the “i designed it myself” effect in mass customization

8/7/2019 The I Designed It Myself Effect in Mass Customization

1/16

MANAGEMENT SCIENCEVol. 56, No. 1, January 2010, pp. 125140issn 0025-1909 eissn 1526-5501 10 5601 0125

informs

doi 10.1287/mnsc.1090.1077 2010 INFORMS

The I Designed It Myself Effect in

Mass CustomizationNikolaus Franke

Institute for Entrepreneurship and Innovation, Vienna User Innovation Research Initiative,WU Vienna University of Economics and Business, A-1090 Vienna, Austria, [email protected]

Martin SchreierMarketing Institute, Department of Management and Center for Research on Organization and Management,

Bocconi University, I-20136 Milan, Italy, [email protected]

Ulrike KaiserInstitute for Entrepreneurship and Innovation, Vienna User Innovation Research Initiative,

WU Vienna University of Economics and Business, A-1090 Vienna, Austria, [email protected]

Many companies offer websites that enable customers to design their own individual products, which the

manufacturer can then produce to order. To date, the economic value of products self-designed using masscustomization (MC) toolkits has been attributed to the two factors of preference fit achieved (which should be ashigh as possible) and design effort (which should be as low as possible). On the basis of literature on behavioraldecision making, we suggest a third factor, namely the awareness of being the creator of the product design.In the course of five different studies, we provide experimental evidence that this I designed it myself effectcreates economic value for the customer. Regardless of the two other factors, self-designed products generate asignificantly higher willingness to pay. This effect is mediated by feelings of accomplishment and moderated bythe outcome of the process as well as the individuals perceived contribution to the self-design process. Thesefindings have important implications for MC companies: It is not enough merely to design MC toolkits in sucha way that preference fit is maximized and design effort is minimized. To capture the full value of MC, toolkitsshould also elicit I designed it myself feelings.

Key words : mass customization; toolkits for user innovation and design; self-design; user design; do it yourself;endowment effect; willingness to pay; psychological ownership

History: Received April 22, 2008; accepted July 23, 2009, by Christoph Loch, R&D and product development.

Published online in Articles in Advance October 16, 2009.

1. IntroductionMany amateur painters hang their pictures on theirwalls, even though the artistic value of such worksmay be questionable in the eyes of others. If one wereto ask them if they would be willing to sell the paint-ings, they would decline or demand exorbitant sums.But if offered another, similar-looking picture createdby another amateur painter, they would hardly beinclined to buy it or would offer only a very small sum.

This is remarkable if the two products are not differentbecause the utility of the process (assuming that theactivity of painting per se generates subjective value)can be considered sunk. But assigning a high subjec-tive value to ones own creations is not a rare case,and it holds for many self-design activities such aspottering, cooking, knitting, building model airplanes,etc. For example, Norton (2009) reports results fromongoing research, where he and his colleagues Arielyand Mochon find that self-folded, amateurish origamiare indeed valued by their originators as highly asorigami made by experts if the labor is considered

fruitful. As Ulrich (2009, p. 10) puts it, A ( ) driverof user design is the utility ( ) users derive from solv-ing their own problems ( ) (I designed it myself!).A user may be willing to accept a lower-quality out-come even at the same cost of expert design. Wedefine the I designed it myself effect as the valueincrement a subject ascribes to a self-designed object,arising purely from the fact that she feels like theoriginator of that object.

This phenomenon may be understood in lightof behavioral decision-making literature that focuseson the many potential biases, heuristics, and otherbounds of rationality impacting actual human deci-sion making in general (e.g., Camerer et al. 2003,Kahnemann and Tversky 2000, Thaler 2000) as well asmany managerial and operations management deci-sions in particular (Bendoly et al. 2006, Loch and Wu2007, Gino and Pisano 2008). This line of research gen-erally maintains that psychological factors play a cru-cial role and subjective attributions sometimes mattermore than objective facts. Research on the endowment

125

Additionalinformation,includingrights

andpermissionpolicies,isavailableathttp://journals.informs.org/.


2/16

Franke, Schreier, and Kaiser: The I Designed It Myself Effect in Mass Customization126 Management Science 56(1), pp. 125140, 2010 INFORMS

effect, for example, suggests that the subjective valuea person attributes to an object is contingent uponwhether she owns the object or not: Goods that areincluded in ones endowment are valued more highlythan identical goods not held in ones endowment(Thaler 1980; Kahneman et al. 1990, 1991). The endow-

ment effect does not rely on legal ownership but isthe result of subjective feelings of ownership (Reb andConnolly 2007), which can vary in intensity. The rea-son for an endowment-like effect in the case of thepainter may be that the person associates feelings ofaccomplishment with the object that arise from theprocess of successfully creating it. Simply buying apicture might lead to a far lower degree of psycholog-ical ownership (Pierce et al. 2003).

The advent of the Internet may well have aug-mented the practical importance of such I designed itmyself effects. Many companies have started to com-mercialize user design by offering websites that enable

customers to design their own individual T-shirts,watches, kitchens, PCs, or sneakers online, which themanufacturer can then produce to order (Dellaert andStremersch 2005, Franke and Piller 2004, Randall et al.2005, Ulrich 2009). This interface between manufactur-ers and customers is known as a mass customization(MC) toolkit, configurator, choice menu, design kit,or toolkit for user innovation and design (Dahan andHauser 2002, Dellaert and Stremersch 2005, Liechtyet al. 2001, Randall et al. 2005, Thomke and von Hippel2002, Ulrich 2009, von Hippel 2001, von Hippel andKatz 2002). These MC toolkits (as we refer to them

throughout this article) dramatically reduce the levelof skill necessary to design a product oneself, as easy-to-use design tools are provided and the intricate pro-cess of physical production is left to the manufacturer.Thus, we can argue on the one hand that I designedit myself effects have the potential to become a massphenomenon.

On the other hand, however, we can also ques-tion the existence of such a value-generating mecha-nism in virtual online design activities with an MCtoolkit. In contrast to physical activities like a paintercreating a picture, a customer building an IKEA book-shelf, a scholar writing a book, an entrepreneur estab-

lishing an organization, or a politician creating a bill(examples taken from Pierce et al. 2003 and Norton2009), designing a product online merely by clicking amouse is not tangible (Peck and Shu 2009). The MCtoolkit provides the user with only simulated feed-back on screen, and the design process might be soeasy that even a novice designer only needs a fewminutes time to create a product she thinks might fither preferences (Franke and Piller 2004, Randall et al.2007). Is this limited role of the originator enough toelicit feelings of accomplishment that may translate

into enhanced subjective ownership and thus also intoan economically relevant effect?

The extant literature on MC toolkits seems to dis-pute this idea. Thus far, research into the reasonswhy products self-designed with MC toolkits maydeliver value to customers and command a price pre-

mium has clearly emphasized the increased prefer-ence fit of the resulting product, that is, the customersassessment of the extent to which the products fea-tures correspond to her preference system (Dellaertand Stremersch 2005, Franke and Piller 2004, Ghoshet al. 2006, Pine 1999, Randall et al. 2007, von Hippel2001). The process experience of self-designing theproduct and the effort involved have been portrayedas a disutility impacting the customers willingnessto use an MC toolkit and the likelihood of aban-doning the customization process without actuallybuying the product (Bendapudi and Leone 2003,Dellaert and Stemersch 2005, Huffman and Kahn

1998, von Hippel 2001). Value-generating psycholog-ical responses evoked by self-designing a product tothe value customers attach to the result have beenneglected, although a number of scholars acknowl-edge that the process of using an MC toolkit mightgive rise to positive emotions. Huffman and Kahn(1998), for example, suggest that some consumersmay find learning their preferences about a product tobe fun (p. 509), and Dellaert and Stremersch (2005)presume that consumers might enjoy mass customiz-ing a product (p. 226).

Only recently, Moreau and Herd (2009) found thatconsumers social comparisons to the designer of

comparable off-the shelf products influence theevaluations of their own self-designed creations,yielding support for our basic premise that psycho-logical factors play a major role in the value cus-tomers derive from MC. There is interesting ongoingresearch emphasizing the potential importance of theI designed it myself effect in MC. In an unpub-lished working paper, Deng and Hutchinson (2009)conclude from patterns in subjects ability to recog-nize their self-created designs when shipped thatinaddition to preference fitperceived authorship andpositive effects arising from the design phase alsoimpact the value they derive from self-design. Their

interpretations call for experimental studies in whichthe I designed it myself effect is disentangled anddirectly measured.

In sum, parallel research (Norton 2009, Moreau andHerd 2009, Deng and Hutchinson 2009) suggests thatbeyond preference fit and effort, there might be athird value-generating effect in MC that arises merelyfrom the fact that the customer is the designer of theproduct. We extend this line of research (1) by provid-ing a clean test for the I designed it myself effect inMC in which we keep preference fit constant; (2) by




3/16

Franke, Schreier, and Kaiser: The I Designed It Myself Effect in Mass CustomizationManagement Science 56(1), pp. 125140, 2010 INFORMS 127

shedding light on the theoretical mechanism underly-ing the effect, that is, the mediator variable of feelingsof accomplishment; and (3) by offering two modera-tors of the effect, namely the quality of the outcomeand the subjective contribution to the self-design pro-cess enabled by the design freedom the toolkit allows.

The relevance of these questions is high; scholarsand practitioners alike have underscored the impor-tance of understanding the mechanisms throughwhich MC generates value for customers (Dellaertand Stremersch 2005, Deng and Hutchinson 2009).After all, developing and implementing such a systeminvolves costs (Piller et al. 2004), and the adoption ofsuch a system only makes economic sense if it gener-ates value by allowing the manufacturer to charge aprice premium or to sell more units (Ansari and Mela2003, Kramer 2007).

In the following, we first develop testable hypothe-ses and then report from five studies in which we

tested them empirically.

2. Development of Hypotheses: Whyand When Self-Design with anMC Toolkit Generates Value

2.1. The I Designed It Myself Effect: DoesSelf-Design Generate Value?

How can we understand the I designed it myselfeffect? Literature on the endowment effect offersone possible explanation: individuals who created anobject interpret it more as theirs than individu-

als who merely bought it, and in turn, subjectiveownership feelings increase the subjective value ofthe product. Although the literature on the endow-ment effect initially focused on factual ownership andthus essentially adopted a binary concept (i.e., eitherone owns an object or not; Thaler 1980; Kahnemanet al. 1990, 1991), later research emphasized psycholog-ical ownership, the state in which individuals feel asthough an object is theirs (Pierce et al. 2003). Thestronger this feeling of psychological ownership is,the higher ones appraisal of an objects value will be(measured as willingness to pay (WTP) or to accept)(Reb and Connolly 2007). A number of researchers

have focused on the factors that lead to enhancedpsychological ownership. For example, Strahilevitzand Loewenstein (1998) found that feelings of owner-ship (and thus the endowment effect and the objectsvalue) are stronger when the person owns the objectfor a longer period of time. More recently, Pierceet al. (2003) proposed a psychological model of theantecedents, experiences, and consequences of psy-chological ownership. They suggest investing the selfin the object as one of the three routes to psycho-logical ownership, in addition to controlling the object

and getting to know the object intimately; they alsostate that the most obvious and perhaps the mostpowerful means by which an individual invests him-self or herself into an object is to create it (p. 93,emphasis by the authors). Similarly, Belk (1988, p. 144)states that the idea that we make things a part of self

by creating or altering them appears to be a universalhuman belief (emphasis by the authors), and Belkand Coon (1993, p. 405) hold that creating the objectis one of the clearest ways of incorporating it into theextended self (emphasis by the authors). This meansthat as a result, the product not only has instrumen-tal value but also (additional) psychological value forits originator (Csikszentmihalyi and Rochberg-Halton1981, Belk 1988, Kleine and Baker 2004, Mittal 2006).However, most of the research on this effect is the-oretical and supported only by anecdotal evidence.Pierce et al. (2003) therefore acknowledge the needfor empirical testing (p. 104) of the factors leading to

psychological ownership.Performing an experimental test of this effect in thefield of products self-designed with MC toolkits canbe seen as a bold step because clearly the degree ofself-investment in this case is relatively small com-pared to the very physical examples used in thisstream of literature, such as a man building a housewith his own hands. Indeed, Pierce et al. (2003) warnthat investment(s) of the self are unlikely to emergequickly (p. 96). Thus, if we find an effect arising fromself-design in the case of subjects using an MC toolkit,we can conclude that this effect is likely to hold ingeneral.

Despite the limited investment of energy involved,MC toolkits still facilitate the act of creating some-thing (von Hippel 2001), and the objects endoge-nous to this process should at least lead to a higherdegree of psychological ownership than similar off-the-shelf products, all other things being equal. Usingan MC toolkit as found on the Web in various con-sumer product categories, customers can select col-ors, designs, and shapes; come up with new andcreative combinations; upload text or images; andso forth. The symbolic enrichment of the productby self-designing it should thus elicit a higher valueamong MC customers than a similar product pur-

chased off the shelf. Usually self-designed productsare customized to ones preferences, which meansthat they are not similar to prefabricated products.However, we propose that self-designing results in ahigher value attribution for the product even if wecontrol for preference fit, thus postulating the exis-tence of a mere I designed it myself effect.

Hypothesis 1 (H1). Beyond the products preferencefit, having designed a product oneself with an MC toolkitdelivers a positive value increment for the respective cus-tomer compared to purchasing a product off the shelf.




4/16


2.2. Is the I Designed It Myself Effect Mediatedby Feelings of Accomplishment?

In the next step, we investigate the mechanism of themain effect more closely. Why does creating an artifactprompt the customer to attribute special value to it?Drawing on Csikszentmihalyi and Rochberg-Halton

(1981), Belk (1988, p. 144) explains, [W]e invest psy-chic energy in an object to which we have directed ourefforts, time, and attention. This energy and its prod-ucts are regarded as a part of self because they havegrown or emerged from the self. Thus, ones ownaccomplishment is embodied in the object endoge-nous to the process. This proud feeling of accomplish-ment serves the need for feelings of competence andefficacy deeply embedded in human nature (Furby1991, Williams and DeSteno 2008). We therefore pro-pose that creating a product oneself elicits feelings ofaccomplishment that are then strongly associated withthe product. In turn, these positive feelings result in a

higher valuation of the object.

Hypothesis 2 (H2). The effect of having designed aproduct oneself with an MC toolkit on the subjective valueof the product is mediated by the feeling of accomplishmentassociated with the object.

2.3. Is the I Designed It Myself EffectModerated by the Outcome of the Process?

We cannot assume that the enhanced valuation ofself-designed objects is a universal law. Rather, it willbe contingent upon certain moderating factors. Onepotential factor is the perceived attractiveness of the

object created. If the artifact fails to meet the sub-jects requirements and the process is perceived asunsuccessful, it is unlikely to satisfy the need forcompetence and efficacy (Pierce et al. 2003). The sub-ject might feel that her investment of time and effortwas not fruitful, which will negate the reasons whyshe should value the outcome (Norton 2009). In theextreme, these feelings may even turn to hatred, asexemplified by a painter who lacerates an unsuccess-ful painting in a sudden fit of anger. In such a case,the object created is more a symbol of incompetencethan competence. This might be an extreme case, as

normally individuals self-protection motive will biastheir perception in a self-flattering manner (Sedikides1993). However, the effect of self-design on perceivedvalue should be enhanced when the subjective prefer-ence fit of the product is higher than when preferencefit is lower. The higher the subjective preference fitachieved, the stronger the main effect should be.

Hypothesis 3 (H3). The effect of having designed aproduct oneself with an MC toolkit on the subjective valueof the product is moderated by the subjective preference fitof the product.

2.4. Is the I Designed It Myself EffectModerated by the Contribution tothe Process?

Assuming that the self-design process is not a blatantfailure, feelings of accomplishment should intensifyin line with the subjects feelings of being the cause

(Pierce et al. 2003, p. 89). Furby (1978) argues that themore the subject is able to exercise control over theobject, the more it will be experienced as part of theself. The reason is that the subject might then attributethe outcome more to her own accomplishment, thussatisfying the need for causal efficacy. The subjectiveassessment of the extent to which one is the causeof the resulting product is likely to be affected bythe subjective contribution enabled by the MC toolkit.The design freedom offered by existing MC toolkitsvaries widely in this respect. Some offer only lim-ited choices (e.g., colors for three product modules) inorder not to overstrain customers (Huffman and Kahn

1998), whereas others offer a virtually infinite solutionspace (e.g., by allowing users to upload self-createdpictures) in order to enable closer preference fit (vonHippel and Katz 2002). We argue that the main effectwill become stronger in those MC toolkits that allowusers to make a larger subjective contribution to theself-designed product. As the subject has a strongersense of being the originator, she will value the result-ing self-designed product more than a subject whofeels that she did not contribute much to the designof the product.

Hypothesis 4 (H4). The effect of having designed a

product oneself with an MC toolkit on the subjective valueof the product is moderated by the subjective contributionenabled by the MC toolkit.

3. Study 1: Qualitative Explorationof I Designed It MyselfFeelings in MC

3.1. MethodIn Study 1, we explore on a qualitative basis whetherI designed it myself feelings actually arise in thecontext of MC, whether they have a value-generatingeffect, and if so, how this effect can be understood. Werecruited a sample of 37 business students (averageage: 23.7 years; 51% females) for a product test studyin separate rooms at the authors university. Subjectswere offered free beverages and snacks to create a nat-ural environment that came close to the experience ofsitting at their own PCs at home. Each subject was ran-domly assigned to one of three MC toolkits and askedto design a product virtually according to her individ-ual preferences. These toolkits allowed them to designan individual (1) T-shirt (http://www.shirtcity.com),(2) scarf (http://www.wildemasche.de), or (3) cell




5/16


phone cover (http://www.designyourhandy.de). TheT-shirt toolkit offered less design freedom than theother two toolkits (in addition to creating some verybasic text elements, users could only place one ofseveral predefined designs on the T-shirt, whereasthe other toolkits allowed users to upload multi-

ple pictures and graphics from any external source).These research objects are typical MC toolkits simi-lar to those common on the Web. As an incentive,we offered participants the opportunity to win eithertheir self-designed product or a comparable off-the-shelf product. This ensured that participants engagedin the self-design processes seriously, similar to the set-ting in reality.

In an in-depth interview approximately 30 minutesin length, each participant was then questioned asto whether she liked her self-designed product andasked to reflect on the reasons why. The interviewswere recorded, transcribed, and analyzed. Two inde-

pendent coders grouped motives into two categories:(1) benefits arising from I designed it myself feel-ings, that is, motives that allow us to conclude that themere awareness of being the originator of the designcreated value for the subject, and (2) other benefits(the latter component in particular comprised state-ments related to the preference fit of the resultingdesign). The values for Krippendorffs alpha are 0.71(I designed it myself feelings) and 0.69 (other bene-fits); values greater than 0.67 are generally taken as anindication of satisfactory agreement among multipleraters (Krippendorff 2004).

3.2. Findings and DiscussionOverall, inquiries as to why participants liked theirself-designed products revealed that 70% of themreported motives related to feeling like the originatorof the design (I designed it myself). Naturally, othermotives also played a role (62% of the cases), withparticipants frequently mentioning preference fit as a

Table 1 Interview Excerpts from Study 1

Reasons for

product value Frequency Examples

Being the originator (I

designed it myself)

70% (scarf 90%, T-shirt 46%,

cell phone cover 79%)

Its definitely not the same as buying a T-shirt in a shop. There is something personal about it.

So its interesting to buy this T-shirt and not a different one created by someone else, wheresomeone else, uh, incorporates his ideas. ( ). It is, its the effort, working and thinking

about what I could do, and this, uh, it is the spirit that is incorporated in it ( ) I think Ive

developed an addiction to the T-shirt ( ) because I designed it, it gained a special, a special

dimension for me.

I think its cool ( ) For me, it has personal value and personal uniqueness ( ) Its from me!

It is, how should I say, something of my own ( ) (cell phone cover)

Well, I fiddled around if someone else had made the exact same one ( ), I would not care,

but if I (scarf)

Other motives (especially

preference fit)

62% (scarf 60%, T-shirt 70%,

cell phone cover 50%)

You can also buy a cell phone cover in a shop, but they are well, they lack the specific ,

you do not get what you want ( )

I could design [the T-shirt] the way I want to and not the way the manufacturers want it.

The scarf looks exactly as I want.

reason for liking the self-designed object. This pat-tern is visible in all three product categories. Table 1lists a number of illustrative statements. It is strikingthat these short and virtual design processes evokedsuch strong emotions. Many participants developeda somewhat personal relationship to the products

although they were only visual representations of dig-ital information and not yet physical objects. Thisis exemplified in statements such as its from meor it is the spirit that is incorporated in it, whichwe heard in many variations. Their accounts alsoindicated that they were proud of the accomplish-ment and the fact that they had given birth to newdesigns. Overall, the findings provide qualitative sup-port for our first two hypotheses.

It also became clear that there is some form ofinteraction between the I designed it myself effectand the effect of the preference fit achieved: 38% ofsubjects revealed that both motives played a role for

them, or they gave answers that made it difficult todisentangle these two motives, for example, whensubjects stated (I like it) because I made it myself. Itpleases me more than a standard product if I madeit myself or I am happy about it! I am happy thatI did a good job, and it simply makes me proudthat I designed something so beautiful, which can beinterpreted as qualitative support for our moderatorhypothesis (H3). Finally, H4 also gained qualitativesupport. The frequency of the I designed it myselfmotive is clearly lowest in the group of subjects whoused the T-shirt toolkit, in which their design free-dom and thus also their subjective contribution to the

result is lowest. In the other two groups, this motivewas indicated roughly twice as often.

In sum, this study provides initial support forour hypotheses. However, the qualitative and cross-sectional setting of the study and the small num-ber of subjects involved warrant further experimentalstudies.




6/16


4. Study 2: The Independent Effectof Self-Design on the ValueAttributed to MC Products (theI Designed It Myself Effect)

4.1. Method

4.1.1. Design. In Study 2, we aim to test whetherhaving designed a product oneself with an MC toolkithas a value-generating effect, that is, whether it leadsto a higher subjective valuation of the self-designedproduct compared to a product obtained off the shelf(H1). The challenge was to rule out the potentiallyconfounding influence of preference fit. To isolatethe I designed it myself effect, we devised a one-factor between-subject experiment that ensured thatthe differences in subjective value attributions (mea-sured as WTP) can only be attributed to the waythe product is obtained (self-design or off-the-shelf

purchase) and not to preference fit. In this setting,subjects in three experimental groups all expressedtheir WTP for an objectively identical object, namelya college T-shirt with a specific design (Design A),printed on American Apparel T-shirts by the com-pany Customink (http://www.customink.com). Onaverage, the preference fit was thus kept identical forthe three groups. The difference between the threegroups (and thus our experimental stimulus) is onlythe process by which they obtained the T-shirt. Nat-urally, our standardization of the outcome restrictsthe creativity involved in the design process and willmost likely reduce the effect size. We will address that

issue further below.For Group 1, we simulated an off-the-shelf buying

situation involving a standard product. This meansthat after inspecting the college T-shirt (Design A) ona poster, the members of Group 1 were asked to indi-cate their WTP for the T-shirt (You can now bid onthis American Apparel T-shirt; if you win it, we willarrange for production by Customink and make itavailable to you.).

Group 2 was instructed to use the toolkit pro-vided by Customink to design a T-shirt with the tar-get design (Design A). The target T-shirt design hadoriginally been created using the Customink toolkit.

Subjects had to design a total of four different textelements, upload one logo, and position each ele-ment properly on the T-shirt, which they managed in23 minutes on average (three subjects failed to repro-duce the T-shirt properly and were subsequently elim-inated from further analyses). A thorough inspectionof the remaining users designs confirmed that an out-sider would not see any difference compared to thetarget design. In this way, preference fit is held con-stant between this group and Group 1. Having fin-ished, subjects were asked to submit a binding bid

for their self-designed T-shirts (You can now bid onthe American Apparel T-shirt you designed; if you winit, we will arrange for production by Customink andmake it available to you.).

If H1 is correct, we should observe a higher WTPamong subjects in Group 2 than in Group 1. Unlike

the others, subjects in Group 2 went through the self-design process and should thus perceive the T-shirtas an object they had created.

Beyond testing the existence of the main effect (H1),we aimed to address three alternative explanationsfor the potential value increment apart from beingthe originator of the T-shirt and the associated feelingof accomplishment (as posited in H2): (1) The sub-jects in Group 2 were exposed to target Design Afor a longer period of time than those in Group 1,because the former were required to reproduce thedesign. Theoretically, this greater familiarity with thedesign might have induced a positive affective reac-

tion, a phenomenon known as the mere exposure effect(Zajonc 1968). In the extreme, this would mean that itis not the feeling of accomplishment associated withbeing the originator of the product that generatesadditional value but only the subjects greater famil-iarity with the design. (2) Another alternative expla-nation would be that the activity of designing and theenjoyment involved have a positive effect on the sub-jects general mood and therefore potentially increasetheir WTP (see Pham 1998 as well as Schwarz andClore 1983 for examples of such a mood effect). (3) Ona related note, a Hawthorne effect is also possible(Roethlisberger and Dickson 1939, Adair 1984): Sub-

jects in Group 2 may have responded to the higherperceived implicit social cognition on the part of theresearchers, who had more frequent and intense con-tact with them than with the subjects in Group 1.

Therefore, Group 3 also received Design A as amodel for the self-design process using the toolkit (asin the case of Group 2), which took them 24 minutes onaverage (difference to Group 2 n.s.). Again, an inspec-tion of the users reproductions confirmed that an out-sider would not see a difference compared to the targetdesign (two subjects failed to reproduce the T-shirtproperly and were subsequently eliminated from fur-ther analyses). After finishing, however, the subjectswere not asked to bid on the T-shirt designs theyhad made themselves (like Group 2) but on a similar-looking target design as an off-the-shelf product (likeGroup 1; pointing to the T-shirt with Design A on aposter: You can now bid on this American ApparelT-shirt; if you win it, we will arrange for productionby Customink and make it available to you.).

If the enhanced value attribution is in fact causedby the alternative explanations of a mere expo-sure effect, a mood effect, or a Hawthorne effect,we should observe a difference in WTP between




7/16


Groups 1 and 3. Both bid on exactly the same off-the-shelf T-shirt shown on the poster, but Group 3 wasexposed to the alternative factors of longer exposuretime to Design A, might have been in a better moodafter the design experience, and might have perceivedan increased level of attention from the researchers.

If there are no differences, then it appears unlikelythat the alternative explanations play a major rolein our setting and thus also in the comparison ofGroups 1 and 2.

If H1 is correct and designing a product oneself hasan effect on the value attached to the product (theI designed it myself effect), we should also observea difference in WTP between Groups 2 and 3. Subjectsin Group 3 did not bid on their own T-shirts, that is,the T-shirts where they were involved in the designprocess. There is no reason for them to feel like theoriginators of the T-shirt. However, prior to the buy-ing process, both groups were engaged in a similar

activity for a similar period of time, namely repro-ducing the T-shirt. Thus, the length of exposure toDesign A, the mood resulting from the design activ-ity, and the attention perceived should be identicalbetween Groups 2 and 3. In the case of differences inWTP, they cannot be attributed to a mere exposureeffect, a mood effect, or a Hawthorne effect.

4.1.2. Procedure. A total of 114 business students(average age: 22.8 years; 47.7% females) recruitedfrom the authors university were randomly assignedto one of three groups. As an incentive to participate,each subject received a small gift (worth 10 euros) and

participated in a raffle for a ski weekend.Each subject was seated in a separate booth, andthe setting ensured that no interaction between sub-jects was possible during the experiment. All instruc-tions were given verbally by the instructors, whoused written checklists to keep the instructions stan-dardized across the experiment and who had beentrained in a workshop prior to the experiment. Asa first task (before being exposed to their individualtreatments), subjects in all three groups completed ashort questionnaire that contained the control vari-ables (1) product interest, (2) purchase intention, and(3) income. After showing subjects the college T-shirt

design (Design A) on a poster (prior to treatment), weasked about (4) their average hypothetical WTP forthe product category and (5) the preference fit of theT-shirt with Design A.

4.1.3. Measurement. WTP was measured bymeans of incentive-compatible BDM auctions (Beckeret al. 1964, Nunes and Boatwright 2004, Wang et al.2007), which have proven to be a highly reliableand valid method of measuring consumers WTPfor consumer goods (Wertenbroch and Skiera 2002).The subjects handed in their binding bids for the

underlying product and then drew a card from anurn on the spot. If a subjects bid was higher thanor equal to the price on the card, she was actuallyrequired to purchase the T-shirt at the price indicated onthe card. If it was lower, she could not purchase theT-shirt. This procedure ensures theoretical incentive

compatibility: prices are exogenous to the subjectsWTP (unlike in other methods such as English auc-tions, bidders cannot directly influence the price paid;Kagel 1995, Wertenbroch and Skiera 2002). Therefore,our dependent variable is not hypothetical but con-stitutes real economic behavior. To avoid anchoringdistortion, subjects were not informed about the pricerange shown on the cards (Wertenbroch and Skiera2002). The prices in the urn ranged from 7 to 25 euros,meaning that they started at a level somewhat belowmarket prices. In twelve cases, the subjects WTPwas higher than the price indicated on the carddrawn. In all of those cases, the subject readily paid

the price indicated on the card on the spot andindicated her name, the desired size, and the addresswhere the T-shirt with Design A should be delivereda few weeks later. As we find that preference fit(measured before treatment) is significantly correlatedwith the dependent variable WTP (measured aftertreatment in the course of the BDM auctions; r= 037;p < 0001, we can assume high levels of validity inour WTP measurement (this procedure is suggestedby Wertenbroch and Skiera 2002).

The control variables we measured were as follows:(1) product interest (My general interest in a collegeT-shirt is high; 3-point scale: 1 = strongly disagree

and 3 = strongly agree); (2) purchase intention (It ishighly probable that I will purchase a college T-shirtwithin the next month; 5-point scale: 1 = stronglydisagree and 5 = strongly agree); (3) income (Howhigh is your disposable income per month?; 6-pointscale: 1 100 euros and 6 500 euros); (4) their aver-age hypothetical WTP for the product category (Howmuch do you usually pay for a T-shirt of compa-rable quality?; amount in euros); and (5) the pref-erence fit of the T-shirt with Design A (I like thedesign of the T-shirt, The T-shirt design comes closeto my idea of a perfect design, The design of theT-shirt looks really great; all three items measured

on 5-point scales, where 1 = strongly disagree and 5=strongly agree; alpha = 0.87; adapted from Randallet al. 2007). Because of the random assignment of sub-jects to groups, there were no significant differencesbetween them with regard to any of these controlvariables.

4.2. Findings and DiscussionThe results support H1: Having made the prod-uct oneself using an MC toolkit increases the valueattributed to the product (measured as WTP), and thiseffect is independent of preference fit (see Table 2).




8/16


Table 2 Value Resulting from Self-Design (Study 2 Findings)

WTP in euros Post hoc tests (LSD)

Treatment Mean (SD) Group 2 Group 3

Group 1

(n= 38)

No activity, bid on

off-the-shelf T-shirt

(Design A)

4.75 (3.04) p= 001 n.s.

Group 2

(n= 33)

Re-design of T-shirt,

bid on self-designed

T-shirt (Design A)

6.85 (3.70) p= 005

Group 3

(n= 38)

Re-design of T-shirt,

bid on off-the-shelf

T-shirt (Design A)

5.26 (3.56)

Note. ANOVA; F2 108 = 3512, p < 005; 2p = 006.

If we first examine the WTP of Groups 1 and 2, wefind that subjects who actively reproduced Design A(Group 2) were willing to pay significantly more fora T-shirt with that design (M= 685 euros; SD= 3.70)

than subjects in Group 1, who submitted a bid fora T-shirt with the same design but as an off-the-shelf product (M= 475 euros; SD = 3.04; p = 001.Although the product designs and the informationabout the manufacturer are objectively identical forboth groups, the difference in WTP is relatively high(over 40%).

As noted above, however, beyond having designedthe T-shirt oneself, there are alternative explanationsfor the value increment observed. We therefore turnto Group 3. In line with our predictions, we find thatsubjects WTP is not significantly different from that ofsubjects in Group 1 (n.s.). Moreover, we find that sub-

jects in Group 3 who self-designed but did not bidon their own T-shirts were willing to pay signifi-cantly less than those in Group 2, who self-designedand bid on their own T-shirts (p = 005. The effectsof the alternative mechanisms of mere exposure, bet-ter mood, and increased attention from the researcherscan therefore be considered negligible in our setting.Overall, we have thus found clear quantitative sup-port for H1: Creating an object oneself with an MCtoolkit will prompt the creator to attribute highervalue to the object endogenous to the process.

When interpreting our findings, however, it isimportant to bear in mind that the experimental set-

ting used in this study has two limitations: (1) Thesetting limited the size of the I designed it myselfeffect. To keep preference fit constant between groups,we standardized the target design. This means thatsubjects did not design as freely as they would whenusing an MC toolkit in real life. In a recent article,Dahl and Moreau (2007) show that tasks in which thetarget outcome is defined (like building model air-planes or painting by numbers) evoke a significantlylower feeling of creativity than unrestricted tasks.We therefore argue that our study is conservative in

nature, and the effect would be even stronger if thecustomer were actually the originator of the designand not merely a reproducer. (2) The second limita-tion is that there is yet another alternative explanationfor the observed value increment in Group 2, namelythat their WTP is higher compared to Group 1 merely

because of the effect of sunk costs (Thaler 1980). Incontrast to Group 1, subjects in Group 2 invested bothtime and effort in the process. Although subjects inGroup 3 were likewise involved in the design process,one might argue that sunk costs only affect the bidsin Group 2 because Group 3 did not actually bid onthe T-shirt they designed, but on the target design onthe poster. We will address these two limitations inStudy 3.

5. Study 3: Feeling ofAccomplishment as a Mediator of

the I Designed It Myself Effect5.1. Method

5.1.1. Design and Procedure. The objective of thisexperiment was to test whether the effect of havingdesigned a product oneself with an MC toolkit on thesubjective value of the product is mediated by thefeeling of accomplishment associated with the object(H2). We asked 116 business students (average age:23.6 years; 54.2% females) to self-design a pair of skisor to choose one of several standard ski designs. Incontrast to Study 2, we did not provide a target designfor those who engaged in the self-design activity (i.e.,

subjects were free to design the skis according to theirpreferences). The subjects behavior therefore emu-lates real MC customer behavior; also, the feeling ofbeing the originator of a design can take full effectwhen subjects are not confined to the task of repro-ducing a target design. We used a different toolkit andproduct category to enhance generalizability.

After answering two control questions (productinterest and income, measured as in Study 2; no dif-ferences between groups), participants were randomlyassigned to one of the following two groups: Group 1engaged in self-design behavior. We instructed them tocreate a ski design according to their own preferences.For this task, we used the toolkit offered by the skimanufacturer Edelwiser (http://www.edelwiser.com).Participants in Group 2 bought off the shelf: We hadthem inspect on a prepared website 28 standard pro-fessional ski designs by the same manufacturer andchoose the one they liked most. All of the standarddesigns used in this study are actually marketed tocustomers by Edelwiser. Immediately after completingthe respective task, subjects handed in binding bidsfor the self-designed (Group 1) or chosen (Group 2)designs and then filled out a questionnaire.




9/16


The questionnaire contained items to measure themediating variable (the feeling of accomplishmentassociated with the task-specific design) and the con-trol variables, namely preference fit (which in this set-ting was likely to be different between the groups)and perceived process costs. We included the latter

as a proxy for sunk costs. If H2 is correct, we shouldfind that a mediator effect arising from the feelingof accomplishment explains why self-design createsvalue for the subjects. The inclusion of perceivedprocess costs will allow us to determine whether thealternative explanation mentioned previously (sunkcost effect) is likely to account for the effects found inStudy 2.

The incentives were a gift bag worth 10 euros anda raffle for 14 pairs of skis. As in Study 2, we ensuredthat no interaction between subjects took place dur-ing the experiment, and once again there was no timelimit for completing the respective tasks. All instruc-

tions were given verbally by the instructors, who usedwritten checklists to keep instructions standardizedacross the experiment.

5.1.2. Measurement. WTP was again measuredusing BDM auctions. The only difference from Study2 is that we decided to have subjects bid not on theentire product but only on the graphic design of theskis. The subjective value therefore corresponds to theprice premium they would pay for having their self-created or chosen graphic design instead of a blankwhite ski design. We did this because subjects mighthave different perceptions of the skis technical qual-ity (which is not affected by self-design), and these

potential differences should not create noise in ourmeasurement of the dependent variable (WTP).

We used a relatively elaborate method to rule outthis effect. Prior to the bidding, we informed par-ticipants that the 14 pairs of skis to be raffled offamong the participants ten days after the experimentwould contain no graphic design from the outset, andwe handed them a blank white ski from the manu-facturer for physical inspection. Then we told themthat they now had the opportunity to bid on theirown design or the chosen graphic design. If they wonthe raffle, they would win one of the 14 pairs of skis.The outcome of the BDM auctions would then deter-mine whether they would (1) get only the pair ofwhite skis for free (if the price drawn was higherthan their bid) or (2) get the pair of skis for free andpay the price drawn for having their own design orthe chosen graphic design printed on the skis (if theprice drawn was less than or equal to their bid). Ifthey did not win the raffle, their bids would haveno consequences. Interviews revealed that the sub-jects clearly understood this principle. As in Study2, subjects were not informed about the price rangeshown on the cards to avoid anchoring distortion

(Wertenbroch and Skiera 2002). The prices in the urnranged from 20 to 120 euros. Of the 14 pairs of skisraffled off after the completion of the experiment, fourpairs were delivered as blank white skis (where thewinners bid was lower than the price indicated onthe card drawn). Overall, the procedure ensured that

the bids only referred to the graphic design and notto the perception of the skis technical quality.The feeling of accomplishment associated with the

self-designed or chosen skis was measured using thefollowing three items (adapted from Louro et al. 2005):When I look at the ski I have self-designed (Group 1;Group 2: I have chosen), (1) the feeling I havecan best be described by the word pride ; (2) I feelproud of having accomplished something; (3) I feelproud because I did a good job (alpha= 0.96). Pref-erence fit was measured using the same three itemsas in Study 2 (alpha = 0.93). The items for perceivedprocess costs (adapted from Dellaert and Stremersch

2005) were The process of getting my ski designwas (1) exhausting and (2) time-consuming(alpha = 0.80). All items were measured on 7-pointscales, where 1 = strongly disagree and 7 = stronglyagree.

5.2. Findings and DiscussionFirst, descriptive findings show that subjects whodesigned their own skis were willing to pay signif-icantly more for their designs (Group 1: M= 7442;SD = 56.85) than subjects who were given the taskof choosing one of the standard professional designs(Group 2: M = 4589; SD = 43.58; F1 116 = 9153;p < 001. This finding replicates those of Franke andPiller (2004) and Schreier (2006) and confirms the pre-diction from Ulrich (2009) cited in the introduction.

In line with earlier findings in MC literature, thosesubjects also reported significantly higher perceivedpreference fit (Group 1: M= 545; SD= 0.57; Group 2:M= 416; SD= 1.27; F1 116 = 15370; p < 0001. Moreimportantly, the feeling of accomplishment associ-ated with the underlying skis is also significantlyhigher for subjects in Group 1 (M= 393; SD = 1.45)versus Group 2 (M= 127; SD= 1.14; F1 115 = 25962;p < 0001; see Table 3). These findings show that sub-jects indeed develop a stronger sense of accomplish-ment with products they design themselves comparedto standard products they merely choose. Finally, per-ceived process costs were also rated higher in the self-design versus standard scenario (Group 1: M= 337;SD = 1.61; Group 2: M= 204; SD = 1.26; F1 115 =243671 115; p < 0001).

We test H2 using four linear models with WTPas the dependent variable (see Table 4; Baron andKenny 1986). In Model 1, we find that self-designinga product has a positive effect on WTP (F1 116 = 9153;p < 001. If we include the feeling of accomplish-ment as a covariate in Model 2, we find this variable




10/16


Table 3 Differences in WTP and Accomplishment (Study 3 Findings)

Mean statistics (n= 116)

Group 2

Group 1 off-the-shelf

self-design design

M (SD) M (SD) F-value 2p

WTP (euros) 7442 (56.85) 4589 (43.58) 9153 007

Feeling of 393 (1.45) 127 (1.14) 25962 019

accomplishment

Preference fit 545 (0.57) 416 (1.27) 15370 012

Perceived process 337 (1.61) 204 (1.26) 24367 018

costs

Notes. Feeling of accomplishment, preference fit, and perceived process

costs are measured on 7-point scales, where 1 = very low and 7= very high.p < 005; p < 001; p < 0001.

to be significantly related to WTP (F1 116 = 16349;p < 0001, and at the same time we find that thetreatment effect becomes insignificant (F1 116 = 0471;

n.s.). A Sobel test supports the idea that the feel-ing of accomplishment mediates the effect of self-design (z = 3822; p < 0001. The results are simi-lar if we also include preference fit as a covariate inthe model (Model 3): both accomplishment (F1 115 =13161; p < 0001 and preference fit (F1 115 = 4225;p < 005 are significantly related to WTP, and thetreatment effect becomes insignificant (F1 115 = 1716;n.s.). We find that both the feeling of accomplish-ment (z = 9550; p < 0001 and preference fit (z =6130; p < 005 mediate the main effect of havingself-designed a product. When we run the modelwith perceived process costs as an additional covari-

ate (Model 4), we find that the group effect remainsinsignificant (F1 115 = 1570; n.s.), and both accom-plishment (F1 115 = 12504; p < 0001 and prefer-ence fit (F1 115 = 4080; p < 005 remain significantlyrelated to WTP. The effect of perceived process costson WTP, however, is not significant (F1 115 = 0032;n.s.). Again, a Sobel test for mediation confirms thesefindings (accomplishment: z = 3419; p < 0001; pref-erence fit: z= 1975; p < 005; perceived process costs:z=0182; n.s.). Overall, we find clear support for

Table 4 Feeling of Accomplishment as a Mediator Variable (Study 3 Findings, Continued)

DV: WTP in euro (ANOVA)

Model 1 Model 2 Model 3 Model 4

F 2p F 2p F

2p F

2p

Group (self-design vs. off the shelf) 9153 007 0471 0004 1716 0015 1570 001

Feeling of accomplishment 16349 013 13161 011 12504 010

Preference fit 4225 004 4080 004

Perceived process costs 0032 000

R2 0.074 0.190 0.220 0.220

Note. n= 116.p < 005; p < 001; p < 0001.

H2, and we do not find evidence for sunk costs as apotential alternative explanation for the I designedit myself effect.

6. Study 4: Preference Fit as aModerator of the I DesignedIt Myself Effect

6.1. Method

6.1.1. Design, Procedure, and Measurement. InStudy 4, we test whether the effect of having designeda product oneself with an MC toolkit on the subjec-tive value of the product is moderated by the subjec-tive preference fit of the product (H3). We adopteda similar experimental approach to that employedin Study 2. The difference is that we manipulatednot only the process of how the subjects obtainedthe product (self-designed T-shirt versus off-the-shelf

T-shirt) but also the preference fit of the product (highversus low preference fit). This led to a 2 2 facto-rial design. In Groups 1 and 2, we simulated an off-the-shelf buying situation with a standard T-shirt. Thepreference fit of the T-shirts target design for Group1 was set to a high level (Designattractive, whereasfor Group 2 the target design was set to a lowlevel (Designunattractive. Groups 3 and 4, on the otherhand, were instructed to reproduce T-shirts: Group3 reproduced Designattractive and Group 4 reproducedDesignunattractive. A total of 129 business students (aver-age age: 22.8 years; 45.7% females) were randomlyassigned to one of the four groups. The procedures

and incentives were similar to those employed inStudy 2: students also received a gift worth 10 euros asan incentive for participation. They were again seatedin separate booths and first answered a set of controlquestions (product interest, purchase intention, andincome, measured as in Study 2). As expected, wefound no significant differences between the groupsregarding these control variables. After exposing themto the target design, we measured the subjects hypo-thetical WTP for the product category and, to check




11/16


the manipulation of the T-shirts attractiveness, thepreference fit of the respective target design. We usedthe same items to measure preference fit as in the pre-vious studies (5-point scale; alpha = 0.89). In a pilotstudy (n = 20) prior to Study 4, we had pretestedten different designs and depicted the most attrac-

tive one as Designattractive and a more mediocre one asDesignunattractive.Groups 3 and 4 were then asked to reproduce the

target designs using the toolkit provided by Cus-tomink and then stated their WTP for the T-shirt theyhad designed themselves. Eventually, we measuredthe perceived process costs (same items as in Study 3;5-point scale; alpha = 0.70). Groups 1 and 2 statedtheir WTP for a T-shirt with the respective targetdesign off the shelf (i.e., without engaging in a self-design process). The procedure used for the BDM auc-tions was identical to the one employed in Study 2. In16 cases, the subjects bids were higher than the price

drawn and they purchased the T-shirt. As in Study 2,we ensured that an outsider would not notice anydifference between the target design and the repli-cated designs (three subjects from Group 3 failed toreproduce the T-shirt properly and were subsequentlyeliminated from further analyses). If H3 is correct, weshould find that the preference fit of the target designmoderates the main effect of self-design on the sub-jective value of the product (i.e., there should be sig-nificant interaction effect).

6.2. Findings and Discussion

6.2.1. Manipulation Check. A mean comparison

between preference fit ratings for Designattractive (M=285; SD = 0.89) versus Designunattractive (M = 207;SD= 0.70; F1 129 = 31323; p < 0001 confirms that themanipulation was successful. There was no differencein the perceived process costs between the two self-design groups (high preference fit: M= 256; SD =1.03; low preference fit: M= 247; SD = 0.72; F1 63 =0221; n.s.), which confirms that we only manipulatedthe outcome, not the process itself.

6.2.2. Findings. First, we were able to replicatethe findings from Study 2 both for the high andlow preference fit scenarios (see Table 5). Subjects inGroup 3 (reproduction of Design

attractive

were willingto pay significantly more for the T-shirt (M= 1025;SD= 6.15) than subjects in Group 1 (off-the-shelf pur-chase of Designattractive (M= 535; SD = 2.40; F1 59 =16501; p < 0.001). Similarly, the bids of subjects inGroup 4 (reproduction of Designunattractive are also sig-nificantly higher (M = 718; SD = 3.70) than thoseof Group 2 (off-the-shelf purchase of Designunattractive)(M= 524; SD = 2.94; F1 70 = 5926; p < 0.05). Thisagain provides sound support for H1.

In line with H3, we see that the difference in WTPis much larger in the high preference fit scenario

Table 5 Positive Interaction Between Preference Fit and Self-Design

Effect (Study 4 Findings)

F-value with

WTP mean (SD) df= 1 (2p

Process 23659 (0.16)

Off-the-shelf design (n= 66) 529 (2.68)

Self-design (n= 63) 859 (5.17)

Outcome 5131 (0.04)

Designattractive (n= 59) 776 (5.22)

Designunattractive (n= 70) 618 (3.44)

Process outcome 4446 (0.03)

p < 005; p < 001; p < 0001.

( WTP Group 1 versus Group 3 = 4.90) than in thelow preference fit scenario ( WTP Group 2 versusGroup 4= 1.94), suggesting that the outcome (subjec-tive preference fit) moderates the magnitude of theI designed it myself effect. We test this interactionusing a linear model with the two factors process(self-design versus off the shelf) and outcome (highversus low preference fit) as well as their interactionas independent variables. First, we find that both fac-tors are significantly related to WTP. Subjects whoactively reproduced T-shirts submitted significantlyhigher bids for the self-designed T-shirt (M= 859;SD = 5.17) than those who submitted a bid for therespective off-the-shelf T-shirt shown on the poster(M= 529; SD = 2.68; F1 129 = 23659; p < 0.001). Wealso find that in the scenarios where the preference fitwas manipulated to a high level (Designattractive, sub-jects submitted significantly higher bids (M= 776;

SD=

5.22) than in the scenarios where preferencefit was manipulated to a lower level (DesignunattractiveM= 618; SD= 3.44; F1 129 = 5131; p < 005). Second,and more importantly, we find that the interactioneffect is significant (F1 129 = 4446; p < 005). This sup-ports H3: The higher the preference fit, the greater theeffect of self-design on the subjective value. Subjectswho manage to design a product they really like mayenjoy a greater feeling of accomplishment than thosewho create a product of mediocre subjective quality.However, it is worth noting that we would expect tofind no value increment (or even a negative effect) ifa customer completely failed to design a product that

matched her preferenceslike a painter who laceratesher unsuccessful painting in a sudden fit of anger.

7. Study 5: The SubjectiveContribution as a Moderator of theI Designed It Myself Effect

7.1. Method

7.1.1. Design and Procedure. In Study 5, we testwhether the effect of having designed a product one-self with an MC toolkit on the subjective value of the




12/16


product is moderated by the subjective contributionenabled by the MC toolkit (H4). We again changedthe underlying product category (this time to wrist-watches) to ensure more generalizable findings. Wemanipulated the design freedom of an MC toolkitfor watch faces developed for the purposes of the

study. A total of 66 business students (average age:23.7 years; 36.9% females) participated in a between-subject experiment and were randomly assigned totwo groups. Group 1 (low subjective contribution)used an MC toolkit that allowed them to design awatch face by configuring it from a set of prede-fined attributes. The toolkit offered six different back-ground colors, six different face designs, six differentnumeral styles, and twenty different numeral colors(thus a solution space of 66620). (Our selection ofspecific attributes was based on attractiveness scoresobtained from 15 students in a pilot study we hadconducted prior to Study 5; attractiveness was mea-

sured with the single item I like this backgroundcolor/face/numeral style/numeral color, where 1 =strongly disagree and 5= strongly agree. The selectedattributes received an average score of > 3.) We reasonthat merely choosing predefined attributes will evokeonly a moderate feeling of contributing to the out-come. For Group 2 (high subjective contribution), weextended the design freedom of the toolkit to enhancethe subjective contribution. Here, subjects receivedthe same predefined attributes but could addition-ally modify the watch face freely, upload pictures,and create new designs. The solution space was thusonly limited by the size and shape of the watch

face. As in the previous studies, we standardizedinstructions using written checklists, and all studentsreceived a gift bag worth 10 euros in return for theirparticipation. After answering two control questions(product interest and income), subjects designed theirown watches. Having completed their self-design pro-cesses, subjects in both groups were asked to bidon their designs. Eventually, they completed a shortquestionnaire. If H4 is correct, we should find that thesubjective contribution enabled by the MC toolkit hasan independent effect on the subjective value of theproduct beyond preference fit and perceived processcosts.

7.1.2. Measurement. As a manipulation check, weincluded the following two items to measure subjectsperceived contribution to the design (adapted fromSpreitzer 1995): I had a great deal of control overthe design process and I had a significant influ-ence over the outcome of the design process (1 =strongly disagree to 7 = strongly agree; alpha = 0.74)and averaged the scores. WTP was again measuredusing BDM auctions (dependent variable), with pricesin the urn ranging from 40 to 80 euros (again, sub-jects were not aware of this price range). In six cases,

the subjects WTP was higher than the price indicatedon the card drawn. In those cases, they purchasedthe watch for the price shown on the card (individ-ual watches obtained from the production companyhttp://www.wmctime.com). Preference fit (alpha =0.82) and perceived process costs (alpha = 0.75; control

variables) were measured on 7-point scales using thesame items as those employed in the previous studies.Because of the random group assignments, we did notfind significant differences between the groups regard-ing product interest and income (measured as in theprevious studies).

7.2. Findings and Discussion

7.2.1. Manipulation Check. As expected, stu-dents who could freely design the watch face(Group 2) reported higher levels of perceived con-tribution (M= 530; SD = 1.15) than participants inGroup 1, who only configured the watch using prede-

fined attributes (M= 404; SD = 1.39; F1 66 = 15752;p < 0001.

7.2.2. Findings. We find that an MC toolkit thatoffers high design freedom generates higher WTP(M= 3034; SD = 23.19) than a toolkit in which thedesign freedom is limited (M = 1921; SD = 10.22;F1 66 = 6502; p < 005 (Table 6). In line with the lit-erature, the subjects in Group 2 reported higher lev-els of preference fit (M= 549; SD = 0.99) than thosein Group 1 (M= 484; SD = 1.10; F1 66 = 6274; p

the “i designed it myself” effect in mass customization

Documents