What is High Quality Content,
and how does it Determine the Success of Content
Marketing?
by
Giedre Remezaite
11107030
Master’s Thesis
Graduate School of Communication
Master’s Programme Communication Science
Professor Ed Peelen
Abstract
The growing intertest in content marketing has required marketing practitioners to rethink the role content plays in attracting consumers and facilitating the buying process. This study is an initial and important step in exploring content quality in the context of content marketing. Using an experimental design, this research examines which aspects of content quality are most important to content users, and explores how content quality affects consumer action and evaluations of a brand. The results indicate that content quality can certainly influence the success of content marketing programs. It was found that content which varies in quality produces significantly different outcomes. Using these insights, this study explores the
opportunities and risks associated with various types of content quality. Moreover, it provides a conceptual framework of content quality that can be used by any content manager to measure and improve their content quality in a valid way. The results are discussed in the light of the theoretical and practical implications for content marketing management.
1 Introduction
The recent growth of the social and mobile web has transformed methods of marketing communication, changing how relationships between companies and consumers are built (Rose & Pulizzi, 2011). Over the past few years, the approach to marketing communication has generally been slowly shifting from push to pull marketing (Chaffey & Smith, 2012). This shift has required companies to employ different communication techniques which focus on helping consumers rather than selling products. Pull marketing involves capturing the interest of consumers who are already looking for information, products or advice. One of the ways to attract audiences is to create valuable content for consumers (Holliman & Rowley, 2014).
Lately, there has been growing interest in content marketing, which is a strategic approach to managing content (Rose & Pulizzi, 2011). Content produced as part of a content marketing strategy can include blogs, e-newsletters, white papers, eBooks, case studies, webinars, mobile applications, etc., which can be produced in various formats, including text, images, audio, video or graphs (Lieb, 2011). Digital content marketing involves:
“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. 23).
The key objectives of content marketing include brand awareness or reinforcement, lead
generation, customer conversation, customer service, passionate subscribers, trust building, and brand positioning (Rose & Pulizzi, 2011). To achieve these goals, companies must develop
2
content which creates value and is of high quality (Johnston, 2016; Lieb, 2011; Rose & Pulizzi, 2011).
Content marketing specialists define high quality content as that which is relevant, unique, clear, useful, timely, and shareable, which has a human voice, connects with people on an emotional level, meets consumers’ immediate needs, and creates value for consumers (Johnston, 2016). Although some practitioners agree on certain aspects of content quality (Lieb, 2011; Rose & Pulizzi, 2011), there is no empirical support for this definition. Indeed, no prior research has yet investigated content quality in the context of content marketing. This knowledge is needed so that content quality can be measured, analyzed, and improved in a valid way. Without this knowledge, companies cannot truly determine the status of their content quality and monitor its improvement.
Numerous studies have investigated data and information quality in both offline and online environments (DeLone & McLean, 1992; Katerattanakul, & Siau, 1999; Maltz, 2000; Rieh, 2002; Wang & Strong, 1996). Although they have provided many insights about information quality, there is no evidence that these insights can be readily applied to content marketing practice. By nature, content is more than simply bare information or data. It is also an engaging form of communication which is addressed to a specific audience (Rose & Pulizzi, 2011). Effective content marketing programs put users at the center of content creation and evaluation, making the user the most important judge of content quality (Linn, 2015). While content quality may vary across audiences, the literature on data quality suggests common elements that are consistent indicators of quality (Wang & Strong, 1996). Using frameworks of data and
information quality, the current study aims to determine the aspects of content quality that are the most important for content users. The first research question is therefore presented below:
3
RQ1. Which aspects of content quality are the most important for content users?
Previous non-academic research has demonstrated that effective content marketing programs can produce the desired results and increase a company’s revenue (Heinz & Abaza, 2016). Given the growing popularity of content marketing, more companies are now producing content; however, not all of it is of a high quality. Companies which produce low quality content often risk damage to their reputation, the loss of consumer trust, a failure to establish
relationships and an inability to close sales (Liubarets, 2016). Even though the risks and opportunities may seem visible, no prior study has investigated how content quality affects the success of content marketing programs. Thus, the second aim of this study is to test the impact of content quality on consumer actions and evaluations of a brand. This knowledge can help companies to increase their consumer engagement, build strong relationships, save resources, and prevent damage to their brand image. The second research question is presented below: RQ2. How does content quality affect consumer action and evaluations of a brand?
As research on content marketing is its infancy, this study will be the first to reveal
indicators of content quality from the user’s perspective. It will develop a conceptual framework that can be used by any content manager to measure and improve content quality in a valid way. Moreover, it will build a theoretical basis for studying content quality in the context of content marketing. Above all, it will demonstrate how various types of content quality affect the success of content marketing programs, and how this knowledge can be applied in practice.
4
Theoretical Background Response Process
In content marketing, turning a prospect into a sales opportunity is a long and continuous journey (Rose & Pulizzi, 2011, Stenitzer, 2016). In order to develop an effective content marketing program, practitioners must understand how consumers respond to brand communication, and how that knowledge can be used to facilitate the consumer journey. Traditional response hierarchy models describe some of these processes and explain how advertising effects occur over time. A few of the best-known response hierarchy models include the AIDA Model (Belch & Belch, 2015), the Hierarchy of Effects Model (Lavidge & Steiner, 1961), the Innovation Adoption Model (Rogers, 2010), and the Information Processing Model (McGuire, 1978). They describe the series of steps potential consumers often take in moving from being unaware of a product to being ready to purchase it. All four models view the
response process as a movement of three consecutive stages: cognitive, affective, and behavioral (Belch & Belch, 2015). The cognitive stage represents the consumer’s knowledge of the brand; the affective stage the consumer’s feelings; and the behavior stage the consumer’s actions towards the brand.
Although response progression may not always operate according to this exact sequence, the models suggest that consumers’ thoughts and feelings about a brand impact upon their behavior, and that response processes can be influenced through communication programs. Given that content marketing is a form of communication, content marketing messages are expected to have an impact on brand evaluations and consumer action. Evidently, consumers cannot absorb all the branded messages they are exposed to, so they use certain criteria to select
5
information and evaluate it (Rieh, 2002). The following section explores these criteria and the impact they have on users’ selection behavior.
Information Selection and Evaluation
Online users often face situations where they must choose information or data among various alternatives, and judge its quality using their own criteria (Rieh, 2002). In traditional information retrieval, this issue is often discussed in the context of topical relevance, which relates to selecting information or data that matches the query. However, prior research has demonstrated that consumers use more diverse criteria for selecting information than mere topicality (DeLone & McLean, 1992; Wang & Strong, 1996). The essential criteria that consistently appears in the literature is information and data quality.
Information and data quality. The concepts of data and information are not identical, as data is a term referring to raw material, while information is data that has been interpreted and placed in context (Knight & Burn, 2005). Nevertheless, a thorough examination of the relevant studies (Eppler & Wittig, 2000; Knight & Burn, 2005) revealed that the literature relating to both data and information quality can provide useful insights, and therefore, these terms are used interchangeably in this study.
The notions of data and information quality are not new, and have been studied by numerous scholars over the years (Katerattanakul, & Siau, 1999; Maltz, 2000; Miller, 1996; Rieh, 2002; Wang & Strong, 1996). Many studies have emphasized the difference between objective and perceived quality (Holbrook & Corfman, 1985, Dodds & Monroe, 1985; Zeithaml, 1988). Objective quality refers to the technical superiority and excellence of information, while perceived quality refers to consumers’ assessment of the salient attributes of information (Dodds & Monroe, 1985). Previous studies have shown that consumers’ evaluations of information
6
quality are based not only on the technical superiority of that information, but also on each user’s perception of its relevance (Alexander & Tate; 1999; Miller, 1996; Rieh, 2002; Wang & Strong, 1996). They emphasize the importance of the user’s perspective, as it is the user who will ultimately judge information quality. Because users are at the center of content creation and evaluation (Linn, 2015), this study takes a consumer viewpoint to discover and understand the underlying aspects of content quality.
Several authors have suggested that information quality is relative, situational and subjective (Miller, 1996; Rieh, 2002; Tayi & Ballou, 1998). Information which is considered appropriate for one task or user may not be sufficient for another use (Tayi & Ballou, 1998). That means that the criteria used to evaluate information quality are also likely to be situational. In her qualitative study, Rieh (2002) found that users searching for information about medicine or computers were much more concerned with the trustworthiness and reliability of the
information than they were when looking for information about travel. Such differences in information evaluation could be explained using the Elaboration Likelihood Model (ELM), which suggests that individuals who are highly involved with the relevant issue assess
information by scrutinizing argument quality (Cacioppo & Petty, 1983). If the issue is important and relevant, then users are likely to be more concerned with the accuracy and believability of the information than, for example, its style of presentation. Although these scholars have emphasized the contextuality of information quality, they have not provided statistically
significant data to support their claims. Thus, the current study examines whether users evaluate content quality differently when performing different tasks.
To answer the first research question, which aims to identify the aspects of content quality that are most important for content users, the present research explores how users judge content
7
quality while performing various tasks1. Based on the regularity of the following tasks, this study explores how users judge content quality when: 1) searching for content about an important subject (e.g. health care, university, finance); 2) seeking entertainment; 3) staying up-to-date with issues; and 4) searching for content about products or services. The present investigation does not aim to reveal how specific quality attributes vary across different tasks; but rather, to provide evidence that content quality should be studied in the context of a known user task. Using the insights from each task, this study aims to determine which aspects of content quality are most important for content users.
Quality dimensions. Defining, assessing and modifying information quality requires an in-depth understanding of this concept. Information and data quality are often studied as multidimensional concepts (Knight & Burn, 2005). Some scholars have built the information quality construct by empirically developing dimensions from the perspective of users (Wang & Strong, 1996) or through literature reviews (DeLone & McLean, 1992). Others have focused on the dimensions that can be objectively assessed (Wand & Wang, 1996). When comparing these studies, the differentiating aspect is whether the viewpoint of the information user has been considered. An examination of information quality from the user’s perspective requires the inclusion of the subjective dimensions which represent the user’s perceptions of how valuable information is to them (Lee, Strong, Kahn & Wang, 2002). Given the identified importance of consumer’s evaluations of content quality, only those frameworks that include the user’s subjective evaluations will be reviewed. Table 1 presents an overview of the previously proposed data and information quality frameworks from the user’s perspective.
8 Table 1
9
Comparisons of information and data quality frameworks have revealed a considerable overlap. All the previous studies identified accuracy as an important quality dimension, which represents the extent to which data or information is correct and reliable (Wang & Strong, 1996). Except for the framework set out by Katerattanakul and Siau (1999), all the other frameworks include a time dimension, reflecting the extent to which information is current or up-to-date. Some studies refer to this dimension as timeliness or currency. Four of the five frameworks emphasized the importance of the context in which the information is used, and referred to this dimension as relevancy. Other synonyms, such as usefulness or target group orientation, were also used. Three of the five frameworks identified objectivity as an important quality criteria. While definitions and conceptualization may vary across studies, this dimension was mostly defined as unbiased representation (Wang & Strong, 1996). Other dimensions such as
consistency (i.e. systematic format and content of information), accessibility (i.e. rapid and easy
access to information), and completeness (i.e. an appropriate and sufficient amount of information) were also mentioned in some studies, but not in all.
Although some agreement is evident among scholars, it appears that the categorization of the dimensions varies across the prior studies. For the sake of the consistent categorization and terminology of quality dimensions, a framework by Wang and Strong (1996) is selected for application in this research, as it includes nearly all the common dimensions. This framework has been recognized and used by numerous previous scholars (e.g. Dadeke, 2000; Eppler & Wittig, 2000; Knight & Burn, 2005), and has been empirically tested in the contexts of
traditional information systems (Lee et al., 2002; Strong, Lee & Wang, 1997), and the internet (Klein, 2002). This study is the first to test it in the context of content marketing. The framework
10
by Wang and Strong includes four data quality (DQ) categories and fifteen dimensions (see Figure 1). Definitions of all the quality dimensions are presented in Appendix A.
Figure 1. A conceptual framework of data quality
Each of the categories in Wang and Strong’s framework are now discussed in turn.
Intrinsic data quality. Intrinsic DQ includes four data quality dimensions: accuracy, believability, objectivity and reputation. The authors found that accuracy and objectivity alone are insufficient for data to be regarded as high quality. Users consider believability and
reputation to be essential parts of intrinsic DQ.
Contextual data quality. Contextual DQ highlights that data must be considered in the appropriate context. It represents the fitness of data for completing the user’s task, and includes five dimensions: value-added, relevancy, timeliness, completeness, and an appropriate amount of data.
11
Representational data quality. Representational DQ includes the dimensions relevant to the format of the data (concise and consistent representation), and the meaning of the data (interpretability and ease of understanding).
Accessibility data quality. Accessibility DQ emphasizes the importance of the ease of accessing the data, and includes two dimensions: accessibility and access security. The authors note that quite often, scholars presume good data accessibly.
Content Quality
Practitioners agree that creating high quality content is a challenging task (Johnston, 2016; Rose & Pulizzi, 2011). In ideal situations, companies have large teams, structures and processes in place, and full resources at their disposal to create high quality content (De Clerck, n.d.). However, in practice, business rarely go from low maturity to a completely integrated content marketing approach. Quite often, they lack writing expertise or full knowledge of their target audience, which mean that their content quality can vary. This study aims to discover the risks and opportunities associated with content that varies in terms of quality. With this knowledge, content managers can plan content marketing programs accordingly, avoid risks, and achieve their desired outcomes. In order to answer the second research question, which explores how
content quality affects consumer action and evaluations of a brand, this study applies the
framework by Wang and Strong and examines four different types of content which vary in their quality.
High overall content quality. Here, content is of high intrinsic, contextual,
representational and accessibility quality. In practice, this type of content is most likely to be produced when a company has complete resources (e.g. experienced writers, and comprehensive information systems in place), high expertise in the field, and substantial knowledge about their
12
consumers. Content managers suggest that high quality content can help brands to break through the clutter, connect with their consumers in meaningful ways, and convert consumer engagement into profitable action (Rose & Pulizzi, 2011). From a practitioners’ perspective, it is anticipated that content which is of high overall quality will produce positive brand evaluations and induce favorable consumer action.
Low overall content quality. Here, content is of low intrinsic, contextual, representational and accessibility quality. In practice, this type of content is likely to be produced when a
company has incomplete resources (e.g. inexperienced writers, no informational systems in place), low expertise in the field, and little or no knowledge about their consumers. This situation can often be faced by start-up businesses. Content managers argue that companies which produce low quality content risk incurring damage to their reputation, loss of consumer trust, and failure to establish relationships (Liubarets, 2016). It is therefore assumed that content which is of low overall quality will result in negative brand evaluations and no action by
consumers.
Low contextual content quality. Here, content is of high intrinsic, representational and accessibility quality, but low contextual quality. This type of content is likely to be produced when a company has complete resources and high expertise in the field, but lacks knowledge of their consumers. This situation can also occur when a company is unable to effectively reach its target audience. Previous research has found that receiving irrelevant content often frustrates users and even motivates them to surf away from the website (Junrain, 2013). Given the negative reactions to irrelevant content, it is expected that content of low contextual value will not
13
is unclear whether content which is mostly of a high quality can produce negative brand evaluations when it is sent to inappropriate audiences.
Low representational content quality. Here, content is of high intrinsic, contextual and accessibility quality, yet low in representational quality. This type of content is likely to be produced when a company has high expertise in the field, substantial knowledge about their consumers, but incomplete resources (e.g. inexperienced writers). This situation can often occur when a company is in an early stage of their content marketing program. Previous research has found that content which is poorly written is perceived to be unprofessional, and to consequently decrease consumers’ trust (Hesse, 2013). Because trust affects consumers’ purchase decisions both directly and indirectly (Kim, Ferrin & Rao, 2009), it is assumed that content of low
representational value will produce low brand evaluations and reduce the possibility of positive brand-related action.
Given that research in the field of content marketing is still in its infancy, and that
practitioners’ insights are not sufficiently complete to predict the effects for each type of content or to foresee the differences among these types, no hypotheses are formulated here; instead, an exploratory approach is taken.
Methods Research Design and Participants
To answer the research questions, an experiment with a between-group design was conducted, which was divided into two independent parts. The first part tested how content quality affects consumer actions and evaluations of a brand, and consisted of four experimental conditions: 1) high overall content quality; 2) low overall content quality; 3) low contextual
14
content quality; and 4) low representational content quality. The second part of the experiment
explored which aspects of content quality are most important for content users, and consisted of four conditions based on common user tasks: 1) looking for content about an important subject (e.g. university, finance); 2) seeking entertainment; 3) staying up-to-date with issues; and 4) searching for content about products or services.
In total, 250 participants took part in the survey. In the first part of the study, the sample was reduced by applying the following criteria: time spent to read the stimulus material, participants’ attention to questions, and their involvement with the given issue, which will be explained later. Eventually, 120 respondents (56% female, 44% male) were included in the sample (Age: M=23.68, SD=3.53). Different sample reduction criteria were used in the second part of the study. Only those respondents who did not answer the questions relating to the second part were removed, leaving a total of 196 respondents.
Stimulus Material
The stimulus material was designed to test how various types of content quality affect users’ actions and evaluations of a brand. The material consisted of four fictitious blogposts, a format which was selected due to being one of the most common and recognizable formats of content marketing programs (Lieb, 2011). The experiment began with the introduction of a fictitious brand, called Amuse Holland, which was described as a non-profit student organization for both international and Dutch students in the Netherlands. The research participants were advised that they would receive a blog post written by Amuse Holland, which would inform them about student finance opportunities in the Netherlands. The blog post designs applied the data quality framework by Wang and Strong (1996). In consideration of the given task and context, and the circumstances in which the content was to be presented, the framework was
15
reduced. Table 2 demonstrates the quality dimensions that were selected or excluded when developing the stimulus material.
Table 2
Conceptual Framework Reduction
The blogposts for the experimental conditions, which are summarized in Table 3.1, were created by applying content attributes assigned to either high or low quality conditions (see Table 3.2). For instance, if the content was intended to be of low intrinsic quality, it did not contain any numbers or titles; or if content was of low contextual value, it was sent to
16
individuals who were not involved with the issue, and so on. The experimental material is included in Appendix B.
Table 3.1
Summary of Experimental Conditions
Table 3.2
Content Attributes in High and Low Content Quality Conditions
Dimensions
Content attributes
High content quality conditions Low content quality conditions Believability
Accuracy
Content refers to (governmental) institutions
Clear and concrete titles, numbers and facts
Content is free of grammar mistakes
Content does not refer to any (governmental) institutions
Vague, no concrete titles or numbers
17 Value - added
Relevancy
Content is distributed to individuals who: 1) are eligible for student finance in the Netherlands; 2) currently do not receive finance; or 3) are interested in finance opportunities for foreign students in the Netherlands
Content is made personal by using words such as “you”; “for you”;
Content is distributed to individuals who either: 1) are not eligible for student finance in the Netherlands; 2) currently receive finance; or 3) are not interested in finance opportunities for foreign students in the Netherlands
Content is written in third person. Use of words such as “a student”; “for them”; Ease of
understanding Concise presentation
Conversational, everyday language
Friendly tone of voice
Content order is easy to follow. It includes bullet points and headings, helping
readers to easily find key points Compact presentation, yet to the point
Institutional and complex language; or sensational and too informal language Authoritative tone of voice; or jargon-filled language
Content order is hard to follow. It does not include bullet points, nor headings, making it hard to find key points Lengthy presentation; content contains unnecessary links and information
Pre-test
A pre-test with 33 participants was conducted to test whether the stimulus material had the intended effects. In it, the participants were exposed to one of the four blog posts, and asked to evaluate the content quality on six Likert-type items, ranging from 1=“strongly disagree” to 7=“strongly agree”. They were asked about the extent to which they agreed that the content was: believable, accurate, valuable, relevant, easy to understand, and concise. One-way Analyses of Variance (ANOVA) revealed that the differences among most experimental conditions were non-significant. To intensify the differences, the stimulus material was adjusted. Then, a small-scale pre-test was conducted, which produced the desired results.
18 Procedures
The study participants were approached in various ways depending on their involvement with student finance opportunities in the Netherlands. Two different links to the online
experiment were created: the first link was distributed to participants who had an involvement with the issue, and the second to those who did not. The first group was approached with the assistance of several student organizations and educational institutions (e.g. International Student Network, Kastu, etc.) who were informed about the study and asked to invite appropriate
individuals to participate in the experiment. The second group was approached using snowball sampling, ensuring that the participants met the required criteria. If the respondents did not meet these criteria, they were later removed from the sample. For instance, if individuals were
involved with the issue but accidentally received the second link, they were excluded from the data analysis. The sample was reduced using the variables outlined in the section discussing covariates.
By opening the link, the respondents were redirected to the online experiment. The participants who received the first link were randomly assigned to either ‘high overall content quality’ or ‘low representational content quality’ conditions. Those who received the second link were randomly assigned to either ‘low overall content quality’ or ‘low contextual content
quality’ conditions. After exposure to the stimulus material, the respondents answered a series of questions related to the effectiveness of the stimulus materials and other measures (see the following section). Following this, in the second part of the experiment, the respondents were randomly assigned to one of the four conditions based on common user tasks: 1) looking for content about an important subject (e.g. university, finance); 2) seeking entertainment; 3) staying up-to-date with issues; and 4) searching for content about products or services. Here, the
19
participants were asked to think of a specific task and then to answer several open and closed questions. Finally, the participants were debriefed and thanked for their participation.
Measures
Brand evaluations. Brand evaluations were measured using two latent concepts common in marketing research: brand attitude, and brand trust (Belch & Belch, 2015). Both concepts were measured using seven-point Likert-type scales, ranging from 1=“strongly disagree” to 7=“strongly agree”. Brand attitude was measured using five items. The participants were asked about the extent to which they agreed with the given statements. The brand was: good,
trustworthy, honest, favorable, and of high quality (Van Noort & Willemsen, 2012). The items formed a reliable scale and were averaged to a single measure (M=5.00, SD=.90, α = .92). Brand trust was measured using three items: “I trust this brand”, “I can rely on this brand” and “This brand is safe” (Chaudhuri & Holbrook, 2001). The items formed a reliable scale, which was averaged to one measure (M=4.64, SD=1.10, α = .91)
Consumer actions. Consumer actions were also measured using two concepts: forwarding intention, and future engagement, which are common goals of content marketing programs (Rose & Pulizzi, 2011). The study measured consumer actions using the same scales as those set out above, with participants asked about the extent to which they agreed with the statements. Forwarding intention was measured using the following two items: ‘I would forward this blog post to my friends’; ‘I think this blog post is worth forwarding’ (Lin et al., 2006). The items were significantly correlated (r=.75, p<.001) and averaged to a single measure (M=4.47,
SD=1.60). Future engagement was measured using one item: ‘In the future, I would like to
20
Covariates. Based on Rieh’s (2002) model of the information quality judgement, user knowledge and issue involvement were measured to assess their impact on brand evaluations and user action, using the same scales as those set out above. The participants were asked about the extent to which they agreed with the given statements. User knowledge was measured using one item: “Before reading this blog post, I had sufficient knowledge about student finance opportunities for foreign students in the Netherlands” (M=3.69, SD=2.02). Issue involvement was measured using three items, as the participants were asked whether student finance for foreign students in the Netherlands was “important”, “relevant”, and “valuable” for them
(Zaichkowsky, 1994). These items formed a reliable scale and were averaged to a single measure (M=5.30, SD=1.64, α=.90).
Content quality measures. All the questions related to the second part of the experiment were formulated using Wang and Strong’s framework. After being instructed to think of a specific task, the respondents were asked how important it was for them that the content was: believable, accurate, objective, of high reputation, valuable, relevant, timely, complete, easy to understand, consistent, concise, and easy to find. All these items were measured on Likert-type scales (1= “not important at all”; 5=“extremely important”).
Results Manipulation Check
Checks like those discussed in the pre-test section above were conducted to test the effectiveness of the stimulus materials. One-way Analyses of Variance (ANOVA) were carried out with experimental conditions as an independent variable and believability, accuracy,
21
The results are summarized in Table 4.1. The analyses are presented for each quality dimension in Appendix C.
A manipulation check revealed that all the experimental conditions had the intended effects, except for two content quality dimensions in one of the experimental conditions. ‘Ease of understanding’ and ‘concise presentation’ were evaluated differently than intended in the ‘low representational content quality’ condition. Because four of the six quality dimensions were manipulated successfully, this experimental condition was included in subsequent analyses for exploratory reasons.
Table 4.1
22 Confound Checks
Confound checks, based on Pearson correlation coefficients, revealed that issue
involvement was not related to brand attitude, brand trust, or future engagement in any of the experimental conditions. The correlation between issue involvement and forwarding intention was also non-significant in two of the four experimental conditions. Furthermore, no significant correlation was found between user knowledge and brand attitude in any of the experimental conditions. User knowledge was also unrelated to brand trust, future engagement and forwarding intention in three of the four experimental conditions.
For exploratory reasons, covariate checks were carried out again for all participants who had answered the attention questions correctly and spent sufficient time reading the stimulus material (N=167). It is assumed that the covariates were not detected in the previous analyses due to a small sample size, so a larger sample was used to test the relation between the
covariates and the dependent variables, including all experimental conditions. The results reveal that both covariates were significantly related to all dependent variables when a larger sample was used (see Table 5). Although these variables should be considered as covariates in future research, the current study excluded them from the subsequent analyses as they were non-significant in the selected sample.
23 Table 5
A Summary of Correlation Coefficients of the Covariates and Dependent Variables
Effect of Content Quality on Brand Evaluations and Consumer Action
The first part of the study explored how content quality affects brand evaluations and consumer action. As was noted earlier, brand evaluations were assessed with reference to two concepts: brand attitude, and brand trust. Consumer actions were also measured via two
concepts: forwarding intention, and future engagement. Before testing the effects, the outliers of each dependent variable were removed and the variance among the groups was tested. The results of the Levene’s F-test showed that it can be assumed that the variances among the groups were equal in all dependent variables, except for future engagement. Appropriate analyses were selected for each dependent variable.
Given that there were no significant covariates, and the assumptions of normal distribution and equal variance were satisfied for all the variables except for future engagement, a one-way ANOVA was used to test the effect of content quality on brand attitude, brand trust and
forwarding intention. Tests were carried out with various types of content quality (i.e.
experimental conditions) as independent variables on each occasion, and brand attitude, brand trust, and forwarding intention as dependent variables. The results are summarized in Tables 6.1 and 6.2.
24
Brand attitude. Based on brand attitude, the outcome of a one-way ANOVA revealed significant differences among the experimental conditions: F(3,114)=6.93, p=<.001, partial
η2
=.15. To discover the differences among the groups, a post-hoc with Bonferroni correction was used. The results showed that brand attitude in experimental condition no.2 was significantly lower than in condition no.1 (Mdifference= -.80, p<.05), no.3 (Mdifference= -.87, p<.05), and no.4
(Mdifference=-.72, p<.05), meaning that content of low overall quality generated the lowest brand
attitude among all types of content quality (see Table 6.1). In a departure from expectations, content which was of low overall quality did not harm users’ attitudes toward the brand;
evaluations were neutral. No other significant differences between experimental conditions were established, meaning that content of high overall, low contextual and low representational quality produced similar attitudes toward the brand.
Brand trust. Based on brand trust, the results of a one-way ANOVA revealed significant differences among the experimental conditions F(3,112)=3.76, p=<.05, partial η2=.09. A post-hoc revealed that brand trust in experimental condition no.2 was significantly lower than in condition no.1 (Mdifference= -.67, p<.05), and no.4 (Mdifference=-.74, p<.05). Although brand trust in
experimental condition no.2 was also lower than in condition no.3, the difference was non-significant (Mdifference= -.64, p=.096). Considering the relatively large standard deviation in
condition no.3 and the small sample size, it is assumed that the results might have been different if a bigger sample had been used. Based on the results, it is concluded that content of low overall quality results in lower brand evaluation than content of high overall and low representational quality. No other significant differences were observed, meaning that content of high overall, low contextual, and low representational quality induce similar levels of brand trust.
25
Forwarding intention. Based on forwarding intention, the results of a one-way ANOVA revealed a significant difference among the experimental conditions F(3,116)=14.60, p=<.001, partial η2=.27. A post-hoc showed that future engagement in experimental condition no.2 was significantly lower than in condition no.1 (Mdifference = -1.94, p<.001), no.3 (Mdifference = -1.17,
p<.05), and no.4 (Mdifference = -1.97, p<.001). A low mean in condition no.2 (M=3.21) indicates
that content which is of low overall quality is unlikely to be forwarded. Although content of low contextual quality was also found to be less likely to be shared than high overall quality
(Mdifference = -.78, p=.219) and low representational quality content (Mdifference = -.80, p<.198),
these differences were non-significant. Based on the relatively large standard deviation in condition no.3, it is assumed that the results might have been different if larger sample had been used. Content of high overall and low representational content quality is similarly likely to be forwarded.
Future engagement. To test the effect of content quality on future engagement, a Welch
F-test was conducted, as this type of test is suitable when the homogeneity of variance
assumption is broken (Field, 2013). The results showed that there was a significant difference among the experimental conditions F(3,59.88)=24.93, p=<.001, ω2=.38. To explore the
differences among the groups, a post-hoc with Games-Howell correction was used, because it is robust to unequal variance among the groups (Field, 2013). It revealed that future engagement in experimental condition no.2 was significantly lower than in condition no.1 (Mdifference= -2.68,
p<.001) and no.4 (Mdifference= -2.44, p<.001). Similar results were found for condition no.3,
where future engagement was significantly lower than for condition no.1 (Mdifference= -1.57,
p<.05) and no.4 (Mdifference= -1.33, p<.05). This means that content which is irrelevant and does
26
were found, meaning that content which is of high overall and low representational quality is similarly likely to increase future engagement.
The results revealed no significant differences between content of high overall and low representational quality for all the dependent variables. This is most likely to be because the manipulation was only partially successful for low representational content quality. Thus, no valid inferences about content which is of low representational quality can be made.
Table 6.1
Means and Standard Deviations for Brand Attitude, Brand Trust, Forwarding Intention, and Future Engagement
27 Table 6.2
Results of a one-way ANOVA for Brand Attitude, Brand Trust, and Forwarding Intention
Predicting Brand Evaluation and Consumer Action
The following analyses were used to explore whether brand evaluation and consumer action can be predicted based on the experimental conditions, and if so, what the differences among the conditions are. By knowing the predictive value of various types of content quality on the desired outcomes, content managers can allocate their resources accordingly. To explore the predictive values, multiple regression analyses were carried out with dummy-coded
experimental conditions as independent variables and brand attitude, brand trust, forwarding intention, and future engagement as the dependent variables. Full details of the analyses and results are provided in Appendix D. The significant predictive values are discussed below.
Brand evaluation. The analyses revealed that compared to content of low overall quality, content of high overall quality can increase brand attitude by 0.80 and brand trust by 0.67 units; content of low contextual quality can enhance brand attitude by 0.87, and brand trust by 0.64 units; and content of low representational quality can improve brand attitude by 0.72, and brand trust by 0.74 units. Given that content of high overall, low contextual, and low representational
28
quality respectively have relatively similar predictive values, resources focused on intrinsic and representational content quality should be sufficient to enhance consumer brand evaluation.
Consumer action. Content of high overall quality can increase forwarding intention by 1.94, and future engagement by 2.68 units, compared to content of low overall quality. When compared to content of low contextual quality, it can increase forwarding intention by 0.78, and future engagement by 1.57 units. The results indicate that the differences among the types of content are relatively large, meaning that content managers who invest in content of high overall quality can significantly increase consumer action.
Content Quality from the User’s Perspective
The second part of the study explored which aspects of content quality were most important for content users. To investigate this, four user tasks were analyzed: 1) looking for content about an important subject; 2) seeking entertainment; 3) staying up-to-date with issues; and 4) searching for content about products or services. It was assumed that users use different criteria to judge content quality when performing different tasks. A Welch F-test was conducted to test this, with user tasks as an independent variable, and believability, accuracy, objectivity, reputation, value-added, relevancy, timeliness, completeness, ease of understanding,
representational consistency, concise presentation, and accessibility as the dependent variables. The results showed that nine of the twelve quality dimensions significantly varied across the tasks (see Appendix E), thus providing evidence that the importance of content quality
dimensions varies across the tasks to some extent. The only dimensions which did not vary were linked to the representation of content.
29
A post-hoc with Games-Howell correction was used to identify the differences among the tasks (see Table 7). A few major observations can be made about the results. First, people judge content quality differently when seeking entertainment in comparison to when they are
performing other tasks. The main difference here is that all the quality dimensions received lower mean scores than in any other task, meaning that content consumed for entertainment might be judged less critically or using different criteria. Second, some quality dimensions were found to be more important than others across all tasks. For instance, the believability, accuracy and reputation of content, which represent intrinsic content quality, were all important across all tasks, whereas objectivity was deemed less important. Relevancy, value-added, timeliness and completeness, which form parts of contextual content quality, were also important factors for each task. Representational content quality dimensions, namely, ease of understanding, representational consistency and concise presentation, were deemed less important across all tasks. The findings indicate that several common elements can be regarded as consistent indicators of content quality.
The answers to open questions revealed that users use even more diverse criteria when judging content quality. They apply criteria such as the aesthetics of the content or website, writing style, external links, references, tone of voice, search engine ranking, and the influence of other users (e.g. likes, number of followers, comments) when evaluating content quality (see Appendix F). The aforementioned quality aspects were noted by numerous participants,
providing evidence that content quality is evaluated using marginally different criteria than data quality.
30 Table 7
31
Conclusions and Discussion
The recent growth of content marketing programs has required marketing practitioners to rethink the role which content plays in attracting consumers and facilitating the buying process
(Holliman & Rowley, 2014). This study is an initial and important step in exploring content quality in the context of content marketing. Using an experimental research design, this research has examined the aspects of content quality which are most important for content users, and how content quality affects consumer action and evaluation of a brand.
Content Quality from the User’s Perspective
Numerous prior researchers have studied data and information quality (Katerattanakul, & Siau, 1999; Maltz, 2000; Miller, 1996; Rieh, 2002; Wang & Strong, 1996), and this study has extended their findings in several ways. The results of the present research show that, as with information quality, content quality is also relative and contextual. It was found that the importance of content quality aspects varies across user tasks. Content which is considered appropriate in one task is most likely to be insufficient for another use. These findings support those of Miller (1996), Rieh (2002), and Tayi and Ballou (1998), who emphasized the
contextuality of information quality. Given that users evaluate content quality differently when performing various tasks, it is concluded that content quality should be studied in a context in which the user’s task is known.
Furthermore, this study provides empirical support for data and information quality
frameworks which have previously been proposed (Alexander & Tate, 1999; Miller, 1996; Rieh, 2002; Wang & Strong, 1996) in the context of content marketing. By taking an empirical
32
content quality, where intrinsic and contextual quality were of the highest importance. Although data quality frameworks have received substantial support, they can only be applied in content marketing to a limited extent. Using a qualitative approach, it was revealed that consumers use highly diverse criteria to judge content quality. The results of this study have enabled a
conceptual framework of content quality to be built (see Figure 2). It was developed on the basis of Wang and Strong’s (1996) framework by identifying those data quality aspects which
received support in the context of content marketing, and by adding new dimensions which were empirically discovered (marked with a star). The additional dimensions are discussed in the following section.
Figure 2. A conceptual framework of content quality
It was found that users pay close attention to the sources used to create content, therefore, a new dimension, references, was added to the framework (see Appendix G for definitions of the new content quality dimension). Reliable references and links for further exploration indicate high intrinsic content quality, because these aspects are linked to the believability and reputation
33
of content or its source. Furthermore, it was found that users apply different criteria when evaluating the representation of content, emphasizing its artistic and emotional value. Although Katerattanakul and Siau (1999) recognized website attractiveness as an important quality indicator, it was not part of Wang and Strong’s framework. Thus, three content quality dimensions were added here: aesthetics, tone of voice, and writing style. Due to the available web tools, content quality is often judged via the consideration of external factors, such as the
influence of other users and the content’s search engine ranking. Older data and information
quality frameworks do not recognize external factors as indicators of quality (Alexander & Tate, 1999; Miller, 1996; Rieh, 2002; Wang & Strong, 1996). The inclusion of external factors
indicates that content is consumed in an environment which can also be influenced by other users. Although external factors are outside the content manager’s control, their impact on overall content quality should be considered in the post-evaluation.
The conceptual framework of content quality illustrates the aspects of content quality that are most important for consumers. This framework is valid for use by any content manager to create, analyze, measure and improve their content. It recognizes content quality attributes that can be assessed both objectively and subjectively. As suggested by Wang and Strong (1996), Miller (1996), and Rieh (2002), the true value of information is determined by the end-user. Thus, content managers should put the consumer at the center of content creation and evaluation. Effect of Content Quality on Consumer Action and Evaluation of a Brand
In order to explore the effects of content quality on consumer action and evaluation of a brand, this study examined four different types of content which varied in terms of quality. Based on the response hierarchy models (Belch & Belch, 2015), it was expected that content quality would influence consumer thoughts and brand-related behavior. The findings indicate
34
that content quality has a significant effect on consumer action and evaluation of a brand. Various types of content quality produced different effects, which are discussed in the following section.
High overall content quality. Although some evidence exists that content marketing programs can produce the desired results (Heinz & Abaza, 2016), this study is the first to discover the role which content quality plays in enhancing brand evaluation and building consumer relationships. It found that content of high intrinsic, contextual, representational and accessibility2 quality can, indeed, increase brand evaluation and encourage consumer action. Content which is of high overall quality produced better results than other types of content. Whether content managers should invest in content of high overall quality depends on their goals. If their aim is to increase brand attitude, brand trust, or the viral spread of content, it is not necessary to allocate their full resources to create content of high overall quality, as content of low contextual quality produces slightly lower, but similar results. Nevertheless, when the goal is to facilitate consumer engagement, creating content of high overall quality is necessary, as this was the only type of content found to enhance consumer engagement.
Low overall content quality. Based on the practitioner’s perspective, it was assumed that content which is of low overall quality would harm a brand’s image and fail to induce brand-related action (Liubarets, 2016). The present research found that content of low intrinsic, contextual and representational quality produced lower brand evaluations than other types of content, and did not induce any favorable brand-related action. Interestingly, content of low overall quality did not harm the brand’s image. This is an essential finding for content managers, as it provides them with the opportunity to explore what high quality content means for their
35
users without damaging their brand’s image. The results, however, should be applied with the consideration in mind that this study did not test the effects of repetitive exposure to poor quality content. It is probable that different effects would occur if users were repeatedly disappointed. Moreover, content which is of low quality can also lower the search engine rank, which can consequently also affect the accessibility of the content (Patel, 2015). Thus, exploring what high quality content means for specific audience has its own risks and opportunities.
Low contextual content quality. It was expected that content of low contextual quality would not produce any positive brand-related action (Junrain, 2013). The results showed that content of high intrinsic, representational, and accessibility quality but low contextual value can increase brand evaluation and slightly enhance forwarding intention, but it is not sufficient to increase user engagement. The practical application of these results is limited to some extent. The participants in the study were exposed to irrelevant content as part of the experiment. In reality, the likelihood of users reading irrelevant content is much lower. The important finding here is that no risks are associated with this type of content. If content is selected and processed, it can increase evaluations and be organically spread among consumers, but it is not adequate in itself to build consumer relationships.
Low representational content quality. In this experiment, content of high intrinsic, contextual and accessibility quality but low representational value was only partially
manipulated. Although the stimulus material was considerably longer than in other experimental conditions and it used complex language, the participants evaluated the representational content quality positively. This may have happened for several reasons. Based on the ELM, users who are involved with the issue are more concerned about argument quality than with the
36
involved, aspects such as complex language and a long text could have been insufficient to lower their evaluations. Moreover, it was discovered that users do not consider representational content quality of the highest importance (see Table 7). Because the content was of high intrinsic and contextual value, the users might have been biased about its representational value. Due to the partially unsuccessful manipulation, no valid inferences about this type of content could be made.
Content Quality Strategy
Based on the findings of this study, it is concluded that no ultimate and best strategy for creating high quality content exists; however, it can be designed by focusing on a few key
points. Because content quality is highly contextual, content managers should ask themselves the following key questions:
What is the purpose of your content? Is it for entertaining users, or informing and educating them? The importance of content quality attributes varies across user tasks, and the aspects which comprise high quality content also depend on the task. For instance, when users look for content about an important subject, they deem content which is believable, accurate, of high reputation, relevant, value-adding, concise, understandable, and accessible to be of high quality. However, when users seek entertainment, they are also concerned with the aesthetics of the content and writing style. Once the purpose of the content is clear, content managers using the conceptual framework of content quality (Figure 2) can explore and discover what high quality content means for their consumers. Luckily, they can do that without damaging their brand’s image.
What are the strategic goals of your content? Evidently, creating high quality content requires considerable time and resources. Whether content managers should invest in high
37
quality content or not depends on their goals. Only content which is of high overall quality can achieve the following: enhance brand evaluation, increase the viral spread of content and build consumer relationships. However, if the goal is to create positive brand evaluation, then content which is believable, accurate, concise, easy to understand, but not necessarily relevant or value-adding is actually sufficient. Depending on the stage of the consumer’s journey, content
managers may wish to invest full or limited resources in content of a certain quality. The predictive values of various types of content quality on desired outcomes (see Appendix D) can be also considered in making informed decisions.
Who are your users? Only content managers in possession of sufficient knowledge about their users, and who are able to reach them effectively, can provide content which is of high contextual value. In order to do so, a practitioner must ensure that the content is relevant, timely, complete, and creates value for the consumer. To gain insights about the target audience, content managers could use primary or secondary research, and experiment with various types of content and monitor their impact. Using these insights, practitioners can improve their content quality accordingly. After all, crafting content and discovering the most suitable approach to key users is crucial to the success of content marketing.
Limitations and Future Research
The results of this research should be considered in the light of several inherent limitations. The possibility of confounding effects was identified in this study. Although the rationale for excluding covariates was justified, future research should consider the confounding effects of issue involvement and user knowledge. Furthermore, the sample size used in this study was relatively small, which limits the generalization of the findings. Because the stimulus material was subjectively evaluated, participants’ opinions regarding content quality varied, causing large
38
standard deviations for the dependent variables. The small sample size and these large standard deviations may have influenced some of the results as well as the partially unsuccessful
manipulation of one of the four experimental conditions. The effects of low representational quality content could not be explored in the study.
Given that the current research has built a theoretical basis for studying content quality in the context of content marketing, future research could explore the effects of other types of content quality. For instance, it could seek to reveal the risks and opportunities associated with content which is of high contextual and representational quality, but of low intrinsic value. Because the present study found that content quality affects brand evaluation and consumer action, future research could focus on the factors which influence content evaluation. For instance, it could focus on the users and test how their personal features, such as involvement, affect content evaluations, or on external factors, such as influence of other users, or the search engine ranking, and test their impact on user content evaluation. Content quality is a complex and highly under researched area. A holistic approach should therefore be taken to better understand the aspects that influence content evaluations.
39 References
Alexander, J. E. & Tate, M. A. (1999) Web wisdom: how to evaluate and create information
quality on the web. Mahwah, NJ: Erlbaum.
Belch, G. E. & Belch, M. A. (2015). Advertising and promotion: An integrated marketing
communications perspective. Singapore: McGraw-Hill Education.
Cacioppo, J. T., & Petty, R. E. (1983). Central and peripheral routes to persuasion: Application to advertising. Advertising and consumer psychology, 1, pp. 3-23.
Chaffey, D., & Smith, P. R., (2012). eMarketing eXcellence: Planning and optimizing your
digital marketing. 3rd ed., Oxford: Butterworth Heinemann
Chaudhuri, A., & Holbrook, M. B. (2001). The chain of effects from brand trust and brand affect to brand performance: the role of brand loyalty. Journal of marketing,65(2), pp. 81-93.
De Clerck, J.P (n.d.). Content marketing: organization and management. I-scoop. Retrieved from
http://www.i-scoop.eu/content-marketing/content-marketing-organization-and-management/
DeLone, W. H., & McLean, E. R. (1992). Information systems success: The quest for the dependent variable. Information systems research, 3(1), pp. 60-95.
Dodds, W. B., & Monroe, K. B. (1985). The effect of brand and price information on subjective product evaluations. NA-Advances in Consumer Research. 12, pp. 85-90.
Eppler, M. J., & Wittig, D. (2000, October). Conceptualizing Information Quality: A Review of Information Quality Frameworks from the Last Ten Years. In IQ, pp. 83-96.
40
Heinz, M. & Abaza, H. (2016, November 17). Uncover the B2B Marketing Trends That Matter [Webinar]. Retrieved from http://www.slideshare.net/Uberflip/uncover-the-b2b-marketing-trends-that-matter
Hesse, L. (2013, November 12). Poor grammar on websites scares 59% away. Real business. Retrieved from http://realbusiness.co.uk/tech-and-innovation/2013/11/12/poor-grammar-on-websites-scares-59-away/
Holliman, G., & Rowley, J. (2014). Business to business digital content marketing: marketers’ perceptions of best practice. Journal of research in interactive marketing, 8(4), pp. 269-293.
Holbrook, M. B., & Corfman, K. P. (1985). Quality and value in the consumption experience: Phaedrus rides again. Perceived quality,31(2), pp. 31-57.
Johnston, M. R., (2016, February 4). 16 Common-Sense Reminders for Uncommonly Good Content. Content Marketing Institute. Retrieved from
http://contentmarketinginstitute.com/2016/02/uncommonly-good-content/
Junrain (2013). Online Consumers Fed Up with Irrelevant Content on Favorite Websites,
According to Janrain Study. Retrieved from
http://www.janrain.com/about/newsroom/press-releases/online-consumers-fed-up-with-irrelevant-content-on-favorite-websites-according-to-janrain-study/
Katerattanakul, P., & Siau, K. (1999). Measuring information quality of web sites: development of an instrument. In Proceedings of the 20th international conference on Information Systems (pp. 279-285). Association for Information Systems.
41
Kim, D. J., Ferrin, D. L., & Rao, H. R. (2009). Trust and satisfaction, two stepping stones for successful e-commerce relationships: A longitudinal exploration. Information systems research,20(2), pp. 237-257.
Klein, B. (2002). When do users detect information quality problems on the world wide web?. AMCIS 2002 Proceedings, 152, pp. 1101-1103.
Knight, S. A., & Burn, J. M. (2005). Developing a framework for assessing information quality on the World Wide Web. Informing Science: International Journal of an Emerging Transdiscipline,8(5), pp.159-172.
Lavidge, R. J., & Steiner, G. A. (1961). A model for predictive measurements of advertising effectiveness. Journal of Marketing. 25, pp. 59–62.
Lee, Y. W., Strong, D. M., Kahn, B. K., & Wang, R. Y. (2002). AIMQ: a methodology for information quality assessment. Information & management,40(2), pp. 133-14.
Lieb, R. (2011). Content marketing: Think like a publisher-How to use content to market online
and in social media. Indiana: Que Publishing.
Linn, M. (2015, May 7). Put Users at the Center of Your Content Strategy: A Look at This American Life. Content Marketing Institute. Retrieved from
http://contentmarketinginstitute.com/2015/05/content-strategy-user-centered/ Liubarets, T. (2016, August, 25). How Low-Quality Copywriting Services Kill Business.
Writtent. Retrieved from
42
Maltz, E. (2000). Is all communication created equal? An investigation into the effects of communication mode on perceived information quality. Journal of Product Innovation
Management, 17(2), pp. 110-127.
McGuire, W. J. (1978). An information-processing model of advertising
effectiveness. Behavioral and management science in marketing, pp. 156-180.
Miller, H. (1996). The multiple dimensions of information quality. Information Systems Management,13(2), pp. 79-82.
Patel, N. (2015, April 9). How Google Judges Quality and What You Should Do About It.
Content Marketing Institute. Retrieved from
http://contentmarketinginstitute.com/2015/04/google-judges-quality/
Rieh, S. Y. (2002). Judgment of information quality and cognitive authority in the Web. Journal of the American Society for Information Science and Technology,53(2), pp. 145-161.
Rogers, E. M. (2010). Diffusion of innovations. New York: Simon and Schuster.
Rose, R., & Pulizzi, J. (2011). Managing Content Marketing: The Real-World Guide for
Creating Passionate Subscribers to Your Brand. Cleveland, OH: CMI Books.
Stenitzer, G. (2016, May 9). Sharpen Your Content Marketing: 4 Ways to Model the Buyer’s Journey. Content Marketing Institute. Retrieved from
http://contentmarketinginstitute.com/2016/05/model-buyers-journey/
Strong, D. M., Lee, Y. W., & Wang, R. Y. (1997). Data quality in context. Communications of the ACM,40(5), pp. 103-110.
43
Tayi, G. K., & Ballou, D. P. (1998). Examining data quality. Communications of the ACM,41(2), pp. 54-57.
Van Noort, G., & Willemsen, L. M. (2012). Online damage control: The effects of proactive versus reactive webcare interventions in consumer-generated and brand-generated platforms. Journal of Interactive Marketing,26(3), pp. 131-140.
Wand, Y., & Wang, R. Y. (1996). Anchoring data quality dimensions in ontological foundations. Communications of the ACM,39(11), pp. 86-95.
Wang, R. Y., & Strong, D. M. (1996). Beyond accuracy: What data quality means to data consumers. Journal of management information systems,12(4), pp. 5-33.
Zaichkowsky, J. L. (1994). The personal involvement inventory: Reduction, revision, and application to advertising. Journal of advertising,23(4), pp. 59-70.
Zeithaml, V. A. (1988). Consumer perceptions of price, quality, and value: a means-end model and synthesis of evidence. The Journal of marketing, pp. 2-22.
44 Appendix A
Definitions of data quality dimensions (Wang and Strong, 1996, pp. 31-32) Intrinsic DQ
Believability - The extent to which data are accepted or regarded as true, real, and credible Accuracy - The extent to which data are correct, reliable, and certified free of error
Objectivity - The extent to which data are unbiased (unprejudiced) and impartial
Reputation - The extent to which data are trusted or highly regarded in terms of their source or content
Contextual DQ
Value-added - The extent to which data are beneficial and provide advantages from their use Relevancy - The extent to which data are applicable and helpful for the task at hand
Timeliness - The extent to which the age of the data is appropriate for the task at hand
Completeness - The extent to which data are of sufficient breadth, depth, and scope for the task at hand
Appropriate amount of data - The extent to which the quantity or volume of available data is appropriate
Representational DQ
Interpretability - The extent to which data are in appropriate language and units and the data definitions are clear
Ease of understanding - The extent to which data are clear and easily comprehended
Representational consistency - The extent to which data are always presented in the same format and are compatible with previous data
Concise - The extent to which data are compactly represented without being overwhelming Accessibility DQ
Accessibility - The extent to which data are available or easily and quickly retrievable Access security -The extent to which access to data can be restricted and hence kept secure
