University of Amsterdam, Amsterdam, The Netherlands Master in Business Administration – Marketing Track
MASTER THESIS
“The Influence of Self-Assembling Intensity and Competence on
Valuation”
Author: Ms J.S. Strebus MSc Student Number: 11417927 Thesis Supervisor: Mr Dr A. Zerres Second Supervisor: Mr J. Demmers MSc Date: 23rd of June, 2017
Statement of Originality
This document is written by Student Janneke Suzan Strebus who declares to take full
responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is solely responsible for the supervision of completion of the work, not for the contents.
Acknowledgement
This Master thesis is the final assignment for my MSc Business Administration – Marketing at the University of Amsterdam. At this point, I would like to thank some people. Completing my studies, including thesis, would not have been possible without the help of those people.
First of all, I am especially grateful to my thesis supervisor Dr Alfred Zerres for his guidance throughout the process of writing this master thesis. Alfred helped me to look critical to the whole process by providing useful discussions and good feedback on how to strengthen this master thesis.
Moreover, I would like to express my gratitude to Romy van Baarsen, Pauli Nijkamp, and Ruben Fuchs who helped and supported me in the process of writing this thesis.
Finally, big thanks to all participants, who were willing to invest time and effort to participate in my online experiment.
Janneke Strebus MSc
Table of Content Statement of Originality ... 2 Acknowledgement ... 3 Abstract ... 6 1. Introduction ... 7 2. Literature Review ... 10
2.1. The Concept Co-Production ... 10
2.2. The IKEA-Effect ... 12
2.3. Equity Theory ... 14
2.4. Theoretical Mechanism Underlying Valuation ... 17
2.4.1. Self-determination theory ... 17 2.4.1.1. Goal relevance ... 19 2.5. Conceptual Model ... 22 3. Method ... 23 3.1. Sample ... 23 3.2. Design... 23 3.3. Procedure ... 24 3.4. Stimulus Materials... 24
3.4.1. Manipulation of goal relevance ... 25
3.4.2. Manipulation of self-assembling intensity ... 25
3.5. Pre-Test ... 27
3.5.1. Sample and procedure pre-test ... 27
3.5.2. Manipulation check pre-test ... 28
3.6. Measures... 29
3.6.1. Dependent variables ... 29
3.6.2. Control and demographic variables ... 31
3.7. Statistical Procedure ... 32
4. Results ... 33
4.1. Descriptive Data of the Sample ... 33
4.2. Reliability of Scales ... 34
4.3. Manipulation Check ... 35
4.3.1. Goal relevance ... 35
4.3.3. Challengingness, believability and reliability ... 36
4.4. Skewness and Kurtosis ... 37
4.5. Correlation Matrix ... 38
4.6. Hypotheses Testing ... 41
4.6.1. Hypothesis 1... 41
4.6.1.1. Influence of self-assembling on willingness-to-pay... 41
4.6.1.2. Influence of self-assembling on satisfaction ... 42
4.6.1.3. Influence of self-assembling on loyalty ... 43
4.6.2. Hypothesis 2... 43
4.6.2.1. Influence of self-assembling intensity on willingness-to-pay ... 43
4.6.2.2. Influence of self-assembling intensity on satisfaction ... 45
4.6.2.1. Influence of self-assembling intensity on loyalty ... 46
4.6.3. Hypothesis 3... 49
4.6.4. Hypothesis 4... 51
4.6.5. Hypothesis 5... 51
5. Discussion ... 52
5.1. Theoretical implications ... 54
5.1.1. Explanation opposed IKEA-effect and self-assembling intensity levels ... 54
5.1.2. Explanation missing link competence ... 57
5.2. Managerial Implications ... 57
5.3. Limitations and Future Research... 57
6. Conclusion ... 63
References ... 64
Appendices ... 72
Appendix A: Manipulation Goal Relevance ... 72
Appendix B: Manipulation Self-Assembling Intensity ... 73
Abstract
This online experiment (n = 389) investigated the influence of self-assembling intensity and feelings of competence on valuation towards the company (loyalty, satisfaction) and product (willingness-to-pay). Self-assembling intensity is manipulated by exposing participants to a scenario in which a product was finished (control condition) or had to be self-assembled (self-assembling condition). The latter condition was subdivided into low, medium and high-intensity levels. A higher valuation was expected for the self-assembling condition compared to the control condition, whereby increased levels of intensity would result in enhanced valuation. However, boundaries were expected, when intensity levels get too high valuation would be reduced to levels that are even lower than the control condition. Furthermore, it was expected that feelings of competence underlie this effect and that making the goal to feel competent again relevant would help to overcome the adverse effects of higher self-assemble intensities on product valuation. Contrary to the expectations, self-assembling products result in lower valuation compared to the obtainment of a finished product. For the different levels of intensity, the expectations were partially met, as expected participants in the high-intensity condition valued the organisation and product less compared to the control condition.
Contradictory to the expectations valuation of the low and medium intensity condition was also lower than the control condition. Furthermore, the expectations about competence were not met. Competence does not seem to influence the effect of self-assembling intensity on valuation. Also, goal relevance appeared to have no impact on valuation. Despite the contradicting results this study generates opportunities for follow-up research.
Keywords: Self-Assembling, Intensity, Valuation, Willingness-to-Pay, Satisfaction, Loyalty, Competence, IKEA-effect
1. Introduction
There is a current trend in which a multitude of companies encourage customers to engage in co-producing products and services. Customers self-assemble new furniture from prefabricated kits, follow the directions on convenience food packages to prepare meals, customise their shoes or scarfs, use online banking and scan their groceries. In this way, customers act as co-producers of the goods and services rather than being passive recipients (Prahalad & Ramaswamy, 2000; Mochon, Norton, & Ariely, 2012; Vargo & Lusch, 2004). This proliferation of co-production has stimulated academic interest in examining the implications of customer participation in co-production processes (e.g., Franke, Schreier, & Kaiser, 2010; Mochon et al., 2012; Norton, Mochon, Ariely, 2012). Research has
demonstrated that people value those self-created products and services more than products finished by the organisation, resulting in among other things a higher willingness-to-pay (WTP; Norton et al., 2012). Whether participants self-assembled (e.g., IKEA boxes, folded origami, LEGO) or self-designed (e.g., T-shirts, scarfs, watches) products, the effect remained the same, labour increases the valuation of co-produced products (Dohle, Rall, & Siegrist, 2014; Franke & Piller, 2004; Franke & Schreier, 2010; Franke et al., 2010; Mochon, et al., 2012; Norton et al., 2012; Schreier, 2006).
Scholars have shown increased interest in the influence of co-production on valuation the last decade, questioning which aspects determine the value customers derive from co-produced products. The purpose of this research is to build on previous literature by
providing new knowledge regarding issues that might influence the valuation of co-produced products. Although previous studies show the beneficial aspects of co-production in
comparison to a traditional organisational production, something that has been largely neglected is the role of co-production intensity. Co-production intensity is the required time and effort that is needed to create a product or service (Haumann, Güntürkün, Schons, &
Wieseke, 2015; Stokburger-Sauer, Scholl-Gissemann, Teichmann, & Wetzels, 2016). Little is known about the consequences of increasing those customer inputs within co-production situations. This neglect is surprising, as customers' perceived effort and time investment are of great importance in their evaluation of a co-production processes. Specifically, there are good reasons to assume that higher perceived intensity may have a negative influence on customer loyalty (Stokburger-Sauer et al., 2016), WTP (Norton et al., 2012) and satisfaction (Haumann et al., 2015) because consumers generally view effort and time as cost factors resulting in a negative influence on those outcomes (Berry, Seiders, & Grewal, 2002; Etgar 2008). Therefore, the focus of this research will be on how different levels of intensity influence the valuation of the product measured as WTP and of the organisation measured as loyalty, and satisfaction, compared to a standardised identical alternative.
To extend the literature further this study takes a mechanism that seems to underlie the value enhancing effect of co-producing products into account. Research has shown a close link with competence, stemming from the self-determination theory, whereby the effect of co-production on product valuation is mediated by feelings of competence (Deci & Ryan, 2011; Franke et al., 2010; Mochon et al., 2012). Co-production increases people’s sense of competence thereby creating a higher WTP (Mochon et al., 2012). Presumably, higher valuation of products is established when the required co-production intensity goes up because higher feelings of competence are linked with the product (Berry et al., 2002). However, when the intensity gets too high, no such feelings will be linked to the product resulting in lower valuation (e.g., Deci & Ryan, 2000, 2011).
Furthermore, studies have shown that when people’s feelings of competence have been threatened the goal to restore those feelings is stimulated thereby creating a preference for products that need to be self-produced over standardised firm productions (Gao, Wheeler, & Shiv, 2009; Mochon et al., 2012). It is unclear whether the valuation of products is affected
by reduced feelings of competence during the co-production process. There are good reasons to assume that lowering customers’ sense of competence enhances valuation of products requiring higher intensity levels, because reducing those feelings creates an intrinsic motivation to increase those competence feelings again (Deci & Ryan, 2000, 2011).
Consequently, tasks requiring more time and effort will be valued to a greater extent because those tasks contribute more to the need to feel competent again (Mochon et al., 2012;
Stokburger-Sauer et al., 2016). Therefore, lowering people's sense of competence might deduct the adverse effect of higher intensity.
In light of these research gaps, this study search answers regarding the following research questions: What is the influence of co-production intensity on valuation, measured
as willingness-to-pay (WTP), satisfaction and loyalty? Specifically, is there a difference between low, medium and high co-production intensity, compared to a standardised firm production? Thereby, is the influence of intensity on product valuation (WTP) influenced by feelings of competence?
Investigating these research questions will result in gaining a deeper understanding of the dynamics between co-production intensity, valuation and competence. For both scholars and practitioners, this study is relevant. Academics and practitioners alike have underscored the importance of understanding the mechanisms through which co-production generates value for customers (e.g., Franke et al., 2010; Mochon et al., 2012; Norton et al., 2012) and the organisation (e.g., Haumann et al., 2015; Pritchard, 1969; Stokburger-Sauer et al., 2016) because of the current trend towards co-creation. This trend makes it interesting to investigate how value creation can be optimised. For scholars, this research will add to the previous literature by providing new scientific insights regarding the influence of different levels of co-production intensity on valuation, compared to a standardised firm production. This study does also provide a more comprehensive overview of different value enhancing outcomes by
taking both the valuation of the product and organisation into account. Thereby, competence is included providing insights into the impact of this factor on the effect. For practitioners, this research will be of economic value since it explores how different levels of intensity influence valuation. Therefore, this study furnishes companies insights to optimise co-production processes to generate customers who are willing to pay more for the product and who are more loyal and satisfied with the company. Accordingly, this study generates new scientific knowledge and might help to increase customers' valuation towards co-producing organisations.
Overall, the current study contributes to marketing research and management in several ways. First, this is the first study that investigates the influence of different levels of co-production intensity on valuation compared to a standardised firm production. The study builds on previous literature by showing that increasing the co-production intensity can result in a reduced valuation. Second, this research focus on both the economic derived value measured as WTP and the subjective value measured as loyalty and satisfaction. Herewith, providing a more comprehensive overview of the effects of self-production on different kind of valuations. Lastly, this research is unique since it explores a mechanism underlying the value enhancing effect of products, that is, the influence of feelings of competence. Thereby, it is researched whether the adverse effect of higher intensities can be averted by making the goal to feel competent more apparent. Taken together, this research provides new scientific knowledge and useful information for practitioners to generate profitable customers.
2. Literature Review
2.1. The Concept Co-Production
Nowadays, a multitude of companies let customers co-produce goods and services. This perspective is rather new and differs from the traditional transactional view in which
firms produce the final products and services which customers then can acquire (Etgar, 2008). In co-production, customers are actively engaged in the creation of the core offering itself within parameters defined by the organisation. In this sense, customers perform part of the traditional functions of the organisation themselves (Lusch & Vargo, 2006; Prahalad & Ramaswamy, 2000) while the institution takes the role of a facilitator providing the toolkits or devices (Troye & Supphellen, 2012). According to this definition, customers can be co-producers of both goods and services (Atakan, Bagozzi, & Yoon, 2014; Auh, Bell, McLeod, & Shih, 2007; Etgar, 2008). The co-production of products is called ‘self-production', and the process of co-producing services is called ‘self-service' (Atakan et al., 2014; Auh et al., 2007; Etgar, 2008). "Self-production is defined as the active engagement in the creation of end products by consumers" (Atakan et al., 2014, p. 39; e.g., watches, furniture). "Self-service is defined as the customer performing all aspects of a specific service encounter. Self-service in its purest form does not involve any assistance from service firm employees" (Meuter, Bitner, Ostrom, & Brown, 2005; e.g., self-serving restaurants, gas stations). In this study, I focus on the co-production of goods, so on the ‘self-production' process.
Self-production can be further subdivided into self-assembly tasks in which products are produced whereby the target outcome is rather fixed (e.g., ready-to-assemble furniture, dinner kits) and into self-design tasks whereby products are mass customised (e.g., design your shoes, skis, watches; Franke et al., 2010; Norton et al., 2012). The former reflects the so-called ‘IKEA-effect' (Norton et al., 2012), named after the Swedish manufacturer of self-assembled furniture. The IKEA-effect pertains the increase in valuation of products that are ‘self-assembled', compared to objectively identical products assembled by others (Norton et al., 2012). Where the latter reflects an ‘I designed it myself' effect, such an effect pertains an increase in valuation of ‘self-designed' products (Franke et al., 2010). This interface is known as a mass customization toolkit. So, with mass customization kits, customers can customise
the design of the end product, wherein the previous example the design is definite. In this study, I focus on the IKEA-effect so on products that need to be
self-assembled. The focus will not be on products that need to be self-designed because in those instances effort and time investments might also influence preference fit with the outcome (Franke et al., 2010). Meaning that when customers can customise products to their tastes, the product is more likely to meet their needs. Besides, with mass customization toolkits, a truly unique asset can be produced resulting in a higher gained utility (Franke & Schreier, 2010). Thirdly, customers may derive utility from participating in mass customization because they find it enjoyable (Franke & Schreier, 2010). Process enjoyment represents a customer's enjoyment experienced during the co-production process; it is a subjective reaction that contributes largely to the value customers derive from self-production (Franke & Schreier, 2010). Lastly, in self-design tasks customers can observe and detect their contribution to the design solution, the customization is visually traceable. With self-assembling, the outcome quality is held constant. Therefore the observed effect of self-production cannot be attributed to observed customization but is rather the result of psychological processes created by the self-assembling process. By focusing on the self-assembling process, those confounding effects can be ruled out, thereby providing a clean test on the influence of intensity on valuation.
2.2. The IKEA-Effect
This research will focus on the IKEA-effect, so on self-assembled products resulting in outcomes that are rather fixed, meaning that the quality of the result is held constant (Norton et al., 2012). Research has shown evidence of the IKEA-effect. Different products have been used to measure this effect. Among other things, it is proven that self-created milkshakes (Dohle et al., 2014 ), LEGO cars (Mochon et al., 2012), IKEA boxes and Origami
(Norton et al., 2012) were valued more compared to a similar product created by someone else. In those studies, the maximum price people were willing to pay (i.e., WTP) for the product was used to conceptualise value (Wertenbroch & Skiera, 2002). So, even for products that are not unique, customised or fun to build (as is the case with the ‘I design it myself' effect) the effect of labour on valuation holds, resulting in a higher WTP (Norton et al., 2012).
In line with earlier research regarding the IKEA-effect, it is expected that self-assembled products, compared to standardised firm productions, are valued economically higher resulting in an increased WTP (Mochon et al., 2012; Norton et al., 2012). Next, to this economic implication, two potential psychological responses are included as well:
satisfaction and loyalty towards the organisation. The former is the overall satisfaction a customer experiences with the company. The latter is a "deeply held commitment to rebuy or re-patronize a preferred product or service consistently in the future" (Michels & Bowen, 2005, p.6). In this study, customer loyalty refers to the customers' re-visit and
recommendation intentions. Those variables are taken into account because customers who are loyal and satisfied towards an organisation are more likely to become regular customers and thus would lead to improved company performance, the same holds for WTP (Skogland & Siguaw, 2004).
Research has shown a strong positive relationship between customer satisfaction and WTP, whereby customers who were willing to pay more for products were also more
satisfied with the firm (Bendapudi & Leone, 2003; Finkelman, 1993; Homburg, Koschate, & Hoyer, 2005). Also, a positive and vigorous relationship was found between loyalty and WTP, whereby people who were more loyal towards the organisation were significantly willing to pay more for the products of this company (Meuter et al., 2005; Stokburger-Sauer et al., 2016). Based on those results it is expected that people who self-assemble products are
not only willing to pay more for the self-produced product but are also more satisfied and loyal towards the organisation providing self-produced goods compared to an organisation providing standardised finished goods. In line with this reasoning the following Hypotheses are set:
H1a: Self-assembling products result in a higher WTP compared to the obtainment of an
identical finished product.
H1b: Self-assembling products result in a higher satisfaction level towards the organisation
compared to the obtainment of an identical finished product.
H1c: Self-assembling products result in a higher loyalty level towards the organisation
compared to the obtainment of an identical finished product.
2.3. Equity Theory
None of the studies that researched the IKEA-effect looked at the consequences of increasing customer inputs within self-assemble situations. Therefore, the central tenet of this study is to address this neglect by investigating how people's required input in
self-assembling tasks affects their WTP, satisfaction and loyalty. I expect that the positive effect of self-assembling on valuation has boundary conditions, wherein higher self-assembling intensity not necessarily result in a higher valuation. Varying levels of intensity are expected to influence valuation differently. The equity theory is used as an overarching theoretical framework to derive a conceptual model that investigates the role of self-assembling intensity on customer's WTP, loyalty and satisfaction compared to a standardised alternative.
The equity theory holds that people strive for a fair distribution of outcomes and inputs between both parties in exchange (e.g., buyer-seller exchange relationships; Adams, 1963). An exchange is perceived as fair when the inputs devoted to achieving an output are
equivalent between both exchange partners (Cook & Emerson, 1978). When a person believes his outcome/input ratio is unfavourable or not equal to the exchange partner, inequity results (Pritchard, 1969). Series of marketing research has shown that equality is important to predict customer behaviour, resulting in, higher payments, loyalty and
satisfaction (Adams, 1963; Haumann et al., 2015; Huppertz, Arenson, & Evans 1978; Oliver & Swan 1989; Swan & Oliver, 1991; Tax, Brown, & Chandrashekaran 1998). Consequently, organisations should make sure customers perceive the outcome/input ratio as fair.
When using the equity theory in the context of self-assembling situations, customers' total input consists of monetary and non-monetary costs (Etgar, 2008). Monetary costs comprise the price customers need to pay for the firm-provided product. Non-monetary costs are the time, and effort consumers must bear to receive the product (Seiders et al., 2002). The outcome is the perceived value customers derive from the self-assembling process, such as economic value by getting a lower price or subjective value by having a positive experience. Conversely, input for the company are all required costs to bring a product to the market, and the outcome is attitudinal and financial performance (Etgar, 2008; Haumann et al., 2015). Financial performance is the price paid by customers. Whereas, attitudinal performance are among other things customers' satisfaction and loyalty levels towards the organisation.
In this research, the focus will be on the customers’ non-monetary costs. Specifically, the focus will be on time and effort requirements to finish a product which together
represents intensity. Intensity is defined as customers' subjective perception of the extent of effort and time invested in the process of assembling a product (Haumann et al., 2015). Time is conceptualised as the required time to finish the product (Anderson & Shugan, 1991), and effort as consumers' energy expenditures (Seiders et al., 2002). According to the equity theory, the more effort and time is required to assemble products the higher customer's total input into the self-assembling process which might diminish the favorability of the
outcome/input ratio (Oliver & Swan, 1989; Pritchard, 1969). So, higher intensity enhances the chance of inequality.
Accordingly, in line with research on the IKEA-effect, a positive relationship between self-assembling intensity and valuation is predicted compared to a standardised firm
production but only up to a certain point. Beyond this point, increasing self-assembling intensity exerts an adverse effect because customers sense an unequal distribution of resources (Haumann et al., 2015; Norton et al., 2012; Stokburger-Sauer et al., 2016). So, when the intensity gets too high cost will outweigh the benefits associated with the self-assembling process as shown in the IKEA-effect, resulting in a lower valuation. This valuation is expected to be even lower compared to an identical finished product because people receiving a finished product do not experience high effort and time costs thereby creating a greater degree of equality as the product does not have to be assembled. The next section explains a particular benefit that might account for the increase in valuation of self-assembled products. Based on the equity theory I propose the following:
H2a: The effect of self-assembling on WTP differs for different levels of assembling
intensity; medium intensity > low intensity > control1 > high intensity.
H2b: The effect of self-assembling on satisfaction towards the organisation differs for
different levels of assembling intensity; medium intensity > low intensity > control > high intensity.
H2c: The effect on self-assembling on loyalty towards the organisation differs for different
levels of assembling intensity; medium intensity > low intensity > control > high intensity.
2.4. Theoretical Mechanism Underlying Valuation
Researchers have shown interest in uncovering the underlying mechanisms of the IKEA-effect. In line with earlier research, this study focusses specifically on a mechanism underlying the elevation of product valuation. Therefore, this sections continues on the dependent variable WTP because this variable represents product valuation. One of the mechanisms that seem to increase the valuation of self-assembled products is the feeling of competence, which stems from the self-determination theory (Deci & Ryan, 2000, 2011; Mochon et al., 2012). This theory is used to uncover the relationship between higher product valuation and self-assembling intensity. Examining if feelings of competence contribute to the effect of self-assembling intensity on WTP.
2.4.1. Self-determination theory
The self-determination theory states that humans have an innate psychological need to feel competent. To feel competent people are willing to successfully produce desired
outcomes in one's environment (Eccles & Wigfield, 2002; White, 1959). Fulfilment of this need is essential for facilitating optimal functioning and personal well-being (Deci & Ryan, 2000, 2011). Therefore, people strive for a competent self-identity, which is in line with the self-affirmation theory which states that people seek to keep a positive view of themselves, which drives self-motivation and subsequent behaviour (Aronson, Cohen, & Nail, 1999; Sherman & Cohen, 2006).
People acquire a competent identity by looking at their behaviour, and the behavioural outcome as this gives information about their competence (e.g. successful fulfilment of a task provides information about a competent identity). Performing the right behaviour is therefore important. Research has shown that actions in which objects and possessions are affected and controlled accomplish a competent identity (Ahuvia, 2005; Belk, 1988; Furby, 1991; Deci &
Ryan, 2000). With higher levels of control, the outcome will be more experienced as part of the self. As a result, the product is attributed more to own accomplishments, thereby
satisfying the need for competence (Deci & Ryan, 2000, 2011). Consequently, people associate feelings of competence with the product, which then functions as a signal of competence to the self, leading to improved evaluations (Mochon et al., 2012).
In self-assembling processes, organisations exert control to customers as they are in charge of assembling the end-product themselves. I expect that with increasing
self-assembling intensities more control will be experienced, resulting in higher feelings of being the originator, which raises customer valuation of the creation because those self-assembled products act as a signal of competence to the self (Chang, Chen, Huang, 2009; Mochon et al., 2012).
In line with the equity theory, feelings of competence associated with the product are perceived as a benefit which improves the input/output ratio (Deci & Ryan, 2000). So, this advantage can offset increasing costs of assembling intensity resulting in improved
evaluations of these products, so in a greater WTP (Bandura, 1977; Franke et al., 2010; Franke & Schreier, 2010; Loewenstein & Issacharoff, 1994; Mochon et al., 2012). However, as stated, I do expect boundaries to the extent in which intensity increases valuation.
The self-determination theory can be used to explain the limits of self-assembling intensity. This theory argues that tasks that contribute towards feelings of competence enhance intrinsic motivation to accomplish those tasks (Deci & Ryan, 2000). Studies have shown that people have greater intrinsic motivation for tasks requiring a higher intensity to complete (Mochon et al., 2012; Goa et al., 2009). However, an optimal level of efficacy is expected; when the intensity gets too high, intrinsic motivation declines (Chang et al., 2009; Meuter et al., 2005). Under such condition, owing to the self-serving bias, customers are less likely to consider the outcome because of their effort which decreases people’s feelings of
ownership of the product resulting in a lower valuation of the end product as feelings of competence are not linked with the product (Chang et al., 2009; Forsyth, 2008; Meuter et al., 2005). It has been proven that easier tasks resulted in increased valuation with the outcome, where higher intensities fail to reach the same level (Chang et al., 2009). So, higher
intensities decrease people’s sense of ownership of the product lowering their WTP. Taken those results together, consistent with previous research I predict that competence mediates the influence of self-assembled intensity on WTP (Mochon et al., 2012). However, differences are expected in WTP. With an ascending intensity (from control to low to medium) the feeling of being the originator of the product is enhanced, which raises customer valuation of the product as those self-assembled products act as a signal of
competence to the self. On the other hand, a high intensity impedes people’s sense of ownership, precluding the link between competence and the product. Consequently costs might outweigh and inequity results, leading to a lower WTP. The WTP is expected to be even lower compared to when a finished product is obtained because less time and effort requirements ar needed with a finished item. Based on this the following hypothesis is set:
H3: Feelings of competence mediate the effect of self-assembling intensity on WTP.
2.4.1.1. Goal relevance
As mentioned in the previous section, I expect that when the self-assembling intensity gets too high, intrinsic motivation goes down resulting in a decreased WTP. Many products that need to be self-assembled might request a lot of time and effort from customers which as a consequence might lead to a lower valuation. Therefore, it is interesting to investigate how to reduce this adverse effect. This section discusses a possible value enhancing tool which could restore the negative effect of higher intensities on WTP. Specifically, this study will
research whether making a high-intensity self-assembling task an important goal can overcome the dampening effect of high-intensity on valuation.
An encounter resulting in failure or ineffectiveness would impinge on the need for competence (Schwarzer, 2014), in such a case a person experiences a low sense of
competence. When people’s feelings of competence are low, intrinsic motivation is created to enhance those feelings again because of people’s innate psychological need to feel competent (Gao, Wheeler, Shiv, 2009; Mochon et al., 2012). To restore those feelings people prefer goals of higher intensities because those goals are relevant for goal pursuit, increasing the positive valence of that goal (Schüler, Sheldon, & Fröhlich, 2010; Sheldon, Elliot, Kim, & Kasser, 2001). In contrast, people who experience a high feelings of competence no such goals of increasing competence are accessible, in such a case lower task intensities will improve evaluation (Mochon et al., 2012; Schwarz, 2004). Therefore, people to whom the high-intensity self-assembling tasks are more relevant for goal pursuit, as is the case when people experience a low sense of competence, are more likely to upgrade outcome
evaluations as a result of self-assembling.
Research has already shown that thwarting feelings of competence induce consumers to choose consumption experiences that require more time and effort by making the goal to feel competent relevant (Gao, Wheeler, & Shiv, 2009; Gibbs & Drolet, 2003; Mochon et al., 2012). Until today, no research investigated to which extent people come to value the process of assembling products when the sense of competence is low. I expect that for individuals induced with low feelings of competence the goal of becoming competent again becomes more relevant resulting in a higher valuation of products with ascending levels of intensity compared to people to whom the relevance of this goal is low as a result of high experienced feelings of competence.
This value enhancing effect is expected because first of all greater goal pursuit relevance should lead to stronger links between the self and the outcome (Deci & Ryan, 2000). Consequently, as mentioned in the previous section, the experienced feelings of
competence will be linked to the product resulting in a higher valuation (Deci & Ryan, 2000). Secondly, higher goal relevance of the outcome may make positive aspects of the product more accessible than negative aspects (Ferguson & Bargh, 2004). Thereby reducing the inequity because the benefits become more apparent resulting in a higher WTP.
Accordingly, I expect that inducing people with the goal to feel competent can overcome the dampening effect of high self-assemble intensity on valuation because higher intensities contribute more to the goal of getting a competent identity. To feel competent again higher intrinsic motivation towards the high-intensity task is established, consequently, feelings of competence after having assembled the product are linked to the product resulting in higher product valuation (Mochon et al., 2012). Thus, the effect of self-assembling
intensity on the WTP depends on whether the perceived intensity provides information to the motivational system regarding usefulness and effectiveness to reach the goal (Labroo & Kim, 2009). Eventually, resulting in a higher WTP of the product (Franke & Schreier, 2010). So, the detrimental effect of higher self-assemble intensity becomes positive, even more positive than the medium self-assemble intensity, as it contributes more to the goal of feeling
competent again (Stokburger-Sauer et al., 2016). Therefore, the following hypothesis is set:
H4: Goal relevance moderates the influence of self-assembling intensity on WTP.
Specifically, people inflicted with a low (high) sense of competence representing high (low) goal relevance express a higher (lower) WTP with high self-assembling intensities.
Specifically, goal relevance is expected to moderate the mediating effect of feelings of competence (James & Brett, 1984; Preacher, Rucker, & Hayes, 2009). Whereby people’s intrinsic motivation to assemble products requiring higher intensities go up when inflicted with high goal relevance, thereby reducing the negative effects of higher intensities as explained in the previous section. So, in those instances, higher levels of self-assembling intensities increase people’s feelings of competence associated with the product resulting in even higher valuations compared to a control, low and medium-intensity tasks. Therefore, the mediating influence of competence on the value enhancing effect of self-assembling intensity depends on goal relevance. Resulting in the following hypothesis:
H5: Goal relevance moderates the mediating effect of competence, where with higher goal
relevance the adverse effect of self-assembling intensity on customer WTP becomes positive: high intensity > medium intensity > low intensity > control
2.5. Conceptual Model
Figure 1 displays the conceptual model and hypothesis that has been drawn up in the literature review.
Figure 1. Conceptual model.
H4
H3
H1a-c, H2 a-c
H5 Self-assembling
intensity (control, low, medium, high)
Feelings of Competence
Valuation (WTP, loyalty, satisfaction)
Goal relevance (high, low)
3. Method
3.1. Sample
An online experiment was conducted to answer the research questions. The experiment was carried out via Qualtrics2, taking the form of an online survey. Thereby, reducing the geographical limitations of recruiting respondents. As long as people spoke English or Dutch they could participate because the study was offered in those two languages; people could select their preferred language. Participants were recruited through the snowball sampling3 method as the selection was based on the pre-existing network of the researcher (Neuman, 2016). This network was reached via online posts on Facebook and LinkedIn (using personal and private messages), emails and by word of mouth. All subjects participated voluntarily in return two participants had the chance to win 25 euro. In total 407 individuals participated in this study of which 387 participants were used for further analysis, for more detailed
descriptives of the sample see the result section.
3.2. Design
The design of this study was a 4 (self-assembling intensity: control, low, medium, vs. high) by 2 (goal relevance: low vs. high) between-subjects design. Participants were randomly assigned to one of the eight conditions. Those conditions were compared on the dependent variables; WTP, loyalty and satisfaction and the mediating variable competence. Thereby, examining the effects, there was controlled for the variables: likability, preference fit, self-efficacy, task enjoyment and hypothetical WTP.
2 Qualtrics provides online survey software tools and solutions (Qualtrics, 2016).
3 Snowball sampling also called, network, chain-referral, or reputational sampling is a special technique in
3.3. Procedure
Data collection took place from the 15th till the 26th of May in 2017. Starting the online experiment, participants got the informed consent, when subjects agreed on the consent the experiment started. Qualtrics randomly assigned participants to one of the eight conditions. The experiment was divided into three parts. In the first part mathematical problems needed to be solved, which manipulated participants goal relevance. Participants were randomly assigned to the high or low goal relevance condition. In the second part, participants were at random exposed to one of the four scenarios manipulating self-assembling intensity. After reading the scenarios, either the low, medium, high self-assemble intensity condition or control condition, participants were solicited to their WTP, satisfaction, loyalty, competence, and intensity. To avoid common method bias, items were randomised within the
questionnaire. The sequence of the items was intentionally designed to be opposite to the causal direction of the hypotheses to minimise any possible demand effect (Grissemann & Stokburger-Sauer, 2012). Specifically, respondents first had to answer questions on their WTP, satisfaction, and loyalty. After that, the survey included items to assess feelings of competence and perceived intensity followed by the control questions. The study closed with a few demographic questions. After collecting those data participants were thanked for participating in the study.
3.4. Stimulus Materials
The following section clarifies how the independent variables were manipulated in this study. Starting with the manipulation of goal relevance followed by the manipulation of the self-assembling intensity.
3.4.1. Manipulation of goal relevance
For the independent variable goal relevance, Qualtrics randomly assigned half of the participants to a condition in which high goal relevance was stimulated. The other half was stimulated with low goal relevance. Low goal relevance was manipulated by presenting participants with four easy math problems (e.g., “How likely is it that a fair coin that is tossed once will come up heads?”) and high goal relevance was created with four difficult math problems (e.g., “Two fair dice are rolled. What is the probability that at least one die come up a 3?” adapted from Mochon et al., 2012). The questions were presented with five possible answers, and participants were told that they could skip questions if they did not know the answer, see Appendix A (Mochon et al., 2012). The difficult questions would result in high goal relevance as those people’s reduced sense of competence would lead to the creation of the goal to enhance those feelings again, where with the easy questions no such goal was created as participants need for competence was affirmed (Mochon et al., 2012).
Additionally, before answering the questions subjects were told that among the participants who answered most questions correctly two times 25 euro would be distributed. By doing this, participants were encouraged to take the questions seriously resulting in more valid and reliable results.
3.4.2. Manipulation of self-assembling intensity
Subjects were randomly assigned to the control, low-, medium- or high- intensity conditions, via Qualtrics. Consistent with prior research, projective scenarios were used to manipulate intensity (Bendapudi & Leone, 2003). A scenario is a short description of a person, object or social situation representing a systematic combination of characteristics of what are thought to be the critical factors in the decision-making processes of respondents (Alexander & Becker, 1978; Atzmüller & Steiner, 2010). The use of projective scenarios is
well established in the psychology and marketing literature and has been shown to minimize social desirability effects and considerable external validity (Bateson & Hui, 1992; Reeder, Hesson-McInnis, Krohse, & Scialabba, 2001; Robinson & Clore, 2001; Voss, Parasuraman, & Grewal, 1998). Therefore, this technique seems an appropriate manipulation of intensity. In this research, the scenario represents a situational description of a person named ‘Bo’ who assembles a side table (self-assemble condition) or got a standardised firm production of the same side table (control condition; Bendapudi & Leone, 2003). The side table was chosen because it can be assembled by one person, as appeared mandatory for the IKEA-effect to occur (Norton et al., 2012). Subjects were asked to put themselves in Bo’s shoes and indicate how they thought Bo would respond in the described situation. The name Bo was chosen to be gender-neutral so that both male and female participants could identify with the character. As mentioned is this technic evaluated as a useful technique. Even when participants adopt an “observer” stance (i.e., analysing Bo from a distance) instead of an “actor” stance (i.e., in which they put themselves in Bo’s shoes), the scenario still represents a conservative test of the hypothesis by reducing the self-serving bias in such a situation (Jones & Nisbett, 1971).
The scenarios were constructed to create a control and self-assembling condition. The latter was further divided into three intensity levels; low, medium, and high. As discussed in the literature review, intensity is defined as customers’ subjective perception of the extent of effort and time invested in the process of self-assembling a product (Haumann et al., 2015). Therefore, different levels of intensity were created by manipulating the required time and effort to complete the assembling task. Higher levels of effort were induced by increasing the necessary steps and fasteners to assemble the product (Johnson & Payne, 1985). Time was manipulated by mentioning how many minutes it took to assemble the table. Furthermore, the different scenarios contained equal amounts of information and a similar format, to make
sure people perceived the task as equally challenging. See Appendix B for an overview of the four scenarios. The different levels of intensity for the scenarios were pre-tested to check whether the manipulation was perceived as expected, the pre-test is discussed in the following section.
3.5. Pre-Test
A pre-test was executed in Qualtrics to test the effectiveness of the self-assembling intensity manipulation. When participants perceive the high-intensity condition as the most intense followed by the medium-condition and then the low condition a proper manipulation is proven. To find those three diversifying scenarios four scenarios were constructed
representing four levels of intensity: low, medium, high, very high. With ascending time and effort requirements representing increased intensities (Haumann et al., 2015; Johnson & Payne, 1985). Three scenarios were selected that appeared to differ significantly from one another in the prescribed direction.
3.5.1. Sample and procedure pre-test
Participants were recruited online through Facebook and LinkedIn, email and by word of mouth. The final sample contained 91 participants of which 38 males (41.76%) with an average age of 26.63 years (SD = 1.26) and 53 women (58.24%) with an average age of 25.74 years (SD = .64). Starting the online survey participants were first given an informed consent, followed by the scenario. Participants were randomly assigned to one of the four scenarios via Qualtrics. After participants had read through the scenario, they were requested to answer questions testing the manipulations effectiveness. Then, some demographical questions were asked. Lastly, participants were thanked for taking part in the pre-test.
3.5.2. Manipulation check pre-test
The final sample was analysed for the extent to which the task was seen as challenging, believable, and relevant. Most importantly, the scenarios were checked for perceived
intensity.
Firstly, participants in the four manipulation conditions saw the task as equally
challenging with an average rating of 3.27 (SD = 1.57), F (3, 87) = 1.21, p = .31, partial η² = .040, measured on a seven-point Likert-scale (1 = not at all challenging and 7 = very
challenging). Conforming the expectation, that only intensity was manipulated. Secondly, the scenario was perceived as equally relevant in all conditions with an average score of 4.01 (SD = 1.62), F (3, 87) = .35, p = .79, partial η² = .012, measured on a seven-point Likert scale (1 = not at all relevant and 7 = very relevant). Thirdly, participants also believed the scenario with an average score of M = 4.64 (SD = 1.55), F (3, 87) = 2.31, p = .082, partial η² = .074,
indicating a medium effect, measured on a seven-point Likert scale (1 = not at all believable, 7 = very believable).
As follows, the intensity was verified using five items, see paragraph 3.6.1. for the items specifications. The Cronbach’s alpha for the 5-item Intensity questionnaire was .88, which is adequate for research purposes (Field, 2009). Therefore, the items were aggregated into one construct representing intensity. A one-way between groups analysis of variance (ANOVA) was used to compare the average intensity score of the four intensity conditions. A large significant effect was found of condition on perceived intensity, F (3, 87) = 9.71, p < .001, partial η² = .251. Post hoc analyses revealed that participants in the low intensity condition (M = 2.03, SD = .69) perceived the intensity as significantly lower than participants in the medium intensity condition (M = 2.82, SD = 1.00), high intensity condition (M = 3.13, SD = 1.08) and the very high intensity condition (M = 3.72, SD = 1.47). Thereby, the medium-intensity condition differed significantly from the very high-medium-intensity condition. The
remaining comparisons were not significant. The effect sizes for these four comparisons were
d = .92, 1.21, 1.47 and 1.15 respectively.
Based on those results, the low, medium and very high-intensity4 conditions were used in the main study because those manipulations turned out to differ significantly from one another in the appropriate direction. Thereby, the different scenarios were perceived as believable, relevant, and equally challenging therefore the pre-tested scenarios were proven to be substantially sufficient for the main study.
3.6. Measures
The items for each construct were selected from existing scales and conceptual papers, modified to fit this study, see Appendix C for an overview of all constructs and used
measurements. First, the dependent variables will be explained, followed by the control and demographic variables.
3.6.1. Dependent variables
Firstly, WTP was measured with a hypothetical direct approach. Subjects were asked: “How much money would you be willing to pay for this side table?” (adapted from Breidert, Hahsler, & Reutterer, 2006; Miller, Hofstetter, Krohmer, & Zhang, 2011). Research has shown that this direct approach generating mean WTP estimates does not significantly differ from real economic behaviour, making this a valid measurement (Miller et al., 2011).
After participants had made their bid, they had to answer one question about their satisfaction with the company: “How satisfied are you overall with the firm which delivered you the side table? Overall I am…”; measured with a 7-point Likert scale (1 = dissatisfied and 7 = satisfied, adapted from Bendapudi & Leone, 2003; Haumann et al., 2015; Voss,
Parasuraman, & Grewal, 1998). This single-item was used to keep the total time to complete the survey reasonable and to prevent respondent fatigue. Thereby, this one questions seems to be valid for the satisfaction construct (Bendapudi & Leone, 2003).
Then, loyalty towards the company was measured by using three items (α = .90, e.g., “I am likely to buy furniture from this company again in the future”; measured with a 7-point Likert scale: 1 = strongly disagree and 7 = strongly agree, adapted from Stokburger-Sauer et al., 2016).
Next, perceived feelings of competence were measured. Competence is synthesised in one construct: personal feelings of pride. As studies have shown that pride is closely linked to feelings of competence (Mochon et al., 2012). Just like competence, pride is important for humans (Tracy & Robins, 2004) and therefore strongly linked to consumers’ evaluation of their identity (Buss, 2001). Therefore, competence was measured with three items (α = .91); those were designed to measure personal feelings of pride associated with the (self-produced) side table: “When I look at the side table I have assembled…” (self-assemble condition; control condition: “...I have ordered”; e.g., “The feeling I have can best be described by the word ‘pride’”; measured with a 7-point Likert scale: 1 = strongly disagree and 7 = strongly agree, adapted from Franke et al., 2010).
Lastly, perceived process intensity was measured with five items (α = .94), the same items were used in the pre-test, “I perceive the process of assembling the product as” (self-assemble condition; control condition: “putting the product in place as…” “effortful”, “demanding”, “exhausting”, “time-consuming”, and “costly (in terms of time and effort)”; measured with a 7-point Likert scale: 1 = strongly disagree and 7 = strongly agree, adapted from Dellaert & Stremersch, 2005; Franke & Schreier, 2010; Franke et al., 2012; Haumann et al., 2015; Troye & Supphellen, 2012).
3.6.2. Control and demographic variables
To test the robustness of the proposed relationships and to control for extraneous influences, control variables were included in this research.
Firstly, preference fit is taken into account. As mentioned in the literature review it is expected that with self-design processes preference fit goes up, resulting in higher valuations (Franke et al., 2010, Franke & Schreier, 2010). To preclude a clean test of self-assembling on valuation is provided this variable is taken into account. Preference fit of the side table is measured with three items (α = .89, e.g., “I like the design of the side-table”; all three items measured on 7-point scales: 1 = strongly disagree and 7 = strongly agree; Franke et al., 2010; Franke & Schreier, 2010).Secondly, I control for the effect of customers’ perceived task enjoyment derived from a particular self-assembling activity as this factor seems to be a major determinant of valuation (Franke et al., 2010). Those positive feelings during self-assembling should overwhelm the adverse effects due to higher self-assemble intensity. Therefore, this study control for this variable. Task enjoyment was measured with four items (α = .94, e.g., “I enjoy assembling furniture”; measured with a 7-point Likert scale: 1 = strongly disagree and 7 = strongly agree, adapted from Franke & Schreier, 2010). Thirdly, self-efficacy is taken as a control variable. Self-efficacy might influence valuation. Research has shown that people high in self-efficacy prefer to perform more challenging tasks (e.g., self-assembling products) which could therefore influence the valuation process (Stokburger-Sauer et al., 2016). This variable was measured with three questions (α = .93, e.g., “I am fully capable of assembling furniture”; measured with a 7-point Likert scale: 1 = strongly disagree and 7 = strongly agree, adapted from Dong et al., 2008). Furthermore, likeability was taken into account because when people like the side table this would automatically result in higher valuation (Franke et al., 2010). Likeability was measured with one item (“I like the side table”; measured with a 7-point Likert scale: 7-point Likert scale: 1 = strongly disagree and 7
= strongly agree, adapted from Franke et al., 2010). Lastly, people’s average hypothetical WTP for the product category was taken into account, as this might influence people’s actual WTP for the side table. This variable was measured with one question (“How much would you normally pay for a side-table of comparable quality? Please indicate the amount of money in euros”, adapted from Franke et al., 2010). By controlling for all those variables, a clean test of the proposed effect is provided.
In line with the pre-test, the scenario was again checked for (a) believability (“How believable was the scenario?”; 7-point Likert scale: 1 = not at all believable and 7 = very believable, adapted from Bendapudi & Leone, 2003); (b) realism (“How realistic was the scenario?”; 7-point Likert scale: 1 = not at all realistic and 7 = very realistic, Bendapudi & Leone, 2003); and (c) challengingness (“To which extent do you think the task in the scenario was challenging?”; 7-point Likert scale: 1 = not at all challenging and 7 = very challenging, Franke et al., 2010).
Finally, the questionnaire ends with demographic characteristics such as sex, age, highest education, income and country of residence and origin.
3.7. Statistical Procedure
Data was collected in Qualtrics. For the analysis, all collected data were downloaded from Qualtrics and exported into the statistical analysis software IBM SPSS Statistics 24. Confidence intervals were set on a 95% interval (Field, 2009). Before testing hypotheses, six preliminary steps were taken. First, data were checked for outliers and extreme scores; those people were excluded from further analysis. Second, descriptive data about the sample were given. Third, frequencies were tested for each item, to identify missing values. Fourth, it was checked whether the manipulations were successful. Fifth, scale reliabilities were assessed for each scale via the Cronbach’s Alpha. Finally, a correlation matrix was provided in which
a summary of variable means, standard deviations, correlations and reliability statistics were given. After those steps the hypotheses were tested, using one-way analysis of covariance (ANCOVA; Allen & Bennet, 2010).
4. Results
For all of the following analysis, a partial eta-squared (ƞ²) of .01 is considered as small, a partial ƞ² = .059 as moderate, and a partial ƞ² = .138 as large (Cohen, 1988). Thereby, Cohen (1988) suggested that as a general rule of thumb, a Cohens d of .20 can be
acknowledged as small, a d of .50 as moderate, and a d of .80 as large. Furthermore, a
correlational coefficient (r) of .10 is recognised as small, a r of .30 as moderate, and a r of .50 as large (Cohen, 1988). Lastly, all effects are reported as significant with a p-value smaller than .05 (Field, 2009).
4.1. Descriptive Data of the Sample
The initial sample size of the experiment consisted of 407 participants. Of this sample, 18 people were excluded from further analysis. Of these exclusions, 17 participants were excluded because of extreme scores on the required time needed to complete the task. Meaning, those people had a duration score which deviated more than three standard
deviations from the average time people took to complete the study (Field, 2009). This long time interval might be problematic because people possibly did not finish the different tasks after each other which could have created distorted results (Allen & Bennett, 2010; Field, 2009). Thereby, one participant was excluded because this person indicated to be willing to pay 850 euro for the side table, this score deviated more than three standard deviations from the average WTP, resulting in distorted results (Allen & Bennett, 2010; Field, 2009).
The final sample consisted of 389 participants, with a mean age of 30.04 years (SD = 13.16), of which 262 women (67.4%) and 127 men (32.6%). Women had an average age of 29.20 years (SD = .77) and men an age of 31.78 (SD = 1.27). The ages ranged between 14 and 76 years. Furthermore, most respondents were Dutch (89.7%) and lived in the
Netherlands (91.3%). Additionally, the sample mostly covered participants with a higher educational background, 83% of the participants had a university bachelor degree or higher. However, although the level of education was predominantly high, the average income did not follow this pattern. Few participants (39, 9.2%) did not want to disclose their yearly income. Of those participants disclosing their income, the average income was between the ten and thirty thousand euro, 268 participants (75.9%) indicate to earn less than 30.000 euro a year. Specifically, 39.70% of the respondents indicate to earn “less than €10.000”, 22.9% indicate to earn between the “€10.000-€19.999”, and 13.3% indicate to earn “€20.000-€29.000” euro this low income might be caused by the many students participating in the sample.
From the final sample, all variables were checked for missing data using frequency test which resulted in 0% missing data. No missing values were observed because of the utilisation of the option ‘forced choice’ for all questions in the Qualtrics.
4.2. Reliability of Scales
Next, the internal consistency of the different measurement constructs was assessed by computing the Cronbach’s alpha. As a general rule, a Cronbach’s alpha between .7 and .8 is acceptable. However, values above .8 are preferred for research purposes (Allen &
Bennett, 2010; Field, 2009). The Cronbach’s alphas, means and standard deviations for the different constructs are shown in Appendix C. Overall, the Cronbach’s alpha for all
the scores of the different constructs were taken together, and all subsequent analysis were based on those aggregated constructs.
4.3. Manipulation Check
In this section, it is examined whether the manipulations of the independent variables were successful. First discussing goal relevance, then intensity is taken into account, ending with the degree to which the scenario was seen as challenging, realistic and believable.
4.3.1. Goal relevance
First, it was examined whether the goal relevance manipulation was successful. So, if participants in the high goal relevance condition answered the difficult questions wrong the manipulation is seen as successful as this would result in a low sense of competence which increases the goal to feel competent again (Mochon et al., 2012). The manipulation of the low goal relevance condition is seen as successful when the easy questions were answered
correctly because this would result in a high sense of competence which would decrease the goal to feel competent. Participants in the low goal relevance condition solved on average 87.16% of the questions correctly, whereas participants in the high goal relevance condition answered on average 20.27% of the questions correctly, which is no better than chance (Mochon et al., 2012). Therefore, the goal relevance manipulation seems to be successful.
4.3.2. Self-assembling intensity
Then, the self-assembling intensity manipulation was verified. A one-way between groups ANOVA was used to compare the average intensity score of the four conditions. A significant strong effect was found, indicating that perceived intensity is influenced by the different conditions, F (3, 385) = 53.54, p < .001, partial η² = .417.
To see which conditions differed significantly from each other post hoc tests were executed. Post hoc analyses revealed that participants in the control group (M = 1.82, SD = .88) perceived the intensity as significantly lower than participants in the low intensity condition (M = 2.98, SD = 1.18), medium intensity condition (M = 3.06, SD = 1.30) and the high intensity condition (M = 4.01, SD = 1.40). Thereby, the low and medium intensity conditions differed significantly from the high-intensity condition. Contradictory to the pre-test, participants in the low and medium intensity condition did not differ in their perceived level of intensity. The effect sizes for these six comparisons were d = 1.03, 1.04, 1.87, 0.65, 0.73 and .06 respectively. Because no differences were observed between participants in the low and medium intensity condition, the intensity manipulation has not been completely successful.
4.3.3. Challengingness, believability and reliability
Next, it was checked if participants saw the given scenario as challenging, realistic, and believable (Bendapudi & Leone, 2003; Franke et al., 2010). Starting with whether the task was seen as challenging. Overall, results revealed that participants did not see the task as very challenging. Scores ranged from 1.78 to 3.82, with an average rating of 2.81 (SD = 1.56). A one-way between groups ANOVA was used to compare the mean challenging score of the four conditions. The ANOVA was statistically significant, indicating that different conditions rated the task not as equally challenging, F (3, 385) = 36.70, p < .001, partial η² = .286. Post hoc analyses revealed that participants in the control group (M = 1.78, SD = 1.22) perceived the scenario as significantly less challenging compared to participants in the low (M = 2.67,
SD = 1.23), medium (M = 3.07, SD = 1.44), and the high-intensity condition (M = 3.82, SD =
1.59). Thereby, did the low and medium-intensity condition differ significantly from the high-intensity condition. Participants in the low and medium intensity condition did not differ
in the extent to which they saw the scenario as challenging. The effect sizes for these six comparisons were d = .73, .97, 1.44, 0.81, 0.49 and .30 respectively. Those results are contradictory to the pre-test in which participants saw the task as equally challenging.
Next, the scenarios were tested for believability and realism. Believability received average ratings from 4.41 to 5.04, and realism received average ratings from 4.41 to 5.35. Based on this it can be assumed that participants in all conditions saw the scenario as realistic and believable (Bendapudi & Leone, 2003).
4.4. Skewness and Kurtosis
Skewness, Kurtosis and the Kolmogorov-Smirnov test were used to verify whether the distribution of scores of the constructs was approximately normal. To test this the aggregated constructs of loyalty and competence were used. The Kolmogorov-Smirnov test and visual inspections of the stem-and-leaf plots showed that none of the constructs was normally distributed, all tests were significant, p < .001. The skewness and kurtosis scores are displayed in Table 1. This table shows that WTP was positively skewed, indicating an
asymmetrical distribution with a long tail to the right. The other three constructs; satisfaction, loyalty and competence were negatively skewed, showing an asymmetrical distribution with a long tail to the left. Thereby, only the constructs WTP and Satisfaction had a high
leptokurtic distribution, as the kurtosis value is bigger than 1 (Allen & Bennett, 2010, Field, 2009). It was examined whether transforming data or changing scores would result in a normal distribution of the constructs (Field, 2009). Those transformations and modifications did not lead to a normal distribution. However, skewness and kurtosis will not make a substantive difference in the analysis when sample sizes are large; containing forty
participants or more per condition (Field, 2009). In this research, every condition contained at least forty participants, so the risk of distorted results was minimised.
Table 1
Skewness and Kurtosis Scores
Skewness SE Kurtosis SE
WTP 1.46 .12 2.93 .25
Satisfaction -.96 .12 1.13 .25
Loyalty -.61 .12 .04 .25
Competence -.71 .12 -.12 .25
Note. n = 389; SE = standard error.
4.5. Correlation Matrix
To assess the size and direction of the linear relationship between the different variables in this research, a bivariate Pearson’s product-movement correlation coefficient (r) was calculated (Allen & Bennett, 2010; Field, 2009). Table 2 provides an overview of the Pearson Correlations between the variables in this research. For this analyses the independent variables (self-assemble condition and goal relevance), mediating variable (competence), dependent variables (WTP, Satisfaction, Loyalty), control variables (hypothetical WTP, likability, preference fit, self-efficacy, and task enjoyment) and demographic variables (gender, age, and income) were included. The analysis presents several significant and relevant correlations between the different variables, notable outcomes are discussed in the following paragraph.
As can be seen from Table 3 the bivariate correlation between self-assemble intensity and all three dependent variables are negative and small to moderate; WTP, r (387) = -.20, p < .001, satisfaction, r (387) = -.28, p < .001, and loyalty, r (387) = -.19, p < .001. Noteworthy, sense of competence does not correlate with any of the variables. Furthermore, a strong positive correlation is observed between loyalty and satisfaction, r (387) = .53, p < .001. WTP has a significant positive but small bivariate correlation with satisfaction, r (387) = .10, p = .05, and loyalty, r (387) = .11, p = .04. As expected WTP was strongly and positively
correlated with the hypothetical WTP, meaning that when people are normally willing to pay more for a side table, they were willing to pay more for it in the experimental condition as well, r (387) = .79, p < .001. Another strong correlation was observed between age and income, r (387) = .61, p < .001. This correlation makes sense because when people grew older, they might have more money due to their jobs. The demographical variables do not have any other strong bivariate correlation with other variables. Therefore, those variables are not taken into account as a covariate when testing the hypotheses. Lastly, some strong
bivariate correlations are observed between the covariates. First, a strong positive correlation is observed between preference fit and likability, r (387) = .83, p < .001. Second, there is a strong positive bivariate correlation between task enjoyment and self-efficacy, r (387) = .57, p < .001. Those covariates are all taken into account in testing the hypotheses.
Table 3
Correlations and Psychometric Properties of Variables
Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1. SA intensity 2.47 1.11 - 2. Goal Relevance 1.57 0.50 -.02 - 3. WTP 103.73 68.75 -.20** -.02 - 4. Satisfaction 5.63 1.02 -.28** .09 .10* - 5. Loyalty 5.10 1.16 -.19** .09 .11* .53** (.90) 6. Competence 4.91 1.49 .37** .04 -.13* .02 .21** (.91) 7. H-WTP 99.93 71.85 -.14** .01 .79** .06 .05 -.10 - 8. Likability 4.25 1.64 -.03 .07 .18** .11* .19** .12* .16** - 9. Preference fit 3.89 1.41 -.04 .05 .11* .11* .22** .13** .10* .83** (.89) 10. Self-efficacy 5.27 1.49 -.07 .00 .00 .24** .14** -.04 .02 .00 .04 (.93) 11. Task enjoyment 3.98 1.67 -.07 -.01 .02 .25** .28** .12* -.01 .15** .15** .57** (.94) 12. Gender 0.33 .47 -.00 -.06 -.01 -.05 -.15** -.21** .04 -.07 -.06 .23** -.06 - 13. Age 30.04 13.16 -.04 .00 .03 -.04 .03 .15** .04 -.30** -.24** .00 -.07 .09 - 14. Income 2.60 2.10 -.02 .01 .05 -.10 -.07 .07 .11* -.23** -.22** .02 -.11* .10 .61** - * p < .05 (2-tailed). ** p < .01 (2-tailed).
Note. n = 389. Cronbach’s internal consistency reliability coefficients appear in parentheses on the diagonal. SA = Self-Assembling, WTP = Willingness to Pay, H-WTP = Hypothetical Willingness to Pay. Gender: 0 = female, 1 = male. SA Intensity: 1 = control, 2 = low intensity, 3 = medium intensity, 4 = high intensity. Goal Relevance: 1 = high goal relevance, 2 = low goal relevance. Yearly income ranging from 1 = “Less than €10.000” to 12 = “More than €150.000”. WTP was measured with an open question how much euro people were willing to pay for the product. Except for SA intensity, Goal Relevance, WTP, H-WTP, gender, age and income, all items were measured on 7-point Likert scales (1 = strongly disagree and 7 = strongly agree).
