MEMBERS WITH DIFFERENT ROLES WITHIN A DECISION
SETTING DIFFER IN INFORMATION NEEDS AT DIFFERENT
POINTS IN THE PURCHASE FUNNEL, BUT DO THEY?
NIENKE HULZEBOS
MSc Marketing Management
Groningen, January 12
th2020
Members with different roles within a decision setting differ in information
needs at different points in the purchase funnel, but do they?
Master Thesis Marketing Management January 12th, 2020
Jeaniena Berdien Nienke Hulzebos Student number: S3456307 Gorechtkade 146B 9713 CL Groningen 06-18682195 E-Mail: j.b.hulzebos@student.rug.nl Internal supervisor: Dr. Martijn Keizer E-mail: m.keizer@rug.nl University of Groningen Faculty of Economics & Business
Department of Marketing
ABSTRACT
This research aimed to find out whether the decision involvement of different members within a decision setting (referred to as different roles) differed in their information needs at different points in the purchase funnel. Additionally, it was expected that loss aversion would moderate the relationship between the members with different roles and their information needs at different points in the purchase funnel. This research focused on people working in secondary educational, since the client of this research was a Dutch educational publisher.
In order to execute this research, data was collected by means of an online survey, which resulted in a sample of 99 participants. The results suggest that members with different decision involvement roles do not differ in their information needs at different points in the purchase funnel. Furthermore, loss aversion did not affect the information needs at different points in the purchase funnel for members with different involvement roles. This research may help marketers to develop their marketing strategy concerning segmentation and targeting.
PREFACE
You are about to read my master thesis about members with different decision roles within a company, and what influence this role has on information needs at different points in the purchase funnel. This thesis was written as the final product to complete the master Marketing Management at the University of Groningen and was commissioned by a Dutch educational publisher. They are developing its website and want to do this on the bases of the needs of its customers. Therefore, I investigated the needs of its customers and wrote this thesis. First of all, I would like to thank my supervisor Martijn Keizer, without his help I could not have written this thesis. Furthermore, I would like to thank all people who took the time to fill out my survey. At last, I would like to thank my family, best friends, and colleagues for their continuous support.
All that remains for me to say is, I hope you will enjoy reading my thesis and will learn something from it.
Nienke Hulzebos
INTRODUCTION
With the rise of the internet, websites have become one of the most important public communication platforms for most, if not all, organizations (Garett, Zhang & Young, 2016). A companies’ website has become an unparalleled platform for people to explore information and acquire knowledge. Thus, a company’s website now is one of the most important touchpoints in the customer journey. Because of this, the way a website is designed plays a critical role in engaging users (Flavián, Guinalíu & Gurrea, 2006).
Furthermore, several studies have shown that website design has a big effect on purchase intention. Ranganathan and Ganapathy (2002) have empirically established that a well-designed website positively affects purchase intention. Moreover, Visineascu, et al. (2015) show that well-designed websites have a positive effect on the immersion a consumer feels, which increases the likelihood that the user stays through conversion. Additionally, Thongpani & Ashraf (2011), state that a companies’ website should provide an online environment which is comfortable and user friendly to the customer, and which features to provide easy access to the information he is looking for. This they state, is in line with information search and risk perception theories.
Thus, a well-designed website can increase purchase intention and can play a critical role in engaging users. For companies, this statement means their website is an important asset in making profit.
from awareness, to consideration, to decision, and at last to post-purchase. So, for companies it is valuable to know which role one has within the buying center of a company and in which stage a customer is located, so they can give them the information they need in this particular stage. In this research, attention will be particularly on the individual differences concerning decision involvement and information needs at different points in the purchase funnel.
One sort of company which heavily relies on its website is an educational publisher. Before, schools received a catalogue with all information about schoolbooks/teaching methods the publisher offered for the current schoolyear. Nowadays, the website of the publishers is where the school/teacher can find all information about the teaching methods, and what’s new. For publishers this means, their website plays an enormous role in selling their offerings.
Schools are the main customers of educational publishers. The norm for schools is to use a teaching method for 4-6 years, after this they can ‘switch’ to a new method. Here they have the choice to stay with the same publisher or they could go for another publisher. For publishers this means they have to stay in the top of the minds of their customers through these years, if they want their customers to stay with them at the ‘switch’ point. One way of doing this is by providing relevant information, and services on their website. Furthermore, since most schools use a teaching method for 4-6 years, this method can be referred to as their status quo. From research we know that most people like to stay with the status quo, rather than changing a situation. One explanation for status-quo bias is loss aversion. We expect that loss aversion in some way influences schools when choosing a new teaching method.
Since the website is one of the most important marketing tools for educational publishers and for companies in general, it is useful for companies to understand the
preferences of their customers concerning the website. Therefore, the goal of this research is to provide the marketing department insights in the different customer segments they can target and what the information needs across different points in the purchase funnel of these customers are. Furthermore, we are interested to see whether loss aversion influences information needs. This knowledge can help the marketing team to design their website and with this eventually increase the purchase intention of their customers. In addressing this problem, the following research question are defined:
Which different information needs across the purchase funnel stages can be defined across the different roles within the buying center?
(1) What is the effect the role one has in a decision setting on information needs at different points in the purchase funnel?
(2) Is this effect moderated by loss aversion?
Relevance and contribution
Many researchers have focused on how companies can best design their website, to increase purchase intention. Furthermore, there is a lot of knowledge on getting to know your customer, to serve them in the best possible way, and on how a customer moves through the purchase funnel. This study contributes to this examining different roles in a decision setting, and whether the information needs at different points in the purchase funnel will differ across these different decision involvement roles. To the best of our knowledge, there is no prior research similar to the construct of the current research, and thus fill a unique gap in literature due to the fact that customers will be segmented on their decision involvement within a decision setting in a company, and information needs at different points in the purchase funnel will be estimated. Ultimately, the outcome of this research helps identifying, defining and discussing the differences in online information needs across different purchase stages for people with different roles. When differences are explored, specific recommendations for these roles can help companies. For example, in targeting customers, and optimizing websites.
LITERATURE REVIEW
Customer segmentationMarket segmentation is central to marketing and a key decision area for organizations in all sectors (Weinstein, 2006). It originally stems from economic pricing theory, which suggests that maximum profits are achieved when pricing levels discriminate between segments (Wind, 1978).
Customer segmentation is grouping together customers with similar product
preferences and buying behaviors. This segmentation helps companies to efficiently allocate their recourse on relatively homogeneous customer segments (Smith, 1956). Furthermore, according to Hunt & Arnett (2004) segmentation is one of the constructs with which
companies can achieve competitive advantage. In their research they state that segmentation generates competitive advantage when (1) segments are identified, (2) these segments are targeted, and (3) customized marketing mixes per segments are defined. Furthermore,
successful customer segmentation helps the firm to determine the allocation of marketing mix resources across customers more precisely. This helps to customize firms’ strategies per customer group. Which has valuable implications, not only for firms, but for customers as well, since their needs are served better and the firm has more insights in their wishes (Kumar, 2018). Subsequently, in this research segmentation is used in order to understand what drives customers.
Moreover, the segmentation of customers can be based on all sorts of variables, regularly the bases for segmentation are geographic, demographic, psychographic, and behavioral variables (Kotler, 1997). Next to these variables, situational (e.g., purchase/use occasion) and customer preferences can be relevant variables in the segmentation of customers (Kotler & Armstrong, 1999).
Once the segments are identified, it is possible for firms to make predictions about the groups’ responses to various situations, to align their marketing strategies and types of policy, and to allow more creative and better-targeted policies to emerge (Teichert et al., 2008). The term segmentation is undisputed, but it includes different approaches (Wedel & Kamakura, 2000). These approaches can be split into two approaches. (1) A priori approach, here groups are selected in advance on the bases of known characteristics like socio-demographic
characteristics or frequency of buying and (2) post-hoc approaches, here empirical
attitudinal characteristics. Furthermore, more recently post-hoc segmentation based on latent variables is used. Here inferences about brand preferences and attribute importance are made (Wedel & Kamakura, 2000; Kamakura & Russell, 1989). The advantage of this last method (i.e., post-hoc segmentation) is that it is more linked to the actual marketplace (Allenby et al., 2002).
In this research customer segmentation will be based on the a-priori approach, this is done on the bases of different levels of involvement. We will do so by segmenting customers on de bases of their involvement in making decisions, which will be discussed later on in the literature review.
Purchase process
Marketeers have tried to understand the purchase behavior of their customers for decades. They tried to define and explain the consumer’s purchase process by using all kinds of models and applying different strategies for improving communication (Ghirvu, 2013). In almost all of these models, the first phase is “awareness”, which refers to consumers
becoming aware of a product/service/brand. The last phase is mostly referred to as “purchase”, “decision” or “action”. Furthermore, the phase between the awareness and decision phases, is mostly referred to as the “consideration” phase, in which consumers consider whether or not they should include a certain product/service/brand in their
consideration set (Hirschmeier & Schoder, 2016). The purchase funnel, proposed by Lewis (1903) is one of these marketing models, which focuses on the consumer and explains how the consumer moves through different stages, from awareness to purchase. Here, the funnel is a metaphor for the attrition that arises as consumers move through stages like awareness, interest, desire and then action (Johson, Lewis & Nubbmeyer, 2016). One of the paths
followed is the use of hierarchy of effect models. Here, the AIDA model is one of the earliest hierarchy of effects models. The stages in this model are Awareness, Interest, Desire, and Action. This model represents a process with multiple stages that describe the different stages that consumers get from being unaware of a brand, to awareness, gets particular preferences, purchases the product, and potentially eventually, develops loyalty for the brand (Ghiryu, 2013).
So, they need information about the advantages of the product, which will help them in their choice to buy. Although the AIDA model is one of the earliest models for showing the buying process, it is still very influential and is linked to more recent models, like the
purchase/marketing funnel (Lemon & Verhoef, 2016).
Moreover, Vázquez, et al. (2014) use a model which is strongly linked to the AIDA model. They use the purchase stages awareness, evaluation, purchase, and post purchase. Although there are similarities between this model and the AIDA model, Vázquez, etal. (2014) take into account the post-purchase stage as well. This is the phase where the customer actually tries the product after he purchased it, and this is the phase where retention/loyalty can arise.
The model which is used in this research is based on the model proposed by Vázquez, et al. (2014). However, in this research the terms for the different stages are somewhat different from the terms Vázquez, et al. (2014) use in their model. The terms which are used are awareness, consideration, purchase, and post-purchase. Although the terms are different, meaning of the terms is very similar to those of Vázquez, et al. (2014).
Although schools are organizations, and thus should not be treated as normal consumers, we expect the purchase funnel of schools to be similar to the one we propose since they as well move through the different stages from awareness to post-purchase.
FIGURE 1
Consumer purchase funnel adopted in this research
Firms make significant marketing investments in online and offline media, and channels to get customers to their website, mobile app, and stores, to effect conversions, or spur them into buying (Kannan, Reinartz & Verhoef, 2016). Because of this, marketers must understand the effectiveness of its investments across those different channels at the different points of the purchase funnel. Does the website convert new customers into making a purchase? Does the website help to increase loyalty with known customers? Understanding which parts of the purchase funnel are affected is relevant for managers in assessing the overall impact of their website and other channels.
Thus, for firms it is good to consider how their website effectiveness may change with customers being in different stages in the purchase funnel. It may very well be that customers in the awareness stage are looking for very different things than customers which are in the decision stage.
Importance of online presence and information search
Yannopoulos (2011), suggest that the internet is the most powerful tool for businesses. Therefore, marketeers should focus and plan strategies for their online presence. One of the most important assets in the online presence of a company is its website. Research suggests that any typical corporate, institutional or private website is built to present specific
information. There are different aspects of web contents. ISO 9241-151 defines content as “a set of content objects”. Subsequently content object is defined as “interactive or
non-interactive object containing information represented by text, video, sound, image, or other types of media (ISO, 2006). Thielsch & Hirschfeld (2019), state that content is of primary importance in the online world, and the subjective perceptions of content are known to influence a variety of user evaluations, and thus change behavioral outcomes and attitudes. Furthermore, how users perceive information on a website, is a very important factor for website success (Thielsch, Blotenberg & Jaron, 2014). Only if readers understand, believe and appreciate the presented information, they are willing to use the website (Lehto & Oinas-Kukkonen, 2011). Thus, content is one of the most important parts of a company’s website. In this research content will be referred to as information.
Moreover, the way information is presented on a website determines the success of the website. The information provided on a website should be easy to understand, and it should be easy to develop an understanding of the information found online (Chen et al., 2013). Furthermore, research shows that when companies put the right information on their website, this can have a positive effect on purchase intention (Visineascu, et al., 2015)
either be goal-directed search or it can be exploratory search. When a consumer performs goal directed information search the goal is to acquire relevant information about a specific product of product category he is already considering or planning to buy. Consumers search for information online in order to acquire relevant information which will help them to make a more optimal product choice (Wolfinbarger & Gilly, 2001; Moe, 2003). Because consumers are actively searching for information, goal-directed information search can be seen as an indicator of consumer interest (Hu et al., 2014), and thus can be interpreted as an indicator of conversion (Shim et al., 2001; Zigmond & Stipp, 2010). Also, Moe (2003) found that consumers with goal-directed information search know what they want and thus have the highest conversion rate. Therefore, consumers who engage in goal directed information search have a positive influence on purchase intention/probabilities (Agarwal, Hosanagar & Smith, 2011). However, instead of having a specific goal in mind while searching for information, consumers can also perform exploratory information search. Here the intention is to acquire information which will potentially be useful in the future and thus not directly aimed at making a purchase (Moe, 2003). Because consumers performing this type of information search have not the specific goal to purchase, the conversion rate is not very high for these consumers (Moe, 2003). However, Dickinger & Stangl, 2011 found that although the motivation of consumers who perform exploratory information search is not specific to make a purchase, it can also result in conversion. This can be explained by the fact that exploratory information search tends to be stimulus-driven, and with the right stimulus, consumers might be triggered to make a purchase.
For our research this distinction indicates that visitors of a website differ in the way they search for information and thus will be triggered by different sorts of information. Furthermore, the different forms of information search can be linked to the different stages of the purchase funnel, which were mentioned in the paragraph above. Meaning that, a visitor who is goal-directed, comes to a website with a specific goal in mind, and may be further in the purchase funnel, than a visitor who is exploratory directed.
Decision involvement
According to Johnston & Bonoma (1981), organizational buying behavior consists of all activities of organizational members as they relate to a buying situation as identification of needs, evaluation of alternatives, and choosing among those alternatives.
(Mallapragada et al., 2016). There are significant differences in the perceived influence of major participants in the buying process. In the business to business market, decisions are mostly made by several member within the firm who differ in their involvement in the decision (Fortin & Brent Ritchie, 1980). Robinson, Faris & Wind (1967), defined this as the buying center. This concept refers to all those members of a firm who become involved in the buying process for a particular product or service.
Furthermore, Webster & Wind (1972) defined the buying center as consisting of five roles: (1) users, those members of the organization who use the purchased products and services; (2) influencers, those who influence the decision process directly or indirectly or indirectly by providing information and criteria for evaluating alternative buying actions; (3) deciders, those with authority to choose among alternative buying actions; (4) buyers, those with formal responsibility and authority for contracting with suppliers; and (5) gatekeepers, those who control the flow of information into the buying center.
Among these different roles, each member has specific and unique interests and expectations, and thus may use different criteria when making decisions (Sheth, 1973). Moreover, members within a company, can have several roles in the buying center.
Typically, empirical studies link the roles of the members within the buying center to different functions within companies, such as buying, finance, operating management, production (Homburg & Rudolph, 2001). Therefore, we predict that the buying center of schools will in some way look like the one mentioned above.
Furthermore, role theory suggests the presence of potential differences between members who have different roles within the buying center. This theory indicates that people learn specific behaviors which are appropriate to the function they employ within the organization (Solomon et al., 1985). Furthermore, it can be assumed that differences exist in perception across the different roles, when making decisions (Tölner, Blutt & Holzmüllee, 2011). In addition, Moriarty & Spekman (1984), have demonstrated that different roles affect the scope and range of one’s information search behavior and needs.
Therefore, we expect members in these different roles to have different information needs, since their involvement, influence, and perception in the buying process is different. We expect decision makers to be more involved with a decision than users.
awareness, consideration, purchase, and post-purchase) of the purchase funnel. Moreover, we expect that segmentation of members in a decision situation can be based on these different roles.
H1: Members with different roles in the buying center, will have different information needs at different points in the purchase funnel.
Status quo
According to Savage (1954), when making decisions, individuals assign probabilities to the possible outcomes, and calibrate utilities to value these outcomes. The decision maker selects the outcome with the highest expected utility. When facing new options, individuals tend to stick with the status quo alternative. Which means they prefer current or previous decisions over new ones. Furthermore Samuelson & Zeckhauser (1988) found that for a variety of decision situations, individuals exhibit a significant and predictable status quo bias. This bias increases with the number of choice alternatives. According to Samuelson &
Zeckhauser (1988), the explanations of status quo bias fall into three main categories. The effect may be seen as the consequence of (1) rational decision making in the presence of transition costs and/or uncertainty; (2) cognitive misperceptions; and (3) psychological commitment stemming from misperceived sunk cost, regret avoidance, or a drive for consistency.
advantages. This phenomenon is also known as loss aversion (Khaneman, Knetsch & Thaler, 1991). Loss aversion is one of the most successful and extensively used explanatory construct in behavioral decision research. In first instance loss aversion was formalized as a component of prospect theory. Which refers to the analysis of decision making under risk (Tversky and Kahneman 1992). Loss aversion was used to explain the common unwillingness to accept gambles offering equal chances to receive or lose a given amount of money. Furthermore, loss aversion can seduce people into judging the same set of alternatives differently depending on whether they are phrased in terms of potential losses or potential gains (Bostrom & Ord, 2006). Thus, phrasing can be used to stir someone into taking a risk or avoiding it. Psychological commitment is the third and final explanation for status quo bias according to Samuelson & Zeckhauser (1988). Once individuals have committed to a certain decision, it is very likely he will stay with this decision because he has invested in it and wants to justify its choice. For schools this commitment means that once they have chosen a certain teaching method, they committed to a method, the less likely they are to switch to a new method. This can further be explained by regret avoidance. Individuals do not like the feeling of regret, so they try to avoid these feelings. Thus, when facing decisions, individuals tend to stick to the status quo and thus do not like change.
We expect that when schools are choosing a new teaching method, disadvantages loom larger than the advantages, and thus it is likely that they in some way are loss averse. We expect this because choosing a new teaching method, can be a risky choice. Teachers mostly use the same method for years and get used to this method. So, when choosing a new teaching method, it is very likely that they want to stay with the same method, since they know how this method works.
We predict that for schools, status quo bias plays a significant role and can be explained by loss aversion. The more particular schools (decision makers) are affected by loss aversion, the more likely this will influence their information needs at different points in the purchase funnel. In this research loss aversion will be used as the predictor for status quo bias.
H2: The effect of the role on information needs at different points in the purchase funnel
CONCEPTUAL MODEL
The hypotheses mentioned above lead to the following conceptual model.
FIGURE 2
CONCEPTUAL MODEL
METHODOLOGY
DesignIn order to answer the research questions of the current research and to provide de marketing board of a Dutch educational publisher recommendations for development of their website, data was collected by means of a survey amongst people working in secondary education. Goal of this survey was to find out whether people with different roles within the buying center of schools differ in their information needs across different points in the purchase funnel when choosing a new teaching method. Moreover, a sub goal of the survey was to discover what the participants think of the current website of the educational publisher. However, the latter will not be discussed in the current research.
Prior to the main study
Prior to the main study 6 interviews were conducted to acquire useful information to set up the survey. This pre-study was conducted during the weeks before the online survey was conducted. The interviews were held with people working in secondary education with different roles within schools.
The first interviewee was a subject teacher, teaching Dutch, who was both a teacher and part of the subject section (i.e., those responsible for choosing new teaching methods for a certain subject). The second interviewee was a subject teacher, teaching French, who was teacher, part of the subject section, and responsible for the flow of information on new methods and such. Furthermore, the third interviewee was a school coordinator, in his role he is responsible for the flow of information on new teaching methods, and the one responsible for making the purchase. The fourth interviewee was a member of the school management, which was responsible for both making the final decision for a new teaching method and for the purchase of new methods. At last, the fifth and last interviewee was a subject teacher, teaching English, who was just a teacher without any other roles within his school.
With this pre-study, the main topics for the main research were tested. The first thing tested was whether the roles of the buying center which were found in theory, match with the actual roles people have in schools in making decisions. Furthermore, interviewees were asked which information they needed most across the different purchase stages.
gave on which information needs they needed most across the stages of the purchase funnel, provided overarching concepts, which was used as input for the survey. To conclude, the interviews gave valuable insights for the main study.
Population and sampling method
After the interviews were analyzed the survey was made and distributed. The survey was distributed by e-mail amongst visitors of the educational publisher website. In order to reach the right audience, a dataset was set up. The participants in this dataset were selected on two criteria (1) they visited the website pages of secondary education in the last 6 months, and (2) left their e-mail address. These were the criteria because these visitors were expected to still remember their visit. Furthermore, the email addresses were needed to reach the visitors. The e-mail was sent in 2 rounds. Additionally, to generate some extra response, a link to the survey was put into all newsletters which were send to secondary educational teachers, in week 50. In the end, the survey was sent to approximately 4000 different email addresses. Lastly, the link to the survey was shared by the writer on LinkedIn and was reshared by 8 other people. In order to make sure only people who had visited the website before were filling out the survey, the link was accompanied with the text that the survey was only for people working in secondary education and had visited/used the website of the educational publisher.
Furthermore, participants were asked if they had visited the website before. If they had not visited the website, they were excluded of the part of the survey where participants had to give their opinion about the current website. Also, participants were asked if they worked in secondary education. If they did not, they were excluded from the survey. In the end, a total of 105 respondents participated in the online survey, of which 6 respondents were eventually deleted from the dataset due to missing values. Further details of the respondents will be discussed in the results section.
Procedure
work at, how many years they are employed in education, how long they use a teaching method of the educational publisher (this question was only shown to the educational publisher clients), and what subject they teach. These questions were used to give the educational publisher an overview of their clients and will be used in a follow-up analysis, which is not included in the current research.
The educational publisher wants to develop their website, they want to do so on the bases of the customer needs. Therefore, questions were included where participants had to give their opinion on the current website of the educational publisher. Furthermore, some questions about what they used on this website, and what information they seek for on the website were included.
In order to answer the research questions, participants received questions on their information needs in the different stages of the purchase funnel (i.e., awareness, consideration, decision, and post-purchase). Here, for every purchase funnel stage a statement was given which indicated the participant to be in a particular purchase stage and had to answer what information they would need in this stage. Furthermore, a question on website functionalities was included. This was included for the recommendations on website design for the educational publisher. The last part of the survey examined loss aversion. Here participants got presented five different loss aversion propositions. Participants were asked to select the answer with which they felt most comfortable. At the end of the survey respondents were given the chance to give suggestions. These suggestions are not used in this research but will be discussed in detail with the educational publisher.
Materials
Independent variable
The independent variable is which role of the buying center a participant has. The five roles of the buying center: user, influencer, gatekeeper, decision maker, and buyer, which were found in literature, were used in the survey to measure the decision involvement of participants. Participants could indicate which role(s) best described them in the decision process in choosing a new teaching method.
Dependent variable
a 5-point Likert scale. With this scale respondents could indicate to what degree they agree with the presented statement from “totally disagree”, to “totally agree” with the different statements which were linked to the stages of the purchase funnel. In total 24 statements were asked to measure information needs. Every stage was introduced with a sentence like the following:
When considering a a new teaching edition/method, I think it is important that the website of a
supplier of educational resources supports me in lesson preparation, teaching, making tests and coaches me with. After which the statements were shown. To measure information needs
in the awareness stage three statements were included. For measuring information needs in the consideration stage eight statements were included. Furthermore, information needs in the decision stage were measured by eight statements, and the information needs in the post-purchase stage were measured by five statements.
To check whether the different questions could be summed into one variable, reliability of the scales was checked. First, we checked for correlation, after which we looked at Cronbach’s Alpha. We did this for all four stages.
a This word differed per stage.
Reliability for awareness
A correlation analysis showed that the first and second questions measuring information needs in the awareness stage did not significantly correlate (r=0,03, p =0,81). The second and third question, also did not correlate significantly (r=0,03, p=0,81), furthermore the first and third question testing awareness, also did not correlate significantly (r = .120, p = .253). These results indicate that the variables testing awareness cannot be summed into a new variable. Additionally, the Cronbach’s Alpha was checked to make sure the variables could not be summed. Here, SPSS, shows a value for Cronbach’s Alpha of 0,114 for the combination of the three questions testing “awareness”. As expected, this does not reach the minimum to allow for summing up the variables, which is a Cronbach’s Alpha of .70 according to Santos (1999). When examining the rest of the output of SPSS, we find the column “Cronbach’s Alpha if item deleted’ which shows the rest of the value of Cronbach’s Alpha if one of the questions is omitted. However, for all of the questions the Cronbach’s Alpha decreases if the questions are deleted. To conclude, the three questions testing awareness cannot be summed. Therefore, the questions measuring awareness will be used individually in every analysis.
A correlation analysis showed that all questions measuring consideration did significantly correlate. These results indicate that we can sum all variables into a new variable which represents consideration. Additionally, we looked at Cronbach’s Alpha to check the reliability of the scale. Here, SPSS, shows a value for Cronbach’s Alpha of 0,83 for the combination of the seven questions testing consideration. And thus, the scale is reliable, and the variables can be summed. Further examination of the output shows the Cronbach’s Alpha if deleted column and shows us that de number only decreases if questions would be omitted. To conclude, the seven questions measuring information needs in the consideration stage are summed.
Reliability for decision
A correlation analysis showed that all questions, except for the first and the fifth question measuring information needs in the decision stage (r=0,12, p=0,26), did significantly correlate, see table in the appendix for all numbers. Additionally, we looked at Cronbach’s Alpha to check the reliability of the scale. The SPSS output shows a value for Cronbach’s Alpha of 0,77 for the combination of the six questions measuring the information needs in the decision stage, which means the scale is reliable. Furthermore, by looking at the Cronbach’s Alpha if item deleted column, it can be seen that the value only decreases when omitting a question. Therefore, it was decided that the scale is reliable although the correlation showed a non-significant value for the first and the fifth question. To conclude, the six questions measuring information needs in the decision stage are summed.
Reliability post-purchase
At last, a correlation analysis was run to check the reliability of the scale testing post-purchase. The output of SPSS showed that all five questions testing information needs in the post-purchase stage did significantly correlate, which means the variables can be summed. Additionally, the Cronbach’s Alpha was checked, which shows a value of 0,87 for the combination of the five questions measuring information needs in the post-purchase stage. Further examination of the Item-Total Statistics table shows that the value of the Cronbach’s Alpha only decreases if questions would be omitted. On the base of these findings, the five questions testing “post-purchase” were summed.
Moderator
statement intended loss aversion; the other statement intended no loss aversion. An example of one of the questions asked was as followed:
(a) You have a 100% chance of winning €100, -
(b) You have 50% chance of winning €200, - or you have 50% of winning €0,-
In order to see whether a participant was loss averse or was not, the results of the statements were summed into a new variable. The sum ranges from 0-5, where 0 indicates total loss aversion and 5 no loss aversion at all. In order to split the variable into two groups, the loss aversion group and the no loss aversion group, the median was calculated (m= 1,00). With the median a new variable was made. Here all participants beneath the median were labelled loss averse and all above the median were labeled not loss averse.
Plan of analysis
After the data collection, the data was transferred to SPSS, in which all analyses were conducted. First, the data was prepared for the analyses, which meant all outliers and errors were removed from the data set. As was mentioned earlier, 6 participants were removed from the dataset because of missing values.
Statistical tests
In order to test the first hypothesis, whether members with different roles differ in their information needs at different points in the purchase funnel, the means of information needs of the different groups were compared. This was done by means of several Mann Whitney U tests, and several Independent Samples t-tests. The Mann Whitney U analyses were done for the questions measuring awareness, because this data was non-parametric.
RESULTS
A total of 105 respondents participated in the online survey, of which 6 respondents were eventually deleted from the dataset because they did not complete the survey. All output of the conducted analyses can be requested from the writer.
TABLE 1
DESCRIPTIVES STATISTICS
Variables Summary Statistics
Using teaching method of the educational publisher
Age
Yes (97%), No (3%)
21-30 (10,1%), 31-40 (16,2%), 41-50 (17,2%), 51-60 (42,2), 61-70 (14,1%)
Role within school Subject teacher (67,9%), Subject section a (22,9%), educational resources coordinator (6,1%), school management (3,1%),
Subject Dutch (12,5%), English (24%), German (21,2), French (%), Math (%), Science (4,8%), Physics (1,0%), Biology (5,7%), NaSk (1,0%), Arithmetic (1,9%), Technic Science (1,0%), Geography (15,2%), History (6,7%), Economy (3,8%), People and Society (3,8%)
Role within buying center a User (75,8%), Influencer (78,8%), Gatekeeper (23,2%), Decision Maker (31,1%), Buyer (8,1%)
Use of method educational publisher
Less than 1 year (3%), 1-3 years (16,2%), 4-6 years (25,3%), 7-9 years (20,2%), over 10 years (35,4%) Use of edition
Frequency usage website
Less than 1 year (20,2%), 1 year (8,1%), 2 years (24,2%), 3 years (17,2%), 4 years (13,1%), 5 years (5,1%), over 5 years (12,1%).
Daily (5,1%), A few times per week (12,1%), A few times per month (17,2%), A few times per year (46,5%), Once a year (9,1%), Less than once per a year (6,1%), Never (4%);
A Those responsible for choosing a new teaching method)
Mean SD Minimum Maximum
Years employed in education 20,30 10,07 1 41
Table 1 Descriptive Statistics shows a summary of the most important descriptives. Most respondents (97%) are using a teaching method from the educational publisher, and only 3% of respondents uses a teaching method from another educational publisher. Furthermore, most respondents (42,2%) are aged between 51 and 60. This can be an explanation for the high mean for years employed in education (M=20,30). Most respondents are loyal customers since they make use of teaching methods of the educational publisher for over 10 years (35,4%). Furthermore, what is interesting to see is that most respondents (46,5%) only make use of the website a few times a year.
Information needs in the different stages of the purchase funnel
In the next paragraph an overview of information needs per purchase funnel is given. Here all statements with their corresponding mean are given. All questions were measured on a 5-point Likert scale, 1 indicating totally disagree and 5 indicating totally agree. The means in the tables give the means for information need per statement in the corresponding purchase funnel stage.
TABLE 2
INFORMATION NEEDS IN THE AWARENESS STAGE
Mean SD Minimum Maximum
Subject specific news messages
TABLE 3
INFORMATION NEEDS CONSIDERATION STAGE
Mean SD Minimum Maximum
Written information Videos & Animations Digital view copy
Demo’s on digital learning platform
Ability to order assessment package
Invite account manager Chat functionality Choice tool 3,36 3,20 3,55 3,39 3,87 3,36 2,71 2,74 0,86 0,99 0,87 0,92 0,63 1,18 1,34 1,30 1,00 1,00 1,00 1,00 1,00 1,00 1,00 1,00 5,00 5,00 5,00 5,00 5,00 5,00 5,00 5,00 TABLE 4
INFORMATION NEEDS DECISION STAGE
Mean SD Minimum Maximum
Interviews Product reviews Catalogue prices Product guide Web shop Calendar 2,91 2,30 3,04 2,82 2,88 2,05 1,04 1,08 0,93 0,91 1,01 1,05 1,00 1,00 1,00 1,00 1,00 1,00 5,00 5,00 5,00 5,00 5,00 5,00 TABLE 5
INFORMATION NEEDS POST-PURCHASE STAGE
Mean SD Minimum Maximum
Decision involvement and information needs
To test the first hypothesis: Members with different roles in the buying center, will have
different information needs at different points in the purchase funnel, several analyses were
carried out of which the results will be discussed in the next section. The results will be discussed per role (i.e., user, influencer, gatekeeper, decisionmaker, and buyer).
TABLE 6
FREQUENCIES ROLES
Frequency Percent Valid Percent
Cumulative Percent One role Two roles Three roles Four roles Five roles Total 26 42 19 10 2 99 26,3 42,4 19,2 10,1 2,0 100,0 26,3 42,4 19,2 10,1 2,0 100,0 26,3 68,7 87,9 98,0 100,0
Table 6 shows that most respondents have more than just one role within the buying center. Most respondents (42,2%) indicated to have two roles, followed by one role (26,3%), followed by three roles (19,2%), followed by four roles (10,15%), and last, 2% indicated to have all five roles. Some respondents are user, influencer, and decision maker, other respondents are influencer and buyer, and some are only a user.
Since most respondents have more than one role, it is hard to compare information needs across the roles individually (e.g., user versus decision maker). Therefore, it was more convenient to compare respondents with a particular role (e.g. user, with the possibility to additionally have another role) with respondents without this particular role (i.e., any role except for the user role). Thus, users are compared to non-users, influencers are compared with non-influencers, gatekeepers are compared with non-gatekeepers, decision makers are compared with non-decision makers, and buyers are compared with non-buyers.
able to make decisions and those who are unable to make decisions. These groups were made on the bases of the different roles and will be discussed later on in this chapter.
It is expected that respondents who are able to make decisions will differ in their information needs at different points in the purchase funnel from respondents who are unable to make decisions.
Overall information needs
TABLE 7
DESCRIPTIVES OVERALL INFORMATION NEEDS
Mean SD Minimum Maximum
Overall information need Valid N (listwise)
2,93 0,50 1,36 4,18
Table 7 shows the mean (M=2,93, SD= 0,50) for overall information needs. Here, all questions concerning information needs across all points of the purchase funnel are summed into one scale. The overall information needs variable will be used to discover whether the overall information needs at different points in the purchase funnel differ across the different roles. This will be tested by means of five independent samples t-tests and will be discussed in the next section.
Overall information needs for non-users versus users
The non-user group (N=22) was associated with an overall information needs value of
M=2,89 (SD=0,54). By comparison, the user group (N=67), was associated with a numerically
bigger overall information needs value M=2,95 (SD=0,49). To test whether non-users and users were associated with statistically significantly different means concerning overall information needs, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a non-significant value for both non-users (p=0,34), and users (p=0,30), furthermore inspection of the Q-Q Plots also reveals the variables to be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(87)=0,05, p=0,83. The independent samples t-tests was associated with a statistically non-significant effect, t(87)=-0,51, p=0,61.
Overall information needs for non-influencers versus influencers
The non-influencer group (N=18) was associated with an overall information needs value of M=3,17 (SD=0,55). By comparison, the influencer group (N=71) was associated with a numerically smaller overall information needs value M=2,87 (SD=0,48). To test the test whether non-influencers and influencers were associated with statistically significantly different means concerning overall information needs, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. Although the Shapiro-Wilk test show a non-significant value for non-influencers (p=0,69), and a significant value for influencers (p=0,03), further inspection of the Q-Q Plots reveals the variables to both be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(87)=1,14, p=0,29. The independent samples t-test was associated with a statistically significant effect, t(87)=2,27 p=0,03. Thus, the non-influencers were associated with a statistically significantly bigger mean than the influencers. Which means non-influencers need more information than influencers.
Overall information needs for non-gatekeepers versus gatekeepers
The non-gatekeeper group (N=67) was associated with an overall information needs value of M=2,99 (SD=0,54). By comparison, the gatekeeper group (N=22) was associated with a numerically smaller overall information needs value M=2,77 (SD=0,32). To test whether non-gatekeepers and non-gatekeepers were associated with statistically significantly different means concerning overall information needs, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a non-significant value for both non-gatekeepers (p=0,19), and gatekeepers (p=0,14), furthermore inspection of the Q-Q Plots also reveals the variables to both be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(87)=2,89, p=0,09. The independent samples t-tests was associated with a statistically non-significant effect, t(87)=1,78 p=0,08. Thus, the non-gatekeepers were not associated with a statistically significantly bigger mean than the gatekeepers. Meaning that non-gatekeepers and non-gatekeepers do not differ in overall information needs.
Overall information needs for non-decision makers versus decision makers
non-decision makers and decision makers were associated with statistically significantly different means concerning overall information needs, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. Although the Shapiro-Wilk test show a significant value for non-decision makers (p=0,03), and a non-significant value for decision makers (p=0,40), further inspection of the Q-Q Plots reveals the variables to both be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(87)=0,07, p=0,79. The independent samples t-tests was associated with a statistically non-significant effect, t(87)=0,96 p=0,34. Thus, the non-decision makers were not associated with a statistically significantly bigger mean than the decision makers. Meaning that non-decision makers and decision makers do not differ in overall information needs.
Overall information needs for non-buyers versus buyers
The non-buyer group (N=83) was associated with an overall information needs value of
M=2,94 (SD=0,48). By comparison, the buyer group (N=6) was associated with a numerically
smaller overall information needs value M=2,85 (SD=0,86). To test whether non-buyers and buyers were associated with statistically significantly different means concerning overall information needs, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a non-significant value for both non-buyers (p=0,41), and buyers (p=0,67), furthermore inspection of the Q-Q Plots also reveals the variables to both be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(87)=3,22,
p=0,08. The independent samples t-tests was associated with a statistically non-significant
effect, t(87)=0,43 p=0,67. Thus, the non-buyers were not associated with a statistically significantly bigger mean than the buyers. Meaning that non-buyers and buyers do not differ in their overall information needs.
Users
Information needs in the awareness stage for non-users versus users
Since the questions measuring information needs in the awareness stage could not be summed into one scale, they were tested individually. The data for these questions were not normally distributed, since the questions were measured on a 5-point Likert scale. Therefore, the data were treated as non-parametric. This applies to all analyses measuring information needs in the awareness stage (i.e., for users, influencers, gatekeepers, decision makers, and buyers).
To test whether non-users and the users were associated with statistically significant different means concerning awareness, a Mann Whitney U analysis was performed.
Question 1 measuring information needs in the awareness stage: The non-user group (N=24) was associated with a Mean Rank of 49,29 (SUM=1095,00) for the first awareness question, by comparison of a Mean Rank of 48,90 (SUM=3658,00) for the user group (N=73). To test the hypothesis that the non-user group and user group were associated with statistically different Means, a Mann Whitney U was carried out. This test showed that there was a non-significant difference (U=869,00, p=0,95) between the non-user group and the user group.
Question 2 measuring information needs in the awareness stage: The non-user group (N=24) was associated with a Mean Rank of 42,33 (SUM=1016,00) for the second awareness question, by comparison of a Mean Rank of 51,19 (SUM=3737,00) for the user group (N=73). To test the hypothesis that the non-user group and user group were associated with statistically different Means, a Mann Whitney U was carried out. This test showed that there was a non-significant difference (U=716,00, p=0,17) between the non-user group and the user group.
To conclude, the above results indicate that non-users and users do not differ in their information needs at the awareness stage.
Information needs in the Consideration stage for non-users versus users
The non-user group (N=23) was associated with a consideration value of M=3,21 (SD=0,88). By comparison, the user group (N=75) was associated with a numerically bigger consideration value M=3,29 (SD=0,66). To test whether non-users and users were associated with statistically significantly different means concerning consideration, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a non-significant value for both non-users (p=0,42), and users (p=0,66), furthermore the inspection of the Q-Q Plots also reveals the variables to be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(96)=1,19, p=0,28. The independent samples t-tests was associated with a statistically significant effect, t(96)=-0,18, p=0,86. Thus, the non-users were not associated with a statistically significantly smaller mean than the non-users. Meaning that non-users and users do not differ in their information needs at the consideration stage.
Information needs in the Decision stage for non-users versus users
The non-user group (N=23) was associated with a decision value of M=2,65 (SD=0,78). By comparison, the user group (N=75) was associated with a numerically bigger decision value
M=2,70 (SD=0,64). To test whether non-users and users were associated with statistically
significantly different means concerning information needs in the decision stage, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a significant value for both non-users (p=0,85), and non-users (p=0,07), furthermore the inspection of the Q-Q Plots also reveals the variables to be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(96)=1,52, p=0,22. The independent samples t-tests was associated with a statistically significant effect, t(96)=-0,30, p=0,77. Thus, the non-users were not associated with a statistically significantly smaller mean than the non-users. Meaning that non-users and users do not differ in their information needs at the decision stage.
Information needs in the post-purchase stage for non-users versus users
post-purchase value M=2,48 (SD=0,77). To test whether non-users and users were associated with statistically significantly different means concerning information needs in the post-purchase stage, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a non-significant value for non-users (p=0,21), and a significant value for users (p=0,00), further the inspection of the Q-Q Plots reveals the variables to be both normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(97)=0,37,
p=0,54. The independent samples t-tests was associated with a statistically non-significant
effect, t(97)=-0,87, p=0,39. Thus, the non-users were not associated with a statistically significantly smaller mean than the users. We can conclude that for users versus non-users we did not find any statistically significant differences in their information needs in the different stages in the purchase funnel.
Based on the findings of the Mann Whitney U analysis, and the Independent Samples t-tests, we can conclude that there was not found any statistical evidence that non-users versus non-users differ in information needs across the different stages of the purchase funnel. Meaning that non-users versus users do not differ in their information needs at different points in the purchase funnel.
Influencers
Information needs in awareness stage for influencers non-versus influencers
To test whether non-influencers and the influencers were associated with statistically significant different means concerning awareness, a Mann Whitney U analysis was performed.
Question 1 measuring information needs in the awareness stage: The non-influencer group (N=21) was associated with a Mean Rank of 52,14 (SUM=1095,00) for the first awareness question, by comparison of a Mean Rank of 48,13 (SUM=3658,00) for the influencer group (N=76). To test whether non-influencer group and influencer group were associated with statistically different Means, a Mann Whitney U was carried out. This test showed that there was a non-significant difference (U=732,00, p=0,55) between the non-influencer group and the influencer group.
awareness question, by comparison of a Mean Rank of 46,68 (SUM=3594,50) for the influencer group (N=77). To test whether non-influencer group and influencer group were associated with statistically different Means, a Mann Whitney U was carried out. This test showed that there was a non-significant difference (U=591,50, p=0,10) between the non-influencer group and the influencer group.
Question 3 measuring information needs in the awareness stage: The non-influencer group (N=20) was associated with a Mean Rank of 44,78 (SUM=895,50) for the third awareness question, by comparison of a Mean Rank of 48,86 (SUM=3664,50) for the influencer group (N=75). To test whether non-influencer group and influencer group were associated with statistically different Means, a Mann Whitney U was carried out. This test showed that there was a non-significant difference (U=685,50, p=0,54) between the non-influencer group and the influencer group.
To conclude, the above results indicate that non-influencers and influencers do not differ in their information needs at the awareness stage.
Information needs in the consideration stage for non-influencers versus influencers The non-influencer group (N=21) was associated with a consideration value of M=3,36 (SD=0,79). By comparison, the influencer group (N=78) was associated with a numerically smaller consideration value M=3,25 (SD=0,70). To test whether non-influencers and influencers were associated with statistically significantly different means concerning consideration, an independent samples test was performed. Before the independent samples t-test was executed, a t-test for normality was done. The Shapiro-Wilk t-test show a non-significant value for both non-influencers (p=0,22), and influencers (p=0,11), furthermore the inspection of the Q-Q Plots also reveals the variables to be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(97)=0,96,
p=0,33. The independent samples t-tests was associated with a statistically non-significant
effect, t(97)=0,58 p=0,56. Thus, the non-influencers were not associated with a statistically significantly bigger mean than the influencers. Meaning that non-influencers and influencers do not differ in their information needs at the consideration stage.
Information needs in the decision stage for non-influencers versus influencers
smaller decision value M=2,61 (SD=0,66). To test whether non-influencers and influencers were associated with statistically significantly different means concerning decision, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a significant value for both non-influencers (p=0,18), and non-influencers (p=0,10), furthermore the inspection of the Q-Q Plots also reveals the variables to be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(96)=0.06, p=0,81. The independent samples t-tests was associated with a statistically significant effect, t(96)=2,39, p=0,02. Thus, the non-influencers were associated with a statistically significantly bigger mean than the influencers.
These results suggest that non-influencers have more need for information in the decision stage, than influencers in this stage. Moreover, the effect size of this result is medium large (Cohen’s d=0,6), according to Cumming & Calin-Jageman (2019).
Information needs in post-purchase stage for non-influencers versus influencers
The non-influencer group (N=21) was associated with a post-purchase value of M=2,56 (SD=0,89). By comparison, the influencer group (N=78) was associated with a numerically smaller post-purchase value M=2,41 (SD=0,73). To test whether non-influencers and influencers were associated with statistically significantly different means concerning post-purchase, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a non-significant value for both non-influencers (p=0,56), and influencers (p=0,00), further the inspection of the Q-Q Plots reveals the variables to be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(97)=0,18, p=0,67. The independent samples t-tests was associated with a statistically non-significant effect, t(97)=0,79
p=0,43. Thus, the non-influencers were not associated with a statistically significantly bigger
mean than the influencers. Meaning that non-influencers and influencers do not differ in their information needs at the post-purchase stage.
the stages no statistical evidence was found. Which suggests non-influencers and influencers in the awareness, consideration, and post-purchase stage do not differ in their information needs.
Gatekeepers
Information needs in awareness stage for non-gatekeepers versus gatekeepers
To test whether non-gatekeepers and the gatekeepers were associated with statistically significant different means concerning awareness, a Mann Whitney U analysis was performed.
Question 1 measuring information needs in the awareness stage: The non-gatekeeper group (N=74) was associated with a Mean Rank of 48,84 for the first awareness question, by comparison of a Mean Rank of 49,52 for the influencer group (N=23). To test whether non-gatekeeper group and gatekeeper group were associated with statistically different Means, a Mann Whitney U was carried out. This test showed that there was a non-significant difference (U=839,00, p=0,92) between the non-gatekeeper group and the gatekeeper group.
Question 2 measuring information needs in the awareness stage: The
non-gatekeeper group (N=74) was associated with a Mean Rank of 49,67 for the second awareness question, by comparison of a Mean Rank of 46,85 for the influencer group (N=23). To test whether non-gatekeeper group and gatekeeper group were associated with statistically different Means, a Mann Whitney U was carried out. This test showed that there was a non-significant difference (U=801,50, p=0,66) between the non-gatekeeper group and the gatekeeper group.
Question 3 measuring information needs in the awareness stage: The
non-gatekeeper group (N=72) was associated with a Mean Rank of 49,68 for the third awareness question, by comparison of a Mean Rank of 42,74 for the influencer group (N=23). To test whether non-gatekeeper group and gatekeeper group were associated with statistically different Means, a Mann Whitney U was carried out. This test showed that there was a non-significant difference (U=707,00, p=0,27) between the non-gatekeeper group and the gatekeeper group.
Information needs in the consideration stage for non-gatekeepers versus gatekeepers The non-gatekeeper group (N=76) was associated with a consideration value of M=3,31 (SD=0,74). By comparison, the gatekeeper group (N=23) was associated with a numerically smaller consideration value M=3,15 (SD=0,63). To test whether non-gatekeepers and gatekeepers were associated with statistically significantly different means concerning consideration, an independent samples test was performed. Before the independent samples t-test was executed, a t-test for normality was done. The Shapiro-Wilk t-test show a non-significant value for both non-gatekeepers (p=0,24), and gatekeepers (p=0,37), furthermore inspection of the Q-Q Plots reveals the variables to be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(97)=0,89, p=0,35. The independent samples t-tests was associated with a statistically non-significant effect, t(97)=0,96
p=0,34. Thus, the non-gatekeepers were not associated with a statistically significantly bigger
mean than the gatekeepers. Meaning that non-gatekeepers and gatekeepers do not differ in their information stage in the consideration stage.
Information needs in the decision stage for non-gatekeepers versus gatekeepers
The non-gatekeeper group (N=75) was associated with a consideration value of
M=2,76 (SD=0,67). By comparison, the gatekeeper group (N=23) was associated with a
numerically smaller decision value M=2,46 (SD=0,61). To test whether non-gatekeepers and gatekeepers were associated with statistically significantly different means concerning decision, an independent samples t-test was performed. Before the independent samples t-test was executed, a test for normality was done. The Shapiro-Wilk test show a non-significant value for both non-gatekeepers (p=0,15), and gatekeepers (p=0,34), furthermore inspection of the Q-Q Plots reveals the variables to be normally distributed. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s test, F(96)=0,17, p=0,68. The independent samples t-tests was associated with a statistically moderately significant effect,
t(96)=1,93 p=0,06. Thus, the non-gatekeepers were associated with a statistically moderately
significantly bigger mean than the gatekeepers. Furthermore, the effect size of these results was medium large (Cohen’s d=0,47), according to Cumming & Calin-Jageman (2019).
