; Jimenez and Mendoza 2013).
This paper sticks to this trend and only focuses on the differentiation between experience attributes and
search attributes. The thi d p odu t-des i i g” ha a te isti credence attributes is thus dropped in this
paper.
5 1.4 Online customer reviews and their role in food purchase decisions
The present research suggests to expand the product range by analyzing the impact of experience attributes and search attributes in OCRs to the category of food products. Food products are very common products that show a high purchase frequency. Battistoni (2012) states that on average more than ten percent of income is regularly spent on food. Furthermore, the purchase of food products is highly dependent on social influences, which makes them a perfect subject for the analysis of OCRs. Katz and Lazarsfeld (1955) demonstrate that WOM and thus the social environment is one of the most influential factors for the purchase of food products. Salmon and Fennis (2014) furthermore prove that depending on the choice and opinion of others people tend to eat healthy or unhealthy.
Past research has used many different dependent variables for measuring the impact of OCRs on the reader (credibility of the OCR, Jimenez and Mendoza 2013; claim recognition, Wright and Lynch 1995;
perceived helpfulness, Mudambi 2010). Since this research focuses on an ingestible product category food preference will serve as dependent variable here. Berridge (1996) shows that two different brain components mediate food reward: a conscious one and an unconscious one. The conscious one describes the hedonic/ affective part that develops out of central processing. It is often referred to as explicit liking.
The unconscious one is the salience/ motivation part that results out of underlying, implicit and automatic behavior, where purchases are done out of a directed impulse. It is therefore named implicit wanting (Finlayson, King and Blundell 2006; Griffieon-Roose et al. 2011). Thus two different components will serve as a measurement for food preference: explicit liking and implicit wanting. Following the reasoning, that at the end of a the purchase funnel a purchase intention is to be found (Kotler 2012), the dependent variable purchase intention will furthermore be included into the definition of food preference.
For a better understanding, two additional remarks need to be made. First, in this research food products will be segmented into hedonic food products (bought for multi-sensory pleasure, e.g. chocolate or wine, Hirschman and Holbrook 1982) and utilitarian food products (bought to fulfill a functional goal, to feel less hungry, e.g. cooking oil, butter, milk, Dhar and Wertenbroch 2000). Past research has proven this to be a meaningful classification that produces significantly different results: hedonic (food) products are preferred when advertised with health/ nutrition claims (Kim, Cheong and Zheng 2009) and in situatio s i hi h the usto e s ognitive capacity is limited (Antonides and Cramer 2013). Utilitarian (food) products on the other hand require taste claims to raise attention (Kim, Cheong and Zheng 2009) and usually receive a higher share of wallet than hedonic (food) products (Antonides and Cramer 2013).
It is therefore only logical to add to this differentiation and make their subjective marketing tactics even
6
more fine-grained. Furthermore, there is a strong correlation in the definitions of hedonic and experience products and in the definition of experience and search products. Hedonic consumption relates to e pe ie i g” the p odu t i a ultise so ” a . That means that consumers of a hedonic product derive enjoyment in ways of tasting, smelling, feeling, fantasizing and thus addressing multiple senses (Hirschmann and Holbrook 1982). It thus shows a close connection to experience attributes, which relate, as explained before, to all those facets of a product experience that cannot be verified prior to purchase – such as taste of a chocolate bar, the feeling/ softness of a fresh tomato or the taste of a wine. Utilitarian e jo e t is de i ed f o the fu tio i g” of a p odu t a d the i st u e tal a d p a ti al e efits of its consumption (Hirschmann and Holbrook 1982; Yim et al. 2014). It is goal oriented and products are bought to serve a specific reason. When a customer is for instance on a diet and thus wants to be sure that the milk purchased is indeed low-fat, he or she will certainly refer to the calorie content as printed on the package – a classical search attribute” as the calorie count just as any other ingredient can easily be verified prior to purchase. These similarities will be used for the development of the hypotheses.
Therefore, this research has two different independent variables: Attributes in an OCR” experience attributes, search attributes, experience and search attributes) and (2) type of food product (hedonic food products and utilitarian food products).
Second, as most Western countries are nowadays facing an increasing ageing of their population
(Grey 2007), it is important to adapt modern media to the ageing customers. Literature provides broad
evidence for people reacting differently to social influences as their age increases. On the one hand, more
mature people show a higher focus and a better understanding of emotions (emotional capacity). On the
other hand, they seem to be more skeptical towards social influences (emotional skepticism). These
opposing observations could either lead to people rejecting experience attributes in an OCR as they
mature or people preferring experience attributes in an OCR. Consequently, the age and thus the two
variables emotional skepticism and emotional capacity of the reader will be included as two important
factors possibly moderating the relationship of att i utes i a OC‘” and the preference of a food
product.
7 1.5 Research question
This paper tries to find an answer to the following research question:
How does the occurrence of key words referring to experience attributes or search attributes in an OCR i flue ce a custo er’s prefere ce of a specific hedo ic or utilitaria food product a d how is this impact
oderated the reader’s age and thus level of emotional skepticism and/or emotional capacity?
Specifically, this research tries to find answers to these sub-questions:
o What are the differences between experience attributes and search attributes?
o Ho does a a ipulatio of these attri utes i OCRs i flue ce the reader’s food preference?
o What role does the type of a food product pla i the i pact of OCRs o the reader’s food preference?
o To hat e te t does the reader’s age and thus emotional skepticism as well as emotional capacity oderate the i pact of OCRs o the reader’s food preference?
1.6 Relevance and uniqueness of the thesis
This paper adds to the current research on search and experience products in two major points.
On the one hand, it offers many more examples on how to classify OCRs. By introducing the concept of fle i le OC‘s”, it a gues that it is possi le to add ess experience attributes, search attributes or even both in one and the same OCR. This paper therefore allows to gain a considerably more detailed view on the i pa t of OC‘s o the i di idual s food p efe e es – overall and by differentiating between hedonic and utilitarian products. On the other hand, the present paper broadens the product range towards food products and thus makes the results more useful for the growing e-food business.
1.6.1 The online food market. This research contributes to the growing insights about the impact
of the Internet o a usto e s o su ptio behavior. More specifically, by focusing on the category of
food p odu ts, it a al zes ho the opi io of othe s i flue es the i di idual usto e s food p efe e es
in an online environment. This is of great importance nowadays, as the food business is currently
experiencing major shifts and developments: Whereas the hassle-free online purchase of books, furniture,
clothes and even cars is taken for granted today, the trading of food products still mainly takes place in
8
offline brick-and-mortar stores (Lebensmitelzeitung.net 2013). However, this situation is rapidly changing, as the food segment is finding its way into the online business at an extremely fast pace. The consulting company Ernst & Young expects the market share of the pure online food retailing segment in Germany to skyrocket from currently 3 percent to 20 percent by 2020 and cross-channel food retailers to raise their share from currently 6 percent to 20 percent by 2020. British consumers are already one step ahead: here 5 percent of consumers use online food retail services (Wagner and Wiehenbrauk 2014). Also the Austrian online food retail business is quickly expanding: researchers expect a market share growth of more than 100 percent from 2006 to 2016 (Warschun, Krüger and Vogelpohl 2013).
The online food business can therefore be seen as a prosperous business sector that does not only deserve, but urgently needs some further research. However, in spite of the mentioned developments, research on food products traded in the online environment is scarce. Most research measures the behavior of shoppers in an online environment with various products, of which food products randomly happen to be included. (Degeratu, Rangaswamy and Wu 2000 . But to the autho s best knowledge, no study has specifically focused on food products and their behavior in an online environment yet. This research will therefore close an open gap and hopefully lay ground for future research on an interesting and future-oriented topic.
1.6.2 The age of the customer as an influencing factor More mature users became an important
Internet user segment that is growing at an extremely fast pace. While in April 2000, 45 percent of the
American 50 – 64 old adults used the Internet, this number increased 12 years later to 77 percent (Zickuhr
and Madden 2012). The amount of the online 65+ American adults showed an even steeper increase: In
April 2000, only 15 percent reported to regularly use the Internet, whereas in April 2012 more than the
half of all American adults aged 65 or older did so and three quarter of them even on a daily basis (Zickuhr
and Madden 2012). But not only on a quantitative basis does the older generation appear as an important
Internet user segment, also the qualitative side is interesting. As people mature, they show certain specific
Internet usage behaviors, as they use it especially for buying products (Wan, Nagayama and Sutcliffe
2010). This is interesting as over 90 percent of American adults above the age of 65 prefer to live
independently (US Census Bureau 2000) and stay in their own home as long as possible (American
Association of Retired Persons . Te h olog pla s a i po ta t ole he e as it o t i utes to agi g
i pla e” a d suppo ts them in any home-based task (Mitzner et al. 2010). The older generation is willing
to learn and use new technologies as long as they help them to remain independent. Consequently, the
older generation can be seen as a prosperous sector for online retailing, as a delivery of food right to their
9
doorstep will significantly contribute to their personal well-being and thus their ability to stay independent. As a consequence, it is important to take a lose look at the usto e s ha gi g behavior towards OCRs as their age increases. The author suggests that people react differently to the influence of OCRs as their age increases due to two characteristical traits: higher emotional capacity and higher emotional skepticism.
10 2 THEORY
In chapter 1, the author gave a general introduction into the topic of OCRs as well as the differentiation between experience attributes, search attributes and credence attributes. He explained that OCRs have a significant impact on readers ’ purchase decisions and presented a research question.
Chapter 2 will go more into detail and explain how experience attributes and search attributes are able to differently influence the reader, what role the type of food product plays and how the age of the reader might influence the impact of OCRs on the variable food preference.
2.1 Experience attributes and search attributes as the basis for opinion shaping
Search products are dominated by attributes that can be verified prior to purchase (Klein 1998;
Nelson 1974). Examples include the price or the package of a product (Zeithaml 1981). Experience products on the other hand are dominated by attributes that can only be verified after the purchase and/or consumption. No o e ill e totall su e hethe a ho olate is eall as ea ” as stated i the commercials, if he or she has never tried it (Klein 1998; Nelson 1974; Zeithaml 1981).
Since experience products therefore create higher pre-purchase information needs (Ford 1990), customers look at sou es of i fo atio a out e pe ie e p odu ts ith diffe e t e es” tha at sources of information about search products. Literature shows that customers react differently to OCRs, depending on whether these talk about a search product or about an experience product. These differences can be subsumed under three bullet points: perceived helpfulness, credibility and their benefi ts f o the i te et, efe ed to as digitaliza ilit ” i the follo i g (Mudambi 2010; Jimenez and Mendoza 2013; Klein 1998).
Mudambi (2010) defines perceived helpfulness of a OC‘ as the diag osti alue” fo the customer in any stage of the purchase decision process. He argues that for experience products, customers prefer moderate ratings to extreme ratings whereas for search products, this so called review extremity has no impact on the perceived helpfulness of an OCR. Review depth on the other hand is important for both search and experience products. OCRs of both product types are perceived as more useful if the word count is high. This effect nevertheless is stronger for search products than for experience products.
The credibility of an OCR is a further factor where search and experience products differ. In the
case of search products, customers assess the credibility of an OCR with the level of detail provided,
11
whereas in the case of experience products, they pay attention to the level of agreement. Thus only experience products ill e efit f o a state e t like user fi d this e ie helpful”. Jimenez and Mendoza 2013).
Apart from that, search and experience products differently benefit from the Internet and thus show a differe t deg ee of digitaliza ilit ”. Klein (1998) argues that in the case of a search product, the ease of access to information, the low cost of that information and the customizable format of that i fo atio leads to a edu ed ost of thi ki g”. I the ase of experience products, the information p o ided i OC‘s a help the eade as e tio ed efo e to i tuall e pe ie e” the p odu t, hi h should significantly lower the risk of purchase.
These examples show that customers indeed recognize the differences between search and experience products in OCRs and then react differently. However, these examples all remain within the ou da ies of i fle i le OC‘s”, he e a OC‘ addresses either the entire search product or the entire experience product. As explained before, Nelson (1974) originally classified products according to a dichotomous scale, where products either belong to the search or to the experience side, based on their dominating attributes. Since then the argumentation however developed towards describing products as a u dle of att i utes”, where a product combines search, experience (and credence) attributes at the same time (Hong, Chen and Hitt 2013). Customers then choose a product according to the importance they attach to each attribute. Considering the aforementioned example, customers can therefore buy a chewing gum due to its taste (experience), its price (search) or its healthiness (credence). Consequently, as stated in the introduction, for one and the same product, OCRs can address experience attributes, search attributes o e e oth. These OC‘s e e des i ed as fle i le OC‘s” efo e.
Therefore the question arises whether customers also react differently to an OCR about one and
the sa e p odu t, depe di g o hethe this fle i le OC‘” add esses only experience attributes, only
search attributes or even both. In fact, literature provides evidence for customers being well aware of
these different attributes and thus reacting differently to any text that addresses them separately. Ford,
Smith and Swasy (1990) state that consumers are more skeptical of claims addressing experience
attributes than of claims addressing search attributes of one and the same product. This is further
supported by the findings of Wright and Lynch (1995) who show that experience claims appear less
believable and less recallable than search claims. Although these findings remain within the boundaries
of subjective advertising, they provide evidence that similar findings might be observed in the case of
OCRs.
12
To put it in a nutshell: The examples presented before show that customers indeed react differently to information in OCRs, depending on their search or experience nature. Mudambi (2010), Jimenez and Mendoza (2013) and Klein (1998) prove differences within the framework of i fle i le OC‘s”. Taki g a o e fi e-grained point of view and allowing to split products into experience attributes and search attributes, Wright and Lynch (1995) and Ford, Smith and Swasy (1990) prove different reactions in the case of subjective advertising and thus set the ground for possible similar findings within the ou da ies of fle i le OC‘s”.
2.2 Food products
This study will be conducted within the category of food products. Taking a look into research, it becomes clear that the majority of authors sticking to the dichotomous scale have a clear opinion regarding the categorization of food products: Wine (Klein 1998, Senecal and Nantel 2004), beer (Clemons, Gao and Hitt 2006), restaurant meals (Zeithaml 1981) and any other food product (Nelson 1970, 1974) are all perceived as experience products. This means that food products create a lot of purchase uncertainty and thus require a lot of pre-purchase information.
Transferring this point of view to the continuum approach allows making the following statement:
Like any other product, food products have search attributes (price, package, design) as well as experience attributes (taste, smell, softness). The experience attributes however seem to be the dominant ones.
Therefore, the following two assumptions can be made. On the one hand, since they can be regarded as a reliable source of information (Decker and Trusov 2010; Moore 2012), OCRs should be able to reduce this mentioned purchase uncertainty, if they specifically address these experience attributes (taste, smell, softness). On the other hand, addressing experience attributes – a d thus the food p odu ts do i a t attributes - should activate feelings of food reward and consequently lead to both higher explicit liking as well as implicit wanting.
Consequently, the author suggests the following hypothesis:
H1: OCRs that address experience attributes will lead to a higher food preference than OCRs that address
search attributes.
13 2.3 The type of product and the need for justification
2.3.1 Hedonic versus utilitarian food products. Food products can be defined in many different ways. They can be classified according to their most obvious characteristics, for example their plain ingredients (low fat versus high fat) or their taste (savory versus sweet), whereas other authors suggest a differentiation according to their impact on health (vices or wants versus virtues or shoulds) (van Doorn and Verhoef 2011).
This research however focuses on a differentiation between hedonic food products and utilitarian food products. This classification assumes that food products are bought for two different goals: to either provide pure enjoyment and multi-sensory pleasure (hedonic) or to fulfill a functional goal, e.g. to feel less hungry (utilitarian) (Hirschmann and Holbrook 1982; Yim et al. 2014).
2.3.2 Hedonic food products and their need for justification. One classification that is closely connected to the hedonic/ utilitarian scale is the virtue/ vice categorization (van Doorn and Verhoef 2011).
Literature categorizes utilitarian products as virtue products, which means that although their consumption is not gratifying, their long term health effect is very positive. Hedonic products are described as vice products, which implies that their consumption is very pleasant, but their long-term health effect detrimental (Ratneshwar and Mick 2005).
Based on this interrelation, the author argues that consumers of hedo i p odu ts” process feelings of guilt after having purchased or consumed them. This creates a situation where the consumer needs to find reasons to justify the purchase and consumption to themselves. The author suggests that search attributes provide a perfect basis for this mentioned justification. He consequently claims that the occurrence of search attributes in OCRs will lead to a positive impact on consumers of hedonic (food) products and thus to a higher food preference.
This will be explained more in detail with the following example: A Chocolate bar is usually bought
for pure enjoyment and multi-sensory pleasure. Fletcher et al. (2006) describe a chocolate bar as a
comfort food ” and Wansink, van Ittersum and Painter (2003, p. 739) further highlight that its
o su ptio e okes a ps hologi all o fo ta le a d pleasu a le state fo the pe so ”. This allo s
classifying chocolate as hedonic food product. On the other hand, chocolate is associated with a bad
image. It has high sugar content and is consequently declared a forbidden product (Knight and Boland
1989). Chocolate can therefore be described as a vice food product. Due to this long-term negative health
14
effect, consumers of chocolate as well as consumers of any other hedonic food product will be faced with a sense of guilt (Kivetz and Simonson 2002; Strahilevitz and Myers 1998). In order to cope with these feelings, consumer of hedonic food products need to find ways of justifying the consumption to themselves (Shafir, Simonson, and Tversky 1993).
One way to do so, is to transform the benefits of a chocolate bar into quantifiable facts, as quantifiable facts can more easily be justified (Hsee 1996). Examples of quantifiable benefits of the purchase and consumption of a ho olate a a e a lo p i e this sa ed e e ts!” , a e pa kage desig this is so p a ti al to a , I ill eed less ti e to alk ho e!” o e tai i g edie ts appa e tl ho olate akes ou feel happ . I eed that!” . All these examples share the fact that they address search attributes (Zeithaml 1981). Consequently, it can be stated that search attributes provide a perfect basis of justification.
Therefore, the following hypothesis can be made:
H2: In the case of hedonic (versus utilitarian) food products, search attributes will lead to a higher food preference (versus no change in the food preference).
2.4 Age leading to emotional capacity and emotional skepticism
The effect of experience attributes and search attributes in an OCR o the eade s food preference is expected to be moderated by the age of the reader and thus by two further variables:
emotional skepticism and emotional capacity. Literature provides broad evidence that as people mature, their processing of information from texts such as OCRs changes. This evidence can be used to develop two competing hypotheses:
(1) On the one hand, as age increases, people seem to focus more on emotions and therefore seem to be more attracted by emotional OCRs. This assumed fact will be referred to as emotional capacity in the following.
(2) On the other hand, more mature people seem to have a higher skepticism towards emotions.
They are therefore more likely to reject OCRs. This trait will be named emotional skepticism in
this paper.
15
2.4.1 Age leading to emotional capacity. As people become older, they seem to have a higher focus on emotions. This assumption is based on the following two facts: First, research has shown that the processing of information changes as people mature. Older people have a better understanding of emotions and want to combine emotions and cognitions (LaBouvie-Vief 1998). They thus concentrate more on emotions (Carstensens, Isaacowitz and Charles 1999), in contrast to young adults who focus more on factual information (Hashtroudi et al. 1994). This differentiation between emotional and factual information can also be transferred to experience attributes and search attributes. The examples of experience attributes brought forward by Zeithaml (1981) – taste, wearability and purchase satisfaction – all show a clear connection to emotions and feelings. Color, style, price, feel, hardness and smell on the other hand as examples for search attributes all bear a factual content that can easily be verified prior to purchase (Zeithaml 1981). The author consequently argues that people will be more influenced by experience attributes as their age increases.
Second, and this works in support of the first statement, more mature people have a higher focus on emotions due to their perceived future life span. As people become older, their focus and goals change.
According to Carstensens, Isaacowitz and Charles (1999), people either devote their lifetime to acquiring new information or balancing their emotional states and maximizing their emotional satisfaction.
Carstensens, Isaacowitz and Charles (1999) showed that when people deem time as limited (in this case life time) they tend to focus more on the presence. People want to enjoy emotions and feelings of the o ” as u h as possi le a d thus a oid the e hausti e p o essi g of futu e pla s. Thei fo us is o e present and less future oriented, more towards achieving emotionally meaningful goals and getting the
a i u out of the o ” due to thei pe ei ed sho t life ti e left.
This leads to the assumption that OCRs addressing experience attributes will lead to a higher food preference as the age of the reader increases.
H3: As people mature, they develop a higher level of emotional capacity.
H4: Experience attributes in an OCR will lead to a higher food preference in the case of more mature
readers due to their increased focus on emotions.
16
2.4.2 Age leading to emotional skepticism. On the other hand, several authors provide evidence of the exact opposite, showing that the occurrence of experience, and thus emotional attributes, in OCRs lead to a higher skepticism for more mature people and ultimately a lower food preference. This assumption is explained in the following.
Although the abilities of dealing with online retailing might be less advanced as people mature, older people have one decisive advantage to the younger generation: Their treasure-like badge of shopping experiences. During their lifetime, older people most likely made many different purchases, which they can then relate to the shopping decisions in an online environment (Drolet, Williams and Lau- Gesk 2006; Williams, Patti and Drolet year, Aimee 2005). Research has shown that they use these experiences and thus rely more on personal experience and tastes when processing information (Williams and Drolet 2005). This behavior can therefore be translated to a lower product related risk ” for older people. Bhatnagar, Misra and Rao (2000) define product related risk ” to e p ese t i situations, where the product pleases the ego-needs of the purchaser or was sold based on its feel or touch. This certainly applies to food products, which are p u hased fo the ego- eeds” and mostly based on the feel or taste.
Since older people have a higher purchase experience and can relate products to their own experiences, they have a lower product-related risk ”. Wan, Nakayam and Sutcliffe (2010) consequently argue that older people rate less products as credence, or, as added by the author, experience. Thus, as people mature, more products are regarded as having more search attributes and less experience attributes, simply due to the fact that these older people can relate their own experience to the evaluation of the product. Thereby they can verify the experience attributes prior to purchase and turn them into search attributes. Consequently, an OCR that addresses experience attributes should be of lower benefit for older people, as this OCR does not contain any new information for them.
This leads to the following hypotheses:
H5: As people mature, they develop a higher level of emotional skepticism.
H6: Experience attributes in an OCR will lead to a lower food preference in the case of more mature readers
due to their increased skepticism towards emotions.
17
2.4.3 Emotional capacity versus emotional skepticism. H4 and H6 therefore argue in two different directions: H4 speaks for a positive effect of the occurrence of experience attributes in OCRs whereas H6 supports a negative effect onto the dependent variable food preference. While both are embedded into extensive literature research, no information can be given which one prevails over the other.
Furthermore, these two effects are not mutually exclusive: H4 can still be true even if H6 is found to be true as well. This would lead to a new situation: Assuming that both emotional capacity and emotional skepticism are equally strong, they would cancel each other out.
2.5 Overview
The author expects that, for one and the same product, an OCR has a different impact on the
reader, depending on whether this OCR addresses (1) experience attributes, (2) search attributes or (3)
both experience and search attributes. Based on the hypotheses, the following table provides an overview
about the expected outcomes:
18 TABLE 1:
Overview Of Expected Results
Type of food product
Attributes in an OCR Expected impact on food preference
Reasons
Hedonic food products
Experience and Search High Combining experience attributes (reduced purchase uncertainty!) with search attributes (reduced feelings of guilt!) = Highest expected food preference
Experience Search
Medium Purchase uncertainty and feelings of guilt both need to be reduced. No information about which feeling is stronger. Consequence:
Impact expected to be the same.
Low -
Utilitarian food products
Experience High Highest reduction of purchase
uncertainty
Experience and Search Medium Medium reduction of purchase uncertainty
Search Low No reduction of purchase
uncertainty.
No need to reduce feelings of
guilt.
19 2.6 Conceptual model
Combining the six mentioned hypotheses leads to the following conceptual model:
FIGURE 1 Conceptual Model
emotional skepticism
INDEPENDENT VARIABLE ATTRIBUTES IN AN OCR Search
Experience
Search & Experience
DEPENDENT VARIABLE FOOD PREFERENCE Explicit Liking Purchase Intention Implicit wanting
EXPERIMENTAL VARIABLE TYPE OF FOOD PRODUCT Hedonic
Utilitarian
H6 H4
H5 H3
emotional capacity
H1
H2
MODERATORING VARIABLE
AGE
20 3 METHODOLOGY
Chapter 3 elaborates on the research design, the participants and the procedure. It furthermore provides information about the scales used and gives a detailed explanation about how each of these was used.
3.1 Research design
A conclusive research was carried out to collect primary, quantitative data for testing the six hypotheses. The experiment can be described as a 2 (hedonic / utilitarian) x 3 (search / experience / search and experience) project, that allowed participants to take part in only one condition (between design).
Therefore, within a descriptive research framework, a single cross-sectional design was chosen: A random sample of participants was drawn, which were then be exposed only once to the research design.
The su e as o du ted ith the help of the o li e su e tool Qualt i s”. Data as olle ted on a total of 16 days in November 2014. Participants received a link, which lead them, by choice, either to the English or the Ge a e sio of the su e . The soft a e of Qualt i s” the a do l assig ed participants to one of the six conditions.
3.2 Participants and descriptives
Participants for this experiment were carefully chosen. Since this project focuses on the impact of
OCRs in the environment of online food retailers, the author decided to increase data reliability by
focusing on participants that are familiar with online food shopping. Through personal contacts, the
customers of an Irish organic salmon online shop (www.gabriele-bergmann.com) were contacted. Apart
from that, the survey was distributed among marketing professionals on linkedin.com and xing.de groups
and furthermore between students of the fields of international business and marketing. To receive a
decent amount of more mature people to account for the hypothesized increased emotional skepticism
and emotional capacity , f ie ds of the autho s fathe e e o ta ted, which resulted in more than 26
percent of participants older than 45 years. Participants joined the survey deliberately, no incentives were
offered. Participants however had the possibility to express their interest in receiving the final results after
this paper had been finalized.
21 3.3 Procedure
In total, the survey link was accessed 485 times. However, before these 485 participants were faced with the actual stimulus, two filter questions were asked. These questions were designed to ensure reliability of the explicit liking and implicit wanting answers in accordance with Finlayson, King, Blundell (2007). In the first filter question, participants had to a s e to the uestio ho hu g do ou feel ight o ”? on a 10-point- et i s ale, a gi g f o (1) not at all hungry” to (10) extremely hungry”.
To avoid unreliable answers that are based on a very low or a very high feeling of hunger, only those participants that i di ated a hu ge -s o e” of – (9) were used for further analysis. The second filter question asked for the personal taste of the participants. Depending on whether they were assigned to the hedo i o the utilita ia o ditio , the e e asked do ou dislike Milka Alpi e Milk Cho olate”
hedo i o ditio / to atoes utilita ia o ditio ”. All participants with a dislike for that food product were excluded.
These two filters reduced the data set to a total of 263 answers. 20 participants had to be further excluded, due to their time to respond (TTR) being greater than 20 seconds in the questions about the implicit wanting. As the majority of their answers could n ot e e o ded Qualt i s”, pa ti ipa ts were furthermore excluded. After having reduced the answers with the mentioned steps, the final data set comprised 218 valid answers. The following two tables give an overview about the mentioned filtering steps and provide further descriptive information on the final data set:
TABLE 2:
Reduction Steps of Initial Data Set
Step Scale Decisive criterion Sample size
Survey accessed - Saw welcome message 485
Filter I: Feeling of hunger 10-point-metric scale S o e – ” 311
Filter II: Dislike food product Binomial No” 263
implicit wanting: TTR Binomial TTR < 20 sec. 243
Completed surveys - Reached end screen 218
22 TABLE 3:
The Final Data Set
Condition Attributes n Average age
C1 Hedonic product + experience claim 37 41
C2 Hedonic product + search claim 42 34
C3 Hedonic product + experience and search claim 28 35
C4 Utilitarian product + experience claim 38 33
C5 Utilitarian product + search claim 33 38
C6 Utilitarian product + experience and search claim 40 40
Total - 218 37
These 218 answers were then used to measure the impact of attributes in an OCR on the three dependent variables: (1) explicit liking, (3) purchase intention and (2) implicit wanting.
After having read the welcome message, participants received the stimulus: They were asked to imagine that their favorite supermarket recently launched an online store, which they are currently browsing through. To help them with their purchase decision, the store presents online customer reviews.
The participants were then faced with an exemplary OCR of positive valence, which they were asked to carefully read through.
In total, there were six different OCRs, which were manipulated as follows: three OCRs were about a hedo i food p odu t Milka Alpi e Milk Cho olate” a d th ee a out a utilita ia food p odu t (tomatoes). Within each category (hedonic versus utilitarian), OCRs differed according to the attributes of the food product they addressed: either (1) only experience attributes, (2) only search attributes or (3) both experience and search attributes. Taste, quality and consistency represented the (1) experience attributes and price, ingredients and package design the (2) search attributes. The (3) experience and search attributes OCRs contained a mixture of both experience attributes and search attributes. Since the text of the OCRs was invented, it was possible to adopt the o di g to the ha a te isti s” of the p odu t.
e.g. ea a d ho olate ” i the ho olate ase, ut f esh a d fi ” i the to atoes ase .
The following table provides a visualization of the OCRs containing (1) only experience attributes:
23 TABLE 4:
Visualization of OCRs Containing (1) only experience attributes
Type of food product OCR Hedonic
Utilitarian
Immediately after having read the OCRs, participants were faced with several questions.
The first part of these questions dealt with the measurement of the dependent variables:
participants first had to indicate their explicit liking and purchase intention towards the food product. The measuring of the implicit wanting score was somewhat more complicated and therefore done at last.
Therefore, after having read certain instruction, participants received in a third step questions regarding the measuring of the implicit wanting variable.
The second part of the questionnaire started with the manipulation check. This study uses the
hedonic and utilitarian extent of the products as an experimental variable that might possibly have an
interaction effect with the dependent variables. It was therefore important to design a test to verify (or
deny) that the products at question were indeed perceived as either hedonic or as utilitarian. Afterwards,
participants were faced with questions regarding emotional skepticism, questions regarding emotional
24
capacity
1and finally a question regarding their age. The survey ended with a thank you message and a note to send the author an email in case of being interested in the final results.
23.4 Detailed description of all variables
The following table provides an overview about all the questions in the survey:
TABLE 5:
Table of Operationalization
Variable Questions Scale Cro a h’s α
Explicit liking Source:
Finlayson, King, Blundell 2007; Dalton, Blundell, Finlayson 2013
Ho pleasa t ould it e to taste
so e of this p odu t o ?” 10-point metric scale, anchored ith e t e el pleasa t” a d
ot pleasa t at all”
-
Implicit wanting Source:
Finlayson, King, Blundell 2007; Dalton, Blundell, Finlayson, 2013
HEDONIC PRODUCTS
Which of the two products do you most want to eat now? ”
(Measure: Time to response)
Milka Alpi e Milk Cho olate” versus (1) Milka Cho o Cookies”
(2) Ha i o Gold Bea s”
(3) Toffifee”
(4) M&Ms”
(5) Milk Wa C isp ‘olls”
(6) KitKat ” (7) ‘affaelo”
(8) Ki de “u p ise”
UTILTIARIAN PRODUCTS
Which of the two products do you most want to eat now?
Tomatoes versus (1) avocados
Fo ed hoi e” / Time to
reponse ” test
.87
.83
31
Please note: The survey contains two different scales measuring the moderating variable emotional capacity. One however was dropped and therefore not further used in the data analysis. It is consequently not mentioned here.
2
Please refer to the appendix for a copy of the complete survey (the copy shows the survey of condition 1 hedonic food products + experience attributes)
3
Please note: Questions 5 (p=.76), 7 (p=.87) and 8 (p=.64) were not significant according to a paired sample t-test
a d the efo e ot i luded i the C o a h s Alpha al ulatio
25 (2) carrots
(3) paprika (4) cucumber (5) salad² (6) radish (7) zucchini² (8) corn² Purchase intention
Source:
Spears and Singh (2004)
How high is your purchase intention for this food product? ”
(1) I would never / definitely buy this product
(2) I definitely intend/do not intend to buy this product
(3) I have a very low / very high purchase interest for this product
7-point-semantic scale.
4.94
Emotional skepticism Source:
Obermiller and Spangenberg (1998)
Please indicate your rate of agreement with the following statements: ”
(1) We can depend on getting the truth in most customer ratings.
(2) Custo e ati gs ai is to i fo the consumer
(3) I believe customer ratings are informative.
(4) Customer ratings are generally truthful.
(5) Customer ratings are a reliable source of information about the quality and performance of product.
5-point-metric scale, anchored
ith I st o gl disag ee” a d I st o gl ag ee”
5.89
Emotional capacity Source:
Larsen and Diener (1987)
EXPERIENCING POSITIVE FEELING“”
(1) When I am happy, I feel very energetic.
(2) When I am happy, I feel like I am bursting with joy.
(3) When I am feeling well, it is easy for me to go from being in a good mood to being really joyful.
6-point-metric scale, anchored
e e ” a d al a s”
6.82
4
Shortened version (3 items) of the original 5 item purchase intention scale (PI)
5
Adopted e sio of the “kepti is to a ds ad e tisi g” s ale. Adoptio as a ied out i a ha ge of o di g:
Ad e tisi g” as epla ed ith usto e ati g”.
6
