F.T. Beke Consumer Behavior During a Promotional Out-of-Stock

(1)

Consumer Behavior During a Promotional Out-of-Stock

How do consumers react to POOS and which antecedents determine that?

F.T. Beke

(2)

23-08-2012

Master Thesis

Consumer Behavior During a Promotional Out-of-Stock

How do consumers react to POOS and which antecedents determine that?

Name: F.T. Beke

Student number: 1681729

Address: Celebesstraat 67a 9715 JD Groningen E-mail: f.t.beke@gmail.com Telephone: +31 (0)6 52111344 1st Supervisor: prof. dr. L.M. Sloot 2nd Supervisor: dr. J.E.M. van Nierop

University of Groningen Faculty Economics & Business

Department Marketing Postbus 800 9700 AV Groningen

(3)

MANAGEMENT SUMMARY

This research finally pays attention to an interesting variation of the widely researched topic of Out-of-Stock (OOS), namely the Promotional Out-Out-of-Stock (POOS). Although similar in reasons why it occurs, consumer behavior following such a POOS differs due to several different antecedents influencing consumers. This research uses an article by Sloot et al. (2005) as its starting point, and applies that to the specific situation of POOS, by changing which options are observed and thus consumers can choose. Furthermore, promotional-related variables are included in extension to the four categories of antecedents already present in Sloot et al. (2005): product-related antecedents, store-related antecedents, situation-related antecedents and consumer-related antecedents. By including a wide array of antecedents, this research pictures a complete framework regarding why consumers behave as they do following a POOS. The dependent variable used in this research consists of six possibilities of consumer behavior: Postpone Purchase, Outlet Switch, Format Switch, Brand Switch, Product Switch Within Brand and Alternative Product. Consumers Behavior is observed and recorded for one of these options, and the influence of the promotional-related variables Absolute Discount, Days Before Expiring of Promotion, Perceived Promotional Frequency and Deal Proneness is the main focus of this research.

Data was collected by means of personal interviews behind the check-out in two stores for every one of the four selected formats. After comparing the sample with the average shopper, the applicability of this research has been confirmed as demographics of both populations proved to be almost equal. For the analysis, a Multinomial Logit (MNL) was used, which allocates utility to every one of the six options, and assumes that consumers select that option with the highest utility.

(4)

(5)

PREFACE

This master thesis was written as an individual project on the University of Groningen, as part of the master program Business Administration (MSc BA), for the specialization Marketing (Strategic Marketing & Marketing Research).

The process of writing this thesis went for the most part rather smoothly. Being the first project of such a magnitude within the master program, deciding about what content to include in and exclude from the report, proved to be the hardest part. While ‘too’ is never a good word, deciding when something was too extensive or too brief was every time a hard decision. Especially regarding the introduction, writing a brief and interesting piece was important. What made this however easier, was the fact that the subject of retailing (supermarkets), and especially the factor promotion within supermarkets, is a really interesting topic. Moreover, finally observing and analyzing everything I learned so far in practice was an exciting process, making it easier to remain motivated even when not everything went according to plan. Regarding this, I would really like to thank my first supervisor Laurens Sloot, who not only gave comments regarding drafts of the thesis, but who also aroused my interest in retailing by describing experiences from his field of work. This was also possible because of the pleasant working relationship we had, based on an almost equivalent position and a relax atmosphere, but also on a demand for hard work and independency. Secondly, I would like to thank my second supervisor Erjen van Nierop, especially with his help regarding the mathematical part of the report, and with his support regarding the analyses. Regarding these analyses, thanks also goes to Marielle Non of the Marketing Department.

Moreover, I would like to thank here my colleague students, especially the four who wrote their thesis simultaneously under supervision of Laurens Sloot: Annemieke, Kasim, Roosmarijn and Victor. Every thesis meeting, they put effort in commenting on drafts, and also outside those thesis meetings, I was able to approach anyone of them for additional questions. Furthermore, during the entire process they operated as my motivators, making that I demanded more than the average from myself. A special thanks goes to Roosmarijn, with whom I co-operated in the process of creating the questionnaire and collecting the data in the supermarkets.

(6)

1. INTRODUCTION

The phenomenon of out-of-stock (OOS) is a wide debated topic and for a good reason. According to research (Grocery Manufacturers of America, 2002), on average 8.3% of the products are OOS, which leads to a direct 3.9% loss of sales for the retailer (Corsten & Gruen, 2002). Furthermore, OOS lead to a higher level of dissatisfaction under consumers, which leads to more consumers opting to switch stores (Anderson, Fitzsimons & Simester, 2006; Jing & Lewis, 2011). According to research in 2011, OOS almost tops the list of annoyances for supermarket consumers (Consumententrends, 2011). Even higher on that list is the specific situation of a Promotional Out-Of-Stock (POOS). This conclusion seems logical, knowing that consumers are more dissatisfied with an OOS when consumers were really planning on buying that specific product (Fitzsimons, 2000). While promotions tend to attract consumers to stores for certain products (Kumar & Leone, 1988), a POOS will lead to higher level of dissatisfaction than an ordinary OOS. Next to the fact that a POOS inflicts more damage to sales than a regular OOS, the average percentage of appearance for a POOS is higher at 10% (Corsten & Gruen, 2002; Wiebach & Diels, 2011) or even 15% (Grocery Manufacturers of America, 2002) compared to the 8.3% appearance of a regular OOS. Therefore, POOS is a severe problem which needs more insight while it has been undeservedly neglected for years.

When dealing with a problem, the first reaction would be solving that specific problem. Also in the particular case of an OOS, a significant amount of research has been dedicated on finding a solution for the occurrence of an OOS. However, solving OOS with the present-day tools appears to be very difficult (Corsten & Gruen, 2003). The introduction of new technologies concerning inventory management has not provided us with the perfect solution: implementing a method to counter an OOS before it happens, in a cost-effective way. Although a possible solution would be simply increasing the stockpile of a company, Nahmias & Smith (1994) show that this only deteriorates the financial performance of a retailer. While an OOS and POOS are based on exactly the same problems in logistics, namely the problem of providing distribution channels with enough products to accommodate demand by consumers, the lack of solution for an OOS means also that there is currently no suitable solution available for POOS.

(8)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 8 behavior in a POOS is very limited, Wiebach and Diels (2011) show us that consumers tend to leave the store more often during a POOS as compared to an OOS. Therefore, a POOS has a more detrimental effect on both direct sales of the unavailable product and on indirect sales of other products. This stronger effect could be explained by the fact that the POOS not only effects regular consumers of the retailer, but also consumers whose only motivation is buying the item on promotion, so-called “cherry-pickers” (Kumar & Leone, 1988). Therefore, the losses for the retailer include the expected increase in amount spend by current consumers, as well as the expected attraction of extra consumers to which promotion tend to lead (Ailawadi & Neslin, 1998).

The reason why research into the specific field of POOS is needed, instead of using the current research on OOS, is that the reactions by consumers cannot be compared. Research on OOS (Sloot, Verhoef & Franses, 2005) uses six different possible reactions for consumers when dealing with an OOS, which need alteration in order to fit the specific POOS situation. The unique factor for POOS is the fact that there is a promotion which is temporary (Kumar & Leone, 1988). This might alter the possibility for the consumer to postpone the purchase, while the promotional period will expire after a certain amount of time (Jones, 1990). Secondly, the fact that there is a promotion alters the situation when switching to a substitute (either a different brand or product, due to differences in price per quantity. Furthermore, the promotion of a certain retailer does not account for other retailers, which is why substituting shops is also different compared to a regular OOS situation. Therefore, this research will contribute to current literature in that it focuses on a POOS situation, instead of an OOS situation.

In that field of research, the amount of articles is very limited, which makes it important to start with a focus on the entire picture, instead of a certain niche within this field of research. Like stated above, understanding consumer behavior is the first and most important step in this process, which is the main focus of this research. In line with previous research concerning OOS (Sloot e.a., 2005), this research will map which characteristics influence consumer reaction towards a POOS. While there is a difference between the two situations, the characteristics will be altered in such a way that they present a good indicator for consumer reaction towards a POOS. The aim of this research is therefore to observe how consumers react to a POOS, and which antecedents affect consumer behavior following such a POOS. The main research question is formulated as follows:

(9)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 9 As will be explained in the remaining part of this report, the antecedents will be divided in several groups. Therefore, the sub-questions on which this research focusses are:

- What is the effect of promotional-related variables on consumer behavior following a POOS? - What is the effect of product-related variables on consumer behavior following a POOS? - What is the effect of store-related variables on consumer behavior following a POOS? - What is the effect of situation-related variables on consumer behavior following a POOS? - What is the effect of consumer-related variables on consumer behavior following a POOS? In order to prevent any misconceptions, several relevant aspects will be defined (similar to Ailawadi, Neslin & Gedenk, 2001). As stated throughout this report, the specific situation of a POOS is the main focus here. A POOS occurs when a product which is on promotion in unavailable in the shelves, which implicates that consumers are unable to grab the product and purchase it. This unavailability is temporary however, and follows Verbeke, Farris and Thurik (1998) in aspects such as remaining presence of labels and empty shelves. A promotion is present when a discount is offered on the price of products. In most cases, this promotion is externally communicated (e.g. through leaflets and/or posters), but this is not always the case as some promotions are only communicated on the shelf or price tag. The discount could arise in many different forms (e.g. buy one, get one free), but the feature which they all have in common that the price per unit decreases due to the discount. This research will be based within grocery retailers or supermarkets, and will include multiple widespread products which enable to compare between these products. Through this method, this article will be able to provide sound conclusions within the field of research concerning POOS and to retailers in practice.

(10)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 10

2. LITERATURE REVIEW

This article will navigate through the nascent field of POOS. While the amount of literature on POOS is almost non-existent, literature concerning the similar event of OOS will serve as a guideline. The development of knowledge concerning OOS will be described, and an overview will be given of the present-day situation of literature on OOS. This will be followed by an insight why adding a promotion changes the situation, and to what extent there has been research in this changed situation.

Regular Out-of-Stock

The literature concerning a regular out-of-stock (OOS) is extensive. Starting with research by Peckham (1963), several researchers (e.g. Schary & Christopher, 1979; Emmelhainz, Stock & Emmelhainz, 1991; Campo, Gijsbrechts & Nisol, 2003; Sloot e.a., 2005) have added to knowledge concerning consumer reaction towards an OOS. In 1963, Peckham measured consumer reaction as two possibilities: buying a substitute brand or postponing the purchase. In order to come up with a complete model, Peckham (1963) included additional factors, such as the possibility to substitute the package size and the possibility to change color. However, the possibilities Peckham (1963) includes in his article are not representative for the reality. Therefore, the possible reactions by consumers were expended, first by Walter and Grabner (1975), and later on in an article by Schary and Christopher (1979), who focused on branded food items. In order to describe consumers’ reactions in more detail, they distinguish more possibilities for consumer reaction and therefore are able to give a more accurate view on how consumers react when dealing with an OOS. According to them, when a consumer encounters an OOS, he will either not buy, postpone the purchase, switch stores or substitute the product for a different size, brand or a totally different product. A additional factor was included by Motes and Castleberry (1985), with the inclusion of multi-brand inventories, and their interest in brand loyalty.

(11)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 11 consumer reactions, Verbeke et al. (1998) return to the basic by only perceiving three types of consumer reaction: postponement of buying, brand switching or switching stores to get the brand. In order to add to the current literature, they included the effect of competing stores, the amount of store loyalty of a consumer and the total household spending and look into the effect of filling the otherwise empty shelf with another product. The most important implication from their research was that loyal consumers showed a higher preference for store switching. Given the fact that store loyal consumers value that certain store more, this is a surprising result. During the same period, Fitzsimons (2000) added to the literature with a focus on possible retailer action to cope with an OOS. By investigating what type of message provides the best results for the retailer concerning consumer reaction towards an OOS, they provide the retailer with well substantiated advice regarding how they should communicate to their consumers.

Then, starting with research by Corstjens and Corstjens (1995), the field of literature in OOS took a huge leap. Where other authors tried to examine how consumers react in practice, instead they tried to explain consumer behavior. According to them, consumers take their decisions based on possible losses concerning with certain decisions: substitution costs (connected to costs of switching), transaction costs (consists of search costs, handling costs and transportation costs) and opportunity costs (loss of utility through deferring or reducing purchases) (Campo, Gijsbrechts & Nisol, 2004). These classifications of consumer reactions based on possible losses were also used in successive research. Campo, Gijsbrechts and Nisol (2000) used the classification in their research, as they try to explain differences in consumer reaction towards the event of an OOS for different products. Based on three types of characteristics, product-, consumer- and situation-related, they not only describe differences in consumer reaction between different products, but also try to explain why consumers react as they do. One of their outcomes was that store characteristics are critical regarding the decision by a consumer to leave the store (so either store switch or postpone/cancel purchase). Furthermore, they also assert the general statement that brand loyal consumers have a smaller tendency to switch products.

(12)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 12 choose to completely drop the purchase, (4) cancel, and continue without buying that specific product. The final two reactions are (5) category switch, which implies that consumers select another type of product within the same brand, and (6) brand switch, in which consumers stick to the product, but instead select another brand. Using antecedents based on Campo et al. (2000), Sloot et al. (2005) explain why consumers show certain behavior. In order to draw a complete picture of influencing factors of consumer behavior, store-related characteristics were also included.

The most recent research concerning OOS has been done by Rani and Velayudhan (2008) and Jing and Lewis (2011). Instead of an examination of the effect of an OOS on an objective measure such as consumer reaction, they focus on the effect of a subjective measure, namely satisfaction. Jing and Lewis (2011) add to present-day literature with their focus on a specific market, the online market, while they try to derive similarities and differences between that market and the offline retail market.

Antecedents

As mentioned above, an important aspect of present-day literature concerning OOS are the antecedents which explain consumer behavior, preferably all of the relevant antecedents (Ailawadi e.a., 2006). Where in the beginning authors only added some or several antecedents, the last couple of years have provided us with more literature, and therefore with an immense amount of antecedents which try to explain consumer behavior (see table 1).

(13)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 13 specific store loyalty. The difference is that specific store loyalty only represents the loyalty to the survey store, while general store loyalty more represents the general loyalty towards a format by the consumer.

Following the research by Campo et al. (2000), Zinn and Liu (2001) come up with additional features and insights towards antecedents explaining consumer behavior. Unlike other literature, they tend to use subjective measures such as perceived inter-store distance (instead of absolute distance), in order to capture consumer behavior more realistically. To explain all these aspects, Zinn and Liu (2001), use a wide array of demographics (see table 2.1).

Like most research regarding OOS, Sloot et al. (2005) expand upon previous literature (Peckman, 1963; Verbeke e.a., 1998; Campo e.a., 2000). In order to bring more clarity to the enormous amount of antecedents, they decide to introduce a fourth category of antecedents, next to the three of Campo et al. (2000): store-related antecedents. This not only affects new antecedents, but Sloot et al. (2005) decide to classify certain antecedents from previous literature in a different category. Until now, the articles by Sloot et al. (2005) and Rani and Velayudhan (2008) are the most recent in their kind and therefore, they represent the most recent list of antecedents as well. While the research by Rani and Velayudhan (2008) is slightly more focused on consumer attitude (see above), the research by Sloot et al. (2005) forms a solid base for future research.

(14)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 14 S = Store-related M = Moment-related (Situational) C = Consumer-related P = Product-related X = Undefined Pec kh am, 1 96 3 W al ter & G ra b n er , 1 9 75 Sc h ar y & C h ri sto p h e r, 1 9 79 M o tes & C as tl eb er ry , 1 98 5 Em mel h ai n z e. a. , 1 9 91 V er b eke e. a. , 1 99 6 C ampo e. a. , 2 0 00 Fitz si mo n s, 2 00 0 Zi n n & L iu , 2 00 1 Sl o o t e. a. , 2 00 5 Ran i & Ve la yu d h an , 2 0 08 Ji n g & Lew is , 2 0 11 Age C C C C Place of living C Type of product P P P

Earlier experience OOS C M C

Brand loyalty X X C P Store perception X Pre-planned shopping list C M Store loyalty X X C C2 C S C Income C C Product quality P Customer Service P

Value for money P

Convenience P

Product variety P

Product availability P

Location P

Complaining behavior C

Perceived product risk M

Intended product usage M M

Urgency of need M M

Repeat brand purchase X

Proximity similar store S S3 _S4 _S

Purchases per trip C C

Assortment S S5 P6

Market share brand P

Type of brand C P AAAI1 P P Item loyalty X P Deal proneness P P Package size P Shopping trip M M M

Table 2.1: Literature Review of the Antecedents

1

: AAAI = Availability of Acceptable Alternative Items

2

: Campo e.a. (2000) distinguish between specific store loyalty and general store loyalty

3

: Zinn & Liu (2001) use perceived measures, a subjective method

4

: Sloot e.a. (2005) mean not necessarily a similar store, just other stores

5

: Fitzsimons (2000) use assortment size as a antecedent

(15)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 15 S = Store-related M = Moment-related (Situational) C = Consumer-related P = Product-related X = Undefined Pec kh am, 1 96 3 W al ter & G ra b n er , 1 9 75 Sc h ar y & C h ri sto p h e r, 1 9 79 M o tes & C as tl eb er ry , 1 98 5 Em mel h ai n z e .a ., 1 99 1 V er b eke e. a. , 1 99 6 C ampo e. a. , 2 0 00 Fitz si mo n s, 2 00 0 Zi n n & L iu , 2 00 1 Sl o o t e. a. , 2 00 5 Ran i & Ve la yu d h an , 2 0 08 Ji n g & Lew is , 2 0 11 Mobility C

General time constraint C C7 C C

Shopping attitude C C C

Specific time constraint M M

Shopping frequency C C C Store distance C C Quantity needed M Product importance P Brand preference X C8 Stock-out message X

Surprise with OOS M

Reason of buying M P

Intention to visit store M

OOS attitude M/C

Price shopper C

Store prices S9

Working outside home C

Members in household C C10 Children at home C C10 Years of education C Ethnic background C Value of purchases C Store type S S11

Part of the week M

Price consciousness C Quality consciousness C Brand equity P Type of product P12 Hedonic level P Brand equity P Transaction history C

Continuation Table 2.1: Literature Review of the Antecedents

7

: Zinn& Liu (2001) use perceived value of time as an indicator if someone is constraint by time

8_{: Zinn& Liu (2001) used perceived difference in brands, which is an indicator for brand preference} 9_{: Fitzsimons (2000) use assortment size as a antecedent}

(16)

Methodology

When using the familiar field of a regular OOS as a benchmark for the unfamiliar field of POOS, multiple techniques are available for research concerning (P)OOS. Several researchers (Walter & Grabner, 1975; Campo e.a., 2000) used a survey to draw conclusions on consumer reaction towards an OOS. However, when applying the classification of costs (Corstjens & Corstjens, 1995), the costs for store switching (substitution costs) in a survey are smaller than during a genuine shopping trip. Therefore, a survey may lack validity, which is why multiple authors used different techniques.

One of these other techniques is a field experiment, used by Emmelhainz et al. (1991) and Verbeke et al. (1998). By co-operating with retailers, they were able to remove several products (Emmelhainz e.a., 1991) or a complete brand (Verbeke e.a., 1998) from the shelves and following this action, they observed consumer reaction. The difficult part of such a setting, as explained by Emmelhainz et al. (1991), is reaching an agreement with retailers to co-operate, which implies deliberately removing products from the shelves (which would create customer dissatisfaction). Therefore, retailers are generally reluctant to do so, especially when the product to be removed is on promotion.

The last suitable option is interviewing consumers after the check-out, used by both Walter and Grabner (1975) and Zinn and Liu (2001). These so-called quasi-experiments (Sloot e.a., 2005) are less biased compared to an ordinary questionnaire, while all reactions by consumers can be observed, and is easier to perform (with support from the retailer) then a true field experiment.

Promotional Out-of-Stock

However, when there is so many literature concerning a regular OOS available, why is additional research towards the special case of a POOS necessary? This can be explained when looking at the literature concerning promotion. The effect of promotion on consumer behavior has been of interest for many years (e.g. Nystrom & Borden, 1936). As proven by Neslin, Henderson and Quelch (1985), consumers tend to purchase larger quantities during a promotion. This is because it changes consumer behavior for one of the following options: brand switching, store switching, category expansion and purchase acceleration (Narasimhan, Neslin & Sen, 1996). Thus, both OOS and promotion affect consumer behavior. But what is the effect when both occur simultaneously? In order to answer this question, it is necessary to first look further into the aspect of promotion.

(17)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 17 1976). Although a price promotion exists in all kind of different forms, such as price deals, coupons and general rebates (Narasimhan, 1988), in general the effect is that consumers receive a discount on the price per article (Walters, 1991). One general remark could already be made, namely the fact that companies tend to use multiple marketing tools at once (Walters & MacKenzie, 1988). By communicating a price promotion via displays or features, the effect of a price promotion is further enhanced. The desired effect of a (price) promotion is attracting extra consumers towards a store or specific brand, and with that increasing the sales. However, the effect of a promotion goes further than just the selected product, while additional sales also might be affected. Consumers who buy a certain product on promotion, tend to buy other products from the same store as well (so-called HALO-effect), which explains why a promotion for one product, can enhance sales of another (Ailawadi e.a., 2006). This signifies the severity when the promotional product is OOS, while also the HALO-effect might be lost.

(18)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 18 One question is to what extent a POOS differs from a regular OOS. According to Campo et al. (2000), including the effect of a promotion into the situation of an OOS changes the costs as mentioned above (Corstjens & Corstjens, 1995). When looking at the different consumer reactions as used in Sloot et al. (2005), the differences are easy to distinguish. With regard to (1) store switch, the effect of a promotion changes the situation because of price differences. When a consumer experiences a POOS in a store, switching towards another store means losing the beneficial effect of the promotion, and therefore an increase in substitution costs (while switching leads to an increase in price). The possibility of (2) item switch is dependent on the type of promotion: is the entire brand on promotion or does it account only for one type within the brand? If the entire brand is on promotion, item switching provides the possibility for the consumer to still benefit from the promotion. However, if only one type of the brand is on promotion, switching item means losing the promotional benefit (and therefore an increase in substitution costs). The third reaction (3) postponement also has a change in perceived costs. While a price promotion tends to be temporary (Charlton & Ehrenberg, 1976), postponing a purchase (which already leads to higher opportunity costs) could lead to a loss of the promotion. The fact that people lose the promotional price also increases the opportunity costs for (4) cancel (canceling the purchase). Canceling might become more attractive, while consumers could enter the store solely for the promotional product (so-called ‘cherry-pickers’). The final two reactions by Sloot et al. (2005) are (5) category switch and (6) brand switch. For these two, the same accounts as for store switch in that consumers lose the promotional benefit, although in this case the increase in substitution costs will be lower while consumers can stay in the same store.

(19)

3. CONCEPTUAL MODEL & HYPOTHESES

As stated throughout the introduction and literature review, where the field of research on a regular OOS is extensive, the specific situation of POOS has not received too much attention. Although both situations have similarities, the main difference is the promotional factor which is included in POOS. Therefore, it is possible to use literature concerning OOS to expand upon and fit it to the POOS situation. In order to do so, this article will continue on the article by Sloot et al. (2005), with respect to the dependent variable consumer behavior and the categories of additional antecedents (product-related, store-(product-related, situation-related and consumer-related). The additional part of this research is, next to the focus on the specific situation of POOS, the inclusion of promotional-related antecedents as independent variables. These will be Absolute Discount, Days Before Expiring of Promotion, Perceived Promotional Frequency and Deal Proneness.

This part is organized as follows. Firstly, the dependent variable Consumer Behavior as used in this article will be described. The possible reactions within Consumer Behavior will be based on previous OOS literature, and when needed altered to fit the specific situation of a POOS. With the introduction of the promotional related independent variables, the second innovative part of this research will be explained. These antecedents will be defined and hypotheses will be drawn regarding the effect of these antecedents on all the possible reactions of consumers following a POOS. Next, using the four categories used in previous OOS literature (Zinn & Liu, 2001; Sloot e.a., 2005), a complete list of additional antecedents will be presented. These antecedents will be based on previous OOS literature, and also on applicability for retailers in practice. Therefore, several variables will be measured as objective as possible. However, while those so-called objective variables not always represent the reality in a reliable manner, also perceived antecedents are included (based on Emmelhainz et al. (1991)). The total array of variables included allows for reliable prediction of the relationship between the independent, and the dependent variable in case of a POOS. As depicted in the conceptual model, the list of antecedents is extensive, as a complete list allows for better prediction of the dependent variable (Sloot e.a., 2005).

Figure 3.1: Conceptual Model

I. Consumer Behavior

(A) Postpone Purchase (B) Outlet Switch (C) Format Switch (D) Brand Switch

(E) Product Switch Within Brand (F)Alternative Product

4. Deal Proneness

2. Days Before Expiring of Promotion 3. Perceived Promotional Frequency

1. Absolute Discount H1A-B-C

H2A-B-C

(20)

3.1 Dependent Variable

I. Consumer Behavior

As stated before, this research will look further into explicit behavior by consumers following a POOS. Analyzing previous literature, (P)OOS has an explicit effect on consumer behavior (Fitzsimons, 2000). Therefore, this research will follow Sloot et al. (2005) and will also use consumer behavior as its dependent variable. The amount of research into consumer behavior following a regular OOS is extensive (see literature review). A wide array of possibilities for consumer reactions have been used, in which every article tried to come up with a complete list of all consumer reactions. While the possibilities for consumer reactions during a POOS have similarities with the one during an OOS, the six possibilities as used in Sloot et al. (2005) are used here as a starting point. However, while promotion tends to alter consumer behavior, the possible consumer reactions are also in need for alteration. Store switch in case of an OOS makes no difference while there is no promotion involved which differentiates between formats. In case of a POOS however, one format offers a certain discount on one of their products (promotion), thus resulting in a difference in prices and thus attractiveness of the offer. While this promotion is only available in one certain format, switching to another outlet of that format still enables consumers to secure the promotion. Therefore, switching stores presents two different possibilities now: switch of store but remaining within the same format and therefore remaining the promotion, or switch of store and format.

Another difference with the research by Sloot et al. (2005) is, that options cancel purchase and postponement of the purchase are merged as one option. The reason to do so is that the result of both options for the retailer is similar, namely that a consumer does not purchase the product at this moment. Furthermore, indicating if consumers really cancel the purchase or rather postpone by not buying now but instead on a next trip, seems rather difficult. This results in the following six possibilities for short term behavior:

(A) Postpone Purchase: No purchase at this moment, thus postponement. (B) Outlet Switch: Switch to another store of the same format.

(C) Format Switch: Switch to another store of a different format. (D) Brand Switch: Switch to another brand with the same product.

(E) Product Switch Within Brand: Switch to another type/size of product of the same brand. (F) Alternative Product: Switch to a totally different product.

(21)

3.2 Independent Variables Promotional-related

The promotional-related antecedents included in this article are one of the additions to present-day literature. As explained before, promotion changes consumer behavior, thus consumer reaction towards a POOS could differ as compared to a regular OOS. In order to completely explain this difference, several additional antecedents will be included which are more related to the aspect of promotion. The concept of promotion could be used however in multiple forms. This article will use promotion as defined as store-level promotion in Volle (2001): it is a real promotion (not only advertising of regular prices), there is more than one product on promotion and external media (feature/display) is used to communicate the promotion. In order to typify the concept of promotion, the antecedents Absolute Discount and Days Before Expiring of Promotion will be used. For differentiating between products regarding promotion, the antecedent Perceived Promotional Frequency (how often consumers think a product is on promotion?) is included and the final antecedent Deal Proneness is included in order to differentiate between consumers.

1. Absolute Discount

Using the definition of promotion by Volle (2001), one (important) aspect of the promotion is the monetary discount. With price discounts being the most used form of promotion (Darke & Chung, 2005), there are other forms of promotion as well. However, although some promotions might not seem as an absolute discount (e.g. buy one, get one free), these offers in essence provide consumers with a monetary discount, either for one product but sometimes also for a bundle of products. The main advantage for consumers in a discount, is thus that the promotional price is lower than the normal price. The difference between both prices, is the benefit for the consumer, the absolute discount. In their research, DelVecchio, Henard and Freling (2006) look at the discount percentage, which basically is the absolute discount relative to the normal price. However, this might not be a fair representation of how much consumers gain from promotions, while Zinn and Liu (2001) have shown that price levels also influences buying behavior by consumers.

(22)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 22 discount), consumers might not be willing to switch store (of a different format) or purchase an alternative product, when that results in not being able to secure the promotion (Fitzsimons, 2000).

Thus, the hypotheses for the effect of Absolute Discount during a POOS are:

H1A: Absolute Discount is positively related to Postpone Purchase in case of a POOS. H1B: Absolute Discount is positively related to Outlet Switch in case of a POOS. H1C: Absolute Discount is negatively related to Format Switch in case of a POOS.

2. Days Before Expiring of Promotion

The variable Days Before Expiring of Promotion is an indicator how long the promotion is valid for consumers, thus whether consumers are able to postpone the purchase and wait for another moment to secure the promotion. When this amount is high, consumers still have the opportunity to secure the advantage the promotion offers at another moment. Thus, one would expect that the higher the amount of days, the more likely consumers postpone their purchase.

Furthermore, when the opposite occurs, thus the promotional period is almost expired, consumers might be willing to search for the product in another outlet of the same format. This still enables consumers to capture the promotional advantage, with the only additional cost coming from the effort it takes of switching outlet. Moreover, consumers might be inclined to switch also to another format, when the product they want to buy is their preferred product. Where in the beginning of the promotional period, consumers might be willing to postpone their purchase, at the end of promotional period (when consumers might expect the promotion is not resupplied) consumers might be inclined to switch format in order to secure their product (without promotion).

While the promotional period is almost always a fixed period (e.g. starting on Monday and valid until Saturday), it is related to the antecedent part of week, as used in the article by Sloot et al. (2005). In their article the effect of the part of the week seems rather logical, namely that consumers are more willing to postpone their purchase at the start of the week as compared to at the end of the week. Therefore, although not directly comparable (due to the included promotion), the effect of the moment during a week might be the similar to effect of Days Before Expiring of Promotion.

The hypotheses concerning Days Before Expiring of Promotion are formulated as follows:

(23)

3. Perceived Promotional Frequency

The effect of (Perceived) Promotional Frequency on consumer reaction towards a promotion has been a point of interest in several articles (e.g. Kahn & Raju, 1991; Kalwani & Yim, 1992; Ailawadi e.a., 2006). While retailers have the possibility to decide how often a certain product or product group is on promotion, this antecedent seems very useful for those retailers.

In previous literature, the fact that promotional frequency affects consumer behavior towards a promotion has been shown (Kahn & Raju, 1991). One of the changes in consumer behavior is the fact that when a product is promoted frequently, consumers start to accept that promotional price as the regular price, and this could decrease the attractiveness of a certain promotion (Kalwani & Yim, 1992). When looking at a reaction concerning a POOS, this could result in the fact that consumers are aware that although the promotion is OOS at this moment, not to long from now the product (group) will be again in promotion. This relationship has been described as well by Ailawadi et al. (2006). Therefore, when promoted frequently, consumers might be willing to postpone the purchase and wait for another promotional period, and not decide to switch product or store.

However, while consumers have a limited array of products and brands they typically buy, not all products are within their consideration set (Roberts & Lattin, 1991). Therefore, an objective measurement of the promotional frequency might seem inappropriate, while some consumers might think a product is never on promotion because they never buy it, while in reality the product is promoted very often. In order to deal with this, perceived promotional frequency is used, as that more is a more realistic representation of the promotional frequency for consumers. This perceived measure might also impact the willingness of consumers to switch brands (Ailawadi e.a., 2006). Similar as with store switch, consumers might be used to experiencing promotions, and might be willing to wait for their preferred brand to arrive again, instead of choosing for a Brand Switch.

Therefore, the hypotheses concerning Perceived Promotional Frequency are:

H3A: Perceived Promotional Frequency is positively related to Postpone Purchase in case of a POOS H3B: Perceived Promotional Frequency is negatively related to Outlet Switch in case of a POOS. H3C: Perceived Promotional Frequency is negatively related to Format Switch in case of a POOS H3E: Perceived Promotional Frequency is negatively related to Brand Switch in case of a POOS.

4. Deal proneness

(24)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 24 this antecedent combines susceptibility for promotions, with willingness to search and browse through shops in search for a nice bargain. This behavior could be explained by the fact that people are searching for variety: the so-called variety seekers (van Trijp e.a., 1996). Variety seekers switch just for the sake of switching (Campo e.a., 2000). This tendency to switch is mainly driven by external factors, of which promotion is one of the most obvious (Gupta, 1988; Campo e.a., 2000). Important to state is that deal proneness is not dependent on income: also people who have a high income have a tendency for deal proneness, it could be even easier for them as they possess the resources (large car, room for storing etc.) to take full advantage of a nice promotion (Blattberg e.a., 1978). Although not used in Sloot et al. (2005), the construct of deal proneness is somewhat related to price consciousness and quality consciousness (Sloot e.a., 2005). Both constructs label consumers with a certain focus towards one aspect of the product: either price, quality or in case of deal proneness, promotion. While most of the time, this promotion tends to result in a price discount, it will attract both price conscious as deal prone consumers. With securing a discount as one of the most important drivers, deal prone consumers might also be labeled as price conscious (Garretson, Fisher & Burton, 2002). The difference however is that deal prone consumers are willing to browse through shops, in search of the best promotion. This could result in buying only the product on promotion in every shop, so-called ‘cherry-pickers’ (Fox & Hoog, 2005). These ‘cherry-pickers’ visit multiple shops, and buy in every shop the best deals (promotions) in order to make full use of these promotions. The fact that consumers are not bounded to one shop, one product or one brand, is also an aspect of deal prone consumers. This results in consumers who are experienced shoppers, as they visited multiple shops and therefore possess a wide array of knowledge concerning shopping and the availability of different brands (DelVecchio e.a., 2006). Deal prone consumers might therefore be willing to switch stores more easily, in search for the nice promotion.

Another aspect of deal prone consumers, is that they are also not attached to a certain brand or product. When having the possibility to switch brands, brand loyalty plays an important role. However, deal prone consumers typically are less brand loyal (Van Trijp e.a., 1996), and thus more willing to switch to another brand in case of a POOS. The same accounts for switching to another product, while deal prone consumers are generally less attached to one certain product or category (Rani & Velayudhan, 2008).

The hypotheses regarding Deal Proneness are thus:

(25)

Product-related

The product-related antecedents are not bounded to just one product; they apply to a product group. A product group is a group of products which share similar characteristics and are mutually interchangeable and comparable(Sloot e.a., 2005). An important aspect is that different products exist: the question is on which characteristics product groups differ?

In order to differentiate between products, different options have been chosen aspects, such as perishability (Krider & Weinberg, 2000) and private label vs. national brand (Bell, Ho & Tang, 1998; Campo e.a., 2000). This article will differentiate products based on “benefits for consumers”, and thus will follow Sloot et al. (2005) by using the product-related antecedent Hedonic Level. Hedonic level is an indicator to what extent a product group applies to hedonic values. The difference between consumer reactions towards a POOS, could be influenced by the hedonic level, while the effect of a promotion tends to be stronger for hedonic products (Chandon, Wansink & Laurent, 2000). Therefore, hedonic products might attract more ‘cherry-pickers’ (Fox & Hoog, 2005), whose only purpose of shopping is buying the product which is on promotion. A POOS of a hedonic product might therefore result in more store switchers than a POOS of a non-hedonic product.

Although the preferences for more hedonic products and more utilitarian products tend to be opposites, the two aspects itself are not opposites (Okada, 2005). Where hedonic products applies to emotional values and utilitarian more to functional values, several products have been shown to possess both hedonic values as well as utilitarian values (e.g. somebody might really like the taste of something (hedonic values) which is also very healthy and full of nutrition (utilitarian values)). Including just hedonic level is thus not enough, while Utilitarian Level is a different concept. Wertenbroch and Dhar (2000) also point to the fact that hedonic values are more important when discarding a product, while utilitarian values are more important when acquiring a product. For utilitarian products, consumers tend to have less time available, thus implying that switching stores for a certain products seems illogical (Okada, 2005). Furthermore, utilitarian products serve a specific function which most of the time can only be fulfilled by that product, thus implying that consumers might be more willing to buy another brand or type in order to fulfill these direct functional demands (Sloot e.a., 2005).

(26)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 26 products (Schary & Christopher, 1979). Although Campo et al. (2000) uses the direct difference between private label and national brand, the difference in preferences for certain products might still be unexplained. Therefore, differentiation between products will be based on brand equity as previously applied by Sloot et al. (2005). Those brands which are valued with a high brand equity are desired, and thus consumers might be willing to put more effort in obtaining those products, especially when they are made more attractive with a promotion.

An antecedent regarding consumer behavior towards both OOS (e.g. Sloot e.a., 2005; Rani & Velayudhan, 2008) as well as promotion (e.g. Chandon, 1995; Garretson e.a., 2002) which has been applied extensively, is Brand Loyalty. The concept of loyalty can be defined as “customers’ psychological disposition toward an object” (Vogel e.a., 2008). When looking at brand loyalty, this “object” is a certain brand, and therefore represents the following question: does the consumer always buy brand X, or does he lack a certain preference towards a certain brand? Guadagni and Little (1983) have shown that loyalty towards a specific brand affects the brand choice, even more so that brand loyalty becomes visible in the previous buying behavior (purchase loyalty) of consumers (Chaudhuri & Holbrook, 2001). When dealing with a POOS, the amount of brand loyalty to a certain brand could have a significant impact on consumer behavior.

Store-related

Although not used in several articles (e.g. Campo e.a., 2000), the category of store-related antecedents has been shown important. Several aspects of stores (e.g. assortment, price level, service quality) have been shown to affect both perceptions and choice of consumers regarding the store. Furthermore, this group of antecedents is important because retailers can influence those directly in practice, enabling them to influence consumer behavior following a POOS.

An aspect used in OOS-literature which covers multiple aspects of the store, is Store Type. Numerous articles have shown that the effect of a promotion, and the effect of an OOS, depends on the type of shop (e.g. Bell & Lattin, 1998; Sloot e.a., 2005). According to Bell and Lattin (1998), the amount of purchases (basket size) and which products are bought, are determined for a large part by the format of the store, as they attract different sets of consumers. The type of store is determined for the most part by two characteristics, which have been used as variables on their own as well: assortment (Sloot e.a., 2005) and store price level (Krider & Weinberg, 2000; Rhee & Bell, 2002). Differing on these aspects, different types of stores exist, e.g. High Service, Value for Money and Discounter. Consumers of every store type enter the store with a certain expectation and perception of the service of a company, and this could very well influence their behavior following a POOS.

(27)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 27 stores has been used in numerous articles with mixed results (Verbeke e.a., 1998; Sloot e.a., 2005). However, the factor promotion alters possible consumer reactions (see above), and this makes differentiating between a store of the same format, or a store of a different format necessary. Furthermore, similar to perceived promotional frequency, objectively determining the proximity of stores to consumers seems hard. Therefore, for these variables perceived levels are used, as that is a better indication of reality (Rani & Velayudhan, 2007). Thus, Perceived Proximity Other Outlet (same format) and Perceived Proximity Other Format (different store and promotions) are included. The location of a store is an important determinant for selecting a certain store (Bell e.a., 1998). When experiencing a POOS, consumers might be willing to switch outlets, but not formats. However, as Krider and Weinberg (2000) point out, those other outlets have to be close enough to be considered as an alternative. Therefore, consumer behavior might differ based on the perceived proximity of those alternative stores.

The final store-related antecedent has also been used in Sloot et al. (2005): Store Loyalty. Just as with brand loyalty, the construct loyalty in itself implies the “customers’ psychological disposition toward an object” (Vogel e.a., 2008). According to Mägi (2003), most consumers tend to have one prominent store which they value more. This is indicated by the number of purchases one makes, as opposed to purchases in another store (Biong, 1993). When a consumer experiences an unpleasant event like POOS, the result on consumer behavior might depend to what extent someone is loyal to that store. Store loyal consumers might balance the negative experience, with positive experiences during previous shopping trips. As Rhee and Bell (2002) also point out, the store which consumers visit normally is of major importance for future store choice, which would mean that store loyal consumers might be more reluctant to switch format or outlet.

Situational-related

A large part of the reason why consumers behave as they do, differs per situation, which makes the situational-related antecedents all the more relevant (Garretson e.a., 2002); some authors even label them as the most important antecedents (Zinn & Liu, 2001).

(28)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 28 which is unavailable might be the most important part of the minor shopping trip. This might result in consumers leaving the store in order to secure this type of product (Campo e.a., 2000).

Connected to how much time a consumer spends in a specific store, is the Specific Time Constraint a consumer has in a certain situation (Rani & Velayudhan, 2008). This antecedent signifies how much pressure consumers feel while doing their groceries, due to time constrain (Park, Iyer & Smith, 1989). The more time pressure there is, the less time the consumer is able to spend on doing their groceries, thus the less likely consumers are willing to switch stores in order to secure one certain product or promotion (Ailawadi e.a., 2001). Furthermore, the more time consumers have for doing their groceries, the more likely consumers switch within the assortment of the store between different brands, products and types of products (Park e.a., 1989).

Consumer-related

The last category of antecedents, consumer-related, might be the most important category. When explaining store selection (why does a consumer chooses store X over store Y), a total of 70% of the variance is explained by this group of characteristics (Bell e.a., 1998). Furthermore, when looking at previous literature concerning a regular OOS, the group of consumer-related antecedents tends to contain the largest number of antecedents (e.g. Campo e.a., 2000; Sloot e.a., 2005). While this research continues on the article by Sloot et al. (2005), the list of consumer-related antecedents used here will also be extensive.

The first antecedent, Shopping Frequency, has been used in several articles. It signifies how often a consumer goes shopping (Campo e.a., 2000), and can be defined as the number of shopping trips by the consumer within a certain period (Liu, 2007). The effect is not only visible in consumer behavior towards an OOS, but also has influence on excluded constructs such as price-sensitivity (Ainslie & Rossi, 1998). Obviously, a high shopping frequency increases the possibilities of postponing the purchase, although this is also dependent on the promotional period. Furthermore, a high shopping frequencies results in experienced shoppers, which might impact the effect of a POOS on consumer behavior as well. Consumers who show a lower shopping frequency and are less experienced, tend to visit one store and are might therefore be more reluctant to switch store (neither same format nor other format) (Rhee & Bell, 2002).

(29)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke 29 same promotions, look for other promotional items in the same store, or in a store of another format.

The decision to visit another outlet or format, as that are two of the options for consumers, differs with respect to the General Time Constraint of a consumer. Where one consumer has other obligations, such as an employment or a large household, and does not has the time to switch stores, other consumers who have nothing on hand and thus have no limitation due to some form of time constraint (Ailawadi e.a., 2001). An objective measure for general time constraint, which has been used in previous literature (Campo e.a., 2000), is the amount of hours dedicated to paid employment.

3.3 Conceptual Model

(30)

Consumer-related characteristics

- Shopping Frequency - Shopping Attitude - General Time Constraint

Situation-related characteristics

- Shopping trip - Specific Time Constraint

Product-related characteristics - Hedonic Level - Utilitarian Level - Brand Equity - Brand Loyalty Store-related characteristics - Store Type

- Perceived Proximity Other Outlet - Perceived Proximity Other Format

- Store Loyalty

I. Consumer Behavior

(A) Postpone Purchase (B) Outlet Switch (C) Format Switch (D) Brand Switch (E) Product Switch Within Brand

(F)Alternative Product

4. Deal Proneness 2. Days Before Expiring of Promotion

3. Perceived

Promotional Frequency 1. Absolute Discount

Figure 2: Full Conceptual Model & Hypotheses H1A-B-C

H2A-B-C

(31)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke Pagina 34

4 METHODOLOGY

4.1 Data Collection

As described in chapter 2 (Literature Review), the data collection offers several possibilities (e.g. survey or experiment). Neither of these possibilities seem perfect, while every one of them has some flaws and not all are suitable for every situation. In this research, characteristics of consumers are very important, which is why a direct contact with them is critical. Furthermore, several store-related and situational-related antecedent might influence consumer behavior in case of a POOS, which the data collection has to take place where the consumer takes it decisions: in the store. Therefore, data collection was done by means of personal interviews. This data collection method has the advantage that every possible reaction by the consumer can be taken into account (Campo e.a., 2000). Where observing objectively which people leave the store as a reaction on POOS seems impossible, consumers could indicate in an interview that leaving the store would be their preferred behavior.

Research Setting

The personal interviews were done in in the Fast Moving Consumer Goods (FMCG)-industry, more specifically in Dutch supermarkets. While consumers behave differently in different type of formats (Consumententrends, 2011), four different formats present in different segments of the market were selected: full service (Albert Heijn and Plus) and value for money (Dekamarkt and Boni). The inclusion of multiple outlets further solidifies the results and enables to draw conclusions on store-related antecedents (Zinn & Liu, 2001; Srinivasan e.a., 2001). In order to decrease the part of the day and week effect, the personal interviews were held on different days of the week, at different times during a day. This enables for generalization of the results, and the possibility to test the effect of the promotional-related variable Days Before Expiring of Promotion.

(32)

Variables

In order to derive the values for the different variables, two methods were used. Some variables were derived by asking questions, other variables were based on observation by the interviewer (see Appendix A). In order to ensure that the questions presented no problems to understand for the respondents, a pre-test was conducted on family and friends of the researcher. The result was that several questions were slightly altered resulting in a comprehensible questionnaire (Malhotra, 2009). In order to further enhance the reliability of the responses by the consumers, the majority of the constructs were measured using multiple questions (Malhotra, 2009). Within these multiple questions, reversed scaling was used, implying that some constructs were derived using questions asked in both a positive form and a negative form. By comparing both responses, the response bias of the respondents was decreased (Mittal & Kamakura, 2001). All antecedents which were asked using multiple questions, were merged into one construct using either correlation-level or Cronbach’s Alpha.

The dependent variable, consumer behavior, was derived using multiple questions. Firstly, consumers were asked how they hypothetically would behave when encountering a POOS (question 44), choosing between postponing the purchase, leaving the store to buy the product or buying another product with the store. Using follow-up questions for the last two options (questions 45 & 46), consumers were able to indicate their behavior with six different possibilities: (A) Postpone Purchase, (B) Outlet Switch, (C) Format Switch, (D) Brand Switch, (E) Product Switch Within Brand and (F) Alternative Product. In order to derive the values for the other (independent) variables, a wide array of questions were included in the personal interviews. The complete list of questions used for the personal interviews can be found in Appendix B, where every question to which is referred in the text is listed.

(33)

Consumer Behavior During a Promotional Out-of-Stock by: F.T.Beke Pagina 34 In order to determine Hedonic Level, Utilitarian Level and Brand Equity, several possibilities exist. In this research, Sloot et al. (2005) is used as an example in the determination of hedonic level and brand equity, and the same technique is applied to utilitarian level. Using a complete list of all products and brands included in this research, five experts rate three statements for every antecedent using 7-point Likert scales (see Appendix C). Brand Loyalty was measured based on Verbeke et al. (1998), who determine loyalty based on previous buying behavior. By asking the number of times a consumer bought the specific brand previously (of the past ten times), brand loyalty can be determined (question 10).

In order to derive the store-related antecedents, both observation and questioning were used. Based on observation, Store Type was tracked by the interviewer using numbers: Albert Heijn (1), Plus (2), Boni (3) and Dekamarkt (4). For both Perceived Proximity of Other Outlet (question 39) and Perceived Proximity of Other Format (question 38),the perceived level was measured using a 7-point Likert Scale (“Not Very Close” = 1; “Very Close” = 7). Measuring the perceived level allows for the inclusion of difference in place of living of the consumer (Rani & Velayudhan, 2007), and thus is a fairer reflection of the reality (Zinn & Liu, 2001). Store Loyalty was measured similar to Brand Loyalty, with one question (question 11) concerning previous buying behavior (Verbeke e.a., 1998), in this case the number of times a consumer visited one certain store (out of the past ten times).

The two situational antecedents were derived in two different manners. Shopping Trip was measured using the number of product purchased (Bucklin & Lattin, 1992), which was observed by the interviewer using the receipt. Specific Time Constraint was derived using two statements (questions 13 & 14), for which consumers could indicate if they agreed using a 7-point Likert Scale. Reversed questioning was used here as well, which as stated before is used to decrease the response bias (Rani & Velayudhan, 2007).

Finally, the consumer-related antecedents were measured using questions in the interview. Shopping Frequency was derived using a fill-in question regarding the number of shopping trips per week of a consumer (question 12). Shopping Attitude was measured using two statements (questions 15 & 16) based on Rani and Velayudhan (2007) containing 7-point Likert Scales. Also regarding this antecedent, reversed coding was used. The final antecedent is General Time Constraint, for which Blattberg et al. (1978) and Campo et al. (2000) serve as an example and number of hours dedicated to paid employment is used as an indicator (question 4).

(34)

4.2 Analysis

Multinomial Logit Model

Consumer Behavior was recorded using six possibilities (see chapter 3): (A) Postpone Purchase, (B) Outlet Switch, (C) Format Switch, (D) Brand Switch, (E) Product Switch Within Brand and (F) Alternative Product. In order to analyze the influence of the independent variables on these dependent variables, a regression is performed. In this case, a certain individual chooses out of n alternatives. In order to do a regression in such a situation, there are multiple options. However, the model most often used is the MultiNomial Logit-model (MNL-model), due to the fact that it allows for the inclusion of independent variables which are not normally distributed (Leeflang e.a., 2000; Malhotra, 2009). The MNL-model is based on the premise that consumers label different alternatives with a different level of ‘utility’ (Campo e.a., 2000). Based on this utility-value, consumers form preferences for a certain alternative. The MNL-model has multiple advantages: not only are the results more robust compared to for instance discriminant analysis, the statistics are also easy to interpret and use (Dawes, Patterson & Midgley, 1998). Therefore, the MNL-model is a powerful technique for analyzing a dependent variable with multiple unordered categories (Moutinho & Hutchinson, 2011).

In this case, the n alternatives are also unordered, similar to the article by Guadagni and Little (1983). This model is based on differences in utility (U) for every option, which determines the probability (P) of selecting alternative j, which in this case is one of the seven consumer reactions mentioned earlier. The consumer chooses the option with the highest utility. The formula for utility is as follows:

With Unj being the utility U for consumer n for alternative j, β being a vector of parameters, xnj being a vector for the explanatory variables and the error term.

While consumers choose the option with the highest utility, the probability P of choosing alternative j can be written as follows (Fiebig e.a., 2010):

F.T. Beke Consumer Behavior During a Promotional Out-of-Stock

Consumer Behavior During a Promotional Out-of-Stock

How do consumers react to POOS and which antecedents determine that?

F.T. Beke

Master Thesis

Consumer Behavior During a Promotional Out-of-Stock

How do consumers react to POOS and which antecedents determine that?

MANAGEMENT SUMMARY

PREFACE

TABLE OF CONTENTS

1. INTRODUCTION

2. LITERATURE REVIEW

3. CONCEPTUAL MODEL & HYPOTHESES

3.3 Conceptual Model

4 METHODOLOGY