The monetary value of transparency in advertising : should brands engage in transparency to increase sales?

(1)

The Monetary Value of Transparency in Advertising:

Should Brands Engage in Transparency to Increase Sales?

Master Thesis

for MSc in Business Studies - Marketing By: Anne Sophie Erbe (10663711) Under supervision of: Dr. J. Demmers Date: August 22th, 2014

(2)

Table of contents Abstract……….page 2 Introduction………...page 3-5 Literature review………...page 6-11 Method………..page 12-20 Results………...page 21-30 Conclusion and discussion………....page 31-33 References……….…page 34-35 Appendix………...page 36-38

(3)

Abstract

The marketing area is shifting into an era of transparency. Due to rapid business change and the explosion of social media it is nowadays no longer possible for marketers to hide certain information about their brand. There is a new tendency where consumers expect the authenticated truth about everything they buy. Consumers claim to expect transparency efforts provided by companies, but does this actually affect their purchase behavior in a positive way? This thesis investigates whether brands should engage in transparency efforts to increase the consumers’ willingness to pay and tries to establish the first findings regarding transparency efforts for business to consumer products in advertising. First of all; transparency, two-sided advertising and disclosure source were discussed as a starting point for the construct of the research. In order to investigate the research question, a field experiment was conducted using a real online Vickery auction where participants were able to provide their highest bid, measuring actual willingness to pay. A 2x3 factorial design was used, where prior knowledge and the disclosure source were the independent variables. Results found that transparency increased willingness to pay, when the information was disclosed by the firm, but only for products where there was low prior knowledge. When there was extensive prior knowledge about a product, transparency had a negative effect on willingness to pay. Overall, transparency should be taken into account when crafting marketing messages in advertising, since it can increase willingness to pay. To conclude with, future research is needed to further investigate transparency in a business to consumer market, when advertising.

(4)

Introduction

“140 happy calories”, claimed Coca Cola in its latest TV commercial. The ad shows multiple activities that can burn off “140 happy calories” in a can of Coca Cola: dancing, walking the dog, doing a victory dance after bowling a strike or sharing a laugh with friends. Why is Coca Cola advertising in such a transparent way? In other words, why is Coca Cola explicitly disclosing information about their product and where does this kind of advertising come from?

Research shows that consumers have become skeptical against advertising over the last few years and that this provides an incentive for advertisers to provide objective and reliable advertising information (Ford, Smith, and Swasy, 1990). However, skeptical consumers might also become skeptical about honest advertising (Pollay and Mittel, 1993). It could be that Coca Cola claims that there are 140 happy calories inside a coke can because they try to be transparent towards consumers. But is this necessary? Does it actually affect consumer perceptions and eventually consumer behavior?

Today’s consumers are demanding more transparent and sustainable products, as they are more aware of society and supporting the environment (Slavin, 2009). According to Achrol & Kotler (2012), consumer behaviour already went through a big change where it was not just about satisfying the consumer anymore, but about creating experiences and stimulating the senses. They refer to it as marketing in the third millennium (Achrol & Kotler, 2012). Through the explosion of social media and rapid business change it is no longer possible to hide crucial information from the consumer. This creates a huge challenge for companies to heavily protect their brands and carefully craft their brand messages (Noble et al., 2012). With the latest social media tools, consumers can criticize brands publically, when and whenever they want. Cohn & Wolfe (2013) refer to a new era where consumers want the

(5)

real and authenticated truth. But to what extent does transparency create value for the consumer? Will they be more willing to pay for the real truth?

There is a huge difference between covering up information and explicitly putting the information out there. Looking at Coca Cola for example, lots of consumers already know that Coca Cola is not the healthiest drink but they drink it anyway. If consumers are already aware, why create an advertisement that keeps telling them how much calories a coke contains? Consumers claim to find transparent efforts important, but research has not investigated the effect on actual consumer behavior yet. In other words, should the negative information be disclosed to increase willingness to pay and can it actually be seen as an effective marketing technique?

Overall, research has revealed that there is not much compliance whether full information disclosure always leads to good results. This thesis will therefore investigate the following research question: to what extent should brands engage in transparency efforts to increase sales, when advertising? The focus of this study will be on the value of transparency and its effect on the actual purchase behavior of products. In other words does transparency actually create added value for the consumer, making them more willing to pay, even when the disclosed information is somewhat negative?

It is likely to expect that consumers are willing to pay for products from which they know exactly what to expect. But whether they are willing to accept the added transparent information is another story. It is therefore interesting for this investigation to see whether it makes a difference when consumers already know a lot about the product or not and how this can influence the acceptance of transparency and thus the willingness to pay. In this thesis this will be referred to as “prior knowledge”. Secondly, it is also important to look at the disclosure source and whether this could have an impact on the consumer’s willingness to pay.

(6)

This research will contribute theoretically because it could provide empirical evidence and insight into the relationship between transparent branding efforts and actual sales, when advertising. Some prior research investigated the relationship between transparency and the intention to buy. This former research was mostly related to the financial industry, and the apparel industry where transparency was more related to sustainability in the supply chain. This study will look at actual purchase behaviour by conducting an experiment where consumers will bid their own real money on products in advertisements. As this matter has not been investigated before, this study will try to fill up the gap in literature where there is no evidence yet regarding the relationship between branding transparency and actual purchase behaviour. Because marketing transparency is a relatively new topic, future research could also profit a lot from this research to provide new studies that investigate transparency.

This study will also provide new knowledge for brand managers and brand marketers in the area of transparent marketing. As it will focus on negative information disclosure that is more common and applicable to numerous brands in this new era of marketing, managers will gain from this research because former research only focused on transparency in the financial area and in the supply chain area. The aim is to provide answers and insights that will help managers and marketers to deal with transparency efforts in advertising and whether it is smart or not to engage in negative information disclosure.

(7)

Literature Review

Transparency in marketing is a rather new phenomenon and there has not been a clear theoretical definition yet to describe this. In a common context, transparency is referred to as “a characteristic of governments, companies, organizations and individuals that are open in the clear disclosure of information, rules, plans, processes and actions.” (Transparency International). One could say that being open and clear about information, relates to being honest about your product or service. As transparency in a branding context has not been researched extensively yet, it will be interesting to see whether transparency efforts actually work in advertising.

Though the field of transparent marketing is quite unexplored and immature, there have been some who constructed requirements for transparent information and also types of transparency. To begin with, Vishwanath and Kaufmann (2001) have constructed five requirements for transparency; accessible, relevant, comprehensive, qualitative and reliable. To shortly explain these concepts; accessible relates to whether the transparent information is easily reached or taken in, relevant relates to whether the information is appropriate, comprehensive refers to whether the information is complete, qualitative refers to the quality of the information and lastly, reliable refers to whether the information is trustworthy. According to Vishwanath and Kaufmann (2001), when using transparent information, all these five requirements have to be applied in order to make the information transparent.

Furthermore, Hultman and Axelsson (2007), investigated four types of transparency that can be applied in the business to business marketing relationships. According to them the relevant types are; supply (chain) transparency, cost/price transparency, organizational transparency and technological transparency. Supply transparency refers to information disclosure about the supply chain, how or by whom the product was made, for example, “Made in China”. Cost/price transparency provides information about where all costs go in

(8)

the production process, for example, “10% of our profit is used to develop our employees’ working skills.” Organizational transparency refers to information disclosure about the company or the seller, for example how the organization is structured or details regarding the company’s employees. Lastly, there is technological transparency, which relates to disclosing information about the attributes of a product such as the different features it has to give. This is the type of transparency applicable to Coca Cola, looking at the given example in the introduction. As technological transparency is also more relevant for business to consumer markets, this type of transparency will be used in this thesis, as it will investigate transparency using consumer-oriented products.

Secondly, prior research investigated two-sided advertising, a discipline that relates to marketing transparency or information disclosure in a certain way. Two-sided advertising implies that both positive and negative messages should be communicated in order to create the most plausible effect in the mind of consumers (Ein-Gar, Shiv, & Tormala, 2012, Eisend, 2006). But, former studies do suggest that this will only work under certain circumstences. Trifts & Haubl (2003) argue that price transparency works, but depends on the trustworthiness and the price attractiveness of the seller. Eisend (2006) also investigated the persuasive impact of two-sided messaging and found that it depended on several variables such as; amount of negative information, attribute quality, marketers’ voluntariness, placement of negative information, and the correlation between the negative and positive attributes. Overall, he found that two-sided messaging significantly amends the perceived credibility of the source and reduces a negative reaction of the customer (Eisend, 2006, p. 194). Also Bohner, Einwiller, Erb & Siebler (2003) found similar results, claiming that two-sided advertising increases the credibility of the seller, regarding product attributes. According to Cohn & Wolfe (2013) the time to embrace full information disclosure is now. Lafferty & Goldsmith (1999) found that consumers were more likely to buy brands from firms that

(9)

engaged in transparency. Not all research complies with this thought. Anderson (1971) also stated that perceived honesty increases trust, which in turn leads to more favorable attitudes. But, Anderson (1971) also found that negative information disclosure leads to a lower valuation of the product, which means that negative information disclosure is not always a good thing to do. Besides that, transparency could also damage a business’ competitive advantage due to the disclosure of information that might be interesting for competitors (Kirby, 2012). On the other hand, the consumer environment has fundamentally changed, as stated before; consumers are skeptical and can criticize brands when and whenever they want which might make these findings more irrelevant than before. Overall, research reveals that there is not much compliance whether full information disclosure always leads to good results.

Thirdly, some underlying mechanisms that can influence the relationship between information disclosure and buying behavior have to be addressed and discussed as well. “Prior knowledge” is one of the underlying mechanisms that could have an influence on the process of disclosing information and buying behaviour. According to Bhaduri and

Ha-Bookshire (2011), prior knowledge related to a firm and its transparency efforts, increased positive attitudes towards a particular firm, which also lead to higher purchase intentions from the particular firm. In other words, when a consumer is familiar with the brand and its transparent actions, this will increase purchase intentions in the end. Although the study limited in quantitative measures regarding “prior knowledge”, the findings were useful to further explore prior knowledge in this study.

In another study, results showed that consumers with extensive prior knowledge evaluated a product more favorably when their information processing stimulated their perception of progress towards a certain goal (Hong & Sternthal, 2010). Consumers with limited prior knowledge showed more favorable evaluation when their processing involved

(10)

“raw” assessment instead of progress towards a goal (Hong & Sternthal, 2010). These results emphasized that consumers with extensive knowledge should be targeted through message content containing the desirability benefits of a product. Consumers with limited knowledge should be addressed by message content that contains “the ease of use” of a certain product. What could be implied from the underlying mechanisms in this literature is that consumers with extensive prior knowledge will be more likely to experience transparency efforts as something positive and do not experience the negative effect of the negative information compared to consumers with limited prior knowledge. It could be that consumers with limited prior knowledge are more neutral regarding the product and therefore might experience the new transparent information as more negative. Prior knowledge will therefore be tested in this thesis because it might be a good predictor for the acceptance of transparency efforts.

But, in this study prior knowledge will be related to the transparent message content regarding the attributes of the product, which can be related to “the ease of use” of the product, making the transparency effort more positive in the mind of the consumer when there is limited prior knowledge. As this is such a specific finding for communicating transparency relating to attributes, this theory will be higher valuated than the contradicting theory claiming positive relationships between prior knowledge and the acceptance of transparency efforts.

What is different in this study is that prior knowledge will be manipulated to control the effect of the negative information on willingness to pay, which makes it possible to single out the effect of transparency. Theoretically, when consumers have extensive prior knowledge, they will adjust their valuation of the product downwards due to the negative information regarding attributes. In the low prior knowledge condition, there could be an upward adjustment of valuation due to the negative information. As they have less prior knowledge, the negative information might be evaluated as more of an effort because of the company providing extra relevant information. This leads to the following two hypotheses.

(11)

H1: Disclosure of product information leads to higher willingness to pay.

H2: The effect of disclosure on willingness to pay is moderated by prior knowledge consumers hold about the product, such that it leads to lower willingness to pay when consumers have extensive prior knowledge about the product but not when they have limited prior knowledge.

A second mechanism that could influence the relationship between full information disclosure and sales is “source credibility”. Prior research indicated that the source becomes more credible when it engages in unique or negative information disclosure (Eisend, 2010).

Former research also indicated that credibility could not be explained on its own. According to Ohanian (1990), credibility perceived by the consumer, can be divided into three sub themes; (1) Attractiveness, which is related to physical appearance, (2) expertise of the source and (3) trustworthiness of the source. Each of these variables was found to be a reliable measure, but as this study will investigate firms as disclosure sources and not persons like Ohanian (1990) did, attractiveness will not be integrated into the research. Expertise will also not be relevant as the disclosure source will be the manufacturer itself or the auction site providing information for the firm. In this thesis, trustworthiness will therefore be integrated to investigate the “pure” effect of transparency provided by the company. If the transparent information always leads to higher willingness to pay, this will be caused by purely the transparent information. If there is a difference in willingness to pay when different sources are used, this might be due to the effect of the company being the discloser of the transparent message.

(12)

In a study by Goldsmith, Lafferty and Newell (2000), corporate credibility was rated towards ad-endorsers, measuring attitudes and purchase intentions towards the ads and the brands. They made a distinction between attitude towards the ad and attitude towards the brand where they found out that credibility towards the brand increases when the company as a source provides the information (Goldsmith, Lafferty & Newell, 2000). Earlier, Lafferty & Goldsmith (1999) also found that corporate credibility seems to have a greater influence on purchase intentions compared to endorser credibility, meaning using a celebrity or a person as the source of information. It is theoretically possible that the firm (the manufacturer of the product itself) will be perceived trust worthier, when disclosing negative information about their own product, than when another source discloses the negative information. This leads to the following hypothesis:

H3: The effect of disclosure on willingness to pay is moderated by the disclosure source, such that it leads to higher willingness to pay when the brand discloses the information, but not when another entity disclosed the information.

Transparency Willingness to pay

Disclosure source Firm vs. other entity

Prior knowledge High vs. low

(13)

Method

Experimental Design

This study conducted a 2x3 factorial design, in which prior knowledge and the disclosure source were the independent variables. The levels of prior knowledge were high and low, manipulated by the variation of the two products chosen for the experiment, namely, (1) no transparency, (2) transparency disclosed by the firm and (3) transparency disclosed by another entity. The levels of the disclosure source were manipulated by using the same advertisement with small adjustments. In one condition, the disclosed information was provided and in the other condition, the disclosed information was left out, all presented in advertisements. Prior knowledge was related to the knowledge about the product. In one condition, a product that was commonly known was used and in the other condition, a product that was rather new and less known was used. This measured whether the participants would be willing to pay more when the transparent claim was provided for a product where there was high prior knowledge versus a product where there was low or no prior knowledge. To test the effectiveness of the research methodology, a pilot study was conducted to do a preliminary analysis before conducting the whole experiment through Veylinx. This measured the likelihood whether the effects were plausible to occur.

This experiment was chosen because it can measure actual willingness to pay instead of just measuring attitudes, which makes the research more reliable. As there are four treatment conditions and 2 measured factors, the factorial design was a logic choice for this experiment.

Measures

Prior knowledge To choose the two different products for the experiment, a pretest conducting a survey was done to determine whether the products had high or no prior

(14)

knowledge in the eyes of the participants. The survey was conducted through Qualtrics, a survey tool licensed by UvA. The two products chosen for the experiments were the Philips hue (no prior knowledge) and the Philips Daily Blender (high prior knowledge). The Philips hue is a rather new and technological product and the Philips Daily Blender is as it says a daily blender, which many people might already have in their kitchen. Both were chosen because they are made by the same brand, both technological products and both durable consumer goods made for usage at home. It might be possible that the eventual effects could not be caused by prior knowledge but could be related to other differences between the products, but the expectation is that this will not be the case due to the careful considerations taken to choose the products.

The expectation was that people would have low or no prior knowledge about the Philips hue and high prior knowledge about the Philips Daily Blender. Several other random products were added to the pretest as well to see if the prior knowledge effect was really there. In the survey there were 2 questions or statements about ten different products: “I know this product” and “I know this product well”, answered by yes or no. Participants were able to answer through yes or no. These statements were chosen because they would give a very clear outcome.

At first, to determine the two products with the most extensive prior knowledge and the least prior knowledge, a one-sample t-test was conducted for the statement “I know this product” with SPSS to measure actual differences. The test value was set to 2, since respondents could answer yes (1) or no (2). Philips hue appeared to be the only product that was not statistically different from the test value, meaning that lots of respondents did not know Philips hue (table 1). All the other products were significant, meaning that the respondents were familiar with the products. Also for the statement “I know this product well”, a one-sample t-test was conducted with a test value of 2 (table 2). Again, only hue was

(15)

not significant meaning that many respondents did not know this product well compared to the other products. These results confirmed the expectation of Philips hue being a product with less prior knowledge, compared to other products.

Table 1. One-sample t-test for statement: “I know this product”

Product T p

Philips hue starterpack -1.453 .163

Philips Daily Blender -19.000 .000

Philips baardtrimmer QT4005 -6.658 .000

Philips Over-Ear koptelefoon -7.550 .000

Philips PerformerPro stofzuiger -8.718 .000

Philips Senseo Twist -5.339 .000

Test value =2

Table 2. One-sample t-test for statement: “I know this product well”

Product T p

Philips hue starterpack -1.453 .163

Philips Daily Blender -7.550 .000

Philips baardtrimmer QT4005 -3.943 .000

Philips Over-Ear koptelefoon -7.099 .000

Philips PerformerPro stofzuiger -19.000 .000

Philips Senseo Twist -5.339 .000

Test value = 2

To see if there were any differences between the remaining products, additional contrast tests were conducted but none of the statements seemed to differ significantly from

(16)

each other so it was decided to stick with the Philips Daily Blender as a product with extensive prior knowledge.

Source type To test whether the type of source would have an impact on the acceptance of transparency, regarding valuation, they had to be clearly manipulated. The advertisement designs for all the treatments were made equal in seize, trying to keep them completely the same accept for the independent variables. Treatment 1 contained the advertisements without transparency (advertisement 1). In treatment 2, information disclosure by a consumer was tested by using the Veylinx logo together with the transparent message disclosed by a random person with a name, naming that she was a customer and naming her age. Also, the transparent message was placed between inverted commas to make it seem more like a consumer response (advertisement 2). Treatment 3 investigated disclosure by the brand where another logo than the main firm logo was used to explicitly show that the information came from a Philips department (advertisement 3).

(17)

Advertisement 2. Transparency – disclosure by another entity

Advertisement 3. Transparency – disclosure by the brand itself

Transparent message The transparent message that would be placed into the advertisements would disclose some kind of information about the attributes, for example whether the product lacks capability in a certain function. This would be considered as an

(18)

honest or transparent claim since the consumer will not be surprised of downfalls after purchasing the product. To test which transparent claim was most neutral and relevant, another pre-test was conducted with Qualtrics. Several statements were measured on positive versus negative and relevance for Philips hue and the Philips Daily blender. The claims for both products that were found most relevant and perceived neutral or slightly negative were chosen for the field experiment. To begin with, a one-sample t-test was conducted for Philips hue to see which statements were most neutral. A test value of 3 (neutral) was applied to check this. The results showed that statement 1, 6 and 8 were not significant, meaning that they did not significantly differed from neutrality (table 3), meaning that they were suitable for the advertisements.

Table 3. One-sample t-test - Philips hue statements

Statement T p

Alleen dimbaar via smartphone, tablet of Philips Hue account -.797 .431

2 jaar garantie op alle verlichting 10.090 .000

Fitting van de lamp: E27-schroefbasis 2.443 .020

Bridge moet aangesloten worden op Wifi 2.372 .023

50 lampen aansluitbaar op 1 bridge 7.560 .000

Gemaakt in China -.850 .401

Verbruikt 8.4 Watt in de felste stand 4.789 .000

De starterpack wordt exclusief smartphone of tablet geleverd 1.763 .086 Test value = 3

Secondly, additional contrast tests were conducted by using a repeated measures test to see if statement 1, 6 and 8 differed statistically from each other. There was no significant difference

(19)

found between 1 and 6 (F = .013, p = .910) and 6 and 8 (F = 2.917, p = .097) Also 1 and 8 showed no significant difference (F = 1.949, p = .171), meaning that each of these statements could work for a neutral statement (table 4.).

Table 4. Tests of Within-Subjects Contrasts – Philips hue statements

Source hue Type ||| Sum

of Squares

Df Mean Square F Sig.

Hue Level 1 vs. Level 6 Level 6 vs. Level 8 .028 4.000 1 1 .028 4.000 .013 2.917 .910 .097

Hue Level 8 vs. Level 1 4.694 1 4.694 1.949 .171

Lastly, the 3 statements were checked on relevance. Statement 1 appeared to be found most relevant compared to statement 6 and 8 (M = 3.71) (figure 1.)

(20)

The same tests were applied to the statements of the Philips Daily Blender, starting with the one-sample t-test for the eight statements. The results showed that all of the statements significantly differed from the neutral value, meaning that all of the statements were either positive or negative (table 5). It was decided to use statement 2: “Niet gemaakt voor het mixen van ijsklontje” for the Philips Daily Blender advertisements, since it was least significant, slightly negative and had quite a high relevance average (M = 3.74). No additional contrast tests were executed since it was clear from the results which statement to pick for the advertisements.

Table 5. One-sample t-test Philips Daily Blender statements

Statement T p

Gemaakt van verstevigd plastic 5.222 .000

Niet gemaakt voor het mixen van ijsklontjes -3.235 .049

Gemaakt in China -8.912 .000

Beschikt over 3 snelheden 8.857 .000

Heeft een inhoud van 1,5 liter 8.952 .000

Gemaakt van roestvrij staal 9.330 .000

Beschikt over een vermogen van 400 Watt 4.446 .000

2 jaar garantie 9.157 .000

Both claims were then executed to the advertisements so that they fitted with the rest of text (see advertisements).

Operations

Willingness to pay The dependent variable, willingness to pay, was measured by the “Veylinx Auction Platform”. Veylinx provides a way to measure real willingness to pay and

(21)

how consumers valuate products. Consumers can make actual bids on products shown by advertisements and answer questions related to the investigation after participating in a bidding. At Veylinx, the participants are free to bid whatever they want for the product. As there is a chance the participants might receive the product, the bids are often realistic and the participants actually really valuate the product. But, there is a tendency that expensive products sometimes get lower biddings than the actual selling price compared to less expensive product that often receive much higher biddings than the actual price. This type of auction can be called a Vickrey auction or a second-price-sealed-bid auction. Participants simultaneously bid sealed bids for the same product. The highest bidder gets the product but pays the amount of the second-highest bid. Due to this rule, the winning bidder can not affect the price it pays, which causes the bidder to provide his true value for the product, making sure that he wins (Ausubel & Milgram, 2004). Therefore, this auction method is suitable for this research, since it provides a way to measure actual willingness to pay based on true product valuation.

Participants To pretest the relevance of the transparent claims and prior knowledge through online surveys, 25-75 participants were needed to make the pretest reliable. As the pretests only asked short simple questions, it was not too much effort for participants to fill in the surveys seriously, which made it more valid. The participants were randomized through Qualtrics. The participants for the real experiment were conducted through Veylinx and were also randomly assigned to each of the advertisements, automatically by Veylinx. Veylinx has a big sample seize that consists of people who subscribed themselves to Veylinx which means they are actually interested in purchasing certain products through bidding which made it favorable for this experiment.

(22)

Results

Demographics

Philips hue – In total, 510 participants participated in the auction and all made a bid on the Philips hue starterpack. 288 out of the 510 participants were men, and 222 were women (table 3). The average age of the respondents was 41,76 ranging from 16 years old to 114 (table 4). 212 respondents seemed to be higher educated (N=340) (table 5), and 76 respondents indicated that they were still studying (N=324) (table 6). Also 333 respondents were not familiar with the Philips hue product (N=510) (table 7).

Table 3. Frequencies of gender

Frequency Percent Valid Percent Cumulative Percent

Valid Male 288 56.5 56.5 56.5

Female 222 43.5 43.5 100.0

Total 510 100.0 100.0

Table 4. Descriptives of age

N Minimum Maximum Mean Std. Deviation

Age 510 1900 1998 1972.24 15.346

Valid N (listwise)

510

Table 5. Frequencies of high education

Valid Male 128 25.1 37.6 37.6 Female 212 41.6 62.4 100.0 Missing Total System 340 170 66.7 33.3 100.0 Total 510 100.0

(23)

Table 6. Frequencies of students

Table 7. Frequencies of familiarity with Philips hue

Frequency Percent Valid Percent Cumulative Percent Valid 1 2 3 4 Total 333 53 44 47 477 65.3 10.4 8.6 9.2 93.5 69.8 11.1 9.2 9.9 100.0 69.8 80.9 90.1 100.0 Missing Total System 33 510 6.5 100.0

Philips Daily Blender - In total, 388 participants participated in the auction and all made a bid on the Philips Daily Blender. 220 out of the 388 participants were men, and 168 were women (table 6). The average age of the respondents was 40,19 ranging from 16 years old to 114 (table 7). 171 respondents seemed to be higher educated (N=258) (table 8), and 77 respondents indicated that they were still studying (N=252) (table 9).

Table 6. Frequencies of gender

Valid Male 220 56.7 56.7 56.7

Female 168 43.4 43.4 100.0

(24)

Table 7. Descriptives of age

N Minimum Maximum Mean Std. Deviation

Age 388 1900 1998 1973.81 16.620

Valid N (listwise)

388

Table 8. Frequencies of high education

Table 9. Frequencies of students

Lastly, the average bid amounts, represented in eurocents, had to be explored for both products (figure 2 and figure 3). The average bid for Philips hue was 1686.43 eurocents (sd = 2563.20) and for the Philips Daily blender the average bid was 1121.56 eurocents (sd = 1155.08). The average bid for Philips hue treatment 1 (no transparency) was 1564.09 eurocents, for treatment 2 (disclosure by consumer): 1510,46 eurocents and for treatment 3 (disclosure by brand): 1973,51 eurocents (table 10). The average bid for the Philips Daily

(25)

blender treatment 1 (no transparency) was 1281.91, for treatment 2 (disclosure by consumer): 1116.56 and for treatment 3 (disclosure by brand): 920.24 (table 11). As seen in the figures, most bids are skewed to the left. Veylinx encourages their participants to bid even when they are not interested in the product, meaning that several participants bid zero, which causes the histogram not to flow normally. The highest bid for Philips hue was €175 and the highest bid for the blender was €99, not even coming close to the average bids.

Table 10. Average bids for Philips hue per treatment

N M Sd

Treatment 1 151 1564.09 2126.063

Treatment 2 182 1510.46 2316.66

Treatment 3 176 1973.51 3085.2

Table 11. Average bids for the Philips Daily Blender per treatment

N M Sd

Treatment 1 125 1281.91 1304.815

Treatment 2 135 1165.56 1170.208

Treatment 3 129 970.24 953.967

(26)

Figure 3. Histogram of bid amount – Philips Daily blender

Tests of normality

To investigate whether the conceived data were distributed normally; tests of normality were executed through SPSS for both the Philips hue starterpack and the Philips Daily Blender. First, The Kolmogorov-Smirnov and the Shapiro-Wilk test.were applied to see if the data were normally distributed. Secondly, the data was transformed, and an algorithm was applied to create lower variance, using the top 50 percent to create a normal distribution (see figures appendix).

Philips hue – The original biddings and the biddings to which an algorithm was applied, both resulted to be significant, F = 509, p = .000 (table 12). Still, the data appeared to be skewed to the left and not distributed significantly, due to biddings of zero cents that can be seen in the histogram provided in the appendix (figure 1). This occurs mainly because of the fact that the Veylinx platform impels their respondents to make a bid even when they do not value the product and are not willing to pay any price for it which creates a lots of zero bid amounts, making the date skew to the left. Veylinx does this to posses a bigger database.

(27)

In order to do parametric test with by using a factorial ANOVA, SPSS assumes that the data is normally distributed, meaning the distribution is symmetrical and not skewed to the left such as this dataset. The skewness and kurtosis are supposed to be zero for normal distribution, but the skew and kurtosis of the bid amount to which an algorithm was applied still appeared to be -.746 and -1.168 (table 13).

Table 12. Tests of normality of original biddings and algorithm biddings

Kolmogorov-Smirnov Shapiro-Wilk

Statistic df Sig. Statistic d Df Sig.

Bid amount .255 509 .000 .643 509 .000 Bid amount with algorythm .262 509 .000 .770 509 .000

a. Lillifors Significance Correction

Table 13. Descriptives bid amount with algorithm applied

Statistic Std, Error Bid amount with

algorythm

Skewness -.746 .108

Kurtosis -1.168 .216

Philips Daily Blender – The original biddings and the biddings to which an algorithm was applied, both resulted to be significant, F = 388, p = .000 (table 14). Again, the data appeared to be skewed to the left and not distributed significantly, due to biddings of zero cents that can be seen in the histogram provided in the appendix (figure 3 and 4). Also, the skew and kurtosis of the bid amount to which an algorithm was applied were not zero and still appeared to be -.746 and -1.168 (table 15).

(28)

Table 14. Tests of normality of original biddings and algorithm biddings

Kolmogorov-Smirnov Shapiro-Wilk

Statistic df Sig. Statistic d Df Sig.

Bid amount .166 388 .000 .827 388 .000 Bid amount with algorythm .290 388 .000 .722 388 .000

a. Lillifors Significance Correction

Table 15. Descriptives bid amount with algorithm applied

Statistic Std, Error Bid amount with

algorythm

Skewness -.949 .124

Kurtosis -.891 -.891

Tobit regressions

Due to the data not being distributed normally, it was decided to test the hypothesis by using the software program Stata. This research method was chosen based on the fact that the skew and kurtosis value of the algorithm biddings were not zero or close to zero for both the Philips hue starterpack and the Philips Daily Blender. Another decision that was made was to run the tests for hue and the blender separately because the two products had such different values and thus very different biddings. When taken together this would overrule other effects.

As the data was censored to the left and cannot go below zero, parametric test could not be used which does not make SPSS suitable for testing. Instead, Tobit regressions were used, since they do have the ability to test data, where the dependent variable is either censored to the left or to the right.

(29)

Philips hue - At first, the relationship between the bids in euro cents and the source of transparency was tested for the Philips hue starterpack (table 16). 141 cases were left-censored, 368 cases were uncensored and zero cases were right-censored (N=509). The probability of Chi2 (p = 0.0000) indicated that the proposed model based on transparency and the source, was significant in comparison to an empty model. There appeared to be no significant effect between transparency and no transparency, which provides no evidence for hypothesis 1: disclosure of product information leads to higher willingness to pay.

But, after conducting an additional contrast test, there seemed to be a significant difference between information disclosure by the consumer and information disclosure by the brand or the manufacturer (p = .0264). Looking at the coefficients, disclosure by the brand (3) had a significant, more positive impact on bid amount then disclosure by the consumer (2), meaning that disclosure by the brand leads to higher willingness to pay which is in line with the third hypothesis: The effect of disclosure on willingness to pay is moderated by the disclosure source, such that it leads to higher willingness to pay when the brand discloses the information, but not when another entity disclosed the information.

Also, some additional results showed that birth year (p = .000) and gender (p = .000) had a significant impact as well on bidding amount. The coefficients showed that when the respondents get older they are more likely to buy the product and that male respondents are likely to bid less than female respondents.

Table 64. Tobit regression, relationship between bid amount, transparency and source type – Philips hue

LR chi2 (4) = 39.63 Prob > chi2 = 0.0000 Pseudo R2 = 0.0055

(30)

Treatment 2 -211.498 360.493 -0.59 0.558 -919.748 496.753 3 556.938 359.298 1.55 0.122 -148.965 1262.841 Birth year 38.424 9.571 4.01 0.000 19.620 57.228 Gender -1338.168 293.370 -4.56 0.000 -1914.545 -761.791 _cons -74254.15 18884.36 -3.39 0.000 -111355.7 -37152.56 /sigma 3111.39 119.4333 2876.743 3346.038

Philips Daily Blender - The relationship between the bids in euro cents and the source of transparency was also tested for the Philips Daily blender (table 17). 98 cases were left-censored, 290 cases were uncensored and zero cases were right-censored (N=388). The probability of Chi2 (p = .0348) indicated that the proposed model based on transparency and the source, was significant in comparison to an empty model. There appeared to be a significant effect between transparency and no transparency for treatment 3, disclosure by the brand or the manufacturer (p = .049). Looking at the coefficients, transparency had a more negative effect on bidding amount than using no transparency which is contradicting hypothesis 1: Disclosure of product information leads to higher willingness to pay. After conducting an additional contrast test, there seemed to be no significant difference between information disclosure by the consumer and information disclosure by the brand or the manufacturer (p < .05), which provides no evidence for hypothesis 3: The effect of disclosure on willingness to pay is moderated by prior knowledge consumers hold about the product, such that it leads to lower willingness to pay when consumers have extensive prior knowledge about the product but not when they have limited prior knowledge. Some additional results showed that birth year (p = .011) had a significant impact as well on bidding amount. Gender

(31)

did not have a significant impact (p < .05). The coefficients showed that when the respondents get older they are more likely to buy the product.

Table 17. Tobit regression, relationship between bid amount, transparency and source type – Philips blender

LR chi2 (4) = 10.36 Prob > chi2 = 0.0348 Pseudo R2 = 0.0020

Bid amount Coef. Std. Err. T P>|t| 95% Conf Interval Treatment 2 -219.461 184.802 -1.19 0.236 -582.811 143.888 3 -367.058 186.136 -1.97 0.049 -733.028 -1.084 Birth year 11.704 4.567 2.56 0.011 2.725 20.683 Gender -52.405 153.894 -0.34 0.734 -354.986 250.176 _cons -74254.15 18884.36 -2.44 0.015 -39699.8 -4267.906 /sigma 1433.96 62.92922 1310.231 1557.689

Overall, there were some significant results found regarding the effect of transparency efforts on willingness to pay. Looking at Philips hue, a product where the respondents had less prior knowledge, transparency provided by the brand or the manufacturer seemed to be significant. Looking at the Philips blender, a product where the respondents had extensive prior knowledge, transparency by the brand was found to be significant but had a negative effect on bidding amount compared to no transparency.

(32)

Conclusion and discussion The main research question investigated in this study was: “should brands engage in transparency to increase sales, when advertising?” The first expectation was that disclosure of product information would lead to higher willingness to pay (hypothesis 1). Also, it was expected that the effect of disclosure on willingness to pay would be moderated by prior knowledge consumers hold about the product, such that it leads to lower willingness to pay when consumers have extensive prior knowledge about the product but not when they have limited prior knowledge (hypothesis 2). When speaking for the Philips hue product, disclosure of product information does lead to higher willingness to pay, but this only accounted for the treatment where the transparent information was provided by the brand, in this case by Philips itself. This is in line with the expectation that when the manufacturer of the product discloses the transparent information, respondents will be more willing to pay than when the auction site discloses the information through a consumer response. This only accounted for Philips hue and not for the blender, where transparency had a negative effect on bidding amount compared to no transparency in the source treatment where the consumer disclosed the information. Besides that the transparency treatments in the Philips hue auction appeared to have more positive biddings, respectively than the transparent treatments in the Blender auction providing support for hypothesis 3, where it was expected that the effect of disclosure on willingness to pay is moderated by prior knowledge consumers hold about the product, such that it leads to lower willingness to pay when consumers have extensive prior knowledge about the product but not when they have limited prior knowledge. Overall, the research question, whether brands should engage in transparency to increase sales when advertising, remains not fully answered. But, for a product like Philips hue, when there is less prior knowledge it is useful to disclose transparent information about the product attributes in an

(33)

advertisement. Being honest and open about your product does definitely create value for consumers but you have to be very careful when crafting the transparent marketing message.

Contributions and implications

This study aimed to create more insight into the field of transparency in a marketing context related to consumers. So far, there have not been many studies to explore the field of transparent marketing. This study has contributed by showing that transparency can lead to higher willingness to pay, when advertising and selling consumer-using products and that there is an actual relationship between disclosing transparent information and consumer behaviour.

For manager and marketers; the use of transparency when advertising can be a valuable asset, but as earlier said, it should be carefully crafted. There is an opportunity for marketers that go about innovative unfamiliar products. Use the firm as a source and provide the customers with extra transparent information that creates added value for them. Do not do this too much when you are selling a product that has extensive prior knowledge, it will probably lead them to the competition. Keep in mind that this study only experimented with technological transparency, meaning disclosing information about the products’ attributes.

Limitations

As transparency has not been investigated extensively yet, the conducted field experiment was very experimental, though based on prior research. The choice to go for technological or attribute transparency was based on the fact that this study used user-products with less and extensive prior knowledge about the user-products. Since the respondents were actually bidding the real amount that they were willing to pay, the results of this study are very reliable regarding other types of studies.

(34)

First of all, what could have been done different was to investigate transparency more to see which transparency claims to use, looking at the negative effect in the transparency treatments of the blender. Maybe, the transparent claim was too negative for the Philips blender. But, the experiment for Philips hue, showed a great significant effect in favor of disclosing transparent information, provided by the company itself.

Secondly, the products could maybe have been chosen more carefully, based on price and brand. Whereas Philips hue is quite an innovative and an expensive product, Philips Daily Blender is quite a regular and product available everywhere. Although the participants were not aware of the price of the two products, it could have been that the participants valued Philips hue as a more expensive product which could have lead to the fact that a transparent message is more appreciated when buying a high priced product.

Lastly, although the products were from the same brand, products without a brand might have lead to more reliable results since Philips might have been highly valuated by the participants, this could have influenced their acceptance of the transparent message and thus their willingness to pay.

Future research

What could be interesting in future research is to test more types of products, since this study only tested two products. Perhaps, testing products without brand names would lead to more reliable outcomes. Also more types of transparency would be interesting to test and compare since this is a relative young research field. It is advised to conduct field experiments, since these results provide reliable results regarding the actual value of transparency. Lastly, it remains important, to adequately manipulate the disclosure source and carefully craft the transparent message.

(35)

References

Achrol & Kotler (2012). Frontier of the Marketing Paradigm in the Third Millennium. Journal of the Academy of Marketing Science, 40(1), 35-52.

Anderson, N.H. (1971). Integration Theory and Attitude Change. Psychological Review, 78(3), 171-206.

Bhaduri, G., & Ha-Brookshire, J.E. (2011). Do Transparent Business Practices Pay? Exploration of Transparency and Consumer Purchase Intention. Clothing and Textile Research Journal, 29(2), 135-149.

Bohner, G., Einwiller, S., Erb, H.-P., & Siebler, F. (2003). When small means comfortable: Relations between product attributes in two-sided advertising. Journal of Consumer Psychology, 13(4), 454-463.

Cohn & Wolfe (2013). From Transparency to Full Disclosure. London

Ein-Gar, D., Shiv, B., & Tormala, Z. L. (2012). When blemishing leads to blossoming: The positive effect of negative information. Journal of Consumer Research, 38(5), 846-859.

Eisend, M. (2006). Two-sided advertising: A Meta Analysis. International Journal of Research in Marketing, p.194

Eisend, M. (2010). Explaining the joint effect of source credibility and negativity of information in two-sided messages. Psychology and Marketing, 27(11), 1032-1049

Ford, G.T., Smith, D.B., and Swasy, J.L. (1990). Consumer Skepticism of Advertising Claims: Testing Hypothesis from Economics of Information. Journal of Consumer Research, 16(4), 433-441.

Goldmsith, R.E., Lafferty, B.A., and Newell, S.J. (2000). The Impact of Corporate Credibility and Celebrity Credibility on Consumer Reaction to Advertisements and Brands. Journal of Advertising. 29(3), 43-54.

(36)

Hong, J. & Sternthal, B. (2010). The Effect of Consumer Prior Knowledge and Processing Strategies on Judgements. Journal of Marketing Research, 47(2), 301-311.

Hultman, J., & Axelsson, B. (2007). Towards a typology of transparency for marketing management research. Industrial Marketing Management, 36(5), 627-635.

Kirby, J. (2012). Trust in the age of transparency. Harvard Business Review, 90 (7/8), 158-160.

Lafferty, B.A. & Goldsmith, R.E. (1999). Corporate credibility in Consumers’ Attitudes and Purchase Intentions When a High versus a Low Credibility Endorser is used in the ad. Journal of Business Research, 44(2), 109-116

Ohanian, R. (1990). Construction and Validation of a Scale to Measure Celebrity Endorsers’ Perceived Expertise, Trustworthiness, and Attrativeness. Journal of Advertising, 19(3), 39-52

Pollay, Richard W. and Banwari Mittal (1993). Here is the Beef: Factors, Determinants, and Segments in Consumer Criticism of Advertising. Journal of Marketing, 57(3), 18-36.

Noble, C. H., Noble, S. M. and Adjei, M. T. (2012). Let them talk! Managing primary and extended online brand communities for success. Business Horizons, 55 (5), 475-

483.

Slavin, M. (2009). Commentary: Transparency increased firms’ credibility. Daily Journal of Commerce.

Vishwanath, T., & Kaufmann, D. (2001). Toward transparency: new approaches and their application to financial markets. The World Bank Research Observer, 16(1), 41-57

(37)

Appendix

Figure 1. Distribution of original biddings – Philips hue

(38)

Figure 3. Distribution of original biddings – Philips Daily Blender

Figure 4. Distribution of original biddings with algorithm – Philips Daily Blender