Abstract
In today’s cluttered media environment, bloggers have the important advantage of the consumer’s attention. This makes blogs a potentially valuable marketing tool for brands, especially within the tourism industry. The experience-based character of most tourism products is highly compatible with the interactivity and multimedia opportunities that blogs have to offer. Moreover, the friendship-like relationship between bloggers and their readers gives blogs a useful publicity advantage over other media channels. However, recent national and international regulations are now obliging bloggers to add sponsorship disclosures to their sponsored content and very little is known about how these sponsorship disclosures might affect the effectiveness of marketing via blogs.
This study explores the effects of different types of sponsorship disclosures in travel blogs on the activation of blog readers’ persuasion knowledge, the brand publicity
effectiveness and blog readers’ blogger attitudes. Moreover, the moderating effects of brand familiarity, perceived blogger credibility and perceived appropriateness of the sponsorship are also explored. An online experiment (N = 161) with three conditions was used to study the different effects and the results reveal an interesting paradox in consumers’ responses to sponsored content on blogs.
Specifically, consumers seem to be aware of the fact that brands regularly pay bloggers in exchange for their endorsement, even without the presence of sponsorship
disclosures. Nonetheless, consumers’ activated persuasion knowledge actually does not affect their brand attitudes and even increases their purchase intentions, especially when they are familiar with the sponsoring brand. This seemingly paradoxical effect can be explained by correspondence bias. Moreover, consumers’ activated persuasion knowledge also does not seem to affect their blogger attitudes, which contradicts earlier findings in this field. Implications for theory and practitioners are discussed, as well as directions for future research.
Introduction
In the tourism industry, the use of Web 2.0 has emerged faster than anywhere else. By 2014, almost 90% of all Dutch tourism-related companies were using some kind of social media, as opposed to only 53% of the companies in other industries (Centraal Bureau voor de Statistiek [CBS], 2015). Moreover, the majority of travellers are using the Internet as a source of pre-purchase information and so-called “Travel 2.0” platforms like blogs and forums have become incredibly popular among this target group (Doolin, Burgess, & Cooper, 2002; Hernández-Méndez, Muñoz-Leiva, & Sánchez-Fernández, 2015).
In today’s cluttered media environment, bloggers have something that is hard to come by: the consumer’s attention. Consequently, it is easy to see why as much as 53% of tourism brands have now incorporated travel blogs into their media mix (CBS, 2015). Even though some brands might be lucky enough to get featured spontaneously every once in a while, many marketing managers are now paying bloggers to write about their products and services, in an attempt to leverage their influence to brand’s advantage (Colliander & Erlandsson, 2015; Hwang & Jeong, 2016).
However, as bloggers start to make more money, their websites become more professional and they start to attract more readers. In response to this development, national and international regulations are now obliging bloggers to proactively inform their readers about any kind of sponsored content. The goal of these sponsorship disclosures is to protect consumers from covert marketing practices by increasing their awareness of commercial content on blogs (Colliander & Erlandsson, 2015; Hwang & Jeong, 2016; Lu, Chang, & Chang, 2014; Reclame Code Commissie [RCC], 2013). But what do these new regulations mean for the publicity effectiveness of the sponsored brand? And how might a blogger’s reputation be affected if their readers feel that they can no longer rely on them to deliver unbiased information?
RQ: To what extents do sponsorship disclosures in travel blogs activate consumers’ persuasion knowledge and how does this affect their perceptions of the blogger and the sponsoring brand?
Following the growing popularity of blogs as an advertising channel and the instalment of these new sponsorship regulations, more research is needed to explore the effects of sponsorship disclosures on the consumer. This information is essential in order to assess the real value of blog marketing for brands in tourism as well as in other industries, but studies in this area have been quite scarce to date. Only a handful of scholars have tapped into the topic of proactive sponsorship disclosures on blogs (e.g., Colliander, 2012; Hwang & Jeong, 2016; Liljander, Gummerus, & Söderlund, 2015) and, until now, none of these studies have focused on comparing different types of sponsorship disclosures. Therefore, the present study will make an attempt to contribute to this relatively new body of research, by
employing an online experiment to compare the effects of different types of sponsorship disclosures for both the brand and the blogger.
Theory The Value of Blogs for Tourism Marketing
Blogs are journal-like websites that use text and multimedia to communicate personal stories and that are frequently updated with new content (Hernández-Méndez et al., 2015). The personal characters of blogs make readers feel like they are getting an inside look into someone else’s life. It can even facilitate parasocial interaction between the blogger and their audience to the point where readers consider bloggers to be knowledgeable friends that can be trusted to deliver unbiased information (Chiang & Hsieh, 2011; Colliander & Erlandsson, 2015; Johnson & Kaye, 2004). According to Doyle (2007) this word-of-mouth like
valuable for brands in the tourism industry.
Tourism-related products (e.g., hotels, rental cars or destinations) are usually
intangible services and experience products. This information-based character makes it hard to assess their quality prior to purchase, which is why consumers like to rely on detailed and first-hand information for their pre-purchase orientation when they are buying experience goods (Park & Nicolau, 2015; Woodside & King, 2001). Consequently, the marketing of tourism products benefits greatly from the interactivity and multimedia opportunities that blogs have to offer. Moreover, blogs usually feature factual information as well as opinion-based reviews. This provides the reader with a very complete picture of the topic of interest and diminishes the chances of post-purchase disappointment (Hernández-Méndez et al., 2015; Park & Nicolau, 2015; Xiang & Gretzel, 2010).
Seeing as there is such a good fit between the medium and their product, it is no surprise that tourism brands are so interested in collaborating with bloggers. However, this commercial interest has radically changed the blogger landscape, making it much more professional. Many bloggers are now working as freelance reporters who frequently post sponsored content on their channels and it comes to wonder whether they can still be considered a trustworthy, independent source of information for consumers (Colliander & Erlandsson, 2015; Jassies, 2013).
Activating Blog Readers’ Persuasion Knowledge
As identified by Chu and Kamal (2008), sponsored content on blogs will generally appear in one of two forms. Often times the brand information will be embedded into the content in a natural way (e.g., when the blogger mentions the brand in a collective post about their favourite products). However, brand information can also be presented as the sole focus of the content (e.g., when a blogger dedicates an entire post to a specific product or service). Either way, these types of sponsored content are actually more similar to publicity than they
are to advertising, as it can be hard for readers to identify the sponsoring brand (Chu & Kamal, 2008; Eisend & Küster, 2011). More specifically, the somewhat covert nature of brands’ marketing activities on blogs makes it less likely that reader’s persuasion knowledge will be activated.
Persuasion knowledge is developed over time and grows as consumers get in touch with different kinds of persuasive efforts by advertisers. It entails knowledge on when, where and how brands might try to influence their attitudes and behaviours. Once consumers learn to recognize persuasive intent, they can use this knowledge to arm themselves against the
advertiser’s influence (Friestad & Wright, 1994). Consequently, persuasion knowledge can help consumers to recognize ulterior motives when they get in touch with an influence agent. However, the accessibility of persuasion knowledge is subject to the strength of the
association between the influence agent and ulterior motives in the consumer’s mind, as well as to the presence or absence of persuasion cues (Campbell & Kirmani, 2001).
Since bloggers tend to be considered a credible, independent source of information, and sponsored content is often discreetly embedded into the blog’s editorial content,
consumers are not very likely to infer ulterior motives when they are reading blogs. Indeed, readers are even known to accept a blogger’s arguments without any careful consideration (Chu & Kamal, 2008; Johnson & Kaye, 2004). Hence, blog readers are particularly vulnerable to the persuasive intent of brands and this is why bloggers are now legally obliged to add sponsorship disclosures to all of their sponsored content (RCC, 2013).
Sponsorship Disclosures on Blogs
Sponsorships on blogs are typically disclosed through disclaimers or so-called advertorial tags (RCC, 2013). Disclaimers are short texts, usually placed at the bottom of an article or embedded into the editorial content, that are often used to disclose smaller-scale sponsorships. For example, a brand might offer the blogger to use their product or service for
free, in exchange for a review on the blog. In this case, the blogger is free to write whatever they want and hence disclaimers will often emphasize the independency of the blogger despite the sponsorship (Hwang & Jeong, 2016; RCC, 2013).
Whenever brands want to exert more influence over the editorial content, they might pay the blogger to write an advertorial. Advertorials will generally allow the blogger to give their own interpretation to a brand-determined briefing describing a specific format. This kind of sponsored content is usually marked with an advertorial tag to notify the readers (Jassies, 2013; RCC, 2013). Ultimately, the purpose of these different types of sponsorship disclosures is to notify the unsuspecting readers of any commercial interests and to activate their
persuasion knowledge.
However, the effectiveness of these different disclosure types has never been studied before. Generally, a disclaimer will be less visible than an advertorial tag and might thus be less effective in activating readers’ persuasion knowledge. Indeed, several studies in the field of covert marketing disclosures have found that more explicit disclosures are more effective in activating persuasion knowledge (e.g., Nelson & Park, 2015; Wei, Fischer, & Main, 2008). In sum, this leads to the following hypothesis:
H1: Sponsorship disclosures in travel blogs will activate readers’ persuasion knowledge, but disclaimers will be less effective in activating readers’ persuasion knowledge than advertorial tags.
The Correspondence Bias Paradox
The obvious goal that brands pursue when they choose to sponsor a blogger is that their readers will be persuaded to try the product for themselves or, at least, that they become aware of its existence and will develop a favourable attitude. The extent to which brand
publicity on blogs or any other channel has positive effects for the brand is referred to as brand publicity effectiveness (Colliander & Dahlén, 2011; Colliander & Erlandsson, 2015). Following Colliander and Dahlén (2011), this entails everything from increased brand awareness or improved brand attitudes to increased purchase intentions or actual purchases among the target group of the brand. However, the activation of persuasion knowledge through sponsorship disclosures could significantly affect the brand publicity effectiveness.
Indeed, previous studies on persuasion knowledge and brand evaluations have found that activated persuasion knowledge can have detrimental effects for brands and that more overt marketing on blogs can negatively affect purchase intentions and blogger perceptions (Liljander et al., 2015; Wei et al., 2008). However, there is also evidence that sponsorship disclosures might not affect the brand publicity effectiveness. Colliander and Erlandsson (2015) recently studied the effects of a third-party disclosure of blog sponsorship and found that even though it had negative effects for the blogger’s reputation, brand attitudes and purchase intentions were not affected.
Thus, even though readers are aware of the relationship between the blogger and a commercial third-party, they might still believe that the blogger would only agree to the sponsored endorsement when they feel that the products are worthy of their attention, leaving the brand publicity effectiveness unharmed. This apparent paradox is known as
correspondence bias: “The tendency to assume that a person’s behaviour is a true reflection of their beliefs or opinions, and thus, their underlying dispositions, when in fact their behaviour could be explained entirely by situational factors” (Cronley, Kardes, Goddard, & Houghton, 1999, p. 627). Accordingly, it is expected that when blog readers’ persuasion knowledge is activated by a proactive sponsorship disclosure, this will not affect the brand publicity effectiveness.
H2: The activation of travel blog readers’ persuasion knowledge will not affect their attitudes towards the brand.
H3: The activation of travel blog readers’ persuasion knowledge will not affect their purchase intentions for the brand.
The blogger might not be so lucky. Previous studies on blog marketing and sponsorship disclosures have found that readers’ perceptions of the blogger are usually negatively affected when marketing practices are brought to light (Colliander & Erlandsson, 2015; Liljander et al., 2015). Liljander et al. (2015) have suggested that these negative effects might be caused by feelings of disappointment. As consumers usually go online to consult genuine word-of-mouth from an independent source, blog readers might feel betrayed by the blogger once it becomes clear that they are affiliated with a third-party.
H4: The activation of travel blog readers’ persuasion knowledge will have a negative effect on their blogger attitudes.
Brand Familiarity
This study explores the effects of different types of sponsorship disclosures in travel blogs on readers’ persuasion knowledge, brand publicity effectiveness and perceived blogger attitude. However, there might be some audience characteristics that moderate these effects. For example, Colliander and Dahlén (2011) studied overt brand publicity on blogs and found that the brand publicity effectiveness is subject to the audience’s existing relationship with the brand. Moreover, Lu et al. (2014) found that when the audience is familiar with the brand in a sponsored blog post, this will result in higher purchase intentions, because consumers tend to have more trust in brands that they are familiar with.
Thus far, there have also been a few studies that explored the relationship between brand familiarity and persuasion knowledge in different settings and these actually yield very interesting results. Specifically, there seems to be a reversed effect of activated persuasion knowledge on brand evaluations for high-familiarity brands (Eagly, Wood, & Chaiken, 1978; Wei et al., 2008). This means that consumers tend to evaluate high-familiarity brands better when their persuasion knowledge is activated to a higher degree, as a result of more explicit sponsorship disclosures (Wei et al., 2008).
H5: Brand familiarity positively affects the relationship between persuasion knowledge and brand publicity effectiveness, such that higher brand familiarity leads to higher brand attitudes and purchase intentions.
Blogger Credibility
Another factor that seems to influence the brand publicity effectiveness in sponsored blog posts is the audience perception of the blogger’s credibility (Colliander & Dahlén, 2011). Source credibility is generally known to be an important factor in attitude and behaviour change research and credibility is also known to play an important role in the relationship between the blogger and their audience (Johnson & Kaye, 2004; Pornpitakpan, 2004). In a similar context, exploring the effects of online hotel reviews, Xie, Miao, Kuo and Lee (2011) have already established that higher perceived source credibility can result in higher purchase intentions for the brand.
Considering sponsored content on blogs, which is often closer to publicity than to actual advertising, brands might also profit from high source credibility. Publicity is often more effective than advertising for the very reason that the source is perceived to be more independent from the commercial party and thus more credible (Eisend & Küster, 2011).
Source credibility can enhance the attitude towards the message, which in turn leads to higher brand attitudes and ultimately even to higher purchase intentions (Pornpitakpan, 2004).
H6: Blogger credibility positively affects the relationship between persuasion knowledge and brand publicity effectiveness, such that higher blogger credibility leads to higher brand attitudes and purchase intentions.
Perceived Appropriateness of the Sponsorship
Along with the development of persuasion knowledge, consumers also develop beliefs about the appropriateness or fairness of persuasive strategies (Friestad & Wright, 1994). Following the instalment of the sponsorship disclosure regulations, and the on-going professionalization of the blogger landscape, some blog readers might have started to
acknowledge that brand sponsorships are a vital part of how blogs work today. Indeed, many Figure 1. Conceptual model of the hypotheses.
big and popular blogs might look quite different if they did not get paid for their endorsement. Wei et al. (2008) studied the relationships between covert marketing disclosures, persuasion knowledge and perceived appropriateness in a different setting and they found that perceived appropriateness could actually mitigate the negative effects of persuasion knowledge. Thus, as long as the blogger proactively informs their readers about the brands involvement, blogger attitudes might not get corrupted when the reader finds the sponsorship to be appropriate.
H7: Appropriateness reduces the negative effect of persuasion knowledge on blogger attitude.
Method Design
To test the hypotheses, which are summarized in Figure 1, this study employed an online, fully randomized between-subjects experimental design with three experimental groups (no sponsorship disclosure, sponsorship disclosure through a disclaimer, sponsorship disclosure through an advertorial tag). This way, the effects of different types of sponsorship disclosures in travel blogs on readers’ persuasion knowledge, the brand publicity
effectiveness and the readers’ blogger attitudes could be compared and any differences could be identified.
Participants and Procedure
The experiment was set up in online research and insights platform Qualtrics and was made accessible through a link that was spread through e-mail and social media, along with the request to share the link with others. Thus, a snowball convenience sample was used to collect participants for this study. This sampling method was used, because it is an effective way to approach a significant number of prospective participants within a relatively short
timeframe. The final sample consisted of 161 Dutch adults, 26% men and 74% women, between the ages of 18 and 76 (M = 31.88, SD = 14.11). In order to take part in the
experiment, participants had to be 18 years or older and users of online travel platforms. This was important, because it increased the chances of the participants being familiar with travel blogs or being potential readers of travel blogs.
Before the link to the experiment was distributed for the actual study, a pretest was conducted among 15 participants, who met the sampling inclusion criteria, in order to check the quality of the questionnaire and make any possible corrections or improvements to the measures or manipulations. The results from the pretest revealed that the wording in some of the questions might be too vague, which made the questions hard to understand for the
participants. This was possibly due to the translation of the pre-existing scales from English to Dutch. This might have also been the reason that as much as five participants had trouble answering the questions on the persuasion knowledge scale and subsequently left these questions open without answering them. To fix these issues, the wording of the questions was adjusted with the suggestions of the pretest participants in mind. Moreover, a response request validation was added to all questions. This validation notified participants when they failed to answer one of the questions and requested them to answer it, which hopefully contributed to the reduction of the number of missing values in the actual study.
For the actual study, participants accessed the experiment by clicking on a Qualtrics link, which led them to a landing page with an explanation about the goal of the study, along with a disquisition of their rights and privacy as a participant in the experiment. Participant’s permission to participate was received actively through the means of a yes/no question. In the introduction to the study, a mild cover story was provided to conceal the actual purpose of the study. This was done to prevent that participants’ suspicions about the experiment’s goal would influence their responses. Participants were told that the goal of the research was to
gain more insight into the way tourists use travel blogs to gather information before they book a vacation or a getaway.
As soon as the participants indicated that they wanted to participate in the experiment, they were referred to the first set of questions on their background variables. Secondly, the participants were asked to read a mock-up of a travel blog that contained either no
sponsorship disclosure, a disclaimer at the end of the text or an advertorial tag. Participants could not skip to the next page for the first 30 seconds, in order to ensure that nobody would skip the manipulation without actually paying attention to the blog post. Finally, after the manipulation, participants were asked to answer some more questions about the blog post that they just saw. The experiment ended with a short text, thanking the participants for their time and informing them about the actual goal of the study. A link to the full Dutch Advertising Code for Social Media was also provided, in case participants wanted to read some more information on these regulations.
Manipulation
The manipulations for this study consisted of mock-ups of a travel blog post that either contained no sponsorship disclosure, a disclaimer at the end of the text or an advertorial tag. The interface of the mock-up blog was created in Photoshop, after the example of an existing popular Dutch travel blog, to ensure external validity (https://www.wearetravellers.nl). After studying the existing blog, the following features were added to the manipulation: a short introduction text on the blogger, accompanied by a picture of the blogger, links to different social media channels, advertising banners on the side of the blog post, a comment section and links to related articles. This way, the mock-up blog “The Travel Diaries” was created, supposedly owned by blogger Lisanne (see Figure 2 for an example).
As a next step, the blog post text was created and Photoshop was used to embed the text into the mock-up interface. The text was the same across all three conditions, as the
manipulation only varied in type of
sponsorship disclosure. The content of the text was based on a small content analysis of 25 sponsored blog posts that appeared across five existing travel blogs, once again to improve the external validity of the manipulation (https://www.travellab.nl, https://www.trip-temptation.com, https://www. reismeisje.nl, https://www.wearetravellers.nl,
https://www.reishonger.nl). Qualitative research from Kozinets, de Valck, Wojnicki and Wilner (2010) has identified four distinct types of narrative strategies in word-of-mouth marketing via bloggers, which can be
identified through several text characteristics. The small-scale content analysis revealed that the most used narrative strategy across all 25 sponsored travel blog posts was the
“evaluation” strategy. This strategy is
characterized by a focus on the product rather than the sponsorship, a minimization of mention of the sponsorship within the text and a strong communal acknowledgement, where the blogger actively addresses their readers and uses terms like “we” and “you”, rather Figure 2. Example of the manipulation.
than only talking from their own perspective (Kozinets et al., 2010). An overview of the results of the content analysis can be found in Appendix A.
Consequently, the final manipulation text was also constructed following an evaluative narrative strategy. It consisted of 328 words in which the blogger, Lisanne, described the first day out of a roadtrip along the Pacific Coast Highway in California. The text contained a short introduction with general information about the Pacific Coast Highway and the places to visit along the way. Furthermore, it contained information about the car she picked up from a car rental company to use for her trip and the great service that she received from this
company. Finally, at the end of the text, the blogger makes a recommendation to the readers about renting a car with this company for their next road trip. For the purpose of authenticity of the blog post, an existing car rental brand was used as the sponsoring brand.
The last thing that was added in Photoshop was a sponsorship disclosure in two of the three conditions, in the form of either a disclaimer text or an advertorial tag. The disclaimer text was based on several examples from existing travel blogs that were investigated for the small-scale content analysis. The three mock-ups were then saved as JPG files to look like screenshots of the actual website and were inserted into the Qualtrics questionnaire. The details and contents of all three manipulations can be found in Appendices B through D. Measures
As this experiment was conducted among Dutch participants, all original scales were translated to Dutch before using them in the questionnaire. The pre-test revealed that these translations had no significant influence on the reliability of the scales. A full overview of the original Dutch questionnaire can be found in Appendix E.
Control variables. Some important demographic variables were measured using a set of four questions about the participants’ age, their gender and their use of online travel
man (1) or a woman (2). The participants’ age was measured using an open-ended self-report question and participants’ use of online travel platforms was measured asking participants: “Do you use online travel platforms?” (1 = yes and 2 = no). Whenever a participant indicated that they use online travel platforms, they were subsequently asked to indicate which
platforms they would normally use (e.g., travel blogs, review websites like TripAdvisor, or travel forums like the LonelyPlanet travellers forum).
Two more measures were added to the questionnaire for variables that were believed to possibly influence the results of the study and that might need to be controlled for later on. First of all, participants were asked whether they had visited the Pacific Coast Highway in California before. This question could be answered with a simple yes (1) or no (2) and was added because this location was an important part of the manipulation and any pre-existing emotions (e.g., happiness linked to memories of a vacation) might have influenced the participants’ scores on the dependent variables. Secondly, the participants were asked to rate the blog post on a scale from 1 to 10, in order to control for any differences in their liking of the blog post.
Finally, following the example of Wei et al. (2008), a measure was added to control for individual differences in pre-existing persuasion knowledge. This was done using a six-item measure by Bearden, Hardesty and Rose (2001), which consisted of the following six-items (1 = strongly disagree and 7 = strongly agree): (1) "I know when an offer is 'too good to be true'," (2) "I can tell when an offer has strings attached," (3) "I have no trouble understanding bargaining tactics used by salespeople," (4) "I know when a marketer is pressuring me to buy," (5) "I can see through sales gimmicks used to get consumers to buy," and (6) "I can separate fact from fantasy in advertising". Principal components factor analysis revealed that these six items formed a single scale (eigenvalue = 4.04) and explained 67.33% of the variance in the participants’ existing persuasion knowledge. The scale displayed excellent
reliability (α = .90) and ranged from 1.50 (little persuasion knowledge) to 7 (great persuasion knowledge), which means that the higher a participant scored on the scale, the more pre-existing persuasion knowledge he or she had (M = 5.49, SD = 0.84).
Manipulation check. In order to check whether the manipulation had been successful, a manipulation check question was added at the very end of the questionnaire. Participants were asked whether they had been notified of the sponsored content in the blog post through means of a disclaimer at the bottom of the text (1), an advertorial tag (3) or whether they had not been notified (3). The manipulation check was added at the end of the questionnaire, to prevent the question from influencing the participants’ answers to the other questions.
Persuasion knowledge. The degree to which the participants’ persuasion knowledge was activated by the blog post, was measured on a 7-point Likert scale using a single question adapted from Wei et al. (2008): “The brand in the blog post was mentioned because it paid to be mentioned” (1 = strongly disagree and 7 = strongly agree).
Blogger attitude. The participants’ attitudes towards the blogger were measured with a three-item measure that was adapted from Mackenzie and Lutz (1989): “My impression of Lisanne, the writer of The Travel Diaries, is good”, “My impression of Lisanne, the writer of The Travel Diaries, is pleasant” and “My impression of Lisanne, the writer of The Travel Diaries, is favourable”. All items could be answered on a 7-point Likert scale (1 = strongly disagree and 7 = strongly agree). Principal components factor analysis showed that the three items loaded on one factor (eigenvalue = 2.71), which explained 90.30% of the variance in the participants’ attitudes toward the blogger. This scale also displayed excellent reliability (α = .95) and ranged from 1 (very negative attitude towards the blogger) to 7 (very positive attitude towards the blogger). Consequently, a higher score on the scale indicated a better attitude towards the blogger (M = 4.48, SD = 1.41).
Lutz (1989) and also consisted of three items that could be answered on a 7-point Likert scale (1 = strongly disagree and 7 = strongly agree): “My impression of Sixt is good”, “My
impression of Sixt is pleasant” and “My impression of Sixt is favourable”. The principal components factor analysis revealed that all three items formed one scale (eigenvalue = 2.72). The three items together accounted for 90.76% of the variance in the participants’ brand attitudes. The scale turned out to be very reliable (α = .95) and ranged from 1 (very negative attitude towards the brand) to 7 (very positive attitude towards the brand), meaning that a higher score on the scale indicated a more positive attitude towards the brand (M = 4.42, SD = 1.14).
Purchase intention. Purchase intention was measured with one item that could be answered on a 7-point Likert scale (1 = strongly disagree and 7 = strongly agree): “For my next vacation, I would like to rent a car at Sixt”. This measure was adapted from Sönderlund and Öhman (2003).
Brand familiarity. The brand familiarity measure was adapted from Kent and Allen (1994). It measured how familiar and experienced the participants were with the Sixt brand and how knowledgeable they were about the Sixt brand (1 = not at all familiar and 7 = very familiar, 1 = not at all experienced and 7 = very experienced, 1 = not at all knowledgeable and 7 = very knowledgeable). Once again, principal components factor analysis was used and showed that all three items loaded on one factor (eigenvalue = 2.34), explaining 77.97% of the variance in the participants’ brand familiarity. This brand familiarity scale turned out to be very reliable (α = .85), ranging from 1 (not at all familiar with the brand) to 7 (very familiar with the brand), which means that the higher a participant scored on the scale, the more familiar he or she was with the brand (M = 2.81, SD = 1.64).
Blogger credibility. The perceived credibility of the blogger was measured with a three-item measure that was once again adapted from Mackenzie and Lutz (1989): “The
blogger is convincing”, “The blogger is believable” and “The blogger is unbiased” (1 = strongly disagree and 7 = strongly agree). Principal components factor analysis showed that these three items loaded on one factor (eigenvalue = 2.39) and explained 79.61% of the variance in the participants’ perceived credibility of the blogger. The scale displayed good reliability (α = .87). It ranged from 1 (low blogger credibility) to 6 (high blogger credibility), which means that the higher a participant scored on the scale, the more they perceived the blogger to be credible (M = 3.77, SD = 1.32).
Table 1
Reliability and Descriptive Statistics for the Scales
Scale Items α M SD
Existing persuasion knowledge 6 .90 5.49 0.84
Blogger attitude 3 .95 4.48 1.41
Brand attitude 3 .95 4.42 1.14
Brand familiarity 3 .85 2.81 1.64
Blogger credibility 3 .87 3.77 1.32
Appropriateness 3 .90 4.54 1.49
Appropriateness. Finally, the perceived appropriateness of the sponsorship on the blog was measured with three items, to be answered on a 7-point Likert scale. This measure was adapted from Wei et al. (2008): “It seems acceptable to me if the brand had paid to be featured on the blog”, “It seems fair to me if the brand had paid to be mentioned on the blog” and “It seems normal to me if the brand had paid to be mentioned on the blog” (1 = strongly disagree and 7 = strongly agree). The last principal components factor analysis revealed that these three items also formed a single scale (eigenvalue = 2.48), explaining 82.72% of the variance in the participants’ perceived appropriateness of the sponsorship. This scale also displayed excellent reliability (α = .90) and ranged from 1 (low perceived appropriateness) to
7 (high perceived appropriateness). This means that the higher a participant scored on the scale, the more they believed the sponsorship to be appropriate (M = 4.54, SD = 1.49). For a summary of all reliability measures and descriptive statistics of the scales, please refer to Table 1.
Results
A first inspection of the data revealed that 39 participants failed to finish the questionnaire. Of those 39 participants, 27 people only finished a small part of the questionnaire and did not answer any questions related to the dependent variables in this study. Thus, their participation was deemed to be irrelevant to the results and these
participants were removed from the dataset. The other 12 people, who did answer at least one of the questions related to the dependent variables, were taken into account and all missing values were recoded to 99.
As a next step, possible outliers were identified that might influence the results of the study. Out of all participants, only 10 people turned out to be outliers on more than one dependent variable or moderator variable (z-score < -2.20 or > 2.20). After an inspection of their division among the different conditions, it was chosen not to remove these outliers from the sample, as most of them were found to be in the advertorial tag condition, which was already the condition with the least amount of participants (n = 42).
Taking a closer look at the final sample, 87.30% used online travel platforms and among the most frequently used platforms were travel review websites (72.70%), corporate travel websites (57.10%), travel blogs (31.10%), and travel social networks (29.80%). Fewer participants used travel forums (3.10%) or other online travel platforms (18.00%). Only 15.50% of the participants had visited the Pacific Coast Highway before and, on average, participants rated the blog post moderately positive, even though the ratings differed quite a bit from participant to participant (M = 5.86, SD = 1.82).
In order to use them in further analyses, several of the variables were recoded into dichotomous variables. The variables that were recoded into dichotomous variables are gender (1 = man, 0 = woman), online travel platform use (1 = yes, 0 = no) and Pacific Coast Highway experience (1 = yes, 0 = no). Also, the categorical variable “condition” was created to indicate the condition that each of the participants had been assigned to (1 = advertorial tag condition, 2 = disclaimer condition, 3 = control condition).
Subsequently, to check whether the manipulation had been successful, a chi-square analysis was performed with condition as the independent variable and the manipulation check measure as the dependent variable. The manipulation check revealed that the
manipulation had been successful, Χ2(4) = 18.50, p < .01. There turned out to be a very weak, positive correlation between the condition that participants had been exposed to and their score on the manipulation check question (τ = .06).
The correlations for all continuous dependent and independent variables are displayed in Table 2. As can be inferred from the Pearson’s correlations, several of these variables are significantly correlated to one another. For example, existing persuasion knowledge is significantly and positively correlated to persuasion knowledge activation. This means that someone’s persuasion knowledge is more likely to be activated by reading a sponsored blog post, when his or her existing persuasion knowledge is already higher than that of someone else. Moreover, blog post rating turns out to be significantly and negatively correlated to the mediator variable persuasion knowledge. Thus, when reading the sponsored blog post activates someone’s persuasion knowledge more, they will rate the blog post less positively than someone whose persuasion knowledge is activated to a lesser degree.
Blog post rating is also significantly and positively correlated to the dependent variables blogger attitude, brand attitude and purchases intention, as well as to the moderator variable blogger credibility. This means that if someone likes the blog post better, this will
also result in better brand attitudes, more positive blogger attitudes and higher purchase intentions. Additionally, someone who finds the blogger to be more credible, will rate the blog post more positively as well. The aforementioned significant correlations between existing persuasion knowledge, blog post rating and the other dependent and independent variables, indicate that these might need to be taken into account as covariates in further analysis, unless the variables were evenly distributed among the different conditions.
Table 2
Pearson’s Correlations Matrix
1 2 3 4 5 6 7 8 9
1. Ex. persuasion knowledge r 1.00 N 155
2. Blog post rating r -.04 1.00 N 155 161 3. Persuasion knowledge r .39** -.33** 1.00 N 155 156 156 4. Blogger attitude r .03 .74** -.18* 1.00 N 155 161 156 161 5. Brand attitude r .08 .29** .02 .23** 1.00 N 155 155 155 155 155 6. Purchase intention r .08 .36** .10 .33** .50** 1.00 N 155 155 155 155 155 155 7. Brand familiarity r .07 .01 .18* -.04 .45** .28 ** 1.00 N 155 155 155 155 155 155 155 8. Blogger credibility r .05 .69** -.36** .70** .34** .39** .02 1.00 N 153 153 153 153 153 153 153 153 9 Appropriateness r .09 .06 .23** .10 .18* .23** .09 .15 1.00 N 152 152 152 152 152 152 152 152 152 Note. *p < .05 (2-tailed), **p < .01 (2-tailed).
To check the randomization of the continuous control variables, a one-way ANOVA was performed. The analysis revealed that none of these variables were significantly related to the condition that the participants had been assigned to (see Table 3). This means that randomization has been successful and age, blog post rating and existing persuasion
knowledge do not have to be taken in to account as covariates in further analysis. Finally, some more randomization checks were performed for the dichotomous background variables like gender and online travel platform use. Chi-square analyses showed that randomization has also been successful for these variables and they do not need to be taken into account as covariates in further analyses (see Table 4).
Table 3
Randomization Checks for the Continuous Variables
F df1, df2 p-value (two-tailed)
Age 0.31 2, 158 .73
Existing persuasion knowledge 0.08 2, 152 .93
Blog post rating 0.25 2, 158 .78
Table 4
Randomization Checks for the Dichotomous Variables
Χ2 df p-value (two-tailed)
Gender 3.16 2 .20
Travel platform use 1.82 2 .40
Travel review websites 1.66 2 .44
Travel social networks 1.09 2 .58
Travel blogs 2.03 2 .36
Corporate travel websites 0.04 2 .98
Travel forums 4.78 2 .09
Different travel platforms 4.96 2 .08
Pacific Coast Highway experience 2.54 2 .28
Regression Analyses
As a first step in the hypotheses testing, three single regression analyses were performed to assess the isolated effects of the mediator variable persuasion knowledge on the dependent variables. The results of these three regression analyses can be found in Table 5. It appears that persuasion knowledge is only significantly related to blogger attitude, F(1, 153)
= 5.27, p < .05, and that it only predicts a small 3% of the variance in this dependent variable (R2 = .03). The correlation appears to be a weak and negative relationship, meaning that someone’s attitude towards the blogger will be more negative as their persuasion knowledge is activated to a higher degree, which appears to be in line with H4. On the other hand, there are no significant correlations between persuasion knowledge and brand attitude, F(1, 153) = 0.06, p = .81, nor between persuasion knowledge and purchase intention, F(1, 153) = 1.45, p = .23. This means that activated persuasion knowledge does not seem to affect the brand publicity effectiveness, which is in line with H2 and H3.
Table 5
Standardized Regression Coefficients for the Effects of Persuasion Knowledge on Blogger Attitude, Brand Attitude and Purchase Intention
Persuasion knowledge Coeff. t 95% CI Blogger attitude -.18* -2.30 -0.29, -0.02 Brand attitude .02 0.24 -0.10, 0.13 Purchase intention .10 1.21 -0.05, 0.20 Note. N = 155. *p < .05 (one-tailed).
Subsequently, to gain a better understanding of the different relationships within the model, the three moderator variables were added to each of the regression models. The first of these three multiple regression models is displayed in Table 6 and is significant, F(4, 147) = 17.81, p < .001. The independent variables explain 33% of the variance in brand attitude (R2 = .33). Taking a closer look at this model reveals that H2 is once again confirmed: Persuasion knowledge activation as a result of reading a sponsored blog post has no significant effect on someone’s brand attitude. On the other hand, both brand familiarity and blogger credibility are positively and significantly correlated with brand attitude, meaning that readers’ brand attitude will improve when they are familiar with the brand or when they perceive the blogger
as credible. Also, it is still possible that the relationship between persuasion knowledge and brand attitude is completely moderated by either one of the moderator variables, which will be explored in further analyses.
Table 6
Standardized Regression Coefficients with Confidence Intervals Estimating Brand Attitude
Brand attitude Coeff. t 95% CI Persuasion knowledge 0.003 0.04 -0.10, 0.11 Brand familiarity 0.43* 6.18 0.20, 0.38 Blogger credibility 0.06* 4.49 0.16, 0.41 Appropriateness 0.09 1.27 -0.04, 0.17 Note. N = 152. *p < .001 (one-tailed).
The multiple regression model with purchase intention as the dependent variable is also significant, F(4, 147) = 14.89, p < .001 (see Table 7). The independent variables explain 29% of the variance in purchase intention and the model displays some interesting results (R2 = .29). Specifically, there appears to be a significant relationship between persuasion
knowledge and purchase intention now that the other independent variables are added to the model. This means that H3 is rejected, which contradicts the results of the single regression analysis. However, there are also significant positive correlations between brand familiarity, blogger credibility and purchase intention and it is possible that these independent variables moderate the relationship between persuasion knowledge and purchase intention, as
hypothesized in H5 and H6.
Finally, the multiple regression model with blogger attitude as the dependent variable is also significant F(4, 147) = 38.21, p < .001 (see Table 7). The independent variables explain 51% of the variance in blogger attitude (R2 = .51). Taking a closer look at the model, it appears that there is no longer a significant correlation between persuasion knowledge and
blogger attitude now that the other independent variables are added to the model. Despite the results of the single regression analysis, this means that one of the other independent variables is a better predictor for blogger attitude and that H4 should be rejected: Activated persuasion knowledge does not have a negative effect on blog readers’ blogger attitude.
Table 7
Standardized Regression Coefficients with Confidence Intervals Estimating Blogger Attitude and Purchase Intention
Blogger attitude Purchase intention
Coeff. t 95% CI Coeff. t 95% CI
Persuasion knowledge 0.11 1.62 -0.02, 0.21 0.20* 2.57 0.04, 0.30 Brand familiarity -0.07 -1.21 -0.16, 0.04 0.24** 3.41 0.08, 0.30 Blogger credibility 0.75** 11.71 0.68, 0.95 0.45** 5.76 0.29, 0.59 Appropriateness -0.04 -0.61 -0.15, 0.08 0.10 1.28 -0.05, 0.21 Note. N = 152. *p < .01 (one-tailed), **p < .001 (one-tailed).
Furthermore, the model also does not show any significant relationship between the perceived appropriateness of the sponsorship and the blogger attitude. This means that H7 is also rejected: Appropriateness does not reduce the negative effect of persuasion knowledge on blogger attitude, as a moderation effect requires a significant effect between the moderator and the dependent variable. There is however, a significant and positive relationship between blogger credibility and blogger attitude, meaning that readers will have more positive blogger attitudes when they perceive the blogger to be more credible.
PROCESS Models for Moderated Mediation
In order to take a closer look at all the different relationships between the variables, the PROCESS macro was used to test the models as a whole (Hayes, 2015). The model with blogger attitude as the dependent variable and blogger credibility as a single possible
mediation (see Figure 3). In addition, both models with purchase intention and brand attitude as the dependent variables and blogger credibility and brand familiarity as possible
moderators will be tested using PROCESS macro model 16 for second-stage moderated mediation with two moderators (see Figure 3).
In order to use the categorical predictor condition in the PROCESS models, it was recoded into two dummy variables for the advertorial tag and the disclaimer conditions, with the control condition being used as the reference group in the analyses (constant). In each of the analyses, one of these dummy variables was added as the predictor, while the other one was added to the model as a covariate.
Figure 3. Conceptual diagrams for PROCESS model 14 and model 16.
Model 14 and 16: estimating persuasion knowledge (first part). The first part of the models describes the estimation of the mediator variable persuasion knowledge as a result of reading a sponsored blog post that contains either no sponsorship disclosure (control condition), a disclaimer or an advertorial tag (see Table 8). On average, readers’ persuasion knowledge was fairly activated by reading the sponsored blog post (M = 5.39, SD = 1.65). However, this part of the model is not significant, F(2, 150) = 0.95, p = .39. Meaning that
none of the conditions had a significant effect on the activation of the readers’ persuasion knowledge and H1 is rejected. Specifically, sponsorship disclosures in blogs do not activate readers’ persuasion knowledge.
Table 8
Unstandardized OLS Regression Coefficients with Confidence Intervals and Their Significance Estimating Persuasion Knowledge Activation After Reading a Sponsored Blog Post
Persuasion knowledge (M)
Coeff. SE t 95%-CI p-value (one-tailed)
Constant (Control) -0.19 .22 -0.84 -0.62, 0.25 .20
Disclaimer (X) 0.19 .32 0.60 -0.44, 0.82 .27
Advertorial (X) 0.43 .31 1.38 -0.19, 1.05 .09
Note. N = 153
Model 14: estimating blogger attitude (second part). The second part of model 14, which estimates blogger attitude, is significant, F(5, 147) = 36.95, p < .001. The type of sponsorship disclosure that someone has been exposed to, the extent to which their persuasion knowledge is activated and their perception of the bloggers credibility account for 52% of the variance in blogger attitude (R2 = .52). Taking a closer look at the model, it reveals that there is no significant effect of persuasion knowledge on blogger attitude, meaning that the
activation of readers’ persuasion knowledge does not affect their blogger attitude. This is in line with the results of the first multiple regression analyses and rejects H4 (see Table 9).
However, there is a significant and positive effect of the disclaimer condition on the blogger attitude, meaning that people have higher blogger attitudes when a blog post
sponsorship is disclosed through a disclaimer text than when a sponsorship is not disclosed at all. Moreover, there is also a highly significant and positive effect of blogger credibility on blogger attitude: If readers perceive the blogger to be more credible this results in more positive blogger attitudes. However, importantly, there is no interaction effect between
persuasion knowledge and blogger credibility. This means that there is no moderation effect and that the effect of blogger credibility on blogger attitude is a direct effect. The results for this PROCESS analysis are also summarized in Figure 4.
Table 9
Unstandardized OLS Regression Coefficients with Confidence Intervals Blogger Attitude with Persuasion Knowledge as a Mediator and Blogger Credibility as a Moderator
Blogger attitude (Y)
Coeff. SE t 95%-CI Constant (Control) 4.34** .14 30.62 4.06, 4.62 Disclaimer (X) 0.35* .18 1.93 -0.01, 0.71 Advertorial (X) -0.06 .24 -0.23 -0.53, 0.42 Persuasion knowledge (M) 0.08 .05 1.55 -0.02, 0.18 Blogger credibility (V) 0.81** .06 13.26 0.69, 0.93 M*V 0.02 .03 0.59 -0.05, 0.06
Note. N = 153. *p < .05 (one-tailed), ** p < .001 (one-tailed).
Model 16: estimating purchase intention and brand attitude (second part). The Figure 4. Statistical diagram for PROCESS model 14 estimating blogger attitude.
PROCESS model 16 with brand attitude as the dependent variable is also significant, F(7, 145) = 9.22, p < .001, and explains 33% of the variance in brand attitude (R2 = .33). Just like in all previous regression analysis, there is no significant effect between persuasion
knowledge and brand attitude, meaning that the activation of readers’ persuasion knowledge does not affect their brand attitude, once again confirming H2 (see Table 10).
Table 10
Unstandardized OLS Regression Coefficients with Confidence Intervals Estimating Brand Attitude with Persuasion Knowledge as a Mediator and Blogger Credibility and Brand Familiarity as Moderators
Brand attitude (Y)
Coeff. SE t 95%-CI Constant (Control) 4.34* .16 27.11 4.02, 4.65 Disclaimer (X) 0.06 .20 0.32 -0.34, 0.46 Advertorial (X) 0.11 .21 0.50 -0.32, 0.53 Persuasion knowledge (M) 0.06 .08 0.80 -0.09, 0.21 Blogger credibility (V) 0.31* .09 3.44 0.13, 0.49 Brand familiarity (Q) 0.30* .05 5.46 0.19, 0.41 M*V -0.04 .06 -0.70 -0.16, 0.07 M*Q 0.11 .04 0.44 -0.32, 0.53 Note. N = 153. *p < .001 (one-tailed).
A closer look at the second part of PROCESS model 16 with purchase intention as the dependent variable, reveals that this model is also significant, F(7, 145) = 8.32, p < .001. It explains 57% of the variance in the dependent variable (R2 = .57). Just like in the earlier multiple regression analysis, persuasion knowledge has a highly significant and positive effect on purchase intention. This means that people whose persuasion knowledge is activated to a higher extent actually have higher purchase intentions, rejecting H3 (see Table 11).
Moreover, looking at the effects of the two moderator variables on brand attitude and purchase intention, which are displayed in Table 10 and Table 11, all correlations turn out to be significant. Thus, people who find the blogger more credible and people who are more
familiar with the brand have higher brand attitudes and higher purchase intentions. However, there are no significant interaction effects between persuasion knowledge and blogger
credibility. This means that the relationships between persuasion knowledge, brand attitude and purchase intention are not moderated by readers’ perceived blogger credibility, which rejects H6.
Table 11
Unstandardized OLS Regression Coefficients With Confidence Intervals Estimating Purchase Intention With Persuasion Knowledge as a Mediator and Blogger Credibility and Brand Familiarity as Moderators
Purchase intention (Y)
Coeff. SE t 95%-CI Constant (Control) 3.99*** .16 24.72 3.67, 4.31 Disclaimer (X) 0.05 .21 0.24 -0.36, 0.46 Advertorial (X) -0.24 .25 -0.96 -0.74, 0.26 Persuasion knowledge (M) 0.23*** .06 3.57 0.10, 0.36 Blogger credibility (V) 0.47*** .07 6.71 0.33, 0.61 Brand familiarity (Q) 0.18** .06 2.88 0.06, 0.31 M*V -0.05 .04 -1.45 -0.13, 0.02 M*Q 0.09* .04 2.42 0.02, 0.17
Note. N = 153. *p < .05 (one-tailed), **p < .01 (one-tailed), ***p < .001 (one-tailed).
Finally, there is a significant positive interaction effect between persuasion knowledge and brand familiarity for the purchase intention variable, but not for the brand attitude
variable. This means that the relationship between persuasion knowledge and purchase intention is moderated by brand familiarity. Hence, H5 is partially confirmed: The positive relationship between persuasion knowledge and purchase intention is reinforced by brand familiarity. However, the correlation between brand familiarity and brand attitude is a direct effect and readers’ brand familiarity does not affect the relationship between persuasion knowledge and brand attitude. The statistical diagrams for both PROCESS models 16 can be found in Figure 5.
Figure 5. Statistical diagrams for PROCESS models 16 estimating brand attitude and purchase intention.
Table 12
Summary of the Hypotheses and Outcomes
Hypothesis # Prediction Outcome
H1 Sponsorship disclosures in travel blogs will activate readers’ persuasion knowledge, but disclaimers will be less effective in activating readers’ persuasion knowledge than advertorial tags.
Rejected
H2 The activation of travel blog readers’ persuasion knowledge will not affect their attitudes towards the brand
Confirmed
H3 The activation of travel blog readers’ persuasion knowledge will not affect their purchase intentions for the brand.
Rejected
H4 The activation of travel blog readers’ persuasion knowledge will have a negative effect on the readers’ blogger attitude.
Rejected
H5 Brand familiarity positively affects the relationship between persuasion knowledge and brand publicity effectiveness, such that higher brand familiarity leads to higher brand attitudes and purchase intentions.
Partially confirmed
H6
Blogger credibility positively affects the relationship between persuasion knowledge and brand publicity effectiveness, such that higher blogger credibility leads to higher brand attitudes and purchase intentions.
Rejected
H7 Appropriateness reduces the negative effect of persuasion knowledge on blogger attitude.
Rejected
Discussion
This study set out to explore the effects of different types of sponsorship disclosures in travel blogs on the activation of blog readers’ persuasion knowledge, the brand publicity effectiveness and blog readers’ blogger attitudes. Moreover, the moderating effects of different audience characteristics were also studied, namely that of brand familiarity, perceived blogger credibility and perceived appropriateness of the sponsorship. A
summarized overview of the findings can be found in Table 12 and the first thing that the results show is that even though readers’ persuasion knowledge was fairly activated after reading the sponsored blog post, this effect was not caused by the blog post manipulation.
Specifically, as there were no significant relationships between either one of the conditions and the persuasion knowledge variable, neither the disclaimer nor the advertorial tag, nor the blog post without a sponsorship disclosure was successful in activating the readers’ persuasion knowledge. This finding surprisingly contradicts the results of existing studies into sponsorship disclosures and persuasion knowledge activation. Even though there have not been any studies in this specific context, studies in both radio and television settings have proven that sponsorship disclosures are usually very effective in activating consumers’ persuasion knowledge (Boerman, van Reijmersdal, & Neijens, 2012; Boerman, van
Reijmersdal, & Neijens, 2014; Wei et al., 2008;).
This leaves to wonder why the current study did not find similar results. One possible explanation is that participants in the sample were already quite high in existing persuasion knowledge, as the findings also show that people who are higher in existing persuasion knowledge are more likely to have their persuasion knowledge activated. Thus, in this case there might be a ceiling effect at play. Namely, when consumers are already very much aware of the fact that blog posts can be sponsored, sponsorship disclosures might not have any added value in warning them of such marketing practices.
However, it is also possible that the results on the persuasion knowledge variable were subject to a test effect. No elaborate cover story was used for this online experiment and the goal of the study might not have been too hard to infer for the participants. Hence, it is possible that when they were asked whether or not they thought that the blog post was sponsored, many of them answered favourably, as they might have thought that was the response that the researcher was looking for. Consequently, future studies in this line of research might want to use a cover story in order to prevent such test effects.
Interestingly, the results of this study also show that sponsorships on blogs actually have very positive effects for brands. Despite the activation of their persuasion knowledge,
blog readers’ brand attitudes were not affected and their purchase intentions even appeared to improve when their persuasion knowledge is activated to a higher degree, especially when the readers are familiar with the brand. Moreover, readers who were familiar with the brand had higher brand attitudes, just like readers who perceived the blogger to be more credible. This is in line with the results that Colliander and Erlandsson (2015) found in their study on third-party disclosure of blog sponsorship and supports the existence of a correspondence bias for blogs. Even though consumers are aware of the fact that a brand might have paid the blogger to be mentioned, they still seem to believe that the blogger would only write about brands that they truly like and are considered worthy of their attention.
Notwithstanding these results, it is interesting that the readers’ purchase intentions were affected, while their brand attitudes were not. A possible explanation for this differential effect might be derived from Nedungadi (1990), who notes that the choice for a particular brand is not always related to consumers’ brand evaluations. Rather, brand consideration or even actual purchase can occur because there are no alternative options available or because the brand in question is highly active in the consumer’s mind (Nedungadi, 1990). Thus, it can happen that consumer purchase intentions increase while their brand attitudes remain stable.
Similarly, it is possible that reading the sponsored blog post activated the existence of the Sixt brand in the readers’ minds. Moreover, the blog post only covered one brand and did not discuss any competitors that the readers could have added to their consideration set. Consequently, when they were asked whether or not they would like to rent a car at Sixt, the circumstances would have made this brand look like a logical choice, resulting in relatively high purchase intentions regardless of the readers’ brand attitudes (Nedungadi, 1990). In any case, this differential effect of persuasion knowledge on the brand publicity effectiveness is an interesting direction for future research.
that the sponsorship does not seem to have any detrimental effects for the blogger, as readers’ attitudes towards the blogger were also not affected by their activated persuasion knowledge and even got more positive as they found the blogger to be more credible. This finding is interesting, as it contradicts the results of Colliander and Erlandsson (2015) who did find negative effects for the blogger. However, Colliander and Erlandsson (2015) studied third-party disclosures and this might have a negative effect for the blogger, because their readers feel betrayed when they find out that the blogger has not been honest with them.
Indeed, the results indicate that readers’ blogger attitudes benefit from a proactive sponsorship disclosure in the form of a disclaimer text and Colliander (2012) has already demonstrated that disclaimers do not tend to decrease readers’ attitudes towards the blog. Hence, sponsored blog posts can still increase blogger attitudes among readers, as long as the sponsorship is proactively disclosed by the blogger, to prevent the readers from feeling betrayed. Ultimately, the audience might even be understanding of the fact that this is how bloggers make their money, they just do not like to be deceived.
Similarly, this might also be the reason that this study did not find any effect of perceived appropriateness of the sponsorship, contrary to earlier research on persuasion knowledge and perceived appropriateness of covert marketing (Wei et al., 2008). There might be another ceiling effect at play, where consumers are simply used to sponsored content on blogs by now and they no longer deem it inappropriate, seeing as it is proactively disclosed. Thus, readers’ blogger attitudes will not be affected by their perceived appropriateness of a third-party sponsorship, as long as they are generally fairly accustomed to these kinds of marketing practices, as was the case with the participants in this study.
Managerial Implications
The results of this study have some important implications for practitioners, especially in tourism marketing. Blogs are an excellent marketing channel for tourism products due to
their word-of-mouth like character, which fits well with the experience-based nature of most tourism products (Park & Nicolau, 2015; Xiang & Gretzel, 2010). The present study shows that sponsoring bloggers can indeed have very positive publicity effects for brands and especially for well-known brands that most consumers are familiar with.
The newly installed regulations for mandatory sponsorship disclosures in sponsored blog posts are meant to warn consumers of sponsored content on blogs. However, the findings of this study indicate that most consumers are already very much aware of sponsored content online and that disclosures do not have any additional effect in making consumers aware of these practices. This is good news for marketing managers, who can keep using blogs as an effective part of their media mix. Again, blog marketing might even lead to increased purchase intentions and can thus be a very valuable tool, especially for well-known brands.
From the blogger’s perspective, the new regulations actually seem to be a positive development. Both the results of this study, as well as those of Colliander (2012) reveal that using disclaimers as a proactive sponsorship disclosure can actually annul the negative effects that third-party sponsorship disclosures might have for the blogger. Consumers have grown accustomed to sponsored content in blogs, just like they have grown accustomed to
advertising in many other aspects of their lives. Nonetheless, this type of marketing is still effective, as consumers seem to believe that bloggers remain independent regardless of their commercial interests. Future research will have to explore this correspondence bias further, in order to outline its determinants and boundary conditions and find out how this mechanism really works.
Limitations and Directions for Future Research
As with all research, this study has some limitations that need to be taken into account when interpreting the results. Both the blog and the blogger used in the manipulation of this study were fictional, which could have affected the results. Even though a content analysis of
existing blogs and an existing brand were used to improve the external validity as much as possible, bloggers usually have a unique and friendship-like relationship with their audience that is almost impossible to replicate in a manipulation. Specifically, bloggers and their readers usually enjoy a degree of parasocial interaction, which is an essential part of why marketing via blogs is so successful (Colliander & Erlandsson, 2015; Hwang & Jeong, 2016).
Moreover, this study employed a snowball convenience sample, which is never ideal. It is thus possible that the participants in this experiment were not typical blog readers and that the results were influenced by personal characteristics that are not usually at play among a blogger’s audience. To tackle these problems, future research into sponsorship disclosures in blogs would benefit from a field-experiment setting. This could be similar to the study of Colliander and Erlandsson (2015), who cooperated with an existing fashion blog to study the effects of third-party disclosures of blog sponsorship and demonstrated that parasocial interaction is indeed an important part of the blogger landscape.
Something else that deserves more attention in future research is that bloggers are increasingly stretching their influence to channels other than just their websites. Nowadays, most bloggers might be seen as “social influencers”, who stay in touch with their audience through every possible social media channel (Freberg, Graham, McGaughey, & Freberg, 2011; Harder, 2015). Consequently, their influence is now even more widespread than before. Bloggers are using Instagram, YouTube and even Snapchat to stay in touch with their readers and to spread sponsored content. However, all of these channels come with their own unique features and the different effects of the sponsored content across these various channels are yet to be explored.
Finally, there is still very limited insight into the way that marketing via blogs is really perceived by the audience and why it tends to have differential effects across situations and contexts. Obviously, there is a need for some clearer outline of the numerous boundary
