Master Thesis Marketing Management

(1)

The gender difference in effectivity of non-permanent displays Volkert Poulie Wilkens 1

Master Thesis Marketing Management

The gender difference in effectivity of

non-permanent displays

(2)

The gender difference in effectivity of

non-permanent displays

What should display communication contain per gender?

Amsterdam, October 27, 2011

Author:

V.J.A. Poulie Wilkens

Department:

Department of Marketing Management,

University of Groningen

Qualification:

Master Thesis BA Marketing Management

Address Author:

Utrechtsestraat 51, 1017 VJ Amsterdam

Phone:

+31 (0)6 26 21 21 68

E-mail:

Volkertwilkens@gmail.com

(3)

Preface

During my 6 months as an intern for Beiersdorf as assistant product manager, I became familiar with the world of cosmetics, drug- and convenience store marketing. Here, the in-store marketing has changed significantly over time. Part of this change are the increased number of in-store marketing activities. One of these activities are non-permanent displays. The usage of non-permanent displays has become more and more frequent, but relevant research on this subject is still lacking. Therefore, this research will be dedicated to create useful information on the usage of non-permanent displays in a shopping environment.

(4)

Inhoudsopgave

Preface ... 3 Abstract ... 5 1. Introduction ... 6 2. Theory ... 9

2.1 Point of Purchase communication ... 9

2.2 Information processing ... 10

2.3 Difference between gender in consumer decision processes and attitudes ... 13

2.4 Differences in response styles ... 16

3 Study ... 18 3.1 Non-Permanent displays ... 18 3.2 Involvement ... 19 4. Hypotheses: ... 22 5. Research Design ... 27 5.1 Questionnaire build-up ... 27

5.2 Restrictions and the dummy variables ... 28

6. Results ... 31

6.1 Pre-analysis ... 31

6.2 Consistency ... 31

6.3 Demographic and general results ... 32

6.4 Regression results ... 34

7. Conclusion and discussion ... 42

7.1 Restrictions ... 42

7.1.1 Display elements construction ... 42

7.1.2 Culture differences ... 44

7.1.3 Sample Size ... 44

7.1.4 Window size of display movie ... 44

7.2 Contributions of the research ... 45

References: ... 47

Appendix:... 50

Appendix A: Questionnaire ... 50

(5)

Abstract

This research concentrates on the moderating effect which gender has on the assessment of display visual. A display is a point of purchase material in the form of an in-store marketing attribute aimed at increasing sales for a particular brand or product. Displays are often used in in-store marketing to combine new product introductions with a price promotion, or price promotion with a premium. The brand and product this research focuses on is NIVEA deodorant.

Through the literature research on print advertisements from Shin (2003), Petrevu (2004), Gailbrath & Stevenson (1993) and others, the constructs on where the differences lie in assessment between gender are identified as: key message placement, message complexity, verbal/pictorial message construction, comparative constructed messages, harmonious constructed messages, category oriented constructed messages and attribute oriented messages. In order to test these elements a model is created for testing the visuals in a regression through data gathered in an online questionnaire.

(6)

1. Introduction

In the world of Fast Moving Consumer Goods the largest parts of the marketing budgets go to numerous Above-The-Line campaigns; the biggest investments are television commercials, big online campaigns and of course, print advertising.

However, in the last few years, Below-The-Line marketing has undergone a rapid change of pace, and is increasingly getting more and more budget (Tyquin, Chapman & Chandra, 2008). One of the big contributors of the rise in the Below-The-Line activities, is the increase in in-store marketing activities. In-in-store marketing activities have more than doubled in the last four years, and this trend is expected to continue in the coming few years.

One of the most important goals of in-store marketing is to influence the impulse buying behaviour of consumers. Stern (1962) performed pioneering research on the factors underlying impulsive buying behaviour and found strong evidence for the use of prominent store displays as one of the major factors influencing this behaviour. This use of displays in stores has also found support in the research of Inman, Winer and Ferraro (2009) on in-store decision making and the effects of category characteristics (such as use of displays) on unplanned purchases. Inman et al found that the use of displays in stores strongly enhances the possibility of consumers making unplanned purchases. These findings were also found earlier in the research of Narasimhan, Neslin, and Sen (1996), who found evidence for impulse buying especially being strongly related with displays presenting price cuts.

Displays are often used in in-store marketing to combine new product introductions with a price promotion, or price promotion with a premium. A price promotion is often part of a non-permanent display. Therefore, non-permanent displays offer an important strategic use. They stand out in the normal shopping environment, which offers a lot of extra exposure time. Also, non-permanent displays are a form of second placement, so it is an extra moment of contact for the consumers. In the drugstore channel, non-permanent displays are an important way of communicating price promotions, product introductions or premium actions in-store.

(7)

Aside from the increasing in-store promotion budget, another change in the cosmetics

retailing industry the past few years is the promotion pressure1. Through internal research of

Beiersdorf it became clear that in the personal care category of the cosmetics industry (deodorant, hairstyling, hair-, shower- and bathcare) the promotion pressure has risen significantly in the last few years. Promotions are becoming an increasingly more important part of the marketing mix. This is also applicable on the deodorant category where this study concentrates on. Here price promotions are becoming increasingly more radical, and frequent. Therefore, new ways of exploiting these promotions provide new challenges and opportunities.

Not only the marketing in-store has changed over the years. Shopping behaviour has also changed in a significant way, especially due to the rise in male shopping. In the last twenty years, male shopping increased significantly, and is still strongly on the rise (Otnesa and McGrath, 2001). Although women still do most of the shopping for the family, men are increasingly doing their own shopping and are contemplating what to buy (Otnesa and McGrath, 2001). This is also seen in the cosmetics industry. Women still do most of the shopping in the drugstores channels, but the male clientele has been, and still is, growing. Furthermore, products are developed in order to meet this phenomenon. Where twenty years ago no specific men hygiene or cosmetics products were available, now entire product lines are available, and this category is still growing. The amount of products offered is also still growing. This development indicates the importance of specific targeting on gender.

Research has been done on the display placement, timings and more. But research concentrating on gender specific product purchases on displays is missing. On displays, the most important indicator for persuasion is the communication on the display. What action is applicable here, which premium do you get, what is the price-off, what are the innovations and what information is communicated through the display. However, information about the effectivity of those elements of in-store communication per gender is lacking, which suggests a gap in the current literature. This is supported by practice, because for multiple

1

(8)

low-involvement categories, promotions and communication are often targeted unisex, or cross categorical.

This research will therefore look into what differences in effectivity will arise on display communication between males and females. This search about the difference in effectivity will lead to the most gender effective way of communicating on displays. Hence, golden rules can be constructed on how to address different genders the best way possible on display communication. This result of the research will have a large value for companies, as they can use it to construct display communication in a more effective way, which will probably result in higher purchase intention and/or purchases for both genders.

This research will show how men and women should be communicated to through displays, so that communication can be gender specific/effective. The research will target male and females in the age segments from 18 to 50+ years.

In order to address the problem this thesis focuses on, the research question of this research will be:

What is the gender difference in effectivity of non-permanent displays and what should display communication contain per gender?

(9)

Theory

In this chapter the relevant literature on the proposed subject of this thesis will be addressed. First, an overview of Point of Purchase communication will be given. Second, information processing and gender specific information processing will be discussed. Lastly, the differences in decision processes of genders will be addressed followed by how gender specific cues and gender specific constructions of print advertisements can be optimized.

2.1 Point of Purchase communication

In the retailing industry, the marketing focus does not lie on in-store marketing. Most of the marketing budget is allocated to media or to other above the line marketing activities. However, in the last few years in-store marketing expenditures as percentage of the entire marketing budget have more than doubled (POPAI, 2004). In 2004 only 3% of the entire marketing budget was allocated to in-store marketing, but in 2006 this percentage rose to 6% and this percentage is expected to keep rising (Tyquin, Chapman & Chandra, 2008). An explanation of the increase in in-store marketing budget are the decision processes of the consumers. A recent study has indicated that 70% of all the purchase decisions are made on the spot, in the retailer itself (Tyquin, et al., 2008). This percentage explains the increase of in-store marketing as a way to influence the consumers decision process in the store.

(10)

Point of Purchase advertisements can be defined as all promotional activities on the shopping floor (de Pelsemacker, 2004). These Point of Purchase advertisement communications can be presented in numerous forms. A few examples are: Displays, Ceiling Communication, Product Positioning, Demonstrations, Scent, Packaging & Interior. This research focuses on only one form of Point of Purchase materials, namely the in-store non-permanent displays (these will be discussed in more detail in the Chapter 3).

The communication on POP advertisements (and other forms of advertisements) can have two goals. These aim at informing and persuading consumers. To persuade a consumer, the information processing has to follow a route in order to have effect. This is also necessary with Point of Purchase communication. Those objectives are (in order of route): (1) Capture Attention, (2) Reminding Consumers of Stimuli , (3) Inform Consumers, (4) Create image, and lastly (5) Persuade Consumers (De Pelsemacker, 2004). So, it can be stated that the ultimate goal of Point of Purchase communication is to positively alter the consumers value sets, through a wide variety of in-store marketing/promotional activities. The Point of Purchase materials are trying to stimulate impulse buying or to enable brand/ product switching or product trial.

2.2 Information processing

To the best of our knowledge, little attention has been paid to the effectivity of communication on Point of Purchase Materials. In order to provide a valid literature research, the most relevant and comparable substitute was used. Print advertisements are processed the same way as communication on Point of Purchase materials. The criteria for assessing and the contact time of print advertisements are also comparable and provide a lot of input for this research. Therefore, print advertisements are the best base for this literature research. And therefore, the following review of processing and assessment of messages is based on research of print advertisements.

(11)

information, this information has to be processed. McGuire (1976) states the following: “Information needs to be processed first, before any changes in attitudes can be realized”. In order to map this process he developed a model from the moment of contact with the information until the attitude change. This model consists of the following steps: Exposure, Processing, Communication Effects and Target Audience Action. This model implicates that in order to effectively influence consumers, readers must give attention to the information followed by accepting the message it is trying to send. When the consumer has given attention to the message and processed it, it can ultimately lead to acceptation of the message and finally to action.

The first three steps that are described in the model of McGuire (Exposure, Processing and Communication Effects), are the steps that are experienced to process the information. The last step is the eventual attitude change the information is responsible for. These steps need to be experienced and undergone by a consumer for the information to reach the desired effect.

(12)

On the peripheral route, little effort is made to interpret and look at the object by the readers of the message. These consumers do not focus or concentrate much on the advertisement. As an effect the motivation, ability and opportunity of those assessors is low (Hoyer and MacInnis, 2008). When the peripheral route is applicable, consumers focus the most on the peripheral cues in an advertisement. These peripheral cues are non-argumentative characteristics of the message which can influence the acceptation process (Petty and Cacioppo, 1983). These can be the attractiveness of colors, the arguments of the message, slogans, words with positive meaning/content for the readers, etc.

When following the central route of the ELM however, the exact opposite of the peripheral route is applicable. People who follow the central route will have a high involvement with the product or advertisement, and the willingness to gather information is higher. Consequently, their motivation, ability and opportunity is high (Hoyer and MacInnis, 2008). Consumers take more time to read and examine the messages. This has implications for the message design. The use of extensive details and more product details are rated as more important by consumers. Also, consumers who follow the central route will most likely be influenced by strong arguments and they will search for more detailed information.

The moderator for the Elaboration Likelihood Model is involvement (Petty and Cacioppo, 1983). Involvement determines which route is most appropriate in the Elaboration Likelihood Model. When involvement of the assessor is high, the central route is taken, so the message has to be constructed according to the central route to achieve the best attitude changes. When involvement is low, the peripheral is most appropriate, and message construction with peripheral cues will achieve the best results.

(13)

consists of three elements; (1) how important is the product, (2) how much risk does the consumer endure when buying a different brand and (3) how long does a consumer think about the purchase (Ratchford, 1987).

When a low involvement category through the Exposure, Processing, Communication Effects and Target Audience action model of Mcguire (1976) is applicable, maximum effectivity will be realised when the advertisement is constructed according to the ELM model. Hence the peripheral route has to be the basis, and a focus should be on peripheral cues. To be most effective, the non-argumentative cues in the advertisement should be focused on, and these have to be optimised. This will have the result that there will be a focus on visual cues, attention to the arguments and on the slogans. When constructed accordingly, there will be more positive attitude changes towards the advertisement (Ratchford, 1987; McGuire, 1976).

2.3 Difference between gender in consumer decision processes and attitudes

There are significant differences between males and females in decision making. Gilligan (1982) argued: “given the differences in women’s conceptions of self and morality, women bring to the life cycle a different point of view and order human experience in terms of different priorities.” The male and the female ethical decision making process differ significantly from each other, male and female use different decision criteria (Galbraith and Stevenson, 1993). The differences are not only bound to ethical decision making. A difference in gender has also been found In overall social norms. Teenage boys were found to undergo more utilitarian decision making, whereas teenage girls are more social/conspicuous conscious in their decision making (Shim, 1993; Petrevu, 2004).

(14)

Also, research indicates females and males process advertisements differently. For example: females are thought to be comprehensive information processors who assimilate all available cues, while males are thought to be selective information processors who assimilate only salient cues (Wolin, 2003). Another indicator is that women process ads more comprehensively under conditions of low/ moderate involvement than men do (Meyers-Levy, 1989; Meyers-Levy and Mahaswaran, 1991). But these differences are expected to evaporate when subtle information cues such as product risk level rises, or the involvement with the message or product increases.

The research examining the selectivity hypotheses suggests that under lower risk conditions, females show equally favourable response to objective and subjective claims. With higher risk conditions, objective claims are more favourable (Wolin, 2003). Another research indicates that men and women react differently to identical advertisements (Petrevu, 2004).

From consumer processing studies it has become clear that there is a difference between interpretation of advertisements and the decision process followed after the interpretation by genders. First, advertisements need to be seen, which will be followed by interpretation and finally it may have an effect on the decision process.

(15)

This suggests that women are more systematic processors than males (Brunel and Nelson, 2003; Petrevu, 2004). This fact indicates that the effectivity per gender of the advertisement is dependent on the placement of the key attributes (Brunel and Nelson, 2003). When the involvement with the advertisement rises, this difference in processing slowly detoriates (Brunel and Nelson, 2003).

The differences in the processing of advertisements between genders results in different assessments of the advertisements. Through the research of Petrevu (2004), it became clear that men have different assessment criteria than women. While male and female find the visual in the advertisement equally important, there are significant differences in processing. Men assess ads better (with as a result, a higher purchase intention) when the advertisements are comparative, simple and attribute-orientated. However, women assess advertisements better when they are verbal, harmonious, complex and category orientated. This means that “each gender reacts better when the advertisements are congruent with their gender-role expectations and information-processing styles” (Petrevu, 2004; Silverman, 2002). Women assess ads better when they are more verbal, and when the message is harmonious, category orientated and complex. Men assess advertisements better when they contain (brand)competition, are more visual, are more simple and focus more on key-attributes of the proposition. Under low-involvement conditions, sex differences emerge because men are more likely to be driven by overall message themes or schemas, whereas women are more likely to engage in detailed elaboration of message content (Meyers-Levy 1989; Meyers-Levy and Maheswaran 1991; Petrevu, 2004). Therefore advertisements targeted at men should focus on why the features distinguish themselves from the competition, or make it unique.

(16)

2.4 Differences in response styles

(17)

Feature Description Details Example

Verbal A bigger focus on verbal keys.

Due to the differences in the brain’s strong points between men and women, women are better processors of verbal information. Hence advertisements which are constructed with more verbal cues will be assessed better by women compared to men (Petrevu, 2004).

Advertisements with more text information and thus more use of words to explain product benefits or reasons to buy the product.

Simple Less information, mainly key attributes.

Due to the constraints of the male brain, males assess information better when there are only a few key attributes in the advertisement.

Less verbal information in ads, keeping the ad simple and more empty, with only the use of keywords and specific unique selling points. Complex More information

available, more detail oriented.

Women assess verbal cues better than males do (Silverman 2002) and are likely to value information-rich resources (Petrevu,2004).

Advertisements with more elaborate visual cues, more detailed information on product aspects and unique selling points.

Comparati ve

Direct product or service comparisons with competitors.

Through sex differences (like aggression, affiliation and social roles), a difference in message assessment by gender is comparison. When the advertisement is constructed with more comparative sentiments men will assess advertisements better. (Petrevu, 2004).

Advertisements which compare the products with the direct competitors, like “the Gillette shaver is 40% more effective than the Wilkinsons Shaver” Harmonio us relationshi ps Harmonious relationships between factors have to be displayed.

Women are socialized to be more concerned about others than themselves (Helgeson and Fritz, 1998). In addition, it has been found that women rank affiliation as more important than men (Petrevu, 2004). So women assess advertisements better when they are focused on harmony in a social construct, and thus rate messages containing communal elements higher.

An ad displaying the balance between nature and the product e.g., “this product is fully in balance with nature”.

Attribute orientated

Objective components stating attribute driven facts (Percy and Elliot, 2008).

Males like to focus on cues that only affect themselves directly (Petrevu, 2004) and thus focus more on the attributes which affect them directly.

Advertisements with enclosed attribute details like in a car advertisement: V6 engine, 10 gallon tank, 5 speed gear, 0-100 kmph in 6 seconds.

Pictorial The usage of pictorial objectives, like diagrams, tables, graphs and pictures.

Males have a tendency to process visual information better than verbal information (Petrevu, 2004). This is due to the characteristics of the male brain. Therefore, more focus must lie in visual information and pictorial information in an advertisement.

Advertisements with graphs, tables, diagrams and pictures as one of the main information distributors.

Category orientated

Advertisements linked to the product category. No direct comparisons with competitors.

Women tend to process relational information better. Hence, they will find category orientated information to be more in line with their way of thought and processing style. So women will assess information with common themes in the claim better (Petrevu, 2004).

An advertisement explaining the position of the product in the product (sub)category.

(18)

3 Study

This research concentrates on a low involvement product group with non-permanent displays. The Beiersdorf deodorant category is a low-involvement product group, and uses non-permanent displays. In this chapter, the factors of low-involvement for deodorant and non-permanent displays will be discussed. First, the non-permanent displays will be discussed, followed by deodorant. We will discuss the relevance for deodorant and non-permanent displays in this part.

3.1 Non-Permanent displays

The Point of Purchase materials this research focuses on are the non-permanent displays. These displays are temporarily placed in a retailer for one to three weeks (varying per action and retailer).

Displays are a form of ‘second-placement’, which means that the products offered in the display are also available on the shelves. So the displays offer an extra moment of contact with the customer in a more attractive manner.

Non-permanent displays are an important factor in the in-store communication. Due to strategic placement, use of the colours, visuals and the text on the display, these displays attract the attention and try to influence the attitude of the consumers

Non-Permanent displays are most often used in three situations; a promotion action on a product (group), a Premium action on a product (group) and most often, a product (group) introduction. Examples of these forms can be seen in the three pictures below.

(19)

For deodorant, all three forms of non-permanent displays are used. But most often they are used to support a new deodorant line in combination with a price promotion. These displays/introductions may vary on the gender targeted. These products could be targeted gender unspecific, male or female.

Displays can be used in multiple ways. From a single product, product lines, product specials to a total category or even a total brand display. So, displays can be used to promote a cross-category product line, a single product, or an entire cross-category and much more. The displays consists of the display itself, the communication of the action and the products which are placed in the display. For this research, a display has been chosen with a product introduction combined with a price-off.

The form of the display used and the products it contains, depends on where the emphasis of the current campaign lies. For example, during product introductions, displays will most primarily be aimed to trial purchases. When an existing product line wants to gain market share, a premium action or bulk promotions on a display will probably have the most effect. Furthermore, combinations or mixtures of all the different forms for non-permanent displays are not uncommon. For example: cross category campaigns (like deodorant combined with body and shower), or introductions in combination with a temporary price promotion. The form and target of a non-permanent display is important for this research. This is due to the differences in involvement between different product groups. In order to make this research relevant and trustworthy, only one product should be used in the displays. Face care (day- and night creams) or body (body lotion, body milk) products have a different level of involvement; they benefit from a higher product involvement. And they would distort the consistency in the research (because deodorant is a low-involvement product group and face/body are high involvement productgroups). Therefore, this research focuses on deodorant.

3.2 Involvement

(20)

the category of low involvement (as can be seen in Figure 3.1 ). This model tells us that deodorant falls in the low involvement – feel category. This category implies that decisions are made to personal taste. The feelings derived from these products can be characterized as “life’s little pleasures” (Ratchford 1987). In order to address this group effectively, Ratchford (1987) claims that advertising needs to emphasize gaining personal satisfaction. But this satisfaction emphasis is not different per gender, and will not differ in importance per gender and is therefore not of influence on this research.

According to Ratchford involvement for the deodorant category is low (Figure 3.1). Deodorant is a low-involvement product for the following reasons: Deodorant is very important to every consumer, most people do not even leave the house without. It therefore can be stated that the risk for using deodorant is high. However, the risk in

(21)

switching between brands is not, and is observed more frequently. Recent developments like rise in promotion pressure and new product launches have encouraged this state.

These market developments suggest a larger focus towards price competitiveness (in the focused target groups for Beiersdorf), and as a result the market is shifting towards a more promotion driven market. As a result the risk of choosing different brands is getting lower and brand switching is becoming more regular.

(22)

4. Hypotheses

After the identification of the research question, hypotheses are needed in order to identify the factors and the assessment of the research question. Therefore, this chapter will concentrate on the construction of the hypotheses and the rationale behind them.

According to the literature described in Chapter 2, there are significant differences between genders in the information processing process and the purchase decision process.

From the literature research it has become evident that women are more systematic processors of advertisements than men. As a result women process advertisements with a Primacy effect, where men process advertisements with a Recency effect (Petrevu, 2004; Nelson, 2003). Hence, placement is crucial for the way each specific gender processes information. Per gender, identification of the most ideal message placement will contribute to the current literature on how to advertise in a shopping environment in low-involvement categories for displays. Therefore Hypothesis 1 can be constructed as follows.

Hypothesis 1a:

Gender has a moderating effect on the assessment of message placement

Hypothesis 1b:

Women process the display communication better when the key element of the message is placed at the beginning

Hypothesis 1c:

Men process the display communication better when the key message is placed at the end of the message

(23)

when they are verbal, harmonious, complex and category orientated. Men however, assess the advertisements more positively when the advertisements are simple, attribute orientated and comparative. In order to investigate if these differences also significantly occur with displays in a shopping environment, these differences will be tested in Hypotheses 2 to 5. In every hypothesis one possible factor will be tested, in order to reach the best possible results for the research.

Petrevu (2004) and Silverman (2002) found that when women assess low-involvement print advertisements, they value complex constructed advertisements significantly better than simple constructed advertisements. Furthermore, the research of Petrevu (2004) concluded that when men assess print advertisements, they favour simple constructed advertisements. Because the processing of information for print advertisements is the most relevant equivalent for assessing non-permanent display communication, Hypothesis 2 can be constructed as follows:

Hypothesis 2a:

Gender has a moderating effect on the assessment of message complexity

Hypothesis 2b:

Women process and evaluate the display communication better when the message is complex

Hypothesis 2c:

Men process and evaluate the display communication better when the message is simple

(24)

contained in a message” (Petrevu, 2004). Because the processing of information for print advertisements is the most relevant equivalent for assessing non-permanent display communication Hypothesis 3 can be constructed as followed:

Hypothesis 3a:

Gender has a moderating effect on the assessment of verbal/pictorial constructed displays

Hypothesis 3b:

Women process and evaluate communication on non-permanent displays better when it is constructed verbal

Hypothesis 3c:

Men process and evaluate the display communication better when the display uses pictorial cues

In the same research of Petrevu (2004), social preferences also played an important role in the assessment of print advertisements. Due to differences in personality traits like aggression, affiliation, harmony and the style in social roles, male and female assess advertisements differently. Men find more value in an advertisement when it is in direct comparison with the category, and thus assess it more positively. On the other hand however, women have another set of social skills and preferences. These contribute to the fact that women assess with a different set of social preferences than males. These factors lead to the facts that women assess advertisements better when they are constructed with harmonious relationships and when they are category oriented. Because the processing of information for print advertisements is the most relevant equivalent for assessing non-permanent display communication Hypothesis 4 can be constructed as followed:

Hypothesis 4a:

Gender has a moderating effect on the assessment of comparative constructed displays

Hypothesis 4b:

(25)

The gender difference in effectivity of non-permanent displays Volkert Poulie Wilkens 25 Hypothesis 4c:

Gender has a moderating effect on the assessment of harmonious constructed displays

Hypothesis 4d:

Men process and evaluate the display communication better when the message is comparative

Hypothesis 4e:

Women process and assess display communication better when it is category oriented

Hypothesis 4f:

Women process and assess display communication better when it is harmonious

Males are known to give a high value to attributes (Schultheiss and Brunstein, 2001). This is also the case in the research of Petrevu (2004). Here it was proven that men assess print advertisements with the use of key attributes significantly higher that advertisements were no attributes were used. Because the processing of information for print advertisements is the most relevant equivalent for assessing non-permanent display communication hypothesis 5 can be constructed as followed:

Hypothesis 5a:

Gender has a moderating effect on the assessment of attribute oriented displays

Hypothesis 5b:

Men process and evaluate the display communication better when the message is attribute orientated

(26)

What is The gender difference in effectivity of non-permanent displays and what should display communication contain per gender?

(27)

5. Research Design

In order to study the proposed hypotheses, a proper research has to be conducted. Therefore, this chapter will focus on the design of this research. First, the questionnaire design be discussed. Followed by the various elements in the displays. Lastly, the tests will be discussed.

In order to test the hypotheses, an online questionnaire has been created. This

questionnaire has been created with www.thesistools.com. The respondents have been

acquired by the personal network of the researcher and through participants in the local drugstores and supermarkets. While the target group of this research are males and females of 30 - 49 years old, the questionnaire will also be available for all other age groups (but they are distinguished in the questionnaire) in order get to get the largest amount of respondents possible. Because five hypotheses will be tested, a minimum of 140 respondents is required to obtain statistical acceptable results.

5.1 Questionnaire build-up

Through the questionnaire the hypotheses will be investigated. The hypotheses will be tested with the help of test displays (constructed with the elements to be tested). But first, the questionnaire will start with some demographic questions (age, education, purchase behaviour and gender) in order to identify the characteristics of the respondents. This provides valuable information about the background of the respondents. To assess the purchase intention and display effectivity, one master display is used and manipulated. This master display can be seen in visual 7 of the questionnaire (see appendix A). For testing, every visual contains some of the elements needed for testing the hypotheses. These variables are elements which will be altered on the master display.

(28)

last page contains the last question which asks the respondents to name the display which they found the most effective in order to look in another way at the effectivity of message placement on gender. This question is at a separate page in order to make sure the respondents will not look back to see which display they found best, and try to recall which display they found the most effective through their memory. The displays have to be interpreted through the peripheral route of the Elaboration Likelihood Model (Petty & Cacioppo, 1983), and thus with low-involvement. Therefore, a time limit of six seconds has been installed for looking at each display. In order to make sure this time limit will be obeyed, the displays have been constructed in a movie of six seconds. The respondent has to press start, and the display will be shown for six seconds.

In order to analyze the results from the questionnaire, the elements in the displays can be handled as dummy variables. For every hypothesis one or several dummy variables are applicable. These dummy variables will be incorporated in a regression model to analyze the data and outcomes (see Table 5.1).

5.2 Restrictions and the dummy variables

In order to conduct a research where the variables do not interfere with the reliability of the research, the sub elements 4a combined with 4b or 4c are not eligible. Because these

messages would be

contradictory to each other. Combining 4c with 4b is eligible however. Another restriction in the model is combining elements 2, 4 and 5. When a display uses one of these elements, the other two cannot be used. This is due

Name Variable Dummy

Moderator Gender 0 = Female 1 = Male

1 Message placement 0 = Beginning 1 = End

2 Message complexity 0 = Simple 1 = Complex

3 Message verbalisation 0 = Verbal 1 = Pictorial

4a Message comparison 0 = Not comparing 1 = Comparing

4b Harmony orientation 0 = Non Harmonious 1 = Harmonious

4c Category orientation 0 = Non category 1 = Category

5 Attribute orientation 0 = Non attribute 1 = Attribute

(29)

to keeping a high clarity of the displays. When these elements would be combined, it would result in displays which are too crowded.

In order for the model to distinguish the variables which have an effect, the elements which are included in the displays will be tested varying two to four times. The result of these restrictions and guidelines is a set-up of the displays and the components they contain per display. This result is shown in Figure 5.2 displayed on the next page.

(30)

Table 5.2 : Visual testing and construction of hypothesis

(31)

6. Results

In this chapter the results of the survey will be analyzed and discussed. First, an overview of the demographic specifics of the respondents will be given. Second, the data will be tested on internal consistency, so a conclusion can be given about the consistency and reliability of the survey. Third, the data will be analyzed in order to investigate the hypotheses.

6.1 Pre-analysis

The survey had a total of 141 respondents. The data was gathered in a four week period,

starting from the 16th of July until the 19th of August. The data was derived mostly from the

online survey (111 respondents). The remaining 30 respondents were gathered from on-site interviews in front of local drugstores in Ermelo. The reason for on-site interviews lied in the difficulty in gathering enough respondents for the online questionnaire and the high drop-out rate of respondents while filling in the questionnaire. The two different interview style are not likely to give any incongruence in the data set, as the on-site and the online questionnaire used exactly the same questions, under the same conditions and requirements. The extra on-site interviews did provide an extra benefit in favour of the consistency. Before the on-site interviews, the largest part of the respondents came from the 30 and younger segment. Thanks to the on-site interviews, this incongruence disappeared, since a large part of the respondents were 30 to 49 segment and the 50 and older segment.

6.2 Consistency

In order to test the internal consistency of the survey and the data, a ‘’ Cronbachs alpha ‘’ reliability analysis has been conducted. It was conducted on the four dependent variables in the questionnaire, both the purchase intention questions (“After seeing the display I have the tendency to purchase the product” and “After seeing the display I would like to try or use the product”) and the display effectivity questions (“This is a good display advertisement” and “This display advertisement contains valuable information”). A cronbachs alpha value

(32)

alpha for this study was 0,938 on the 4 independent elements. So the internal consistency of the questionnaire is high and thus

valid.

The questionnaire tried to test the hypotheses from the four multiple choice questions. However, due to the high internal consistency of the dependent variables, the four variables have been merged into one variable by using the mean value of all the four variables combined. This will decrease the amount of regressions needed to test the hypotheses and makes the analysis easier to interpret and more efficient.

6.3 Demographic and general results

The questionnaire contained some demographic variables in order to identify the characteristics of the sample. The frequencies of respondent details of the survey are shown in the graphs on the right side of the page. The sample in this questionnaire consisted of 52% females and 48% males.

(33)

(derived om statline.cbs.nl on 8-10-2011). So this sample is not entirely representative for the society, because the mean education level in The Netherlands is MBO. This disproportionate number is probably due to the use of the characteristics of the network used for obtaining respondents. The household size mean is 2,59, which comes near the national average of 2,2 (CBS.nl, derived on 3-9-2011). Also, the variance in the mean comes near the average of the Dutch society.

From our sample it can be stated that most of the respondents purchase their own personal hygiene products, but this score is not representative with the Dutch society. While younger people (male and female) buy their personal hygiene products themselves, in families of a higher age category most often only one person does all the shopping (which includes personal hygiene products). Therefore it can be stated that this figure is not completely representative.

The ratings of the display scores were set up within a 7 points Likert scale. So the most neutral score of assessing the displays is 4. However, in the sample the mean scores of all the displays was lower than the neutral score (3,4961).

It becomes clear that none of the nine displays meet the neutral display score of 4. So all nine displays are evaluated negatively. There also is a

lot of variance between the ratings in and between the displays, but none of the mean scores drop below three or rise above four. Display nine scored the best, with a mean of 3,9220.

Display 7 (the master, neutral display) scored second best, with a mean of 3,8387. Display 6 was scored relatively high with 3,7161. Contradictory to display 7 and 6, displays 3 and 4 where the most mediocre rated displays. Display 1 is indisputably the worst

rated, with a mean score of 3,1571. Displays 2, 8 and 5 also received relatively more negative assessments and thus had a relatively lower rating. But the ratings of these three displays do not differ much from each other (difference is <0,1).

(34)

6.4 Regression results

All the p-values in the following regressions are two-tailed. Therefore, a test result will be significant if it has a p-value of <0,05. However, the hypotheses to be tested are one-tailed. Therefore each score on the moderator elements will be found significant if it is <0,1. For optimum clarity, the elements which are tested one-tailed are marked blue in each regression model. These blue elements will be tested at a 10% level.

In order to investigate the preliminary results of the regression, the first regression does not include the moderating elements for the hypotheses. In the regressions, the mean of the four display assessment variables acts as the independent variable. The preliminary regression focused on display elements, and the valuation on the displays. In addition to these elements, gender has also been included in the regression model in order to counter-act the Extreme Response Style (ERS). Without gender in the regression, the valuations of males are significantly lower. This implicates a different response style for gender, and this is counteracted by including gender in the regression. The ERS and therefore including gender in the regression is supported by the significance of gender in the regression (which is 0,009) Therefore it can be said that males rate the displays significantly lower than females (with -0,216). The R squared for this regression is .034, which means that this model predicts 3,4% of all future values.

All the elements in the regression are significant. Because a 7 point Likert-scale was used, the neutral value is 4. The constant value is

3,943, which is almost near the neutral value. From the regression, it becomes clear that when the key message is placed at the beginning of a display, it is rated better when it is placed at the end (-.941). When a message is complex constructed, the displays are significantly rated better (1,891) compared to when the displays are constructed simple. The biggest positive

(35)

element is the pictorial construction. When a display is pictorial constructed, it is rated 2,242 points higher compared to when it is constructed verbally. The sample was negative towards comparative statements in the displays. When the displays contained comparative elements, they scored -1,630 points. A harmonious constructed display had the strongest negative associations. When displays contained harmonious elements it was rated 1,922 points lower compared to when it was not harmonious. When a display had some category oriented elements it was rated slightly less than when it does not have category elements (-.866). When the display contains attribute oriented details it is rated 1,456 points higher compared to when there are no attribute elements in the display.

In order to test the hypotheses and the moderating effects of gender on the assessment of the display visuals, a linear regression with a moderator analysis was conducted on all elements jointly. However, due to the limited amount of respondents, this analysis did not have any workable results.

Therefore, 7 different regression analyzes have been conducted, one regression for each hypothesis. The dependent variable of every regression is the combined valuation data, and the independent variables consist of the elements used in the display. A hypothesis is proven when the sign of the estimate is correct on <0,10 for moderating elements. The <0,10 value is applicable on the moderator elements because the regression is testing a one-tailed hypothesis. Because all other elements are tested two-tailed, all other elements are tested at the <0,05 line.

For hypotheses 1a, 1b and 1c, the regression results can be seen in the following table. The regression had a R square of 0,034, which means that this model will predict 3,4% of all future values.

(36)

(males rate a display better when the key message is placed at the end than women). This difference is minimal (0,049) and not significant (0,778). Furthermore, this change in

gender evaluation does not mean men rate last placed key messages better. Because the Beta of the message placement is -1,251 and significant (with 0,000), the ratings of both gender are better when the key message is shown at the beginning, which suggests that placement of the key message at the beginning of an display is more effective.

For hypothesis 2 the R² is 0,034, which means that this model will predict 3,4 % of all future values. The regression results can be seen in the table to the right. Hypothesis 2a (“Gender has a moderating effect on the assessment of message complexity”) should be rejected with a significance of 0,398.

Hypothesis 2b (“Men process and evaluate the display communication better when the message is simple”)

should be rejected. This is due to the 1,972 higher rating on when messages are constructed

Elements Beta Standard

Error Significance Constant 4,183 ,145 ,000 MSGPLACE -1,251 ,267 ,000 COMPLEXITY 2,141 ,609 ,000 PICTORIAL 2,839 ,703 ,000 COMPARATIVE -2,241 ,433 ,000 HARMONIOUS -2,564 ,452 ,000 CATEGORY -1,119 ,292 ,000 ATTRIBUTE 1,637 ,484 ,001 MSGPLACExGENDER ,049 ,175 ,778 Gender R² = 0,034 -,364 ,131 ,005

(37)

complex compared to simple constructed messages, and this is significant on a 0,001 level. Hypothesis 2c (“Women process and evaluate the display communication better when the message is complex.”) should be accepted with a 1,972 higher rating on complex constructed messages and taking the significance of 0,001 in regard. While the effect of gender points the way the hypothesis predicted (males rate a simple display better than a complex display, and women rate a complex display better than a simple one). This difference is minimal (-,169) and not significant (0,398). So there is no significant influence of the moderator, which defies hypothesis 2a. Because the Beta of the message complexity is 1,972 and significant (with 0,001) the display assessments of both genders are higher when the display construction is simple, and therefore Hypothesis 2b should be rejected.

The regression on Hypothesis 3 had a R squared of 0,034 which means that the model predicts of 3,4% of all the future values. The results can be seen in the table below. And from those results, we can conclude that hypothesis 3a (“Gender has a moderating effect on the assessment of verbal/pictorial constructed displays”) has an effect of 0,064 and a significance of 0,717 which should result in rejection.

So there is no moderating effect of gender in the assessment of pictorial or verbal constructed displays. Hypothesis 3b (“Men process and evaluate the display communication better when the advertisement uses pictorial cues”) should be accepted because of the Beta (2,211) and the significance (0,001). So men and women significantly assess pictorial constructed displays better than verbally constructed displays. Therefore, hypothesis 3c (“Women process and evaluate communication on non-

permanent displays better when it is constructed verbal”) should be rejected.

While the effect of gender points at the effect the hypothesis predicted (males rate a pictorial constructed display better than a verbally constructed display than women), this

(38)

difference is minimal (0,064) and not significant (0,717). Furthermore, this change in gender evaluation does not mean that women rate verbally constructed displays better. Because the Beta of a pictorial construction is 2,211 and significant (with a 0,001 significance) the ratings of both genders are higher when the display is constructed pictorial. Therefore a pictorial construction of a display seems to have the highest effect on the assessment of the displays for both genders.

For hypothesis 4a until 4f the regression results will de described individually, because three regressions were necessary to test

hypotheses 4. All three regressions had a R square of 0,034, so the regressions predict 3,4% of the future values.

The first regression for hypothesis 4 is applicable on hypotheses 4a (“Gender has a moderating effect on the assessment of comparative constructed displays”) and 4d (“Men process and evaluate the display

communication better when the message is comparative’’). The regression results can be found in the table to the right. From these results, it becomes clear that both hypotheses 4a and 4d should be rejected. For 4a, the moderating effect points at the prediction of the hypothesis (0,086), but, this difference is minimal and not significant (0,625).

Because the Beta of a comparative construction is -1,672 and significant (with 0,000) the ratings of both genders are better when the display is constructed non-comparative, so both genders rate non-comparative displays significantly better than comparative displays. Therefore hypothesis 4d should also be rejected

The second regression for hypothesis 4 is applicable on hypotheses 4b (“Gender has a moderating effect on the assessment of category oriented constructed displays”) and 4e (“Women process and assess display communication better when it is category oriented”).

(39)

The regression results can be found in the table on the top right of this page. From these results, it becomes clear that both hypotheses 4b and 4e should be rejected. For 4b, the effect of gender points towards the prediction of the hypothesis

(women rate a categorical

constructed display better than men). But, this difference is minimal (-0,127) and not significant (0,471).

Because of the negative beta (-0,805) and its significance (0,006), a display which is not categorical constructed is

rated significantly higher than one which is categorical constructed. Therefore, hypothesis 4e should also be rejected.

The last regression for hypothesis 4 is applicable on hypotheses 4c (“Gender has a moderating effect on the assessment

of harmonious constructed displays”) and 4f (“Women process and assess display communication better when it is harmonious”). The regression results can be found in the table to the right.

From the regression, it becomes clear that both hypotheses 4c and 4f should be rejected, 4c can instantly be rejected because of the significance level of the gender moderator. This level is 0,598, so here gender is not a

Error Significance Constant 3,922 ,134 ,000 Gender -,174 ,102 ,088 MSGPLACE -,942 ,241 ,000 COMPLEXITY 1,891 ,580 ,001 PICTORIAL 2,243 ,670 ,001 COMPARATIVE -1,631 ,413 ,000 HARMONIOUS -1,923 ,431 ,000 CATEGORY -,805 ,291 ,006 ATTRIBUTE 1,457 ,461 ,002 CATEGORY x Gender R²= 0,034 -,127 ,176 ,471

(40)

significant moderator. Furthermore, the effect the moderator has is not significant and small and even points the opposite way of the proposition of the hypothesis. 4f can also be rejected. When looking at the beta (-1,967) it becomes clear that non-harmonious constructed displays will be rated significantly better (with a significance of 0,000), contradictory to what the hypothesis predicted. When the absence of the moderating effect for a harmonious constructed display is taken into account, combined with the high Beta of the harmonious element, both genders will assess a non-harmonious constructed better than a harmonious constructed one.

For hypothesis 5 the regression results can be seen in the table to the right. The R square of this regression is 0,034, which means that this model predicts 3,4% of all future values. Hypothesis 5a (“Gender has a moderating effect on the assessment of attribute oriented displays”) should be rejected. While the effect of gender points the way the hypothesis predicted (males rate a attribute oriented display better than women), this difference is minimal (0,173) and not significant (0,326). Hypothesis 5b (“Men process and evaluate the

display communication better when the message is attribute orientated.”) can be accepted. The Beta of an attribute oriented construction is 1,373 and significant (with 0,003). So the ratings of both genders are better when the display is constructed attribute oriented.

Although the gender elements have proven not to have a moderating effect on the assessment of display visuals, the last question of the questionnaire was trying to double test the first hypothesis “Gender has a moderating effect on the assessment of the message placement”. This double test was set up to test if the message placement of the visuals is also gender dependable. If the test would be successful, it would show that males would rate the displays shown last better than they would rate the displays they saw at the beginning (even when considering males should prefer a display at the beginning over one

(41)

shown at the end). If this is the case, it shows that males store information which was given first significantly worse than information last. This phenomena is suggested by the literature, it states that males interpret the information given last more effective and remember that information more profoundly, compared to the information given at the beginning (Petruvu, 2004).

In order to test this, the last question of the questionnaire was created. This question asked the favorite display of each respondent (without looking back). So when the data was gathered, the favourite displays of males should have been the displays more towards the end compared to females. A significant and positive beta would prove this fact.

In order to support this construction, the order of the displays in the questionnaire was not set up randomly. The first three visuals were constructed so they should be the most effective for males (according to the hypotheses two to five). In these displays, there were more elements present which males should evaluate more positively according to the literature research. In the rest of the six displays, the displays were constructed more neutral (male effective elements combined with more female effective elements according to the literature) towards males.

Unfortunately, the moderating effect on gender has been proven not to exist in this sample, and was therefore rejected. But this test is trying to double test this phenomenon and if it should be positive, should encourage

(42)

7. Conclusion and discussion

This chapter will focus on the results and the reasons for these results. Also limitations of the research will be discussed, and possible future directions for this line of research will be adressed

The research question of this research is: “What is the difference between genders in effectivity of non-permanent displays and what should display communication contain per gender?”. Through testing this question, this research aimed to look at the effectivity of gender specific displays in a low-involvement product category and the possible construction of those gender specific displays. This possible difference was tested through seven elements which (according to the literature discussed in the theory chapter) would undergo different assessment for males and females. Unfortunately, all of these elements have proven to be not significant. Therefore, it has become clear that gender does not have a moderating role on the assessment of displays in this research. The differences the literature suggested in the interpretation of print advertisements between males and females, are not eligible for displays. However, the assessment differences in gender of points towards the prediction of the hypotheses, but this effect was small and more important, not nearly significant.

Although this research could not prove the effect of gender as a moderator, gender could still have a moderating effect on the assessment on visuals. The reasons for these beliefs can be found in some crucial restrictions this research came across.

7.1 Restrictions

7.1.1 Display elements construction

(43)

more information than others, which resulted in lower ratings.

In hindsight, the construction of the elements was not optimal. In order to test the displays effectively on the construction, every display should have contained the same information so respondents would only have assessed the display on the form the information was presented. However, because the information between displays differed, respondents judged on the amount of information presented instead. Respondents found some displays far too

crowded, which resulted in lower ratings (with sometimes a difference of almost 0,8 points). This can be seen in the table presented to the right. Here the mean scores of all the respondents can be seen per visual. One piece of evidence is provided by the mean results on display 7. This display was included for the mere fact of conducting the regression. This display contained no information as it was functioning as the master display. During the street interviews, the reason of the high

ratings for this display was identified as “the display is not as crowded as the other displays”. Another proof is that displays which have more (crowded) elements, are rated more negative (as can be seen by the table on

the right).

Therefore in order to conduct this research properly, a different research design should be used. To test the moderating role of gender in this form, a