Product satisfaction in food choice is multiple-reference dependent: Evidence from an in-store non-hypothetical consumer experiment on bread

Food Quality and Preference 56 (2017) 8–17

Contents lists available at ScienceDirect

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

Product satisfaction in food choice is multiple-reference dependent: Evidence from an in-store non-hypothetical consumer experiment on bread Carl Johan Lagerkvist a,⇑, Anne Normann b, Annika Åström b a b

Department of Economics, Swedish University of Agricultural Sciences, PO Box 7013, 750 07 Uppsala, Sweden SP Food and Bioscience, PO Box 5401, Frans Perssons väg 6, 402 29 Göteborg, Sweden

a r t i c l e

i n f o

Article history: Received 1 June 2016 Received in revised form 20 September 2016 Accepted 20 September 2016 Available online 21 September 2016 Keywords: Goals Multiple reference Consumer behaviour Field experiment Decision systems

a b s t r a c t Consumer behaviour is goal-orientated. In food product research, goals as cognitive constructs have been shown to translate through the product into an evaluation of product attributes and onto actual choice. In relation to food consumer behaviour and food product choice, however, the manner by which goals operate on post-purchase affective states (need fulfilment) has been largely unexplored. This study examined how food product attributes relate to consumer satisfaction and how this association differs along the goal gradient. We posited that goals are translated through the target object (the product) into a satisfaction representation of product attributes of the identified product. Based on tri-reference point (TRP) goal dependency and the Kano approach to satisfaction measurement, we then analysed the product attribute satisfaction that characterised different goal levels using data collected in an in-store, non-hypothetical consumer experiment with a random sample of 229 consumers. The existence of TRP dependence on product attribute satisfaction was strongly supported, indicating that need fulfilment was directed by transitions across goal reference states. Moreover, a lack of direct proportionality between goal valuation and the instrumentality of the product attributes as means to need fulfilment was identified. These results have normative implications for food product development and research in terms of targeting consumer needs. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction It is well known that people’s behaviour, including that relating to their role as a consumer, is goal orientated. The relationship between motivation and goals has long been documented within the marketing area (e.g. Pieters & Wedel, 2007) and in the field of consumer decision making (Bettman, Luce, & Payne, 1998). However, goals involve no specific action, although goal attainment may ultimately be attempted by performing a variety of reasoned and unreasoned actions. For food consumers, goals as motivators can refer to what the product has (or does not have) in terms of features, but it can alternatively appeal to wide ranges of standards of consumer expectations in terms of need fulfilment and, ultimately, the satisfaction derived from aligning goal attainment and behavioural action. Consumers are therefore exposed to advertisements that appeal to wide ranges of standards of expectations, such as ‘‘Boost yourself to perfection”, ‘‘. . .for a special ⇑ Corresponding author. E-mail address: [email protected] (C.J. Lagerkvist). http://dx.doi.org/10.1016/j.foodqual.2016.09.006 0950-3293/Ó 2016 Elsevier Ltd. All rights reserved.

occasion”, ‘‘Enjoy plant power” and ‘‘Good for you”, each with a connotation with product performance, ultimately aiming to generate an post-purchase affective state of satisfaction to induce brand or product loyalty. Consumers’ product evaluation for low-involvement products is known to be dynamic and determined by habitual goals (e.g. Verplanken & Aarts, 1999) or based on goals triggered by information and context immediately available to the consumer in the form of arrays of cues (Hamlin, 2010) such as brand, colour, price, etc. Together with product conceptualisations in the form of conscious or unreasoned meaning assigned to what is perceived (Thomson, 2010), such considerations of goals influence the desirability of products and eventually become manifested through intentions to attain the goal (Baumgartner & Pieters, 2008). Goal-based choice studies, where in a connectionist network consumers evaluate products, have emerged recently (Stijn, van Osselaer, & Janiszewski, 2012). In an editing and evaluation process, goals as cognitive constructs are translated through a target object or a set of target objects as sub-goals (the product) into a representation of identified product attribute preferences

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17

(instrumental means) and on to actual choice (Lagerkvist, Normann, & Åström, 2015). According to goal setting theory (Kruglanski et al., 2002), goals represent cognitive constructs having distinct motivational meaning. Goal systems are then defined to represent systematic networks of mentally desirable end-states (super-ordinate goals) which are inter-connected with their means of achievement and to alternative super-ordinate goals. Goal-attainment as a motivational construct thereby involve management of goal conflicts as well as means choice and substitution that might be activated during goal pursuit. The concept of means relates to ‘‘any activity, event or circumstance perceived as likely to contribute to goal progress” (Kopetz, Kruglanski, Arens, Etkin, & Johnson, 2012, p. 212). A single goal might be associated with multiple instrumental means, or multiple simultaneous goals sharing some means may exist at the same time, thus implying existence of means substitution effects. Moreover, alternative goals might exist with or without a common set of means, thus leading to potential cognitive conflicts (Hull, 1938). For instance, the goal of buying bread could include various sub-goals such as a specific purpose (e.g. bread for breakfast) or bread with certain functional (i.e. what the bread can do for the consumer), emotional and/or hedonic features. The various instrumental means would then represent certain product salient features or attributes. Fig. 1 depicts a possible multiple goal system with two alternative goals and a set of interrelated and inhibitory means. Goal setting theory posits that goal activation triggers activation of specific goal-means configurations (i.e. actions). Such configurations can be unique and stable in being formed by goalmeans associations established by the individual to be instrumental (i.e. habits), or non-unique, leading to alternative pursuits to its attainment. The extent of desirability and the strength of goalmean associations in goal considerations then come to relate to a value attached to the goal itself and a transfer of value from the goal to the means, with the consequence of establishing a preference structure for certain means over others (Zhang, Fishbach, & Kruglanski, 2007). Existing research confirms that a goal might serve as a reference point, as it divides potential outcomes into regions of success or failure (Heath, Larrick, & Wu, 1999), and that goals inherit the properties of loss aversion and diminishing sensitivity to value changes as proposed by prospect theory (Kahneman & Tversky, 1979). Analogously, a range of multiple goals (i.e. on a goal gradient or a range of motivational phenomena) would generate multiple reference points. This is consistent with the fundamental allocational property of goal systems by which more or less mental

9

resources are allocated to a given goal attainment depending on where on the goal gradient the goal is located (Kruglanski et al., 2002). Recently, Lagerkvist et al. (2015) found that the characteristics of the goal-value function extend to a food product evaluation situation with multiple goals, where three goals serve as reference points (a base level as status quo (SQ)), striving to reach a target level (TL), while simultaneously seeking to avoid a bottom line (referred to as the minimum requirement, MR). The subjective value function was found to be asymmetric in changes of goals along the goal gradient, which then become partitioned into four regions: Success (x > TL), gain (x > SQ), loss (MR < x < SQ) and failure (x < MR). Multiple goal reference points thus imply that consumers who are below a certain goal, say the MR level, will work harder to attain certain product features or to obtain a more favourable overall impression to increase their achievement than consumers who are above this MR level. Goals as reference points then implies that mean choices and preference formation among competing products (as sub-goals and means) that are considered acceptable but which differ in some means to attainment are formed differently along the continuum from bottom line to target level and beyond, and therefore should differ in desirability. This refers to the substitutability relations between various means along the goal gradient. Additional criteria for the identification of goal reference points are that they reflect the mental values of specific end-states rather than general desires and depend on the choice environment (Förster, Liberman, & Friedman, 2007). Together, these criteria suggest that need fulfilment as motivation and need satisfaction as attainment are ‘‘inextricably intertwined” (Oliver, 2010, p. 140). Moreover, the criteria suggest that people may have a set of expectations ranging from the worst acceptable to the ideal, and a corresponding set of post-purchase affective states to such desired standards (Santos & Boote, 2003).

1.1. Present study There has been little research on goal-directed product satisfaction, but it is widely accepted that if people commit to goals, the goal alters their values (e.g. Locke & Latham, 1990). Therefore, we considered the following question for multiple goals as reference points for satisfaction: How are different attributes (as means) for a given food product differently related to customer satisfaction and how does this relation differ along the goal gradient? The sense of achievement can be expected to differ along the goal gradient, as consumers may have different goal-directed

Fig. 1. A system of multiple goals and means (own adaption to a multiple goal system from Kruglanski et al., 2002). Facilitative links relate vertically connected elements (i.e. goals and their means of attainment). Lateral links exists between competing (substitution) and/or common (complementary) sub-goals and/or means. Note: dashed lines indicate the systemic structural effect from having goal 2 to represent an alternative to goal 1, sharing three sub-goals and with indirect relationships to certain means to attainment. Sub-goal 4 and Mean 6 is unique to goal 2 (i.e. inhibitory to the goal 1 system). Several goals, sub-goals and means can be added to reflect an even more elaborate goal system hierarchy.

10

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17

standards for both product attributes and overall impressions relating to the characteristics of the goal-value function. The manner by which goals operate on post-purchase affective states (need fulfilment) in relation to food behaviour and food product choice has been largely unexplored in previous consumer research. In this study, we posited that goals are translated through the target object (the product) to reflect the goal-value function, but then also further reflected into a satisfaction representation of product attributes of the identified product. The first contribution of this study therefore lies at the intersection of goal setting theory (multiple reference points in food choice) and product research in the form of need fulfilment formation for food products. The second contribution is to examine whether the effect of goal pursuit on consumer satisfaction of food choice is consistent with the predictions of a multiple-reference value function. This part of the study questions the prevailing assumption about direct proportionality between goal valuation and the instrumentality of the product attributes as means to goal attainment. 2. Material and methods This study draws on decision-based data from an in-store, nonhypothetical experiment on bread choices which was conducted during May-June 2013. The study was part of a larger project on goal-means configurations and testing of tri-reference point (TRP) theory for food choice. The pre-decision data were published by Lagerkvist et al. (2015). An in-store or similar location provides several advantages in food choice and marketing research because of the realism introduced by using a real shopping environment, real goods and real payments (Jaeger et al., 2005; Xue, Mainville, You, & Nayga, 2010). Bread was selected for the purpose of the larger study, as it is a product category with a wide variety of products ranging from basic and inexpensive to high-quality items. 2.1. Participants and design Participants were recruited from the national consumer database at the Swedish Institute for Food and Biotechnology (SIK). This database is maintained on a rotational basis to provide a representative sample of Swedish food consumers. An e-mail invitation to participate in the study was sent to all consumers listed (approximately 1500). The invitation was brief and asked addressees about their interest in participating in a study at their choice of supermarket outlet. Respondents were initially screened on their purchasing frequency for four product categories (bread, fruit & vegetables, dairy, and cold cuts) and on whether they alone or together with someone else in the household are responsible for food purchases (purchasing bread at least once per month and being at least partly responsible was used as the cut-off). An online background questionnaire to collect demographic information was

activated for those consumers who agreed to take part in the study. Consumers were also asked to select an appointment time for participation in the experiment that coincided with their usual food shopping from a set of alternatives that included weekdays (between 10.00 and 19.00) and Saturdays (14.00–19.00). Supermarkets within the same chain have a similar store layout and typically keep the same range within product lines. Hence, it was deemed reasonable to expect any given consumer participating in the study to have a fair chance of being familiar with the range of bread sold at their usual supermarket. Three supermarket alter natives were provided, corresponding to the three major food retailers in Sweden (ICA, COOP and Axfood). The average time taken for the experimental session was approximately 25 min. A total of 229 complete responses were obtained. 2.2. Procedure After developing and pre-testing instructions and the experimental procedure with a group of 10 consumers (data not included in this study due to their preliminary nature), the experiment was divided into five stages. The priming to TRP ‘mere’ goals (i.e. goals that are specific and challenge levels of performance) through the instructions (Table 1) meant that participants were provided with self-control over the choice of stimuli to be used in the subsequent elicitation of salience. Therefore, the reference points in the experiment were endogenous in that the consumers themselves were asked to identify and characterise each bread of their choice based on their aspirations, and in doing so they were not directed towards any specific type of bread. Recent research has demonstrated that goal priming influences decision making during a choice process (Carlson, Tanner, Meloy, & Russo, 2014). The procedure used in the experiment was as follows: (i) On arrival at the supermarket, participants were met and told that they would be provided with further instructions when they came to the bread section within the store. They were also told to proceed with their shopping as normal meanwhile. (ii) At the beginning of the actual experiment, participants were provided with the instruction, depending on manipulation (Table 1), to select their Bread 1. Bread 1 was to be identified so as to represent the baseline alternative (i.e. Status Quo, henceforth SQ) from which transitions on the goal gradient were to be initiated. In order not to induce forced choice, participants were free to take the time they needed to find the bread associated with this first instruction. (iii) Participants were instructed that they were to make two subsequent choices of bread (Bread 2 and Bread 3) (Table 1) and that they would then be asked to make a random 50/50 draw to decide which of the two breads they would be eligible to purchase. Participants were informed at this point that

Table 1 Operationalisation of multiple reference points within the field experiment by samples. Breadchoice

Treatment 1 Manipulation (SQ induced to be below MR)

Treatment 2 Manipulation (SQ induced to be above MR, but below TL)

Bread 1 (Status quo, SQ)

Please find a bread which you would normally not buy. This should be the bread that you regard as definitely not ‘‘fitting your standards” and therefore does not meet the requirements you have for any bread that you buy

Please find the bread that you typically buy – this is your ‘‘usual” bread!

Bread 2 (Minimum requirement, MR)

Please now find a bread you consider slightly better than the one that you just selected – Select a bread that you would be willing to buy, but it should not be better than just qualifying for purchase

Please now find a bread that you are willing to buy but which is just being acceptable for purchase, but no more

Bread 3 (Target level, TL)

Finally, please choose the bread that fully meets all your requirements for bread – this is bread for which you feel that ‘‘little extra something” and which you are more than happy to buy Note: There might be other breads that you consider better than the one you have actually chosen, but the bread of your choice has everything that you really are looking for and it is such that you are willing to buy it

11

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17

(iv)

(v) (vi)

(vii)

on completion of the choice of the two additional breads, they would be provided with a gift voucher for SEK 100 as part of the total compensation for this part of the study and that they could decide for themselves how much of this amount they would use to pay for the bread of their choice or for any other shopping. At this stage participants were then informed that after their final draw, they were free to keep and buy Bread 1 instead. Participants were presented with instructions about selecting Bread 2 (MR level). Instructions for the MR level were framed in the parlance of the direction of each transition. This meant that in treatment 1, the wording asked participants to search for the bread that just qualified as purchasable, whereas in treatment 2 the wording asked participants to search for bread that was just acceptable for purchase. Participants were instructed about selecting Bread 3 (TL level). Upon completion of choice of Bread 3 and after the draw to determine which bread was eligible for purchase, participants were told that they would be met after the cashier and asked to show the bread which they decided to purchase and the receipt for the purchase of their choice. This was to ensure that the experiment actually led to a real purchase. Participants were also told that they should be ready to answer a final questionnaire, thus qualifying for the full SEK 300 compensation for their participation. Based on the final choice of bread and after purchase (actual payment), a Kano questionnaire was used to measure product satisfaction.

During the experiment, enumerators took notes about the breads that each subject selected; this information included type of bread (six categories), price per pack and pack weight. Participants were not asked to self-report this information, because this could have generated anchoring effects. In the pre-decision study dealing with pre-purchase product evaluation and choice features, check-all-that-apply (CATA) forms were collected after each completion of actual choice (i.e. steps ii, iv and v). However, the Kano questionnaire (post-purchase satisfaction set) was based on product features not used in the CATA form, as it is prudent to assume that choice criteria differ from features that determine satisfaction (Oliver, 2010). 2.3. Measures In an early study, Hull (1938) noted that goals are positive or negative reinforcers (depending on whether there is an enhancement or avoidance of a contentment state) in the sense that there could be a motivational response from a goal-directed pursuit because of a sense of achievement. In other words, consumer needs acts as motivators because the individual is driven by a desired end-state (i.e. need fulfilment) with a corresponding outcome in form of contentment (extent of satisfaction) or resentment (dissatisfaction). Satisfaction and dissatisfaction as the extent of need fulfilment then serve as a motivational classification in relation to the rewarding or deficit property in response to a stimulus (positive or negative reinforcer). The need satisfaction framework established by Maslow (1943) and, in particularly the reinforcement theory approach by Herzberg (1968) suggest that satisfaction and dissatisfaction are unipolar but independent and simultaneous (at least to some extent). From a product perspective, this means that the extent to which product attributes acts as need fulfilling, or not, can be studied from the perspective of which need category that each attribute falls into. Attributes can then be classified as bivalent satisfiers (can cause both satisfaction and dissatisfaction, monovalent dissatisfiers (causing dissatisfaction only when not

being present in the product), or as monovalent satisfiers (corresponding to a higher need state which can only relate to satisfaction) (Oliver, 2010, p.149). From a product optimization perspective there might then exist dissatisfying attributes for which an independent removal would be possible. It might also be possible to add positively reinforcing attributes. Oliver (1993) first suggested that product quality is an antecedent to satisfaction. From this, it follows that the satisfaction response should be associated with a particular focus (e.g. a product, an expectation, etc.) and that the response should then be viewed either in structural terms of the type or number of events or product features leading up to an outcome or, alternatively, as an overall impression. For the former part, the focus is on uniqueness and strengths of associations within the goal system. This focus allows for the satisfaction/dissatisfaction classification of each product attribute in relation to its need fulfilment. For the latter part, established definitions of satisfaction in relation to a product or a service propose that it arises from a comparison between expectations and outcomes, and therefore depends on the gap between the perceived and experienced quality of the product to which these aspects relate (e.g. Cardello, Schutz, Snow, & Lesher, 2000; Oliver, 1980). Satisfaction was measured in two ways. First, the relationship between need-fulfilling product attributes and customer satisfaction was evaluated as originally detailed in the quality management literature. The seminal work by Kano, Seraku, Takahashi, and Tsuji (1984), which corresponds to the reinforcement theory approach by Maslow (1943) and Herzberg (1968), acknowledges the independent ability of product attributes to satisfy and dissatisfy. This is attractive when seeking to understand the structure by which need-related product attributes affect satisfaction and dissatisfaction along a goal gradient, as it will detail the direction and strength of the satisfaction process. A Kano questionnaire (Berger et al., 1993; Kano et al., 1984; Rivière, Monrozier, Rogeaux, Pagès, & Saporta, 2006; Sauerwein, Bailom, Matzler, & Hinterhuber, 1996) was developed to measure how different food product attributes differently related to customer satisfaction. A focus group discussion with a group of 10 consumers generated inputs on the product associations consumers have when buying bread in supermarkets and on the product features that would characterise bread along the quality gradient. The Kano questionnaire was then developed by the research team together with representatives from two major bread-baking companies in Sweden for identification of need-fulfilling product features. The Kano questionnaire provides data to classify each product feature according to the Kano categories (attractive (A); must-be (M); one-dimensional (O); indifferent (I); reverse (R); and questionable (Q)) according to each reference point.

Table 2 Product features included in the Kano questionnaire. Product feature 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Unsweetened With fruits and/or nuts With additives/preservatives High sugar content With whole grain Rich in fibre Based on sourdough Labelled with bake date Labelled with best-before date Pre-sliced Sold in large pack Sold in small pack Compact texture Airy texture Filling

12

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17

Please use the tables below and answer each question in relation to the bread that you have just chosen: If the bread is baked  I like it that way on sourdough, how  This is a basic requirement for me Functional form of do you feel?  I am neutral the question  I can live with it that way  I dislike it that way

Dysfunctional form of the question

If the bread is not baked on sourdough, how do you feel?

    

I like it that way This is a basic requirement for me I am neutral I can live with it that way I dislike it that way

How do you rank this bread in relation to its sourdough content?















1

2

3

4

5

6

7 Excellent

Totally unsatisfactory

Fig. 2. Illustration of a functional and dysfunctional question in the Kano questionnaire, followed by the ranking of the product feature.

Table 2 shows the 15 product features that were included, in both their functional and dysfunctional forms. Features were presented in a randomised order without the possibility to go back to revise responses to earlier questions. An example of a product feature task used for the Kano questionnaire is given in Fig. 2. Second, to obtain a measure of the gap between the perceived and experienced quality of the product to which these aspects relate, participants were asked to rank their satisfaction with each product feature using a 7-item feature satisfaction score (FSi,) (1 = totally unsatisfactory; 7 = excellent). This was the first step in measuring how the product feature associations differ along the goal gradient. This measure is similar to the overall food satisfaction measure recently used by Vad Andersen and Hyldig (2015). However, in order to obtain a comparative standard for the gap analysis, upon completion of the Kano questionnaire participants in the present study were also asked to assign importance to each of the 15 product features in relation to the bread that they regularly buy, using a 7-item feature importance score FIi (1 = completely unimportant; 7 = extremely important). Apart from providing demographic data, in order to test whether participants differed across treatments (sampling bias) and whether final choices could be expected to depend on key drivers to goal-means pursuit, the first online questionnaire also included the self-reported habit index measure (SRHI) (Verplanken & Orbell, 2003) and the personal involvement scale (PI) (Zaichkowsky, 1994). The SRHI measure was framed as ‘‘buying bread is something that. . .”, while the PI measure was framed as: ‘‘for me bread is. . .”. Furthermore, five of the sources from the product involvement measure in Mittal and Lee (1989) that refer to an interest in a goal-object were included. These were: (a) product involvement, (b) brand decision involvement, (e) product hedonic value, (g) product utility and (h) brand risk, each relating to utilitarian, sign-value and hedonic goals. Each source had three statements which were answered on a 7-item agreement range.

2.4. Data analysis First, independent sample t-tests (for differences in mean age) and chi-square tests (for gender and self-reported purchase frequencies of bread) were applied to examine differences between

treatment groups. Differences in self-reported habits, personal involvement and product involvement were examined through a series of independent-samples Kruskal-Wallis tests. Data from the Kano questionnaire were organised according to the satisfaction category scheme developed by Oliver (2010) as detailed in Appendix 1. For each product feature, Kano satisfaction/dissatisfaction indices were then computed following Oliver (2010, p. 154) as:

ðyÞ ¼ Satisfaction potential ¼

AþO AþOþMþI

ðxÞ ¼ Dissatisfaction potential ¼ 

OþM AþOþMþI

ð1Þ ð2Þ

These indices are obtained on the ranges ½0; 1 and ½1; 0, respectively and can therefore be plotted as coordinates on a two-dimensional (x,y) graph. The satisfaction and dissatisfaction potentials provide information that permits comparison of the ability of product features to satisfy or dissatisfy, or the degree to which they can do both. Next, for the purpose of examining how the product feature satisfaction differed across the goal-reference points, we calculated for each feature the pair-wise Euclidean distance between the satisfaction/dissatisfaction coordinates following the Pythagorean Theorem. In the Euclidian plane, if p0 and q0 represent two coordinates in the (x,y) space so that p0 ¼ ðx1 ; y1 Þ and q0 ¼ ðx2 ; y2 Þ, then qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the distance is given by dðp; qÞ ¼ ðx2  x1 Þ2 þ ðy2  y1 Þ2 . A quality index gap measure, Qi, i = 1,. . .,15, was calculated for each feature as the difference between scores on the feature satisfaction measure FSi, and the feature importance score FIi. 3. Results The outcome of the experiment was that 75 participants selected the bread chosen as their minimum requirement, while 61 and 91 participants selected their status quo and target levels choices, respectively. Two participants in treatment 1 were found to have chosen their status quo choice and were therefore not included in the following analysis (as this type of bread was meant to indicate that it was not acceptable for purchase). Furthermore,

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17

six participants failed to report their purchases. Data were pooled across treatments, as the results from the pre-purchase product evaluation study by Lagerkvist et al. (2015) could not reject the assumption that the goal value functions were equal across treatments. Hence, there is no theoretical reason why product satisfaction would differ across treatments. 3.1. Socio-economic comparison across treatments Before analysing the results related to the Kano satisfaction measure and the quality index gap, we first examined whether there were any systematic differences between participants in each treatment by age, gender and self-reported purchase frequencies for bread. We chose to examine this by treatment categories instead of by actual choices, as mean age was 44.6 (SD = 13.3) and 43.1 (SD = 12.2) in treatment 1 and 2, respectively, and equality of means could not be rejected (t = 0.87, p = 0.384). Moreover, there were 70.1% and 77.8% women in treatment 1 and 2, respectively, but gender differences in the proportions could be rejected (v2 = 1.749, p = 0.186). Finally, 82.4% and 86.7% of the participants in treatment 1 and 2, respectively, reported that they typically purchased bread at least once per week. Differences between treatment by purchase frequencies could be rejected (v2 = 5.82, p = 0.121). The results for the self-reported habit index indicated high levels of internal reliability (Cronbach’s alpha 0.91) and a mean score of 49.99 (SD = 15.4), which is indicative of a rather medium level of habitual bread purchases. For the personal involvement, there was high internal reliability (Cronbach’s alpha 0.90) and the mean of 49.4 (SD = 10.3) is indicative of higher than neutral assessment of involvement. For the product involvement, internal reliability ranged from 0.71 (brand risk) to 0.89 (brand decision involvement) and the mean score when pooling the five sources was 70.8 (SD = 13.9), which indicates a rather high level of involvement. 3.2. Satisfaction and quality gaps Fig. 3 shows the satisfaction/dissatisfaction potential by product features and reference points. The feature classification is detailed in Table A2 (Appendix). The five product features ((1) unsweetened; (2) with fruits and/or nuts; (5) with whole grain; (6) rich in fibre; and (7) based on sourdough) were categorised as attractive across the three goal-reference points. As monovalent satisfiers, these features can only satisfy and thus act as delighters, or product attribute wishes, at the three goal levels. It can be

13

expect that absence of these attributes would not reduce satisfaction. Labelling with best-before date was found to have a onedimensional character across the goal-reference points and thus this type of labelling will always contribute to satisfaction. However, the responses indicated a close tie to a must-be requirement for the minimum requirement goal level. The labelling with bake date had a similar response, but here the must-be character was more pronounced for the minimum requirement level. These results suggest that these types of labelling change character from monovalent dissatisfiers (i.e. are taken for granted and for which an absence can only dissatisfy) at lower levels of the goal gradient to bivalent satisfiers (one-dimensional) of essential but less sought-for attributes at higher goal levels. The must-be character (here only for bake date) is related to the distinction between primary and secondary product attributes within the marketing literature (e.g. Keller, 2003). Primary product attributes are defined as those that are essential in providing a solution to a problem (need deficit), while non-essential attributes are secondary. The results from this study can be interpreted as indicating that labelling with bake date works as a positive reinforcement which helps the consumer to surpass the minimum requirement goal level in providing some minimal quality assurance. The results also show that the satisfaction categorisation is more nuanced for some features. Bread sold pre-sliced is attractive at the minimum requirement (MR) and status quo (SQ) levels, but becomes one-dimensional at the target level (TL). A similar pattern was found for the filling feature. There was also an indication that the ‘airy texture’ feature was attractive at both the MR level and the TL, but that this feature had an ambiguous nature at the SQ level. Furthermore, the results show that presence of additives/ preservatives and a high sugar content were characterised as reverse features, meaning that they are not wanted and also that the participants actually expect to have their bread with the opposite characteristics. Bread sold in large packs also had a reverse categorisation at the MR and SQ levels, but for bread at the TL this was an indifferent (i.e. neutral) feature. Finally, the ‘compact texture’ feature was either of an indifferent character or the results were ambiguous. Together with the characterisation of the airy texture feature, these results suggest that the texture of bread might reflect a segmentation of participants. Table 3 shows the Euclidian distances between satisfaction/ dissatisfaction potentials and allows comparison between goalreference points with respect to the location of the product features on the unity space in Fig. 3. The results show that the satisfaction/dissatisfaction potentials are not stable states.

Note: Numbers refers to product features as provided in Table 2. Kano categories: Attractive = A; Must-be = M; One-dimensional = O; Indifferent = I; Reverse and Questionable = R. Fig. 3. Kano satisfaction/dissatisfaction potential of product features and feature classification by goal-reference points.

14

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17 Table 3 Euclidian distance between product feature satisfaction/dissatisfaction by goal-reference points. Feature No.

Product feature

d(MR;SQ)

d(MR;TL)

d(SQ;TL)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Unsweetened With fruits and/or nuts With additives/preservatives High sugar content With whole grain Rich in fibre Based on sourdough Labelled with bake date Labelled with best-before date Pre-sliced Sold in large pack Sold in small pack Compact texture Airy texture Filling

0.177 0.075 0.074 0.036 0.122 0.150 0.118 0.214 0.162 0.118 0.070 0.115 0.232 0.088 0.139

0.147 0.072 0.053 0.223 0.055 0.115 0.086 0.196 0.044 0.054 0.189 0.131 0.126 0.175 0.110

0.160 0.097 0.118 0.258 0.162 0.049 0.108 0.147 0.118 0.159 0.207 0.064 0.137 0.127 0.152

Note: MR = minimum requirement, SQ = status quo; TL = target level.

2.5 2

Quality Index Gap

1.5 1 0.5 0

MR

-0.5

SQ

-1

TL

-1.5

Fig. 4. Relationship between feature satisfaction and feature importance as quality index gap.

The most pronounced differences in locations between the MR and the SQ levels were found for the labelling with bake date (increase in satisfaction potential, with dissatisfaction potential largely constant) and compact texture (increase in both dimensions), respectively. For the comparison between the MR and TL, the satisfaction potentials increased for sugar content and labelling with bake date at the TL. Higher sugar content also increased satisfaction between the SQ level and the TL. Furthermore, bread sold in large packs deducted satisfaction at the TL in comparison with the MR and SQ levels. The unsweetened feature also differed between goal levels, with increasing satisfaction and decreasing dissatisfaction potentials towards the SQ level and both decreasing satisfaction and dissatisfaction potentials towards the TL. Some moderate differences in locations were also found for fibre content (increase in dissatisfaction potential) and labelling with best-before date between the MR and SQ levels (increase in satisfaction and decreased dissatisfaction). The airy texture feature showed a reduction in both the satisfaction and the dissatisfaction potentials from the MR to the SQ level and then from the SQ level

to the TL. Similar types of symmetric reductions were found for the whole grain and filling features. Fig. 4 shows the quality index per product feature and by each goal-reference point as the difference between the ranking of satisfaction with the bread that was chosen and paid for within the experiment and the importance given to each feature for the regular purchases. Interestingly, the labelling with best-before and bake dates were the only two features which participants selecting their choice at the TL ranked on a lower level of satisfaction than their typical choice of breads. These features were also the only two for which the quality index gaps were lower for the TL choices than for the MR and SQ choices. More importantly, Fig. 4 shows that the quality index gap was not constant across the goal-reference points. This is a novel result contradicting the transfer principle in means valuation, which suggests that the magnitude of valuations or devaluations of goal attainments is in direct proportion to the value of the goal itself (Kopetz et al., 2012). Instead, the results in Fig. 4 reflect an important motivational property of goal-means association in that the

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17

instrumentality of means differs in magnitude within a given goal level and, more importantly, between goal levels.

4. Discussion It is widely acknowledged that product development, particularly for fast-moving consumer products such as food, should be based on consumers’ needs and desires. A particular challenge in meeting these demands and wishes is understanding how product choice turns into product satisfaction, typically regarded as the difference between perceived and expected quality of the product, which has become the major measure used for assessing consumer product experience (Tzokas, Hultink, & Hart, 2004). Following Giese and Cote (2000), there are three issues to consider when identifying the components of satisfaction for such an assessment. First, satisfaction is an emotional or cognitive response. Second, the response is related to a certain focus, such as an expectation or an experience. Third, the response refers to a given point in time, which means that the response can be seen as an attainment after consumption or after choice. These three criteria inevitably imply that need fulfilment as motivation and need satisfaction as a response should be interdependent. The present study investigated the manner by which multiple reference-dependent goals operate on post-purchase satisfaction as affective states (need fulfilment) in relation to food behaviour and food product choice. Analysis of how multiple goals along the goal gradient operate on satisfaction is relevant because it establishes a relationship between goals as the driver, or motivation, of behaviour by being the mechanism that directs the selection of need-fulfilling product attributes and governing product evaluation, on the one hand, and experience evaluation in the form of satisfaction, on the other. Goals as reference points in themselves reflect the mental values (emotional and cognitive) of specific desired end-states, through an underlying goal value function. Goals, goal systems and satisfaction therefore translate into a common affective denominator, so the units of analysis should be comparable. This study provides two important results in relation to satisfaction response from a goal-orientated product choice. First, the response focus by Giese and Cote (2000), when interpreted from a goal system analysis perspective, suggests that the satisfaction response may be viewed in structural terms of the type or number of events or product features that relate to a desired outcome. The focus is on uniqueness and strengths of associations between the goal system and the satisfaction response. The Kano approach to satisfaction measurement is then relevant to apply because of its recognition of the independent ability of product attributes to satisfy and dissatisfy. This is attractive when seeking to understand the structure by which need-related product attributes affect satisfaction and dissatisfaction along a goal gradient, as it details the direction and strength of the satisfaction process. Our results suggest that product attribute satisfaction is not stable along the goal gradient. This means that transition along the goal gradient will induce dynamic changes in attribute satisfaction for each attribute. Moreover, the results suggest that there are not just transitions to consider, but also changes in the relative importance of product attributes from a satisfaction and dissatisfaction perspective along the goal gradient. These results have practical implications and implications for food consumer behaviour research. For product developers and alike, the results suggest that product optimisation and marketing can be directed and focused on changes in relative attribute satisfaction and dissatisfaction, instead of considering the product as an aggregate. The goal levels can then be seen as the context in which the behaviour is analysed. In this sense, for example, key features of bread to be bought for a special occasion (target level) can be made more easily

15

distinguishable for the consumer through product messages. The research implications derive from goal-directed pursuit and product attribute satisfaction being intertwined. As the findings from this study and in the recent goal-based choice literature suggest, not only are goals translated into product evaluation and choice, and then further onto the satisfaction response, but there is also a hierarchical nested structure to consider when aiming to understand how consumer preferences are formed, maintained and updated into repeated behaviour. In choice studies, for example, there is a need to consider the role of goal levels as determinants as well as affective expectations and aspirations as antecedents insofar as goals continue to operate onto satisfaction. Second, the most established definition of satisfaction as a hedonic measure in relation to a product or a service rests on the attainment criteria in the definition of satisfaction by Giese and Cote (2000). This suggests that satisfaction as an overall response arises from a comparison between expectations and outcomes, and therefore depends on the gap between the perceived and experienced quality of the product to which these aspects relate (e.g. Cardello et al., 2000; Oliver, 1980). In the present study, a measure of the gap between the perceived and experienced quality of the product to which these aspects relate was obtained by asking participants to rank their perceived satisfaction for a set of key product attributes after purchase. This part of the gap measure was then similar to the overall food satisfaction measure recently used by Vad Andersen and Hyldig (2015), except that we kept the measure at the level of specific product attributes. Our other component, referring to experienced quality, was related to the accumulated experience (i.e. the product the consumer regularly buys). Based on goal setting theory, we expected to find a direct proportionality between goal valuation and the instrumentality of the product attributes as means to goal attainment. This suggests that the gap would be constant along the goal gradient and in direct proportion to the magnitude of goal attainment. This was not what we found. Instead, our results suggest that the effect of goal pursuit on consumer satisfaction of food choice is consistent with the predictions of a multiple-reference value function (Lagerkvist et al., 2015; Wang & Johnson, 2012). Multiple goalreference point dependence explicitly predicts that people in a comprehensive evaluation of objective outcomes will not only draw on changes in subjective values from a status quo goal level, but also strive to reach a target level, while simultaneously seeking to avoid reaching a bottom line. Thus the theory rests on the assumption that the goal value function is asymmetric, meaning that transitions towards the bottom line with be associated with larger loss aversion than the gains from approaching the target level. Our results suggest that the instrumentality of product attributes as means in the goal system differs, because the magnitude of the gap showed non-proportionality along the goal gradient. The gap was more negative, or less positive, for all attributes for bread selected at the minimum requirement goal level, in comparison to the gaps for bread selected as status quo or target level. An exception to this was the gap for the bake date and the best-before date attributes, for which there were very small gap differences across goal levels. Since the gap for these two attributes was close to zero, this means that breads chosen had these features and that the participants usually bought breads with these labels. At another level, the gap analysis suggested that the proportionality due to the goal levels differed across product features. This was especially the case for the comparison between breads at the status quo and target levels. The gap was larger for the attributes related to ‘unsweetened’, ‘with fruits and/or nuts’, ‘with whole grain’, ‘rich in fibre’, and ‘based on sourdough’. For the remaining attributes, there were small discrepancies between

16

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17

attributes depending on whether the goal level was at the status quo or target level. This finding is interesting for at least two reasons: It depicts the attributes that became activated with the goal system network for transitions between the status quo and the target levels and the attributes that did not became differentiating. This is relevant for the motivational property in the goalsystem network and can be used to direct product positioning. Following on from this, another consequence of the results is that they question the approach of using an overall satisfaction measure at the product level. Using such an approach must be seen as a crude approximation, based on the results in this study. An overall product satisfaction measure provides less information, as it is usually not related to the gap between perceived and experienced quality and therefore only includes one side of attainment. The hedonic approach also fails to recognise that as an outcome measure, it is most likely determined, or driven by, not only a sub-set of product attributes but also from the perspective that this sub-set is endogenously determined as an effect of goal pursuit.

5. Conclusions This study shows that goal setting and product evaluations are related and that product attribute satisfaction is goal-dependent. The manner by which goals as reference points operate on consumer behaviour and how satisfaction in relation to a food product is formed and influenced by the mental representation of the rea-

son why that product was chosen are relevant for consumer and product research and also for product development. The reference-dependent goal-mean approach developed in this study is also relevant in food choice research because it integrates the choice itself with the reason or purpose for which the choice was made or considered. The comparative and evaluative aspect of consumer satisfaction induced by multiple reference points has implications that can be exploited within practical marketing when positioning a new product or re-positioning an existing product in order to maximise market penetration. Acknowledgements This research was performed with economic support from the Swedish Research Council FORMAS. We would like to express our particular appreciation to the managers of the three supermarkets in Gothenburg, Sweden, for allowing us to run the experiment at these locations. In addition, we gratefully acknowledge the support in data collection provided by nine staff members of SP Food and Bioscience. This study was presented at the 11th Pangborn Sensory Science Symposium in Gothenburg. We are grateful for the comments and suggestions that we received. Appendix See Table A1.

Table A1 Sample Kano evaluation table. Consumer requirements

Functional

Dysfunctional

I like it that way This is a basic requirement for me I am neutral I can live with it that way I dislike it that way

I like it that way

This is a basic requirement for me

I am neutral

I can live with it that way

I dislike it that way

Q Q R R R

Q Q R R R

A M I I R

A M I I I

O M M M Q

Note: A = attractive; M = must-be; O = one-dimensional; I = indifferent; R = reverse; Q = questionable. (Adopted from Oliver, 2010, p. 153).

Table A2 Satisfaction categorisation of product features. Feature No.

Product feature

MR Type

SQ Type

TL Type

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Unsweetened With fruits and/or nuts With additives/preservatives High sugar content With whole grain Rich in fibre Based on sourdough Labelled with bake date Labelled with best-before date Pre-sliced Sold in large pack Sold in small pack Compact texture Airy texture Filling

A A R R A A A M O (or M) A R (or I) I I (or R) A A (or O)

A A R R A (or Q) A A O O A R I (or A) A, I, or R A, I, R or Q O

A A R R A A A O O O I I A, I, or R A A or O

Note: Classification as indicated in Fig. 3.

C.J. Lagerkvist et al. / Food Quality and Preference 56 (2017) 8–17

References Baumgartner, R. F., & Pieters, R. (2008). Goal-directed consumer behaviour: Motivation, volition, and affect. In C. P. Haugtvedt, P. M. Herr, & F. R. Kardes (Eds.), Handbook of consumer psychology (pp. 367–392). New York: Lawrence Erlbaum Associates. Berger, C., Blauth, R., Boger, D., Bolster, C., Burchill, G., DuMouchel, W., et al. (1993). Kano’s methods for understanding customer-defined quality. Center for Quality Management Journal, 4, 3–36. Bettman, J. R., Luce, M. F., & Payne, J. W. (1998). Constructive processes in consumer choice. Journal of Consumer Research, 4(2), 75–85. Cardello, A. V., Schutz, H., Snow, C., & Lesher, L. (2000). Predictors of food acceptance, consumption and satisfaction in specific eating situations. Food Quality and Preference, 11, 201–216. Carlson, K. A., Tanner, R. J., Meloy, M. G., & Russo, J. E. (2014). Catching nonconscious goals in the act of decision making. Organizational Behavior and Human Decision Processes, 123, 65–76. Förster, J., Liberman, N., & Friedman, R. S. (2007). Seven principles of goal activation: A systematic approach to distinguishing goal priming from priming on non-goal constructs. Personality and Social Psychology Review, 11, 211–233. Giese, J. L., & Cote, J. A. (2000). Defining consumer satisfaction. Academy of Marketing Science Review, 00(1), 1–24. Hamlin, R. P. (2010). Cue-based decision-making. A new framework for understanding the uninvolved food consumer. Appetite, 55, 89–98. Heath, C., Larrick, R. P., & Wu, G. (1999). Goals as reference points. Cognitive Psychology, 38, 79–109. Herzberg, F. (1968). One more time: How do you motivate your employees. Harvard Business Review, 46(1), 53–62. Hull, C. L. (1938). The goal-gradient hypothesis applied to some ‘field-force’ problems in the behaviour of young children. Psychological Review, 45(4), 271–299. Jaeger, S., Lusk, J. L., House, L. O., Valli, C., Moore, M., Morrow, B., et al. (2005). The use of non-hypothetical experimental markets for measuring the acceptance of genetically modified foods. Food Quality and Preference, 15, 701–714. Kahneman, D., & Tversky, A. (1979). Prospect theory. Econometrica, 47, 263–292. Kano, N., Seraku, N., Takahashi, F., & Tsuji, S. (1984). Attractive quality and must-be quality. The Journal of the Japanese Society for Quality Control, 39–48. Keller, K. L. (2003). Strategic brand management: Building, measuring, and managing brand equity (2nd ed.). Upper Saddle River, NJ: Prentice Hall. Kopetz, C. E., Kruglanski, A. W., Arens, Z. G., Etkin, J., & Johnson, H. M. (2012). The dynamics of consumer behaviour: A goal systemic perspective. Journal of Consumer Psychology, 22, 208–223. Kruglanski, A. W., Shah, J. Y., Fishbach, A., Friedman, R., Chun, W., & Sleeth-Keppler, D. (2002). A theory of goal systems. In M. P. Zanna (Ed.). Advances in experimental social psychology (Vol. 34, pp. 331–378). San Diego. CA: Academic Press. Lagerkvist, C. J., Normann, A., & Åström, A. (2015). A theoretical description and experimental exploration of tri-reference point theory with respect to food choice. Food Quality and Preference, 41, 60–74.

17

Locke, E. A., & Latham, G. P. (1990). Work motivation and satisfaction: Light at the end of the tunnel. Psychological Science, 1, 240–246. Maslow, A. H. (1943). A theory of human motivation. Psychological Review, 50(4), 370–396. Mittal, B., & Lee, M.-S. (1989). A causal model of consumer involvement. Journal of Economic Psychology, 10, 363–389. Oliver, R. L. (1980). A cognitive model of the antecedents and consequences of satisfaction decisions. Journal of Marketing Research, 17(4), 460–469. Oliver, R. L. (1993). A conceptual model of service quality and service satisfaction: Compatible goals, different concepts. In T. A. Swartz, D. E. Bowen, & S. W. Brown (Eds.), Advances in marketing and management (pp. 65–85). CT: JAI Press, Inc.. Oliver, R. L. (2010). Satisfaction: A behavioural perspective on the consumer (2nd ed.). Armonk, NY: M.E. Sharp. Pieters, R., & Wedel, M. (2007). Goal control of attention to advertising: The Yarbus implication. Journal of Consumer Research, 34(2), 224–233. Rivière, P., Monrozier, R., Rogeaux, M., Pagès, J., & Saporta, G. (2006). Adaptive preference target: Contributions of Kano’s model of satisfaction for an optimal preference analysis using a sequential consumer test. Food Quality and Preference, 17, 572–581. Santos, J., & Boote, J. (2003). A theoretical exploration and model of consumer expectations, post-purchase affective states and affective behaviour. Journal of Consumer Behaviour, 3(2), 142–156. Sauerwein, E., Bailom, F., Matzler, K., & Hinterhuber, H. H. (1996). The Kano model: How to delight your customers. In Preprints Volume I of the IX International working seminar on production economics (pp. 313–327). Austria: Innsbruck. Stijn, M., van Osselaer, J., & Janiszewski, C. (2012). A goal-based model of product evaluation and choice. Journal of Consumer Research, 39(2), 260–292. Thomson, D. M. H. (2010). Reaching out beyond liking to make new products that people want. In H. J. H. MacFie & S. R. Jaeger (Eds.), Consumer driven innovation in food and personal care products. Cambridge: Woodhead Publishing. Tzokas, N., Hultink, E. J., & Hart, S. (2004). Navigating the new product development process. Industrial Marketing Management, 33(7), 619–626. Xue, H., Mainville, D., You, W., & Nayga, R. M. Jr., (2010). Consumer preferences and willingness to pay for grass-fed beef: Empirical evidence from in-store experiments. Food Quality and Preference, 21, 857–866. Vad Andersen, B., & Hyldig, G. (2015). Food satisfaction: Integrating feelings before, during and after food intake. Food Quality and Preference, 43, 126–134. Verplanken, B., & Aarts, H. (1999). Habit, attitude, and planned behavior: Is habit an empty construct or an interesting case of goal-directed automaticity? European Review of Social Psychology, 10(1), 101–134. Verplanken, B., & Orbell, S. (2003). Reflections on past behavior: A self-report index of habit strength. Journal of Applied Social Psychology, 33(6), 1313–1330. Wang, X. T., & Johnson, J. G. (2012). A tri-reference point theory of decision making under risk. Journal of Experimental Psychology: General, 141(4), 743–756. Zaichkowsky, J. L. (1994). The personal involvement inventory: Reduction. Revision. and application to advertising. Journal of Advertising, 23(4), 59–70. Zhang, Y., Fishbach, A., & Kruglanski, A. W. (2007). The dilution model: How additional goals undermine the perceived instrumentality of shared path. Journal of Personality and Social Psychology, 92(3), 389–401.