Examining the Performance of Brand-Extended Thematic-Content: The Divergent Impact of Avid-and Skim-Reader Groups

.


Highlights:
 Brand-extended thematic-content creates mutual benefits for readers and the brand  Skim reading on social media provides only limited and more fragile brand response  Reading frequency is central in affecting brand response for avid engagers  Enjoyable content may support avid reading over functional content

Introduction
Online media use trends, such as short periods of engagement by users and simultaneous usage of multiple forms of media (Newman & Levy, 2014), have resulted in a lack of cognitive focus on ads and have affected the way marketers and advertisers pursue consumers (Malthouse, Haenlein, Skiera, Wege, & Zhang, 2013).To succeed in delivering their brand messages in this era, brands have adopted a new mindset in the creation of content (Malthouse et al., 2013;Pulizzi, 2012), making the attractive content a priority in digital marketing actions (Accenture, 2016).Digital content marketing can be defined generally as "creating, distributing and sharing relevant, compelling and timely content to engage customers at the appropriate point in their buying consideration processes, such that it encourages them to convert to a business building outcome" (Holliman & Rowley, 2014, p. 285).These practices are often characterized by regular content creation aimed at longterm relationship building (Holliman & Rowley, 2014).
As a result of this development, users of online media are now encountering increasing quantities of brand-created digital content covering broader themes and topics rather than focusing on content about the brands.Gao and Feng (2016) have defined content, which has a broader product category-level focus instead of a more limited brand focus as brand-extended content.With respect to the context of this study, we elaborate this definition and categorize brand-extended thematic-content produced by content marketers as thematically-bound, nonbrand focused, and frequently-produced messages with a purpose of repeatedly driving consumers toward the brand's online sphere of influence.The rationale for differentiating brand-extended thematic-content from pure brand-extended content is that brand-extended thematic-content does not limit merely to product category-level, broadening the focus for example to lifestyles or societal issues.The thematic focus of brand-extended content makes it possible for users also to expect content from specific thematic areas, supporting their further intentions to engage with content that they find personally relevant.
Academic research related to content marketing is still scarce (Holliman & Rowley, 2014;Järvinen & Taiminen, 2016).The importance of considering brand-extended thematiccontent in digital content marketing rises especially due to the fact that prior literature has mostly been interested in users who interact with brands online because of their personal relationship with brands.In these conditions, voluntary social media interaction between users and brands is argued to have important implications for branding (e.g., Gao & Feng, 2016;Kim & Drumwright, 2016) and in building customer-brand relationships (e.g., Hudson et al., 2016;Malthouse et al., 2013;Habibi, Laroche, & Richard, 2014).Similarly, the literature often assumes that users engage with online brand communities and their members because they are already involved with a brand and want to strengthen their relationship with that brand (e.g.Habibi et al., 2014).Hence, more research is needed to explore the impact of brand-extended thematic-content for readers.
To that end, we contribute to the evolving discussion of content marketing by empirically testing the proposed behavioral and psychological impact of brand-extended thematic-content on readers and their brand response.To do this, we take into account elements of media user profiling like frequency of usage, platform, and content preferences (see Brandtzaeg, 2010), which we consider as being important for understanding engaged readers of brand-created thematic-content.In addition, as all readers cannot be assumed to be equally interested in the brand (Habibi et al., 2014), the impact of brand-extended thematiccontent for readers may differ based on individual reader's brand connection.Without a strong existing brand relationship, brand-extended thematic-content can be a key driver for readers to develop an interest in the brand.Specifically, when readers' brand involvement remains low, but involvement and engagement to brand's content is higher, readers are more likely to pay attention also to the actual brand (Wirtz et al., 2013).Hence, we study readers who are not existing customers of the brand but who still have varying levels of engagement with brand-extended thematic-content.
This study highlights the role of more "avid readers" who read brand-extended thematic-content on brand website-based blogs, compared to "skim readers" who are satisfied to read a more concise version of this content on the popular social networking site Facebook.
Specifically, we argue that the impact of brand-extended thematic-content on positive brand response will be stronger for avid readers.As content marketing is often characterized by continuously offering new content (Holliman and Rowley, 2014), we explore whether reading frequency of brand-extended thematic-content impacts readers' brand responses beyond the previously-explored effects, such as the affect transfer.
The article proceeds as follows.We first explain the logic behind brands deciding to focus on creating brand-extended thematic-content.Then, we discuss the formation of a positive brand attitude through attitude towards the brand-extended thematic-content, the reading frequency of this content, and brand familiarity.Third, to empirically test our research model, we use data from two brand-extended thematic-content reader groups (avid and skim) who are not existing customers of a brand.Finally, we discuss our results and generate relevant suggestions for future research.
2 Theoretical Background and Hypotheses

Logic of Brand-Extended Thematic-Content in Attracting and Engaging Readers
Content marketing has become a widely-used digital marketing tactic due to its potential to create positive influence on business (Holliman & Rowley, 2014;Järvinen & Taiminen, 2016;Pulizzi, 2012).Content marketing objectives seem to be in line with general digital marketing objectives, such as creating awareness, changing brand attitudes, and building brand associations (Järvinen, Töllinen, Karjaluoto, & Jayawardhena, 2012;Yan, 2011).
Along with those objectives, marketing practitioners have stated brand trust and more directly sales-focused objectives such as increased website traffic and lead generation important (Holliman & Rowley, 2014).Many content marketing critics state that there is nothing new in the idea of content marketing; for example, customer magazines implementing similar ideas have existed for a long time.The crucial difference, however, may not lie in the execution of the content itself.Instead, the rationale for differentiating content marketing from more traditional forms is that, today, this content delivery underlines the role of digital channels and emphasizes the aspect of pulling readers to brand's online channels with interesting content (Holliman & Rowley, 2014;Järvinen & Taiminen, 2016)."Pulling" refers to attracting readers who already are involved with a particular issue or a brand and who, therefore, seek content related to that issue or brand (Holliman & Rowley, 2014).
Consequently, thematic involvement, in terms of personal relevance and interest in a topic (see Zaichkowsky, 1985) can be seen as a prerequisite for readers to seek, consume, and become engaged with online content.
Beyond the thematic involvement, the value of content marketing for readers is argued to arise from reading experiences, mostly relating to the provided information or entertainment (Holliman & Rowley, 2014).From a theoretical stance, uses and gratification (U&G) theory is a relevant approach for explaining media use behavior.Specifically, U&G theory suggests that gratification sought from media use defines readers' further usage of that media (e.g., Katz et al., 1973Katz et al., , 1974).Indeed, if readers do not find themselves motivated to use a medium, they will stop using it (Joines, Scherer, & Scheufele, 1999).More evidence is found in information systems literature, where Bhatterjee (2001) found satisfaction and perceived usefulness as the primary drivers for continuous use of an information system.
However, whereas satisfaction encourages readers to return to read more content, in the long run it is the more specific experiences that define future content reading (Mersey, Malthouse, & Calder, 2012).To ensure continuous reading behavior, brands' content must support the creation of these experiences.
The U&G approach has successfully been adopted into marketing (e.g., Jahn & Kunz, 2012;Muntinga et al., 2011) as well as into information systems studies (e.g., Gao & Feng, 2016).In the field of marketing, Jahn and Kunz (2012) identified hedonic and functional content as the biggest reasons for users to engage to brand fan pages.Similarly, Shi, Chen, and Chow (2016) found that information value, followed by entertainment value, has the strongest effect on continued interaction intention with social media brand pages.In other online media studies, these motivations are usually called "perceived enjoyment" (Calder et al., 2009;Heinonen, 2011) and "entertainment" (Heinonen, 2011;Muntinga et al., 2011;Mersey et al., 2012).In addition, utilitarian motivations are often listed as "information" or "learning" (Brodie et al., 2013;Heinonen, 2011;Mersey, Malthouse, & Calder, 2012;Muntinga et al., 2011).In the same vein, information systems literature, specifically the unified theory of acceptance and use of technology (UTAUT) and UTAUT2 models (see Venkatesh, Thong, & Xu, 2012), highlight the role of utilitarian and hedonic value in influencing behavioral intentions to use a technology.Similarly, Gao and Feng (2016) identified information-seeking and entertainment among the more socially-driven gratifications as major usage motivations for social media.While admitting that social gratifications may have a part in our study in defining media usage, here we only focus on information and enjoyment, which are especially relevant for readers to engage with content created by content marketers (Holliman & Rowley, 2014).Based on this rationalization and the substantial existing research on media gratifications driving continuous media usage in different media contexts, we hypothesize that: H1: Information gratification significantly drives frequent reading behavior.
H2: Enjoyment gratification significantly drives frequent reading behavior.
However, as brands cannot ultimately be assumed to create content without at least a partial self-interest, brand-extended thematic-content should also have some impact on readers' brand responses.For those who are not existing customers of a brand, content marketing may particularly affect brand awareness and attitudes (Malthouse et al., 2013).
While brand awareness basically relates to the ability to recognize and recall the brand, brand attitude refers to a favoring the brand and providing positive evaluations of the brand (Schmitt, 2012).Brand attitude is often used as a measure in understanding brand responses in other kinds of marketing communication tactics like advertising (MacKenzie et al., 1986;Muehling & Lazniak, 1992) or advergames (Van Reijmersdal et al., 2010).Related to online content strategies, brand-extended content like knowledge, news, and information from the product category was recently identified as affecting brand attitude better than a content strategy using only brand-related content (Gao & Feng, 2016).
To further explore the impact of reading of brand-extended thematic-content on brands, we next identify two important mechanisms that occupy the center of our conceptual research model.

The Impact of Reading Frequency on Brand Attitude
So far, academic content marketing literature has focused more on understanding content marketing from the perspective of marketers.The literature has also overlooked the aspect of a brand's continuous content creation and discusses content marketing at specific points in time, especially related to the buying process (see Järvinen & Taiminen, 2016).However, limited support for the perspective of continuous content creation exists.Holliman and Rowley (2014) state that content marketing should be viewed "as an ongoing cultural stance where the focus is on building an authentic relationship over the longer term, rather than conducting a series of short-term campaigns" (p.287).The importance of focusing on relationship is also supported by brand community literature.For example, Habibi et al.
(2014) highlight the consumer's ongoing relationship with the brand community enabling to create more supportive brand responses through continuous exposure to the brand and enhanced communication possibilities with the brand.Continuous usage behavior has also been noted as a central concept in information systems literature (see Shaikh & Karjaluoto, 2015).Consequently, it seems that, similar to information systems and for many online-based businesses' success (Bhatterjee, 2001), frequent reading behavior should have a central position in successful content marketing strategies.
Consumers' repetitive interaction with brand-related content has been studied particularly in social media context.Related to the evaluation of those interactions, frequency and the amount of interaction by the customer are seen central (Hudson et al., 2016;Shi, Chen, & Chow 2016).In addition, Jahn and Kunz (2012) found that the intensity, which they use as a concept similar to frequency, of a customers' brand page usage is more important in affecting their brand loyalty than their participatory interaction behaviors on the brand page.
Frequency of online interaction has also been found in a non-brand context affecting the preferences towards the other entity.Ledbetter and Mazer (2014) found that social media interactions can strengthen relationship outcomes like relationship tie strength, in terms of affecting other persons' cognitions, affection, and behaviors (i.e.interdependence).Similarly, continuing their work relating to social media user relationships, Ledbetter, Taylor, and Mazer (2016) found a significant relationship between the frequency of online media interaction and relational closeness towards another person.While the previous effects are discussed related to personal relationships, we argue that similar effect may exist related to brands, due to the connection enabled by the online platforms (Kaplan & Haenlein, 2010).
Thematic content has the ability to create common ground and reveal shared interests between a brand and content readers (Mangold & Faulds, 2009;Wirtz et al., 2013).
Supporting this, Hudson et al. (2016) found that frequent interaction with a brand on social media fosters emotional attachment and intimacy.Attachment translates into self-centered connection with a brand (Schmitt, 2012), at least partially reflecting the same aspects as relational closeness (Ledbetter, Taylor, & Mazer, 2016) and interdependence (Ledbetter & Mazer, 2014).Furthermore, albeit weaker than attachment, brand attitude is also a construct that represents consumers' connection to a brand (Schmitt, 2012).We argue that frequent interaction with brand-extended thematic-content should affect positively on the brand and, therefore, hypothesize the following: H3: Reading frequency of brand-extended thematic-content has a significant positive effect on brand attitude.

The Impact of Affect Transfer Effect for Brand Attitude
To be a relevant concept in a content marketing context, reading frequency of brand-extended thematic-content should affect readers' brand responses in the presence of previously highlighted factors.An affect transfer effect impacts consumers' brand responses when their motivations to evaluate the brand are low (Muehling & Lazniak, 1992).Affect transfer indicates a direct positive effect from attitude toward a brand-associated object (e.g. an ad) to attitude toward the brand (Machleit & Wilson, 1988;MacKenzie, Lutz, & Belch, 1986;Muehling & Lazniak, 1992).This indicates that consumers' cognitive processing is often focused on related contextual factors instead of directly processing the brand (MacKenzie, Lutz, & Belch, 1986).Brand-extended thematic-content may produce brand responses the same way.This is because processing the content may produce a positive attitude towards this content (Mersey et al., 2012) as readers may consider this content relevant and likable.
However, while the affect transfer effect has mainly been studied experimentally related to content, like a single advertisement (e.g., Muehling & Lazniak, 1992) or advergame (e.g., Van Reijmersdal et al., 2010), we argue that the affect transfer effect may remain over time, when brand relationship is primarily determined by the brand-extended thematic-content.In line with this, we hypothesize: H4: Information gratification has a significant positive effect on content attitude.
H5: Enjoyment gratification has a significant positive effect on content attitude.
H6: Attitude towards content has a significant positive effect on brand attitude.
Affect transfer is noted to occur especially when the brand is relatively unfamiliar (Machleit & Wilson, 1988).Brand familiarity is thus an important concept to take into account when assessing brand attitude formation.Brand familiarity can be defined as all the experiences with a brand, which includes all the cognitions (Alba & Hutchinson, 1987) that affect a consumer's brand knowledge (Campbell & Keller, 2003).Previous research has made evident that consumers' positive attitudes to brand-created content, like ads, increase their cognitive brand evaluations (MacKenzie et al., 1986).This kind of attention makes it also harder to influence consumers directly through affect transfer (Campbell & Keller, 2003;Machleit & Wilson, 1988).However, positive brand-related experiences generated by brandextended thematic-content should mediate the affect transfer effect in the absence of other equally important experiences with a brand.Hence, we hypothesize: H7: Brand familiarity mediates the effect of attitude towards content on brand attitude

Hypothesizing Divergent Brand Impacts for Skim Readers and Avid Readers
Today, users are engaging in numerous online platforms, which offer them different content consumption possibilities.For brands, different platforms also require special planning (Gao & Feng, 2016).Brand blogs are particularly noted to be important for content marketing (Pulizzi, 2012).On the other hand, social networking sites are argued as platforms, which leverage the reach of content marketing (Järvinen & Taiminen, 2016).It is common that this kind of content on social media is more concise, whereas blog posts are longer.In a similar vein, short version of the brand-extended thematic-content on a blog is often posted on other social media sites to introduce the blog content to the larger audience.Typically, a link and an invitation to read the entire content from the blog are attached to support website traffic.
Following these presumptions, we next identify two potential differences in mechanisms that lead to the formation of a positive brand attitude.
Compared to social network sites, such as Facebook, the blog environment generally differs due to the presence of just one content creator at the time, which enables greater focused attention to specific content as well as to the content creator.Readers mostly already know to expect specific content theme from the particular blog.Social network sites, on the other hand, are full of frequently changing content from a variety of sources (Gao & Feng, 2016) including the users themselves, their friends and brands followed.Naturally, social networking sites also contain numerous distinct content topics to focus on.During social networking site browsing, consumers are not likely to have interest in or even the resources to focus equally on all content, which creates a platform characterized by a constant rivalry for users' attention.Therefore, in a general manner, social media browsing behavior can be seen as skim reading.Skim reading relates to a behavior in which a reader chooses the most relevant and important content or content parts to focus on, instead of reading all the content available (e.g.Reader and Payne, 2007;Duggan and Payne, 2009).Skimming makes it possible to assess the available content and make a decision about whether more focused reading is preferred.
Presuming that a blog posts about brand-extended thematic-content offers more encompassing content than the concise version on a social networking site, the level of engagement between the group reading solely social networking site content and the group also reading content on a blog should differ.To rationalize this, skim reading literature (e.g.Reader and Payne, 2007;Duggan and Payne, 2009) argues that readers who find content more relevant to focus on are likely to invest their time and effort to read that content.
Similarly, in the context of this study, the behavioral difference could be explained by the different level of intentional engagement, defined as the "interest in devoting energy, effort, and time to a brand activity" (Andrine, Solem, & Pedersen, 2016, p. 448) which is then actualized as a behavioral manifestation (van Doorn et al. 2010).This way, skim readers who solely read the concise brand-extended thematic-content on a social networking site are not equally engaged compared to those travelling from a social networking site to the brand blog or who otherwise visit the blog because of the brand-extended thematic-content.Based on this, we call these more engaged readers as "avid readers".Highlighting differences between these two groups, we next propose the different impact of skim readers and avid readers for brand response.
First, as revealed from the brand relationship context, consumers who invest time and other resources may be more focused on long-term relationship maintenance (e.g.Sung & Choi, 2011).Similarly, for more engaged readers, reading frequency of brand-extended thematic-content may strengthen their positive brand-related responses.This may particularly occur because of the shared thematic interest between the brand and readers, which may make the brand stand as something more important for the reader (Wirtz et al., 2013), strengthening the connection with a brand (Hudson et al., 2016;Ledbetter et al., 2016).Based on this, we argue that more frequent reading of brand-created thematic-content may also impact brand attitude, especially in the case of the avid-reader group.Hence, we hypothesize that: H8: Reading frequency has a stronger effect on brand attitude for avid readers than skim readers.
Second, drawing from the discussion related to the hypothesis H8, in the case of skim readers the brand may remain more distant.This may be because for less engaged readers, content may not be as valuable as for more engaged readers.In this case, the brand may be seen less important (Wirtz et al., 2013) and brand attitude is more likely being shaped by perceptions related to positive attitude to brand-extended thematic-content.In the same vein, affect transfer should determine the emergence of skim readers' brand attitude formation more than within the avid-reader group.Therefore, we hypothesize: H9: Attitude towards the content has a stronger effect on brand attitude for skim readers compared to avid readers.Hudson et al. (2016) noted the importance of also taking into account the amount of consumer-brand interaction.Similarly, we argue that the results may vary with respect to whether brand-extended thematic-content is totally read only a few times or numerous times.
Particularly, the number of positive exposures to content may have an effect on more positive brand attitudes (Machleit & Wilson, 1988).Hence, in this study, we control the effect of total number of reads on brand attitude measured by whether the content has been read between one and ten times or more than ten times.Finally, we also control for the effect of age on brand attitude.Figure 1 illustrates the proposed research model.

Data Collection and Sample
To test the hypotheses, data was collected in cooperation with a brand that was relying heavily on content marketing and social media marketing to reach especially female readers.
The brand updated weekly its blog purely consisting of brand-extended thematic-content and also delivered most of its content via its Facebook page.In detail, the content was not brandrelated, but focused on providing inspiration, stories, and ideas related to house construction, design, and decoration.Compared to blog content, Facebook content was often presented in shorter versions or as "introductions," from which the consumer could then continue to read the content through the blog.Many times this conversion from Facebook to blog content was pursued through adding a suggestion, like "Read more from our blog!" on Facebook posts, with a link included to the full blog post.During the time the data was gathered, the brand had been operating in the home construction business about a year and a half, and it was not that familiar in its product category.Nonetheless, the brand had been "liked" by over 7,000 people on Facebook.
To reach thematically-engaged Facebook followers, as well as brand-website-based blog readers, the survey was administered through both the company blog and the company Facebook page.To verify that respondents were consuming brand-extended thematic-content, the question "How do you follow brand-created content related to decoration and design?" was included as a background question.The question was implemented using yes/no options related to reading brand-created content on Facebook and yes/no options related to reading brand-created content in website-based brand blog.
A total of 213 acceptable responses were received, resulting in an effective response rate of 45.5%.After removing the respondents who were not reading brand-extended thematic-content on either the website-based blog or on Facebook, as well as a small proportion who only read the blog (N=16), the final data set consisted of 189 respondents.
Most of the respondents were female (92%), which supports the goal of the creation of brandextended thematic-content, reaching the demographic segment that the brand was targeting.
As web technology use and motives are noted to vary between genders (Shi, Chen, & Chow, 2016), this data only reveals effects related to females.Most of the respondents were between 26-35 (39%) and 36-45 (32%) years of age.While 19% of the respondents had just drifted to the brand website, and 4% noticed the brand because of some other reasons, the great majority (90%) reported that they had noticed the brand because of its brand-extended thematic-content, although 15% of those who just drifted to the brand website also chose this answer, giving support to our contextual starting point.Furthermore, 74% of the respondents reported they had read brand-extended thematic-content more than 10 times.Platformspecific descriptions are found in Table 1.To ensure that common method bias did not seriously affect the results (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003), we enforced a procedural remedy at the data collection stage to ensure that respondents' identities remained confidential.Furthermore, item ambiguity was reduced, and the items were mixed in the questionnaire.We also ran a model with a common method factor to examine the role of common method variance in the model (Podsakoff et al., 2003).Whereas the average variance explained 0.74 of the indicator variance, the average method-based variance explained just 0.01.The results support the insignificant effect of the method variance, and so common method bias is unlikely to be of serious concern for this study.

Measurement
For measuring media gratifications, we adapted an information gratification measure (e.g."I get good tips from the content") from Mersey et al. (2012) and an enjoyment gratification measure ("I find following content enjoyable") from Calder et al. (2009).Both gratifications were measured with five indicators and were modified for the study context.Related to the enjoyment measure, we replaced one item with the indicator "I find content entertaining," as entertainment is one of the dimensions closely related to enjoyment.To measure thematic content consumption frequency, we modified the usage intensity measure used by Jahn and Kunz (2012), making it more content-oriented.Two modified items were: "I frequently follow the content" and "I regularly read the content."The former items were measured on a five-point Likert-type scale ranging from "totally disagree" to "totally agree".
In addition, the measure for brand attitude (i.e.negative/positive, unpleasant/pleasant, do not like/like, bad/good) was taken from MacKenzie et al. (1986).The measure for attitude towards content (i.e.bad/good, do not like/like, irritating/not irritating, uninteresting/ interesting) was adapted from Mitchell and Olson (1981).The measure for brand familiarity (i.e.familiar/unfamiliar, recognized/unrecognized, had heard/had not heard) was taken from Simonin and Ruth (1998).These items were all measured on a five-point scale.

Measurement Model Assessments
The hypotheses were tested with SmartPLS 3 (Ringle, Wende, & Becker, 2015).We found partial least-squares structural equation modeling (PLS-SEM) especially suitable for our testing, as it is practical when the data sample size is small and the tested model is explorative (Henseler et al., 2009).Both these conditions stand in this study.To reveal the differences between the avid-reader group who read brand-extended thematic-content both on Facebook and on brand website-based blog and the skim-reader group who solely read brand-extended thematic-content on Facebook, the two groups were analyzed separately.However, to make group comparison possible, the data sets had to meet the partial measurement invariance (Henseler et al., 2016).To verify the homogeneity of the data groups, the MICOM procedure (Henseler et al., 2016) with 3000 permutations was implemented.As the MICOM procedure detected no significant differences between the groups, the group comparison could be made.
However, to sustain high factor loadings (> .7) in both conditions, two indicators from the information gratification measure and two indicators from the enjoyment gratification measure were eliminated.The measurement model results are shown in Table 2.Moreover, internal reliability was established with high composite reliabilities ranging between 0.82 and 0.96.To assess the discriminant validity of our model, we used Fornell and Larcker's (1981) criterion.The results showed that the square root of AVEs were greater than cross-correlations between factors, indicating that the model had acceptable discriminant validity (see Table 3).

Structural Model Assessments and Hypothesis Testing
To evaluate the differing effects of skim readers versus avid readers in the research model, the moderation analysis was performed using partial least-squares multi-group analysis (PLS-MGA) (Henseler et al., 2009) with a bootstrapping of 2000 subsamples.In addition, the mediation effect of brand familiarity on the relationship between attitude towards content and brand attitude (H7) were assessed using bootstrapping sampling, as it is suggested to be more robust compared to Baron and Kenny's (1986) four-step approach (Preacher & Hayes, 2008).
Complete results of the hypotheses testing are collected in Table 4. Next, these results are discussed in detail.
To start, our results showed that information gratification was significantly related to reading frequency for skim readers (β = 0.358, p < .01)but not for avid readers (β = 0.168, ns).Thus, H1 is only confirmed within the skim-reader group.Similar results occurred related to the effect of information gratification on attitude toward content, providing only partial support for H4.Relatedly, information gratification was the primary influencer of attitude toward content (β = 0.433, p < .001)for skim readers, but the effect was not significant for avid readers (β = 0.193, ns).Enjoyment was found to have an effect on reading frequency within the skim-reader group (β = 0.272, p < .05),as well as avid-reader groups (β = 0.304, p < .01).Enjoyment also had a significant role in positive content attitude formation in both cases (β skim = 0.305, p < .001;β avid = 0.411, p < .001).Hence, both H2 and H5 are supported.While the PLS-MGA did not find significant differences between the groups related to these hypotheses, it is interesting to note that for skim readers, a search for information seems to be a more important motivation for reading, whereas for avid readers, information-seeking does not support more frequent reading behavior.Therefore, it seems that only hedonic gratification is able to sustain the avid readers' continuous brand-extended thematic-content reading in this context.Our data also strongly supports differing mechanisms for brand attitude formation between the skim-reader and avid-reader groups.PLS-MGA revealed significant differences between the two groups within both hypothesized effects H8 (Δβ = 0.253, p < .05)and H9 (Δβ = .340,p < .01),strongly supporting both these hypotheses.The differences were even found to be so strong that H3, related to the frequency effect, was rejected within the skim-reader group (β = .064ns), and H6, related to affect transfer, was rejected within the avid-reader group (β = 0.141 ns).To illustrate the differences between models, significant group-specific results are found in Figure 2 and Figure 3. Next, these differences are examined more closely.
Within the avid-reader group, we found reading frequency to be a significant predictor of positive brand attitude (β = 0.318, p < .01).This result indicates that, in the case of avid readers, reading frequency of brand-extended thematic content means that the brand represents something positive to the readers.Furthermore, our results reveal brand familiarity as the primary predictor of positive brand attitude (β = 0.407, p < .001),which also partially mediates (VAF = 37%) the effect from attitude towards content to brand attitude for avid readers.This supports the increased attention given to the brand as well as the manner by which the positive attention based on brand-extended thematic-content affects brand attitude through some kind of brand-relevant experiences.Together with these results, and the insignificant result of the affect transfer effect, it seems that, within the avid-reader group, affect transfer is not any more important in defining the formation of brand attitude; the connection to a brand behind the brand-extended thematic-content matters more.Finally, the results reveal that younger avid readers have better brand attitudes compared to older avid readers.This result also significantly differed from the skim-reader group, in which age did not have any effect on brand attitude.Following similar logic, in the case of skim readers, who are more less engaged to brandextended thematic content, reading frequency had no effect on brand attitude (β = 0.064 ns).
Instead, we found that attitude towards the content was a primary reason for positive brand attitude within skim-reader group (β = 0.483, p < .001).These results support the existence of strong affect transfer from content to brand.Within the skim-reader group, brand familiarity was not found to be the most important aspect (β = 0.289, p < .001),as skim readers more likely give only limited attention and importance to the brand.This is also supported by the considerably high effect (β = 0.456, p < .01)from attitude towards content on brand familiarity.Similarly, as the results revealed barely a partial mediation effect of brand familiarity between attitude towards the content and brand attitude (VAF = 22%), it can be said that in the case of skim readers, brand-extended thematic-content builds on the affect transfer effect in influencing brand attitude.Finally, we were able to explain 33% of the brand attitude variance in the avid-reader group and 56% of the variance in the skim-reader group.
To make the results more interpretable, we added an f 2 value that explained the relative power of the constructs in the model, which could be divided into small (>0.02), medium (>0.15), and large (>0.35) (Hair et al., 2014, pp.176-178).We found that, within the skimreader group, content attitude (f 2 = 0.381, high) and brand familiarity (f 2 = 0.153, medium) had a considerable effect size, affecting brand attitude.In the avid-reader group, brand familiarity was found to be the only concept having a medium effect size (f 2 = 0.241, medium), whereas reading frequency was established close to the limit of medium effect size (f 2 = 0.132, small) and therefore can be considered an important concept.As a conclusion, the effect sizes verify that the two different mechanisms are contextually relevant.

Discussion
In this study we examined the role of brand-extended thematic-content in content marketing both for the reader and the brand.We defined brand-extended thematic-content from content marketers' perspective as thematically-bound, non-brand focused, and frequently-produced messages with a purpose of repeatedly driving readers to brand's online sphere of influence.
Reflecting the reviewed literature, we argued that reading of this content differs based on consumers' level of "interest in devoting energy, effort, and time" (Andrine et al., 2016, p. 448), which is actualized in this study as differing behaviors of skim readers or avid readers towards the brand-extended thematic-content.Specifically, we suggested that in this study, blog reading was characterized by a deeper level of interest and effort compared to social networking site readers.
In the light of this study, brand-extended thematic-content can be seen as mutually beneficial for both the reader and the brand.By creating thematically-relevant and interesting content, brands can offer enjoyment and information for readers on a continuous basis.For brands, brand-extended thematic-content creation offers possibilities for positive brand visibility and enables the building of relationships with consumers.From a theoretical perspective, our results indicate differing mechanisms for positive brand attitude formation between the two thematically-engaged content reader groups.
The findings of this study support earlier literature (Barreda et al., 2015;Gao & Feng, 2016) regarding the ways that social media can be used to make users more brand-aware and also to affect positive perceptions of a brand.However, when readers are only engaging with more concise version of the brand-extended thematic-content on a social networking site, their brand attitude is largely defined by the affect transfer from content attitude.In the same vein, for skim readers more frequent interaction with the brand-extended thematic-content was not found to affect brand attitude.This indicates that for skim readers, frequent interaction does not make them feel more connected to the brand, and the brand may remain distant to the consumer.
Our study indicates that the brand attitude of avid readers is formed because of their frequent reading of brand-extended thematic content and their familiarity with the brand.In addition, the findings indicate that only enjoyment gratification is strong enough to influence this reading frequency.Based on these findings, reading frequency supports readers connecting with a brand.Specifically, the brand may be appreciated more as a content creator, and, in this way, make the brand more important to avid readers.This supports the arguments by Wirtz et al., (2013).This finding also supports previous literature related to the effects of interaction frequency on the perceived connection with the object of interaction (Hudson et al., 2016;Ledbetter et al., 2016), revealing these effects in the context of brandextended thematic-content.Similarly, avid readers' increased connection to the brand behind the content offers an explanation for the insignificant effect of content attitude on brand attitude.The results related to reading frequency offer evidence for the importance of ongoing brand-extended thematic-content creation by the brand.This supports Holliman and Rowley's (2014) view of seeing content marketing as building a long-term relationship with consumers.

Limitations and Recommendations for Future Research
The main limitation of this study is its small sample size (N skim = 76; N avid =113) and the use of data from just one brand.It is typical in the implementation of content marketing that a short version of blog content is "introduced" in social media to attract readers to view the entire blog; this study provides evidence in the case of a skim-reader group and an avidreader group only.Another data-related limitation concerns gender.Gender specification is

Figure 2 .
Figure 2. Significant structural model results for avid readers.

Figure 3 .
Figure 3. Significant structural model results for skim readers.

Table 1
Respondent Profiles

Table 2 .
Measurement Model Results

Table 3 .
Composite Reliability, AVE, Correlations and Square Root of AVE on the Diagonal Note.n.a.= not applicable, 1 Correlation coefficient (Due to measuring CF with two items, CR was not calculated)

Table 4 .
Structural Estimates and Group Comparison (PLS-MGA)Notes.N skim skim reader group, N avid avid reader group, β standardized coefficient, f 2 the relative impact a construct has in producing the R 2 value of the endogenous construct, |Δβ| the absolute value of the difference in coefficients between groups, * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed test).