<%BANNER%>

Best of Both Worlds? Consumer Inferences about the Benefits of Hybrid Products

Permanent Link: http://ufdc.ufl.edu/UFE0019761/00001

Material Information

Title: Best of Both Worlds? Consumer Inferences about the Benefits of Hybrid Products
Physical Description: 1 online resource (63 p.)
Language: english
Creator: Bilgin, Baler
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: attribute, category, hybrid, plausability, products, similarity, value, variability
Marketing -- Dissertations, Academic -- UF
Genre: Business Administration thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: A common form of product innovation involves combining two existing products to form a hybrid product. Hybrid products differ from other types of product innovations (e.g., incrementally new products, radically new products) in that they do not introduce novel benefits. Instead, a hybrid product portends to offer the best features of two existing constituent product categories, without the weaknesses of either. This research aims to identify conditions that influence a consumer's willingness to accept claims that the hybrid product will deliver a benefit that is characteristic of a constituent product category. Results from four experiments indicate that consumer acceptance of benefit claims about hybrid products depends on (1) the perceived distribution of attribute values within each constituent product category and, (2) the similarity between the constituent product categories.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Baler Bilgin.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Janiszewski, Chris A.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0019761:00001

Permanent Link: http://ufdc.ufl.edu/UFE0019761/00001

Material Information

Title: Best of Both Worlds? Consumer Inferences about the Benefits of Hybrid Products
Physical Description: 1 online resource (63 p.)
Language: english
Creator: Bilgin, Baler
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: attribute, category, hybrid, plausability, products, similarity, value, variability
Marketing -- Dissertations, Academic -- UF
Genre: Business Administration thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: A common form of product innovation involves combining two existing products to form a hybrid product. Hybrid products differ from other types of product innovations (e.g., incrementally new products, radically new products) in that they do not introduce novel benefits. Instead, a hybrid product portends to offer the best features of two existing constituent product categories, without the weaknesses of either. This research aims to identify conditions that influence a consumer's willingness to accept claims that the hybrid product will deliver a benefit that is characteristic of a constituent product category. Results from four experiments indicate that consumer acceptance of benefit claims about hybrid products depends on (1) the perceived distribution of attribute values within each constituent product category and, (2) the similarity between the constituent product categories.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Baler Bilgin.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Janiszewski, Chris A.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0019761:00001


This item has the following downloads:


Full Text





BEST OF BOTH WORLDS? CONSUMER INFERENCES ABOUT THE BENEFITS OF
HYBRID PRODUCTS




















By

BALER BILGIN


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2007
































O 2007 Baler Bilgin




























To my family and beloved uncle









ACKNOWLEDGMENTS

I would first like to thank my advisor, Chris Janiszewski, for his sincere commitment to

my development as a researcher. I am grateful to Chris not only because of his patient guidance

of my doctoral work over the years, but also because of his unrelenting dedication to marketing

research that will inspire me for the rest of my career as a marketing researcher. I would also like

to express my gratitude to my committee members Joe Alba, Lyle Brenner, Robyn LeBoeuf, and

Ira Fischler for their valuable advice on my work. I owe Lyle Brenner special thanks for his

significant contribution to my training by being a supportive co-author and a good friend. I thank

Alan Cooke for his insightful comments on my research proposals. I thank Gary Hunter and

Steve Taylor who encouraged me to pursue a PhD at the University of Florida.

I am deeply indebted to my parents, Nurten and Bahri; and my sister, Bahar; who have not

only wholeheartedly supported my decision to pursue a PhD in the U.S., but also have managed

to make me feel at home despite the 5,000 miles that separated us. Finally, I would like to thank

Kryslaine Lopes whose joyful presence brightened my days and motivated me to get back to

work after disappointing experiments. It frightens me to think how miserable my PhD years

would have been without her.













TABLE OF CONTENTS




page


ACKNOWLEDGMENTS .............. ...............4.....


LIST OF TABLES ................ ...............7............ ....


LIST OF FIGURES .............. ...............8.....


AB S TRAC T ......_ ................. ............_........9


CHAPTER


1 INTRODUCTION ................. ...............10.......... ......


2 NEW PRODUCT S .............. ...............12....


Communicating Hybrid Product Benefits .............. ...............13....
Structural Mapping .................. ........ .... .... .................. .............1

Category Plausibility Determines Hybrid Attribute Values ................. .....................16
Attribute Value Plausibility Determines Hybrid Attribute Values............... ................20


3 EXPERIMENT 1 .............. ...............23....


Method ................. ...............23.................

Design ................. ...............23.................
Procedure ................. ...............23.................
Stim uli .............. ...............25....
Predictions ................ ...............26.................
Re sults ................ ...............27.................
Discussion ................. ...............27.................


4 EXPERIMENT 2 ................ ...............29........... ....


Method ................. ...............29.................

Design ................. ...............29.................
Procedure ................. ...............30.................
R e sults................ ..... ........ ...............32.......

Manipulation Check .............. ...............32....
Analy si s ................. ...............3.. 2..............
Discussion ................. ...............33.................


5 EXPERIMENT 3A ................ ...............34................


M ethod ................... ........... ...............3.. 4....

Design and Stimuli .............. ...............34....













Procedure ................. ...............35.................
R e sults................ ..... ........ ...............36.......

Manipulation Check .............. ...............36....

Analy si s ................. ...............36.......... .....
Discussion ................. ...............37.................


6 EXPERIMENT 3B .............. ...............39....


M ethod ................... ........... ...............3 9....

Design and Stimuli .............. ...............39....
Procedure ................. ...............39........ ......
Re sults ................ ..... ...__. ................39......

Manipulation Check .............. ...............39....

Analy si s ................. ...............40._._._......


7 EXPERIMENT 4 ................. ...............42..............


Method ......__................. .......__. .........42

D esign .........._..... ............ ...............42....
Procedure and Stimuli .............. ...............42....

Re sults................ ...............43........ ......
Discussion ................. ...............44........ ......


8 GENERAL DI SCUS SSION ........._.__............ ...............46...


Theoretical Implications .............. ...............47....
Limitations ......__................. .......__. .........51

Managerial Implications .............. ...............52....
Future Research .............. ...............54....


APPENDIX EXPERIMENTAL STIMULI ......__................. .......__. ......... 5


LIST OF REFERENCES ......__................. .......__. .........5


BIOGRAPHICAL SKETCH .............. ...............63....










LIST OF TABLES

Table page

3-1 Experiment 1 design .............. ...............25....

3-2 Experiment 1 means............... ...............27.

4-1 Illustration of attribute presentation in Experiment 2 ................ .......... ................3 2











LIST OF FIGURES

FiMr page

2-1 Structural mapping paradigm............... ...............14

2-2 Categorization under uncertainty paradigm and hybrid products ................. ................. 15

2-3 Category plausibility (variability-based) judgment ................ .............................18

2-4 Distributions of a target attribute's values in two constituent categories with equal
variability. ............. ...............20.....

2-5 Value plausibility (di stribution-based) judgment. ................ ..............................21

4-1 Experiment 2 results. ............. ...............33.....

5-1 Experiment 3A procedure ................. ...............35........... ...

5-1 Experiment 3A results............... ...............37

6-1 Experiment 3B results ................. ...............40................

8-1 Experiment 4 results .............. ...............44....









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

BEST OF BOTH WORLDS? CONSUMER INFERENCES ABOUT THE BENEFITS OF
HYBRID PRODUCTS

By

Baler Bilgin

August 2007

Chair: Chris Janiszewski
Major: Business Administration

A common form of product innovation involves combining two existing products to form a

hybrid product. Hybrid products differ from other types of product innovations (e.g.,

incrementally new products, radically new products) in that they do not introduce novel benefits.

Instead, a hybrid product portends to offer the best features of two existing constituent product

categories, without the weaknesses of either. This research aims to identify conditions that

influence a consumer's willingness to accept claims that the hybrid product will deliver a benefit

that is characteristic of a constituent product category. Results from four experiments indicate

that consumer acceptance of benefit claims about hybrid products depends on (1) the perceived

distribution of attribute values within each constituent product category and, (2) the similarity

between the constituent product categories.









CHAPTER 1
INTTRODUCTION

A common form of product innovation involves combining two existing products to create

a hybrid product. A hybrid product is appealing because it uses the benefits of one product to

compensate for the deficits of the other product and vise-versa. For example, extended stay

hotels combine a hotel (convenient location, uncomfortable living space) and an apartment

(inconvenient location, comfortable living space) to create a product that has the convenience of

a hotel and the comfort of an apartment. Yet, not all hybrid products are successful. For

example, the Ecomobile is a combination of a car and a motorcycle. The Ecomobile claims to

combine the benefits of a car (e.g., weather protection) with the benefits of a motorcycle (e.g.,

low fuel consumption).2 The Ecomobile is just one example of a hybrid product that has won an

engineering award (others include BMW' s C1, NCR' s microwave bank), yet has experienced

limited commercial success.

One source of a hybrid product' s success may be the consumer' s willingness to accept that

the constituent categories' benefits can generalize to the hybrid product. Inferences about benefit

generalization will influence the consumer' s willingness to gather additional information (e.g.,

search, trial) about the hybrid product.3 At present, our understanding of the factors that

encourage benefit generalization is limited (but see Gregan-Paxton, Hoeffler, and Zhoa 2005).

Understanding why certain category benefits generalize to a hybrid product, whereas others do

not, might increase the manager' s ability to screen hybrid product ideas. Such insight might also




SOther existing examples include toaster ovens, sofa beds and camera phones.

2 Ecomobile is currently manufactured by the Swiss company Peraves (http://www.peraves.ch).

3 The term "benefit generalization" will refer to a consumer' s willingness to generalize the better one of two
constituent category attribute values to the hybrid product.









aid in the development of promotional strategies that encourage the generalization of benefits

from the constituent categories to the hybrid product.

This research investigates the potential influence of two processes that could determine the

consumer' s willingness to generalize benefits from the constituent categories to the hybrid

product. Category plausibility (i.e., an assessment of the product category that is most likely to

generate the hybrid product' s performance) and attribute value plausibility (i.e., an assessment of

the most likely hybrid attribute value given the distribution of attribute values in each constituent

product category attribute) are both legitimate strategies for benefit generalization. Four studies

show that attribute value plausibility guides benefit generalization. When there is a small amount

of overlap in the attribute value distributions of the constituent product categories, one of these

overlapping values is generalized to the hybrid product. When the overlap in the attribute values

of the constituent product categories is not diagnostic (i.e., no overlap, considerable overlap), the

perceived similarity of the constituent category distributions determines benefit generalization.

Chapter 2 discusses how hybrid products differ from other types of new products, and how these

differences may affect the processes underlying consumer inferences about hybrid products.









CHAPTER 2
NEW PRODUCTS

One of the most fundamental activities in marketing is the development and introduction of

new products. A number of strategies have been proposed to facilitate the successful

development of new product ideas. These strategies include brainstorming (Ulrich and Eppinger

1995; Srinivasan and Lovejoy 1997), morphological analysis (Urban and Hauser 1993), user

observation (Leonard and Rayport 1997), lead user analysis (von Hippel 2005), visual imagery

(Dahl, Chattopadhyay, and Gorn 1999), template application (Goldenberg, Mazursky, and

Solomon 1999), analogical thinking (Dahl and Moreau 2002), and product combination (Gregan-

Paxton et al. 2005).

Factors that influence the assessment of the value of a new product idea can be classified

under three broad categories. First, managers assess the market potential for the hybrid product

by considering consumer needs, as well as the size and growth rate of the target market. Second,

managers assess the feasibility of the proj ect from an engineering and marketing perspective.

Can the product be produced and delivered at a price that the target market will find appealing?

Is the technology required for production based on the firm's core competencies? Third,

managers assess the feasibility of a proj ect from a communication perspective. Can new product

benefits be effectively communicated to consumers? Will the consumer be able to appreciate the

benefits of the product? For example, the difficulty TiVo has experienced communicating its

benefits to consumers, as indicated by its rather slow adoption rate, illustrates the importance of

marketing communications in a new product' s success (Wathieu and Zoglio 2005).



It can thus be reasonably concluded that helping consumers understand and accept the

benefits of a new product is one of the largest obstacles to a successful new product introduction










(Hirshman 1980; Gatignon and Robertson 1985; Gregan-Paxton et al. 2002). This research

focuses on the communication aspect of hybrid product benefits to consumers. Specifically, I

investigate factors that marketers can use in their promotions to influence a consumer' s

expectation that a hybrid product will combine the best attributes of its constituent categories.

Communicating Hybrid Product Benefits

The effective communication of new product benefits is particularly difficult for products

that are a hybrid of existing products. Hybrid products are often positioned as the aggregate

benefits of the two existing products. Yet, the consumer' s acceptance of this positioning claim

requires a suspension of his/her beliefs about what defines one category or the other. For

example, consider the air freshening light bulbs produced by Ozonelite (ozonelite.com) and

Technical Consumer Products (www.fresh2.com). Consumers may find it difficult to believe that

an air freshener could light a room or that a light could remove odors. What follows is a review

of processes that have traditionally been used to explain consumer inferences about the benefits

of hybrid products.

Structural Mapping

The structural-mapping perspective (Gentner and Markman 1994, 1997) is commonly used

to predict how consumers will make inferences about the benefits of new products. The

structural-mapping perspective assumes that inferences about the characteristics of a new

product are made by drawing an analogy to a familiar, host category. Consumers access

information related to a host category, map properties of the host category onto the target

product, and make inferences about the target product's characteristics using a process of

knowledge transfer (Gentner and Markman 1994, 1997; Gregan-Paxton and John 1997; Moreau,

Markman, and Lehmann 2001). In general, people are more likely to map the properties of a









relational system that are relevant to the analogy (i.e., people map alignable features) (Gentner

and Markman 1997; Gregan-Paxton et al. 2002).

The structural mapping perspective has been particularly effective at predicting consumer

inferences about new-to-the-world products, which defy classification by virtue of offering novel

benefits or new uses for existing benefits. For example, prior to its introduction, a digital camera

was a radical innovation because consumers were not only unfamiliar with its benefits but also

with the product itself, making it rather difficult to categorize it to an existing product category.

In situations like these, consumers can make inferences about the benefits of new-to-the-world

products via analogical reasoning (Figure 2-1). An analogous host category is often selected

because it has benefit dimensions that align with the target product, hence the knowledge can be

easily transferred (Moreau et al. 2001). Marketers can try to influence the inference process by

suggesting a category (e.g., camera, scanner) that will encourage the most beneficial set of

inferences about the new product (e.g., digital camera) (Moreau et al. 2001; Yamauchi and

Markman 2000).

Host Catenory Tarnet Product

Benefit 1 ?

Benefit 2 -?

Figure 2-1. Structural mapping paradigm and product innovations introducing novel benefits.

The structural mapping perspective may be inappropriate for predicting consumer

inferences about hybrid products. Hybrid products often combine two existing products with

non-overlapping benefits, and thus do not involve introduction of new benefits nor new usage

contexts for known benefits. For example, miles per gallon (mpg) is an important attribute for a

car-motorcycle hybrid, yet consumers know this attribute well. Thus, the critical consumer

process in understanding a hybrid product is not learning its benefits but making inferences about









its attribute values. What is peculiar about the inference process in a hybrid product context is

that it requires reconciliation of contradictory information from the two constituent categories.

As a result, the process is unlikely to be one where knowledge is transferred from a single host

category to the target product. Instead, the process involves an inference about which of two

conflicting values is more plausible for a given attribute (Figure 2-2). Take the car-motorcycle

hybrid as an example. Cars perform poorly on the mpg attribute relative to motorcycles. The

question the consumer needs to answer is which of these two conflicting mpg values (i.e., good

or bad) the car-motorcycle hybrid will inherit.

Host Catenory A -Target Host Catenory B

Benefit 1 ? Deficit 1

Benefit 2 ? Deficit 2

Deficit 3 ? Benefit 3

Deficit 4 ? Benefit 4

Figure 2-2. Categorization under uncertainty paradigm and hybrid products.

Two competing processes may be responsible for consumer inferences about hybrid

product attribute values. First, it may be the case that consumers try to categorize the hybrid

product into the more plausible of two constituent categories on a given attribute and infer the

attribute value from the distribution of that category (i.e., category plausibility determines hybrid

attribute values). This process is similar to structural mapping, but allows a consumer to choose

the relevant constituent category for making inferences about each attribute. Second, it may be

that consumers try to determine a value that is most plausible given the distribution of values in

the two constituent categories (i.e., attribute value plausibility determines hybrid attribute

values). What follows is a discussion of each of these processes.









Category Plausibility Determines Hybrid Attribute Values

Judgments about the attribute values of a hybrid product could depend on a categorization

process. Evidence for this link comes from research in categorization under uncertainty (Rips

1989; Rips and Collins 1993; Smith and Sloman 1994; Sloman and Rips 1998). According to this

literature, categorizing an ambiguous obj ect as a member of category A or B is a judgment that

depends on two sources of information: (1) the relative variability of each constituent category

and (2) the similarity of the ambiguous obj ect to each constituent category. Rips (1989) has

found that people consider the relative variability of each constituent category to be the most

diagnostic information for a categorization judgment. For example, consider the quarter

category, the pizza category, and an ambiguous obj ect that is three inches in diameter. The

ambiguous obj ect is more similar to the quarter category than to the pizza category. Yet, given

the variability in quarter and pizza diameters, the ambiguous obj ect is more plausibly a pizza.

Smith and Sloman (1994) found that people use category variability, rather than similarity, to

categorize ambiguous stimuli only when their verbal protocols showed awareness of differential

variability across categories.

The impact of category variability on categorization judgments has also been studied in the

perceptual domain. A notable study by Fried and Holyoak (1984) has revealed that people are

sensitive to the relative variability of perceptual categories when they are making categorization

judgments. They showed that participants classified some checkerboard patterns that were more

similar to the prototypes (or mean) of the lower variability category as members of the high-

variability category. In a subsequent study, Stewart and Chater (2002) identified a boundary

condition for the use of variability-based categorization in perceptual classification. In their

experiments, Stewart and Chater used outline circles each with a single solid dot somewhere on

their circumference as their stimuli. The circles varied only in the position of the dot around the









circumference and this position was diagnostic of category membership. The low-variability

category distribution had a standard deviation of 110 and the high- variability category had a

standard deviation of 280. The critical exemplar fell in between the nearest members of these two

categories. Results showed that people used variability-based categorization to classify the

critical exemplar when the experimental procedure sensitized them to differences in the

variability of category members by presenting all members at the same time. When item

presentation was sequential, on the other hand, participants failed to notice differences in the

variability of the category members, making it impossible to use variability-based categorization.

In this condition, participants used similarity-based categorization. This finding parallels Smith

and Sloman' s (1994) finding that variability-based categorization precedes similarity-based

categorization when people notice differences in the relative variability of the candidate

categories on a critical dimension.

Inferring the value of a hybrid product attribute is similar to the categorization of an

ambiguous obj ect with one caveat. In a typical categorization under uncertainty task, the features

values of the to-be-categorized stimulus are known. Thus, the consumer can use these values to

assess whether the stimulus was drawn from the distribution of obj ects that represent category A

or B. In a hybrid product categorization task, the features values of the hybrid product are

unknown. Thus, the consumer can not use these values to categorize. Instead, the variability of

the distribution of obj ects that represent category A or B is potentially diagnostic. In line with

previous research (e.g., Rips 1989; Stewart and Chater 2001), people should expect that the

hybrid product attribute value is more likely to be drawn from the distribution of values in the










high variability category (Figure 2-3). As such, the hybrid should take on the mean / modal

value from this category.

Hla: When constituent product category variability is diagnostic, people are more

likely to generalize the modal value of the higher variability category to the

hybrid product.

A B A B







Plausible Plausible

Figure 2-3. Category plausibility (variability-based) judgment.

When differences in the variability of the two constituent categories are similar or difficult

to notice, differences in the category variability are not diagnostic for inferring hybrid product

attribute values. In this situation, people should rely on similarity information, as has been shown

by Smith and Sloman (1994; also see Stewart and Chater 2002). However, unlike a

categorization task, there is no similarity between the hybrid product and a constituent product

category because the hybrid product' s values are not known. Thus, the similarity between the

two constituent categories of a hybrid product becomes potentially diagnostic. As constituent

product categories become more similar, it becomes more plausible that the hybrid product can

bridge the benefit gap between the categories and provide the benefit (Figure 2-4).




SOne could argue that people infer the average of the two constituent category values for the hybrid product on a
given attribute. This averaging argument would not apply to discrete attributes, which require an inference of
whether an attribute is present or not. Also, the appeal of hybrid products for marketers lies in their ability to
combine the best attributes of two products in one product, and not so much in their ability to offer the average value
of the attributes of their constituent categories. Hence, from a substantive standpoint, it is important to understand
the factors that affect consumers' inferences regarding which of the two constituent category values the hybrid
product is likely to inherit on a given attribute.










Similar types of plausibility judgments have been observed in the induction literature (e.g.,

Osherson, Smith, Wilkie, Lopez, and Shafir 1990). For example, people find it more plausible

that an attribute of category A will generalize to category B as the perceived similarity between

the two categories increases. Further evidence for the role of similarity on plausibility judgments

comes from the conceptual combination literature. Wisniewski (1996, 1997) reported a number

of experiments showing that as the similarity of constituents of a noun-noun compound increased

(e.g., painter photographer), people tended to use a hybridization strategy to interpret the

combination, in which two concepts are combined so that each acquires properties of the other.

Wisniewski explained this finding using the alignment theory. That is, because highly similar

constituents tend to have more alignable features, people are more willing to transfer multiple

properties from both constituents when interpreting a compound (see Costello and Keane 1997,

2000 for an alternative account).2 Attesting to the difficulty of encouraging people to adopt a

hybridization strategy, several experiments have shown that participants rarely use this strategy

(e.g., only about 3% of the word compounds were interpreted using hybridization (Wisniewski

1997)).

Consistent with prior research, people should become more likely to accept that the hybrid

product' s performance is consistent with the better performing category when the two constituent

categories are perceived to be more similar.

Hlb: When constituent product category variability is non-diagnostic, people are more

likely to rely on the similarity of the constituent categories to make an inference

about a hybrid product' s performance. As similarity increases, people become


SUnlike Wisniewski who used different compounds to manipulate constituent similarity, the experiments reported
here keep the constituent categories of a hybrid product constant and manipulate their similarity between subjects.
By keeping constituent categories constant, these experiments control for the number of alignable attributes across
different levels of similarity. Hence, alignability does not explain the reported pattern of findings.









more likely to generalize the mean (modal) value of the constituent category that

has the highest performance level.

Dissimilar Categories Similar Categories
~~A B A







Not plausible Plausible

Figure 2-4. Distributions of a target attribute's values in two constituent categories with equal
variability.

Attribute Value Plausibility Determines Hybrid Attribute Values

Judgments about the attribute values of a hybrid product could depend on an assessment of

the plausibility of specific values. Consumers may assess the distributions of attribute values in

each constituent category and, based on the overlap in these distributions, make an inference

about the attribute value of the hybrid. Category distributions have been shown to influence

people's stimulus estimates. For example, Huttenlocher, Hedges, and Vevea (2000) varied the

distribution of stimuli in their experiments such that participants in one condition saw a uniform

distribution while participants in another saw a normal distribution. Huttenlocher et al. then

asked their participants to judge category membership for a new series of stimuli where the range

of values was extended to include stimuli outside the range of the initial set. Their results showed

that the probability of judging extreme values to be members fell off more rapidly for normal

than uniform distributions of the same range. This was because participants considered the fact

that there are fewer stimulus values near the tails of a normal vs. uniform distribution. This result

replicated when participants were asked to reproduce a stimulus (e.g., fish) after seeing it.

Participants' size estimations were more biased towards the category prototype when the










category distribution was uniform than normal. The authors argued that people use distributional

information automatically when categorizing stimuli and predicting their values.

A B A B







Plausible Plausible

Figure 2-5. Value plausibility (distribution-based) judgment.

Based on the evidence that people use distributional information in various category

judgments, I expect that constituent product category distributional information will be

diagnostic when there is a small amount of overlap in distributions (Figure 2-5). In these cases,

people should infer that the hybrid product will have the overlapping value, as it is most

plausible. More specifically, for each attribute, people compare the existing distributions from

the two constituents. When this comparison yields a small overlap (i.e., few observed attribute

values common to both constituents), the overlap becomes diagnostic and overlapping attribute

value is inferred to the hybrid product. For example, say, 99% of the cars possess airbags

whereas 100% of motorcycles do not. Because the overlap in this example is in the deficit region

(i.e., 1% of the cars and 100% of motorcycles do not possess airbags), people should infer

"deficit" on this attribute for the hybrid product.

H2a: When constituent product category distributions are diagnostic (i.e., distributions

marginally overlap), people are more likely to generalize the overlapping value.

When there is no overlap (i.e., no common attribute value), or significant overlap (i.e.,

numerous common attribute values) in constituent category distributions, then the overlap is not

diagnostic. In this case, people should use the distance between the means of the two constituent










category distributions to infer the hybrid product attribute value. As the mean values become

more similar, the differences between the means of the attribute distributions become smaller and

it becomes more plausible that the hybrid product can bridge the benefit gap and provide the

benefit. It is noteworthy that this hypothesis differs from Hypothesis lb in that people must be

able to compare distributions, not simply assess category similarity, for similarity to exert an

influence on inferences about hybrid product attribute values.

H2b: When constituent product category distributions are not diagnostic (i.e.,

distributions do not overlap, distributions significantly overlap), people are more

likely to rely on the similarity of the distributions to make an inference about a

hybrid product' s performance. As similarity increases, people become more likely

to generalize the mean (modal) value of the constituent category that has the

highest performance level.

These hypotheses are tested using four experiments. Experiment 1 is a direct test of

whether category plausibility/ versus attribute value plausibility guides people' s inferences about

hybrid product benefits. Experiment 2 manipulates constituent category similarity and the

diagnosticity of the value distributions independently to test the hypothesis that category

similarity influences people's hybrid product inferences when the value distributions are not

diagnostic. In Experiments 3A and 3B, accessibility of the distributional information is

manipulated to examine whether people use constituent category similarity when the

distributional information is inaccessible. Finally, Experiment 4 replicates findings from

Experiment 3A studies in a more ecologically valid experimental context.









CHAPTER 3
EXPERIMENT 1

Experiment 1 was an initial investigation into the processes that determine consumer

expectations about the attribute values of hybrid products. The challenge was to manipulate

constituent category attribute value distributions in a way that (1) allowed participants to assess

category plausibility (H1) and attribute value plausibility (H2) and (2) allowed each of these

hypotheses to make unique predictions. With this constraint in mind, the experiment involved an

independent manipulation of the perceived attribute value variability in the constituent category

that provided the benefit and in the constituent category that did not provide the benefit. This

allowed for independent manipulations of distribution variability and overlap.

Method

Design

The design was a deficit category variability (low, high) and benefit category variability

(low, high) by replicate order counterbalance factor (fast food and casual restaurant, car and

motorcycle, sports car and station wagon) by category order counterbalance factor (category

named first when describing the hybrid product) mixed design with only the first factor

manipulated within subj ects. The order in which the four different combinations of deficit

category variability (low, high) and benefit category variability (low, high) were presented was

determined by a Latin square design, which is illustrated in Table 3-1.

Procedure

Participants entered a behavior lab and were seated at personal computers. The instructions

stated that they would have to assess several new products (services) that were a combination of










two existing products (services). Participants were warned that the new products might seem

novel because these products were only recently introduced into the market. Then, a verbal

description of the first hybrid product was introduced (e.g., "The new service is a combination of

a fast food and a casual restaurant") followed by the presentation of the first attribute (e.g., fast

service). Each constituent category attribute distribution was described side-by-side using a

single sentence (e.g., "None of the casual restaurants have fast service."; "All of the fast food

restaurants have fast service."). While this information remained on the screen, the participant

was asked to predict the value of the hybrid product on this attribute by clicking on one of two

options provided (e.g., Yes No). After participants reported their predictions for the first

attribute, they repeated the procedure for the second attribute. There were four attributes to be

predicted for each hybrid product replicate, and participants had to predict all four of them before

the next hybrid product was considered. The key dependent measure was the proportion of the

four benefits generalized to the hybrid product.

The four treatment conditions were created by altering the order in which descriptions of

the attribute value distributions of the constituent product categories was presented. For example,

when attribute value variability was low in each constituent product category, it was indicated

that "none" of the deficit category members had the benefit, whereas "all" of the benefit category

members had the benefit (e.g., "None (all) of the casual (fast food) restaurants have fast

service."). These descriptors where changed to "none" and "most", "few" and "all", and "few"

and "most" for the subsequent attributes of the same replicate (Table 3-1 ).






SIn the remainder of the paper, I will use the term product to refer to both hybrid product replicates (e.g., vehicles)
and hybrid service replicates (e.g., restaurants).










Table 3-1: Experiment 1 design for the Casual Restaurant Fast Food Restaurant replicate
illustrating the order in which the constituent category variability information was
presented in each of the four conditions.

Constituent Category Variability Information
Casual Restaurant Fast Food Restaurant Replicate
Experimental High Quality Inviting
Condition Value-Priced Fast Service Food Ambiance
A None All None Most Few All Few -Most
B None Most Few All Few Most None All
C Few All Few -Most None All None Most
D Few Most None All None Most Few All

Stimuli

The hybrid products and their attributes were selected based on a series of protests, which

resulted in a total of nine hybrid product replicates that varied in the similarity of their

constituents to each other. These nine replicates were divided into three groups, each containing

three replicates, based on the degree of similarity between their constituents. In all experiments

except Experiment 4, three hybrid product replicates were used whose constituent categories

were moderately similar to each other. This would help reduce the chance of a floor or ceiling

effect when perceived similarity was manipulated in subsequent experiments. Second, protests

were used to select the sets of attributes used to test consumer inferences about the hybrid

product. Pretest participants listed the first Hyve benefits that came to mind for each constituent

category. These lists were used to identify three benefits for each constituent category that (1)

ranked high in popularity (i.e., listed frequently), (2) were an expected benefit in one of the

constituent categories but not the corresponding category. Two of these three benefits were used

in Experiment 1, which were selected randomly, due to design restrictions.2 All three attributes




2 The Latin square design employed in experiment 1 dictated that one attribute corresponded to each of the four
distribution description conditions (i.e., "none"-"most", "few"-"most", "few" -"all", "none" "all"), limiting the
number of attributes to four for each hybrid product replicate.










were used in the subsequent experiments that employed a different design. The hybrid product

replicates and benefits used in the experiments are listed in the Appendix.3

Predictions

Although the experiment is a two-by-two factorial design, the predictions of the category

plausibility hypothesis and the attribute value plausibility hypothesis do not align with main

effect or interaction tests. The category plausibility hypothesis (H1) predicts that benefits should

be more likely to generalize to the hybrid when the benefit category has more variability than the

deficit category, as opposed to vise-versa. Thus, there should be more benefit generalization in

the "none" / "most" condition than in the "few" / "all" condition. The attribute value plausibility

hypothesis (H2) predicts that benefits should be more likely to generalize to the hybrid when the

overlap in constituent product categories is in the benefit domain. Thus, there should be more

benefit generalization in the "few" / "all" condition than in the "none" / "most" condition.

The category plausibility hypothesis and value plausibility hypothesis make similar

predictions when constituent category variability is not diagnostic (i.e., "none" / "all" and "few"

/ "most" conditions). Benefit generalization should depend on the similarity of the distributions.

Because constituents with distributions of "few" / "most" have means closer to each other than

do constituents with distributions of"none" / "all," there should be more benefit generalization

in the "few" / "most" condition than in the "none" / "all" condition. Ninety undergraduate

students participated in the experiment in return for class credit.








3 A final pretest showed that all 12 attributes had different constituent attribute values for at least 75% of the
participants and that 10 of the 12 attributes had different constituent attribute values for at least 75% of the
participants. Further refinement of the attributes was limited by concerns about using unimportant benefits.










Results

The means for the generalization scores by condition are reported in Table 3-2. These

means were computed by averaging the proportion of benefits generalized in each of the four

category variability descriptions, collapsing over participants. Consistent with attribute value

plausibility hypothesis, participants generalized more benefits in the "few" / "all" (M~= .87)

condition than in the "none" / "most" condition (M~= .69, t (179) = 6.25, p < .001). Consistent

with both hypotheses, participants generalized more benefits in the "few" / "most" (M~= .83)

condition than in the "none" / "all" condition (M~= .70, t (179) = 4.47, p < .001). Supplemental

analyses showed that the replicate factor did not interact with any of the critical tests. The

presentation order of the constituent product categories in the hybrid product description (e.g., "a

combination of a fast food and casual restaurant" versus "a combination of a casual and fast food

restaurant") also did not interact with any of the critical tests.

Table 3-2: Experiment 1 means: The influence of constituent category attribute value variability
on benefit generalization.

Constituent category descriptions
Defieit category Benefit category Results
"None" "All" 0.70
"None" "Most" 0.69
"Few" "All" 0.87
"Few" "Most" 0.83

Discussion

Experiment 1 provides an initial test of the category plausibility hypothesis and the

attribute value plausibility hypothesis. The results are more consistent with the attribute value

plausibility hypothesis. People infer that a hybrid product will have a value that is common to the

overlap in the distributions of the constituent product categories. Experiment 1 also suggests that

people are sensitive to two types of distributional information. People use the overlap in

distributions to infer hybrid values when the overlap is diagnostic (i.e., small), but rely on the









similarity of the distributions when the overlap is not diagnostic (i.e., no overlap or significant

overlap).

The results of the first experiment are inconsistent with part of the category plausibility

hypothesis (Hla), but not the entire category plausibility hypothesis (Hlb). The category

plausibility hypothesis contends that (1) constituent category attribute value variability is

diagnostic of hybrid attribute values and (2) when attribute value variability is not diagnostic,

constituent category similarity is diagnostic. This hypothesis could be amended to argue that

category similarity is diagnostic for all benefit generalization judgments. For example, it may

have been the case that a "none" / "most" description of constituent category distributions

resulted in a perception of less category similarity than a "few" / "all" description of constituent

category distributions. As a consequence, the "few" / "all" condition resulted in more benefit

generalization.









CHAPTER 4
EXPERIMENT 2

Experiment 1 used distributional descriptors (e.g., "none" / "most") to manipulate the

distributional properties of the constituent product categories that comprise the hybrid product.

These distributional descriptors also created a concurrent manipulation of the similarity of the

constituent product categories that comprise the hybrid product. To determine whether the

similarity of the constituent product categories could be solely responsible for benefit

generalization, or whether attribute value plausibility is also diagnostic, Experiment 2

manipulated category similarity and the diagnosticity of the vahue distributions independently.

Method

Design

The design was an attribute value diagnosticity (diagnostic, not diagnostic) by constituent

category similarity (low, high) by hybrid product replicate (fast food and casual restaurant, car

and motorcycle, sports car and station wagon) by category order counterbalance factor mixed

design with only the replicate factor being manipulated within-subject. The order in which the

hybrid product replicates was presented was randomized. The attribute value diagnosticity

manipulation was the "none" / "most" (diagnostic) and "few" / "most" (non-diagnostic)

conditions. The "none" / "most" condition was diagnostic because it entails a small distributional

overlap in the deficit region. The "few" / "most" condition was non-diagnostic because the

distributional overlap is large and could be framed as either in the deficit or benefit region. The

constituent category similarity manipulation encouraged participants to perceive the constituent

product categories as more similar or different prior to making the benefit generalization

judgments.









Procedure

The procedure was similar to that of Experiment 1 with two exceptions. First, there was a

similarity manipulation. Before expressing expectations about the hybrid product, participants in

the low and high similarity conditions completed a categorization task that involved organizing

four categories into two pairs. Two of the four categories presented were the constituent

categories for the hybrid product while the other two were decoy categories. The two decoy

categories in the low similarity condition were selected so that they would be paired with one of

the constituent categories, thus reducing the perceived similarity between the constituents. For

example, in the fast food casual restaurant hybrid, the two decoy categories in the low

similarity condition were a sandwich restaurant and an upscale restaurant. If the fast food

restaurant is paired with the sandwich restaurant and the casual restaurant is paired with an

upscale restaurant, then the fast food restaurant and casual restaurant categories should be rated

less similar. The two decoy categories in the high similarity condition were a bakery and a coffee

shop. In this context, the fast food restaurant should be paired with the casual restaurant and the

categories should be rated more similar. After pairing the four categories, participants were

asked to explain in detail why they paired the specific items together. Participants then rated the

similarity of the constituent categories to each other.

The other change in the procedure involved the presentation of the constituent product

category attribute value distribution information. First, unlike Experiment 1, there were three

rather than two attributes for each constituent category. Second, in Experiment 1, each

constituent category attribute distribution was described using a single sentence. These sentences

were presented in pairs, one attribute at a time (e.g., "None (All) of the casual (fast food)

restaurants have fast service."). Since constituent category similarity was manipulated before the

attribute presentation in this experiment, the attribute value distribution description procedure










needed to be more efficient to maximize the effect of the similarity manipulation on participants'

predictions. Thus, in Experiment 2, all six attributes that had to be predicted were presented on

the same page (Table 4-1 illustrates the attribute presentation in Experiment 2).

After completing the categorization task used to manipulate constituent category

similarity, and before expressing expectations about the hybrid product, participants in the

diagnostic attribute value condition saw:

"Below, you see a distribution of the values of several attributes for two constituent

product categories. A "Most" means that most products sampled from the corresponding

category possess the attribute, while "None" means that none of the products sampled from the

corresponding category possess the attribute."

The respondents then saw a listing of the six attributes along with two column headings

that were the constituent product category names. Each entry in the column was "most" or

"none". For example, in the fast food causal restaurant replicate, the fast service, value-priced,

and drive-through service attributes were listed as "most" in the fast food category and "none" in

the casual restaurant category. Similarly, the inviting ambiance, high quality food, and attentive

service attributes were listed as "none" in the fast food category and "most" in the casual

restaurant category. The non-diagnostic value condition differed from the diagnostic value

condition only in the substitution of "few" for "none". Thus, each attribute was listed as "most"

for one constituent category and "few" for the other constituent category. Participants indicated

whether or not a hybrid product would possess each of the attributes while this information

remained on the screen. One hundred ninety-six undergraduate students participated in the

experiment in return for class credit.









Table 4-1: Illustration of attribute presentation in the non-diagnostic condition of Experiment 2
for the Fast Food Restaurant Casual Restaurant replicate.

AttriutesConstituent Categories
Fast Food Restaurant Casual Restaurant
Fast service Most Few
Value-priced Most Few
Drive-through service Most Few
Inviting ambiance Few Most
High quality food Few Most
Attentive Service Few Most

Results

Manipulation Check

The responses of those participants who paired the four categories in the expected fashion

for all three replicates were included in the analysis (n = 120). The constituent category

similarity ratings differed significantly for the low and high similarity conditions (Miow, = 3.05;

M~high= 3.92, F(1, 118) = 23.50, p < .001).

Analysis

The means for the generalization scores by condition are reported in Figure 4-1. The

predicted interaction between attribute value diagnosticity and constituent category similarity

was significant (F(1, 116) = 4.07, p < .05). When the constituent category value distributions

were diagnostic, participants were insensitive to the category similarity manipulation (M1ow =

.69; Mhigh = .71, F(1, 119) = .15, p = .69). When the constituent category value distributions

were non-diagnostic, participants increased their willingness to generalize benefits as constituent

category similarity increased from low to high (M1ow = .64; Mhigh = .76, F(1, 119) = 10.00, p <

.01). Supplemental analyses showed that the order of constituent category presentation did not

exhibit a main effect (F(1, 112) = .00, p = .98) or interact with similarity (F(1, 112) = .05, p =

.82), attribute value diagnosticity (F(1, 112) = 3.21, p = .08), or a combination of the two (F(1,

112) =.10, p =.75).











1-
S0.9
0.76
F90.8 -.7
0.69 0.71




a Diagnostic: "None"/"Most" Non-diagnostic:
"Few"/"Most"

Constituent Product Category Values

O Low Category Similarity High Category Similarity

Figure 4-1. Experiment 2 results: The influence of constituent category similarity and
distribution diagnosticity on expectations of constituent benefits in a hybrid product.

Discussion

The results of Experiment 2 provide further evidence that expectations about the benefits

of a hybrid product depend on the attribute values of the constituent categories. When the

attribute value distributions of the constituent categories had a diagnostic overlap, the common

value was generalized to the hybrid product. Importantly, this generalization process was not

sensitive to category similarity. When the attribute value distributions of the constituent

categories had a non-diagnostic overlap, participants relied on the similarity of the constituent

categories. Taken together, the first two experiments suggest that people anticipate hybrid

product performance by assessing the plausibility that a particular level of performance could be

achieved.









CHAPTER 5
EXPERIMENT 3A

If consumers are using information about the target attribute distributions of the constituent

product categories to make inferences about the performance of a hybrid product, then benefit

generalization judgments should be sensitive to the accessibility of the distributional information.

This prediction also follows from existing research that shows that people use distributional

information in a categorization task when they are aware of differences in the variability across

categories (Smith and Sloman 1994).

Experiments 1 and 2 used distributional information descriptors to ensure that participants

were cognizant of the distributional information (i.e., accessibility was guaranteed). In

Experiment 3, the salience of the distributional information was manipulated. In one condition,

participants experienced the "few" / "most" non-diagnostic condition procedure of Experiment 2.

In a second condition, participants saw the benefits of each constituent category product, but

these benefits were not accompanied by distribution information. Consistent with Hypothesis 2b,

similarity should only exert an influence on benefit generalizations when the distributional

information is salient.

Method

Design and Stimuli

The design was a distributional information accessibility (inaccessible, accessible) by

constituent category similarity (low, medium, high) by hybrid product replicate (fast food and

casual restaurant, car and motorcycle, sports car and station wagon) by category order

counterbalance factor (two orders) mixed design with the replicate factor manipulated within-

subj ect. The order in which the hybrid product replicates was presented was randomized. A










medium similarity condition (i.e., no constituent category similarity manipulation) was included

in the design as an additional control condition.

Procedure

The procedure was identical to Experiment 2. The distributional information accessibility

manipulation was achieved by altering the "few" / "most" non-diagnostic condition procedure of

Experiment 2 (i.e., accessible condition) to create the inaccessible condition. Participants in the

inaccessible distribution information condition were told, "Below you see the attributes of two

constituent categories." Each category listed the three attributes. For example, the fast food

category listed the fast service, value-priced, and drive-through service benefits. The casual

restaurant category listed the inviting ambiance, high quality food, and attentive service benefits.

The benefit lists were displayed side-by-side to discourage participants from thinking about

distributional information in each of the constituent product categories.

The accessible condition used the "few" / "most" non-diagnostic condition procedure of

Experiment 2. One hundred ninety-eight undergraduate students participated in the experiment in

return for class credit.

Constituent Categories
Fast Food Restaurant Casual Restaurant
Attributes
Fast service Inviting ambiance
Value-priced High quality food
Drive-through service Attentive Service

Figure 5-1. Experiment 3A procedure: Illustration of attribute presentation in the inaccessible
distributional information condition.














Manipulation Check

The responses of those participants who paired the four categories in the expected fashion

for all three replicates were included in the analysis (n = 145). The constituent category

similarity ratings were significantly different by similarity condition (2Mow = 3.03; 2Anedium

3.33, 2Migh= 4.12, F(2, 142) = 12.39, p < .001). Planned contrasts showed that the low and

medium similarity conditions did not differ (F(1, 142) = 1.96, p = .16), but that the medium and

high similarity conditions did differ (F(1, 142) = 13.43, p < .001). The similarity rating did not

depend on the accessibility of the distributional information (F(1, 133) = 0.01, p > .05) or an

interaction of the accessibility and category similarity information (F(2, 133) = .56, p > .05).

Analysis

The means for the generalization scores by condition are reported in Figure 5-1. The

predicted interaction between similarity and distributional information accessibility was

significant (F(2, 139) = 4.40 p .02). When distributional information was inaccessible,

participants did not increase their willingness to generalize attribute benefits as constituent

category similarity increased from low (2Mow = .70) to medium (2nediuin= .74; (F(1, 139) = 1.60,

p = .21) and from medium (2nedium = .74) to high (M~igh = .71, F(1, 139) = .78, p = .38). When

distributional information was accessible, participants increased their willingness to generalize

attribute benefits as constituent category similarity increased from low (2Mow,= .61) to medium

(2Enedium = .69, F(1, 139) = 4.50, p .04) and from medium (2nedim = .69) to high (M~igh = .77,


SIt is possible that the lack of a difference between the low and medium similarity conditions arose because of the
wording of the manipulation check question. Asking respondents in the low similarity condition to assess the
similarity of the constituent product categories may have encouraged them to focus on similarities, rather than the
dissimilarities thev just considered.


Results










F(1, 139) = 6.26, p < .02). Supplemental analyses showed that the order of constituent category

presentation did not exhibit a main effect (F(1, 133) = 1.19, p = .28) or interact with similarity

(F(2, 133) = .22, p = .80), alignability (F(1, 133) = .12, p = .73), or a combination of the two

(F(2, 133) = 2.98, p = .06). Finally, there was no difference between the accessibility conditions

in the moderate similarity condition (M~inaccessible = .74, Maccessible = .69; F(1, 144) = 1.91, p > .05).

This suggests that the accessibility manipulation did not alter the participants' implicit

assumptions about the constituent category distributions.




= 0.9
F9 0.77
S0.8 -0.74
0.70 0.71 0.69
0 0.7 -
0.6 -0.61

P~0.5
Inaccessible Accessible "Few"/"Most"

Non-diagnostic Distributional Information

O Low Similarity O Medium Similarity 5 High Similarity

Figure 5-1. Experiment 3A Results: The accessibility of constituent category distributional
information moderates the influence of category similarity on benefit generalization.

Discussion

The results of Experiment 3A suggest that consumers rely on constituent category

similarity to infer the value of a hybrid product attribute value when distributional information is

accessible, but not diagnostic. Unlike Experiment 2, in which salient but nondiagnostic

distributional information encouraged the use of constituent category similarity, participants

were not sensitive to the perceived similarity of the constituent product categories when

distributional information was not salient. These results provide further evidence that inferences









about hybrid product performance are sensitive to the attribute value distributions of the

constituent product categories.

One might be concerned that the side-by-side presentation of the product attributes in the

inaccessible distribution information condition encouraged participants to assume that the

distribution for the constituent categories was "none all" for the product attributes. If this is

indeed the case, then explicitly presenting the distribution information of "none all" should not

change the pattern of results observed in the inaccessible distribution information condition. If,

however, the side-by-side presentation of the stimuli successfully reduced the salience of the

distributions in Experiment 3A, then explicitly presenting the nondiagnostic "none all"

distributional information should encourage participants to use constituent similarity in

accordance with Hypothesis 2b.









CHAPTER 6
EXPERIMENT 3B

In this study, the distributional information accessibility manipulation was achieved by

explicitly providing the "none" / "all" non-diagnostic distribution information. This explicit

distribution presentation should encourage participants to rely on constituent category similarity

when making inferences about the hybrid product given the accessible but nondiagnostic nature

of the distribution information.

Method

Design and Stimuli

The design was a constituent category similarity (low, medium, high) by hybrid product

replicate (fast food and casual restaurant, car and motorcycle, sports car and station wagon) by

category order counterbalance factor (two orders) mixed design with the replicate factor

manipulated within-subj ect.

Procedure

The procedure was identical to the accessible distribution information condition in

Experiment 3A except that the "few" / "most" non-diagnostic category variability description

was replaced with the "none" / "all" non-diagnostic category variability description. Ninety-two

undergraduate students participated in the experiment in return for class credit.

Results

Manipulation Check

The responses of those participants who paired the four categories in the expected fashion

for all three replicates were included in the analysis (n = 70). The constituent category similarity

ratings were significantly different by similarity condition (2Mow, = 2.72; 2nedim = 3.34M~high=

3.79, F(2, 67) = 8.11, p < .001). Planned contrasts showed that the low and medium similarity










conditions did differ (F(1, 67) = 6.37, p .02), and that the medium and high similarity

conditions differed marginally (F(1, 67) = 13.43, p = .06).

Analysis

The means for the generalization scores by condition are reported in Figure 6-1. The

predicted main effect of similarity was significant (F(2, 64) = 3.73 p .03). As constituent

similarity increased from low (2Mow,= .67) to medium (2nedium = .72) to high (M~high = .78), people

became more willing to generalize attribute benefits to the hybrid product. Supplemental

analyses showed that the order of constituent category presentation did not exhibit a main effect

(F(1, 64) = .04, p = .85) or interact with similarity (F(2, 64) = .59, p = .56).

Cumulatively, the results of Experiments 3A and 3B furnish further support for the

hypothesis that non-diagnostic but accessible distribution information encourages participants to

rely on constituent category similarity when inferring hybrid product benefits.




0.9
F9 0.78
"o0.8 .7


06 -06
0.5
Accessible ("None"/"All")

Non-Diagnostic Distributional Information

O Low Similarity O Medium Similarity High Similarity


Figure 6-1. Experiment 3B Results: The accessibility of constituent category distributional
information moderates the influence of category similarity on benefit generalization.

A question that the results reported here do not address pertains to the informational inputs

people use when distributional information is inaccessible. For example, what information did










the participants in the inaccessible distribution information condition in Experiment 3A use to

infer hybrid product benefits? Although several processes may underlie these inferences,

Experiment 4 explored whether people's theories about new product success in the marketplace

govern their predictions in the absence of distribution information.









CHAPTER 7
EXPERIMENT 4

Experiments 2 and 3 manipulated the perceived similarity of the constituent product

categories independently of the categories themselves. This procedure was necessary so as to

unconfound the similarity and the attribute value distributions of the two constituent product

categories. A more ecologically valid approach to investigating these factors is to select hybrid

products that vary with respect to their constituent categories' similarity to each other.

Replicating the results of Experiment 3A, benefit generalization should increase as similarity

increases when distributional information is accessible, but not when distributional information

is inaccessible.

Method

Design

The design was a distributional information accessibility (inaccessible, accessible) by level

of constituent category similarity (low, medium, high) by hybrid product replicate (three

replicates per level of similarity) by category order counterbalancing factor (two levels) mixed

design with the constituent category similarity and replicate factors manipulated within-subj ect.

The order in which the nine hybrid product replicates was presented was randomized.

Procedure and Stimuli

The procedure was the same as in Experiment 3A, except that constituent product category

similarity was manipulated using different hybrid product replicates. After a series of protests

explained in Experiment 1, three hybrid products expected to have low constituent category

similarity (e.g., light bulb and air freshener, pen and calculator, restaurant and movie theater),

moderate constituent category similarity (e.g., fast food and casual restaurant, car and

motorcycle, sports car and station wagon), and high constituent category similarity (e.g., jet ski









and snowmobile, mountain bike and road race bike, TiVo service and Movies-on-Demand) were

selected. A pretest confirmed that the similarity (1= not similar at all, 7= very similar) of the

categories varied as intended (F(2, 38) = 52.40, p .001). The difference between the low

similarity (Miow, = 2. 17) and moderate similarity (Mmod =3.65; F(1, 19) = 35.3 p .001) and the

moderate similarity and high similarity (M~high = 4.82, F(1, 19) = 21.8, p .001) pairs was

significant. The same set of protests was also used to select the sets of attributes used to test

inferences about the hybrid product (see the Appendix for the complete list of hybrid product

replicates and their attributes).

Results

Ninety-two undergraduate students participated in the experiment in return for class credit.

The data were analyzed using a repeated measure MANOVA with similarity as a within-subj ect

factor and distributional accessibility as a between-subject factor. The means for this analysis are

reported in Figure 8-1. The predicted interaction between distributional information accessibility

and similarity was significant (F(2, 180) = 3.12, p .05). When distributional information was

inaccessible, participants did not increase their willingness to generalize the benefits as

constituent category similarity increased from low (M~= .70) to moderate (M~= .72; F(1, 41) =

0.85, p = .36), but did increase their willingness to generalize benefits as constituent category

similarity increased from moderate (M~= .72) to high (M~= .82; F(1, 41) = 25.06, p .001).

When distributional information was accessible, participants increased their willingness to

generalize benefits as constituent category similarity increased from low (M~= .65) to moderate

(M~= .74; F(1, 49) = 14.44, p .001) and from moderate (M~= .74) to high (M~= .82; F(1, 49 =

13.65, p .001). The order of constituent category presentation did not have a main effect (F(1,

88) = 2. 16, p = .15) or interact with constituent category similarity (F(2, 176) = .11, p = .90),










distributional information accessibility (F(1, 88) = 0.48, p = .49), or a combination of the two

(F(2, 176) = 0.0, p = .99).




o 0.9
a 0.82 0.82




Incesil Accessibl







Discussion cesil


attribute~No-dagosi vau istributional informatinwsacsile u o igo tic ninraen

constituent prdc aeory similarity resuled in a iincreaeinthe likelhoo ofgenealizin

thgue benfit fxpromen theut: h cesiiiy constituent categorie.We trbt au distributionalinomtnwa
not ccesible, an inces in onets theituent p oduc category similarity didno result ien ran zincras


in t e likelihoo ofgeer aldiziong theenefts foro the contribte uen clategorieaty lyowto s moderat







levels of similarity. The unexpected finding in the high similarity condition may be an artifact of

the hybrid product replicates (e.g., jet ski and snowmobile, mountain bike and road race bike,

TiVo service and Movies-on-Demand) used in this condition. Alternatively, it may be that highly

similar constituent categories make distributional information easy to access. In other words, the









information presentation format in the inaccessible condition can discourage the access of

attribute value distribution information, but it can not prevent access to this information.









CHAPTER 8
GENERAL DISCUSSION

Marketing research on new products has mainly focused on two types of product

innovations: enhancement and new-to-the-world products. This focus has been accompanied by

an emphasis on knowledge transfer as the process underlying the consumer' s learning of the

novel benefits that these product innovations introduce. Because knowledge transfer facilitates

consumer learning of novel product benefits by utilizing consumers' existing product knowledge,

it is an efficient and effective marketing tool to communicate the novel benefits of new products.

Not all new products offer novel benefits to consumers, however. For example, hybrid

products combine two existing products with known benefits to create a new product that offers

the best features of these two products, without the weaknesses of either. Because consumers

already know about the benefits of a hybrid product (e.g., mpg in the car-motorcycle hybrid),

knowledge transfer is unlikely to assist in their understanding of a hybrid product. The

consumer' s challenge with hybrid products is to determine which of the two conflicting

constituent category values to accept on an attribute. This research provides evidence for the

hypothesis that an attribute plausibility judgment resulting from a comparison of constituent

distributions drive people's hybrid product attribute value predictions. People compare the

existing distributions from the two constituents each time an attribute has to be predicted. If there

is a small distributional overlap, then people generalize the overlapping attribute value for the

hybrid product, which is the most plausible value for a distribution resulting from a combination

of the two constituent distributions. When there is no overlap, or significant overlap in

constituent category distributions, then the overlap is not diagnostic. In this case, people compare

the constituent category distributions to judge their similarity: the more similar the constituent

category distributions are, the more likely the hybrid product is to bridge the attribute gap and









generalize the benefit. Experiments 3A and 3B demonstrate that distributional information must

be salient in order for people to use attribute plausibility or distribution similarity in their benefit

generalization judgments. Experiment 4 uses real world hybrid products to manipulate

constituent category similarity and provides evidence supporting the attribute plausibility



Theoretical Implications

There are at least two important streams of psychology research that can shed light on how

consumers evaluate hybrid products: the categorization under uncertainty literature (e.g., Rips

1989; Smith and Sloman 1994; Murphy and Ross 1994) and the conceptual combination

literature (e.g., Costello and Keane 1997, 2000; Wisniewski 1997). Results reported here

corroborate Eindings from the categorization literature that people use distributional information

as input to their judgments only when they are aware of it (Smith and Sloman 1994) or when it is

made salient (Stewart and Chater 2002). In the current experiments, only when the experimental

procedure highlighted distributional information did participants in the current studies take it into

consideration when making inferences about the hybrid product.

Another set of Eindings from the categorization literature has shown that people exhibit a

strong tendency to base their inferences and predictions on a single category when faced with

categorization ambiguity (Malt, Ross, and Murphy 1995; Murphy and Ross 1994, 1999A; Ross

and Murphy 1996). Only under rare circumstances are people shown to use information from

more than one candidate category to make inferences about an ambiguous obj ect (i.e., multiple

category strategy; Moreau et. al. 2001; Gregan-Paxton et. al. 2005). For example, in marketing,

Moreau et al.'s (2001) participants used a multiple category strategy to make inferences about a

new product (digital camera) only when they were explicitly informed of the relationship

between the digital camera and the two candidate categories (film-based camera and a scanner).









In the context of the experiments reported here, a single category strategy would predict

that participants would generalize the benefit to the hybrid product only half the time since a

given constituent category performs better than the other only on half of the attributes. If benefit

generalization exceeded 50%, this would provide evidence for the multiple category strategy.

Because participants generalized benefits to the hybrid product more than half the time across all

experiments, current studies provide evidence for the use of multiple category strategy when

making inferences about the hybrid product. How can this finding be reconciled with the

robustness of the single category strategy in the literature?

One procedural aspect common to studies finding single category strategy is that the

presentation of one of the candidate categories precedes that of the other. For example, Moreau

et al. (2001) found that their participants used only the first category cued when making

predictions about the performance of the new product, thereby ignoring the alternative category

cued subsequently. The order of category presentation had an important influence on new

product evaluations even when explicit mappings from the two candidate categories were

provided. Specifically, 57% of subj ects who saw the camera ad first categorized the new digital

camera as a camera, compared to 3 1% of subj ects who saw the scanner ad first. In another study,

Ross and Murphy (1996) presented participants with a story containing a reference to a person

whose identity was uncertain. The text of the story (e.g., realtor) cued the person' s identity, but

an alternative identity was also subsequently cued (e.g., burglar). Participants were asked to

predict the probability that the ambiguous person would engage in certain category-consistent

and inconsistent behaviors (e.g., for burglar, pay attention to the sturdiness of the doors). Results

showed that the impact of the alternative category cued later in the text was limited to only those










questions highly associated with that category. That is, significant contextual support was

necessary to induce multiple category strategy.

The implication is that cuing alternative categories sequentially impedes people's ability to

use multiple category strategy by allowing them to structure the ambiguous obj ect according to a

single category. Once an initial representation of an obj ect is formed based on the first category

cued, restructuring it to include new category information becomes a challenging task. It is

possible that simultaneous representation of the constituent categories in the current experiments

discouraged participants from forming an early representation of the hybrid product according to

one of the constituent categories. This observation is supported by the finding that changing the

order in which constituent categories was presented did not influence the proportion of benefit

generalization in the current studies. Further evidence for this proposition comes from Gregan-

Paxton et al. (2005) who showed their participants the category cues at the same time and

observed multiple category strategy use.

Current results have implications for the conceptual combination literature as well. Several

models of conceptual combination have been recently proposed (e.g., Costello and Keane 2000;

Murphy 1988; Wisniewski 1997). Wisniewski's (1997) dual process model is one of the more

prominent of these models. An important pillar of Wisniewski's (1997) model is that combining

two concepts (e.g., cactus carpet) involves a structural alignment process by which people

compare the two concepts. According to this model, structural alignment underlies

interpretations of combinations whose concepts are highly similar to each other. Because highly

similar concepts tend to be more alignable than dissimilar concepts (Gentner and Markman

1997), they highlight the alignable differences that subsequently govern the combination's

interpretation. Consistent with this contention, Wisniewski (1997) showed that people are more









willing to transfer multiple properties from both constituents (i.e., hybridization) when the

constituents are highly similar to each other.

Given Wisniewski's results, the impact of constituent category similarity on hybrid

product evaluation evinced by the current set of experiments may come as no surprise. What is

intriguing, however, is that the between subj ects manipulation of constituent category similarity

in the experiments reported here controls for the degree of alignability between the constituent

categories. In other words, the degree of alignability between the constituent categories cannot

account for the increased benefit generalization at higher levels of similarity unless one argues

that the between subj ects similarity manipulation affected the degree of alignability between the

constituent categories in a consistent way. The implication is that although alignability is an

important determinant of hybridization, similarity between constituent categories may influence

the interpretation of the combination via a route other than alignability. The findings reported

here support the contention that similarity affects hybrid product evaluation by bringing its

constituent categories' distributions closer to each other. That is, because means of two

categories approach to one another on a given attribute as their perceived similarity increases, it

becomes more plausible that the hybrid product can successfully offer that attribute. This

differential route is plausible in a hybrid product context in which technical plausibility (i.e.,

whether it is technically feasible for the hybrid product to overcome the trade-off on an attribute)

interacts with conceptual plausibility (i.e., whether it is easy to imagine the hybrid product

possessing a given attribute) to influence hybrid product evaluations.

One intriguing finding was that participants used constituent category similarity to make

inferences about the hybrid product only when the procedure made category distributional

information salient. It is possible that there are other, perhaps more pervasive mechanisms than









constituent category similarity that people employ when making inferences about hybrid product

benefits. Unless its use is encouraged, as the current studies do by cuing distributional

information, constituent category similarity may be dominated by other sources of inputs to the

hybrid product inference process. One such mechanism can be people's intuitive theories about

the marketplace (Chernev and Carpenter 2001). Further research is needed to identify possible

mechanisms consumers may employ when distributional information is not salient.

Limitations

This research is subj ect to a limitation common to almost all consumer behavior research,

which is using an undergraduate student subj ect pool in experiments. Because this is a relatively

homogeneous group and not representative of the typical American consumer, it is difficult to

assess the generalizability of the reported results to the population.

Another limitation involved the dichotomous nature of the dependent variable employed in

the current experiments. Participants indicated whether or not the hybrid product would have

certain attributes of its constituent categories. This served to force participants to generalize the

value of either one of the constituent categories to the hybrid product, which allowed me to

directly analyze the impact of the independent variables on the choice of which constituent

category would drive hybrid product inferences. However, this aspect of the procedure possibly

limited the generalizability of the results in the following way. By forcing inferences, the

dependent measure may overstate the degree to which consumers in real life engage in

spontaneous inferences about hybrid product attributes. This concern is less valid for experiential

attributes whose value can only be inferred before one actually uses the product (e.g., how

thrilling it will be to ride a car-motorcycle hybrid) than for search attributes whose value is easier

to assess through search. Although the set of attributes used in the current experiments contained

experiential attributes, there were search attributes as well. This was an outcome of the criteria










employed that guided the selection of attributes in the pretest. Specifically, only those attributes

listed most frequently for each constituent category by pretest subj ects were selected to ensure

that the attributes used were important enough to encourage inference making in the real world.

The downside of using the most important attributes was that the Einal attribute list contained

both search and experiential attributes.

The extent of this limitation in generalizability is mitigated by the following factors. First,

consumer inferences about the hybrid product that occur prior to information search can affect

the likelihood of actually engaging in such search. Positive inferences about hybrid product

attribute values, for example, can increase the consumer' s willingness to gather additional

information about the product. Such initial inferences can also serve as expected performance

criteria for the hybrid product against which its real performance can be judged, which in turn

influences the consumer' s overall evaluation of the product. Finally, there may be search

attributes that managers may be unwilling to advertise to consumers due to hybrid product's low

performance on these attributes. It may be advantageous to have consumers infer the values of

these attributes for the hybrid product.

Furthermore, since the current experiments employed a dichotomous dependent measure,

one should be cautious in generalizing the Eindings to continuous attributes. As will be discussed

in the Future Research section, alternative processes may be available to consumers to infer the

value of a hybrid product on a continuous attribute.

Managerial Implications

From the manager's perspective, hybrid products will be successful to the extent that

consumers are willing to generalize benefits from both constituents to the hybrid product. Prior

research has shown that only in rare circumstances do consumers transfer knowledge from

multiple categories (e.g., Moreau et al. 2001). The implication is that encouraging consumers to









transfer knowledge from both constituent categories is a highly risky managerial proposition in

the case of hybrid products. Furthermore, such a promotional strategy may be ineffective with

hybrid products since they do not involve the learning of novel product benefits. A better

promotional strategy can thus be designed by exploiting the factors identified in this work that

affect consumers' inferences at the attribute level, which cumulatively determine the evaluation

of the hybrid product.

One such factor is constituent category similarity. Emphasizing the similarity between the

constituents of a hybrid product increases favorable inferences about it. As such, although it may

be tempting for hybrid product managers to promote the technical appeal of the hybrid product

by emphasizing how it combines two very dissimilar existing products, this research and

development focus in promotion may not induce the desired consumer response. Furthermore,

product communications should also involve attempts to increase the perceived similarity of the

constituent categories to each other. For example, juxtaposing the constituent products in print

ads (e.g., a car and a motorcycle) showing them perform the same function (e.g., transportation)

may increase favorable inferences about their combination. Put differently, using super-ordinate

category labels in product promotions that encompass both constituent categories may increase

the perceived similarity of the constituent categories to each other, resulting in more favorable

inferences about the hybrid product.

As the current experiments indicate, the role of constituent similarity is not

straightforward. Category distributional information must be salient for similarity to influence

hybrid product evaluations. Thus, it is important for hybrid managers to include such information

in their promotions. In order to do so, phrases that may cue distributional information such as the










ones used in current experiments (e.g., few, most, some) may be utilized in product

communication to convey attribute information (e.g., some motorcycles have airbags).

The second factor that affects benefit generalization is the degree of overlap of the

constituent categories on important attributes. The managerial implication is that those important

product attributes on which the two constituent categories have overlap in the benefit category

should be advertised to encourage benefit generalization. Again, the degree of overlap can be

emphasized using phrases that cue distributional information.

Finally, although not tested directly in this research, current results and existing research

provide sufficient evidence to suggest that advertising the constituent products simultaneously

(e.g., in a print ad) may increase the use of information from both constituent categories (i.e.,

multiple category strategy), resulting in more favorable inferences about the hybrid product.

Future Research

Existing marketing research paints a very broad brush of new products in general by

failing to consider the characteristics of new products that have conceptual implications for how

consumers evaluate them. By focusing on peculiar aspects of hybrid products, this research

shows that processes underlying the consumer evaluation of hybrid products are different from

those underlying other types of product innovations that have hitherto been investigated. Future

research will benefit from identifying peculiar aspects of different types of product innovations,

which will lead to a more nuanced and effective approach to studying the consumer evaluation of

new products.

One interesting extension of the current findings is examining whether simultaneous

presentation of category cues encourages a multiple category strategy. It is also worthwhile

investigating how the length of time after structuring a new product representation on the basis

of one category can influence people's propensity to use multiple category strategy. Existing









research shows that people ignore the second category cued a few minutes after the first category

cue. Will this strong tendency to use single category strategy become more or less pronounced

over time (e.g., presenting the second category cue one day after the presentation of the first

one)?

Additionally, the Theoretical Implications section distinguished between technical and

conceptual plausibility in the hybrid product concept. Conceptual plausibility is necessary but

not sufficient to understand consumer inferences about hybrid products. That is, it may be

conceptually plausible for a hybrid product to have a certain feature (e.g,. a motorcycle with

retractable wheels that can be used when slowing or stopping) yet not technically plausible to

effectively offer it (e.g., whether retractable wheels will work effectively). Results from the

current studies support this distinction by showing that it is possible to influence technical

plausibility while keeping attribute alignability constant, a factor that has been shown to

influence conceptual plausibility (Wisniewski 1997). More research is needed however to

establish that these two plausibility judgments indeed rely on different processes.

As mentioned before, current experiments employed a dichotomous dependent measure. It

is possible that this limited the processes available to consumers when evaluating hybrid

products. It will be interesting to investigate alternative processes that may underlie consumer

inferences when the dependent measure is a continuous attribute. It is plausible that an averaging

model may approximate people' s inferences in this case (Anderson 1967). Given two constituent

categories, consumers may simply average the mean values of the two categories on an attribute

to predict the value for their combination. The interesting question is under what circumstances

will consumers diverge from this effortless yet possibly error prone averaging strategy? What

factors will encourage the use of a weighted average model in which the two constituent










categories differentially contribute to what the hybrid product' s value will be on a given

attribute? Given the theoretical and managerial importance of how consumers evaluate hybrid

products, I believe these questions warrant further research.









APPENDIX
EXPERIMENTAL STIMULI
















Low Constituent Category Similarity Hybrid Products
Pen / Calculator (2.25) Light bulb / Air freshener (2.00) Restaurant / Movie theater (2.25)
Pen Calculator Light bulb Air freshener Restaurant Movie theater
Disposable Programmable Lasts about 5000 hours ^` Can be refilled Gourmet food Dark
Inexpensive Graphing feature Provides pure light Circulates fragrance Easy to converse Stadium seating
Wide selection of Outstanding
Super-fine tip Has memory Reveals natural colors Consistent frag. delivery food sound system
Medium Constituent Category Similarity Hybrid Products
Fast food / Casual restaurant (3.85) Car / Motorcycle (3.55) Sports car / Station wagon (3.55)
Fast food Casual restaurant Car Motorcycle Sports car Station wagon
Value-priced' High quality food Weatherproof Low fuel consumption Aerodynamic Ample trunk space
High parking
Fast Service La, as a, ambiance Airbag convenience Low weight Low engine noise
Drive-through
service Attentive service Air-conditioning Low Emissions Fast Family Car
High Constituent Category Similarity Hybrid Products
Jetski / Snowmobile (4.4) Mountain bike / Race bike (5.15) TiVo / Movies-on-Demand (4.9)
Movies-on-
Jetski Snowmobile Mountain bike Race bike TiVo Demand
Rides on water Heated seats All-terrain Fast ^` Pause broadcasts New releases
Life jacket Head and tail lights
compartment ^\ Rugged Weight ^` Records shows Pay-per-view ^`
3600 spins in its own Many hours of Instant access to
length^/ Snow beams Shock absorbers Aerodynamic ^` recording time movie library
^` Over 25% of respondents thought both constituents possessed benefit.


Only the attributes in italics were used in experiment 1.









LIST OF REFERENCES


Anderson, Norman H. (1967), "Averaging Model Analysis of Set-Size Effect in Impression
Formation," Journal ofExperintental Psychology, 75 (October), 158-165.

Chemnev Alexander and Gregory S. Carpenter (2001), "The Role of Market Efficiency Intutions
in Consumer Choice: A Case of Compensatory Inferences," Journal of2arketing
Research, 38 (August), 349-361.

Costello, Fintan J. and Mark T. Keane (1997), "Polysemy in Conceptual Combination: Testing
the Constraint Theory of Combination," in Nineteenth Annual Conference of the Cognitive
Science Society, Hillsdale, NJ: Erlbaum.

Costello, Fintan J.and Mark T. Keane (2000), "Efficient Creativity: Constraint Guided
Conceptual Combination," Cognitive Science, 24 (April/June), 299-349.

Dahl, Darren W. and Page Moreau (2002), "The Influence and Value of Analogical Thinking
During New Product Ideation," Journal of2arketing Research, 39 (February), 47-60.

Dahl, Darren W., Amitava Chattopadhyay, and Gerald J. Gomn (2001), "The Importance of
Visualisation in Concept Design", Design Studies, 22 (January), 5-26.

Fried, Lisbeth S. and Keith .J. Holyoak (1984), "Induction of Category Distributions: A
Framework for Category Leamning," Journal ofExperintental Psychology: Lea-l~rning.
M~enory and Cognition, 10 (April), 234-257.

Gatignon, Hubert and Thomas S. Robertson (1985), "A Propositional Inventory for New
Diffusion Research," Journal of Consumer Research, 11 (March), 849-867.

Gentner, Dedre and Arthur B. Markman (1994), "Structural Alignment in Comparison: No
Difference without Similarity," Psychological Science, 5 (May), 152-158.

Gentner, Dedre and Arthur B. Markman (1997), "Structure Mapping in Analogy and Similarity,"
American Psychologist, 52 (January), 45-56.

Goldenberg, Jacob, David Mazursky, and Sorin Solomon (1999), "Toward Identifying the
Inventive Templates of New Products: A Channeled Ideation Approach," Journal of
Marketing Research, 36 (May), 200-210.

Gregan-Paxton, Jennifer and Deborah Roedder John (1997), "Consumer Leamning by Analogy: A
Model of Internal Knowledge Transfer," Journal of Consumer Research, 24 (December),
266-284.

Gregan-Paxton, Jennifer (2001), "The Role of Abstract and Specific Knowledge in the
Formation of Product Judgments: An Analogical Learning Perspective, Journal of
Consumer Psychology, 11 (3), 141-158.










Gregan-Paxton, Jennifer, Jonathan D. Hibbard, Frederic F. Brunel, and Pablo Azar (2002), "' So
That' s What That Is': Examining the Impact of Analogy on Consumers' Knowledge
Development for Really New Products," Psychology and Marketing, 19 (June), 533-550.

Gregan-Paxton, Jennifer, Steve Hoeffler, and Min Zhao (2005), "When Categorization is
Ambiguous: Factors that Facilitate the Use of a Multiple Category Inference Strategy,"
Journal of Consumer Psychology, 15 (2), 127-140.

Gregan-Paxton, Jennifer and Page Moreau (2003), "How Do Consumers Transfer Existing
Knowledge? A Comparison of Analogy and Categorization Effects," Journal of Consumer
Psychology, 13 (4), 422-430.

Hirshman, Elizabeth (1980), "Innovativeness, Novelty Seeking, and Consumer Creativity,"
Journal of Consumer Research, 7 (June), 283-295.

Huttenlocher, Janellen, Larry V. Hedges, and Jack L. Vevea, (2000), "Why Do Categories Affect
Stimulus Jud gment?" Journal ofExperintental Psychology: General, 129 (June), 220-241.

Leonard, Dorothy and Jeffrey F. Rayport (1997), "Spark Innovation through Empathic Design,"
Harvard'Business Review, 75 (November/December), 102-113.

Malt, Barbara C., Brian H. Ross, and Gregory L. Murphy (1995), "Predicting Features for
Members of Natural Categories When Categorization is Uncertain," Journal of
Experimental Psychology: Lea-l~rning. Mentory, and' Cogrition, 21 (May), 646-661.

Moreau, C. Page, Arthur Markman, and Donald R. Lehmann (2001), "'What Is It?'
Categorization Flexibility and Consumers' Responses to Really New Products," Journal of
Consumer Research, 37 (March), 489-498.

Murphy, Gregory L. (1988), "Comprehending Complex Concepts," Cogritive Science, 12 (4),
529-562.

Murphy, Gregory L. and Brian H. Ross (1994), "Predictions from Uncertain Categorizations,"
Cogritive Psychology, 27 (October), 148-193.

Murphy, Gregory L. and Brian H. Ross (1999), "Induction with Cross-Classified Categories,"
M~enory & Cogrition, 27 (November), 1024-1041.

Murphy, Gregory L. and Brian H. Ross (2005), "The Two Faces of Typicality in Category-Based
Induction," Cogrition, 95 (March), 175-200.

Osherson, Daniel N., Edward E. Smith, Ormond Wilkie, Alej andro LC~pez, and Eldar Shafir
(1990), "Category-Based Induction", Psychological Review, 97 (April), 185-200.

Park, Whan C. and Daniel C. Smith (1989), "Product-Level Choice: A Top-Down or Bottom-Up
Process," Journal of Consumer Research, 16 (December), 289-299.










Rips, Lance J. (1989), "Similarity, Typicality, and Categorization", in S. Vosniadou & A. Ortony
(Eds.), Similarity andAnalogicalRea;soning, 21-59. Cambridge, MA: Cambridge
University Press.

Rips, Lance J. and Allan Collins (1993), "Categories and Resemblance," Journal of
Experimental Psychology: General, 122 (December), 468-486.

Ross, Brian H., and Gregory L. Murphy (1996), "Category-Based Predictions: Influence of
Uncertainty and Feature Associations," Journal ofExperintental Psychology: Lea-l~rning.
M~entoy and Cognition, 22 (May), 736-53.

Sloman, Steve A. and Lance J. Rips (1998), "Similarity as an Explanatory Construct," Cognition,
65 (January), 87-101.

Smith, Edward E., and Steven A. Sloman (1994), "Similarity- versus Rule-Based
Categorization," M~entoy & Cognition, 22 (June), 377-386.

Srinivasan, V. and William S. Lovej oy (1997), "Journal of2arketing Research, 34 (February),
154-163.

Stewart, Neil and Nick Chater (2002), "The Effect of Category Variability in Perceptual
Categorization, "Journal ofExperintental Psychology: Lea-l~rning. Mentory, and Cognition,
28 (September), 893-907.

Takashi, Yamauchi and Arthur B. Markman (2000), "Inference Using Categories," Journal of
Experimental Psychology: Lea-l~rning. Mentory, and Cognition, 26 (May), 776-795.

Ulrich, Karl T. and Steven D. Eppinger (1995), Product Design and Development. New York:
McGraw-Hill Book Company.

Urban, Glen L. and John R. Hauser (1993), Design and2arketing of New Products, 2nd edition,
Englewood Cliffs, NJ: Prentice Hall.

Verde, Michael F., Gregory L. Murphy, and Brian H. Ross (2005), "Influence of Multiple
Categories on the Prediction of Unknown Properties," M~entoy & Cognition, 33 (April),
479-487.

Viswanathan, Madhubalan and Terry L. Childers (1999), "Understanding How Product
Attributes Influence Product Categorization: Development and Validation of Fuzzy Set-
Based Measures of Gradedness in Product Categories," Journal of2arketing Research, 36
(February), 75-94.

von Hippel, Eric (2005), Dentocratizing Innovation, Cambridge, MA: MIT Press.

Wathieu, Luc and Michael Zoglio (2005), Harvard Business School Case Study, October, (No:
9-501-038).










Wisniewski, Edward J. (1996), "Construal and Similarity in Conceptual Combination," Journal
of2~emory and Language, 35 (June), 434-453.

Wisniewski, Edward J. (1997). "When Concepts Combine", Psychonomic Bulletin and Review, 4
(December), 167-183.









BIOGRAPHICAL SKETCH

Baler Bilgin earned his bachelor degree in political science and international relations from

Bogazici University in Istanbul, Turkey. In 2000, he moved to the United States to pursue a

Master of Business Administration at the Illinois State University in Normal-Bloomington. After

completing his MBA, Baler entered the Ph.D. program in marketing at the University of Florida.

He joined the University of California-Riverside as an Assistant Professor of marketing in July

2007.





PAGE 1

BEST OF BOTH WORLDS? CONSUMER IN FERENCES ABOUT THE BENEFITS OF HYBRID PRODUCTS By BALER BILGIN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007 1

PAGE 2

2007 Baler Bilgin 2

PAGE 3

To my family and beloved uncle 3

PAGE 4

ACKNOWLEDGMENTS I would first like to thank my advisor, Chris Janiszewski, for his sincere commitment to my development as a researcher. I am grateful to Chris not only because of his patient guidance of my doctoral work over the years, but also be cause of his unrelenting dedication to marketing research that will inspire me for the rest of my career as a marketing researcher. I would also like to express my gratitude to my committee memb ers Joe Alba, Lyle Brenner, Robyn LeBoeuf, and Ira Fischler for their valuable advice on my work. I owe Lyle Brenner special thanks for his significant contribution to my training by being a supportive co-author and a good friend. I thank Alan Cooke for his insightful comments on my research proposals. I thank Gary Hunter and Steve Taylor who encouraged me to pursue a PhD at the University of Florida. I am deeply indebted to my parents, Nurten and Bahri; and my sister Bahar; who have not only wholeheartedly supported my decision to purs ue a PhD in the U.S., but also have managed to make me feel at home despit e the 5,000 miles that separated us. Finally, I would like to thank Kryslaine Lopes whose joyful presence brightened my days and motivated me to get back to work after disappointing experiments. It fright ens me to think how miserable my PhD years would have been without her. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT.....................................................................................................................................9 CHAPTER 1 INTRODUCTION................................................................................................................. .10 2 NEW PRODUCTS.................................................................................................................12 Communicating Hybrid Product Benefits..............................................................................13 Structural Mapping..........................................................................................................13 Category Plausibility Determines Hybrid Attribute Values............................................16 Attribute Value Plausibility Dete rmines Hybrid Attribute Values..................................20 3 EXPERIMENT 1................................................................................................................. ...23 Method....................................................................................................................................23 Design..............................................................................................................................23 Procedure.........................................................................................................................23 Stimuli........................................................................................................................ .....25 Predictions.......................................................................................................................26 Results.....................................................................................................................................27 Discussion...............................................................................................................................27 4 EXPERIMENT 2................................................................................................................. ...29 Method....................................................................................................................................29 Design..............................................................................................................................29 Procedure.........................................................................................................................30 Results.....................................................................................................................................32 Manipulation Check........................................................................................................32 Analysis....................................................................................................................... ....32 Discussion...............................................................................................................................33 5 EXPERIMENT 3A................................................................................................................ .34 Method....................................................................................................................................34 Design and Stimuli..........................................................................................................34 5

PAGE 6

6 Procedure.........................................................................................................................35 Results.....................................................................................................................................36 Manipulation Check........................................................................................................36 Analysis....................................................................................................................... ....36 Discussion...............................................................................................................................37 6 EXPERIMENT 3B................................................................................................................ .39 Method....................................................................................................................................39 Design and Stimuli..........................................................................................................39 Procedure.........................................................................................................................39 Results.....................................................................................................................................39 Manipulation Check........................................................................................................39 Analysis....................................................................................................................... ....40 7 EXPERIMENT 4................................................................................................................. ...42 Method....................................................................................................................................42 Design..............................................................................................................................42 Procedure and Stimuli.....................................................................................................42 Results.....................................................................................................................................43 Discussion...............................................................................................................................44 8 GENERAL DISCUSSION.....................................................................................................46 Theoretical Implications....................................................................................................... ..47 Limitations.................................................................................................................... ..........51 Managerial Implications........................................................................................................ .52 Future Research......................................................................................................................54 APPENDIX EXPERIMENTAL STIMULI.............................................................................57 LIST OF REFERENCES...............................................................................................................59 BIOGRAPHICAL SKETCH.........................................................................................................63

PAGE 7

LIST OF TABLES Table page 3-1 Experiment 1 design........................................................................................................ ..25 3-2 Experiment 1 means......................................................................................................... ..27 4-1 Illustration of attribute presentation in Experiment 2........................................................32 7

PAGE 8

LIST OF FIGURES Figure page 2-1 Structural mapping paradigm.............................................................................................14 2-2 Categorization under uncertainty paradigm and hybrid products......................................15 2-3 Category plausibility (v ariability-based) judgment...........................................................18 2-4 Distributions of a target attributes values in two cons tituent categories with equal variability...........................................................................................................................20 2-5 Value plausibility (distribution-based) judgment..............................................................21 4-1 Experiment 2 results....................................................................................................... ...33 5-1 Experiment 3A procedure..................................................................................................35 5-1 Experiment 3A results...................................................................................................... ..37 6-1 Experiment 3B results...................................................................................................... ..40 8-1 Experiment 4 results....................................................................................................... ...44 8

PAGE 9

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy BEST OF BOTH WORLDS? CONSUMER IN FERENCES ABOUT THE BENEFITS OF HYBRID PRODUCTS By Baler Bilgin August 2007 Chair: Chris Janiszewski Major: Business Administration A common form of product innovation involves combining two existing products to form a hybrid product. Hybrid products differ from other types of pr oduct innovations (e.g., incrementally new products, radically new products ) in that they do not introduce novel benefits. Instead, a hybrid product portends to offer the best features of two existing constituent product categories, without the weaknesses of either. This research aims to identify conditions that influence a consumers willingness to accept claims that the hybrid product will deliver a benefit that is characteristic of a c onstituent product category. Results fr om four experiments indicate that consumer acceptance of benefit claims a bout hybrid products depends on (1) the perceived distribution of attribute values within each cons tituent product category and, (2) the similarity between the constituent product categories. 9

PAGE 10

CHAPTER 1 INTRODUCTION A common form of product innovation involves combining two existing products to create a hybrid product. A hybrid product is appealing b ecause it uses the benefits of one product to compensate for the deficits of the other produc t and vise-versa. For example, extended stay hotels combine a hotel (convenient location, uncomfortable living space) and an apartment (inconvenient location, comfortable living space) to creat e a product that has the convenience of a hotel and the comfort of an apartment.1 Yet, not all hybrid products are successful. For example, the Ecomobile is a combination of a car and a motorcycle. The Ecomobile claims to combine the benefits of a car (e .g., weather protection) with the benefits of a motorcycle (e.g., low fuel consumption).2 The Ecomobile is just one example of a hybrid product that has won an engineering award (others incl ude BMWs C1, NCRs microwav e bank), yet has experienced limited commercial success. One source of a hybrid products success may be the consumers willingness to accept that the constituent categories benefits can generalize to the hybrid pr oduct. Inferences about benefit generalization will influence the consumers willingness to gather additional information (e.g., search, trial) about the hybrid product.3 At present, our understand ing of the factors that encourage benefit generalization is limited ( but see Gregan-Paxton, Hoeffler, and Zhoa 2005). Understanding why certain category benefits gene ralize to a hybrid product, whereas others do not, might increase the managers ability to screen hybrid product ideas. Such insight might also 1 Other existing examples include toaster ovens, sofa beds and camera phones. 2 Ecomobile is currently manufactured by the Swiss company Peraves (http://www.peraves.ch). 3 The term benefit generalization will refer to a cons umers willingness to generalize the better one of two constituent category attribute values to the hybrid product. 10

PAGE 11

11 aid in the development of promotional strategies that encourage the gene ralization of benefits from the constituent categories to the hybrid product. This research investigates the potential influe nce of two processes that could determine the consumers willingness to generalize benefits fr om the constituent categories to the hybrid product. Category plausibility (i.e., an assessment of the produc t category that is most likely to generate the hybrid prod ucts performance) and attribute value plausibility (i.e., an assessment of the most likely hybrid attribute value given the distribution of attribute valu es in each constituent product category attribute) are both legitimate stra tegies for benefit gene ralization. Four studies show that attribute value plausi bility guides benefit generalizati on. When there is a small amount of overlap in the attribute valu e distributions of the constituen t product categories, one of these overlapping values is generalized to the hybrid product. When the overlap in the attribute values of the constituent product categories is not diagno stic (i.e., no overlap, cons iderable overlap), the perceived similarity of the constituent category distributions determines benefit generalization. Chapter 2 discusses how hybrid products differ from other types of new products, and how these differences may affect the pr ocesses underlying consumer infe rences about hybrid products.

PAGE 12

CHAPTER 2 NEW PRODUCTS One of the most fundamental activities in mark eting is the developmen t and introduction of new products. A number of strategies have been proposed to facilitate the successful development of new product ideas. These strategi es include brainstormi ng (Ulrich and Eppinger 1995; Srinivasan and Lovejoy 1997), morphological analysis (Urb an and Hauser 1993), user observation (Leonard and Rayport 1997), lead user analysis (von Hippe l 2005), visual imagery (Dahl, Chattopadhyay, and Gorn 1999), templa te application (Goldenberg, Mazursky, and Solomon 1999), analogical thi nking (Dahl and Moreau 2002), and product combination (GreganPaxton et al. 2005). Factors that influence the asse ssment of the value of a new product idea can be classified under three broad categories. First, managers assess the market potential for the hybrid product by considering consumer needs, as well as the size and growth rate of the target market. Second, managers assess the feasibility of the project fr om an engineering and marketing perspective. Can the product be produced and delivered at a price that the target market will find appealing? Is the technology required for production base d on the firms core competencies? Third, managers assess the feasibility of a project fr om a communication perspective. Can new product benefits be effectively communicated to consumers? Will the consumer be able to appreciate the benefits of the product? For example, the difficulty TiVo has experi enced communicating its benefits to consumers, as indicated by its rather slow adoption rate, illu strates the importance of marketing communications in a new products success (Wathieu and Zoglio 2005). It can thus be reasonably concluded th at helping consumers understand and accept the benefits of a new product is one of the larges t obstacles to a successf ul new product introduction 12

PAGE 13

(Hirshman 1980; Gatignon and Robertson 1985; Gr egan-Paxton et al. 2002). This research focuses on the communication aspect of hybrid pr oduct benefits to consumers. Specifically, I investigate factors that market ers can use in their promotions to influence a consumers expectation that a hybrid product will combine the best attributes of its constituent categories. Communicating Hybrid Product Benefits The effective communication of new product benef its is particularly difficult for products that are a hybrid of existing products. Hybrid products are often posit ioned as the aggregate benefits of the two existing products. Yet, the consumers acceptance of this positioning claim requires a suspension of his/her beliefs about what defines one category or the other. For example, consider the air freshening light bulbs produced by Ozonelite (ozonelite.com) and Technical Consumer Products (www.fresh2.com). Consumers may find it difficult to believe that an air freshener could light a room or that a lig ht could remove odors. What follows is a review of processes that have traditionally been used to explain consumer infere nces about the benefits of hybrid products. Structural Mapping The structural-mapping perspective (Gentn er and Markman 1994, 1997) is commonly used to predict how consumers will make inferen ces about the benefits of new products. The structural-mapping perspective assumes that in ferences about the characteristics of a new product are made by drawing an analogy to a familiar, host categor y. Consumers access information related to a host category, map properties of the host category onto the target product, and make inferences about the target products characteristic s using a process of knowledge transfer (Gentner and Markman 1994, 1997; Gregan-Paxton and John 1997; Moreau, Markman, and Lehmann 2001). In general, people ar e more likely to map the properties of a 13

PAGE 14

relational system that are relevant to the anal ogy (i.e., people map aligna ble features) (Gentner and Markman 1997; Gregan-Paxton et al. 2002). The structural mapping perspective has been pa rticularly effective at predicting consumer inferences about new-to-the-world products, which defy classification by virtue of offering novel benefits or new uses for existing benefits. For ex ample, prior to its intr oduction, a digital camera was a radical innovation because consumers were not only unfamiliar with its benefits but also with the product itself, making it rather difficult to categorize it to an existing product category. In situations like these, consumers can make inferences abou t the benefits of new-to-the-world products via analogical reasoning (Figure 2-1). An analogous host category is often selected because it has benefit dimensions that align with the target product, hence the knowledge can be easily transferred (Moreau et al 2001). Marketers can try to in fluence the inference process by suggesting a category (e.g., camera, scanner) that will encourage the most beneficial set of inferences about the new product (e.g., digital camera) (Moreau et al. 2001; Yamauchi and Markman 2000). Host Category Target Product Benefit 1 ? Benefit 2 ? Figure 2-1. Structural mapping paradigm and product innovations introducing novel benefits. The structural mapping perspective may be inappropriate for predicting consumer inferences about hybrid products. Hybrid product s often combine two existing products with non-overlapping benefits, and thus do not involve introduction of new benefits nor new usage contexts for known benefits. For example, miles pe r gallon (mpg) is an important attribute for a car-motorcycle hybrid, yet consumers know this at tribute well. Thus, the critical consumer process in understanding a hybrid product is not learning its be nefits but making inferences about 14

PAGE 15

its attribute values. What is pe culiar about the inference proce ss in a hybrid product context is that it requires reconciliation of contradictory information from the two constituent categories. As a result, the process is unlikely to be one where knowledge is transf erred from a single host category to the target product. Instead, the pr ocess involves an infere nce about which of two conflicting values is more plausible for a given attribute (Figure 2-2). Take the car-motorcycle hybrid as an example. Cars perform poorly on th e mpg attribute relative to motorcycles. The question the consumer needs to answer is whic h of these two conflicting mpg values (i.e., good or bad) the car-motorcycle hybrid will inherit. Host Category A Target Host Category B Benefit 1 ? Deficit 1 Benefit 2 ? Deficit 2 Deficit 3 ? Benefit 3 Deficit 4 ? Benefit 4 Figure 2-2. Categorization under uncertain ty paradigm and hybrid products. Two competing processes may be responsible for consumer inferences about hybrid product attribute values. First, it may be the case that consumer s try to categor ize the hybrid product into the more plausible of two constituent categories on a given attribute and infer the attribute value from the distri bution of that category (i.e., category plausibility determines hybrid attribute values). This process is similar to st ructural mapping, but allows a consumer to choose the relevant constituent category for making infe rences about each attribute. Second, it may be that consumers try to determine a value that is mo st plausible given the di stribution of values in the two constituent categories (i.e., attribute value plausibility determines hybrid attribute values). What follows is a discussion of each of these processes. 15

PAGE 16

Category Plausibility Determines Hybrid Attribute Values Judgments about the attribute values of a hybrid product could depend on a categorization process. Evidence for this link comes from res earch in categorization under uncertainty (Rips 1989; Rips and Collins 1993; Smith and Sloman 1994; Sloman and Rips 1998). According to this literature, categorizing an ambiguou s object as a member of categor y A or B is a judgment that depends on two sources of information: (1) the re lative variability of each constituent category and (2) the similarity of the ambiguous object to each constituent category. Rips (1989) has found that people consider the relative variability of each constituent category to be the most diagnostic information for a categorization j udgment. For example, consider the quarter category, the pizza category, and an ambiguous object that is three inch es in diameter. The ambiguous object is more similar to the quarter category than to the pizza category. Yet, given the variability in quarter and pizza diameters, the ambiguous object is more plausibly a pizza. Smith and Sloman (1994) found that people use cate gory variability, rather than similarity, to categorize ambiguous stimuli only when their verbal protocols showed awareness of differential variability across categories. The impact of category variability on categoriz ation judgments has also been studied in the perceptual domain. A notable st udy by Fried and Holyoak (1984) has revealed that people are sensitive to the relative variability of perceptu al categories when they are making categorization judgments. They showed that participants classi fied some checkerboard pa tterns that were more similar to the prototypes (or mean) of the lowe r variability category as members of the highvariability category. In a subs equent study, Stewart and Chater (2002) identified a boundary condition for the use of variabil ity-based categorization in per ceptual classification. In their experiments, Stewart and Chater used outline circles each with a single solid dot somewhere on their circumference as th eir stimuli. The circles varied only in the posi tion of the dot around the 16

PAGE 17

circumference and this position was diagnostic of category membership. The low-variability category distribution had a standa rd deviation of 11 and the hi ghvariability category had a standard deviation of 28. The critical exemplar fell in between the nearest members of these two categories. Results showed that people used va riability-based categoriz ation to classify the critical exemplar when the experimental pro cedure sensitized them to differences in the variability of category members by presenting all members at the same time. When item presentation was sequential, on th e other hand, participants failed to notice differences in the variability of the category memb ers, making it impossible to use variability-based categorization. In this condition, participants us ed similarity-based categorization. This finding parallels Smith and Slomans (1994) finding that variability-b ased categorization precedes similarity-based categorization when people notice differences in the relative variability of the candidate categories on a critical dimension. Inferring the value of a hybrid product attribute is similar to the categorization of an ambiguous object with one caveat. In a typical categorization under uncertainty task, the features values of the to-be-categorized stimulus are known. Thus, the consumer can use these values to assess whether the stimulus was drawn from the distribution of objects th at represent category A or B. In a hybrid product categorization task, the features values of the hybrid product are unknown. Thus, the consumer can not use these values to categorize. Instea d, the variability of the distribution of objects that re present category A or B is potenti ally diagnostic. In line with previous research (e.g., Rips 1989; Stewart and Chater 2001), people should expect that the hybrid product attribute value is more likely to be drawn from the distribu tion of values in the 17

PAGE 18

high variability category (Figure 2-3).1 As such, the hybrid should take on the mean / modal value from this category. H1a : When constituent product category variab ility is diagnostic, people are more likely to generalize the modal value of the higher variability category to the hybrid product. A B A B Plausible Plausible Figure 2-3. Category plausibility (variability-based) judgment. When differences in the variability of the tw o constituent categories are similar or difficult to notice, differences in the category variabil ity are not diagnostic for inferring hybrid product attribute values. In this situation, people should rely on similarity information, as has been shown by Smith and Sloman (1994; also see Stew art and Chater 2002). However, unlike a categorization task, there is no similarity be tween the hybrid product and a constituent product category because the hybrid products values ar e not known. Thus, the similarity between the two constituent categories of a hybrid product becomes potentially diagnostic. As constituent product categories become more similar, it beco mes more plausible that the hybrid product can bridge the benefit gap between the categorie s and provide the bene fit (Figure 2-4). 1 One could argue that people infer the average of the two constituent category values for the hybrid product on a given attribute. This averaging argument would not apply to discrete attributes, which require an inference of whether an attribute is present or not. Also, the appeal of hybrid products for marketers lies in their ability to combine the best attributes of two products in one product, and not so much in their ability to offer the average value of the attributes of their constituent categories. Hence, from a substantive standpoint, it is important to understand the factors that affect consumers inferences regardin g which of the two constituent category values the hybrid product is likely to inherit on a given attribute. 18

PAGE 19

Similar types of plausibility judgments have been observed in the induction literature (e.g., Osherson, Smith, Wilkie, Lopez, and Shafir 1990) For example, people find it more plausible that an attribute of category A will generalize to category B as the perceived similarity between the two categories increases. Furthe r evidence for the role of simila rity on plausibility judgments comes from the conceptual co mbination literature. Wisniews ki (1996, 1997) reported a number of experiments showing that as the similarity of constituents of a noun-noun compound increased (e.g., painter photographer), people tended to us e a hybridization strategy to interpret the combination, in which two concepts are combined so that each acquires properties of the other. Wisniewski explained this finding using the alig nment theory. That is, because highly similar constituents tend to have more alignable featur es, people are more willing to transfer multiple properties from both constituents when interp reting a compound (see Costello and Keane 1997, 2000 for an alternative account).2 Attesting to the difficulty of encouraging people to adopt a hybridization strategy, several experiments have show n that participants ra rely use this strategy (e.g., only about 3% of the word compounds were interpreted using hybr idization (Wisniewski 1997)). Consistent with prior research, people should become more likely to accept that the hybrid products performance is consistent with the better performing cat egory when the two constituent categories are perceived to be more similar. H1b : When constituent product category variab ility is non-diagnostic, people are more likely to rely on the similarity of the constituent categories to make an inference about a hybrid products performance. As similarity increases, people become 2 Unlike Wisniewski who used different compounds to ma nipulate constituent similarity the experiments reported here keep the constituent categories of a hybrid product c onstant and manipulate their similarity between subjects. By keeping constituent categories constant, these experiments control for the number of alignable attributes across different levels of simila rity. Hence, alignability does not explain the reported pattern of findings. 19

PAGE 20

more likely to generalize the mean (modal) value of the constituent category that has the highest performance level. Dissimilar Categories Similar Categories A B A B Not plausible Plausible Figure 2-4. Distributions of a target attributes values in two constituent categories with equal variability. Attribute Value Plausibility Determines Hybrid Attribute Values Judgments about the attribute values of a hybrid product could depend on an assessment of the plausibility of specific valu es. Consumers may assess the distri butions of attribute values in each constituent category and, based on the overlap in these distributions, make an inference about the attribute value of the hybrid. Category distributions have been shown to influence peoples stimulus estimates. For example, Huttenl ocher, Hedges, and Vevea (2000) varied the distribution of stimuli in their experiments such that participan ts in one condition saw a uniform distribution while participants in another saw a normal distribu tion. Huttenlocher et al. then asked their participants to judge category membersh ip for a new series of stimuli where the range of values was extended to include stimuli outside the range of the initial set. Their results showed that the probability of judging extreme values to be memb ers fell off more rapidly for normal than uniform distributions of th e same range. This was because participants considered the fact that there are fewer stimulus valu es near the tails of a normal vs. uniform distribution. This result replicated when participants were asked to reproduce a stim ulus (e.g., fish) after seeing it. Participants size estimations were more bias ed towards the category prototype when the 20

PAGE 21

category distribution was uniform than normal. Th e authors argued that pe ople use distributional information automatically when categoriz ing stimuli and predicting their values. A B A B Plausible Plausible Figure 2-5. Value plausibility (distribution-based) judgment. Based on the evidence that people use dist ributional information in various category judgments, I expect that constituent product category distributional information will be diagnostic when there is a small amount of overlap in distributions (Figur e 2-5). In these cases, people should infer that the hybrid product will have the overlapping value, as it is most plausible. More specifically, fo r each attribute, people compare the existing distributions from the two constituents. When this comparison yields a small overlap (i.e., few observed attribute values common to both constituents), the overlap becomes diagnostic and overlapping attribute value is inferred to the hybrid product. For example, say, 99% of the cars possess airbags whereas 100% of motorcycles do not. Because the ove rlap in this example is in the deficit region (i.e., 1% of the cars and 100% of motorcycles do not possess airbags), people should infer deficit on this attribut e for the hybrid product. H2a : When constituent product category distribu tions are diagnostic (i.e., distributions marginally overlap), people are more lik ely to generalize the overlapping value. When there is no overlap (i.e., no common attr ibute value), or signi ficant overlap (i.e., numerous common attribute values) in constituent category distributions, th en the overlap is not diagnostic. In this case, people should use the distance between the means of the two constituent 21

PAGE 22

22 category distributions to infer the hybrid product attri bute value. As the mean values become more similar, the differences between the means of the attribute distributions become smaller and it becomes more plausible that the hybrid produc t can bridge the benefit gap and provide the benefit. It is noteworthy that th is hypothesis differs from Hypothes is 1b in that people must be able to compare distributions, not simply assess category similarit y, for similarity to exert an influence on inferences about hybrid product attribute values. H2b : When constituent product category dist ributions are not diagnostic (i.e., distributions do not overlap, distributions significantly overlap), people are more likely to rely on the similarity of the distributions to make an inference about a hybrid products performance. As similarity increases, people become more likely to generalize the mean (modal) value of the constituent category that has the highest performance level. These hypotheses are tested using four experime nts. Experiment 1 is a direct test of whether category plausibility versus attribute value plausibility guides peoples inferences about hybrid product benefits. E xperiment 2 manipulates constituent category similarity and the diagnosticity of the value distributions independently to test th e hypothesis that category similarity influences peoples hybrid product in ferences when the value distributions are not diagnostic. In Experiments 3A and 3B, accessibi lity of the distributional information is manipulated to examine whether people use constituent category similarity when the distributional information is inaccessible. Fi nally, Experiment 4 replicates findings from Experiment 3A studies in a more ecologically valid experimental context.

PAGE 23

CHAPTER 3 EXPERIMENT 1 Experiment 1 was an initial investigation in to the processes that determine consumer expectations about the attribut e values of hybrid products. The challenge was to manipulate constituent category attribute value distributions in a way that (1) allowed participants to assess category plausibility (H1) and at tribute value plausibility (H2) and (2) allowed each of these hypotheses to make unique predictions. With this constraint in mind, the experiment involved an independent manipulation of the pe rceived attribute value variabil ity in the constituent category that provided the benefit and in the constituent category that did not provide the benefit. This allowed for independent manipulations of distribution variability and overlap. Method Design The design was a deficit category variability (low, high) and benefit category variability (low, high) by replicate order c ounterbalance factor (fast food and casual restaurant, car and motorcycle, sports car and station wagon) by cat egory order counterbala nce factor (category named first when describing the hybrid product) mixed design with only the first factor manipulated within subjects. The order in whic h the four different co mbinations of deficit category variability (low, high) and benefit catego ry variability (low, high) were presented was determined by a Latin square design, which is illustrated in Table 3-1. Procedure Participants entered a behavior lab and were seated at personal computers. The instructions stated that they would have to assess several new products (servi ces) that were a combination of 23

PAGE 24

two existing products (services).1 Participants were warned that the new products might seem novel because these products were only recently introduced into the market. Then, a verbal description of the first hybrid product was introduced (e.g., The new service is a combination of a fast food and a casual restaurant) followed by th e presentation of the fi rst attribute (e.g., fast service). Each constituent cate gory attribute distribut ion was described side-by-side using a single sentence (e.g., None of the casual restaurant s have fast service.; All of the fast food restaurants have fast service.). While this in formation remained on the screen, the participant was asked to predict the value of the hybrid product on this a ttribute by clicki ng on one of two options provided (e.g., Yes No). After particip ants reported their pr edictions for the first attribute, they repeated the procedure for the sec ond attribute. There were four attributes to be predicted for each hybrid product replic ate, and participants had to pr edict all four of them before the next hybrid product was considered. The key dependent measure was the proportion of the four benefits generalized to the hybrid product. The four treatment conditions were created by altering the order in wh ich descriptions of the attribute value distributions of the constituent product categories was presented. For example, when attribute value variability was low in each constituent product category, it was indicated that none of the deficit category members had th e benefit, whereas all of the benefit category members had the benefit (e.g., None (all) of th e casual (fast food) re staurants have fast service.). These descriptors where changed to none and most, few and all, and few and most for the subsequent attributes of the same replicate (Table 3-1 ). 1 In the remainder of the paper, I will use the term product to refer to both hybrid product replicates (e.g., vehicles) and hybrid service replicates (e.g., restaurants). 24

PAGE 25

Table 3-1: Experiment 1 design for the Casual Restaurant Fast Food Restaurant replicate illustrating the order in which the constitu ent category variability information was presented in each of the four conditions. Constituent Category Variability Information Casual Restaurant Fast Food Restaurant Replicate Experimental Condition Value-Priced Fast Service High Quality Food Inviting Ambiance A None All None Most Few All Few Most B None Most Few All Few Most None All C Few All Few Most None All None Most D Few Most None All None Most Few All Stimuli The hybrid products and their attr ibutes were selected based on a series of pretests, which resulted in a total of nine hybr id product replicates that vari ed in the similarity of their constituents to each other. These nine replicat es were divided into three groups, each containing three replicates, based on the degree of similar ity between their constitu ents. In all experiments except Experiment 4, three hybrid product replic ates were used whose constituent categories were moderately similar to each other. This woul d help reduce the chance of a floor or ceiling effect when perceived similarity was manipulated in subsequent experiments. Second, pretests were used to select the sets of attributes used to test cons umer inferences about the hybrid product. Pretest participants listed the first five benefits that came to mind for each constituent category. These lists were used to identify three benefits for each constituent category that (1) ranked high in popularity (i.e., liste d frequently), (2) were an e xpected benefit in one of the constituent categories but not th e corresponding category. Two of thes e three benefits were used in Experiment 1, which were selected randomly, due to design restrictions.2 All three attributes 2 The Latin square design employed in experiment 1 dict ated that one attribute corre sponded to each of the four distribution description conditions (i.e., none-most, few-most, few -all, none all), limiting the number of attributes to four for each hybrid product replicate. 25

PAGE 26

were used in the subsequent experiments that employed a different design. The hybrid product replicates and benefits used in the experiments are liste d in the Appendix.3 Predictions Although the experiment is a tw o-by-two factorial design, the predictions of the category plausibility hypothesis and the at tribute value plausibi lity hypothesis do no t align with main effect or interaction tests. The category plausibility hypothesis (H1) predicts that benefits should be more likely to generalize to the hybrid when the benefit category has more variability than the deficit category, as opposed to vise-versa. Thus, there should be more benefit generalization in the none / most condition than in the few / all condition. Th e attribute value plausibility hypothesis (H2) predicts that bene fits should be more likely to ge neralize to the hybrid when the overlap in constituent product cate gories is in the benefit domai n. Thus, there should be more benefit generalization in the f ew / all condition than in the none / most condition. The category plausibility hypothesis and valu e plausibility hypothesis make similar predictions when constituent category variability is not diagnostic (i.e., none / all and few / most conditions). Benefit gene ralization should depend on the si milarity of the distributions. Because constituents with distribut ions of few / most have m eans closer to each other than do constituents with distributions of none / all there should be more benefit generalization in the few / most condition than in the none / all condition. Ninety undergraduate students participated in the experi ment in return for class credit. 3 A final pretest showed that all 12 attributes had diffe rent constituent attribute values for at least 75% of the participants and that 10 of the 12 attributes had diffe rent constituent attribute values for at least 75% of the participants. Further refinement of the attributes was limited by concerns about using unimportant benefits. 26

PAGE 27

Results The means for the generalization scores by condition are reported in Table 3-2. These means were computed by averaging the proportion of benefits generalized in each of the four category variability descriptions, collapsing over participants. Consistent with attribute value plausibility hypothesis, participants generali zed more benefits in the few / all ( M = .87) condition than in the none / most condition ( M = .69, t (179) = 6.25, p < .001). Consistent with both hypotheses, particip ants generalized more benefits in the few / most ( M = .83) condition than in the none / all condition ( M = .70, t (179) = 4.47, p < .001). Supplemental analyses showed that the replicate factor did not interact with any of the critical tests. The presentation order of the constituent product categories in the hybrid product description (e.g., a combination of a fast food and casual restaurant versus a combination of a casual and fast food restaurant) also did not interact with any of the critical tests. Table 3-2: Experiment 1 means: The influence of constituent category attribute value variability on benefit generalization. Constituent category descriptions Deficit category Benefit category Results "None" "All" 0.70 "None" "Most" 0.69 "Few" "All" 0.87 "Few" "Most" 0.83 Discussion Experiment 1 provides an initial test of the category plausibility hypothesis and the attribute value plausibility hypothesi s. The results are more consistent with the attribute value plausibility hypothesis. People infe r that a hybrid product will have a value that is common to the overlap in the distributions of the constituent product categories. Experiment 1 also suggests that people are sensitive to two t ypes of distributional information. People use the overlap in distributions to infer hybrid values when the overl ap is diagnostic (i.e., small), but rely on the 27

PAGE 28

28 similarity of the distributions when the overlap is not diagnostic (i.e., no overlap or significant overlap). The results of the first experime nt are inconsistent with part of the category plausibility hypothesis (H1a), but not the en tire category plausibility hypo thesis (H1b). The category plausibility hypothesis contends that (1) constituent category attribute value variability is diagnostic of hybrid attribute valu es and (2) when attribute valu e variability is not diagnostic, constituent category similarity is diagnostic. Th is hypothesis could be amended to argue that category similarity is diagnostic for all benef it generalization judgments. For example, it may have been the case that a none / most desc ription of constituent category distributions resulted in a perception of less cat egory similarity than a few / all description of constituent category distributions. As a consequence, the f ew / all condition resu lted in more benefit generalization.

PAGE 29

CHAPTER 4 EXPERIMENT 2 Experiment 1 used distributional descriptors (e.g., none / most) to manipulate the distributional properties of the constituent product categories that comprise the hybrid product. These distributional descriptors also created a concurrent manipulation of the similarity of the constituent product categories that comprise th e hybrid product. To determine whether the similarity of the constituent product categorie s could be solely responsible for benefit generalization, or whether attribute value plau sibility is also diagnostic, Experiment 2 manipulated category similarity and the diagnosticity of the value distributions independently. Method Design The design was an attribute va lue diagnosticity (diagnostic, not dia gnostic) by constituent category similarity (low, high) by hybrid product replicate (fast food and casual restaurant, car and motorcycle, sports car and station wagon) by category order counterbalance factor mixed design with only the replicate f actor being manipulated within-s ubject. The order in which the hybrid product replicates was presented was randomized. The attribute value diagnosticity manipulation was the none / most (diagnos tic) and few / most (non-diagnostic) conditions. The none / most condition was diagnos tic because it entails a small distributional overlap in the deficit region. The few / most condition was non-diagnostic because the distributional overlap is large and could be framed as either in the deficit or benefit region. The constituent category similarity manipulation encour aged participants to perceive the constituent product categories as more similar or different prior to making the benefit generalization judgments. 29

PAGE 30

Procedure The procedure was similar to that of Experime nt 1 with two exceptions. First, there was a similarity manipulation. Before e xpressing expectations about the hybrid product, pa rticipants in the low and high similarity conditions completed a categorization task that involved organizing four categories into two pairs. Two of the four categories pr esented were the constituent categories for the hybrid product while the othe r two were decoy categories. The two decoy categories in the low similarity c ondition were selected so that they would be paired with one of the constituent categories, thus reducing the perc eived similarity between the constituents. For example, in the fast food casual restaurant hybrid, the two decoy categories in the low similarity condition were a sandwich restaurant and an upscale restaurant. If the fast food restaurant is paired with the sandwich restaurant and the casual restaurant is paired with an upscale restaurant, then the fast food restaurant and cas ual restaurant categor ies should be rated less similar. The two decoy categories in the high similarity condition were a bakery and a coffee shop. In this context, the fast f ood restaurant should be paired with the casual restaurant and the categories should be rated more si milar. After pairing the four categories, participants were asked to explain in detail why th ey paired the specific items together. Participants then rated the similarity of the constituent categories to each other. The other change in the pro cedure involved the presentation of the constituent product category attribute value distribut ion information. First, unlike E xperiment 1, there were three rather than two attributes for each constituent category. Second, in Experiment 1, each constituent category attribute di stribution was described using a single sentence. These sentences were presented in pairs, one at tribute at a time (e.g., None (All) of the casual (fast food) restaurants have fast service.) Since constituent category simila rity was manipulated before the attribute presentation in this ex periment, the attribute value distribution description procedure 30

PAGE 31

needed to be more efficient to maximize the eff ect of the similarity manipulation on participants predictions. Thus, in Experiment 2, all six attributes that had to be predicted were presented on the same page (Table 4-1 illustrates the at tribute presentation in Experiment 2). After completing the categorization task used to manipulate constituent category similarity, and before expressing expectations about the hybrid product, participants in the diagnostic attribute value condition saw: Below, you see a distribution of the values of several attributes for two constituent product categories. A "Most" means that most products sampled from the corresponding category possess the attribute, wh ile "None" means that none of the products sampled from the corresponding category po ssess the attribute. The respondents then saw a listing of the six attributes along with two column headings that were the constituent product category names. Each entry in the column was most or none. For example, in the fast f ood causal restaurant replicate, the fast service, value-priced and drive-through service attributes were listed as most in the fast food category and none in the casual restaurant category. Similarly, the inviting ambiance high quality food and attentive service attributes were listed as none in the fast food categor y and most in the casual restaurant category. The non-diagnostic value condition differed from the diagnostic value condition only in the substitution of few for non e. Thus, each attribute was listed as most for one constituent category and few for the other constituent category. Participants indicated whether or not a hybrid product would possess each of the attributes while this information remained on the screen. One hundred ninety-six undergraduate students participated in the experiment in return for class credit. 31

PAGE 32

Table 4-1: Illustration of attribute presentati on in the non-diagnostic c ondition of Experiment 2 for the Fast Food Restaurant Casual Restaurant replicate. Constituent Categories Attributes Fast Food Restaurant Casual Restaurant Fast service Most Few Value-priced Most Few Drive-through service Most Few Inviting ambiance Few Most High quality food Few Most Attentive Service Few Most Results Manipulation Check The responses of those particip ants who paired the four cate gories in the expected fashion for all three replicates were included in the analysis (n = 120). The constituent category similarity ratings differed significantly for the low and high similarity conditions ( Mlow = 3.05; Mhigh= 3.92, F (1, 118) = 23.50, p < .001). Analysis The means for the generalization scores by condition are reported in Figure 4-1. The predicted interaction between attribute value diagnosticity and constituent category similarity was significant (F (1, 116) = 4.07, p < .05). When the constituent category value distributions were diagnostic, participants were insensit ive to the category similarity manipulation (Mlow = .69; Mhigh = .71, F (1, 119) = .15, p = .69). When the constituent category value distributions were non-diagnostic, participants increased their willingness to ge neralize benefits as constituent category similarity increased from low to high (Mlow = .64; Mhigh = .76, F (1, 119) = 10.00, p < .01). Supplemental analyses showed that the orde r of constituent category presentation did not exhibit a main effect ( F (1, 112) = .00, p = .98) or interact with similarity ( F (1, 112) = .05, p = .82), attribute value diagnosticity ( F (1, 112) = 3.21, p = .08), or a combination of the two ( F (1, 112) = .10, p = .75). 32

PAGE 33

High Category Similarity Low Category SimilarityProportion of Benefits.Constituent Product Category Values N on-diagnostic: "Few"/"Most" Diagnostic: "None"/"Most" 1 0.9 0.8 0.7 0.6 0.5 0.76 0.71 0.64 0.69 Figure 4-1. Experiment 2 results: The infl uence of constituent category similarity and distribution diagnosticity on e xpectations of constituent benefits in a hybrid product. Discussion The results of Experiment 2 pr ovide further evidence that exp ectations about the benefits of a hybrid product depend on the attribute values of the constituent categories. When the attribute value distributions of the constituen t categories had a diagnostic overlap, the common value was generalized to the hybrid product. Im portantly, this generali zation process was not sensitive to category similarity. When the a ttribute value distributions of the constituent categories had a non-diagnostic overlap, participants relied on the similarity of the constituent categories. Taken together, the first two experi ments suggest that people anticipate hybrid product performance by assessing the plausibility th at a particular level of performance could be achieved. 33

PAGE 34

CHAPTER 5 EXPERIMENT 3A If consumers are using information about the targ et attribute distributi ons of the constituent product categories to make inferences about the performance of a hybrid product, then benefit generalization judgments should be sensitive to the accessibility of the distributional information. This prediction also follows from existing resear ch that shows that pe ople use distributional information in a categorization task when they are aware of differences in the variability across categories (Smith and Sloman 1994). Experiments 1 and 2 used distributional informa tion descriptors to ensu re that participants were cognizant of the distributional information (i.e., accessibility was guaranteed). In Experiment 3, the salience of th e distributional information wa s manipulated. In one condition, participants experienced the f ew / most non-diagnostic condition procedure of Experiment 2. In a second condition, participants saw the bene fits of each constituent category product, but these benefits were not accompanied by distribution information. Consistent with Hypothesis 2b, similarity should only exert an influence on benefit generalizations when the distributional information is salient. Method Design and Stimuli The design was a distributional information accessibility (inaccessible, accessible) by constituent category similarity (low, medium, high) by hybrid product replicate (fast food and casual restaurant, car and motorcycle, spor ts car and station wa gon) by category order counterbalance factor (two orde rs) mixed design with the replic ate factor manipulated withinsubject. The order in which the hybrid product replicates was presented was randomized. A 34

PAGE 35

medium similarity condition (i.e ., no constituent category similar ity manipulation) was included in the design as an addi tional control condition. Procedure The procedure was identical to Experiment 2. The distributional information accessibility manipulation was achieved by altering the few / most non-diagnostic condition procedure of Experiment 2 (i.e., accessible condi tion) to create the inaccessible condition. Participants in the inaccessible distribution informati on condition were told, Below you see the attributes of two constituent categories. Each cat egory listed the three attributes For example, the fast food category listed the fast service, value-priced, and drive-through servic e benefits. The casual restaurant category listed the in viting ambiance, high quality food, and attentive service benefits. The benefit lists were displayed side-by-side to discourage participants from thinking about distributional information in each of the constituent product categories. The accessible condition used the few / mos t non-diagnostic condition procedure of Experiment 2. One hundred ninety-eight undergraduate students participated in the experiment in return for class credit. Constituent Categories Fast Food Restaurant Casual Restaurant Attributes Fast service Inviting ambiance Value-priced High quality food Drive-through service Attentive Service Figure 5-1. Experiment 3A proce dure: Illustration of attribute presentation in the inaccessible distributional information condition. 35

PAGE 36

Results Manipulation Check The responses of those particip ants who paired the four cate gories in the expected fashion for all three replicates were included in the analysis (n = 145). The constituent category similarity ratings were significantly different by similarity condition (Mlow = 3.03; Mmedium = 3.33, Mhigh= 4.12, F (2, 142) = 12.39, p < .001). Planned contrasts showed that the low and medium similarity condi tions did not differ ( F (1, 142) = 1.96, p = .16), but that the medium and high similarity conditions did differ ( F (1, 142) = 13.43, p < .001).1 The similarity rating did not depend on the accessibility of the distributional information ( F (1, 133) = 0.01, p > .05) or an interaction of the accessibility and category similarity information ( F (2, 133) = .56, p > .05). Analysis The means for the generalization scores by condition are reported in Figure 5-1. The predicted interaction between similarity and distributional information accessibility was significant ( F (2, 139) = 4.40 p < .02). When distributional information was inaccessible, participants did not increase their willingness to generalize attr ibute benefits as constituent category similarity increased from low ( Mlow = .70) to medium ( Mmedium = .74; ( F (1, 139) = 1.60, p = .21) and from medium ( Mmedium = .74) to high ( Mhigh = .71, F (1, 139) = .78, p = .38). When distributional information was accessible, particip ants increased their willingness to generalize attribute benefits as constituent category similarity increased from low ( Mlow = .61) to medium ( Mmedium = .69, F (1, 139) = 4.50, p < .04) and from medium ( Mmedium = .69) to high ( Mhigh = .77, 1 It is possible that the lack of a difference between the low and medium similarity co nditions arose because of the wording of the manipulation check ques tion. Asking respondents in the lo w similarity condition to assess the similarity of the constituent product categories may have enco uraged them to focus on similarities, rather than the dissimilarities they just considered. 36

PAGE 37

F (1, 139) = 6.26, p < .02). Supplemental analyses showed that the order of constituent category presentation did not exhibit a main effect ( F (1, 133) = 1.19, p = .28) or interact with similarity ( F (2, 133) = .22, p = .80), alignability ( F (1, 133) = .12, p = .73), or a combination of the two ( F (2, 133) = 2.98, p = .06). Finally, there was no difference between the accessibility conditions in the moderate similarity condition ( Minaccessible = .74, Maccessible = .69; F (1, 144) = 1.91, p > .05). This suggests that the accessibility manipula tion did not alter the pa rticipants implicit assumptions about the constituent category distributions. High Similarity Medium Similarity Low SimilarityProportion of Benefits.Non-diagnostic Distributional Information Accessible "Few"/"Most" Inaccessible 1 0.9 0.8 0.7 0.6 0.5 0.77 0.71 0.69 0.74 0.61 0.70 Figure 5-1. Experiment 3A Results: The accessi bility of constituent category distributional information moderates the influence of cate gory similarity on benefit generalization. Discussion The results of Experiment 3A suggest th at consumers rely on constituent category similarity to infer the value of a hybrid product at tribute value when distributional information is accessible, but not diagnostic. Unlike Experi ment 2, in which salient but nondiagnostic distributional information encouraged the use of constituent category similarity, participants were not sensitive to the perceived similarity of the constituent product categories when distributional information was not salient. These re sults provide further evidence that inferences 37

PAGE 38

38 about hybrid product performance ar e sensitive to the attribute value distributions of the constituent product categories. One might be concerned that the side-by-side presentation of the product attributes in the inaccessible distribution information condition enc ouraged participants to assume that the distribution for the constituent cat egories was none all for the product attributes. If this is indeed the case, then explicitly presenting the di stribution information of none all should not change the pattern of results observed in the inaccessible distribution information condition. If, however, the side-by-side presen tation of the stimuli successfully reduced the salience of the distributions in Experiment 3A, then explic itly presenting the nondiagnostic none all distributional information should encourage part icipants to use const ituent similarity in accordance with Hypothesis 2b.

PAGE 39

CHAPTER 6 EXPERIMENT 3B In this study, the distributional information accessibility manipulation was achieved by explicitly providing the none / all non-diagnostic distribution information. This explicit distribution presentation should en courage participants to rely on constituent category similarity when making inferences about the hybrid produc t given the accessible but nondiagnostic nature of the distribution information. Method Design and Stimuli The design was a constituent category simila rity (low, medium, high) by hybrid product replicate (fast food and casual restaurant, car an d motorcycle, sports car and station wagon) by category order counterbalance fact or (two orders) mi xed design with the replicate factor manipulated within-subject. Procedure The procedure was identical to the accessi ble distribution information condition in Experiment 3A except that the few / most non-diagnostic category variability description was replaced with the none / all non-diagnostic category vari ability description. Ninety-two undergraduate students participated in the experiment in return for class credit. Results Manipulation Check The responses of those particip ants who paired the four cate gories in the expected fashion for all three replicates were included in the analys is (n = 70). The constituent category similarity ratings were significantly differe nt by similarity condition (Mlow = 2.72; Mmedium = 3.34, Mhigh= 3.79, F (2, 67) = 8.11, p < .001). Planned contrasts showed th at the low and medium similarity 39

PAGE 40

conditions did differ ( F (1, 67) = 6.37, p < .02), and that the medium and high similarity conditions differed marginally (F (1, 67) = 13.43, p = .06). Analysis The means for the generalization scores by condition are reported in Figure 6-1. The predicted main effect of similarity was significant ( F (2, 64) = 3.73 p < .03). As constituent similarity increased from low ( Mlow = .67) to medium ( Mmedium = .72) to high ( Mhigh = .78), people became more willing to genera lize attribute benefits to th e hybrid product. Supplemental analyses showed that the order of constituent ca tegory presentation did not exhibit a main effect ( F (1, 64) = .04, p = .85) or interact with similarity ( F (2, 64) = .59, p = .56). Cumulatively, the results of Experiments 3A and 3B furnish further support for the hypothesis that non-diagnostic but ac cessible distribution information encourages participants to rely on constituent category similarity when inferring hybrid product benefits. High Similarity Medium Similarity Low SimilarityProportion of Benefits.Non-Diagnostic Distributional Information Accessible ("None"/"All") 1 0.9 0.8 0.7 0.6 0.5 0.78 0.72 0.67 Figure 6-1. Experiment 3B Results: The accessi bility of constituent category distributional information moderates the influence of cate gory similarity on benefit generalization. A question that the results reported here do not address pertains to the informational inputs people use when distributional information is in accessible. For example, what information did 40

PAGE 41

41 the participants in the inaccessible distribution information condition in Experiment 3A use to infer hybrid product benefits? Although severa l processes may underlie these inferences, Experiment 4 explored whether peoples theories about new pr oduct success in the marketplace govern their predictions in the abse nce of distribution information.

PAGE 42

CHAPTER 7 EXPERIMENT 4 Experiments 2 and 3 manipulated the perceived similarity of the constituent product categories independently of the categories themselves. This procedure was necessary so as to unconfound the similarity and the attribute valu e distributions of the two constituent product categories. A more ecologically va lid approach to investigating th ese factors is to select hybrid products that vary with respect to their cons tituent categories sim ilarity to each other. Replicating the results of Experiment 3A, benefit generalizat ion should increase as similarity increases when distributional info rmation is accessible, but not when distributional information is inaccessible. Method Design The design was a distributional information acce ssibility (inaccessible, accessible) by level of constituent category similarity (low, medium, high) by hybrid product replicate (three replicates per level of similarity) by category order counterba lancing factor (two levels) mixed design with the constituent category similarity an d replicate factors manipulated within-subject. The order in which the nine hybrid product replicates was presented was randomized. Procedure and Stimuli The procedure was the same as in Experiment 3A, except that constituent product category similarity was manipulated using different hybrid product replicates. After a series of pretests explained in Experiment 1, three hybrid products expected to have low constituent category similarity (e.g., light bulb and air freshener, pen and calculator, restaurant and movie theater), moderate constituent category similarity (e .g., fast food and casual restaurant, car and motorcycle, sports car and station wagon), and hi gh constituent category si milarity (e.g., jet ski 42

PAGE 43

and snowmobile, mountain bike and road race bi ke, TiVo service and Movies-on-Demand) were selected. A pretest confirmed that the similarity (1= not similar at all, 7= very similar) of the categories varied as intended ( F (2, 38) = 52.40, p < .001). The difference between the low similarity ( Mlow = 2.17) and moderate similarity ( Mmod =3.65; F (1, 19) = 35.3 p < .001) and the moderate similarity and high similarity ( Mhigh = 4.82, F (1, 19) = 21.8, p < .001) pairs was significant. The same set of pretests was also used to select the se ts of attributes used to test inferences about the hybrid product (see the Ap pendix for the complete list of hybrid product replicates and their attributes). Results Ninety-two undergraduate students participated in the experiment in return for class credit. The data were analyzed using a repeated measur e MANOVA with similarity as a within-subject factor and distributional accessibility as a betweensubject factor. The means for this analysis are reported in Figure 8-1. The predicted interaction between distributional information accessibility and similarity was significant ( F (2, 180) = 3.12, p < .05). When distributional information was inaccessible, participants did not increase thei r willingness to generalize the benefits as constituent category similarity increased from low ( M = .70) to moderate ( M = .72; F (1, 41) = 0.85, p = .36), but did increase their willingness to ge neralize benefits as constituent category similarity increased from moderate ( M = .72) to high ( M = .82; F (1, 41) = 25.06, p < .001). When distributional information was accessible, participants increased their willingness to generalize benefits as constituent category similarity increased from low ( M = .65) to moderate ( M = .74; F (1, 49) = 14.44, p < .001) and from moderate ( M = .74) to high ( M = .82; F (1, 49 = 13.65, p < .001). The order of constituent category pr esentation did not have a main effect ( F (1, 88) = 2.16, p = .15) or interact with cons tituent category similarity ( F (2, 176) = .11, p = .90), 43

PAGE 44

distributional information accessibility ( F (1, 88) = 0.48, p = .49), or a combination of the two ( F (2, 176) = 0.0, p = .99). 1 Proportion of Benefits.0.9 0.82 0.82 0.8 0.74 0.72 0.7 0.65 0.7 0.6 0.5 Inaccessible Accessible Non-diagnostic Distributional Informaiton Low Similarity Medium Similarity High Similarity Figure 8-1. Experiment 4 Results: The accessib ility of constituent category distributional information moderates the influence of cate gory similarity on benefit generalization. Discussion Experiment 4 offers additional support for the attribute value plausibi lity hypothesis. When attribute value distributional information was ac cessible, but not diagnostic, an increase in constituent product category similarity resulted in an increase in the likelihood of generalizing the benefits from the constituent categories. When attribute value distributional information was not accessible, an increase in constituent product category similarity did not result in an increase in the likelihood of generalizing th e benefits from the constituent categories at low to moderate levels of similarity. The unexpect ed finding in the high similarity condition may be an artifact of the hybrid product replicates (e.g., jet ski and snowmobile, mountain bike and road race bike, TiVo service and Movies-on-Demand) used in this condition. Alternatively, it may be that highly similar constituent categories make distributional information easy to access. In other words, the 44

PAGE 45

45 information presentation format in the inacc essible condition can disc ourage the access of attribute value distribution information, but it can not prevent access to this information.

PAGE 46

CHAPTER 8 GENERAL DISCUSSION Marketing research on new products has mainly focused on two types of product innovations: enhancement and new-to-the-world products. This focus has been accompanied by an emphasis on knowledge transfer as the proc ess underlying the consumers learning of the novel benefits that these product innovations intr oduce. Because knowledge transfer facilitates consumer learning of novel product benefits by u tilizing consumers ex isting product knowledge, it is an efficient and effective marketing tool to communicate the novel benefits of new products. Not all new products offer novel benefits to consumers, however. For example, hybrid products combine two existing products with known benefits to create a new product that offers the best features of these two products, without the weaknesses of either. Because consumers already know about the benefits of a hybrid pr oduct (e.g., mpg in the car-motorcycle hybrid), knowledge transfer is unlikely to assist in their understand ing of a hybrid product. The consumers challenge with hybrid products is to determine which of the two conflicting constituent category values to accept on an attrib ute. This research provides evidence for the hypothesis that an attribute plausibility judgm ent resulting from a comparison of constituent distributions drive peoples hybrid product at tribute value predictions. People compare the existing distributions from the two constituents each time an attribute has to be predicted. If there is a small distributional overlap, then people ge neralize the overlapping attribute value for the hybrid product, which is the most plausible value for a distribution resulting from a combination of the two constituent distributions. When there is no overlap, or significant overlap in constituent category distributions, then the overlap is not diagnostic In this case, people compare the constituent category distributions to judge th eir similarity: the more similar the constituent category distributions are, the more likely the hy brid product is to bridge the attribute gap and 46

PAGE 47

generalize the benefit. Experiments 3A and 3B de monstrate that distributional information must be salient in order for people to us e attribute plausibility or distri bution similarity in their benefit generalization judgments. Experiment 4 uses real world hybrid products to manipulate constituent category similarity and provides evidence supporting the attribute plausibility hypothesis. Theoretical Implications There are at least two important streams of psychology research that can shed light on how consumers evaluate hybrid products: the categori zation under uncertainty literature (e.g., Rips 1989; Smith and Sloman 1994; Murphy and Ross 1994) and the conceptual combination literature (e.g., Costello a nd Keane 1997, 2000; Wisniewski 1997). Results reported here corroborate findings from the categ orization literature that people use distributional information as input to their judgments only when they are aw are of it (Smith and Sloman 1994) or when it is made salient (Stewart and Chater 2002). In the current experiment s, only when the experimental procedure highlighted distributional information did pa rticipants in the current studies take it into consideration when making inferenc es about the hybrid product. Another set of findings from th e categorization literature ha s shown that people exhibit a strong tendency to base their inferences and pr edictions on a single category when faced with categorization ambiguity (Malt, Ross, and Murphy 1995; Murphy and Ross 1994, 1999A; Ross and Murphy 1996). Only under rare circumstances are people shown to use information from more than one candidate category to make inferences about an ambiguous object (i.e., multiple category strategy; Moreau et. al 2001; Gregan-Paxton et. al. 2005). For example, in marketing, Moreau et al.s (2001) participants used a multip le category strategy to make inferences about a new product (digital camera) only when they we re explicitly informed of the relationship between the digital camera and the two candidate categories (film-based camera and a scanner). 47

PAGE 48

In the context of the experiments reported here, a single category strategy would predict that participants would genera lize the benefit to the hybrid pr oduct only half the time since a given constituent category performs better than the other only on half of the attribut es. If benefit generalization exceeded 50%, this would provide evidence for the multiple category strategy. Because participants generalized benefits to the hybrid product more than half the time across all experiments, current studies provide evidence for the use of multiple category strategy when making inferences about the hybrid product. Ho w can this finding be reconciled with the robustness of the single category strategy in the literature? One procedural aspect common to studies finding single category strategy is that the presentation of one of the candidate categories pr ecedes that of the other. For example, Moreau et al. (2001) found that their participants used only the fi rst category cued when making predictions about the performance of the new pr oduct, thereby ignoring the alternative category cued subsequently. The order of category presentation had an important influence on new product evaluations even when explicit mappings from the two candidate categories were provided. Specifically, 57% of subjects who saw the camera ad first categorized the new digital camera as a camera, compared to 31 % of subjects who saw the scanner ad first. In another study, Ross and Murphy (1996) presented pa rticipants with a story cont aining a reference to a person whose identity was uncertain. The text of the st ory (e.g., realtor) cued the persons identity, but an alternative identity was also subsequently cued (e.g., burglar). Participants were asked to predict the probability that the ambiguous person would engage in certain category-consistent and inconsistent behaviors (e.g., for burglar, pay attention to the st urdiness of the doors). Results showed that the impact of the a lternative category cued later in the text was limited to only those 48

PAGE 49

questions highly associated with that category. That is, signi ficant contextual support was necessary to induce multiple category strategy. The implication is that cuing alternative cate gories sequentially impedes peoples ability to use multiple category strategy by allowing them to structure the ambiguous object according to a single category. Once an initial re presentation of an object is formed based on the first category cued, restructuring it to include new category information becomes a challenging task. It is possible that simultaneous representation of the c onstituent categories in the current experiments discouraged participants from fo rming an early representation of the hybrid product according to one of the constituent categories. This observatio n is supported by the finding that changing the order in which constituent categories was presen ted did not influence th e proportion of benefit generalization in the current studies. Further ev idence for this propositi on comes from GreganPaxton et al. (2005) who showed their participan ts the category cues at the same time and observed multiple category strategy use. Current results have implications for the con ceptual combination literature as well. Several models of conceptual combination have been recently proposed (e.g., Costello and Keane 2000; Murphy 1988; Wisniewski 1997). Wisniewskis ( 1997) dual process model is one of the more prominent of these models. An important pillar of Wisniewskis (1997) m odel is that combining two concepts (e.g., cactus carpet) involves a structural alignment process by which people compare the two concepts. According to this model, structural alignment underlies interpretations of combinations whose concepts ar e highly similar to each other. Because highly similar concepts tend to be more alignable th an dissimilar concepts (Gentner and Markman 1997), they highlight the alignable differences that subsequently govern the combinations interpretation. Consistent with this contention, Wisniewski (1997) showed that people are more 49

PAGE 50

willing to transfer multiple prope rties from both constituents (i.e., hybridization) when the constituents are highly similar to each other. Given Wisniewskis results, the impact of constituent category similarity on hybrid product evaluation evinced by the cu rrent set of experiments may co me as no surprise. What is intriguing, however, is that the between subjects manipulation of constituent category similarity in the experiments reported here controls for th e degree of alignability between the constituent categories. In other words, the degree of alignability between the constituent categories cannot account for the increased benefit generalization at higher levels of sim ilarity unless one argues that the between subjects similarity manipulation affected the degree of a lignability between the constituent categories in a consis tent way. The implication is that although alignability is an important determinant of hybridization, similar ity between constituent categories may influence the interpretation of the combination via a rout e other than alignability. The findings reported here support the contention that similarity a ffects hybrid product evaluation by bringing its constituent categories distributions closer to each other. That is, because means of two categories approach to one another on a given attr ibute as their perceived similarity increases, it becomes more plausible that th e hybrid product can successfully offer that attribute. This differential route is plausible in a hybrid product context in whic h technical plausibility (i.e., whether it is technically feasib le for the hybrid product to overcome the trade-off on an attribute) interacts with conceptual plausibility (i.e., wh ether it is easy to im agine the hybrid product possessing a given attribute) to in fluence hybrid product evaluations. One intriguing finding was that participants us ed constituent category similarity to make inferences about the hybrid pr oduct only when the procedure made category distributional information salient. It is possible that there ar e other, perhaps more pervasive mechanisms than 50

PAGE 51

constituent category similarity that people employ when making inferences about hybrid product benefits. Unless its use is encouraged, as the current studies do by cuing distributional information, constituent category similarity may be dominated by other s ources of inputs to the hybrid product inference process. One such mech anism can be peoples in tuitive theories about the marketplace (Chernev and Carpenter 2001). Furt her research is needed to identify possible mechanisms consumers may employ when distributional information is not salient. Limitations This research is subject to a limitation common to almost all consumer behavior research, which is using an undergraduate student subject pool in experiments. Because this is a relatively homogeneous group and not representative of the typical American consumer, it is difficult to assess the generalizabilit y of the reported results to the population. Another limitation involved the dichotomous nature of the de pendent variable employed in the current experiments. Partic ipants indicated whether or not the hybrid product would have certain attributes of its constituent categories. This served to force participants to generalize the value of either one of the constituent categor ies to the hybrid product, which allowed me to directly analyze the impact of the independent variables on th e choice of which constituent category would drive hybrid produc t inferences. However, this aspect of the procedure possibly limited the generalizability of the results in the following way. By forcing inferences, the dependent measure may overstate the degree to which consumers in real life engage in spontaneous inferences about hybrid product attributes. This concern is less valid for experiential attributes whose value can only be inferred be fore one actually uses the product (e.g., how thrilling it will be to ride a car-motor cycle hybrid) than for search at tributes whose value is easier to assess through search. Although the set of attri butes used in the current experiments contained experiential attributes, there were search attributes as well. This was an outcome of the criteria 51

PAGE 52

employed that guided the selection of attributes in the pretest. Sp ecifically, only those attributes listed most frequently for each constituent category by pretest subjects were selected to ensure that the attributes used were important enough to encourage inference making in the real world. The downside of using the most im portant attributes was that the final attribute list contained both search and experiential attributes. The extent of this limitation in generalizability is mitigated by the following factors. First, consumer inferences about the hybrid product that occur prior to information search can affect the likelihood of actually engaging in such search. Positive inferences about hybrid product attribute values, for example, can increase the consumers willingness to gather additional information about the product. Such initial infere nces can also serve as expected performance criteria for the hybrid product against which its real performance can be judged, which in turn influences the consumers overall evaluation of the product. Finally, there may be search attributes that managers may be unwilling to ad vertise to consumers due to hybrid products low performance on these attributes. It may be advantageous to have consumers infer the values of these attributes for the hybrid product. Furthermore, since the current experiments employed a dichotomous dependent measure, one should be cautious in generalizing the findings to continuous attributes. As will be discussed in the Future Research section, alternative processes may be ava ilable to consumers to infer the value of a hybrid product on a continuous attribute. Managerial Implications From the managers perspective, hybrid products will be successful to the extent that consumers are willing to generalize benefits from both constituents to th e hybrid product. Prior research has shown that only in rare circumstances do cons umers transfer knowledge from multiple categories (e.g., Moreau et al. 2001). The implication is that encouraging consumers to 52

PAGE 53

transfer knowledge from both constituent categorie s is a highly risky ma nagerial proposition in the case of hybrid products. Furthermore, such a promotional strategy may be ineffective with hybrid products since they do not involve the learning of novel product benefits. A better promotional strategy can thus be designed by exploi ting the factors identified in this work that affect consumers inferences at the attribute level, which cumulatively determine the evaluation of the hybrid product. One such factor is constituent category sim ilarity. Emphasizing the si milarity between the constituents of a hybrid product increases favorab le inferences about it. As such, although it may be tempting for hybrid product managers to prom ote the technical appeal of the hybrid product by emphasizing how it combines two very dissimilar existing products, th is research and development focus in promotion may not induce the desired consumer response. Furthermore, product communications should also involve attempts to increase the perceived similarity of the constituent categories to each othe r. For example, juxtaposing th e constituent products in print ads (e.g., a car and a motorcycle) showing them perform the same functi on (e.g., transportation) may increase favorable inferences about their co mbination. Put differently, using super-ordinate category labels in product promotions that en compass both constituent categories may increase the perceived similarity of the c onstituent categories to each other, resulting in more favorable inferences about the hybrid product. As the current experiments indicate, the role of constituent similarity is not straightforward. Category distribut ional information must be salient for similarity to influence hybrid product evaluations. Thus, it is important for hybrid managers to include such information in their promotions. In order to do so, phrases that may cue distri butional information such as the 53

PAGE 54

ones used in current experiments (e.g., few, most, some) may be utilized in product communication to convey attribute informati on (e.g., some motorcycles have airbags). The second factor that affects benefit genera lization is the degree of overlap of the constituent categories on important attributes. The managerial implication is that those important product attributes on which the two constituent cat egories have overlap in the benefit category should be advertised to encour age benefit generalization. Again, the degree of overlap can be emphasized using phrases that cu e distributional information. Finally, although not tested directly in this re search, current results and existing research provide sufficient evidence to suggest that adve rtising the constituent products simultaneously (e.g., in a print ad) may increase the use of info rmation from both constituent categories (i.e., multiple category strategy), resulting in more favorable inferences about the hybrid product. Future Research Existing marketing research paints a very broad brush of new products in general by failing to consider the characteri stics of new products that have conceptual implications for how consumers evaluate them. By focusing on peculiar aspects of hybrid pro ducts, this research shows that processes underlying the consumer evaluation of hybrid products are different from those underlying other types of pr oduct innovations that have hither to been investigated. Future research will benefit from identifying peculiar aspects of different t ypes of product innovations, which will lead to a more nuanced and effective approach to st udying the consumer evaluation of new products. One interesting extension of the current findings is examining whether simultaneous presentation of category cues encourages a mu ltiple category strategy. It is also worthwhile investigating how the length of time after structuring a new pr oduct representation on the basis of one category can influence peoples propens ity to use multiple ca tegory strategy. Existing 54

PAGE 55

research shows that people ignore the second category cued a few minutes after the first category cue. Will this strong tendency to use single cat egory strategy become more or less pronounced over time (e.g., presenting the second category cu e one day after the presentation of the first one)? Additionally, the Theoretical Im plications section distingu ished between technical and conceptual plausibility in the hybrid product conc ept. Conceptual plausibility is necessary but not sufficient to understand consumer inferenc es about hybrid products. That is, it may be conceptually plausible for a hybrid product to ha ve a certain feature (e.g,. a motorcycle with retractable wheels that can be used when slowi ng or stopping) yet not te chnically plausible to effectively offer it (e.g., whether retractable wheels will work effectively). Results from the current studies support this dis tinction by showing that it is po ssible to influence technical plausibility while keeping attri bute alignability constant, a f actor that has been shown to influence conceptual plausibility (Wisniewski 1997). More research is needed however to establish that these two plausibility judg ments indeed rely on different processes. As mentioned before, current experiments em ployed a dichotomous dependent measure. It is possible that this limited the processes available to consumers when evaluating hybrid products. It will be interesting to investigate alternative processes that may underlie consumer inferences when the dependent measure is a continuous attribute. It is plau sible that an averaging model may approximate peoples inferences in this case (Anderson 1967). Given two constituent categories, consumers may simply average the mean values of the two ca tegories on an attribute to predict the value for their co mbination. The interes ting question is under what circumstances will consumers diverge from this effortless yet possibly error prone averaging strategy? What factors will encourage the use of a weighted average model in which the two constituent 55

PAGE 56

56 categories differentially contri bute to what the hybrid produc ts value will be on a given attribute? Given the theoretical and managerial importance of how consumers evaluate hybrid products, I believe these questi ons warrant further research.

PAGE 57

APPENDIX EXPERIMENTAL STIMULI 57

PAGE 58

58 Low Constituent Category Similarity Hybrid Products Pen / Calculator (2.25) Light bulb / Air freshener (2.00) Restaurant / Movie theater (2.25) Pen Calculator Light bulb Air freshener Restaurant Movie theater Disposable Programmable Lasts about 5000 hours ^ Can be refilled Gourmet food Dark Inexpensive Graphing feature Provides pure light Circulates fragranc e Easy to converse Stadium seating Super-fine tip Has memory Reveals natural colors Consistent frag. delivery Wide selection of food Outstanding sound system Medium Constituent Category Similarity Hybrid Products Fast food / Casual restaurant (3.85) Car / Motorc ycle (3.55) Sports car / Station wagon (3.55) Fast food Casual restaurant Car Motorcycle Sports car Station wagon Value-priced1 High quality food Weatherproof Low fuel consumption Aerodynamic Ample trunk space Fast Service Inviting ambiance Airbag High parking convenience Low weight Low engine noise Drive-through service Attentive service Air-conditioning Low Emissions Fast Family Car High Constituent Category Similarity Hybrid Products Jetski / Snowmobile (4.4) Moun tain bike / Race bike (5.15) TiVo / Movies -on-Demand (4.9) Jetski Snowmobile Mountain bike Race bike TiVo Movies-onDemand Rides on water Heated seats All-terrain Fast ^ Pause broadcasts New releases Life jacket compartment Head and tail lights ^ Rugged Weight ^ Records shows Pay-per-view ^ 360 spins in its own length ^ Snow beams Shock absorbers Aerodynamic ^ Many hours of recording time Instant access to movie library ^ Over 25% of respondents thought both constituents possessed benefit. 1 Only the attributes in italics were used in experiment 1.

PAGE 59

LIST OF REFERENCES Anderson, Norman H. (1967), "Averaging Model An alysis of Set-Size Effect in Impression Formation," Journal of Experimental Psychology 75 (October), 158-165. Chernev Alexander and Gregory S. Carpenter (2001) The Role of Market Efficiency Intutions in Consumer Choice: A Case of Compensatory Inferences, Journal of Marketing Research 38 (August), 349-361. Costello, Fintan J. and Mark T. Keane (1997), Polysemy in Conceptual Combination: Testing the Constraint Theory of Combination, in Nineteenth Annual Conference of the Cognitive Science Society, Hillsdale, NJ: Erlbaum. Costello, Fintan J.and Mark T. Keane (2000) Efficient Creativity: Constraint Guided Conceptual Combination, Cognitive Science 24 (April/June), 299-349. Dahl, Darren W. and Page Moreau (2002), The Influence and Value of Analogical Thinking During New Product Ideation, Journal of Marketing Research 39 (February), 47-60. Dahl, Darren W., Amitava Chattopadhyay, and Ge rald J. Gorn (2001), The Importance of Visualisation in Concept Design, Design Studies 22 (January), 5-26. Fried, Lisbeth S. and Keith .J. Holyoak ( 1984), Induction of Cate gory Distributions: A Framework for Category Learning, Journal of Experiment al Psychology: Learning, Memory and Cognition, 10 (April), 234-257. Gatignon, Hubert and Thomas S. Robertson (1985), A Propositional Inventory for New Diffusion Research," Journal of Consumer Research 11 (March), 849-867. Gentner, Dedre and Arthur B. Markman (1994) Structural Alignmen t in Comparison: No Difference without Similarity, Psychological Science 5 (May), 152-158. Gentner, Dedre and Arthur B. Markman (1997), Structure Mapping in Analogy and Similarity, American Psychologist, 52 (January), 45-56. Goldenberg, Jacob, David Mazursky, and Sorin Solomon (1999), Toward Identifying the Inventive Templates of New Products : A Channeled Ideation Approach, Journal of Marketing Research 36 (May), 200-210. Gregan-Paxton, Jennifer and Deborah Roedder John (1997), Consumer Learning by Analogy: A Model of Internal Knowledge Transfer, Journal of Consumer Research 24 (December), 266-284. Gregan-Paxton, Jennifer (2001), The Role of Abstract and Specific Knowledge in the Formation of Product Judgments: An Analogical Learning Perspective Journal of Consumer Psychology, 11 (3), 141-158. 59

PAGE 60

Gregan-Paxton, Jennifer, Jonathan D. Hibbard, Frd ric F. Brunel, and Pablo Azar (2002), So Thats What That Is: Examining the Impact of Analogy on Consumers Knowledge Development for Really New Products, Psychology and Marketing 19 (June), 533-550. Gregan-Paxton, Jennifer, Steve Hoeffler, a nd Min Zhao (2005), When Categorization is Ambiguous: Factors that Facilita te the Use of a Multiple Category Inference Strategy, Journal of Consumer Psychology 15 (2), 127-140. Gregan-Paxton, Jennifer and Page Moreau (2003 ), How Do Consumers Transfer Existing Knowledge? A Comparison of Analogy and Categorization Effects, Journal of Consumer Psychology, 13 (4), 422-430. Hirshman, Elizabeth (1980), Innovativeness, Novelty Seeking, and Consumer Creativity, Journal of Consumer Research 7 (June), 283-295. Huttenlocher, Janellen, Larry V. Hedges, and Ja ck L. Vevea, (2000), Why Do Categories Affect Stimulus Judgment? Journal of Experiment al Psychology: General 129 (June), 220-241. Leonard, Dorothy and Jeffrey F. Rayport (1997), Spark Innovation thro ugh Empathic Design, Harvard Business Review, 75 (November/December), 102-113. Malt, Barbara C., Brian H. Ross, and Gregory L. Murphy (1995), Predicting Features for Members of Natural Categories When Categorization is Uncertain, Journal of Experimental Psychology: Learning, Memory, and Cognition, 21 (May) 646-661. Moreau, C. Page, Arthur Markman, and Dona ld R. Lehmann (2001), What Is It? Categorization Flexibility and Consumer s Responses to Really New Products, Journal of Consumer Research, 37 (March), 489-498. Murphy, Gregory L. (1988), Comprehending Complex Concepts, Cognitive Science, 12 (4) 529-562. Murphy, Gregory L. and Brian H. Ross (1994), Pre dictions from Uncertain Categorizations, Cognitive Psychology 27 (October), 148-193. Murphy, Gregory L. and Brian H. Ross (1999), I nduction with Cross-Classified Categories, Memory & Cognition 27 (November), 1024-1041. Murphy, Gregory L. and Brian H. Ross (2005), The Two Faces of Typicality in Category-Based Induction, Cognition 95 (March), 175-200. Osherson, Daniel N., Edward E. Smith, Ormond Wilkie, Alejandro Lpez, and Eldar Shafir (1990), Category-Based Induction, Psychological Review 97 (April), 185-200. Park, Whan C. and Daniel C. Smith (1989), P roduct-Level Choice: A Top-Down or Bottom-Up Process, Journal of Consumer Research 16 (December), 289-299. 60

PAGE 61

Rips, Lance J. (1989), Similarity, Typicality, an d Categorization, in S. Vosniadou & A. Ortony (Eds.), Similarity and Analogical Reasoning 21-59. Cambridge, MA: Cambridge University Press. Rips, Lance J. and Allan Collins (1993), Categories and Resemblance, Journal of Experimental Psychology: General 122 (December), 468-486. Ross, Brian H., and Gregory L. Murphy (1996), Category-Based Predictions: Influence of Uncertainty and Feature Associations, Journal of Experimental Psychology: Learning, Memory and Cognition, 22 (May), 736-53. Sloman, Steve A. and Lance J. Rips (1998), Similarity as an Explanatory Construct, Cognition 65 (January), 87-101. Smith, Edward E., and Steven A. Sloman (1994), Similarityversus Rule-Based Categorization, Memory & Cognition, 22 (June), 377-386. Srinivasan, V. and William S. Lovejoy (1997), Journal of Marketing Research 34 (February), 154-163. Stewart, Neil and Nick Chater (2002), The Effect of Categor y Variability in Perceptual Categorization, Journal of Experimental Psychology: Learning, Memory, and Cognition 28 (September), 893-907. Takashi, Yamauchi and Arthur B. Mark man (2000), Inference Using Categories, Journal of Experimental Psychology: Learning, Memory, and Cognition 26 (May), 776-795. Ulrich, Karl T. and Steven D. Eppinger (1995), Product Design and Development. New York: McGraw-Hill Book Company. Urban, Glen L. and John R. Hauser (1993), Design and Marketing of New Products, 2nd edition, Englewood Cliffs, NJ: Prentice Hall. Verde, Michael F., Gregory L. Murphy, and Br ian H. Ross (2005), Influence of Multiple Categories on the Prediction of Unknown Properties, Memory & Cognition 33 (April), 479-487. Viswanathan, Madhubalan and Terry L. Ch ilders (1999), Understanding How Product Attributes Influence Product Categorization: Development and Validation of Fuzzy SetBased Measures of Gradedness in Product Categories Journal of Marketing Research 36 (February), 75-94. von Hippel, Eric (2005), Democratizing Innovation, Cambridge, MA: MIT Press. Wathieu, Luc and Michael Zoglio (2005), Harvar d Business School Case Study, October, (No: 9-501-038). 61

PAGE 62

62 Wisniewski, Edward J. (1996), Construal and Similarity in Conceptual Combination, Journal of Memory and Language, 35 (June), 434-453. Wisniewski, Edward J. (1997). When Concepts Combine, Psychonomic Bulletin and Review 4 (December) 167-183.

PAGE 63

BIOGRAPHICAL SKETCH Baler Bilgin earned his bachelor degree in political science and international relations from Bogazici University in Istanbul Turkey. In 2000, he moved to the United States to pursue a Master of Business Administrati on at the Illinois State Universi ty in Normal-Bloomington. After completing his MBA, Baler entered the Ph.D. program in marketing at the University of Florida. He joined the University of California-Riverside as an Assistant Professo r of marketing in July 2007. 63


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101220_AAAAJC INGEST_TIME 2010-12-21T01:40:29Z PACKAGE UFE0019761_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 4609 DFID F20101220_AACNAE ORIGIN DEPOSITOR PATH bilgin_b_Page_06thm.jpg GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
52c7f2d16a7e87530e250be6972655fc
SHA-1
0b1b9e8705fd359ca9b3fba816bc2bba31500102
25271604 F20101220_AACMTH bilgin_b_Page_10.tif
3930637430cf5d5cc586cbb54cf5158e
94160be5a4e1a64f63d9cef1829aa1a634dd58d9
1883 F20101220_AACMYE bilgin_b_Page_20.txt
5c0666fc2cced62236788580ea2ba1d9
00ed99a0a9dcee0ab274c1bfaaaf8f43b709dd49
5728 F20101220_AACNAF bilgin_b_Page_07.QC.jpg
54eb40c4a0188c301a70fc8b2a4173c3
ed86237d97a3d9597b372df97f9adb5cd5a2129e
F20101220_AACMTI bilgin_b_Page_11.tif
873767d2856b13aa737d1f3e54fb27f8
599ca1361779d32bce11192e7e9b360d4b2f107b
26721 F20101220_AACMOK bilgin_b_Page_01.jpg
edba85cc252ff58ef2812b5a295f37ea
2244801bb759ec7fdd34b18605e7630666264b30
1874 F20101220_AACMYF bilgin_b_Page_21.txt
0fa7fa6b311ebce54b3293949c61a5c0
3895757284d91bfd23c03dbf86ac6e5e5fa1dda2
F20101220_AACMTJ bilgin_b_Page_12.tif
af437fdc0cde754e9d1390d8f76bbe4c
7df928cca56c422e5458e5ea33c25de5d882a2f7
3691 F20101220_AACMOL bilgin_b_Page_02.jpg
8573474ab7536dfd168798ee2047c370
a1d1b4bf23d3413d4934e850f83248cfd2dcd8bc
1934 F20101220_AACMYG bilgin_b_Page_22.txt
6c9432914f747ba94f789bcce5382639
18ebb60bdb69e0c02f429cbb7e3817756e42bde6
1589 F20101220_AACNAG bilgin_b_Page_07thm.jpg
ca674dec4eec6b0c2be0a2297a70aa77
6bec06144339e7ba4152f239274de74a00cd931c
F20101220_AACMTK bilgin_b_Page_13.tif
314ed2d3015f3fa9c7e5356f2cbb3c5f
76fec64b0651137d2e566d5e56afff03ac8a1757
4324 F20101220_AACMOM bilgin_b_Page_03.jpg
2276938ec3ca6ecdb7dbdff06cab284e
08f18193d819cf2007c30ebd73b0d48d0c2aa599
1820 F20101220_AACMYH bilgin_b_Page_23.txt
1201451d0569ef455bd11ca2e3509ff9
dad8e69ab1681c0ccf8acb0311123aa94e028448
14253 F20101220_AACNAH bilgin_b_Page_08.QC.jpg
d711bcef885c87ccafdf5d664dfb8d06
d63016dcd17909f40753905f9756b8f07799a4f4
F20101220_AACMTL bilgin_b_Page_14.tif
1251e882ace64e297e4e27ac17678566
ec83c67c577ac631d420a79d4f55093db39f5729
75402 F20101220_AACMON bilgin_b_Page_04.jpg
831b64407eb004835bda1c91c0444190
ffcab059b18f9d2d186c228be5e1e8ebac9c8d41
2117 F20101220_AACMYI bilgin_b_Page_24.txt
3a0205f0f564ad19947a23f8ccbeac4f
53e14b2c6a4f44bb19d359e08096132f7d05c0f6
3782 F20101220_AACNAI bilgin_b_Page_08thm.jpg
7ef1b60b443ef6dec15f89dd88b921ca
141d009b546375d5500c4737232825dc498e9b60
98849 F20101220_AACMOO bilgin_b_Page_05.jpg
857f6fb2141f66fd9e4d1e83e55e2b07
05972c6ce699ad705e5e9dda4291d1e1d9de5197
2124 F20101220_AACMYJ bilgin_b_Page_26.txt
afb5ea90ae23ec64f741511786fa0408
ead278cc6cedb3c464883b0148d4d4918cfc7dc5
21780 F20101220_AACNAJ bilgin_b_Page_09.QC.jpg
b7010ef45665256726a27e851eab61d6
4a77ec5f19527e2a435c041a2eb4fbb983aa9a18
F20101220_AACMTM bilgin_b_Page_15.tif
d7ac65557eef8b0b9df10824dab382f6
d205549a55716f72000ba6a9506067410f8d54e0
86544 F20101220_AACMOP bilgin_b_Page_06.jpg
84a837859334251d18d968bed7709203
a8af7e04a277b5bb57a528bd406fc07e44286969
2174 F20101220_AACMYK bilgin_b_Page_27.txt
f1df4a6113bf18c611ee0cc84a4a02e4
dc792682cf16eeef961ace505b4614de7a542f40
5578 F20101220_AACNAK bilgin_b_Page_09thm.jpg
3d96da4c40e78445656f87af2e4f9e32
192ba2ce48a60ed4eb53ce3a6a07664d29cbbe0b
F20101220_AACMTN bilgin_b_Page_16.tif
13ac8f36bacad98aac7f509da72c389d
3f3572c1bc8bdfe162a01458434d13c7db284810
18426 F20101220_AACMOQ bilgin_b_Page_07.jpg
39c8b189269cbcf57a0fc5682425e475
7bfbe2bd80d2cb202224ea927d89ab0c2d0dc102
1033 F20101220_AACMYL bilgin_b_Page_28.txt
f06446519f86e9908f9f61f613970ea4
25f89d69f869cf8225082335b94e222cdcdbceab
8501 F20101220_AACNAL bilgin_b_Page_10thm.jpg
96e5a226f4374ef91056f2fa3be5dc05
6a83536efa99e3571fffd0d9ed04d7ecf3837b47
F20101220_AACMTO bilgin_b_Page_17.tif
a949e88a6ca971b9da3f16b78c26d962
0749bc230e34c5e8d3497a482e0c38f2b8b3f987
51125 F20101220_AACMOR bilgin_b_Page_08.jpg
21e382a13dc4310248a882b4e7051164
ff57fe659f07e382100738e5388027c0a4d9690f
1937 F20101220_AACMYM bilgin_b_Page_29.txt
84a21134c722fa8fba963e7d62824a34
1c16ebbe716565d48cb14d147f5dd0940bd28d69
6061 F20101220_AACNAM bilgin_b_Page_11thm.jpg
48d7b41f644293ee160d459012b8606e
0dbc62bce985449b14171ef7a4833038bc25ec3f
F20101220_AACMTP bilgin_b_Page_18.tif
81e43894a9e831419f54e335ad0f428f
11545170bcf130a3fcd9debfb90b245c608bfee0
70648 F20101220_AACMOS bilgin_b_Page_09.jpg
87d4e9e30fd4d4d2227b85899c6c27b1
c3948d754b1079b7fb070ac889361043adc51904
2252 F20101220_AACMYN bilgin_b_Page_30.txt
0abac0347be01f98b1a34d4df63c942b
f65ae7606cf56c3f9cc306206d884c4abfd801fe
33508 F20101220_AACNAN bilgin_b_Page_12.QC.jpg
d52aa297eda780d8e0f48c1391fcd405
07ccd733de4d1c061b78c5586a48ef9da15a4c20
F20101220_AACMTQ bilgin_b_Page_19.tif
807f65e1341422176e202107bc6ae77e
f0d3e94636ee1730321e1e9112e671c85100ae57
102759 F20101220_AACMOT bilgin_b_Page_10.jpg
dd9ed080bde280cdaa4593915127ef04
9b4ad248ef6d776db022027731cb8c9a775b62bf
2027 F20101220_AACMYO bilgin_b_Page_32.txt
6ad677f2bc7e1122eb9c7ee7b953bee5
35ebd8caab5b5a5660bceadc07c35166c427e64d
8210 F20101220_AACNAO bilgin_b_Page_12thm.jpg
dfc4fa2ea15af17e61c3e17c1506e776
0e4ee5b69bd6f9b4e4a6ddf9b5882b9ffc6bd254
F20101220_AACMTR bilgin_b_Page_20.tif
e1ff45eafc5a3f2664a5e3af8735cf27
24ec7fd0147fda2fded1efd5bcfef6100e076ad2
75775 F20101220_AACMOU bilgin_b_Page_11.jpg
711db65daee57e9430b00481f242a021
df6ae578b60326e1e56f0790299763938cfa0995
1584 F20101220_AACMYP bilgin_b_Page_33.txt
307f7057d111ee5c340b12bf64f93e18
c0f7171ec20d31214d2cd29aebb7a655e6111e72
33874 F20101220_AACNAP bilgin_b_Page_13.QC.jpg
204cdf2311e8775220c5e0e299aac051
f6be1a9ccc725d41e0281c6665f5e3f07e2f1cfc
F20101220_AACMTS bilgin_b_Page_21.tif
3e10d6f5fd1d600021b10433fc34ec3f
1ed7b44f5686734dbd7ff3e1b98796e3c14b25b7
103075 F20101220_AACMOV bilgin_b_Page_12.jpg
4b02125e146e755bd293834fdadade87
d808fdbf50a1efea99548c4b4f19286e614a1a2b
1922 F20101220_AACMYQ bilgin_b_Page_34.txt
4af86d68b8ee57379008285b1b5a95f8
3acd18b48a906624ce92902456b39c4e64d3b8b8
8538 F20101220_AACNAQ bilgin_b_Page_13thm.jpg
2257e78d1b1caeebe67c4ef3515f4280
514ea2bd2b6b264d28fc1342206f56d9a9ad266d
F20101220_AACMTT bilgin_b_Page_22.tif
43e2a9aad3f3f5931e95c85eb56506ea
f74a6f68aec610952e630e5204d8d0e930114e22
105133 F20101220_AACMOW bilgin_b_Page_13.jpg
dafa976cdac9f4959cf10edf84e647f8
8d9a97e71407c70b560f2163107604102737c68b
33941 F20101220_AACNAR bilgin_b_Page_14.QC.jpg
7e66762221cf163d5a330166dd709437
734de49333fcc36bea6beb5759279b52eb9dfb09
F20101220_AACMTU bilgin_b_Page_23.tif
48bb9f635e4fc308d3f8b57054c80e1d
782db5c3f8411a2f56e27a5179c9e00236d21abf
103872 F20101220_AACMOX bilgin_b_Page_14.jpg
2400cff58ef15c31bc2bcb6e5d39ba56
40159ef3faecf719e1e97bb2d65203b8107e7e9a
1786 F20101220_AACMYR bilgin_b_Page_35.txt
9c5b4c28e98978b52830b1e15ff8e345
359569bba7c4bf5303695a975fa8e54e85bad933
8232 F20101220_AACNAS bilgin_b_Page_14thm.jpg
220489055fd88273cc6eeff7710cd373
229018693a83e8794266c9bc2b3a7325886bcca2
F20101220_AACMTV bilgin_b_Page_24.tif
5e991cb17ca5f5a341a048f8e74f7d4a
7a2e1c94c4d1c540a2c3e563cbb194e7086e0f2c
94016 F20101220_AACMOY bilgin_b_Page_15.jpg
dee3db86b583773cd5edc43edef5017a
d9a8448a4b8a131f81fbd752dba8a1b57963340a
2121 F20101220_AACMYS bilgin_b_Page_36.txt
e6419a3072724d8914df38d320a652cb
f7faa4d42fbad1c37fac7fde8a89dbe440dadbb0
31330 F20101220_AACNAT bilgin_b_Page_15.QC.jpg
64c9e128d99499acb0626073da38de08
f28f36b921da30a97f4d9250d3117c1a9af53835
F20101220_AACMTW bilgin_b_Page_25.tif
1a2320af295725bc0debf27470e1753b
b3e993aa3f3632786624c87633b4c88f17f54309
116631 F20101220_AACMOZ bilgin_b_Page_16.jpg
00b2272e11a5d460da97e6c3115ac252
03834010206b494a8817022dcf6586ecf30e8b74
2037 F20101220_AACMYT bilgin_b_Page_37.txt
1c2904a3f166424c8f614741aec424a5
b65bea7d06e7177cdf068020285df729bf31822e
37736 F20101220_AACNAU bilgin_b_Page_16.QC.jpg
b5fe703f070cc92bfc99604d6358165f
dab3e05598cc21195e38061b853aecf3c7339229
F20101220_AACMTX bilgin_b_Page_26.tif
09915c577436f7a8017e9b298f583b77
733f13db85aaace3e72b77eb88178f49a6170edb
962 F20101220_AACMYU bilgin_b_Page_38.txt
c72dc80c277aa508c1865dacc611f9bc
76820007229aa5c5e5def2a7ec7b0b1291345d92
9011 F20101220_AACNAV bilgin_b_Page_16thm.jpg
55005ec2a804575022de9ce3c73a7a25
0347df72ab548e80684598524ba0773669fdde11
F20101220_AACMTY bilgin_b_Page_27.tif
657eefbee90986036464d7968594cc71
07d5091b2154b4525f6033107b2d0af9dc57b895
1807 F20101220_AACMYV bilgin_b_Page_39.txt
52c22ff8d8e5d04767939d49adc706c2
8252d4e835be5ee2b6c285c092210f0aeee0f3a8
33437 F20101220_AACNAW bilgin_b_Page_17.QC.jpg
82fcc31840fa821dae6b0dc6f1a80328
af05e1445503097aed3950319e37e82cf7842bb5
839088 F20101220_AACMRA bilgin_b_Page_11.jp2
40805935a08f9e3a6d72341dcc59fc11
eeff170ada9fb6eb6c712698140ce7fd2776fc80
F20101220_AACMTZ bilgin_b_Page_28.tif
e7ae5d6109dc79cdefd94f3c8ca63bb2
c4ead67147759317aec95b053aba11474248ae06
1856 F20101220_AACMYW bilgin_b_Page_40.txt
04855ff4d0dc785598b87b0eedf55e20
1685eb66727b64760ec6f459a02f105a58a382aa
8091 F20101220_AACNAX bilgin_b_Page_17thm.jpg
e1e41d85c6fee54322779fd4e0144626
baf49f48b0c509000fbcffc39f25095aeaf917ba
1051918 F20101220_AACMRB bilgin_b_Page_12.jp2
27e0ffad4455cb605e8efdd78d748363
62042665153f8f35b61bd3d9bd1eaa2ac74cca6e
374 F20101220_AACMYX bilgin_b_Page_41.txt
c80a27cb2879e91e0c56e1cf1081e450
31ee6c7a6cfa9a4f16042976d5ab055e5dd85a67
1051916 F20101220_AACMRC bilgin_b_Page_13.jp2
db10073c037466e33650ed98a7a4d8b1
7e430046f9a03b54a2526dd46938ad0bc27a6e80
56628 F20101220_AACMWA bilgin_b_Page_25.pro
5817d7ce30d290c9c1805b5456443f59
a90f07edf5d27fe364520b8b90b1029b8f4a95ef
1978 F20101220_AACMYY bilgin_b_Page_42.txt
8472b5c38c8a09722174e1946037a795
4a1edc0cd4b22f79c8a1776e5f0bda8f247965f7
30414 F20101220_AACNAY bilgin_b_Page_18.QC.jpg
811ae7a152369f91bca15e45b295724d
f3e17dee59951661f35a6c1681692d5ea8ae7fe1
1051950 F20101220_AACMRD bilgin_b_Page_14.jp2
640a508aeba9c646b765d20c24dd2c9c
b812d927129a35ec589173794389d98d0504d1c7
53786 F20101220_AACMWB bilgin_b_Page_26.pro
f485e12b6a2d17eaea0d8c4ac94c34d1
e23cb973545e7ee3f18d984b60d9c38e4e2b9090
2162 F20101220_AACMYZ bilgin_b_Page_43.txt
05ba4926637837891e97275fba147f2f
be7a23d49e70e54707266b4848c97ca58f2f6c6c
7772 F20101220_AACNAZ bilgin_b_Page_18thm.jpg
02c04be94b7e2903cb7423e8693d70e6
5d4f8a7949d4dce7eaf6d80598187a45fe07527b
1034770 F20101220_AACMRE bilgin_b_Page_15.jp2
1cf8a3b76c56308055d55b7dfe1696a4
bf3306cc7db218552f3077f9e534f76fa6be890f
50462 F20101220_AACMWC bilgin_b_Page_27.pro
3f3615b6ba288d9ac070bbdafbf056bf
121b2dca765562778553f6292e4a8fe7b5ebb5cc
1051951 F20101220_AACMRF bilgin_b_Page_16.jp2
4b85ead4c846d7c41b13afe3f211d090
26a69c31103902f0d8d76b9f79a02d2e2b7a4a41
8915 F20101220_AACNDA bilgin_b_Page_47thm.jpg
3127122eefb08456c5566ffac4238183
d5ed3439466ac236aceb4dc4474ff68f0cf4450b
25849 F20101220_AACMWD bilgin_b_Page_28.pro
27942e4ce3ab58b4fd4db407e9d3414c
c7491e038dd6a2eccf6c7790c164d99e492b8e43
1051969 F20101220_AACMRG bilgin_b_Page_17.jp2
0df930c6c05a7927e39e88b484bb0b3b
20b496848fd7133fa449f30c9e1ab999d2cdf367
35206 F20101220_AACNDB bilgin_b_Page_48.QC.jpg
beaf87f60b6fa1d74f0ac1be93b57a43
085803dad0050d9e8900c26d5e49f6e9bacc44d1
45834 F20101220_AACMWE bilgin_b_Page_29.pro
8c983a4846e191d769fd5727f4cbb247
c15252fdbafc00c3df44efbc3aaf73ff9fd36898
1051957 F20101220_AACMRH bilgin_b_Page_18.jp2
ab30b970fe68a0177d79aec715f35969
80fab73065a26c4574b3b989782ae46c968dd038
8618 F20101220_AACNDC bilgin_b_Page_48thm.jpg
5b5dd22f04d16229916715c0bd2bdeef
f9232609e59397c087c300bee979f338fbe089af
57505 F20101220_AACMWF bilgin_b_Page_30.pro
84f40ef9fd07d1a0baae35ad6fb3a717
8012116abf2804933b7895a8266e5d8b2b723ad9
1051958 F20101220_AACMRI bilgin_b_Page_20.jp2
1ffdbe9f0665ecf05302652d7571d409
edb3ab61e6c93e4fa56d973f84558a59c307a315
35622 F20101220_AACNDD bilgin_b_Page_49.QC.jpg
da31a19ff09ba774e142aab9a73ecf8b
1d2d198e80c01bce8f5b1a646a79d192796d6263
49955 F20101220_AACMWG bilgin_b_Page_31.pro
b798ce5af09939c0a0c662dce502be85
f8c5083b811a8c063ae523e50cf31835f6d786b6
1051979 F20101220_AACMRJ bilgin_b_Page_22.jp2
63923062d4d93ddcee0a7c5ac4c95461
3f0341c180f61007b3ca7259b5764f0b55a0e278
35362 F20101220_AACNDE bilgin_b_Page_50.QC.jpg
74507c9103c642284642613373b5e545
3cba9fe92b05472cafc140deeb0d094bf32041de
29856 F20101220_AACMWH bilgin_b_Page_33.pro
1d5bb20b741e03d31ed82c4baf79e898
9dc654f1d79ae42c18b6af70363817b09bc29d4d
37663 F20101220_AACNDF bilgin_b_Page_51.QC.jpg
aae69eda6eb32cd05362ecc96fa17d6a
9696547d822f3a51d5ca4a37f4f0025b5385339b
45456 F20101220_AACMWI bilgin_b_Page_34.pro
e3806899fefd1474fddfddedfe1b0c26
2681f75bcc6e6fe31f819fe8d71b6610ba903c41
962500 F20101220_AACMRK bilgin_b_Page_23.jp2
b90c056021cf7da0bb366da1cb354eb6
9ad9b03e05881fc57a09d8e4322a2cfd22269d70
8759 F20101220_AACNDG bilgin_b_Page_51thm.jpg
1b20079215742cf28d88e4b47bd2559d
5adbe6be4b7f936a4e05fc7a4524bae1b0322efe
39984 F20101220_AACMWJ bilgin_b_Page_35.pro
51c8f3b7777a1b7c991d887fac2f4f72
eab64c51a21f921047055c1c4b7b5a6858a9a59c
1051965 F20101220_AACMRL bilgin_b_Page_24.jp2
d007bafd85143ac7c16a609d00bd2818
4282768afffa14d2c95356f8c0a392689be3448c
35135 F20101220_AACNDH bilgin_b_Page_52.QC.jpg
8b8ba45b8250d4fb6a9db3b35326374b
0d53606d4ee52c6349567b45c5b7376c7230eb3e
54047 F20101220_AACMWK bilgin_b_Page_36.pro
ec97452305777ad34d5a3ffbd0543e77
14724c523ae10c9ab5aa70e1eba5a0d4cc280d3d
1051923 F20101220_AACMRM bilgin_b_Page_25.jp2
4948c25d786f97decc7cdeed8cf864c5
5f9c65509ecb86f34474f0d8b4d1a9f8bc150c16
33287 F20101220_AACNDI bilgin_b_Page_53.QC.jpg
9adefffa31b784467abd4942135660bb
a9f20c839c3764c6f8af47fbb5496a7c3a690306
42950 F20101220_AACMWL bilgin_b_Page_37.pro
3e628b62b28e94f6e07ed3a812ab709a
02f3c3250b52ea713c3c80c2acd73ad36fa2bd1a
1051968 F20101220_AACMRN bilgin_b_Page_26.jp2
3d1e9ff77c8c03a6ac87789933e44459
e33ede9487894519eb39243bbcd7ddaeed7bb738
8288 F20101220_AACNDJ bilgin_b_Page_53thm.jpg
beebf7428bac4d17736984dc80124e82
6858be16d586de9c4b2b000c842d097abc16e2f2
24065 F20101220_AACMWM bilgin_b_Page_38.pro
2d66716788f1a0b6f2106079e7b9c3f8
a0fa13ddd06f4b4a953934bff9a800d1a9dad6ab
1051984 F20101220_AACMRO bilgin_b_Page_27.jp2
2ab0788ce2c6879ff82228d9ef7c5963
ffc99dddb9e5931f204ab594531fc6e7e610e632
35829 F20101220_AACNDK bilgin_b_Page_54.QC.jpg
f605aaeceba6b9c4819fcac1e2015356
6b613a3b2c580f27bf69d1264180dbf5d155152f
41350 F20101220_AACMWN bilgin_b_Page_39.pro
37b69bdda4e9600c8e1cb49b9bdf2e2a
5b14cc464e3df8335f70396a2b6ab0b9eddf1df4
574392 F20101220_AACMRP bilgin_b_Page_28.jp2
189a8afeaf0b8ce542dd97fdc0b937d3
0c29224371e24f78600164d1c1cfa793e0287fa5
8389 F20101220_AACNDL bilgin_b_Page_54thm.jpg
e4b45191c1d8a88a41e2b357d2bfc134
683a711ecb711edd34fd7dbdc34f08afa32f29b2
40265 F20101220_AACMWO bilgin_b_Page_40.pro
d2085d48f19169c7d71bf4221b170bea
41979c673cfc37b69a014dd465a1e70e88a13e8f
1034863 F20101220_AACMRQ bilgin_b_Page_29.jp2
bdc3f262f3d399ee1c58b73c63c12f23
af1c44253820eec1164df934a5c61d3039c4199f
35928 F20101220_AACNDM bilgin_b_Page_55.QC.jpg
81a563e3890065d2325183361dd75d0c
2c00eb7cbb51d49731d66e2b2ad1ce25b0b7c928
1051936 F20101220_AACMRR bilgin_b_Page_30.jp2
cb5a7765fab95fd7227e51e136c211e3
6328f24cfccbf5469d1787a524cae39fe51c73b5
8813 F20101220_AACNDN bilgin_b_Page_55thm.jpg
981391282262d4760a29a4734fe5b45f
caf088ee80395a40570b59a49e1c300635569e84
9321 F20101220_AACMWP bilgin_b_Page_41.pro
e9e39e2a78ef798710011ebc625948c6
fde5643ea1f59086b51afa88eb8f716cd06de144
1051955 F20101220_AACMRS bilgin_b_Page_31.jp2
ef07019c3595707bd2f9becab6911274
990e7ecff1284defeaa8bcd732b740f137414c7c
5878 F20101220_AACNDO bilgin_b_Page_56.QC.jpg
70ad1ac7c9a4eabc795a0f0afba702a9
eeaba2080c1d83794f11d9e35509b9225aa5d0c1
47034 F20101220_AACMWQ bilgin_b_Page_42.pro
d6af383a47607fc210ef0bd3b23d842c
e88e8b2e47b18637f53d6606f1811d94fdd34e77
1012290 F20101220_AACMRT bilgin_b_Page_32.jp2
4faab45d4a4ccea79693cd7d0183e486
99c721d2422bee1adb32fcac64f545d80ff6b368
1591 F20101220_AACNDP bilgin_b_Page_56thm.jpg
7da1e7f39d309d3b195d686b2a1a1126
87e2f24770250635d8d8bc6c473047451210a240
53979 F20101220_AACMWR bilgin_b_Page_43.pro
ccbd399f9c1075d23f98161cff6761fe
f7fa94dbce86bbdb9d2e3d6563bfde659d52bf33
703646 F20101220_AACMRU bilgin_b_Page_33.jp2
415bd3e9f1366ae4e754e70fa859a680
b84c27b1cfca8eec6e66ccd1a8f3223d0fcba7b6
1846 F20101220_AACNDQ bilgin_b_Page_57.QC.jpg
c4ff90a6fd63c04df157ecec3675593b
0dc69c84422b5ca029eb5466f94fe22a61dba8c5
28317 F20101220_AACMWS bilgin_b_Page_44.pro
1a5370c2cc1b33cd4d0fb0401e1a9cb5
5662be3e61abf0d790882f6e1e4bcebf59203a35
1022337 F20101220_AACMRV bilgin_b_Page_34.jp2
775cd5eb382fbf5160edbdc58420e02b
e75336080be2db23c5d334baf3783ba875820e9a
46125 F20101220_AACMMY bilgin_b_Page_32.pro
73cea798b651a2b0aeade6ee9c6b1051
dc32259fd79705c3c0ed9d66534e6f9e809590e2
16702 F20101220_AACNDR bilgin_b_Page_58.QC.jpg
3da5b31fbbd3c50ce4624b50ae9422de
3005cbc0437a84eda1466e8d188ec257fad8f206
5056 F20101220_AACMWT bilgin_b_Page_45.pro
968073e56df6f7c6659d5ddc34d6ae00
7b29ca13f001248353f42d401ddf2b63c211d4a6
904055 F20101220_AACMRW bilgin_b_Page_35.jp2
8322679e67edc061600d928af38539ca
f248fd968cfb682935303f3aaa01ad43b259bed1
28446 F20101220_AACMMZ bilgin_b_Page_40.QC.jpg
2e5a5730bf13787ec9faf6eb9044619b
863cdc7747cf602e8221bc24468f030fe5043d6b
4515 F20101220_AACNDS bilgin_b_Page_58thm.jpg
672c3de586d56dcae6acd301ebb3617c
d5a57464671233941b82f052c6533ee36480250c
55768 F20101220_AACMWU bilgin_b_Page_46.pro
e010b9d321ba98cceb726f6b5b46c413
2ceb336c6f9c199a0e95e3a49c8cfbc042822703
F20101220_AACMRX bilgin_b_Page_36.jp2
7325a7b38fdb1981524c6155ccce58f6
68540a025304fe926c0bec80eff5a3eaaf22db1b
35183 F20101220_AACNDT bilgin_b_Page_59.QC.jpg
cbaca7e0053b83c7de64a78ee0ae7d2b
43ffb7d4ef565bd01ee1a5efc7876c59f0b4000e
54768 F20101220_AACMWV bilgin_b_Page_47.pro
874bc6b47f4ca59b23d4d8dc6d59a707
7e62ec0d228bb8844b2a06836c3b3a893cdf049d
101786 F20101220_AACMPA bilgin_b_Page_17.jpg
47154de59c14eaf11a64e9608399494c
4f5a4878f614c190ee52829fd9b2cea1831d1991
941867 F20101220_AACMRY bilgin_b_Page_37.jp2
54253a3a96059d5f8f39838adb2c3dc2
d998cb8358c995c53ef28eceb40256210553b395
39016 F20101220_AACNDU bilgin_b_Page_60.QC.jpg
af50fa26288432ffbadc5b8666e6db86
aebf1f2401e94776bd76304d9bb18557575a7200
54728 F20101220_AACMWW bilgin_b_Page_49.pro
f7f19b48b215c89d47cbb792b7bb6d8a
5cfd877f4fad892e30dbe7b6e3171a4d3668a2c4
97975 F20101220_AACMPB bilgin_b_Page_18.jpg
dc0463f86b9d9aa254c09e02dc1ff598
d5c16a2b3a2f42694d708758e1999da2cda9f853
534274 F20101220_AACMRZ bilgin_b_Page_38.jp2
8c051c1af4faa8ba789daf75bd0db470
5a6894aa75894d34050cef60a507f087736e4cf9
9148 F20101220_AACNDV bilgin_b_Page_60thm.jpg
43f51a60c428054e770231586fdb1451
e46944ba06dd02f69da82817fb6517090d2d0f7d
114851 F20101220_AACMPC bilgin_b_Page_19.jpg
7ef6f543a8cbb12580fbd7e32bc470c7
1ecd963bdd3c48ea040d44881efc0800613941f1
56247 F20101220_AACMWX bilgin_b_Page_50.pro
35be261ec8c1376f201ff21387ffd407
6f761e8a4a3409a6d8f9419bd93ab8ebe19c39b6
34648 F20101220_AACNDW bilgin_b_Page_61.QC.jpg
eb83b4540292700ea49909e41ac02221
9fec39deb7569e650376262347b0868e5bdff27b
97852 F20101220_AACMPD bilgin_b_Page_20.jpg
e321386b6fa4ac9e0bb2d22e323b063c
e7432a2b3eb7a49e655cb7ea20e7ce9e95872a5f
F20101220_AACMUA bilgin_b_Page_29.tif
d510c98ae11fa3a7acd9939c1accadb5
215a448b159f16ea47fa5422a7e5999d05fd3859
57421 F20101220_AACMWY bilgin_b_Page_51.pro
895ff885cf63e3da009649bfda8179c8
2445d1a5c8053f4101bd95dec7a6bed714371250
8764 F20101220_AACNDX bilgin_b_Page_61thm.jpg
fcc70a21e74248fa79d3584834f6d31d
e8d95c0bf9693f98c5911a665cca618c04061ab5
96312 F20101220_AACMPE bilgin_b_Page_21.jpg
98beec55bd699f0f4cfd11824598d223
eb1ac0485d1768037176c7d9992148bb14d50d58
F20101220_AACMUB bilgin_b_Page_30.tif
79ff1f0c1a022a0ae4ef92820810a4db
05e4af96a11399fd5fb4463524bcb3e9bd929ebc
54595 F20101220_AACMWZ bilgin_b_Page_52.pro
18e1059bc1ff2d577d8d84bea363b8e1
1a830eeea1a3b2cf3e8659333a0698f803e49bb7
1575 F20101220_AACNDY bilgin_b_Page_62thm.jpg
ed73069de3e664c8be80212af7650b7b
3682865c6f6d07d88ff43715a8360e79718d5735
95213 F20101220_AACMPF bilgin_b_Page_22.jpg
661307e1fb025282eeb85a7140c81249
74c2957fc2cace8527663764a6387cd34e5afb43
F20101220_AACMUC bilgin_b_Page_31.tif
e4cba37a81c9929508472ca641170514
bb7e35f22f4db5e7932123aed8c13f0d7f66e793
10517 F20101220_AACNDZ bilgin_b_Page_63.QC.jpg
e92ca01f0c9e7abdf38beac18585c0a4
492fe92df302fe119259bb54a75dc45b4bc65762
86855 F20101220_AACMPG bilgin_b_Page_23.jpg
6b205d6450d16f1f6332f24e62cb0aba
8df92d860403f019ccf87765fc7f0b49fedc6ed0
1269 F20101220_AACMZA bilgin_b_Page_44.txt
3fd02bc931b60d402e30b35f4f62b899
428675f3e8e91367757bdda1ec1e3e33d0e001c6
35858 F20101220_AACNBA bilgin_b_Page_19.QC.jpg
1d82a18063a01305e98ef5b486a098b5
7b5a7650ee8597b3336a1f46a440187bc1b2cbe5
F20101220_AACMUD bilgin_b_Page_32.tif
4bc497563c677a661906516ec69a779b
de9068abdaaad8b619752a2f9dd48b0e0edf03d5
104668 F20101220_AACMPH bilgin_b_Page_24.jpg
23c8b16c65a27a9a569653c98d354512
567f885656932d563332b822e44a4dae1b1608c0
203 F20101220_AACMZB bilgin_b_Page_45.txt
8f2db5aafc3a9b293bb6ae8c5b229f8e
cdde14faa0f5a0455632df46db3aac8ede1147c2
8788 F20101220_AACNBB bilgin_b_Page_19thm.jpg
18c773e1a2bcb175e80668871d5edb59
0059ccf6887e512345680e381a4c8236de7e252c
F20101220_AACMUE bilgin_b_Page_33.tif
c4d1ab3035f01f314f5c1adbac985f92
524528421af9e19a8093c679903f2dee3c7ee898
2259 F20101220_AACMZC bilgin_b_Page_46.txt
7c5e45e1b96890ed5d837c1d76a3949e
a5bb384fee830d0c175d1910619e78dfc44545c8
32454 F20101220_AACNBC bilgin_b_Page_20.QC.jpg
8c33b634f9f1d0edfd8b28d13b84f5fd
c29810ec9043343f0c6298caab63e19e9129dea9
F20101220_AACMUF bilgin_b_Page_34.tif
3a03f078df9aeb838fc01edff7357145
57bf74cb7b2d4fe39a1030c9b5285131ef79ed5d
112798 F20101220_AACMPI bilgin_b_Page_25.jpg
0809c1d8534d95bfb20c83b150ce950a
275b08ff5d9bb0a216cac08542c9bd021eb12d4f
2182 F20101220_AACMZD bilgin_b_Page_47.txt
c2260ea7ded5f648c5b6c05900bb778f
c462a10b449c675eec9757fc8d413f3878bbb3a4
8271 F20101220_AACNBD bilgin_b_Page_20thm.jpg
69e3adc7a84fb984eec2bd902db61ec0
d5c8790763637c865229d41b54be7a6453cdc576
F20101220_AACMUG bilgin_b_Page_35.tif
4793af84bae520eab59ac35de7edabfd
cec2819ec817e68a21da47e19dc9437a30e38775
102152 F20101220_AACMPJ bilgin_b_Page_26.jpg
9a827f3c2f34a7ebb0a28af030d7a3a7
ed20f97480d1a184b362644581487425c5f0dc99
2136 F20101220_AACMZE bilgin_b_Page_48.txt
5dd1a51eb21322794c94918e98b5c723
bc7457efc09aba8f63d3d3b4a5df19a10b126817
32270 F20101220_AACNBE bilgin_b_Page_21.QC.jpg
3564b0a95fa0fa2f7d5f5cbaed54e081
23a9080a25f88ccf3ebb54459de38ad1312de866
F20101220_AACMUH bilgin_b_Page_36.tif
9f991a4ed1b23a8babb35dd2e3ddf50e
79df1978b562085df27b6fcc4daa3dfa321c7afa
2151 F20101220_AACMZF bilgin_b_Page_49.txt
6275d8364d23baa83202adf38136203f
fcfb9615a2e071f891913f57a52cbb154a264a4a
8306 F20101220_AACNBF bilgin_b_Page_21thm.jpg
eb3618129cb8f2ef14ac162d8a20e948
eed6675a7bf378d25119c2178a5cf27ec121625c
F20101220_AACMUI bilgin_b_Page_37.tif
f7257a82dd0bbeab9e131e21ed7c7ccb
4a73cfd150ba0e2a87d60c22142997d0f91b528a
101072 F20101220_AACMPK bilgin_b_Page_27.jpg
a9c0050b18894bf390acc5503f8ceebe
17d1f3e78f5a3f3e7da8256ed5066138b47b4587
2209 F20101220_AACMZG bilgin_b_Page_50.txt
966fe08212a9b8d424fe961727df6235
f4659400021318d1a79d2e1dfc8cffcbd1d63c04
30902 F20101220_AACNBG bilgin_b_Page_22.QC.jpg
03356d88cbade4d30505d487dac16495
f1f0fa362ee992748695c85b0a2ce5e32f11fd39
F20101220_AACMUJ bilgin_b_Page_38.tif
91bfaa0bb55bfdabbedd1a1529d5336f
9b98656bad3175097dbe3cfb6bc4ee08296556b9
53582 F20101220_AACMPL bilgin_b_Page_28.jpg
b936d972e76025f0e677daf63bee1f37
6953bfde84bc9a75e65b2a1cd25ff6cae09644c0
2288 F20101220_AACMZH bilgin_b_Page_51.txt
4ac5f34ccf0f2e992eafc76d0fca92b4
52d3b57da404e3118a598c57be331994cf2dd5aa
7487 F20101220_AACNBH bilgin_b_Page_22thm.jpg
678c56908ecdc06acab4d866e4599881
71ce1087191e69cd0068d3bbfdaf20ad02600200
F20101220_AACMUK bilgin_b_Page_39.tif
86b318b4d056e2471d6d77a63e171172
6649f5c6052b91d859c95b50dfd2762a29587395
93762 F20101220_AACMPM bilgin_b_Page_29.jpg
0089fd6b48a335f0240c772c37f9f437
4de5d8cd2cb01df5db6e84527f978f853ed9a120
2183 F20101220_AACMZI bilgin_b_Page_52.txt
3d7ab74afe8af29bf1d3b60806456804
122f207d03b7d2b6f1683fb949a3447a1ad08102
28494 F20101220_AACNBI bilgin_b_Page_23.QC.jpg
7ae0cb4778e66f9c9f3c86fa301dda08
42f4fd37169636f0813f7b38ab30741a6d6ae8cd
F20101220_AACMUL bilgin_b_Page_40.tif
b4c0ee11eb1bf3115f24bfbf46ceb4dc
d2975bb03985210e7dbb89d5278248b57b32e949
113050 F20101220_AACMPN bilgin_b_Page_30.jpg
184f87b492a18f68d27a022a6a1a7254
5e47f1002cced65b9bf2e0ed2ec5f097dfa0e546
2041 F20101220_AACMZJ bilgin_b_Page_53.txt
75525d6ec98b8cd5b0511b6b5295ed84
7c3f44744144aaea0fe758af2f6812ea6b1d4e8d
7193 F20101220_AACNBJ bilgin_b_Page_23thm.jpg
9d4fc63faf68036cc046b4a851404e8c
782f3bff5b0092a18ab53724920b8bd461396d53
F20101220_AACMUM bilgin_b_Page_41.tif
276f2b96a1a66c1b03ad0c383c8e21be
fbcc8ce0e4603b3dc1d59fe0856323a30abcf62c
99961 F20101220_AACMPO bilgin_b_Page_31.jpg
550294035c25027c27b13a8a1b09afd5
f822dd5d7dd8364b4b3d9ec572b1601992fdb2cf
2194 F20101220_AACMZK bilgin_b_Page_54.txt
7288c36e8340415d552e2d097a74bb31
696b0be4f81ffafee5ec5f726d4cd4ce36f0b5fe
8253 F20101220_AACNBK bilgin_b_Page_24thm.jpg
253989ebc72598ab3d5be89fd43f371e
e92b08985555bfd194cd1eec4f8e8ebb134bdc8b
94358 F20101220_AACMPP bilgin_b_Page_32.jpg
139a1154fa5a7fc5a7eae4d0f3d44e24
066a10ef125cdf78905b0f431e49bcf998340778
2148 F20101220_AACMZL bilgin_b_Page_55.txt
0b0730f9e357ae4836185616884101a1
2fccec339687cd01b5a68f132d8e721425518f9b
34730 F20101220_AACNBL bilgin_b_Page_25.QC.jpg
bae3decffe8304cfeba8607d47a7b926
03581671c9fcb0a1d827c7815711e54fe5815489
F20101220_AACMUN bilgin_b_Page_42.tif
ac124e581cbb32405c0f2f47fb589190
7c85b77bd8d8c6c2cf6ee183bb09b36ee2e074be
66032 F20101220_AACMPQ bilgin_b_Page_33.jpg
1762122a67dab7daf544f0de94f7fdc0
46c3da4d1fa373418f94776381bc6a4f5afb508b
270 F20101220_AACMZM bilgin_b_Page_56.txt
011cbfea6b512f87e2b32d74881a6bec
f9ba823153c1b412ecf76bd482c2b2f353f8de94
8810 F20101220_AACNBM bilgin_b_Page_25thm.jpg
cf43de206b362652d0bcbc470328c30e
61c1a6aa49020e01f149d81446c7e656a378088b
F20101220_AACMUO bilgin_b_Page_43.tif
951985ae7e2c3fff92df4b3332e75ca1
c8474d9ef1e37750f50e26779ce6f7dfeaac6466
84361 F20101220_AACMPR bilgin_b_Page_35.jpg
556a0e242298860de0743834545e338b
a951defb39cd69b3bbeb7fc269ef091b52ed6c3d
53 F20101220_AACMZN bilgin_b_Page_57.txt
e9800014cf9e6191ed748f1c68fbef51
68ed00d46bb03f8b073a574a2672b6fa31aa8d64
33163 F20101220_AACNBN bilgin_b_Page_26.QC.jpg
d9fcf1251017f06a4f1ad1eadfaacbfc
cf9934e5f27773ee4ac3c662a4c6c2f917948e12
F20101220_AACMUP bilgin_b_Page_44.tif
3041adc2a10914489a0cde49b138fcd9
d3b7d1a70a2545991834216ea011487ff4fe62f2
100661 F20101220_AACMPS bilgin_b_Page_36.jpg
b631612c63d29af49dcd9c7c0459d0d3
9dc91028653620a21198273f35e7504d6ad1b299
2217 F20101220_AACMZO bilgin_b_Page_58.txt
b7769f54a178a7f25bc104d821dc3ba5
8bec936b29380cfcfd664bd841344258c14178d7
8085 F20101220_AACNBO bilgin_b_Page_26thm.jpg
63a901c2ab4e2fead182a9846938d369
84a9c63319f167e021891d7fb322084c9884d3cc
F20101220_AACMUQ bilgin_b_Page_45.tif
1ffd92f3e0146d88e8793a7e61350b89
15a8dbd83ea80939c553aac0c8201220be97da1c
89731 F20101220_AACMPT bilgin_b_Page_37.jpg
d4248e4ae432b1eaa6119106d28b4366
bf0bd68035df4ffd306a59908edf4410907b3748
2507 F20101220_AACMZP bilgin_b_Page_59.txt
4cc059bd32371442b01c68bc74fc5e5d
48cdd0c0b8981afa6dd69c392c7171dcd20add8c
32903 F20101220_AACNBP bilgin_b_Page_27.QC.jpg
6f915f68ea5161c7bd8ce36b52ff0976
5c5578f0ae9210766bb9ff19188e9448fe327309
F20101220_AACMUR bilgin_b_Page_46.tif
424cbf1d5eb42deb5fe26d7ce45f631e
bd7ca46224fb450a354ed4980b2026b59202b580
50229 F20101220_AACMPU bilgin_b_Page_38.jpg
f8869c7f0c5b2a62a06cd124e092592a
2185a57f6aed49a43937356cb1fe7148469d3dd6
2596 F20101220_AACMZQ bilgin_b_Page_60.txt
3f3fd789ae37eea110aa863237a3d0ef
906cf37075efb877c0d331c1150ca9523a9b0f89
7942 F20101220_AACNBQ bilgin_b_Page_27thm.jpg
93da1ff9c5a384271be1b194223c16b5
f8370b6e6eeaa50f1bc412064bad1b43f7473b17
F20101220_AACMUS bilgin_b_Page_47.tif
600bd2baac03c4a5cd5f5a96f7370465
41f7b8edfcb529e7a7db2e0f7236a5956ec75041
84296 F20101220_AACMPV bilgin_b_Page_39.jpg
6bfa5404edd1631d6000cda512c4b83a
5594604025566f68ab6671fa54892070ef89ff0a
2330 F20101220_AACMZR bilgin_b_Page_61.txt
5750ad72ca90ee604ebc89df26d7bdd8
4c60e12f7a8962ccbc1c8caee16c79cedb6b2af0
17727 F20101220_AACNBR bilgin_b_Page_28.QC.jpg
a12053d8f93a83242f6dbdef84b1efc3
1ce9c867e5fae49e2846219516a8457be00443c5
F20101220_AACMUT bilgin_b_Page_49.tif
9618709be3205ade9535fbc854d210a0
8b69e967556fb5b37887efb5e11f2cbcbe8725db
86042 F20101220_AACMPW bilgin_b_Page_40.jpg
5bd831a230fccc9eb2c9b78c3ecd7015
fe25c86f1370b83afb76b8b2c15f1145f882e2f2
4305 F20101220_AACNBS bilgin_b_Page_28thm.jpg
5eb91d123ee52c3077a82422622c0e5d
2c25104bc13e3721fb6d60df667e2f57e3ff5799
F20101220_AACMUU bilgin_b_Page_50.tif
2c543ee1d05dae5249514671b083d3ff
1007537ae71aa3d23007bf0b301ef34dd0649e35
21995 F20101220_AACMPX bilgin_b_Page_41.jpg
3761e95ddaee974b579378ba52adc4aa
e5b60d4dfea4e9135d47ba9f6f184198b89b3086
280 F20101220_AACMZS bilgin_b_Page_62.txt
c8abb16b2aaaa5550fe3f6f1b296af4f
cadde09426a9f958a41ab56bf5d802be3fa72764
29103 F20101220_AACNBT bilgin_b_Page_29.QC.jpg
80909ba56963474f667b2c822a9e96d5
2d91466ab7a722a959ed8e51ba9e0256fab732c5
F20101220_AACMUV bilgin_b_Page_51.tif
617b4176fe36fc5a31a282a6287819c8
2a60f339eb051afa3b5628ae39e2e7f0a8b62a58
35670 F20101220_AACMNA bilgin_b_Page_43.QC.jpg
93752c75ad1eeb5d8e6cf84ce7587e40
4ef66202148b946012cfdb2a39d663d0c9e2d055
93239 F20101220_AACMPY bilgin_b_Page_42.jpg
de9c4182e181f40632ed307fbe3d881f
f2ebdfbbb067a5e1f6a550687060434f631d7a5f
7210 F20101220_AACNBU bilgin_b_Page_29thm.jpg
c1c6e0dad06d96c87baa735ae776092c
c5898efdaed969854cc9e40c11cd0058fc1b9de4
F20101220_AACMUW bilgin_b_Page_52.tif
9c2643a0e746ec656b9dac63e148596a
1d5f618582a4b9ac1b8caa4e53c3afe57332b5e4
8772 F20101220_AACMNB bilgin_b_Page_50thm.jpg
28f6f320dd4fc751f3bed31241d23f17
f155b3c0beda34f818fd705f68a28a0fc0497a2c
108106 F20101220_AACMPZ bilgin_b_Page_43.jpg
b52681f8080b20a17ca5547b3f7cb75b
aa2ee3d9c4737a06f2b7dd281fed03618de2f3f2
569 F20101220_AACMZT bilgin_b_Page_63.txt
05adf0466be75231ce721b8700de0e9c
f2287a8e93ce619beb887590e7d49bc7b05d204a
36829 F20101220_AACNBV bilgin_b_Page_30.QC.jpg
f41331ca651215039cdd26b55c640b01
bc341f092a9c2251a040899422d607503323e4d2
F20101220_AACMUX bilgin_b_Page_53.tif
ec3d204365d4d93cbf4c17f5af43a49b
59625a7fecf863370a2f20b42783b020aafd7792
24795 F20101220_AACMNC bilgin_b_Page_11.QC.jpg
3446009d17ac09adc0f9a6d64b14caba
8525584de6b21be9d7bf3ba4562ea79b612b4878
1959 F20101220_AACMZU bilgin_b_Page_01thm.jpg
a2068909378130c9722069209926f11d
cdf313bf401f769201901339ab9efed18a5675b8
F20101220_AACNBW bilgin_b_Page_30thm.jpg
9653332a78591605639157591a856731
938e60622e798429942024f444f866a3e7a4d086
926561 F20101220_AACMSA bilgin_b_Page_39.jp2
e7fc54b3dc7cdc3297a05e4f8e79ca04
262528b204eb8e7ed4f4ff6974dd57bc1a15791c
F20101220_AACMUY bilgin_b_Page_55.tif
cece129e50b434b28b91e70188b1dfe2
776a5b127564be425ba9ebce60f80b68d0329a78
1044618 F20101220_AACMND bilgin_b_Page_21.jp2
34942081c5af967c35ffc014041bb00f
19877bacfaf7e1bf42f6fc2e6e1d1b2ed705649d
8413 F20101220_AACMZV bilgin_b_Page_01.QC.jpg
32beb03464754d560496d8f81be68d28
99ad98a8604aacf31fb96eb2aaa35c2b687d5243
32349 F20101220_AACNBX bilgin_b_Page_31.QC.jpg
5406d79fd4629a3e2e451425e7beddf2
980c560e61aa9f2f88001974180252b6f79cb5f3
901992 F20101220_AACMSB bilgin_b_Page_40.jp2
e355843d837410c41cba3682232e2f07
8cfac51a109ac24a0281ab80f38c08815820d454
F20101220_AACMUZ bilgin_b_Page_56.tif
b1a1ce61548e1abab928833acdf6848b
a2679a47600bc4cb72bf49d55386f9dd8b720c49
57873 F20101220_AACMNE bilgin_b_Page_16.pro
4bc28a8976ec31eba522870cdd63074d
8a93497089b3385d4b20778d9d845253583953ee
1090 F20101220_AACMZW bilgin_b_Page_02.QC.jpg
c405afd69c8e60a80655ac51cdf04e53
98aa6b23109d4746184f474a8d9dc3f1ce954606
8150 F20101220_AACNBY bilgin_b_Page_31thm.jpg
cc8867dc0fe596e470f12fd2b74af8d2
dbe8659146a8d0d9334a4cc78041c1d6cdff2dfb
216841 F20101220_AACMSC bilgin_b_Page_41.jp2
e4c279bc465f074976000e8700af95fb
69709b0af91b4ed6cc3be5ed70c0e6bdff3a0664
8557 F20101220_AACMNF bilgin_b_Page_52thm.jpg
36fcb0affdd10d32365d74a5bab45c65
fff619eb6bc89465f96ac1a54673b69185d60497
1347 F20101220_AACMZX bilgin_b_Page_03.QC.jpg
8e1725ff02ec99ea23f4b5c0c46d69ac
f8e03b27a1d169b0270375ff4d2a682cb1e692ff
1049095 F20101220_AACMSD bilgin_b_Page_42.jp2
ad885332c17bf8af80deebea8b029920
41cd294362bc8ae60363094fb1aa312a8b45e824
51823 F20101220_AACMXA bilgin_b_Page_53.pro
3b54bb538f65b74733d0595d5deb47ed
384808133b4e543e52e0531a140a7f32668e5840
536 F20101220_AACMZY bilgin_b_Page_03thm.jpg
a47de9058fd561865ce78e02de0ca5c8
6ffe07a0bf4a5f7f3e3fc47c4e0a017fb2a24ba2
31419 F20101220_AACNBZ bilgin_b_Page_32.QC.jpg
e7b4d9b1061560178b22ed025faed2da
1e7faf9ec1ac174fb566762cb6c3196b64e2bbb9
1051986 F20101220_AACMSE bilgin_b_Page_43.jp2
14989eb150428d9ee95dc8c1744c9227
b8c3117c80c2f8b71ea0d68ce22c24ca68e5337c
33411 F20101220_AACMNG bilgin_b_Page_10.QC.jpg
d8c7421fd497fbf54973fe568155da2b
3cb2c1620c6e50e519feb19beb406389229c1d09
54521 F20101220_AACMXB bilgin_b_Page_54.pro
c8826a812c0f5b57b218838cdc918a00
19147206995ff3123f8d31c47b446f7c04a21ffb
24495 F20101220_AACMZZ bilgin_b_Page_04.QC.jpg
24ad3c71c1cb3d8a5672487b8bead10e
baf0ccc4e69c4ed53a052c68976029f9c1e6d24e
887511 F20101220_AACMSF bilgin_b_Page_44.jp2
7ea29209ddd1af0c9b54cb5e608acd16
a85d532ec66ce5eb6fc1041e2b1672522e45d495
8640 F20101220_AACMNH bilgin_b_Page_59thm.jpg
49da5a480361fcb1b61e109601319ca9
53214289cfcfe8c07e4dce51a0766217be857ba3
54720 F20101220_AACMXC bilgin_b_Page_55.pro
b923c891b45cac30c9c78920686e2a0c
327cdfdeacf25e0d181039a1587664c8e7920175
2513 F20101220_AACNEA bilgin_b_Page_63thm.jpg
aca06e1af8e309e188493524a49053b5
f434de9e2d44b4c9275baaba4dc5012326072ee7
116174 F20101220_AACMSG bilgin_b_Page_45.jp2
6205de201939f11c0c9ceba84aead780
807b8d88ee65391f756e8148ca6f48186fec5092
103964 F20101220_AACMNI bilgin_b_Page_53.jpg
e33849b9ec45e6661d34245cdcb72e5e
55aa530d979c30e44a4f94cdfbbacde05f043982
6683 F20101220_AACMXD bilgin_b_Page_56.pro
2bfc6ee08ff05cba3ae19406c991cca0
abbd70e820215d2fcfe02d124f0f91fd7de54950
74978 F20101220_AACNEB UFE0019761_00001.mets FULL
2d821d35d3770d8a258652a604b5224f
56f759d9f3c6bbad7833ce22bbbe8fba6ebae352
34425 F20101220_AACMNJ bilgin_b_Page_24.QC.jpg
5d4bc1fb5eaae538499debf51fa80ab4
fac8c2671f7f00a9e9fc22de668bcafc85167439
1052 F20101220_AACMXE bilgin_b_Page_57.pro
8ba30ef856cd894b47d72c19886bed8f
a7e3223025b8fbd1c2d322423a59f05c24c78994
1051944 F20101220_AACMSH bilgin_b_Page_46.jp2
9b64ae728270ee8021b31f5b2495a1d7
fb4607f224a01ec0ce541c3745eeef45764fbc8f
2325 F20101220_AACMNK bilgin_b_Page_25.txt
8a3ac287708ccef0599ef76d553371bd
d3f88b5e022d63f61ca65e62e339b7215ae092bd
44314 F20101220_AACMXF bilgin_b_Page_58.pro
c4630cf1b0b45d5aa4eabede99a1536e
57c0cb1df4b4aabe8cff4e2779554ecd67e12901
F20101220_AACMSI bilgin_b_Page_47.jp2
408b5b781a5946c1e459e2848f56c799
2a7804aeb2780dc07ec67bb67380fa7c675663dc
560 F20101220_AACMNL bilgin_b_Page_57thm.jpg
889e8d1388c0055c58f2a6a7058c4048
caf647ca1baa9cf188357977c6cba0b9821dcfcc
62202 F20101220_AACMXG bilgin_b_Page_59.pro
a4e1247a1399d1cbd5f7d34865b9e684
5fa4a3278103fd8ecff7832ded8b4df54c27806e
1051956 F20101220_AACMSJ bilgin_b_Page_48.jp2
d99f6c5ce1d7d9885841012b48767006
d6f71c158c61cb98758fbb0f1c59a4dcb7a3e315
91613 F20101220_AACMNM bilgin_b_Page_34.jpg
aacaf7af509898f9bff0a2ff006ec84d
6befcf4583ede0779170e7f0fca05cd320d0327f
64467 F20101220_AACMXH bilgin_b_Page_60.pro
8e703d08ec6a4a8462bf00ef448a2c13
a35d4b115d531d55adf443fbadedcad9607de022
F20101220_AACMSK bilgin_b_Page_49.jp2
0a7ff6f912d0d0a304eac678cec963a4
c24fb307871b6a6298c2d543d71e8eb97a0dab58
112436 F20101220_AACMNN bilgin_b_Page_46.jpg
a23b22e560e472b56437b650c1436fa5
a15e1b219fa0c160433c64f4a65d607b53d7b5e2
57301 F20101220_AACMXI bilgin_b_Page_61.pro
6af70a8d05cc8ba34bcf5541259ade0e
3a25e0b09be9b6fb1dfb0e188a70cfa5fdcddb48
53608 F20101220_AACMNO bilgin_b_Page_24.pro
30f70d39ed9acf0c652083d4b6c07853
fa0ac9df87cb6ee2e9adda35f769a385d9913541
6724 F20101220_AACMXJ bilgin_b_Page_62.pro
eaddecdd3938b64a26f8e81c94060fe2
387e5366da31ecb19d4e301bd5464a0e7381426b
F20101220_AACMSL bilgin_b_Page_50.jp2
2f340726fe153f866797477550388e30
1ae85e0895ca4cb499bed265e5ac3d9ca88d22c6
F20101220_AACMNP bilgin_b_Page_48.tif
1864d50efbe3c734d1f78da3d37c41cf
941f9de26aa7efdc076bd4a4784af76df5686bd6
13321 F20101220_AACMXK bilgin_b_Page_63.pro
906d2a2b4ae90afae8c87ac6f43b36c5
25c40b035a02e1f219195839edf973e219ec7bcb
1051959 F20101220_AACMSM bilgin_b_Page_51.jp2
2a5464d45c7e0716bc592a3a3aa9d1b9
fd8966ebec699d1b49204a69f5e41a72b8798194
F20101220_AACMNQ bilgin_b_Page_01.tif
57fc6c77f741ec7e985d1fcbd330737b
4b1acf1b3bb16ea57962c961043c450c1eaa4328
482 F20101220_AACMXL bilgin_b_Page_01.txt
6dd72537fa793b97be8a77bad1b37944
a17c3b274c63db14cb40b62e33c8c0a624a9aff2
1051978 F20101220_AACMSN bilgin_b_Page_52.jp2
281ba94981018c4cd5824782991137c2
6dd584eaaad85a46a5ff80a320eab769aa300715
67364 F20101220_AACMNR bilgin_b_Page_05.pro
266237f7124f926b835b1e14a4962ec9
1eb21bf699ff0a79711958a35b7303d14f4d80d5
82 F20101220_AACMXM bilgin_b_Page_02.txt
84653b5a183e645acea87051b99d5fa6
e95e8efe73bbc5b63da5c2934f329d42773d3da2
1051948 F20101220_AACMSO bilgin_b_Page_53.jp2
b19d0ed951d53d36e206c0ec54c6e22d
2fed36dadb91ccc64fd327dcbc5ada50656004dc
30003 F20101220_AACMNS bilgin_b_Page_42.QC.jpg
673a815e3427f31108b2192304dd274a
5ec653b1bee4e637cf3e5779c65b07ae1e5bcd0d
85 F20101220_AACMXN bilgin_b_Page_03.txt
8624251ba4dc4405cd95c4c2cb666c9b
c08f76cf95a46ba4d6ffb1b3c53b29b1b7541741
1051967 F20101220_AACMSP bilgin_b_Page_54.jp2
727625eba10283e3f8f278025dfabc7a
28337a7e6eaa40f040f20ecb1331f85a70b01f32
5518 F20101220_AACMNT bilgin_b_Page_62.QC.jpg
4a0579bf0ee050c04b4ad79ba5ef70ac
d152522f772df7a07e6dc0a4aa6d8a5d6a4b441b
1438 F20101220_AACMXO bilgin_b_Page_04.txt
668135cd83ccad872f70daddacdd70dc
b9091220815b2853ba73331137878ec4efbcbb90
1051981 F20101220_AACMSQ bilgin_b_Page_55.jp2
eb9c6529398047b5e56bb27e189b8770
f6f3bee8d3dda1cd9a688015f5e18313a540286b
1051966 F20101220_AACMNU bilgin_b_Page_19.jp2
77c9f0b36ae7ae3ac94e0e259ea49910
8ffab9dc856ea41cd60368d4dfc6b845604944ac
2839 F20101220_AACMXP bilgin_b_Page_05.txt
73e8a50f247057fa9da1bc4235c60c19
a44f554abce6fb87af5bec06d37212d267c26e89
155110 F20101220_AACMSR bilgin_b_Page_56.jp2
4e9582754c33bb7de35463ee4001f322
d27301e848c1d135fd7a4942c66d586bdf0fc650
275328 F20101220_AACMNV bilgin_b.pdf
06e097d6d8fab28a74ab9b06a592106d
8bbc85abcff4de58a23cd617e54035fcc86ae27a
36612 F20101220_AACMSS bilgin_b_Page_57.jp2
36462b981b31afd47d6b298ccf8c1f94
8edc7325a82a37a7eb1e18d3a5b66c805c933f26
F20101220_AACMNW bilgin_b_Page_61.tif
6b10de71b1651f41f24b2041842b68c5
5a834f067cfb00fa84f3251ede8065271333ba29
2436 F20101220_AACMXQ bilgin_b_Page_06.txt
709496d271903ca061e0e0a51d1134f1
6d8a7cd00135b86334a0af7403b0aa4f4784233c
906893 F20101220_AACMST bilgin_b_Page_58.jp2
440afdb3ae20ae7c0e2e1dffa180be1d
e597711e5b8e9f51b533a5a4cc7f4a6fbef17309
1983 F20101220_AACMNX bilgin_b_Page_31.txt
b7b85e6c5681084216a14af07b214de7
4bf8d57f7378998a5e1c136cf2bc65e1dee00027
356 F20101220_AACMXR bilgin_b_Page_07.txt
28bb0cc31f180b9bb0b165a395dfa976
d1ef296786cffabfa8a102cf4ce24fe456a400d0
1051983 F20101220_AACMSU bilgin_b_Page_59.jp2
28897332da4546a5ee14fbb63d872b8d
47c7a883d2712c76bcce06d0e04db6820b406363
8721 F20101220_AACMNY bilgin_b_Page_43thm.jpg
787b1ff02737445b441422af1f1f0c01
e1eaca7184133d5ad07285206922b7049942e879
1231 F20101220_AACMXS bilgin_b_Page_08.txt
7064587aaf0fd5977ab2c1da878abc6c
55b8d35f91294a369f0657296c7f1021da3f5775
1051971 F20101220_AACMSV bilgin_b_Page_60.jp2
ca13e4ab086e32ca509cd3f0ea518fa5
2645557be58c6e6b264f25f2611e1d39f04ff906
266280 F20101220_AACMNZ bilgin_b_Page_07.jp2
49204c9cbb9d6c1bfbf35a11588ad626
f36c93f250e2a5c999f8d70b686069386c70c474
1507 F20101220_AACMXT bilgin_b_Page_09.txt
1ab554167af23a4b4451db9d5e6e5915
9d6217120b1f1de6eba02b491524c36ba9e1ab61
1051963 F20101220_AACMSW bilgin_b_Page_61.jp2
e8075dff8495f37d6698ceb4b2aea70c
ee51a5505baa44e0cd2418fba4ed1a284195a8c3
F20101220_AACMXU bilgin_b_Page_10.txt
6ad2ef0ffe634734ed9202669802f9d4
a8c82c351d3a3781da17ac05ade1d6bc941c426c
171799 F20101220_AACMSX bilgin_b_Page_62.jp2
3027b1c604bc1bdaf6835c20aabce6d3
4d2819f0bd133a0d08ecd3e8e4044417824a972f
1485 F20101220_AACMXV bilgin_b_Page_11.txt
5bd526f545bd1487f571864d498a38b8
c18038026d104da14388738195cd69219ef5f821
83988 F20101220_AACMQA bilgin_b_Page_44.jpg
80b5e28b01a098f5b84db120e728dc26
8d73ec69cf38c7a21adfee634d41a3dbb12c4d8a
312708 F20101220_AACMSY bilgin_b_Page_63.jp2
752142f7fe5545b2ab0a79a59cfdf9fc
e246e3b3496a028c7939545b7b0c1f583ebf4083
2060 F20101220_AACMXW bilgin_b_Page_12.txt
920aab85e0f5d50981103b9db4cc9080
2bef44fbda6eae953cc99289b5b343373959fa00
13131 F20101220_AACMQB bilgin_b_Page_45.jpg
8d90b1380c9ed67b55d70b4722ac2342
7c383857546ffaf8e4241dff09f56f40e06e6b9f
F20101220_AACMSZ bilgin_b_Page_02.tif
6187b96a43a0a82f96b3423d4b206c8d
97a273ee70818d1437a721d3762dd5c2cd02ac1d
2026 F20101220_AACMXX bilgin_b_Page_13.txt
e147534508593eb9e0f6df93ed3cc497
078050103fb98abe12a9a72a85fded34ed4b7da6
111466 F20101220_AACMQC bilgin_b_Page_47.jpg
26ec70e31f39f64b3f8fe682c00d6e21
6210c0537bae798293c8d37a2acb314fcafe409c
F20101220_AACMVA bilgin_b_Page_57.tif
02f82af1d0b21fb5aa42b8fe763910fe
1a7d13ac200bdfb08e51a437b37b5041c2b3df4c
2105 F20101220_AACMXY bilgin_b_Page_14.txt
26e74825954203f067be4b08bd271741
942e9624bf7ddab6737e2da32fe378ae3eac417c
108383 F20101220_AACMQD bilgin_b_Page_48.jpg
c63512db6a6886df3e2662dd70dff979
17a9403b1a5bde77044f1b07f30744e756ff767e
25265604 F20101220_AACMVB bilgin_b_Page_58.tif
30e50722e45b9c102eff3a4805e91fe8
b2b5486593d06ff05fb51205b9b069d1569fa446
2065 F20101220_AACMXZ bilgin_b_Page_15.txt
0d1d5aa1de4b3ac09585241fcacee0f0
7c715f6c9d28fa5355f89fc3fd815af67a88432b
110348 F20101220_AACMQE bilgin_b_Page_49.jpg
57bf8f61c281665d117ea66144c86a6e
ec471244dab3943f1b2d06b475d3b8e6c9d68bc4
F20101220_AACMVC bilgin_b_Page_59.tif
2b9e487aeabae703dfa0e61f2f8cbf2c
dd67dcda0584aac93b4c58c4f454069f7200e92a
110521 F20101220_AACMQF bilgin_b_Page_50.jpg
6801c4343f285785a7b894d7049b7a16
4db689bd5533b338b95cf93884cba3373564784e
114459 F20101220_AACMQG bilgin_b_Page_51.jpg
794eaf8d755d7cb613f01d259d915cea
bdadbcd8000a00d66012effb103e614adef94c7a
7347 F20101220_AACNCA bilgin_b_Page_32thm.jpg
66f93eea2ffa5244c7bf51d1804ea178
32c8eae14de525693f93118be56e56941eeffa28
F20101220_AACMVD bilgin_b_Page_60.tif
b55f124535eed9950dfcfac3b60e08eb
aef7edd7e3fcdf7c0aa224702a47fba987b03f08
108239 F20101220_AACMQH bilgin_b_Page_52.jpg
1372d7a8bffe2ecad5b5df9e27742586
b46aa4ffdba9b695ec50ce8ec54e3c89987d0957
21866 F20101220_AACNCB bilgin_b_Page_33.QC.jpg
df381d0d5827c72a7374ed0ab6011820
0724395b90a8ac3c204809ed8d1956d6b7d554aa
F20101220_AACMVE bilgin_b_Page_62.tif
2b1e0238ec89ff74d66771417c963a47
b58cc9300669fe14c629d4684dcaa8d533871d86
110283 F20101220_AACMQI bilgin_b_Page_54.jpg
eabcfa956697faa5db809d27f760cfa1
7b939a7ead028d4fffb14fe9b117493760fffe1f
5919 F20101220_AACNCC bilgin_b_Page_33thm.jpg
a4ebe5ed041e7096ed798a15dee86074
6185c59bd23d4e3f22e4ab1485e6b760724278c8
F20101220_AACMVF bilgin_b_Page_63.tif
57a906d63b0d922d79bce9fdd91500cb
30ebc54a9baec8634b8882736c26f2792bb1c601
30148 F20101220_AACNCD bilgin_b_Page_34.QC.jpg
427214b257cb5463d7bd55146c02a634
9fb96a912c4946710c821027b2f0637e77c88c22
7836 F20101220_AACMVG bilgin_b_Page_01.pro
5efe72cb76913b52dcfd7d4ca180ea1b
e0a928d94f92ba41b6322fce8e2333ad385ca0a2
109164 F20101220_AACMQJ bilgin_b_Page_55.jpg
f6faaca99ce99184b60ac51711da77dd
aab21f427aea2d6b69baffbdaf0640c3f259652f
7473 F20101220_AACNCE bilgin_b_Page_34thm.jpg
e526772e6faeec6a5fef8fe9f1e003db
8c93b1045e49f81c95ba7966041f47be9bf226a0
777 F20101220_AACMVH bilgin_b_Page_02.pro
35d91fdf797152e3feddceb847600b34
7f0c7099bcf3743230d3a77e91f77504976106eb
16264 F20101220_AACMQK bilgin_b_Page_56.jpg
ee8a12ab2070daa81cd048e5f7175776
e5dc3f8b9818b6b28e076ad92705644152f43db2
27338 F20101220_AACNCF bilgin_b_Page_35.QC.jpg
a568d486585d63e50f0a2a6a0b366ff6
5e4709c02fd082c13059679ca34ffed22eef3f90
1063 F20101220_AACMVI bilgin_b_Page_03.pro
fde46aa9bcaaa021e193c0fcf07d15f5
6944779a1ad364ac7ad65edd994bef21f1e72674
5241 F20101220_AACMQL bilgin_b_Page_57.jpg
8b97154c1412ef1e22e059bf415a12b2
d178705d35b91c7beb12fd2adbb42f48f5cbcff4
6963 F20101220_AACNCG bilgin_b_Page_35thm.jpg
a82d51897ab8b78e2a054e3e649a1326
138a13cfdef27f97cc7b7b13820f7c30eeaf251c
35191 F20101220_AACMVJ bilgin_b_Page_04.pro
65771f48d943ab33dfd2e4902e2470af
4f09fe924d548e529a8ed8366e19b76a00d60ff8
56754 F20101220_AACMQM bilgin_b_Page_58.jpg
5f0e9c2689c5de7949bebdc9d167e04b
e01e8219eba1dfa48a14683dfd833560021ef7a4
31951 F20101220_AACNCH bilgin_b_Page_36.QC.jpg
302a0b9b7f2034bb0aecb868163dbbb7
76192c04d97d1281620c322d58b2294011ec101b
7596 F20101220_AACMVK bilgin_b_Page_07.pro
4c1a4615d067feafd2bc4d5b899d8bb2
646454cb7fab9d61e8dc7b76c6b3c08f60453167
120133 F20101220_AACMQN bilgin_b_Page_59.jpg
fbb483e7f2c7ad6efa11d94a8a667b25
c9c30894f8db2f71b9a30e2f55213a0955ac8a91
8052 F20101220_AACNCI bilgin_b_Page_36thm.jpg
2a8a967f9d89afad21e17b4801f6afba
8e9396c562a0bf16eb956993d4ad7439cf5480a8
27234 F20101220_AACMVL bilgin_b_Page_08.pro
7e896d844d99c49fd685c90c6ba2ea2e
4efd66d35f3e0854ee7cc9445aaa594173752222
139620 F20101220_AACMQO bilgin_b_Page_60.jpg
008d9b89f6de506b340935ad5592f189
c3ee59cda9f0642f2f9725a63706e85a88b0e843
28357 F20101220_AACNCJ bilgin_b_Page_37.QC.jpg
614062021adea7747b60b02261b258c3
d565dbfe6197956eea5b6271c5a35803be40f7d3
32416 F20101220_AACMVM bilgin_b_Page_09.pro
6392bd3df9e34ae0322b6f3925dec694
de11399d33844e34c85c263363182c283b566d62
7636 F20101220_AACNCK bilgin_b_Page_37thm.jpg
c204246f90d4914e9a8926d14a8e9732
8533aea6b67499a1c949a26b8145a29344713242
120494 F20101220_AACMQP bilgin_b_Page_61.jpg
0ef630fd86743e8c86e05c6ffcecdfd9
f7d21a0e33cbace7f1e7740d9ad42ed89a284fd5
52939 F20101220_AACMVN bilgin_b_Page_10.pro
c42db683383959d99203e8820186ee5e
30b9112d937362232c2ee858a930da10e8252498
16400 F20101220_AACNCL bilgin_b_Page_38.QC.jpg
ca142a417bedf3faa13cc66f2ca59916
9bce9a688b3ea29c0701e7c3fb3ac1f83835a2e5
17414 F20101220_AACMQQ bilgin_b_Page_62.jpg
17fa66c374c67f97e2eae8eee580abf7
647328e4a03bc959369254d7de4f1ec1e83c6376
4177 F20101220_AACNCM bilgin_b_Page_38thm.jpg
4b48d6004b18bce1e2950c726f7d66b2
12562a0a054506e12b7670119987552438c71aff
31964 F20101220_AACMQR bilgin_b_Page_63.jpg
da85c0ffe6de611b8f27fbe3ca42c8ef
7abdc9f6f0db26f966097f8f9c826fbf67b2896b
37414 F20101220_AACMVO bilgin_b_Page_11.pro
c4e552444c8a4fd6bd96d8ea187705d3
f7ecd9c16501501e7a448624990bfb89ce325792
26969 F20101220_AACNCN bilgin_b_Page_39.QC.jpg
3c4f730f29766c91f470139197c03ce4
cab276e47759980a35ea5ee5aa539a37b369d037
240916 F20101220_AACMQS bilgin_b_Page_01.jp2
2d9f1df5e8dc4660a83e61037a807ff2
dc23ed9b32d37dc702fcb25fc07efacadd98888c
50150 F20101220_AACMVP bilgin_b_Page_12.pro
02a62c1b2678232cf57215f1e8ffd1b3
45cfaea1a21322a1a6ecd858c019079d6407ebf8
7467 F20101220_AACNCO bilgin_b_Page_39thm.jpg
a190946cdfbd27302e2f607284a18ebd
ac79125de5af8225f4ebfb779d350de03b1b89b1
22536 F20101220_AACMQT bilgin_b_Page_02.jp2
d6fba677d43f820730a3ca19d1e1b8e4
6df0440f8698f33dc6e4cdad395d965efe8b4d17
50772 F20101220_AACMVQ bilgin_b_Page_13.pro
aa9660ba6ac60f7f0e0246fc3cf043c3
caa89a118e2def79800e53ac63ae59fcd9a05dbf
7376 F20101220_AACNCP bilgin_b_Page_40thm.jpg
ce94365ee20f007e386d8802496528a9
0c7c67c8cfce75e415af5654aa3337460b66591f
29076 F20101220_AACMQU bilgin_b_Page_03.jp2
d29532061f8f49819283f77910ce0eb4
77ca434d2bf7f809e48e88465fa9e13047bf6490
51148 F20101220_AACMVR bilgin_b_Page_14.pro
2cb4272af3775d5204ab16d5b2cad64d
aae322e9e03a7affad03c7e677feae6a3be34697
7344 F20101220_AACNCQ bilgin_b_Page_41.QC.jpg
15da7b7f2cb38747c8b4dd7aaaf6be07
8ec51656028302bd4190085dbb7f66fe099edfa8
800326 F20101220_AACMQV bilgin_b_Page_04.jp2
d395d7f8e912e9346f9b702efbd144c7
9f3a2fc527860b3b5de5b5d8a980ccb8a0daf28e
47735 F20101220_AACMVS bilgin_b_Page_15.pro
8e6b904ceae170008df60147ade72f62
13047013c883cead8ad68603c76aac2a646290b3
F20101220_AACNCR bilgin_b_Page_41thm.jpg
8cffe8ea35a47582dbc1d3d662f16d97
001853d73ba58a8a0fa78ba8474a93666a19712f
F20101220_AACMQW bilgin_b_Page_05.jp2
3409d21e5a2c66ddc811579219290b2c
fe65c1fcd76e72100cf45b542af8c75dd96d0ee7
50973 F20101220_AACMVT bilgin_b_Page_17.pro
7f9d5691f43dcf0f8fb45c381c58edb6
91778a7a3c5fcf0cbc24a541dd2c2536ee0ca2b0
7866 F20101220_AACNCS bilgin_b_Page_42thm.jpg
a7a53783391c44435dc3f840b30da088
02b3176f53da52463492ecf25c45de992efabfb5
1051985 F20101220_AACMQX bilgin_b_Page_06.jp2
5dc05219fde4ff2a2c94c79905121e38
e8dcbe817035897ec1001c571d64d3db119fd8da
53230 F20101220_AACMVU bilgin_b_Page_18.pro
7ae529fe531b69c0d98cc31dc8d3ae2f
df864b9dc0d14685f03dc4157424b20ca746624f
26676 F20101220_AACNCT bilgin_b_Page_44.QC.jpg
e027827580d9e2eaa68b00a4ac1976f7
71536831aaaefb23e55df493c06173b581b1f6ad
752060 F20101220_AACMQY bilgin_b_Page_09.jp2
f9dcdd0c0426b19e78870f85497f90ad
3b80f5dc804fa0dd637644c719772593310cf839
60719 F20101220_AACMVV bilgin_b_Page_19.pro
77a54b2fc5edeb222731b86758452442
0d816ee2e3a457a4f97c376894912f1dba805b32
59642 F20101220_AACMOA bilgin_b_Page_06.pro
394b2a5b22a79071b029f122cdf82f94
58c89f48dab1dd07da1c8864e4501231f6d3af3b
6781 F20101220_AACNCU bilgin_b_Page_44thm.jpg
5431feadfceec54157111b45839f676c
146794cf504065c28e94f9f8cebd8bd154069c8c
1051954 F20101220_AACMQZ bilgin_b_Page_10.jp2
14fe9d16aa452b542d0686ac0dc62068
b7297dbc878b2a822f2b14221bbf5b6c15af2a14
45530 F20101220_AACMVW bilgin_b_Page_20.pro
95050c51779dbcd1f9eff50423e41e3e
a472a5f54c182c41a03082e92e7f255a897924f7
485 F20101220_AACMOB bilgin_b_Page_02thm.jpg
9137f59e5e416ea8741fd6c48cdede31
cd562899ab56e6c155685b70b3bef1e70ba7344d
4843 F20101220_AACNCV bilgin_b_Page_45.QC.jpg
8e489baba9a560d8f6bf9f6bcabc243c
c2649e5602c67999bc80b4bceecbcf84166a315e
45591 F20101220_AACMVX bilgin_b_Page_21.pro
8d6ae779e33551ff0cc8d1e478e9fd33
f0c2e9695d74dc168862f004ee9f086308dcfef5
F20101220_AACMOC bilgin_b_Page_54.tif
a6c8753d4b464ae9224afcb4450b6ca2
42c14accbbeaee8ba80ba359205d23d493f8cd99
1275 F20101220_AACNCW bilgin_b_Page_45thm.jpg
0c9ddae4eef96cb1913c38c01e508799
1c6c814ed5c3ef361101ec74c4128f45077e5846
F20101220_AACMTA bilgin_b_Page_03.tif
4a30897ea238168c8a574d312b0a7691
8b829569a7ed14367afaa77a7fa6a1eabe6aaaf6
46941 F20101220_AACMVY bilgin_b_Page_22.pro
4f8b3c5cfe3c10fa7719841804a78913
407332809f68d2a18675aae83c7d6170ed270db5
7758 F20101220_AACMOD bilgin_b_Page_15thm.jpg
40ffd466261decd3253c1fb7a00c0c4f
129b88366415eee2c087558e613582ae9f692559
36271 F20101220_AACNCX bilgin_b_Page_46.QC.jpg
9be5b0a992a6ec481dd9e104988b6ef5
3995fb91f44dabca2ae12d91b475fd45de5e74b0
F20101220_AACMTB bilgin_b_Page_04.tif
27d5ccc84f00500e43ccec83b9bb1713
c6901b0d27817f3d4ebb1bd7259a9cc599b2bb7c
42678 F20101220_AACMVZ bilgin_b_Page_23.pro
9247786ad8072e25a6735d3821323d8c
ae1cb81c21b718252ba78cfe27d19870a0aff8a6
8667 F20101220_AACMOE bilgin_b_Page_49thm.jpg
ee55455fe12e48a894ada2a7081e5c44
6ca7e44112d6596429d377cc006cb7094be5855c
8727 F20101220_AACNCY bilgin_b_Page_46thm.jpg
c2e0045d112d82266494fd71ed946ffd
aca150bcbcb10ccbfa32bc42483b280c2407c88c
F20101220_AACMTC bilgin_b_Page_05.tif
793ef68fa017f9bf184e2036676dc404
75e6c7998b5baa8cc4d71a641609eb5bb9cafc1b
930919 F20101220_AACMOF bilgin_b_Page_08.jp2
2b06edb72873c0ee664091f143208c2e
f9f9f214edfbac68578a7af5039b0fac38688b48
36398 F20101220_AACNCZ bilgin_b_Page_47.QC.jpg
f94eed7874d15e9a5054a6f88de9326f
5b028d09aa08ceaf52c236d0166f696b76f0bae6
6059 F20101220_AACNAA bilgin_b_Page_04thm.jpg
7bbbd98f622031849569e50d5aeb7cdf
92625c530b163dd5e9575cc240b4ae53990fbd33
F20101220_AACMTD bilgin_b_Page_06.tif
4ddc25326b53f831d01bcd364b951e54
405ab1438cf93705cab09ce18950c3b4e46c4c48
54300 F20101220_AACMOG bilgin_b_Page_48.pro
bacbba0ed6c1615801d0f7f1e22cd15c
3e76996b5cafe0943a1885f97bd1cd0aa01acc04
2266 F20101220_AACMYA bilgin_b_Page_16.txt
8505758423775dc6b13f0c1edfa3d55b
7ec9ef1d0f8bd4aad47fc33f26da678a552df582
21122 F20101220_AACNAB bilgin_b_Page_05.QC.jpg
e538011a04ff40e496311111a74c8694
f91295bda3604e605cbbadc0c5a55c133b8422f7
F20101220_AACMTE bilgin_b_Page_07.tif
a8b47ecec14932b26e25e65c162eb2d6
d11aad96704e817f247254a871b6e97f5e64ba0f
2008 F20101220_AACMYB bilgin_b_Page_17.txt
2cb53be46aa76fdea08d7831ff4a422e
d874516365dcb7103dedc0c3af9f99e365c914ce
5426 F20101220_AACNAC bilgin_b_Page_05thm.jpg
0abffd177666c5f4bd812353c323d733
6dccdffeebccf9233e2fe63423c15f754caff489
F20101220_AACMTF bilgin_b_Page_08.tif
554e7f29b4be8b8c703484ec022f5898
c6371e6ff9bd02289357fee7e96f9bbd7c11632d
96605 F20101220_AACMOH UFE0019761_00001.xml
fbbf242a84da5926077448bd18aec752
19e1352f6d8e022b88c963e2e8f869d49eba281d
2146 F20101220_AACMYC bilgin_b_Page_18.txt
209e23c8e90bbe90e8db300e6fac2130
ccf94957e1fbbe200f7e22f1d94a45a3fc3a45b8
18048 F20101220_AACNAD bilgin_b_Page_06.QC.jpg
fdbe9edbac4cd5744014048e0c9819cf
1cf5f610eaef6907cd9910bff2f5cbe3b5a24a53
F20101220_AACMTG bilgin_b_Page_09.tif
51861a3614b548fb2d1a873d30f9caed
185cb3a59cec69d2e1844d0c22d2ea252d45318f
2403 F20101220_AACMYD bilgin_b_Page_19.txt
3d8f95592ae23da27fd57a667213637b
8b22df3209ba5ad717ff11ad165e01c6d5ae018f