INTRODUCTION
Internet penetration among adult population in the US, people 18 years of age and above, is estimated at 73% or 147 million people (Madden, 2006). While rural adults lag behind urban and suburban adults in Internet usage as much as 8% (Horrigan and Murray, 2006), there is general consensus that Internet can improve perceptions about quality of life (Madden, 2006).
Research also suggests that while rural users pursue many of the same online activities as their urban / suburban counterparts, they are less likely to engage in transactions (Bell, Reddy & Rainie, 2004). One reason could be that non-metro residents lack high-speed Internet facilities at home. In fact, Horrigan (2006) highlights a 15% gap in home broadband connection between rural and other residents.
Another explanation for less use of the Internet by rural residents could be that they, being older, are more loyal to local land-based businesses with which they are more familiar and comfortable in transactions. In these cases, rural residents may buy essential, day-to-day products locally from land-based businesses and purchase hard to access products such as software, books, and plane tickets on-line.
The literature generally lacks an analysis, rather than description, of Internet purchases by non-metro residents. This study contributes to the literature by addressing the following questions:
i) Can the Internet purchasing behavior of non-metro residents be explained by type of good or service selected?
ii) Is quality of life in non-metro communities influenced by access to high-speed Internet at home and, if so, how?
THEORETICAL FOUNDATIONS AND RESEARCH HYPOTHESES
Purchase Behavior
It is a well-established fact that simple descriptive matrices, such as the Boston's Consulting Group's growth share matrix, are often used by business managers to analyze or predict business units' performance. Witness the popularity of "grid" analyses and prescriptions in publications such as the Harvard Business Review that are often read by business managers. One such grid analysis that summarizes four prominent theories of consumer behavior: utility maximizing economic theory, affective or emotions-laden behavioral theory, social theory, and stimulus-response psychology, is the Foote, Cone & Belding (FCB) matrix (Vaughn, 1986; Figure 1). The FCB matrix contends that purchase behavior varies for high versus low involvement products and those which require predominantly thinking or feeling judgment.
Involvement is defined as a consumer's perceived riskiness of the purchase; a low risk purchase is categorized as low involvement and a "risky" purchase as high involvement (Rossiter, Percy & Donovan, 1991). The "think" and "feel" dimensions relate to motives for purchasing. Specifically, "thinking" products tend to solve or remove a problem; for example, Aspirin removes aches and pains. By contrast, "feeling goods" such as food and sweets stimulate and satisfy one's taste sensations and address sensory enjoyment needs.
While space does not permit an explanation of all four theories included in FCB, the grid has been validated in 24 different nations, with general agreement among marketers that products such as shown in Figure 1, on average, represent each of the four cells.
How can the FCB matrix explain and predict Internet purchase behavior? The Internet can provide both iconic (visual; relevant for "feeling" products; mainly used to elicit emotions) and echoic (oral; relevant for "thinking" products) learning in consumers. However personal income, or financial resources, strongly affects how much non-metro residents buy on the Internet. On this topic, it is well-documented that rural Americans, on average, have lower incomes than people living in metro areas (Horrigan & Murray 2006).
Since financial resources often determine, or limit, consumption of "feeling" goods, we hypothesize that:
H1: Non-metro residents will purchase more "thinking" goods than "feeling" goods on-line.
From the FCB matrix, these are likely to be goods that do not require close inspection, touch, or handling.
[FIGURE 1 OMITTED]
By the same token, since "risky" or high involvement purchases tend to be expensive purchases, we posit that:
H2: Non-metro residents will purchase more low involvement than high involvement products over the Internet.
Having highlighted the relevance of FCB grid in explaining Internet purchases, we now turn to delineating the relationship between high-speed Internet connection and quality of life.
Quality of Life
Quality of life (QOL) is a higher-order construct that encompasses concepts such as work satisfaction, and family-life satisfaction (McCrea et al, 2006). This paper examines residents' perceptions about satisfaction with various facets of their community: community quality of life. Briefly, a community provides public necessities such as law enforcement, water, sewage, and other services. High-speed Internet is part of these "essential" services. In the next section, we use this assumption to theorize the effects of high-speed Internet connection on residents' community quality of life perceptions.
High-speed Internet connection at homes is expected to reach 50% this year (Horrigan, 2007) after only nine years to achieve this market-penetration rate. In contrast, it took 15 years for cell phones to achieve the 50% market penetration. Given this rapid diffusion of home broadband technology, it is reasonable for rural Americans to expect their community to have broadband connection. However, as aptly observed by Bell, Reddy & Rainie (2004), there is an economic disincentive for building broadband infrastructure in rural areas; there are not enough paying users to overcome the initial investment. Hence, it is likely that many rural communities are not broadband enabled: a recent survey suggests that 27% of non-metro regions have no broadband access (Pew Internet Project, 2004).
According to cognitive dissonance theory (Festinger, 1957), an unconfirmed expectancy; in our case, the lack of high-speed Internet connection in the community, would create a state of "psychological discomfort" for the non-metro resident. This is because the reality contradicts the resident's expectation. In this situation, the resident stimulated to reduce the psychological tension lower her perceptions about quality of life in the community. In fact, research suggests that the resident would exaggerate or magnify the disparity (Hovland et al, 1957). This line of reasoning leads to the following prediction:
H3: Availability of high speed Internet connection at home positively affects quality of life perceptions.
METHODOLOGY
Data from the Illinois Rural Life Panel 2000 and the Illinois Rural Life Poll 2005 are used to address the research questions. The rural-life surveys, conducted by the Illinois Institute for Rural Affairs, target non-metro residents and include questions about quality of life issues, public services, and Internet usage behavior. The 2000 survey, mailed to 4,877 residents, had a response rate of 31% or 1510 responses. The 2005 mail survey targeted 2000 residents and had a 32% response rate; had 640 usable responses.
Responses to the questions given in Table 1 are analyzed to profile the Internet purchases of non-metro residents and to examine hypotheses 1 and 2. The "Table SALT" methodology (Schenker, et. al. 2007) which combines frequency table (Table) with the stem-and-leaf plot (stem-and-leaf type or SALT) was used. Since responses are simply counts, statistical tests concerning proportions were used to test hypotheses 1 and 2.
The relationship between high-speed Internet connection and quality of life in the community is assessed using an index constructed as a measure of high speed Internet connection. Specifically, the 2005 Poll asked respondents to specify the type of Internet connections in their home. The response categories were: dial-up, wireless, DSL, and cable. Table 2, columns 2 and 3, list the data download and upload speed of these connections. Based on this information, we assigned scale values for the connections (Table 2, column 4).
The QOL measure is a perceptual measure from the 2005 Illinois Rural Life Poll.
Question Response Categories During the past Much Worse Stayed the Better Much five years has Worse Same Better the quality of (1) (2) (3) (4) (5) life in your community become
The hypothesis, "the higher the speed of Internet connection, the higher the quality of life", was empirically examined using an exponential model of the form:
QOL = [a.sub.0] [1-[e.sup.[-a.sub.1]x]
Where, QOL = Quality of life in the community; [a.sub.0] = Upper limit for QOL, and x = Speed of Internet connection
This model formulation relaxes the "constant returns to scale" assumption of the linear model (Figure 2). Specifically, the model has regions of increasing, decreasing, and constant returns to scale. In addition, it captures the threshold and the saturation effects that many theorists believe exist in predictors of QOL (Bognar, 2005). To elaborate, consider a person with a low-speed dial-up Internet connection (28kbps). It is likely that she will find a gradual increase in Internet connection speed, to, say, up to 200kbps level, increasingly satisfactory (increasing returns to "x"). Now, the same person may feel equally satisfied with connection speeds in the range of 200kbps to 768kbps (constant returns to scale). Speeds above 768kbps to 2mbps may produce decreasing satisfaction (decreasing returns to "x"). It is also reasonable to assume that at some point in time, increases in Internet speed will not produce any satisfaction (saturation effect). The threshold for QOL highlights the fact Internet adds to an already existing, base level of QOL.
[FIGURE 2 OMITTED]
Since [a.sub.0] = 5 in the QOL measure, the model reduces to:
QOL * = [[a.sub.0]-QOL]/[a.sub.0]= [1-[e.sup.[-a.sub.1]x]
which can be expressed as:
ln QOL * = [a.sub.1]x
The model was calibrated using least-squares procedures.
RESULTS
Figure 3 provides frequencies with tallies of the 2000 and 2005 purchases, replacing counts. The use of tallies not only facilitates visual processing of information, but also provides information about the "year of activity"; note that the tally symbols "0" represent year 2000 and "5" denote 2005.
The Table SALT display (Figure 3) reveals that books were the most often purchased product, followed by airline tickets, and clothes. While purchases of electronics, and software decreased in 2005, purchases of games and music CDs increased.
[FIGURE 3 OMITTED]
Do the findings support hypotheses 1 and 2 in Figure 3? H1 predicts more "thinking" product purchases. Of the 1169 purchases during 2000 and 2005, a majority involves "thinking" type products: 72% or 841 of the total purchases were of the "problem-solving" type. These purchases are not chance happenings. The binomial test validates this conclusion (b(n;.5, x=841) results in z = 14.98, significant at the p<.05 level). Therefore, since the Internet has been used predominantly to purchase products that included medicines and software, we conclude that H1 is supported by the "counts" data.
Hypothesis 2 (H2) predicts more low involvement purchases but Figure 3 suggests the opposite. In fact, 55% or 643 of the 1169 purchases were of high involvement type-- products such as electronics, and clothing. Considering 51% as a majority number, then b(n;.51, 526) results in z = -3.84 (p<.05). This suggests that more high involvement product purchases have occurred during the two time periods rejecting H2.
Hypothesis 3 predicts higher quality of life perceptions for people with high-speed, home internet connections. The parameter a1 which highlights the relationship between the variables was statistically insignificant (Table 3). However, measurement errors could be involved. Specifically, "speed of Internet connection at home" is measured using an index number; an indirect approach to measurement. A perceptual measure could have returned a different result. For example, a direct questioning approach that requests respondents to state beliefs about the speed of home Internet connection could have resulted in a stronger relationship between the predictor and the criterion. Another reason for the insignificance could be that mediating variables such as "income" and "education" could have masked the relationship between quality of life and high speed Internet connection. Models were run including income, age, and other socioeconomic characteristics but without significant results. Thus, whatever the reason, further theorizing is needed to delineate the relationship.
DISCUSSION AND IMPLICATIONS
Implications for E-Business Explanations about e-commerce focus on two concepts: technology, and economics (Rosenbloom, 2003). Technology, especially information technology, is considered the "transformation" force for commerce. Virtually any transaction from ordering groceries to reading a book could be completed over the Net. But, as aptly observed by Morrison (2007), few people own an electronic book and not many consumers order groceries online (E-book, introduced in the year 2000, was expected to achieve at least $251 million in sales. Today, the market for e-book is so small that Forrester Research; a firm that tracks Internet sales, doesn't even track the category.)
"If technology was not a salient determinant of the e-commerce revolution, what else other than sheer economics (read "cost-savings") to explain its growth", touted analysts such as Carr (2000). The belief is that e-commerce would cut the need for physical infrastructures such as retail stores and producers could deal with consumers directly over the cyberspace thus bypassing expensive wholesalers or middle establishments. The expected "disintermediation" did not happen, however; instead, "infomediaries" such as Yahoo and eBay emerged to connect buyers and sellers.
Thus, we contend that it is not the technology and /or cost reductions that determine the competitiveness of e-business, but it is consumer behavior that is the salient predictor of e-commerce success and growth. Put simply, a demand side approach to e-commerce which requires a thorough understanding of online consumer behavior would provide competitive advantage to e-businesses (Korgaonkar & O'Leary, 2006).
Examining Internet purchases using the FCB grid was used as a start to understanding consumer behavior using the Internet. The results of the previous analyses suggest that products which solve a problem, "thinking" products in FCB terminology, are predominantly bought on-line. So on-line businesses dealing with consumers, commonly referred to as b2c business, then, ceteris paribus, can generate more sales by offering "thinking" products than "feeling" products. Note that this prescription is in line with the management principle of "selective concentration" which suggests that firms not try to be everything to everyone.
An e-business, like land-based businesses, must enhance customer value to gain competitive advantage. Essentially, value represents a trade-off of salient "get-and-give components" which are perceived as benefits and sacrifices respectively (Parasuraman, 1997). Research suggests that all factors, qualitative and quantitative, subjective and objective, that make up the complete shopping experience must be considered in order to understand what value means to a customer (Zeithaml, 1988). In a "distillation" of literature related to customer value, Chen and Dubinsky (2002) list the following factors as germane to enhancing customer value:
(i) Relevant information: Quick and easy access to useful information is one of the attractiveness or benefits of the Internet. However, customers may not want too much information. To make the on-line shopping a pleasant experience, e-businesses should offer pre-screened alternatives geared to customer need. A case in point is dress shirts sold by the online retailer Paul Fredrick (see www.PaulFredrick.com). This web site offers six collar choices, button or French cuff, and various sizes from which to choose.
(ii) Ease-of-use of the web site: E-business should provide consumers with a web site that makes users perceive a sense of control over the interaction. In addition, the web site should gain the interest of the customer thus making the shopping experience favorable.
(iii) Customer Service: Evidence exists to show that Internet shoppers have positive feelings about Web assistants who assist customers in Web shopping. A simple gesture such as a "hot linked" email address for customer assistance could influence Internet users to shop on-line more frequently.
Implications for Non-Metro, Small and Medium Brick-and-Mortar Retailers
While the implications for e-businesses to sell certain types of products are supported by previous analyses, less clear are approaches that can be used by small and medium-size businesses in small rural communities. While the advent of high-speed Internet is often welcomed as the future of a rural community, it may also mean a potential loss of business in some main street businesses selling traditional products. Loss of sales to e-businesses such as mail-order companies plus competition from regional shopping centers can mean store closures in these communities which are of special concern to elderly populations who are less mobile and probably less familiar with purchasing on the Internet.
So, what are some options for small main street stores? First, simply obtaining a web site without an increase in merchandise may provide more exposure but not have much effect on sales because the small store can not compete with large companies who offer more options as well as lower prices. They may also pay shipping on initial purchase or returns which reduces the delivered prices even more. Second, E-businesses, especially mail-order companies, may be open 24 hours per day compared with a traditional eight hour day in small retail stores.
One approach that seems to have worked is for stores to take on specialty lines such as running shoes and accessories. These stores offer specially designed merchandise which appeals to a special group of customers. The merchandise is not available in land-based stores so the competition is not intense. The specialty store may not even carry the merchandise in inventory; rather it orders the merchandise when contacted by a customer. These customers may also buy accessories that are not readily available in land-based businesses because the volume is too small.
A second approach is for a store in a rural area to work with local providers of unique products but who, by themselves, are unable or unwilling to market over the Internet. For instance, craftspeople may produce unique items that have a market elsewhere. By working with several products, and marketing collectively, it may be possible for the consortium or cluster to be profitable. These could involve a wide range of products rather than those identified earlier because the purchases are likely to live in metro, rather than rural, areas. In fact, rural residents may find similar products in land-based businesses. Thus, this cluster approach is essentially marketing a sense of "rural".
Third, there is a role for public agencies such as cities and/or county governments to provide an infrastructure that allows local establishments to gain exposure and markets through a collaborative approach. In Illinois, for instance, several counties with assistance from a Rural Community Development Initiative project funded by USDA-Rural Development created a community web page that was then expanded to include private businesses. The businesses received funds to build their capacity to manage their own web page and to make their pages consistent with the overall community page. This approach provides excellent exposure for these businesses and allows them to capture economies of scale in web page design and management that otherwise would not be available.
Fourth, small rural stores may find it useful to take a lesson from the ACE Hardware model and pool their funds to purchase merchandise that is stored and distributed from a central warehouse. Individual stores can manage a web page that displays a wide assortment of merchandise and they could price according to local conditions. Customers could purchase the item on-line with the local business specializing in providing service for the product.
CONCLUSION
E-commerce offers few or no barriers to entry which has intensified competition in the business-to-consumer (b2c) sector. Since consumers have multiple choices for Internet purchases, it is essential that e-business owners understand consumer behavior and the types of goods or services likely to be made on the Internet. Likewise, the Internet is a major competitor for land-based businesses, especially those with a limited selection and higher prices. Small land-based businesses in rural areas must recognize the competition and devise innovative ways to market their products. Simply adding a web page is unlikely to be sufficient to successfully market their products. They may also have to identify new merchandise, make arrangements to expand the selection, and/or otherwise capture additional markets. This research is a first step in that direction.
REFERENCES
Bell, P., P. Reddy, & L. Rainie (2004). Rural areas and the internet. Washington, D.C.: Pew Internet & American Life project.
Bognar, G. (2005). The concept of quality of life. Social Theory and Practice, 31(4), 561-580.
Carr, N. G. (2000). Hypermediation: Commerce and clickstream. Harvard Business Review, 78, 46-47.
Chen, Z. & A. J. Dubinsky (2003). A conceptual model of perceived customer value in E-commerce: A preliminary investigation. Psychology & Marketing,
Festinger, L. (1957). A theory of cognitive dissonance. New York: Harper & Row.
Horrigan, J. B. (2007). US lags behind: Why it will be hard to close the broadband divide. Retrieved (February 16, 2008) from www.pewinternet.org.
Horrigan, J. & K. Murray (2006). Rural broadband internet use. Washington, D.C.: Pew Internet & American Life project.
Hovland, C. I., O. J. Harvey & M. Sherif (1957). Assimilation and contrast effects in reactions to communication and attitude change. Journal of Abnormal and Social Psychology, 55, 244-252.
Korgaonkar, P., & B. O'Leary (2006). Management, market, and financial factors separating winners and losers in ebusiness. Journal of Computer-Mediated Communication, 11, 1128-1149.
Madden, M. (2006). Internet penetration and impact. Washington, D.C.: Pew Internet & American Life project.
McCrea, R., T. Shy & R. Stimson (2006). What is the strength of the link between objective and subjective indicators of urban quality of life? Applied Research in Quality of Life, 1, 79-96.
Morrison, C. (2007, October). Taking a page out of the e-book. Business 2.0., 34.
Parasuraman, A. (1997). Reflections on gaining competitive advantage through customer value. Journal of the Academy of Marketing Science, 25, 154-161.
Rosenbloom, B. (2003). Behavioral dimensions of e-commerce: Augmenting technology and economics. Psychology & Marketing, 20(2), 93-98.
Rossiter, J. R., L. Percy & R. J. Donovan (1991). A better advertising planning grid. Journal of Advertising Research, October/November, 11-21.
Schenker, N., K. L. Monti, G. W. Cobb, R. S. Fesco & J. S. Chmiel (2007). Combining features of a frequency table and stem-and-leaf plot to summarize the American Statistical Association's strategic activities. The American Statistician, 61(3), 245-247.
Tukey, J. (1977). Exploratory data analysis. Reading, Mass: Addison-Wesley.
Vaughn, R. (1986). How advertising works: A planning model revisited. Journal of Advertising Research, February, 57-66.
Zeithaml, V. A. (1988). Consumer perceptions of price, quality, and value: A means-end model and synthesis of evidence. Journal of Marketing, 52, 2-22.
Adee Athiyaman, Western Illinois University
Table 1: Measures of Internet Purchases: Questions from Illinois Rural Life Panel - 2000, and Illinois Rural Life Poll - 2005 The question read, "Please indicate which of the following items you have purchased in the last six months using the Internet". Item Check All That Apply Airline tickets Books Clothes Computer / Electronic equipment Software Drugs/Pharmaceuticals/Medicine Games/Music CD's Food items Services (Licenses) Auctions (eBay) Table 2: Measure for High Speed Internet Connection Type of Internet Download Speed Upload Speed Scale Connection Value Dial-Up 56kbps 28 to 30kbps 1 Wireless More than 200kbps At least 200kbps 2 DSL 768kbps to 6mbps 128 to 768kbps 3 Cable 4mbps to 15mbps 384kbps to 1.5mbps 4 Note: Information about connection speed was obtained from: www.high-speed-internet-access-guide.com Table 3: Empirical Estimates: The Exponential Model Relating Quality of Life With Speed of Internet Connection at Home Un-standardized Coefficients Standardized B Std. Error Coefficients Beta Intercept .885 .029 Speed of Internet .010 .016 .035 t Sig. B Std. Error Intercept 30.092 .000 Speed of Internet .648 .518
