Quantifying effects of convenience and product packaging on consumer preferences and market share of seafood products: The case of oysters

Quantifying effects of convenience and product packaging on consumer preferences and market share of seafood products: The case of oysters

Food Quality and Preference 28 (2013) 492–504 Contents lists available at SciVerse ScienceDirect Food Quality and Preference journal homepage: www.e...

1MB Sizes 1 Downloads 21 Views

Food Quality and Preference 28 (2013) 492–504

Contents lists available at SciVerse ScienceDirect

Food Quality and Preference journal homepage: www.elsevier.com/locate/foodqual

Quantifying effects of convenience and product packaging on consumer preferences and market share of seafood products: The case of oysters Simone Mueller Loose a,b,⇑, Anne Peschel c, Carola Grebitus d a

MAPP Centre for Research on Customer Relations in the Food Sector, Aarhus University, Haslegaardsvej 10, 8210 Aarhus V, Denmark Ehrenberg-Bass Institute for Marketing Science, University of South Australia, Adelaide 5000, SA, Australia c Institute for Food and Resource Economics, Department of Agricultural and Food Market Research, University of Bonn, Germany d Morrison School of Agribusiness and Resource Management, Arizona State University, 7231 E. Sonoran Arroyo Mall, Mesa, AZ 85212, United States b

a r t i c l e

i n f o

Article history: Received 14 June 2012 Received in revised form 29 October 2012 Accepted 8 November 2012 Available online 28 November 2012 Keywords: Visual shelf presentation Market share simulation Discrete choice experiment Retail packaging Health claim Carbon zero claim

a b s t r a c t This study analysed the relative importance of product packaging format and preparation convenience for oysters on consumer choice and market share. Consumer preferences for opened versus unopened oyster preparation formats and the provision of easy-to-prepare accompaniments with or without visual serving suggestions were assessed in a choice experiment. The impact of product packaging and preparation convenience on consumer choice was analysed relative to the traditional demand factors of price, region of origin, oyster species, health, environmental and quality claims. A total of 1718 Australian oyster consumers participated in an online choice experiment with visual product stimuli to simulate their choice of ready-packaged oysters in a retail store. Considering preference heterogeneity respondents’ choices were analysed with a scale adjusted latent class model and six different consumer segments were identified. Market share simulations illustrate the impact of oyster product variations on consumer choice. Overall, price, preparation format and species were the most important choice drivers, followed by region of origin and accompaniments, while packaging format and claims only had a minor influence on consumer choice. Consumer differences in price sensitivity and preferences for species and different oyster accompaniments provide scope for consumer oriented product differentiation with the potential to increase oyster demand and healthy seafood consumption. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction Seafood is considered to be a healthy food choice (Kris-Etherton, Harris, & Appel, 2002; Schmidt, Skou, Christensen, & Dyerberg, 2000) and several countries recommend to consume at least two servings of seafood per week (NHMRC, 2003; USDA, 2010; WHO, 2012). Because these dietary guidelines are not met by the majority of people (Welch et al., 2002), consumer food research has been investigating drivers and perceived barriers of seafood consumption. At the same time producers of aquaculture seafood species such as salmon and oysters aim to better understand consumer preferences to market their products more efficiently (Gempesaw, Bacon, Wessells, & Manalo, 1995; Liu, Kow, Grewal, & FitzGerald, 2006). Among several other factors, previous studies identified perceived inconvenience as substantial barrier to seafood consump⇑ Corresponding author at: MAPP Centre for Research on Customer Relations in the Food Sector, Aarhus University, Haslegaardsvej 10, 8210 Aarhus V, Denmark. Tel.: +45 89486852. E-mail addresses: [email protected], [email protected] (S. Mueller Loose), [email protected] (A. Peschel), [email protected] (C. Grebitus). 0950-3293/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.foodqual.2012.11.004

tion. Consumers either believe seafood preparation to be too time consuming or they lack knowledge thereof (Altintzoglou et al., 2010; Olsen, Scholderer, Brunsø, & Verbeke, 2007; Rortveit & Olsen, 2009). Also, consumers are inexperienced in judging the freshness of seafood products (Pieniak, Verbeke, Scholderer, Brunsø, & Olsen, 2007). These barriers can potentially be overcome by smart retailing and packaging solutions making transport, preparation and storage easier for customers, but their relative importance and impact on market share are so far largely unknown. This study focuses on oysters which constitute a low fat, low cholesterol, high quality protein source containing essential vitamins (FRDC, 2004). Based on these attributes, oysters qualify as healthy seafood product which is desired to be increased in consumption from a public health perspective. They were chosen in this study because they are more difficult to handle in terms of assessing freshness and safety, preparation (opening the shells) and transportation compared to more common seafood species like salmon and cod. In Australia oysters rank as eighth most widely consumed seafood species (Danenberg & Mueller, 2011) and their production represents an economically important branch of the Australian aquaculture sector (ABARES, 2011). Even though oysters can be considered a highly recommendable seafood

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

product because of the nutritional value, it is still considered an undervalued seafood species (FRDC, 2002). Oysters are known by 99% of Australian consumers, liked by 60% (FRDC, 2002) but only frequently consumed by 37% (in 2010) (Danenberg & Mueller, 2011). A number of potential barriers to oyster consumption were previously identified in studies such as Kow, Yu, FitzGerald, and Grewal (2008), and Liu et al. (2006). Consumers find them hard to open and potential meal usages are perceived as limited. Their distribution in Australia is so far mainly limited to specialty fish stores, narrowing wider availability. Hence, deeper understanding of determinants of oyster purchase choices is crucial for seafood producers and public policy makers in order to increase product sales and healthy seafood consumption. This article attempts to address the gaps outlined above through two research questions. First, what is the relative importance of convenience factors, such as food preparation format, product packaging, serving suggestions and preparation accompaniments, for consumer choice of oysters relative to traditional economic factors such as price, region of origin, health, quality and environmental claims? Second, to what degree do consumer segments differ regarding the importance of oyster characteristics on choice behaviour? Taking into account findings for both research questions, market share simulations provide insights into the existing market potential for product differentiation by packaging and convenience oyster characteristics. 2. Determinants of oyster and seafood consumption According to the focus of this study, this section first reviews general food consumer studies on the influence of preparation convenience and packaging. Then a number of different drivers identified in the literature as influencing oyster consumption specifically and seafood consumption more generally are outlined.

493

research (Geeroms, Verbeke, & Van Kenhove, 2008). Here easyto-prepare seafood meals could provide an alternative as healthy ready-to-eat meals (Olsen et al., 2012). With respect to oysters Liu et al. (2006) suggested introducing preparation advice to reduce perceived inconvenience and to attract a broader range of consumers, who have little experience with oyster preparation. As for preparation format, the effect convenience has on consumers’ choice of oyster and seafood products so far has not been quantified. 2.2. Packaging Food packaging was repeatedly found to be a strong driver for consumers’ food choice and packaging characteristics were observed to demand significant market price differences (Mueller Loose & Szolnoki, 2012). Thereby packaging format not only triggers consumers’ subconscious symbolic associations and valuations (Becker, van Rompay, Schifferstein, & Galetzka, 2011) but also affects consumers’ ability to inspect food characteristics and to transport the product safely. So far there is no research available on the impact of different oyster or seafood packaging formats on consumer value perception and product choice. Seafood packaging is generally strongly related to retail distribution formats. In Australia, the pre-dominant share of oysters are sold through specialty seafood retailers, where oysters are offered on pre-moulded black PET trays that are wrapped individually after consumers selected the amount they want to purchase. To increase the availability of oysters, producers aim for a wider distribution through other retail formats, such as grocery stores, where oysters have to be offered completely pre-packed, such as on glad-wrapped deli-trays. According to industry advice from the Australian Oyster Council it is unknown if consumers would accept such a closed packaging solution compared to the traditional offering. 2.3. Price

2.1. Convenience of food preparation For food products in general, convenience is related to time savings and reduced efforts regarding preparation, cooking and cleaning (de Boer, McCarthy, Cowan, & Ryan, 2004). Convenience in food preparation has recently received increasing attention (Bava, Jaeger, & Park, 2008; Brunner, van der Horst, & Siegrist, 2010; Mahon, Cowan, & McCarthy, 2006; Olsen, Menichelli, Sørheim, & Næs, 2012; Scholderer & Grunert, 2005) in line with the growing market share of ready-to-eat and half-prepared meals. Due to consumers perceiving seafood preparation as time consuming, convenience has been identified as an important factor to expand general seafood consumption (Altintzoglou et al., 2010; Olsen et al., 2007). In addition to perceived difficult preparation of seafood, many consumers feel insecure about preparation methods (Brunsø, Verbeke, Olsen, & Jeppesen, 2009; Hicks, Pivarnik, & McDermott, 2008; Olsen, 2003, 2004; Rortveit & Olsen, 2009). The physical preparation format of seafood (e.g. deboning or filleting) is expected to allow easier preparation. For oysters, the half shell preparation format does not require consumers to open oysters with a knife, which is related to risk of injuries. Also, half-shell oysters offer the highest degree of convenience because they are ‘‘ready-to-eat’’ or to be further processed (e.g. broiling or barbecuing). Although food and particularly oyster preparation was identified as a consumption barrier, to our best knowledge its relative importance is so far still unexplored. Besides the physical preparation format of seafood products any other factor that reduces preparation time has the potential to increase convenience. Particularly ready-to-eat meals and convenience food have recently gained attention in food consumer

Traditional economic factors such as price or perceived cost were found to be strong barriers to seafood consumption because seafood is generally perceived to be more expensive than other protein sources (Brunsø et al., 2009; Olsen, 2004; Verbeke & Vackier, 2005). This perception was pronounced for oysters, which are seen as a relatively expensive product that is mainly limited to consumption on special occasions (Girard & Mariojouls, 2003). Price sensitivity for seafood generally depends on income (Myrland, Trondsen, Johnston, & Lund, 2000) and it is therefore expected to be lower for higher income segments. 2.4. Region of origin Region of origin is generally an important aspect in food purchase decisions because it jointly signals a range of different credence product characteristics. Region of origin can function as a quality indicator for region-specific taste characteristics and as an indicator of food safety risks (Kim, 2008; Scarpa, Philippidis, & Spalatro, 2005). In addition to the quality inference, consumer ethnocentrism and regionalism (Sharma, Shimp, & Shin, 1995) were identified as drivers for regional preferences. For instance seafood consumers in coastal areas are generally more willing to buy products which have been caught or farmed in their local area than being transported from another seafood producing region (Bose & Brown, 2000; Brécard, Hlaimi, Lucas, Perraudeau, & Salladarré, 2009). This also applies to Australian oyster consumption, where domestic products are preferred over imported products (FRDC, 2002). Besides ethnocentrism, preference for local production is attributable to perceived freshness due to shorter transportation

494

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

times (Kow et al., 2008; Liu et al., 2006). The fact that consumers paying attention to geographical origin were also observed to be sensitive to eco-labels and whether the product was caught wild or farmed (Brécard et al., 2009; Fernández-Polanco, Mueller Loose, & Luna, in press) suggests a generally higher interest of region sensitive consumers towards credence attributes.

2.5. Food safety Freshness and quality considerations are closely related to food safety, an important barrier to seafood consumption. Seafood was repeatedly related to mercury and antibiotics contamination as well as the presence of parasites like Anisakis, leading consumers to develop negative quality associations (Verbeke, Sioen, Pieniak, Van Camp, & De Henauw, 2005; Wessells, Kline, & Anderson, 1996). Consumers’ perceived low ability to identify whether a product is safe or not and their limited seafood knowledge were found to constitute main barriers for seafood consumption (Birch & Lawley, 2012; Hicks et al., 2008). With respect to oysters it was repeatedly observed that consumers, who are particularly concerned about food safety issues preferred half shell pre-opened oysters over closed shell format (Kow et al., 2008; UNIC, 2003) – even though the closed shell format is safer. This can mainly be related to the trust consumers have in the seafood salesperson, who examines the oyster and is more experienced in deciding if it is safe to eat (Birch & Lawley, 2012; Pieniak et al., 2007).

2.6. Health motivation and health claims Taste and nutritional value are among the main drivers for purchasing seafood (Birch & Lawley, 2012; Olsen, 2004; Verbeke & Vackier, 2005). Health motivation is a particularly strong driver for older consumers, who have stronger health concerns and are, at the same time, more experienced in eating seafood (Olsen, 2003). For oysters in particular, taste and variety in the diet were identified as main consumption factors for US consumers (Gempesaw et al., 1995; Hanson, House, Sureshwaran, Posadas, & Liu, 2003). Although seafood is generally perceived as healthy, with the exception of Grisolía, López, and Ortúzar (2012) it is largely unknown, if health claims are able to reinforce consumers’ associations of healthiness with oyster consumption as was observed for other food categories (Lähteenmäki, 2013).

2.7. Environmental and quality claims A number of environmental and health claims have recently gained research attention with expectations to increase consumer choice and willingness to pay for different seafood species (Fernández-Polanco et al., in press; Johnston & Roheim, 2006; Wessells, Johnston, & Donath, 1999). Carbon claims are specific environmental claims for which positive consumer response was observed for other food products (Onozaka & McFadden, 2011; Vanclay et al., 2011), but to our best knowledge there are no insights available for oysters or seafood. Diverging findings exist regarding the potential of quality claims or certifications to increase seafood consumption. No effect of quality claims could be observed in an in-home test involving preparation and consumption of real cod samples (Kole, Altintzoglou, Schelvis-Smit, & Luten, 2009). A positive impact of seafood quality certifications was observed for consumers’ stated preferences in a choice experiment (Jaffry, Pickering, Ghulam, Whitmarsh, & Wattage, 2004), but verbal and not visual product alternative presentation were used, limiting the predictive ability of a choice experiment (Mueller, Lockshin, & Louviere, 2010).

2.8. Research questions The literature review identified a large number of drivers that were found to determine choice and perceived value of seafood in general and of oysters in particular. While the relative quantitative impact of traditional choice factors such as price, food safety and region of origin has been already examined in several studies, the relative importance of various convenience and packaging factors is still unknown. Such knowledge would provide important insights to public policy makers, seafood producers and marketers on the relative importance they should attach to these attributes compared to determinants such as price, region of origin, health, quality and environmental claims. Hence, the underlying research questions of this study are: (1) To what degree can the choice of oysters (as one specific seafood category) be influenced by product packaging and convenience factors, such as preparation format, serving suggestions and preparation accompaniments? This question also explores the relative importance of oyster packaging and preparation convenience relative to traditional demand factors. Seafood consumers were repeatedly found to differ in several dimensions, such as their consumption level (Girard & Mariojouls, 2003), their trust and confidence into seafood quality (Pieniak et al., 2007; Verbeke, Vermeir, & Brunsø, 2007) and their sensory preferences (Semenou, Courcoux, Cardinal, Nicod, & Ouisse, 2007). Taking heterogeneity into account allows producers to identify and target consumer segments, which differ in their needs and choice drivers for oyster consumption. This is investigated by a second research question: (2) To what degree do consumer segments differ in their preferences for oyster characteristics and the relative attribute importance on oyster choice? Finally, market share simulations will combine results from both research questions to provide insights into the existing oyster market potential for product differentiation by packaging and convenience characteristics.

3. Materials and methods 3.1. Discrete choice experiment with visual shelf simulation In order to analyse the preferences for oysters at the point of sale we used a stated preference approach, utilising a discrete choice experiment (DCE) with visual shelf simulation (Mueller, Lockshin, & Louviere, 2010). By presenting respondents with a set of alternatives from which they can choose, the attribute preferences are investigated indirectly without asking the participant directly about the value of certain product attributes. In doing so, one obtains results closer to real in-store decision making processes compared to questionnaire surveys, limiting social desirability bias (Norwood & Lusk, 2011). Discrete choice experiments have been frequently applied to examine seafood preferences, for a recent overview see Grisolía et al. (2012). Visual presentation of packaging attributes was shown to trigger unconscious processing of product cues in consumer choices and consumers are not able to introspect or validly evaluate the importance of these cues when they are presented only verbally (Mueller, Lockshin, & Louviere, 2010). Verbal presentation and framing of attributes was found to be particularly inappropriate when the attributes being evaluated are non-utilitarian, abstract, require a sensory experience or are used subconsciously (Fitzsimons et al., 2002). Product retail presentation format, product

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

packaging and visual serving suggestions are product attributes that provide a sensory experience at the point of purchase and therefore require visual presentation to be validly measured in choice experiments. To our best knowledge this study is one of the first to utilise shelf simulation discrete choice experiments for seafood products. 3.2. Product attributes and levels Attributes and levels for the choice experiment were selected to cover the most important choice drivers in guiding consumers’ retail purchase decision for oysters. The selection was based on previously identified drivers from the literature review and in consultation with the Australian Oyster Consortium which shared with us prior corporate qualitative research on Australian consumers’ oyster purchase behaviour and advice from seafood marketing experts. Price was included with four price levels ($4.50, $5.50, $6.50 or $7.50 per half dozen), representing the range of market prices at the time of the study. We included origin with two different national origins (Australia and New Zealand), one state origin (New South Wales, NSW) and one region within the state (Corrie Island, NSW), which was potentially less familiar to consumers from other states. Thereby our aim was to test if states or specific regions would be preferred by consumers over national origin declarations. Farmed oysters in Australia are mainly limited to two species, Pacific Oysters native in Asia and the endemic Sydney Rock Oysters. Because these two species are not necessarily marketed by their names and differ in their size and visual appearance, it was important to present them visually in realistic photographs. The widely distributed Australian ‘heart tick’ (Rayner, Boaz, & Higginson, 2001) and the carbon zero claim were included in the

495

experiment as health and environmental claims respectively. While the ‘heart tick’ is a standardised third party certified label there is no uniform carbon label available in the food market. This study utilised a carbon zero logo from the company ‘Atlas Copco’, which was previously used in other food choice experiments (Mueller Loose & Remaud, 2013). As for the accompaniments, the incidence of the ‘none’ level was increased to prevent an over-representation of choice options with claims. Finally, a quality award (Sydney Royal Fine Food Show Gold Medal) was selected as an experimental factor to test the impact of a quality claim on consumers’ oyster choice. Closed and half-shell open oysters are traditionally available in most international seafood markets. A third format of pre-shucked oysters, which is already available at the west coast US market, was included as a third attribute level. Because the oysters were already pre-opened, consumers do not have to open the oyster with a knife. Also, in the pre-shucked but closed shell format oysters are prevented from spilling or loosing liquid by keeping both halves together, also protecting them from any potential spoilage. Oysters were either shown as whole shell unopened or half-shell open oysters in the photograph. Before the experiment respondents were informed about the new pre-shucked oysters, which in the choice sets were identified by a purple ‘‘NEW easy open’’ claim in the visual product presentation (see upper left choice alternative in Fig. 1). We hypothesise this to have a positive effect on choosing oysters and to increase market share relative to the closed format. Two oyster packaging tray formats were selected for the experiment. The black pre-moulded PET tray is mainly available at fish mongers, securing the half-shell opened oysters and preventing them from slipping. The glad-wrapped deli tray is new to the market and is more suitable for supermarket retailers, because unlike the pre-moulded trays they do not require further wrapping and

Choose the alternative you prefer most.

Would you realistically purchase your most preferred alternative?

Yes

Fig. 1. Example choice set with visual shelf simulation.

No

496

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

Table 1 Attributes and levels in discrete choice experiment. Attribute

# levels

1

2

3

4

Price per half dz. Origin

4 4

$4.50 Australia

$5.50 New Zealand

$6.50 New South Wales (Australia)

$7.50 Corrie Island (NSW)

Species

2

Pacific Oyster

Marketing claim

4

None

None

Quality award Packaging

4 2

None

None

Preparation format

4

Sydney Rock Oyster Health claim (heart tick) Gold medal PET moulded tray black Unopened

Opened half shell

Opened half shell

Accompaniments (w/wo visual serving suggestion)

8

Kilpatrick

Champagne Sabayon

Champagne Sabayon + visual serving suggestion

Environmental claim (carbon zero) None Glad wrapped white deli tray Unopened – prechucked easy open Kilpatrick + visual serving suggestion

are ready for check-out. We hypothesise that glad wrapped deli packaging is less preferred by consumers than the black PET packaging because pre-opened oysters can tilt, the wrapping partially impedes inspection of the content and consumers are unfamiliar with this packaging form. The literature review and advice from the industry consortium identified accompanying oyster preparation sachets, allowing a quick and convenient preparation of broiled or barbecued oysters, as a potential factor to increase the likelihood of purchasing oysters. Two preparation styles, the well-known Kilpatrick recipe and a less known Champagne Sabayon preparation were both included in the experiment with and without a visual serving suggestion, showing an example of the prepared oyster meal. It was expected that a visual serving suggestion should reduce information asymmetry and provide consumers with an example of the prepared meal. To prevent an over-representation of oysters with accompaniments in the choice experiment, four ‘none’ levels were included for this eight-level attribute. Overall, eight attributes with two to eight levels were included into the choice experiment (see Table 1). By including the majority of the relevant oyster product attributes in the experiment, we avoided a potential over-estimation bias from neglecting important choice drivers (Gao & Schroeder, 2009; Islam, Louviere, & Burke, 2007). 3.3. Experimental design The overall experimental design contained 128 sets that consisted of 64 sets from a D-optimal 8  4^5  2^2 orthogonal main effects plan (Street & Burgess, 2007) and 64 sets randomly drawn from the full factorial. The 128 sets were blocked into eight versions of 16 choice sets and respondents were randomly allocated to one of the eight versions. As consumption occasion respondents were asked to choose oysters for a dinner with friends and family. In the experiment respondents were presented with photo-realistic images showing 16 successive choice sets of four alternatives and a no-choice option (negative purchase intent) in each set (Fig. 1). 3.4. Consumer sample The study was conducted in early 2011 in Australia. 1718 participants, representative of Australian seafood consumers, were recruited by an Australian online panel provider to take part in an extensive online survey regarding their seafood consumption patterns and completed a DCE investigating their preferences for oysters. The panel provider actively manages a panel of 300,000

5

6

7

8

None

None

None

None

members and is certified by ISO 26362 and the Quality Standard for Online Access Panels (QSOAP) ‘Gold Standard’. For their participation respondents received a remuneration of about AUD 1.30. To qualify respondents had to be oyster consumers (eating oysters at least several times a year). Sample characteristics are provided in the left column of Table 5. There were slightly more female than male respondents and in line with prior seafood consumption research (Olsen, 2003) older consumers were overrepresented in the sample. The relative share of Australian states in the sample is largely identical to their proportion of inhabitants. Respondents were asked about their seafood consumption in the last week, including all occasions where seafood was part of a meal, such as filling in sushi, as ingredient in a soup, as part of a sandwich or salad or as a main meal component of a seafood dish. Fifty per cent of the sample had some kind of seafood component as part of their diet at least twice in the last week and more than a quarter of respondents more than five times a week. The survey was conducted in summer, the main barbecue and outdoor season in Australia, where seafood is more likely to be served as part of a larger variety of food components. It is therefore likely that seafood consumption is less frequent during the winter season. 3.5. Scale adjusted latent class model Respondents’ choices were analysed with a scale adjusted latent class model (SALCM), identifying segments of consumers, differing in their oyster preferences and thereby accounting for differences in response consistency (Mueller, Lockshin, Saltman, & Blanford, 2010). A SALCM separately considers two types of response heterogeneity by identifying independent utility and error variance latent classes (Vermunt & Magidson, 2008). The overall preference distribution is assumed to consist of a combination of unobservable, latent groups or classes that differ in their utility between the groups but are similar within (Boxall & Adamowicz, 2002). Similarly, distinct groups that differ in their choice consistency (error variance) are modelled separately. As the error variance is inversely related to the scale parameter k, certain or consistent respondents are assigned a higher k than uncertain or inconsistent respondents. In the model by Magidson and Vermunt, (2007) the random utility of alternative j for individual n depends on the latent preference class q = 1, . . ., Q and the unobserved scale factor class kd, d = 1, . . ., D, and each individual is expressed in terms of the likelihood that they belong to each class. For identification, one scale factor is normalised to unity and the remaining scale factors represent ratios of the reference scale class. The probability of choice i by individual n, conditional on preference class q and scale factor class d is:

497

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504 0x

expðkd bq n;i Þ Pðni jq; dÞ ¼ X 0x expðkd bq n;j Þ

ð1Þ

Table 2 Estimates for aggregated multinomial logit model (n = 1601).

We assume that there are Q discrete latent preference classes and D scale classes, but that class membership is hidden. The overall log likelihood is:

" # Q N D X X X ln L ¼ Pn ¼ ln Pnijq;d n¼1

t-Stat

b

j

ð2Þ

d¼1 q¼1

All choice models were estimated in Latent Gold Choice Syntax 4.5 (Statistical Innovations Corp.). To account for algorithm randomness we followed the approach suggested by Dolnicar and Leisch (2010), where 20 randomised start seeds were used and the solution with the best model likelihood is selected to increase the chance for an identified global maximum. Attribute importance was approximated with the share of variance explained by each attribute, assuming that the presented attributes determine 100% of the choice process (Lancsar, Louviere, & Flynn, 2007; Louviere & Islam, 2008). 3.6. Market share simulation From the estimated part worth utilities class specific market shares were simulated according to Eq. (1). One attribute at a time was varied according to the examined attribute levels and all other attributes were fixed to a base level: origin (Australia), species (Sydney Rock), preparation format (open half-shell), packaging (black pre-moulded PET tray), accompaniments (none), claim (none), award (none). The number of alternatives assumed to be present in the market was varied according the number of levels of the attribute under consideration. Additionally, a no-choice alternative was included in all simulations, capturing the share of consumers not willing to buy any of the available product alternatives. Aggregated market shares were derived by class size weighted averages across latent class specific market share predictions.

Constant Price $4.50 $5.50 $6.50 $7.50 Preparation format Unopened Unopened – preshucked Opened half shell Species Sydney Rock Pacific Origin Australia New Zealand New South Wales (Aus) Corrie Island (NSW) Accompaniment Kilpatrick Kilpatrick + serv sugg Champagne Sabayon Champagne S. + serv sugg None Packaging PET black tray pre-moulded Glad wrapped white deli tray Claim Heart tick Carbon zero None Quality award

0.40

18.00

Wald-stat Attribute importance (%) 323.96

0.54 46.04 2758.98 0.17 13.45 0.17 12.71 0.55 36.47

34.4

0.37 28.05 2820.40 0.18 13.75 0.55 52.83

35.2

0.17 22.68 0.17 22.68

514.48

6.4

0.19 15.44 1334.88 0.53 36.07 0.22 17.32 0.12 9.31

16.7

0.13 6.93 0.18 9.76 0.23 11.70 0.18 8.82 0.10 7.91

319.42

4.0

0.09 11.76 0.09 11.76

138.28

1.7

0.10 0.01 0.09 0.06

115.50

1.4

8.85 0.70 9.35 3.27

10.72

0.1 2

Notes: LL = 36,527.9, BIC(LL) = 73,188.6, N = 1601; df = 1583, q = 0.102. All part worth utility estimates are significant at p < 0.001 except for quality award (p = 0.001) and carbon zero. All Wald statistics are significant at p < 0.001. Attribute importance as attribute’s partial contribution of an attribute to model fit (log likelihood).

4. Results 4.3. Selection of optimal number of latent classes 4.1. Reasons for no-choice From the originally 1718 respondents who qualified for the survey, n = 117 (6.8%) always chose the none-option in the choice sets and were therefore excluded from further analysis. When asked for reasons of not accepting any of the choice alternatives, 23% stated that they cannot eat raw oysters, 23% only consumed oysters when eating out on premise because of the risk of purchase and preparation, 16% only bought locally and would never buy pre-packed oysters, 14% only bought larger quantities than half a dozen, 7% only bought certain local origins and 7% considered all options as too expensive. Choices of the remaining 1601 respondents, who were willing to buy at least one of the offered alternatives, are analysed in the following sections. 4.2. Aggregated MNL model An aggregated multinomial logit model was estimated as reference model, assuming independence of irrelevant alternatives and homogeneous consumer preferences. The part worth utility estimates of the reference model in Table 2 are highly significantly different from zero, confirming an influence of all attributes selected on consumer choice. The rightmost column provides the attribute importance derived from the partial contribution of each attribute to the model fit (effect size). Accordingly, on the aggregated level oyster preparation format and price have the strongest impact, where pre-opened half-shell and lower prices are preferred, followed by a preference for domestic origins and oyster species.

A SALCM with two scale classes was fit to respondents’ choice data by iteratively estimating a number of models to ensure a reliable number of classes and final model specifications. The Bayesian Information Criteria (BIC) statistic decreased continuously when increasing the number of preference classes. Although we could not observe a minimum BIC value there was a distinct ‘kink’ for the BIC statistic, when increasing from six to seven classes, suggesting that there was only a minor improvement of model fit beyond six preference classes. Also, increasing the number of preference classes to seven resulted in a solution with very small segments with less than 5% of respondents, corroborating that six preference classes provided the optimal fit. Increasing the number of scale classes from one to two significantly improved the BIC fit statistic, but when allowing three scale classes the model did not converge. Accordingly, a model with six preference and two scale classes was selected. Models specifying price as continuous and categorical variable were compared with the log-likelihood-ratio-test and resulted in a significant improvement of the categorical specification (Chi2 = 57.68, df = 12, p < 0.001). 4.4. Utility estimates and attribute importance for oyster attributes Consumer preferences for each segment follow from the part worth utilities in Table 3 and the effect size based attribute importance in Table 4 (Lancsar et al., 2007; Louviere & Islam, 2008). The scale classes S1 and S2 captured respondents, who differed in choice consistency. Because the scale parameter k is inversely

498

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

Table 3 Scale adjusted latent class model estimates (6 preference classes and 2 scale classes). Latent class N Relative size

C1 419 37% 0.260

q2

t-Stat

C3 307 27% 0.430

C6 305 27% 0.433

b

t-Stat

b

t-Stat

4.21

17.67

1.61

7.60

3.59

24.31

3.91

33.78

1.62

10.58

0.21 0.37 0.01 0.59

4.70 8.29 0.26 12.19

0.17 0.28 0.00 0.45

3.03 5.15 0.00 7.51

0.32 0.38 0.03 0.73

5.02 5.73 0.56 11.07

2.73 1.38 0.70 3.41

28.58 15.81 7.68 24.45

1.17 0.15 0.24 1.09

12.33 1.39 2.03 7.41

4.07 1.50 1.12 4.45

47.57 17.96 10.78 23.32

0.38 0.13 0.24

9.33 3.37 7.28

0.38 0.12 0.50

6.96 2.18 11.89

2.84 1.89 4.73

18.80 13.66 44.78

2.97 1.44 4.41

16.29 9.30 35.15

0.28 0.21 0.49

2.69 2.05 5.92

0.26 0.07 0.19

4.71 1.28 3.87

1.32 1.32

43.54 43.54

1.03 1.03

27.88 27.88

0.58 0.58

17.25 17.25

0.10 0.10

2.22 2.22

0.40 0.40

6.40 6.40

0.19 0.19

5.92 5.92

0.81 2.41 1.00 0.61

18.59 36.23 22.11 13.66

0.01 0.23 0.21 0.04

0.26 4.26 3.56 0.70

0.47 1.44 0.43 0.53

7.38 20.19 6.86 8.43

0.44 1.21 0.38 0.39

4.71 12.38 4.59 4.35

0.21 0.57 0.34 0.02

2.06 4.88 3.41 0.16

0.42 1.17 0.60 0.15

6.12 17.58 10.34 2.44

0.34 0.24 0.65 0.43 0.50

5.23 3.70 9.10 6.11 11.44

0.20 0.30 0.28 0.27 0.05

2.41 3.67 3.13 3.11 0.87

0.57 0.59 0.71 0.41 0.04

6.05 6.49 7.69 4.06 0.66

0.33 0.20 0.12 0.09 0.73

2.48 1.53 0.91 0.67 8.35

0.53 0.83 0.69 1.12 0.46

3.50 5.76 3.49 5.03 4.31

0.39 0.54 0.45 0.35 0.13

4.30 6.06 5.04 3.82 2.21

0.15 0.15

5.95 5.95

0.13 0.13

4.21 4.21

0.53 0.53

13.39 13.39

0.07 0.07

1.23 1.23

0.34 0.34

5.41 5.41

0.07 0.07

1.90 1.90

0.22 0.03 0.25 0.21 S1 40.4%

5.50 0.73 7.65 3.50

0.40 0.16 0.24 0.10 S2 59.6%

8.29 3.25 5.82 1.29

0.13 0.13 0.26 0.17

2.35 2.25 4.98 2.08

0.12 0.05 0.07 0.03

1.53 0.64 1.03 0.25

0.10 0.16 0.06 0.06

1.07 1.75 0.69 0.41

0.25 0.15 0.10 0.20

4.63 2.68 2.22 2.47

k

b

t-Stat

k

t-Stat

1.00

0.45

44.71

t-Stat

C5 159 14% 0.334

21.58

Scale factor

t-Stat

C4 186 16% 0.371

4.02

b Constant Price $4.50 $5.50 $6.50 $7.50 Preparation format Unopened Unopened – preshucked Opened half shell Species Sydney Rock Pacific Origin Australia New Zealand New South Wales (Aus) Corrie Island (NSW) Accompaniment Kilpatrick Kilpatrick + serv sugg Champagne Sabayon Champagne S. + serv sugg None Packaging PET black tray Glad wrapped deli tray Claim Heart tick Carbon zero None Quality award Scale class Class size

C2 225 20% 0.159

b

t-Stat

b

Notes: LL = 29,463.2, BIC(LL) = 59,774.9, N = 1601; df = 1486, class error = 0.08, q2 = 0.344.

Table 4 Attribute effect size for latent classes in per cent (n = 1601). Segment # Segment characterisation Segment size

C1 Patriotic and species 26%

C2 Species and format 14%

C3 Half shell 19%

C4 Half shell and price sensitive 12%

C5 Price sensitive and accompaniments 10%

C6 Price sensitive 19%

Weighted average

Price Preparation Species Origin Accompaniments Packaging Claim Quality award

3.6 2.0 48.9 38.6 4.7 0.7 1.2 0.2

5.6 11.2 72.3 1.7 2.3 1.3 5.5 0.1

1.9 81.5 4.6 7.0 1.8 2.7 0.4 0.1

33.9 60.7 0.1 3.6 1.6 0.0 0.1 0.0

43.5 9.3 10.9 7.9 19.9 7.8 0.8 0.0

91.4 0.4 0.5 6.1 1.1 0.1 0.3 0.1

27.8 25.8 25.0 14.1 4.3 1.7 1.3 0.1

Notes: Sorted in descending order for weighted average. Attribute importance as attribute’s partial contribution of an attribute to model fit (log likelihood).

related to error variance, a lower scale parameter indicates a higher error variance and less certain and consistent choices. Hence, the scale class S2, consisting of about 60% of the sample, did choose less consistently compared to scale class S1. The latent class choice model also reveals that most consumer segments only rely on a few product characteristics when making a choice, agreeing with findings on the prevalence relevance of choice heuristics in consumers’ choice decisions (Hensher, 2010).

For instance for segment C6, price explains 91% of choice variance, followed by origin with 6%. Similarly, C3 used one main characteristic of format to make oyster choices, explaining 82% of variance. The two segments C1 and C4 used mainly two attributes for their choice decisions (species and origin for C1 and format and price for C4). The smallest segment C5 is characterised by the most complex decision strategy with three major and three minor choice drivers.

499

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

For a number of attributes we observed agreement between the segments in which attribute levels are most preferred, although the relative importance of attributes differed strongly across segments. Generally all segments agree in that they prefer pre-opened half-shell oysters and this is followed by pre-shucked and unopened formats. With a weighted average of 26% preparation format was the second most important attribute across all segments. For respondents in segments C3 and C4, comprising about a third of the sample, preparation format was the dominant choice criterion. All segments strongly prefer domestic origins, except for a small preference for New Zealand origin of segment C2 for which origin was not very important (1.7%). Although on average origin only accounted for 14% of choice variance, it was particularly important for about a quarter of respondents (segment C1), for which it was the second largest choice driver after species. Similarly, all segments prefer the black PET packaging format over glad wrapped deli-trays, but overall packaging had only a minor impact on consumer choice (on average 1.7%), except for a higher importance (7.8%) for segment C5. The majority of segments had the highest preference for the health claim followed by the carbon zero and no claim, but the overall share of choice variance explained by claims was only 1.3%, ranking as second least important attribute. The quality award ‘gold medal’ did have a significantly positive impact on choice for three of the six segments, but with an average of only 0.1% explained choice variance it had the lowest importance across all attributes. Across segments we observed considerable choice variance for the attributes species and accompaniments, which most likely relate to variance in sensory seafood preferences (Birch & Lawley, 2012; Semenou et al., 2007). While the aggregated choice model in Section 4.2 suggested an overall preference for Sydney Rock Oysters and a low importance of species, the SALCM clearly reveals that consumers are strongly polarised in their species preferences, where all segments except for C2 preferred Sydney Rock Oysters, while segment C2 had a strong preference for Pacific Oysters. This preference heterogeneity did cancel out in the aggregated model, thereby strongly underestimating the effect of species on consumer choice (only 6.4% instead of 25.0% in the latent class model). For accompaniments segments C4 and C1 preferred oysters without pre-prepared flavour sachets, while the other segments did have a positive utility for Kilpatrick preparations. Accompaniments had the highest importance for segment C5 (19.9%), but were overall of minor importance (4.3%). All segments show a dislike of the rather unfamiliar Champagne Sabayon preparation. It is also interesting to observe that three of the six segments did not act according to a linear price-demand function but preferred the second lowest over the lowest price (segments C1, C2 and C3). This behaviour has repeatedly been observed for other food with significant quality uncertainty and a high importance of experience and credence attributes (Bredahl, 2004; Mtimet & Albisu, 2006; Mueller, Lockshin, & Louviere, 2010). In the case of quality uncertainty higher prices are used as signal for higher quality and this effect can over-compensate the negative utility of higher prices (Mueller, Osidacz, Francis, & Lockshin, 2010; Plassmann, O’Doherty, Shiv, & Rangel, 2008).

4.5. Characterisation of latent classes by covariates Choice segments were characterised by sociodemographic variables and their average seafood consumption in Table 5. Analysis of variance (ANOVA) only revealed few significant differences across segments. This is in accordance with previous findings that differences in revealed food preferences are only weakly related to sociodemographics (Corsi, Mueller, & Lockshin, 2012; Mueller, Lockshin, Saltman, et al., 2010).

Table 5 Sociodemographic characterisation of sample (n = 1601) and latent classes (all shares in per cent).

N Relative size

Sample 1601

C1 419 26%

Gender Females 57 59 Males 43 41 Age (in years) 18–24 2 4 25–34 11 13 35–44 16 18 45–54 20 18 55–64 20 18 65+ 32 30 Children at home (<12 years old) No 86 79* Yes 14 21* State NSW 31 26 Victoria 24 23 Queensland 21 25* Western Australia 9 13* South Australia 9 8 ACT 2 2 Tasmania 3 2 Northern Ter. 1 1 Income per year n 1245 167 6$35,000 28 32 $35,001–$65,000 27 26 $65,001–$95,000 19 16* $95,001+ 26 27 Seafood consumption p. week 0–1 times 22 23 2–4 times 50 51 5+ times 28 27 * **

C2 225 14%

C3 307 19%

C4 186 12%

C5 159 10%

C6 305 19%

54 46

67** 33**

64** 36**

49* 51*

50 50

1 11 18 21 20 29

1 11 22* 21 19 26

2 10 17 29* 19 24*

2 15 13 21 15 35

1 8 12 15 23 43*

85 15

84 16

86 14

88 13

91* 9*

28 26 21 8 11 2 3 1

30 26 24 8 5* 4 3 0

31 23 19 8 12 1 6 1

29 22 18 13* 13* 2 4 0

36* 23 19 7 9 3 3 1

256 34 26 20 21

229 23* 24 20 33**

152 19* 31 20 30

113 28 22 20 29

328 31 28 17 24

26 50 24

20 50 30

24 50 27

24 44* 32

19 51 30

ANOVA significantly different across preference classes, p < 0.10. ANOVA significantly different across preference classes, p < 0.05.

Female consumers are slightly over-represented in those segments that have a particularly high preference for pre-opened half-shell oysters, suggesting that females dislike having to open closed oysters more than male consumers. Males were over-represented in segment C5 that had the strongest preference for readyto-use Kilpatrick accompaniments. Older consumers tended to be over-represented in the most price sensitive segment C6, concurring with their on average lower incomes. There was hardly any relationship between price preferences and income, only the price insensitive segment C3 had a significantly higher income.

4.6. Market share simulations Fig. 2 shows market share simulations for seven different product attributes. Thereby the product alternatives competing in the market at a certain price point (horizontal axis) only differ in the levels of one attribute, indicated by the different lines in each chart. The total consumer demand for oysters represents a market share of 100%. Because consumers differ in their oyster preferences the maximum demand can only be met by differentiated oyster products. For instance, the total market share can be maximised by simultaneously offering those ‘optimal’ product specifications that maximise each segments’ utility. Vertically, at each price point the market shares of the competing products and the share of consumers not willing to buy any oysters of this particular attributeprice specification add up to 100%. The no-choice share is not shown in the charts but follows as the remainder of 100% minus the product alternatives’ market shares. When comparing market share differences across attributes in this market share simulation

500

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

Notes: Market shares of product alternatives and no-choice option add up to 100%. Number of product alternatives considered: 1) x=5, 2) x=2, 3) x=3, 4) x=4. Figures show market shares for variation of price and one product attribute at a time. Base values of other attributes (not varied): Australian origin, Sydney Rock Oysters, opened halfshell oysters, PET black tray packaging, no accompaniments, no claims, no award.

Fig. 2. Market share simulations.

it should be considered that absolute differences depend on the number of attribute levels that represent different alternatives available on the market. For instance, the impact on market share of identical utility differences for only two attribute levels (e.g. species, packaging, and award) is therefore higher than for three, four or five choice alternatives. Because of heterogeneous price preferences the market shares only marginally decrease when the price increases from $4.50 to $5.50 per dozen. Segments C4–C60 decreasing consumer demand is compensated by the increase of demand by the medium-price preferring segments C1–C3. For prices above $5.50 there is a steeper decline in demand because all consumer segments concur in declining utility for price increases. The no-choice share increases as the demand for oysters falls through price increases. For instance for preparation format, the no-choice share increases from 28% at $4.50 to 52% at $7.50.

A second characteristic that is common across all market share simulations is the convergence of product alternatives’ market shares for higher prices. This convergence is a result of the increasing share of consumers not buying any oysters as prices increase. The added value from oyster characteristics then only affects a decreasing share of consumers. For example, at $4.50 oysters in black PET pre-moulded packaging attract a 12% higher market share (of 37%) compared to glad wrapped oysters (25%). This advantage decreases to only 8% at $7.50, where the market share for black PET reduced to 21% and that for glad wrapped oysters to 13%. The impact of a product attribute on market share is reflected by the degree of separation of the lines of different product alternatives. As expected from Table 4, this difference is largest for preparation format, where at $4.50 pre-opened half-shell oysters attract a market share of 48% compared to only 14% for preshucked and 10% for unopened oysters. A slightly smaller

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

difference can be observed for the second most important attribute species, where at $4.50 a market share of 46% is predicted for Sydney Rock Oysters compared to 22% for Pacific Oysters. The market shares simulated for regions of origin visualise the strong discrimination by Australian consumers between domestic and imported oysters, where at $4.50 a market share of 23% is predicted for Australian oysters relative to 4% for New Zealand oysters. By comparison different domestic regions of origin only achieve a maximum market share differentiation of 4.6%. According to Table 4 accompaniments explained more choice variance than packaging and claims, but simulated market shares in Fig. 2 show that this variance mainly stems from consumers’ low preference for the unfamiliar Champagne Sabayon flavouring, for which at $4.50 a market share of 8% and 9% is predicted without and with visual serving suggestion. By comparison to other attributes, the value adding potential to oysters without accompaniments (18%) is rather limited, as Kilpatrick flavouring only adds 1% without and 3% with visual serving suggestion. Packaging only explains little choice variance (Table 4) but for two assumed packaging alternatives available in the market, at $4.50 black pre-moulded PET packaging attracts a 13% higher market share than glad wrapped deli trays, suggesting that the role of packaging cannot be neglected. Compared to oysters without claim, both the carbon zero claim and the health claim ‘heart-tick’ suggest potential to increase the market share by 4% and 8% respectively. Finally, in the case that two identical half dozen of oysters are offered on the market, those with a quality award are predicted a 5% higher market share compared to those without.

5. Discussion 5.1. Importance of convenience and packaging attributes Seafood consumption convenience and packaging were previously identified as barriers to seafood consumption (Brunsø et al., 2009; Olsen et al., 2012; Olsen, 2003, 2004; Rortveit & Olsen, 2009), but their importance relative to traditional factors such as price, origin and species remained largely unexplored. Our results for the first research question identified preparation format as the strongest barrier to oyster consumption, where offering unopened oysters had an extensively detrimental effect on consumer preferences and strongly decreased market share compared to opened half-shell oysters (Fig. 2). Overall, preparation format, which is related to convenience of ready-to-eat, concerns about food safety and potential injury risks, was the second most important driver after price, accounting for 26% of choice variance. Concurring with findings by Kow et al. (2008) and Liu et al. (2006) our findings confirm that Australian consumers show a strong tendency to reject unopened oysters, seeking the convenience of already opened oysters. Although all consumer segments preferred the innovative market preparation format of pre-shucked, easy-open oysters over unopened oysters, our market share simulations at this stage indicate still a limited ability to overcome consumption barriers. Respondents’ unawareness and unfamiliarity with this new format is presumably the main reason for the limited acceptance. If preshucked (easy-open) oysters were introduced to the market, concerted marketing campaigns would be needed in order to build awareness for the new processing and attached benefits. It is also likely that direct experience with this new format, such as wordof-mouth, in-store presentation and sales-person advice, could increase consumer awareness. Furthermore, the appearance of easyopen and unopened oysters is very similar. In order to avoid confusion among shoppers clear and distinct differentiations for example in form of labels are vital at the point of sale (see for example the ‘Gold Band’ approach currently used in the US).

501

Providing consumers with accompaniments such as flavour sachets increase convenience by allowing consumers to quickly prepare variations to standard seafood meals and can help to overcome uncertainty about applicable preparation methods (Olsen et al., 2012). Consumer preferences for two different accompaniment types for Kilpatrick and Champagne Sabayon oyster preparations were tested in the experiment with and without visual serving suggestions. Although the absolute impact on market share was moderate, visual meal presentation generally had a positive impact on consumer choice, suggesting that it can reduce uncertainty and information asymmetry (Brunsø et al., 2009). Generally, Australian oyster consumers agreed in a higher preference for the classical Australian Kilpatrick recipe compared to the less familiar Champagne Sabayon recipe. Different accompaniments and recipes could be tested for their potential to increase consumer preference. Packaging only explained little choice variance but market share simulations indicated that the traditional black PET pre-moulded tray available in traditional seafood stores, preventing oysters from tilting and slipping, captured a larger market share than glad wrapped white deli trays. This suggests that consumers are moderately affected by the extrinsic appearance of seafood packaging and that glad wrapped trays, suitable for self-servicing retail distribution in supermarket shelves, would have to be offered at a price discount compared to traditional packaging presentation. Nevertheless, it should be considered that large-scale distribution of oysters in supermarkets could increase purchase frequency, when oysters are easily available and bought at the weekly grocery shopping (Dellaert, Arentze, Bierlaire, Borgers, & Timmermanns, 1998; Scholderer & Grunert, 2003). This would represent an increase of the total oyster market volume, not yet represented by the market share simulations in Fig. 2. Considering the high seafood consumption frequency of the consumer sample, it could be expected that an increase in oyster consumption is likely to at least partially substitute the consumption of other seafood species. Although seafood products are generally considered to be a healthy option, we observed a small effect of the ‘heart-tick’ health claim on consumer choice and market share. Using this claim in supermarket settings, where seafood strongly competes with other protein sources, seems more reasonable than in traditional specialty stores, where only seafood is sold. Though potential users of this claim will need to consider if the advantages of the claim provide adequate compensation for the claim licence fees charged by the Heart Foundation. An environmental ‘carbon zero’ claim had a smaller positive influence on consumer choice than the health claim and was even less preferred than no claim by the most price sensitive segment C6. Carbon zero claims currently have a low market penetration in Australia and generally tend to suffer from consumer distrust (Dhanda & Hartman, 2011; Mueller Loose & Remaud, 2013). A quality award in form of a food show medal explained the lowest share of choice variance and only had a small effect on market share, which is in line with previous findings for seafood (Kole et al., 2009) and meat (Grebitus, Menapace, & Bruhn, 2011), but differs from a stronger effect observed for wine (Lockshin, Jarvis, d’Hauteville, & Perrouty, 2006; Lockshin, Mueller, & Louviere, 2010). This small effect for oysters is likely due to the lack of familiar context of consumers seeing oysters sold with such Gold Medals from tasting events, where such events are relatively rare. Results suggest that such medals should be displayed where they are available. However, the investment of resources needed in the attempt to win such gold medals should be well reasoned as our results indicate that the resources needed to attain a medal are rather unjustified. The face validity of our finding is strengthened by its alignment with previous research, suggesting that price is the strongest driver

502

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

for seafood choice (Brunsø et al., 2009; Olsen, 2004; Verbeke et al., 2005). Adding to existing research, our market share simulations revealed that price only had a small detrimental effect on consumers’ oyster choice up to a certain critical price (here $5.50 per half dozen), after which overall market price sensitivity increased sharply. This suggests an optimal revenue maximising market price if products are not targeted to different segments by product differentiation (see next section). Species explained 25% of choice variance, where the majority of consumers prefer Sydney Rock Oysters. We cannot exclude that the species names might have been associated with region (Sydney and Pacific), but we could not observe a significant over-representation of eastern Australian states for the preference segment C1 that showed the highest preference for Sydney Rock Oysters. Region of origin per se was the fourth most important choice driver. Australian consumers were found to strongly prefer Australian regions over New Zealand origin, concurring with prior findings (FRDC, 2002). Other than expected, we found no general preference for more specific regions of origin (e.g. NSW or Corrie Island), which also hardly affected simulated market share. Most likely, sub-regions would have to build up a national reputation of particular quality associations first, in order to gain an advantage and a price premium in the market. Compared to consumers’ low differentiation across Australian regions, our results suggest that market shares can be more successfully increased by product differentiation with convenience and packaging attributes. Overall, the findings quantified that convenience and packaging are important factors for seafood choice, where preparation format, packaging format and accompaniments with serving suggestions jointly explained 31.7% of choice variance. 5.2. Consumer heterogeneity and product differentiation Regarding the second research question of consumer preference differences, we identified some scope for oyster product differentiation. First of all, consumers differed in their price sensitivity and three segments representing more than half of the population preferred a medium over the lowest price. Producers and retailers could target these consumers by offering those combinations of product characteristics preferred by these segments at higher prices. These less price sensitive segments did react stronger than average to Australian origin, no accompaniments and quality award (C1), Pacific oyster species and health claim (C2), and opened half-shell preparation and black PET tray (C3). Retailers offering these particular product specifications could accordingly charge higher prices. Contrary, accompaniments and carbon claim are unlikely to appeal to these less price sensitive consumers. About 14% of the sample (segment C5) reacted stronger than average on accompaniments and for this segment the visual serving suggestion also had an above average effect on utility. Generally there was strong agreement in Australian consumers’ preferences for domestic origins, open half-shell oysters and black pre-moulded PET packaging trays, suggesting that product differentiations deviating from these concordantly preferred oyster characteristics have to be offered at a price discount. Because of few sociodemographic differences between the preference segments, targeting appears to be mainly limited to offering different product specifications in the same retail outlet. 5.3. Limitations and future research This study was limited to oyster consumers in Australia and future research should aim at validating and generalising the findings for other species and different seafood markets. While the

pre-opened half-shell preparation format is particular to oysters, factors such as deboning, ready-prepared fillets versus whole fish and pre-seasoning would be particularly interesting for other seafood species. The small effect of the heart tick might be attributed to the fact that once participants were exposed to the information for the first time (i.e. the first choice set they encountered that included the heart tick) they memorised the information of oysters being a healthy choice for the heart. This might have led to a learning effect and as such the preference for a product carrying the heart tick cannot be further singled out by our design and modelling. This is a limitation of the paper and future research should address the impact of health claims by allowing substitution between different product categories. The research design did not fully integrate potential substitution effects with other seafood species and an increase of the availability and market share of oysters could come at the expense of other seafood species. This study is limited to the initial purchase decision and sensory attributes have to be included to model the repurchase decision. Although discrete choice experiments with visual shelf simulations are characterised by a high external validity (Mueller, Osidacz, et al., 2010), our findings were derived from stated preferences without actual purchase transactions. Future research should aim at further validation, for instance using nonhypothetical choice experiments or by analysis of real market data (Mueller Loose & Szolnoki, 2012). 6. Conclusion This study contributed to consumer food behaviour research by quantifying the relative importance of several convenience and packaging attributes for the specific seafood category of oysters. Choice experiments with visual shelf simulations were found to be a suitable methodology to assess the impact of convenience and packaging attributes relative to other oyster attributes during the initial purchase decision. Market share simulations were used to effectively visualise and summarise the quantitative effects of the examined seafood attributes. After price, preparation format was the strongest barrier to oyster consumption with an unanimous consumer preference for pre-opened half-shell oysters. However, retailers considering to offer pre-shucked, easy-open oysters are likely to gain market share compared to unopened oysters, particularly if consumers can experience or are informed through marketing activities about this product innovation. Australian consumers expect oysters to be offered in the traditional black PET pre-moulded tray, they are used to from traditional seafood stores, and alternative self-servicing retail distribution friendly glad wrapped packaging options would have to be offered at a price discount to be competitive. Market potential was shown to exist for adding pre-prepared accompaniments in form of convenient flavour sachets and photographic meal presentations can further increase their preference, addressing consumers’ lack of preparation knowledge. Considering potential licensing costs, health and carbon claims tend to have limited market potential. Differences in consumer preferences allow producers and retailers considerable potential for product differentiation by oyster species and accompaniments. Acknowledgements The authors acknowledge project funding from Australia’s Fisheries Research and Development Cooperation (FRDC). We gratefully acknowledge Nick Dannenberg (former Senior Research Associate at the Ehrenberg-Bass Institute for Marketing Science) for his valuable help during the empirical study.

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

References ABARES (2011). Australian fisheries statistics 2010. Canberra: Australian Government Department of Agriculture, Fisheries and Forestry. Altintzoglou, T., Birch Hansen, K., Valsdottir, T., Odland, J. Ø., Martinsdóttir, E., Brunsø, K., et al. (2010). Translating barriers into potential improvements: The case of new healthy seafood product development. Journal of Consumer Marketing, 27(3), 224–235. Bava, C. M., Jaeger, S. R., & Park, J. (2008). Constraints upon food provisioning practices in ‘busy’ women’s lives: Trade-offs which demand convenience. Appetite, 50(2–3), 486–498. Becker, L., van Rompay, T. J. L., Schifferstein, H. N. J., & Galetzka, M. (2011). Tough package, strong taste: The influence of packaging design on taste impressions and product evaluations. Food Quality and Preference, 22(1), 17–23. Birch, D., & Lawley, M. (2012). Buying seafood: Understanding barriers to purchase across consumption segments. Food Quality and Preference, 26(1), 12– 21. Bose, S., & Brown, N. (2000). A preliminary investigation of factors affecting seafood consumption behaviour in the inland and coastal regions of Victoria, Australia. Journal of Consumer Studies & Home Economics, 24(4), 257–262. Boxall, P. C., & Adamowicz, W. L. (2002). Understanding heterogeneous preferences in random utility models: A latent class approach. Environmental and Resource Economics, 23(4), 421–446. Brécard, D., Hlaimi, B., Lucas, S., Perraudeau, Y., & Salladarré, F. (2009). Determinants of demand for green products: An application to eco-label demand for fish in Europe. Ecological Economics, 69(1), 115–125. Bredahl, L. (2004). Cue utilisation and quality perception with regard to branded beef. Food Quality and Preference, 15(1), 65–75. Brunner, T. A., van der Horst, K., & Siegrist, M. (2010). Convenience food products. Drivers for consumption. Appetite, 55(3), 498–506. Brunsø, K., Verbeke, W., Olsen, S. O., & Jeppesen, L. F. (2009). Motives, barriers and quality evaluation in fish consumption situations: Exploring and comparing heavy and light users in Spain and Belgium. British Food Journal, 111(7), 699–716. Corsi, A. M., Mueller, S., & Lockshin, L. (2012). Let’s see what they have. . .: What consumers look for in a restaurant wine list. Cornell Hospitality Quarterly, 53(2), 110–121. Danenberg, N., & Mueller, S. (2011). Omnibus consumer research findings – Wave 2. Australian Seafood Cooperative Research Centre and the UniSA Ehrenberg-Bass Institute for Marketing Science. de Boer, M., McCarthy, M., Cowan, C., & Ryan, I. (2004). The influence of lifestyle characteristics and beliefs about convenience food on the demand for convenience foods in the Irish market. Food Quality and Preference, 15(2), 155–165. Dellaert, B. G. C., Arentze, T. A., Bierlaire, M., Borgers, A. W. J., & Timmermanns, H. J. P. (1998). Investigating consumers’ tendency to combine multiple shopping purposes and destinations. Journal of Marketing Research, 35(May), 177– 188. Dhanda, K., & Hartman, L. (2011). The ethics of carbon neutrality: A critical examination of voluntary carbon offset providers. Journal of Business Ethics, 1–31. Dolnicar, S., & Leisch, F. (2010). Evaluation of structure and reproducibility of cluster solutions using the bootstrap. Marketing Letters, 21(1), 83–101. Fernández-Polanco, J., Mueller Loose, S., & Luna, L. (in press). Are retailers’ preferences for seafood attributes predictive for consumer wants? Results from a choice experiment for Seabream (Sparus aurata). Aquaculture Economics & Management. Fitzsimons, G. J., Hutchinson, J. W., Williams, P., Alba, J. W., Chartrand, T. L., Huber, J., et al. (2002). Non-conscious influences on consumer choice. Marketing Letters, 13(3), 269–279. FRDC (2002). Retail sale and consumption of seafood. Revised edition. Deakin, ACT, Australia: Fisheries Research Development Corporation. FRDC (2004). What’s so healthy about seafood? A guide for seafood marketers. Deakin, ACT, Australia: Fisheries Research Development Corporation. Gao, Z., & Schroeder, T. C. (2009). Effects of label information on consumer willingness-to-pay for food attributes. American Journal of Agricultural Economics, 91(3), 795–809. Geeroms, N., Verbeke, W., & Van Kenhove, P. (2008). Consumers’ health-related motive orientations and ready meal consumption behaviour. Appetite, 51(3), 704–712. Gempesaw, C. M. I., Bacon, J. R., Wessells, C. R., & Manalo, A. (1995). Consumer perceptions of aquaculture products. American Journal of Agricultural Economics, 77(5), 1306–1312. Girard, S., & Mariojouls, C. (2003). French consumption of oysters and mussels analysed within the European market. Aquaculture Economics & Management, 7(5–6), 319–333. Grebitus, C., Menapace, L., & Bruhn, M. (2011). Consumers’ use of seals of approval and origin information: Evidence from the German pork market. Agribusiness, 27(4), 478–492. Grisolía, J. M., López, F., & Ortúzar, J. d. D. (2012). Sea urchin: From plague to market opportunity. Food Quality and Preference, 25(1), 46–56. Hanson, T., House, L., Sureshwaran, S., Posadas, B., & Liu, A. (2003). Opinions of U.S. consumers toward oysters: Results of a 2000–2001 survey. Office of Agricultural Communications, Division of Agriculture, Forestry, and Veterinary Medicine, Mississippi State University.

503

Hensher, D. A. (2010). Attribute processing, heuristics, and preference construction in choice analysis. In S. Hess & A. Daly (Eds.), Choice modelling: The state-of-theart and the state-of-practice. UK: Emerald Bingley. Hicks, D., Pivarnik, L., & McDermott, R. (2008). Consumer perceptions about seafood – An internet survey. Journal of Foodservice, 19(4), 213–226. Islam, T., Louviere, J. J., & Burke, P. F. (2007). Modeling the effects of including/ excluding attributes in choice experiments on systematic and random components. International Journal of Research in Marketing, 24(4), 289–300. Jaffry, S., Pickering, H., Ghulam, Y., Whitmarsh, D., & Wattage, P. (2004). Consumer choices for quality and sustainability labelled seafood products in the UK. Food Policy, 29(3), 215–228. Johnston, R. J., & Roheim, C. A. (2006). A battle of taste and environmental convictions for ecolabeled seafood: A contingent ranking experiment. Journal of Agricultural and Resource Economics, 283–300. Kim, R. (2008). Japanese consumers’ use of extrinsic and intrinsic cues to mitigate risky food choices. International Journal of Consumer Studies, 32(1), 49–58. Kole, A., Altintzoglou, T., Schelvis-Smit, R., & Luten, J. (2009). The effects of different types of product information on the consumer product evaluation for fresh cod in real life settings. Food Quality and Preference, 20(3), 187–194. Kow, F., Yu, L., FitzGerald, D., & Grewal, D. (2008). Understanding the factors related to the consumers’ choices of oysters in Australia: An empirical study. Journal of Foodservice, 19(4), 245–253. Kris-Etherton, P. M., Harris, W. S., & Appel, L. J. (2002). Fish consumption, fish oil, omega-3 fatty acids, and cardiovascular disease. Circulation, 106(21), 2747–2757. Lähteenmäki, L. (2013). Claiming health in food products. Food Quality and Preference, 27(2), 196–201. Lancsar, E., Louviere, J. J., & Flynn, T. N. (2007). Several methods to investigate relative attribute impact in stated preference experiments. Social Science & Medicine, 64(8), 1738–1753. Liu, Y., Kow, F., Grewal, D., & FitzGerald, D. (2006). Consumer purchase behaviour for oysters: An empirical study in some state capital cities of Australia. International Journal of Consumer Studies, 30(1), 85–94. Lockshin, L., Jarvis, W., d’Hauteville, F., & Perrouty, J.-P. (2006). Using simulations from discrete choice experiments to measure consumer sensitivity to brand, region, price, and awards in wine choice. Food Quality and Preference, 17(3–4), 166–178. Lockshin, L., Mueller, S., & Louviere, J. (2010). The influence of shelf information on consumers’ wine choice. In 5th international Academy of Wine Business Research conference, Auckland, NZ. Louviere, J. J., & Islam, T. (2008). A comparison of importance weights/measures derived from choice-based conjoint, constant sum scales and best-worst scaling. Journal of Business Research, 61, 903–911. Magidson, J., & Vermunt, J. K. (2007). Removing the scale factor confound in multinominal logit choice models to obtain better estimates of preference. In Sawtooth Symposium Conference Proceedings, Santa Rosa, California. Mahon, D., Cowan, C., & McCarthy, M. (2006). The role of attitudes, subjective norm, perceived control and habit in the consumption of ready meals and takeaways in Great Britain. Food Quality and Preference, 17(6), 474–481. Mtimet, N., & Albisu, L. M. (2006). Spanish wine consumer behavior: A choice experiment approach. Agribusiness, 22(3), 343–362. Mueller, S., Lockshin, L., & Louviere, J. J. (2010a). What you see may not be what you get: Asking consumers what matters may not reflect what they choose. Marketing Letters, 21(4), 335–350. Mueller, S., Lockshin, L., Saltman, Y., & Blanford, J. (2010b). Message on a bottle: The relative influence of wine back label information on wine choice. Food Quality and Preference, 21(1), 22–32. Mueller, S., Osidacz, P., Francis, I. L., & Lockshin, L. (2010c). Combining discrete choice and informed sensory testing in a two-stage process: Can it predict wine market share? Food Quality and Preference, 21(7), 741–754. Mueller Loose, S., & Remaud, H. (2013). Impact of corporate social responsibility claims on consumer food choice. British Food Journal, 115(1). Available from: . Mueller Loose, S., & Szolnoki, G. (2012). Market price differentials for food packaging characteristics. Food Quality and Preference, 25(2), 171–182. Myrland, Ø., Trondsen, T., Johnston, R. S., & Lund, E. (2000). Determinants of seafood consumption in Norway: Lifestyle, revealed preferences, and barriers to consumption. Food Quality and Preference, 11(3), 169–188. NHMRC (2003). Dietary guidelines for Australian adults. Commonwealth of Australia. Norwood, F. B., & Lusk, J. L. (2011). Social desirability bias in real, hypothetical, and inferred valuation experiments. American Journal of Agricultural Economics, 93(2), 528–534. Olsen, N. V., Menichelli, E., Sørheim, O., & Næs, T. (2012). Likelihood of buying healthy convenience food: An at-home testing procedure for ready-to-heat meals. Food Quality and Preference, 24(1), 171–178. Olsen, S. O. (2003). Understanding the relationship between age and seafood consumption: The mediating role of attitude, health involvement and convenience. Food Quality and Preference, 14(3), 199–209. Olsen, S. O. (2004). Antecedents of seafood consumption behavior. Journal of Aquatic Food Product Technology, 13(3), 79–91. Olsen, S. O., Scholderer, J., Brunsø, K., & Verbeke, W. (2007). Exploring the relationship between convenience and fish consumption: A cross-cultural study. Appetite, 49(1), 84–91. Onozaka, Y., & McFadden, D. T. (2011). Does local labeling complement or compete with other sustainable labels? A conjoint analysis of direct and joint values for

504

S. Mueller Loose et al. / Food Quality and Preference 28 (2013) 492–504

fresh produce claim. American Journal of Agricultural Economics, 93(3), 693– 706. Pieniak, Z., Verbeke, W., Scholderer, J., Brunsø, K., & Olsen, S. O. (2007). European consumers’ use of and trust in information sources about fish. Food Quality and Preference, 18(8), 1050–1063. Plassmann, H., O’Doherty, J., Shiv, B., & Rangel, A. (2008). Marketing actions can modulate neural representations of experienced pleasantness. Proceedings of the National Academy of Sciences of the United States of America, 105(3), 1050– 1054. Rayner, M., Boaz, A., & Higginson, C. (2001). Consumer use of health-related endorsements on food labels in the United Kingdom and Australia. Journal of Nutrition Education and Behavior, 33(1), 24–30. Rortveit, A. W., & Olsen, S. O. (2009). Combining the role of convenience and consideration set size in explaining fish consumption in Norway. Appetite, 52(2), 313–317. Scarpa, R., Philippidis, G., & Spalatro, F. (2005). Product-country images and preference heterogeneity for Mediterranean food products: A discrete choice framework. Agribusiness, 21(3), 329–349. Schmidt, E. B., Skou, H. A., Christensen, J. H., & Dyerberg, J. (2000). N-3 fatty acids from fish and coronary artery disease: Implications for public health. Public Health Nutrition, 3(1), 91–98. Scholderer, J., & Grunert, K. G. (2003). Promoting seafood consumption: An evaluation of the Danish campaign for fresh fish. In J. B. Luten, J. Oehlenschläger, & D. Ólafsdóttir (Eds.), Quality of fish from catch to consumer: Labelling, monitoring and traceability. Wageningen: Wageningen Academic Publishers. Scholderer, J., & Grunert, K. G. (2005). Consumers, food and convenience. The long way from resource constraints to actual consumption patterns. Journal of Economic Psychology, 26(1), 105–128. Semenou, M., Courcoux, P., Cardinal, M., Nicod, H., & Ouisse, A. (2007). Preference study using a latent class approach. Analysis of European preferences for smoked salmon. Food Quality and Preference, 18(5), 720–728.

Sharma, S., Shimp, T. A., & Shin, J. (1995). Consumer ethnocentrism: A test of antecedents and moderators. Journal of the Academy of Marketing Science, 23(1), 26–37. Street, D., & Burgess, L. (2007). The construction of optimal stated choice experiments: Theory and methods. Hoboken, NJ: John Wiley & Sons, Inc.. UNIC (2003). New Brunswick oyster aquaculture industry market study. Moncton, CA: Unic Marketing Group Ltd.. USDA (2010). Dietary guidelines for Americans. Washington, DC: U.S. Government Printing Office. Vanclay, J. K., Shortiss, J., Aulsebrook, S., Gillespie, A. M., Howell, B. C., Johanni, R., et al. (2011). Customer response to carbon labelling of groceries. Journal of Consumer Policy, 34(1), 153–160. Verbeke, W., Sioen, I., Pieniak, Z., Van Camp, J., & De Henauw, S. (2005). Consumer perception versus scientific evidence about health benefits and safety risks from fish consumption. Public Health Nutrition, 8(4), 422–429. Verbeke, W., & Vackier, I. (2005). Individual determinants of fish consumption: Application of the theory of planned behaviour. Appetite, 44(1), 67–82. Verbeke, W., Vermeir, I., & Brunsø, K. (2007). Consumer evaluation of fish quality as basis for fish market segmentation. Food Quality and Preference, 18(4), 651–661. Vermunt, J. K., & Magidson, J. (2008). LG-Syntax (TM) users’s guide: Manual for latent GOLD 4.5 Syntax module. Belmont, MA: Statistical Innovations Inc.. Welch, A., Lund, E., Amiano, P., Dorronsoro, M., Brustad, M., Kumle, M., et al. (2002). Variability of fish consumption within the 10 European countries participating in the European Investigation into Cancer and Nutrition (EPIC) study. Public Health Nutrition, 5(6b), 1273–1285. Wessells, C. R., Johnston, R. J., & Donath, H. (1999). Assessing consumer preferences for ecolabeled seafood: The influence of species, certifier, and household attributes. American Journal of Agricultural Economics, 81(5), 1084. Wessells, C. R., Kline, J., & Anderson, J. G. (1996). Seafood safety perceptions and their effects on anticipated consumption under varying information treatments. Agricultural and Resource Economics Review, 25(1), 12–21. WHO (2012). Promoting a healthy diet for the WHO Eastern Mediterranean region: User-friendly guide. Cairo: World Health Organisation.