 |
Internet Marketing · eMarketing · Internet Advertising · Online Branding |
|
Internet Marketing · eMarketing · Internet Advertising · Online Branding |
|
|
|
||
| ADVERTISING BY ICONOCAST |
||
INTERNET SHOPPING, CONSUMER SEARCH AND PRODUCT BRANDING
forthcoming Journal of Product and Brand Management
Michael R. Ward
Assistant Professor
Department of Agricultural and Consumer Economics
INTERNET SHOPPING, CONSUMER SEARCH AND PRODUCT BRANDING
forthcoming Journal of Product and Brand Management
Michael R. Ward
Assistant Professor
Department of Agricultural and Consumer Economics
University of Illinois, Urbana-Champaign
Illinois, USA
ward1@uiuc.edu
and
Michael J. Lee
Graduate Student
Department of Economics
University of Illinois, Urbana-Champaign
Illinois, USA
mjlee@uiuc.edu
First Draft: April, 1999
This Draft: July, 1999
ABSTRACT: Recent interest in the Internet as a medium for commerce has raised questions about
the usefulness of branding on the World Wide Web. In this paper we examine whether consumers
use brands as sources of information when shopping on the Internet. Applying theory from the
economics of information, we predict that recent adopters of the Internet will be less proficient at
searching for product information and will rely more on brands. As they gather more experience on
the Internet, their search proficiency should rise and their brand reliance should fall. These
hypotheses are tested and confirmed using usage and opinion survey data from the Internet
community. These results suggest that branding can facilitate consumers’ acceptance of electronic
commerce.
KEYWORDS: electronic commerce, advertising, consumer search
1
INTRODUCTION
Commerce on the Internet, or e-commerce, has experienced rapid growth during its infant
years. The pace is not expected to slacken. Forrester Research estimates that online sales in the U.S.
amounted to $7.8 billion in 1998 and forecasts that this form of electronic commerce will reach $108
billion by 2003. While this would still amount to under 5% of all retail sales in 2003, it would
represent a dramatic increase in Internet retailing. Investors seem to believe that the volume of ecommerce
will grow considerably. For example, Amazon.com, a leader in electronic retailing, now
has a market capitalization of $22 billion, greater than either Sear’s or all of America’s bookstores
put together.
Online shoppers appear to be attracted to the ease with which they can find products on the
Internet, the detailed product information available and the variety of choices offered. Because of
the relative ease of vendors setting up shop, myriads of smaller retailers have embraced the Internet.
However, with the proliferation of online retailers, sellers are having difficulty distinguishing their
products or services from their competitors’, especially those of unscrupulous fly-by-night
companies. Consumers often bypass these problems by relying on branded products. Ernst and
Young recently reported that 69% of those surveyed stated that brand names play a significant role
in their online buying decisions. As a result, marketing through established brands may be required
on the Internet, even though consumers’ cost of information gathering seems quite low.
We investigate the ability of brand names to convey product information to potential buyers
as a substitute for consumer’s own information gathering activities. It has long been debated
whether advertising is used solely to promote brand loyalty (Dixit and Norman, 1978) and thus tends
to be anticompetitive (Comanor and Wilson, 1974) or if it conveys information more efficiently than
alternative mechanisms (Nelson, 1970, Nelson, 1974). We find evidence that suggests that
consumers with more years of Internet experience are more efficient at gathering product
2
information on their own and that they also tend to rely less on brand names when purchasing. We
infer that brand names are substitutes for consumers’ direct information gathering—at least on the
Internet—and thus may contribute to market efficiency.
Our findings are based on data from Georgia Institute of Technology’s Graphics,
Visualization and Usability Center (GVU) Eighth Survey of Internet Usage. The GVU has
conducted semiannual surveys of Internet usage since 1994. GVU’s Eighth Survey, conducted in
October 1997, includes information about respondents’ product search behavior, brand reliance and
a measure of their Internet experience. We use these differences in Internet experience to track a
natural progression from a relatively naive consumer who relies on brand information to a relatively
savvy consumer who has less need of brand information. Unlike the near ubiquitous experiences
and understanding consumers have with traditional retailing, knowledge of the Internet and how to
shop using it varies widely. The newness of the Internet and the relationship between Internet
experience and shopping proficiency allow us to identify effects that may not be apparent for other
forms of retailing.
THE INTERNET: INFORMATION AND UNCERTAINTY
The Internet was originally designed for the exchange of data between decentralized
computers and has evolved into the World Wide Web.1 The ease of publishing on the Web has
facilitated the adoption of this technology by consumers and producers of goods alike. With the help
of search engines like Altavista and Excite or portals like Yahoo! and AOL, consumers can obtain
product information and often make purchases with much less effort than through other distribution
channels. Likewise, with the low cost of Web publishing, firms can offer more product information
through this medium than most others. This results in more product information, on balance, being
supplied to consumers than ever before.
3
Plentiful product information may not alleviate all the problems of consumer search for two
reasons. First, despite the increased availability of product information, it is still not costless to
obtain (Brynjolfsson and Smith, 1999). On the Internet, search for information may involve a nontrivial
navigation of hyperlinks between Web sites and an intelligent usage of the search engines and
directories. For many users, especially those inexperienced to the Internet, finding product
information may be frustrating. Indeed, forty-six percent of those surveyed in the GVU’s Ninth
Survey in 1998 indicated that they had trouble finding new information. Thus, although consumers
may often like to obtain all available information, they may not practically be able to do so.
Second, even with the information available, some uncertainty about product quality is likely
to linger. Although some product characteristics can be easily illustrated or described on a Web site,
other product characteristics require consumption before their quality are known. For example,
firms can and do sell food products from their Web sites. A firm could state the price, ingredients,
and availability of its product, but it would have difficulty in both verifying the truthfulness of this
objective information and describing subjective information, such as flavor or feel. As a result, some
residual uncertainty about the product features is likely to remain.
The costs of search and the unverifiable nature of some product characteristics pose
challenges to consumers. Both problems limit the amount of confidence a consumer may have about
a product’s quality. These problems apply to all forms of retailing, but have specific consequences
when applied to the Internet. Because of the low costs of setting up a Web site, unreputable firms
offering low quality products could potentially claim their products are of high quality, earn a profit
before the ruse is uncovered, and then quickly disappear. Thus, even though the Internet can easily
provide more information than other distribution channels, the ease with which scams can develop
may induce consumers to require more information in order to purchase.
4
BRANDING AS AN ASSURANCE OF QUALITY
Information asymmetries between buyers and sellers can result in market failure (Akerlof,
1970). Buyers are only willing to pay the expected value of the products offered for sale. Sellers
of a high quality product, however, may withhold their product if their costs exceed this price. Since
this biases their expectations upward, buyers revise their expected value of products offered for sale
downward. This, in turn, could deter sellers from offering relatively higher quality, more expensive
products. The resulting adverse selection on the part of sellers can ultimately lead to a market
failure, one in which only the lowest quality products are offered for sale. For example, in GVU’s
Eighth Survey, the second leading reason (38% of all respondents) why consumers do not purchase
more products and services on the Internet is because they believed product quality is difficult to
judge. In order for markets to work, or work more efficiently, some mechanism must be adopted to
relieve the information asymmetry.
Nelson argues that brand advertising may be such a mechanism (Nelson, 1970, Nelson,
1974). He makes the distinction between two types of goods: search and experience. A search
good’s quality is verifiable upon inspection, whereas an experience good’s quality is difficult to
judge upon inspection. Therefore, only upon the purchase and usage of an experience good can its
true quality be revealed. A firm advertising a search good can directly inform its customers of its
product’s quality. In contrast, information regarding the merits of experience characteristics are
inherently unverifiable and may not seem credible to consumers. Where it is available, producers
of experience goods can seek credibility from third-party sources, such as Underwriters Laboratory
and Consumer Reports. But third-party information may be hard to come by for the many new
products offered by smaller firms operating on the Internet.
Economic models based on Nelson’s work show that an established brand name can signal
high product quality (Klein and Leffler, 1981, Kihlstrom and Riordan, 1984, and Milgrom and
5
Roberts, 1986). Essentially, high quality producers advertise their brand heavily, but only expect
to recoup the cost of the advertising from many repeat purchases. Low quality producers cannot
mimic this behavior because the true product quality will be revealed before enough purchases have
been made to recoup its investment in advertising. If a seller chooses to produce a high quality
product, it can overcome the asymmetric information problem and differentiate itself from the low
quality producer by developing a brand name and advertising more.2 In these models, companies
create brand-name equity to assure consumers of high product quality.3
RELYING ON BRANDS VERSUS SEARCHING
Brand names are just one source of information; most consumers also conduct some form
of product search (Stigler, 1961, Carlson and MacAfee, 1983, Benabou, 1990, and Benabou, 1993).
A general conclusion is that, since search is costly, in terms of time and effort, consumers will stop
short of becoming perfectly informed. If brand advertising signals useful product information,
consumers may rely on it as an “expensive” source of information (Butters, 1977, Pashigian and
Bowen, 1994, and Png and Reitman, 1995).
We test for this substitutability between consumers’ use of brands and search by relating it
to a measure that should be associated with exogenous changes in search costs. Specifically,
consumers should become more proficient at searching for information on the Internet as they gain
more experience with it. Navigating and evaluating information found on the World Wide Web can
be daunting. Looking for information from search engines, directories and portals are skills
developed with use. Indeed, authors have noted that a certain amount of experience is needed before
one develops the proper skills to “flow” in an intermediated environment such as the Internet
(Hoffman and Novak, 1996). Moreover, beyond merely finding relevant information, one must
evaluate its credibility. Time and experience are required to learn the credibility assuring institutions
6
that have developed on the Web (e.g., site of the day, moderated rather than unmoderated
newsgroups).
If proficiency in searching the Internet increases with experience or over time, then reliance
on brand names should likewise decrease. Internet search proficiency is likely to increase as users
gain more experience with the medium. Increased proficiency decreases the cost of gathering and
evaluating information, specifically product information. Alternatively, consumers can rely on well
-known brand names as shortcuts in evaluating the merits of different products. Unlike search costs,
increased Internet experience is not likely to make consumers more proficient at inferring product
quality from brand names. Therefore, as the “price” of searching, relative to using brands, falls with
increasing Internet experienceSand if they are substitutesSmore experienced consumers should rely
on brands less.
Applying this argument to the Internet at its current point of development is likely to be
more fruitful than applying it to other distribution channels. This is because current Internet users
differ widely in their level of their experience with the medium and this experience is likely to have
significant impacts on their usage proficiency. Other media used for obtaining consumer product
information, such as traditional retailing, telemarketing and catalog shopping, are all developed
enough that variation in consumers’ search abilities is not likely to be linked with identifiable
measures of consumers’ experience with the media. Thus, we are exploiting a natural experiment
that occurs because the adoption of the Internet is not yet ubiquitous.
To empirically test the hypothesis of brand reliance, we use survey data collected
semiannually by the Graphics, Visualization, and Utilization (GVU) Center at the Georgia Institute
of Technology. Though there are data limitations (discussed later), we will statistically analyze the
relationship between brand reliance, search proficiency and experience on the Internet. Using cross
tabulations and ordered logistic regressions, we find evidence of increased search proficiency and
7
decreased brand reliance as experience on the Internet increases. We infer from this that, on the
Internet at least, use of brands and search are substitutes and, therefore, brands convey useful
information.
DESCRIPTION OF THE DATA
The results of GVU’s World Wide Web User Surveys, begun in 1994, have been made
publicly available for academic research. These semiannual surveys record users opinions and usage
patterns for a large number of Web-based activities. These survey data, however, pose two general
problems to researchers. First, collected at high-exposure sites, the data may not fully represent the
characteristics of the population of interest—all individuals who use the World Wide Web. Only
those with a disposition toward highly trafficked sites such as Netscape and Yahoo! are likely to be
included in the samples. Second, only those willing to spend the time filling out the questionnaires
were included. This creates a self-selection problem: those who answer the surveys may not
represent the population of interest. However, self-selection should not affect within sample
comparisons as much as comparing the GVU sample to other similar data. Comparisons between
less experienced users to more experienced users, as in our analyses, do not depend much on whether
the sample of less experienced users are different, on average, from the population of less
experienced users.
We employ GVU’s General Information and Opinions on Internet Commerce Questionnaires
from the Eighth Survey (October 1997). GVU’s survey strategy is to require all respondents to
answer general demographic questions and then proceed to specialized surveys on particular
Internet-related topics (e.g., privacy, publishing, shopping, commerce, etc.). General demographic
information is available for about 10,000 respondents for each survey, with 1,500 to 4,000
respondents answering questions from specialized surveys. For our analyses, we used answers
8
related to online shopping proficiency from one specialized survey and brand reliance from another,
as well as general demographic information. However, since respondents were not required to
answer all the specialized surveys, only 1,671 of the 2,946 respondents in the shopping survey are
among the 1,987 respondents in the Internet commerce survey that asks about brand reliance.4
From the general demographics survey, the variables obtained from all respondents include
gender, race, marital status, educational status, income, age, and the number of years the respondent
has been using the Internet. Summary statistics of the demographic variables for both the shopping
and branding surveys are reported in Table I. It is clear from this table, that the GVU samples do
not reflect the population at large. In particular, survey respondents include a smaller proportion of
blacks, a larger proportion of males, individuals with a college education or an advanced degree,
people with higher incomes, and younger people than the general population. It is likely, however,
that these samples better reflect the population of people who use the Internet.
[take in Table I]
From the Internet commerce survey, a question is asked whether a Web site having a wellknown
brand name is an important determinant in product-purchasing decisions. The eight possible
responses can be grouped into four broad categories that indicate different levels of brand reliance
(see Table II). Our measure of brand reliance differs from the raw responses in two ways. First,
while it is clear that a brand being “necessary” is more stringent than a brand being “preferred,” it
is not clear if any of the “depends” choices are more stringent than the others. Therefore, we
aggregate all of the “depends” choices into one category. Second, while respondents were able to
select more than one category, few did (see Table III).5 For those who did select more than one
category, we put them into a category between the two they had selected. A series of questions from
the online shopping survey measure how successful respondents are at shopping online, how long
9
it takes respondents to find what they are looking for and how long it takes for them to give up their
product search. These questions are asked separately for both personal and professional shopping.
[take in Table II and Table III]
TESTS AND RESULTS
Our two hypotheses are that more experienced Internet users are more proficient at seeking
product information and purchasing items on the Web and that more experienced Internet users rely
less on brand names. The data that indicate experience, proficiency and brand reliance are
categorical variables that take on a small number of discrete values. Therefore, we test these
hypotheses with cross tabulations and test for a random allocation across cells.
Tables IV and V report cross tabulations for the percentage of the time respondents claim
success in finding the product they looking for and the time spent searching for a product online with
Internet experience. There is a general trend towards both more success and quicker searches for
more experienced people. For example, the percentage of people who are successful shopping more
than half the time (the Most and All categories) rises from 61% for those with under 6 months of
Internet experience to 72% for those with over 7 years experience. Likewise, the percentage of
people requiring less than 15 minutes (the top two categories) rises from 68% to 75%. Differences
across individuals with different amounts of Internet experience are statistically significant at the
2% level in Table IV and the 1% level in Table V. These results suggest that search proficiency
increases with experience.
[take in Table IV and Table V]
Tables VI report the cross tabulation for the time before one gives up product search and
Internet experience. The theory previously outlined has more ambiguous implications for this table.
More proficient searchers may be willing to search longer if they are more confident of eventual
10
success, but may give up sooner if they expect results more quickly. It is not clear a priori which
effect should dominate. Nonetheless, the table indicates statistically significant differences across
individuals with different levels of experience. For example, the percentage of people giving up
within 30 minutes (the top three categories) falls from 77% for those with under 6 months of Internet
experience to 68% for those with over 7 years experience. These results are consistent with the
hypothesis that search proficiency increases with experience.
[take in Table VI]
Table VII reports the cross tabulation of Internet experience and measures of brand reliance.
Again the differences across respondents with different levels of experience are statistically
significant. The table indicates a rather steady decline in brand reliance with experience. For
example, the percentage of people ‘requiring’ a brand name falls from 11% to 7% between the least
and most experienced Internet users, the percentage who at least ‘prefer’ a brand name falls from
48% to 34% and the percentage who ‘don’t care’ rises from 16% to 27%. These results suggest that
reliance on brand information falls as people gain experience on the Internet.
[take in Table VII]
It is possible that the above findings are a result of a relationship between demographic
variables and experience. For example, it is possible that more educated people both have more
experience and are more proficient at searching, which leads them to be less brand reliant. In this
case, the effect of experience net of education could be negligible. In order to control for this, we
also ran ordered logit regressions of the search proficiency and brand reliance variables against
variables measuring gender, race, gender, marital status, educational status, income, age, and
experience. Doing so requires that we impose a parametric structure on the data rather than the nonparametric
cross tabulations above. Specifically, the categories for product search and brand
11
reliance are assigned ordinal values from 1 to 5 and 1 to 7 in the case of the brand reliance
regression.
The results of the ordered logit regressions are reported in Table VIII. In general, they
confirm the findings from the cross tabulations. Internet experience tends to increase measures of
search proficiency and decrease brand reliance, all else held constant. We found only a few of the
demographic variables to have significant effects on shopping behavior. Men both spend less time
searching and give up searching more quickly than do women. Race has no effect and marital status
only has a marginal effect. Education appears to make people less successful at search. People with
higher incomes may be more successful searchers, even though they spend less time searching and
give up their searches sooner. Senior citizens seem to be particularly poor online searchers.
Nevertheless, none of the demographic variables—other than Internet experience—significantly
affects brand reliance.
[take in Table VIII]
Our previous discussion hypothesized a link between brand reliance and experience on the
Internet. Since brands are used as time-saving devices to signal quality, the opportunity cost of time
should be related to the reliance of brands. More specifically, those with a higher opportunity cost
of time should rely more on brands. And since individuals with greater incomes generally have
higher time costs, they should be more brand reliant. Other research finds evidence to confirm that
higher incomes are associated to a greater dependence on brand names as a source of information
(Pashigian and Bowen, 1994 and Png and Reitman, 1995).
Our ordered logit regressions provide a direct test of the relationship between income, a
measure of opportunity cost of time and brand reliance. These tests failed to find a significant
relationship. We suspect this was due to measurement problems in the income variable. First,
income represents an imperfect measure of an individual’s opportunity cost of time. Other
12
idiosyncratic factors affecting opportunity cost, such as fondness for the Web, may be related to
higher incomes. Second, while the respondents’ current income is reported on the surveys, no
information can be obtained for the respondents’ permanent income which more closely related time
costs. Those individuals with a much higher permanent income than current income—such as
college students—may be more reliant on brand names. If this is the case then the expected
relationship between brand reliance and income will be confounded.
In summary, our results indicate that a significant statistical relationship exists between
Internet experience and both search proficiency and brand reliance. Thus, we find some evidence
to validate the hypothesis that as individuals gain more search experience on the Internet, these
individuals will be less reliant on brand names as a signal for product quality. Instead, more
experienced searchers may opt to find more direct product information to discern product quality.
CONCLUSION
Our results indicate that as individuals gain more experience using the Internet, they are
more likely to search for alternative sources for information and be less reliant on product branding.
This finding is consistent with the substitutability of brand advertising for search, especially for
consumers with relatively high search costs. We infer that this supports for the notion that branding
does not merely promote product loyalty. It also conveys useful product information that tends to
make markets more efficient. Our results suggest a number of possible hypotheses for further study.
First, we conjecture that as the Internet population matures, brand reliance to assure product
quality may give way to reliance on direct product information, more easily found because of the
decreasing costs of search. Our findings could have implications for the future of branding and the
level of advertising on both the Internet and in general. As more consumers obtain access to the
Internet and gain proficiency at searching for product information, producers may find that they need
13
not advertise as heavily to signal their products’ features. The advertising that producers do
purchase is likely to be directed toward consumers with higher search costs, or those without Internet
access. If so, advertising on the Internet, where consumers have relatively low search costs, may not
reach levels comparable to other media, e.g. television, newspapers, magazines.
Second, the Internet may lead to a general increase in the level of quality of consumer
products.6 Brands are an imperfect mechanism for assuring product quality. In particular, we found
evidence that direct product search may be a more efficient mechanism, at least for experienced
Internet users. If so, the total cost of assuring product quality may fall due to the Internet, making
firms’ investment in quality more lucrative.
Third, we might expect that markets for consumer goods will become more efficient because
of the commercialization of the Internet. Our results also suggest that consumers are more informed
about the products they search for on the Internet than if they had to rely information gathered
through traditional means. Otherwise, they would not be willing to forego reliance on information
conveyed through brand advertising. Models of product search (e.g., Carlson and McAfee, 1983)
suggest that inefficient firms are viable only because consumers lack information about more
efficient and less expensive alternatives. Lower search costs due to the Internet may lead to a
weeding out of these inefficient firms.
In general, many of the existing models of decision making under imperfect information may
apply to the advent of the Internet. Whether or not our specific results hold up under scrutiny, it is
generally believed that the use of Internet will decrease in the cost of gathering and conveying
information. If so, models that rely on information costs could yield interesting differences between
Internet and non-Internet based activities. Thus, future research that applies models based on
adverse selection, moral hazard, free-riding or costly search to the Internet could be especially
fruitful.
14
REFERENCES
Akerlof, G. (1970), “The Market for Lemons: Quality Uncertainty and the Market Mechanism”,
Quarterly Journal of Economics, Vol. 84 No. 3, pp. 488-500.
Benabou, R. (1990), "Search Market Equilibrium, Bilateral Heterogeneity, and Repeat Purchases",
mimeo, MIT.
Benabou, R. (1993), "Search Market Equilibrium, Bilateral Heterogeneity, and Repeat Purchases",
Journal of Economic Theory, Vol. 60 no. 1, pp. 140-158.
Brynjolfsson, E. and Smith, M., “Frictionless Commerce? A Comparison of Internet and
C o n v e n t i o n a l R e t a i l e r s , ” w o r k i n g p a p e r J a n u a r y , 1 9 9 9 ,
<http://ecommerce.mit.edu/papers/friction/friction.pdf>
Butters, G. (1977), "Equilibrium Distribution of Sales and Advertising Prices", Review of Economic
Studies, Vol. 44, pp. 465-491.
Carlson, J. and McAfee, R. P. (1983), “Discrete Equilibrium Price Dispersion”, Journal of Political
Economy, Vol. 91 No. 3, pp. 480-493.
Comanor, W., and Wilson, T. ( 1974), Advertising and Market Power, Harvard University Press,
Cambridge, Massachusetts.
Dixit, A. and Norman, V. (1978), "Advertising and Welfare", Bell Journal of Economics, Vol. 9
No. 1, pp. 1-17.
Ernst & Young, LLP. (1998), Internet Shopping: An Ernst & Young Special Report, January 1998.
<http:/www.ey.com/shopping.html>
Fournier, S. (1998), “Consumers and Their Brands: Developing Relationship Theory in Consumer
Research”, Journal of Consumer Research, Vol. 24 No. 4, pp. 343-373.
Graphics, Visualization, and Utilization Center. (1998), GVU’s WWW User Surveys.
<http://www.gvu.gatech.edu>
Hoffman, D. and Novak, T. (1996), “Marketing in Hypermedia Computer-Mediated Environments:
Conceptual Foundations”, Journal of Marketing, Vol. 60, pp. 50-68.
Kihlstrom, R. and Riordan, M. (1984), “Advertising as a Signal”, Journal of Political Economy, Vol.
92 No. 3, pp. 427-450.
Klein, B. and Leffler, K. (1981), “The Role of Market Forces in Assuring Contractual
Performance”, Journal of Political Economy, Vol. 89 No. 4, pp. 615-641.
Lynch, D. and Rose, M. (1993), Internet System Handbook, Addison-Wesley Publishing, Reading,
Massachusetts.
Lynch, J. and Ariely, D. “Electronic Shopping for Wine: How Search Costs for Information on
Price, Quality, and Store Comparison Affect Consumer Price Sensitivity, Satisfaction with
Merchandise, and Retention,” working paper November, 1998
<http://ecommerce.mit.edu/forum/papers/ERF104.pdf>
Milgrom, P. and Roberts, J. (1986), “Price and Advertising Signals of Product Quality”, Journal of
Political Economy, Vol. 94 No. 4, pp. 796-821.
Nelson, P. (1970), “Information and Consumer Behavior”, Journal of Political Economy, Vol. 78
No. 2, pp.311-329.
Nelson, P. (1974), “Advertising as Information”, Journal of Political Economy, Vol. 82, No 4., pp.
729-754.
Nichols, M. (1998), “Advertising and Quality in the U.S. Market for Automobiles”, Southern
Economic Journal, Vol. 64 No. 4, pp.922-939.
Pashigian, B. and Bowen, B. (1994), “The Rising Cost of Time of Females, the Growth of National
Brands, and the Supply of Retail Services”, Economic Inquiry, Vol. 32, pp. 33-65.
15
Png, I.P.L. and Reitman, D. (1995), “Why Are Some Products Branded and Others Not?”, Journal
of Law and Economics, Vol. 38, pp. 207-224.
Stigler, G. (1961), “The Economics of Information", Journal of Political Economy, Vol. 69 No. 3,
pp. 213-225.
16
Table I
Descriptive Statistics - Percent of Sample with Characteristic
Searching
Sample
Branding
Sample
Male 62.7% 67.2%
Black 1.8% 1.5%
Married 45.1% 43.4%
Divorced 11.5% 11.9%
Living Together 10.1% 10.8%
In College 5.5% 5.3%
Some College 28.3% 27.4%
College Grad 28.8% 29.4%
Post Grad 20.1% 21.8%
Income $20,000-$40,000 26.6% 25.6%
Income $40,000-$50,000 12.1% 11.9%
Income $50,000-$75,000 19.4% 19.5%
Income Over $75,000 19.7% 19.6%
Income Not Say 11.5% 11.5%
Age 25-39 39.8% 40.9%
Age 40-64 43.9% 43.5%
Age Over 64 3.6% 3.3%
Internet Exp. 6-12 Months 18.7% 16.7%
Internet Exp. 1-3 Years 35.1% 34.0%
Internet Exp. 3-6 Years 20.5% 22.4%
Internet Exp. Over 7 Years 8.7% 11.3%
Observations 2,829 1,954
17
Table II
Electronic Commerce Questions from GVU 8
Q: How important is each of the following when you consider ordering a product/service over the
web (even if you have never done so). (Please check all that apply.)
That the company and/or products have a well-known brand name:
Possible Answer: Coded as:
Site must have this Require
I prefer sites that have this Prefer
Doesn't matter to me Don’t Care
Depends on what I'm ordering Depends
Depends on how much I'm spending Depends
Depends on how well I know the company Depends
Depends on what information is being collected Depends
Don't know Dropped from sample
Q: When you are intentionally searching for product/service information, what percentage of the
time do you find what you are looking for?
Possible answers are: All (close to 100%), Most (close to 75%), Half (close to 50%), Few
(close to 25%), None (close to 0%), Not Applicable
Q: On average, how many minutes do you spend searching before you find the first piece of useful
product/service information?
Possible answers are: Less than 5 minutes, 5 - 15 minutes, 15 - 30 minutes, 30 - 60 minutes,
More than 60 minutes, Don't know, Not Applicable
Q: How many minutes on average does it take you to give up a search if you cannot find the
product/service information you were looking for?
Possible answers are: Less than 5 minutes, 5 - 15 minutes, 15 - 30 minutes, 30 - 60 minutes,
More than 60 minutes, Don't know, Not Applicable
18
Table III
Description of Branding Responses
and
‘Require’
and
‘Prefer’
and
‘Depends’
and
‘Don’t Care’
Total
Require 130
68.4%
33
17.4%
26
13.7%
1
0.5%
190
100.0%
Prefer 33
3.8%
609
70.8%
245
28.5%
6
0.7%
860
100.0%
Depends 26
3.3%
245
31.4%
560
71.7%
50
6.4%
781
100.0%
Don’t Care 1
0.2%
6
1.3%
50
10.8%
408
87.7%
465
100.0%
Total 190
100.0%
860
100.0%
781
100.0%
465
100.0%
1,954
19
Table IV
The Relationship between Internet Experience
and Online Shopping Success Rate
Internet Experience
Success Rate
Under 6
Months
6-12
Months
1-3
Years
4-6
Years
Over 7
Years
Total
None
(0% of the time)
5
1.07%
2
0.39%
4
0.41%
1
0.17%
2
0.82%
14
0.50%
Few
(25% of the time)
56
11.94%
50
9.67%
75
7.61%
43
7.52%
13
5.35%
237
8.50%
Half
(50% of the time)
122
26.01%
120
23.21%
222
22.52%
134
23.43%
54
22.22%
652
23.39%
Most
(75% of the time)
222
47.33%
261
50.48%
548
55.58%
300
52.45%
125
51.44%
1,456
52.24%
All
(100% of the time)
64
13.65%
84
16.25%
137
13.89%
94
16.43%
49
20.16%
428
15.36%
Total 469
100.00%
517
100.00%
986
100.00%
572
100.00%
243
100.00%
2,787
100.00%
Each cell contains both the count of respondents and the column percentage. The P2 value for
differences across columns is 29.9 which, with 16 degrees of freedom, is significant at the 2%
level.
20
Table V
The Relationship between Internet Experience
and Time Spent Searching for Products Online
Internet Experience
Time Searching
Under 6
Months
6-12
Months
1-3
Years
4-6
Years
Over 7
Years
Total
Less than 5 Minutes 101
22.95%
132
26.29%
236
24.89%
177
31.72%
84
35.00%
730
27.16%
5-15 Minutes 199
45.23%
213
42.43%
432
45.57%
242
43.37%
97
40.42%
1,183
44.01%
15-30 Minutes 93
21.14%
103
20.52%
194
20.46%
93
16.67%
39
16.25%
522
19.42%
30-60 Minutes 27
6.14%
35
6.97%
69
7.28%
33
5.91%
14
5.83%
178
6.62%
More than 60
Minutes
20
4.55%
19
3.78%
17
1.79%
13
2.33%
6
2.50%
75
2.79%
Total 440
100.00%
502
100.00%
948
100.00%
558
100.00%
240
100.00%
2,688
100.00%
Each cell contains both the count of respondents and the column percentage. The P2 value for
differences across columns is 33.2 which, with 16 degrees of freedom, is significant at the 1%
level.
21
Table VI
The Relationship between Internet Experience
and Time Until Online Search Given Up
Internet Experience
Time to Give Up
Under 6
Months
6-12
Months
1-3
Years
4-6
Years
Over 7
Years
Total
Less than 5 Minutes 41
8.91%
30
5.92%
53
5.53%
26
4.63%
17
7.17%
167
6.13%
5-15 Minutes 138
30.00%
174
34.32%
256
26.69%
157
27.99%
72
30.38%
797
29.26%
15-30 Minutes 172
37.39%
186
36.69%
340
35.45%
205
36.54%
72
30.38%
975
35.79%
30-60 Minutes 77
16.74%
88
17.36%
235
24.50%
139
24.78%
46
19.41%
585
21.48%
More than 60
Minutes
32
6.96%
29
5.72%
75
7.82%
34
6.06%
30
12.66%
200
7.34%
Total 460
100.00%
507
100.00%
959
100.00%
561
100.00%
237
100.00%
2,724
100.00%
Each cell contains both the count of respondents and the column percentage. The P2 value for
differences across columns is 47.3 which, with 16 degrees of freedom, is significant at the 1%
level.
22
Table VII
The Relationship between Internet Experience
and Reliance on Brand Names
Internet Experience
Brand Reliance Under 6
Months
6-12
Months
1-3
Years
4-6
Years
Over 7
Years
Total
‘Don’t Care’ 49
16.01%
54
16.56%
156
23.46%
93
21.28%
60
27.27%
412
21.08%
Between ‘Don’t
Care’ and ‘Depends’
5
1.63%
3
0.92%
22
3.31%
10
2.29%
7
3.18%
47
2.41%
‘Depends’ 64
20.92%
76
23.31%
171
25.71%
117
26.77%
55
25.00%
483
24.72%
Between ‘Depends’
and ‘Prefer’
40
13.07%
33
10.12%
69
10.38%
63
14.42%
24
10.91%
229
11.72%
‘Prefer’ 107
34.97%
127
38.96%
195
29.32%
127
29.06%
63
28.64%
619
31.68%
Between ‘Prefer’ and
‘Require’
6
1.96%
7
2.15%
2
0.30%
2
0.46%
1
0.45%
18
0.92%
‘Require’ 35
11.44%
26
7.98%
50
7.52%
25
5.72%
10
4.55%
146
7.47%
Total 306
100.00%
326
100.00%
665
100.00%
437
100.00%
220
100.00%
1,954
100.00%
Each cell contains both the count of respondents and the column percentage. The P2 value for
differences across columns is 60.7 which, with 24 degrees of freedom, is significant at the 1%
level.
23
Table VIII
Ordinal Logit Estimates of Consumer Internet Search Behavior
Search
Success
Time
Searching
Time to Give
up Search
Brand
Reliance
Male 0.057 -0.219* -0.222* -0.010
Black 0.149 0.239 0.239 0.210
Married 0.144 0.019 0.031 -0.023
Divorced 0.040 0.285+ 0.044 0.088
Living Together 0.053 -0.050 0.221+ -0.148
In College -0.206 -0.203 -0.221 0.147
Some College -0.136 0.104 0.038 -0.062
College Grad -0.364* -0.077 -0.120 -0.031
Post Grad -0.464* 0.051 -0.092 -0.010
Income $20,000-$40,000 0.218+ -0.523* -0.326* -0.013
Income $40,000-$50,000 0.391* -0.623* -0.463* 0.119
Income $50,000-$75,000 0.177 -0.426* -0.203 0.003
Income Over $75,000 0.165 -0.634* -0.516* 0.050
Income Not Say 0.160 -0.426* -0.555* 0.191
Age 25-39 0.031 -0.160 -0.054 -0.133
Age 40-64 0.013 -0.094 -0.241+ 0.057
Age Over 64 -0.638* -0.023 -0.633* 0.548+
Internet Exp. 6-12 Months 0.289* -0.042 0.058 -0.057
Internet Exp. 1-3 Years 0.364* -0.026 0.407* -0.498*
Internet Exp. 3-6 Years 0.457* -0.241+ 0.381* -0.442*
Internet Exp. Over 7 Years 0.646* -0.299+ 0.355* -0.621*
Observations 2,787 2,688 2,724 1,954
Concordant 55.7% 53.1% 55.7% 55.4%
The table does not report the various intercept coefficients. Asterisks and plus signs denote
statistical significance at the 1% and 10% levels.
24
1. For a brief and fascinating look at the early of history of the Internet, read pages 3-13 of the
Internet System Handbook by Daniel Lynch and Marshall Rose. For a good (and probably the most
authentic) history of the World Wide Web, visit the birthplace of the Web—CERN—at www.w3.org.
2. See Nichols, 1998 for a particularly robust test of this theory.
3. There are obviously many non-economic reasons for creating brand equity. A recent paper argues
that brands are instrumental in creating personality-specific relationships between a firm and its customers
(Fournier, 1998).
4. We limited our sample to those over 17 years of age because we expected more respondent error
among those who claim to be 17 or under. Had we included these respondents, 1,756 of the 3,144
shopping survey respondents would be among the 2,107 individuals who answered the Internet commerce
survey.
5. Only 311 out of a sample of 2,072 selected more than one category for this question.
6. But see Lynch and Ariely, 1998 for the view that increased availability of price information may
not make demand more price sensitive if information about quality is more important.
ENDNOTES