Evidence and impact of consumer human capital in e-commerce transactions
Consumer purchasing processes are undergoing a fundamental change due to the dramatic growth in the utilization of the Internet as resource in making purchasing decisions. Yet at this stage, the picture of online information sharing structures remains murky. An understanding of investment and ownership patterns in “human consumption capital” developments is of great utility to marketing practitioners and policymakers who must respond to a rapidly changing marketplace. We study these developments through a survey questionnaire administered to 200 students and conclude that, although significant differences exist among online consumers, there is a high individual payoff for information shanng. Consequently, we can expect investments in organizations and structures that facilitate this sharing to increase and marketers to pay increasing attention to these developments in the future.
For marketers, the fundamental way that consumers make purchasing decisions is of great interest. The advancement of information technologies (IT), such as the Internet, has brought the marketplace to the doorstep of the average consumer. Aside from shrinking the physical distance between the consumer and the marketer, the Internet and new IT are bridging a fundamental gap that has existed for ages. This gap is the differential between the knowledge or human capital required for consumption decisions, and the knowledge possessed by the marketer. There are symptoms of a fundamental transformation everywhere. For the first time in the history of marketing, consumers now can coordinate and share consumption information on a large scale that is sure to alter consumption decision processes in a fundamental way. Additionally, the traditional skills that have made a consumer a “smart shopper” are no longer as effective as they were in the past, and these skills are increasingly being supplanted by newer skills needed to adopt Internet and IT in routine consumption decisions.
The acceleration to such a wholesale transformation of consumer skills is boosted in many ways, and ironically, simultaneously choked, by the proliferation of new products and the rapid advancement of technology. Cutting edge technology that has made the blinking VCRs an anachronism could not have been possible without a corresponding overhaul of consumption skills that makes such technologies a working reality. However, newer technologies have also compounded the information processing load of consumers. Rapid advances in technology cause newly acquired knowledge and skills to become obsolete as fast as they are acquired. Consequently, many consumers often play a catch-up game to keep abreast of the latest products. In this context, the consumer knowledge and information processing skills will take an increasingly important place on the agenda of marketing theoreticians and practitioners. This consumer human capital, defined as the knowledge and the information processing skills required to make consumption decisions and labeled “consumption human capital”, will be an important dimension guiding consumer behavior for the near future. Our concept of human capital is closely aligned with the idea of human capital developed by G.S. Becker (Becker 1993).
In this research, we investigate the consequences of the interaction between consumption human capital and advances in IT. The research is motivated by the recognition that investments in consumption knowledge affect future consumption behavior (Ratchford, 2001). The theoretical problems that may arise if we recognize that consumer human capital investment decisions are critical, especially in the Internet mediated environment, are the following: a) Who should invest in consumption human capital? b) If one invests, for what products / services should such investments be made? c) What institutional arrangements are good for such investment purposes? and d) How do advances in IT and the Internet affect all the above?
We use a theoretical framework for analyzing consumer human capital as exposited in (Richards, 2002); however, detailed and conclusive empirical validity of the criticality of consumer human capital investment is yet to be demonstrated. The main aim of this research paper is to establish empirical support for recognizing the criticality of consumer human capital investment in an Internet mediated consumption environment. Once established, answers to the theoretical questions posed above will provide a rubric and general guidance for marketing practitioners and public policy makers.
2. BACKGROUNDS AND SIGNIFICANCE
There is now an increasing trend among consumers to search for information and purchase products and services over the Internet, and a growing realization by online marketers that consumers have become more Internet savvy. Some recent survey results support this contention, for instance:
* 68 percent of all U.S. web users shopped online in 2000, with more than three-fourth of all users projected to purchase online by 2003. (eMarketer, July 2000).
* 45 percent of U.S. consumers who intend to buy a car carry out research on the Internet. (Diameter, 2001).
* 93 percent of consumers surveyed have researched products online. (Yankelovich, 2001).
The increasing use of the Internet for consumption decisions is also accompanied by changes in the established institutional arrangements for facilitating communication and information dissemination. For example, consumers can form coalitions or networks for sharing and trading consumption-related expertise. Many online information networks and discussion groups exist for this purpose. For potential consumers, formation of such coalitions changes the incentives to invest in human capital by themselves. A coalition or a network enables sharing and accessing resources for consumption experience by any consumer, including novices. Consumers who have surplus consumption expertise can market their services to other members; therefore, they have greater incentive to invest compared to novice consumers needing similar human capital. For example, an experienced used-car purchaser can offer valuable service to inexperienced buyers in the market. Once we assume that a mechanism such as the Internet exists for sharing and trading consumer human capital, the task is to chart out the likely institutional structures (from the view of improving social welfare) for ownership and investment of resources required for trading among members. To the best of our knowledge, no research work has yet been published investigating how consumer human capital investments and the institutions undertaking such investments are impacted by the increased use of the Internet in consumption decisions.
3. RESEARCH METHOD
Our method is based on sample survey administered to students from a large South-Western university. The sample is a convenience sample of 200 respondents. Such a sample size is necessary for doing between-group analyses with the data. The questionnaire contained sections on a variety of topics, including the following:
1. General demographics characteristics
2. Internet Skills
2. Purchasing habits over the Internet
3. Finding product Information characteristics
Responses to the Internet skills were nominal data and therefore correspondence analysis (using the optimal scaling utility of SPSS 12.1)was used to extract the main expertise dimensions of consumers in using Internet technology. These extracted dimensions and their scores will be used as proxies for measuring consumption human capital, in effect grouping consumers into experts and novices with respect to their Internet consumption human capital. We then study the differences between experts and novices in their purchasing behavior as well as finding product information over the Internet. The differences, if observed, would help to answer the following questions:
1. Do experts process more “tacit” information compared to novices?
2. Do experts process consumption information more efficiently and effectively compared to novices?
3. Do experts purchase high human capital input goods more than novices do?
To answer these questions, the following hypothesis is necessary: consumers endowed with differing amounts of human capital would tend to perceive products and services as requiring different amounts of information processing for their consumption decisions. Even if some consumers are the same with respect to their consumption human capital, certain goods and services such as investment services, insurance services, and electronics are likely to incur complex and involved information processing because of the inherent complexity of such products. This is similar to the concept of “tacitness” (Nonaka and Takeuchi 1995, Polyani 1966) referred to in the literature; therefore, in our study, such products requiring complex information processing are labeled as “high-tacit” goods. Other products like books, groceries, and even movie tickets are relatively less intensive in information processing, and therefore they do not possess the same level of “tacitness” as an insurance policy, and, accordingly, they are labeled as “low-tacit” goods.
This research project, in addition to gathering empirical evidence regarding a fast developing and highly relevant issue for the new economy, the findings would also attempt to dispel some of the doubts lingering about the impact of the Internet in affecting traditional marketing models and theories. Most importantly, a definite conclusion that the Internet has demonstrable consequences in consumption decisions would be reassuring to many new economy entrepreneurs.
4. DESCRIPTION OF THE DATA
The surveys were conducted with a questionnaire that addressed the following aspects:
* General demographic questions, consisting of questions such as gender, income, and age.
* Level of ability to perform a variety of Internet skill-related tasks.
Descriptive statistics of the sample are given in Table 1. There were 209 total respondents. The percentage of females in the sample was 44.5% and males 55.5%. 30.9% of the respondents have indicated residence in an urban area and 64.7% in a suburban area. A small percentage (4.3%) indicated residence in a rural area. As expected in surveys done on college students, the majority of the respondents (67%) belonged to the 21-27 age group. The income distribution showed some degree of spread all over the ranges with 30.2% of the respondents indicating yearly family income of over $49,000.
In addition, the “Purchasing over the Internet” section had questions that asked the general shopping behavior of the respondents such as type the goods they usually buy, the amount of money spent on average for shopping, and the general reasons for purchasing over the internet. The “Finding Product Information” section had questions such as the amount of time spent searching for information, the kind of good information is typically sought, and frequency of searching for information. The reader may note that the questionnaire for our study consisted of sections on a variety of topics and questions that closely parallel the 10th GVU WWW survey (Georgia Tech corporation, 1998).
5.1 Skill Level of the Respondents
In the general demographics section of the questionnaire, one question was asked whether or not the respondent had performed the following activities online:
1. Ordered a product/service by filling out a form on web
2. Made a purchase online for more than $100
3. Created a web page
4. Customized a web page for yourself (e.g. MyYahoo, CNN Custom News)
5. Changed your browser’s “startup” or “home” page
6. Changed your “cookie” preferences
7. Participated in an online chat or discussion
8. Listened to a radio broadcast online
9. Made a telephone call online
10. Used nationwide online directory to find an address or telephone number
11. Downloaded music from the Internet
12. Downloaded software from the Internet
13. Used the Internet to read news
14. Sent bulk email or spam
15. Used an online auction
16. Used websites like priceline.com, etc to book airline, hotel, and rental car.
The above skill variables were frequency analyzed and the percentage of respondents indicating in affirmative to each skill type is given in Table 2.
Based on the frequency analysis shown in Table 2, it is evident that the respondents show a good degree of variance in responding to the above skill dimensions of using the Internet. The key idea of our analysis is to separate the buyers into experts and non-experts on the basis of the above variables. As a naive solution, we can take a linear combination of the above variables equally weighted so as to capture the “expert–non-expert” dimension. However, this assumption rests on the idea that all of the above variables essentially capture different orthogonal dimensions of the expertise. However, some of the skill variables might be correlated with each other; therefore, this procedure would result in giving extra weight to the dimension captured by the correlated variables. Ideally we would want to give extra weight to uncommon variables so that experts are sufficiently separated from non-experts. This is because, experts and non-experts would be equally well-versed in ‘basic’ internet skill dimensions. The only differentiator in this instance would be the uncommon variables such as “sending bulk email.” To overcome this limitation of a naive grouping, as well as to gain a clearer understanding of various expertise dimensions that may underlie among users, further analysis may be needed. We discuss this in the next section.
5.2 Initial Steps for Grouping Users into Expertise Groups
A correspondence analysis (homogeneity analysis) was conducted on the above Internet skill dimension variables using the optimal scaling utility available in SPSS 12.1 version. The eigen values corresponding to the two principal axes are 0.27 and 0.116 respectively. Thus, these two dimensions explain a total of 38.6 % of the variance in the data.
The first two dimensions, as shown in Table 3, reveal two distinct types of expertise. The first dimension weighs heavily on the Internet web tools and page-related expertise (variables such as web page creation, changed browser’s start-up page, etc.) based on observing the loadings on variables for dimension 1. The second component weighs heavily on the purchase-related expertise (variables such as ordered a product, used online auction websites, etc.). Therefore, the data suggest two basic underlying expertise dimensions among the users. It may be cautioned that these two dimensions explain only 38.6% of the sample variance, and there could also be many other important dimensions. To check for this possibility, we did the homogeneity analysis for three, four, and five dimensions. The proportion of variance extracted by these dimensions was progressively smaller and there was difficulty in interpreting additional dimensions. To keep our analysis simple, and considering this study is only a preliminary study, our analysis proceeds with two dimensions.
The object scores or loadings representing the variables shown in Table 3 can be utilized akin to factor scores, and these scores can be considered as interval scaled data. A preliminary grouping of users based on expertise dimension could be based on the number of different skill level items they have accomplished on the items described in the above table. Tentatively, we clubbed together users as experts if they had more than ten of the above skills, and the rest as novices. To confirm whether this type of grouping represent significant differences in the object scores of the two dimensions, a t-test was conducted. The t-test results are shown in Table 4.
The t-test shows significant differences between “experts” and “novices” in their two Internet skill dimensions. The results of the t-test are important from two angles. First, the naive grouping of users into “experts” and “novices” is validated by further evidence from the homogeneity analysis. Second, by classifying expertise into two distinct dimensions, it may enable further analysis into factors that my aid one dimension and not the other.
5.3 Relationship of Expertise Dimensions with Demographic Variables
It may be interesting to also identify the relationship of the demographic variables with the two expertise dimensions. The two dimensions and their object scores were regressed with the following independent variables: age, gender, family income, length of time using the Internet, and the respondent’s living area whether urban or suburban. The regression of the both the dimensions were significant (p<.001) with F values of F=13.5, d.f 5, 200; and F=4.2, d.f 5, 200 for dimension 1 and 2 respectively. The regression results are shown in Table 5.
As shown in Table 5, dimension 1, which represents the Internet related technical skills and which also explains almost 27% variance in the data, is obviously our main interest. Gender and Length of time using the Internet have significant relationship with (p<.001). For the second dimension, which is the Internet purchase related expertise, is significant with all variables except age (at alpha level of 5%).
5.4 Do Experts Process More Tacit Information?
In the purchasing information section of the questionnaire, the users were asked about how often they search for information about products and services, which they intend to buy at some point in the future. They were also asked about which products that they have researched. There were 26 product/service categories ranging from generic grocery items to complicated services like insurance and investment choices. As a first step, we can group certain type of products like grocery, books, flowers, videos, magazines, home electronics, music CDs, stock quotes, and clothing items, as belonging to “low-tacit” category whereas items like insurance services, investment choices, banking and financial services, travel arrangements, autos, and legal services can be grouped into ‘high-tacit’ category. Our intent is to see if expert users comparatively search more ‘high-tacit’ products/services than novices. To do so, we proceed by enumerating the responses of “low-tacit” and “high-tacit” products/services separately for each user. This aggregate score would then indicate the level of search for “low-tacit” and “high-tacit” product categories, respectively, for each user. A preliminary grouping considered is as follows: evidence lends credence to the hypothesis that investment in Internet-related skills is a definite plus for increased use of the Internet for consumption decisions.
A second hypothesis that may be investigated is whether the experts search for high-tacit information proportionately more than novices. To test this, we differenced the scores for high-tacit and low-tacit goods for each expert and novice respectively. This difference is in effect measuring how much more high-tacit information compared to low-tacit information is processed by the user. This difference, if found to be significantly higher for experts than for novices, would then indicate that experts process proportionately more tacit information compared to novices. However, the t-test results turned to be insignificant and therefore the second hypothesis was not supported.
5.5 Do Experts and Novices differ in their communications with Vendors and other Users?
It may be interesting to observe whether “experts” and “novices” differ in their communication with internet vendors and other consumers over the Internet. In Table 8, the variable COMMN_V measures the frequency of such communication with vendors and the variable COMMN_U measures the frequency of communication with other consumers and users over the Internet.
The t-test shows a significant difference between experts and novices for the COMMN_U variable. The observation that highly skilled users tend to communicate more with other consumers is significant because it would make it more difficult for vendors to hide their deficiencies and continue to take advantage of unsuspecting consumers.
5.6 Do Experts Buy More Complicated Goods and Services Compared to Novice?
An interesting question is to see if “experts” with their greater endowment of human capital, buy more complicated goods and services compared to novices. It is possible that experts buy more than novices both low-tacit goods as well as high-tacit goods. Now, a closely related question is that whether experts show greater likelihood of buying high-tacit goods compared to low-tacit goods. If this is true, there is no better evidence than this to show that increased internet-related human capital of consumers, not only enables consumers to buy more products over the Internet, but it also enables a shift to buying more and more complicated goods over the Internet. To analyze this, we compare the means of the number of low-tacit goods purchased (L_Tacit_Pur), the number of high-tacit goods purchased (H_Tacit_Pur), and the difference between the number of low-tacit and high-tacit goods purchased (Diff_Tacit_Pur) for both experts and novices. The results given in Table 9 and 10 are significant, answering our question in the affirmative.
5.7 Are Experts More Efficient in Finding Information?
It could be hypothesized that experts with their larger endowment of consumption human capital may be more efficient than novices in finding consumption related information. There were questions in our survey that asked the amount of time spent in searching before useful information is found, the amount of time spent before giving up the search when useful information is not found, and the proportion of occasions when the search becomes successful in finding the right product information. The responses were compared between experts and novices to identify whether experts are more efficient than novices in finding information. The test results in Table 11 show that experts and novices differ in their success in finding information, with the mean percentage of success for experts higher than novices. Accordingly, there is some evidence to show that experts are more efficient in finding information.
6. CONCLUSION AND DISCUSSION
The results of the data analysis described in previous sections demonstrate conclusively that investment in consumption human capital pays off in many ways. Perhaps the most important result is that the promise of the Internet in replacing traditional channels and the promise of selling more complicated and tacit products and services, when consumers invest sufficient quantities of human capital. The investment by consumers is multi-dimensional and marketers need to be cognizant of such characteristics when online channels are designed.
This study, of course, comes with many limitations. First, it may be argued that the results may not generalize over a broad spectrum of the population because of the student sample used in this study. However, it is also true that people in the age group 21-27 form a major number of Internet users and e-commerce adopters in this country. So in many respects, our study is, in fact, capturing the correct population of interest.
It may also be argued that the classification of users into experts and novices is quite subjective based on the count of skills. Although we have overcome this objection to a certain extent by testing the difference between experts and novices in their object scores from the homogeneity analysis, and finding evidence that this grouping has some validity, it may be true that our classification is still on an ad-hoc basis. Future studies should develop measures that separate experts and novices in a more refined way. However, this separation will not be easy given the fact that the Internet and the skills required to successfully navigate online transactions are constantly changing, and, because of this, developing more valid measures is a problem by itself. With that said, however, the evidence outlined in our study would go a long way in enabling more focused research in this fast developing area.
TABLE 1: DESCRIPTIVE STATISTICS
VARIABLES Count Column
AGE 18-21 49 23.4%
(years) 21-24 85 40.7%
25-27 55 26.3%
27-30 9 4.3%
30-33 4 1.9%
33-36 4 1.9%
Over 36 3 1.5%
SEX Male 116 55.5%
Female 93 44.5%
INCOME Cannot say 25 12.0%
Under $10,000 14 6.7%
$10,000-$19,000 33 15.9%
$19,000-$29,000 39 18.8%
$29,000-$39,000 20 9.6%
$39,000-$49,000 14 6.7%
> 49,000 63 30.2%
LIVING Urban 64 30.9%
LOCATION Suburban 134 64.7%
Rural 9 4.3%
TABLE 2: TABLE OF FREQUENCIES OF SKILL TYPES
Variables Count %
Ordered a product over Internet 148 70.8%
Made a purchase online over $100 146 69.9%
Created a webpage 80 38.3%
Customized a webpage 97 46.4%
Changed the browser’s start-up page 141 67.5%
Changed the “cookie” preferences 112 53.6%
Participated in online chat 189 90.4%
Listened to radio online 163 78.0%
Made a telephone call online 84 40.2%
Used a telephone directory online 134 64.1%
Downloaded music online 167 79.9%
Downloaded software online 156 74.6%
Used the Internet to read news 197 94.3%
Sent bulk email 40 19.1%
Used online auction websites 97 46.4%
Used auction websites like Priceline, etc 100 47.8%
TABLE 3: HOMOGENEITY ANALYSIS OF SKILL VARIABLES
Ordered a product over Internet .128 .482
Made a purchase online over $100 .161 .382
Created a webpage .350 .003
Customized a webpage .382 .092
Changed the browser’s start-up page .485 .040
Changed the “cookie” preferences .478 .086
Participated in online chat .224 .004
Listened to radio online .304 .036
Made a telephone call online .281 .001
Used a telephone directory online .448 .005
Downloaded music online .251 .083
Downloaded software online .441 .027
Used the Internet to read news .202 .024
Sent bulk email .008 .036
Used online auction websites .077 .343
Used auction websites like priceline, etc .096 .209
TABLE 4: T-TEST DIFFERENCES BETWEEN TENTATIVE GROUPINGS OF “EXPERTS”
Dependent Mean Std. Error
Variable t p value Difference Difference
Dimension 1 20.409 .000 1.61053 .07891
Dimension 2 -1.935 .054 -.25773 .13317
TABLE 5: REGRESSION OF EXPERTISE DIMENSIONS WITH DEMOGRAPHIC VARIABLES
Independent Variable Model 1: Dimension1 as
Beta t p value
Age -.022 -.365 .716
Gender .365 5.931 .000
Family Income .014 .228 .820
Length of using Internet -.290 -4.707 .000
Area of Living -.102 -1.647 .101
Independent Variable Model 2: Dimesnion2 as
Beta t p value
Age -.007 -.108 .914
Gender .163 2.399 .017
Family Income .151 2.193 .029
Length of using Internet .168 2.478 .014
Area of Living -.170 -2.489 .014
TABLE 6: GROUPING OF “LOW-TACIT” AND “HIGH-TACIT” PRODUCTS/SERVICES
Low tacit Categories High Tacit categories
1 Generic Grocery 1 Travel arrangements
2 Branded grocery 2 Home electronics
3 Concert Plays 3 Autos
4 Stock quotes 4 Investment choices
5 Clothing Shoes 5 Banking
6 Flowers 6 Insurance
7 Video/ Movies 7 Legal services
8 Music CD’s, Tapes 8 Real Estate
9 Magazines/Newspapers 9 Computer Hardware
10 Jewelry 10 Computer Software
TABLE 7: T-TEST RESULTS
Dependent for Equality
Variable of Variances
F Sig. t value
Score variances 1.456 .229 -6.14 .000
variances not -6.07 .000
Score variances .606 .437 -7.35 .000
variances not -7.28 .000
Dependent Mean Std. Error
Variable Difference Difference
Score variances -1.83158 .29832
variances not -1.83158 .30133
Score variances -1.98246 .26951
variances not -1.98246 .27197
Variables F Sig.
COMMN_V Equal variances 1.217 .271
COMMN_U Equal variances 11.598 .001
t-test for Equality of Means
Dependent Mean Std. Error
Variables t p value Difference Difference
COMMN_V Equal variances -1.108 .269 -.26605 .24005
Equal variances -1.120 .264 -.26605 .23754
COMMN_U Equal variances -3.121 .002 -.71277 .22837
Equal variances -3.035 .003 -.71277 .23481
Expertise N Mean
1:novice Std. Std. Error
Variables 2: exert Deviation Mean
Diff_Tacit_pur 1.00 114 -.1140 1.30186 .12193
2.00 95 0.4737 1.88974 .19388
L_Tacit_Pur 1.00 114 1.0351 1.23324 .11550
2.00 95 1.8211 1.70093 .17451
H_Tacit_Pur 1.00 114 0.9211 1.12214 .10510
2.00 95 2.2947 1.69400 .17380
Variables of Variances
Diff_Tacit_pur Equal variances 21.78 .000
L_Tacit_Pur Equal variances 15.63 .000
H_Tacit_Pur Equal variances 8.35 .004
t-test for Equality of Means
t p value Difference
Diff_Tacit_pur Equal variances -2.65 .009 -.58772
Equal variances -2.56 .011 -.58772
L_Tacit_Pur Equal variances -3.86 .000 -.78596
Equal variances -3.75 .000 -.78596
H_Tacit_Pur Equal variances -7.00 .000 -1.37368
Equal variances -6.76 .000 -1.37368
Variables of Variances
Minutes Equal variances
spent assumed 11.71 .001
Minutes Equal variances
spent before assumed 4.05 .046
search Equal variances
% success Equal variances
in finding assumed 1.48 .225
Variables t-test for Equality of Means
p Mean Std. Error
t value Difference Difference
Minutes Equal variances
spent assumed -.40 .684 -.110 .270
not assumed -.38 .700 -.110 .286
Minutes Equal variances
spent before assumed -1.19 .234 -.254 .213
search Equal variances
not assumed -1.15 .252 -.254 .221
% success Equal variances
in finding assumed 3.70 .000 .335 .090
not assumed 3.71 .000 .335 .090
We thank the College of Business Administration, California State University, Sacramento for providing financial support for collecting the data used in this project.
Becker, G. S. , Human Capital, Chicago: University of Chicago Press (for the National Bureau of Economic Research), 1964.
Nonaka, Ikujiro and Takeuchi, Hirotaka, The knowledqe-creating company: How Japanese companies create the dynamics of innovation, Oxford University press, Inc., 1995.
Polyani, Michael, The Tacit Dimension, London: Routledge & Kegan Paul, 1966.
Ratchford, Brian T., The Economics of Consumer Knowledge, Journal of Consumer Behavior, 27 (March), 2001, 397-411.
Richards, Joseph B., Impact of the Internet on Consumer Human Capital, Unpublished Doctoral Dissertation, Syracuse University, New York, 2002.
Joseph Richards earned his Ph.D. at Syracuse University, New York in 2001. He is currently an Assistant Professor of Marketing at California State University-Sacramento.
Laura Riolli earned her Ph.D. at University of Nebraska, Lincoln in 1998. She is currently an Associate Professor of Organizational Behavior at California State University-Sacramento.
Jordan T.L. Halgas earned her J.D. from The Ohio State University in 1994. She is currently an Assistant Professor of Organizational Behavior at California State University-Sacramento.
COPYRIGHT 2004 International Academy of Business and Economics
COPYRIGHT 2005 Gale Group