The Dimensions Of Commercial Cyberspace

William D. Wells, Qimei Chen

University of Minnesota

Abstract

In this study, the authors identify twelve dimensions of cyberspace on the Web through factor analyses of data generated by coders of web sites. A lengthy list of Web site attributes is summarized in twelve dimensions. As in previous analyses of differences among entities, overall evaluation accounts for the largest share of variance. The cognitive and affective elements of this dimension distinguish Web sites that attract from Web sites that alienate potential users. The remaining dimensions — Outreach, Expertise Requirements, Completion, Local vs. International, Power Consumption, Accessibility, Contact Options, Text vs. Pictures, Complexity, Foreign vs. Domestic Origin, and Pop-up Windows – summarize objective properties of Web sites themselves. These properties may be desirable or undesirable depending on the background, needs and purposes of the individual Web user, and upon the nature of the usage situation. All twelve factors have managerial and theoretical implications that are discussed by the authors.

Introduction

Depth interviews with Net users show that search engines omit information that could be very useful. For example:

I went to this search engine, looking for a site offering clothes that I could wear at casual times. I hoped that the result of the search would bring me to the kind of site where I could purchase casual clothes online. I typed the keywords casual clothes into the search box and submitted my request. The search engine came back with a shockingly large quantity of results. Overwhelmed by these thousands of recommendations, I casually entered a site called casualday.com which sounded like a good place to shop. Apparently, this is a site selling casual clothes with clear pictures to illustrate the products. I browsed the site a little bit, and found some items that I needed. It was not until the moment that I tried to check out some products that I realized that this was a B-to-B site, i.e. they only wholesale casual clothes to other companies in large quantities. … Why can’t the search engine tell me this is a B-to-B site in the first place? Or at least, the site itself could have told me that upfront. I wasted about ten minutes.

Obviously, if this search engine had informed this user that casualday.com was a B-to-B site, the Net experience would have been more productive.
Other respondents complained that they were led to sites where they could not find what they needed. This experience was even more frustrating:

When I was led to this site called officeguide.com, I thought that finally I might find the office information I was looking for. You know, I’m starting my own business and would like to find a possible space in the next one or two weeks. This site offers a guide to offices all over the world. I selected the relevant country and city information, hoping that I would be able to get some options of office space for rent in the district I defined. To my great disappointment, all I was shown was a form that asked for personal information and contact information, and a note telling me that I’ll be contacted within the next 30 days. I was totally frustrated. If I had a month to wait, I would be driving around to find a space myself. Why didn’t the search engine hint that I actually could not find out about the space online? I decided to use the newspaper’s classified section next time instead.

Again, an edit – immediately available information versus you will be contacted – would have assisted this Web user.

Other respondents said that search engines led them to Web sites that are under construction:

Web sites under construction are very annoying. Those are the sites that you can’t help leaving as quickly as possible. However, even if I exit the site immediately, I still have to spend about one to three minutes just to wait for my browser to go back. I did a little calculation one day. If I encounter, say 10,000 sites like that, I have spent around 30,000 minutes surfing these meaningless things. That’s like spending a month’s working hours doing nothing. I do have other responsibilities in my life. I asked myself, why have I wasted time on all these sites I have no intention to visit? I should have spent this time with my daughter and my wife, or to exercise and appreciate nature…

These impatient examples, and many others, show that current search engines omit filters that could improve the search experience. Presumably, engines that provide this valuable service would attract customers and repeat business. But what filters are needed? To what attributes should Net users be alerted? To study this issue, we approached it from the customer’s point of view. We first asked a diverse sample of Web users to describe their Web search experiences. We next asked a larger sample of Web users to identify similarities and differences among corporate Web sites. We then coded a representative sample of corporate Web sites and factor-analyzed the codes to identify a smaller, more parsimonious set of Web site differences. As the irritated quotes above suggest, these differences could make search engines more effective. Knowledge of perceived differences can also be useful to Web site builders. With millions of Web sites lingering in cyberspace (Hamm-Greenawalt 2000), all sites must compete for visitors. But compete on what basis? What differences among Web sites are most likely to be noticed by Web users? Finally, differences among Web sites are important to Web researchers. We know that Web sites differ in many ways. We also know that findings based on one kind of site might not apply to others. But what is meant by kind of site? On what dimensions do Web sites differ?

Study Design

To answer these questions, we started with a randomly generated pool of Web sites retrieved from the Comprehensive Dot Com Directory(tm) offered by Google.com. This site allies with Network Solutions (one of the largest web hosts) to provide easy access to millions of business Web addresses, and it has enriched its database through relationships with such companies as infoUSA, NewsReal, GTE Directories, Looksmart, and VeriSign. Among the 1,060,000,000 Web sites listed in this directory, we randomly selected 800 as the population for this study. The study unfolded in three phases. The object of the first phase was to obtain a preliminary but comprehensive inventory of variables along which Web sites differ. The object of the second phase was to code a large, representative sample of .com Web sites on these perceived differences. The object of the third phase was to reduce the ratings to a smaller, more general set of factors. These factors represent dimensions of cyberspace – differences among Web sites that merit early warning. Respondents for the first phase were recruited by poster. The posters described topic, length and location of the interview, offered $10 compensation and encouraged those who saw them to refer family members and friends to join the study. Fifty posters were distributed throughout a university campus and nearby neighborhood. Potential respondents who volunteered were selected so as to provide a broad range of Web users. Ten of the respondents in the depth interview phase were male, and five were female. Their ages ranged from 18 to 50. Nine were college students from different majors: finance, statistics, human resources, anthropology, computer science, psychology, history, etc. Three were Ph.D. students from philosophy, strategic management, and conservation biology. The remaining three were technicians, administrators, and sales representatives. Their ethnic backgrounds and Internet experiences also varied. While most were American, some were Asian, African, Arabian, or European. Some were veteran Net users (8-9 years); others had comparatively little experience (10 months was the shortest). All were Internet users, with surfing frequencies ranging from several times a day, to less than once a week.

Interview Procedure

During three-hour interviews, respondents were provided with Internet connections, food, drink and a natural Web surfing environment. From the 800 Web site pool, we randomly generated four triples of Web sites for each respondent, and encouraged respondents to surf each triple naturally. Respondents were asked to describe their first impressions of each site and then asked which two sites in each triple were most similar and which two were most different. Then, they were asked to describe the ways in which the similar sites were similar and the different sites were different. This procedure–subjective clustering of stimuli followed by free response labeling –has previously proved useful for obtaining similarities judgments (Kelly 1963; Green, Carmone and Fox 1969; Green, Wind, Krieger and Saatsoglou 2000). Finally, respondents were asked why they categorized the similarities and differences the way they did. All responses were tape recorded and transferred into text transcript for analysis. In the depth interviews, 15 respondents analyzed 39 triples of 117 web sites. This phase was discontinued when additional respondents produced no new discriminations. The process had reached its saturated point and new interviews were not yielding new items.

Depth Interview Findings

The depth interviews provided a preliminary inventory of perceived differences among Web sites. Respondents identified some familiar marketing factors: Target Size ( general population, specific segment or ultra-tightened niche ). Origin of the Business ( domestic versus foreign ). Type of Site ( B-to-B, B-to-C or Mixed ). They also identified design factors such as demands large system requirement versus demands little system requirement; lots of pop-up windows versus no pop-up windows; and amateurish design versus professional design. Many identified e-commerce related variables. For instance, respondents distinguished sites that permitted online ordering from sites that did not, and sites sponsored by a single company from sites sponsored by a community of companies. They considered content factors such as number of links to other sites and whether the site was multilingual or in one language only. They noticed whether the site enhanced the company image or damaged the company image, whether it required users to have lots of expertise or little expertise, and whether it is primarily oriented toward the past, present or future. A subjectively classified list of the discriminations the depth interview respondents made is shown in Figure 1.

Figure 1. Preliminary Inventory of Perceived Differences Among a Large Representative Sample of .Com Web Sites

Business:

Design:

E-commerce:

Content:

Value:

Consumer:

Theme:

Some interesting distinctions were deleted at this point because they seemed too unique to be of general value. For instance, one respondent said he would distinguish sites that revealed the business owners’ educational backgrounds from sites that did not. Another separated sites that were reached from other URLs from sites that were reached first. These discriminations seemed too idiosyncratic to be of general interest.

Coding and Factor Analysis

The 45 preliminary discriminations were placed in a coding form (Appendix A). The form also included a 6-item scale that measures overall Attitude Toward the Web Site (Chen and Wells 1999). Three coders used this form to code a new sample of Web sites generated from the 800 Web site pool. The random selection process yielded 5 pairs of duplicating sites. The inter-coder reliability on these sites was 90%. The ratings of the 150 unique Web sites were factor analyzed using an eigen value of 1.00 as the stopping criterion and principle component analysis with Oblimin rotation. This procedure yielded 14 factors. After dropping items that had very low loadings (<.40) and items that cross-loaded on more than one factor, a second analysis of the purified data produced twelve factors (Table 1) which explained 72% of the total variance. These twelve factors represent an initial outline of the dimensions along which Web sites are perceived to differ. As will be discussed in the Managerial and Theoretical Implications section of this article, these factors can guide search engines designers. They also have managerial and theoretical implications for Web site builders and Internet researchers.

Web Worthiness

The first of the 12 factors (Table 1) explains approximately one fourth of the total variance. It is a general good site-bad site factor that taps into overall evaluation – whether the layout and navigation structure of the site is clear, whether its appearance is appealing, whether it is well-organized, and whether it is considerate of the user. This factor also includes assessment of whether the site enhances the sponsoring company’s image, whether it is professionally designed, whether it is trustworthy, and whether its purpose is easily identifiable. This dimension might be called Web Worthiness, by analogy to sea worthiness. It correlates .85 with the Attitude Toward the Site scale previously shown to measure overall subjective evaluation (Chen and Wells 1999). Web Worthiness resembles the Evaluation dimension found by Osgood, Suci, and Tannenbaum (1957) in their pioneering research on the dimensions of meaning. In Osgood, Suci and Tannenbaum’s investigations, Evaluation accounted for the largest portion of the variance among entities that ranged from sonar sounds to political candidates (Brewer 1994). Evidently, Web sites are no exception.

Table 1: Dimensions of Cyberspace:

Factor #

Factors

% of Var

Items

Loadings

1

Web Worthiness

23

Navigation

.85




Layout

.83




Purpose

.81




Organization

.81




Image Enhancing

.75




Ease of Getting Lost

-.73




Considerateness

.68




Appearance (Appealing vs. Boring)

.67




Friendly

.66




Professional Design

.61




Trustworthiness

.58






2

Outreach

11

Updates

.73




Size of the Business

.71




Many References

.61




Many Links

.57




Banner Ads

.42






3

Expertise Requirements

6

Technical

.86




Expertise Required

.85






4

Completion

6

Shopping Online

.75




Not Under Construction

.50






5

Local vs. International

4

Market (local-international)

.64




Type of Site (B2B-B2C)

.63




Target size (small niche-general population)

.47






6

Power Consumption

4

High System Requirement

.86




Low Site Load Speed

.68




Interactivity

.43






7

Accessibility

4

Registration Required

.86




Pay

.82






8

Contact Options

3

Number of Options

.84






9

Text vs. Pictures

3

Graphics

.83




Information Convey via Pictures

.80






10

Complexity

3

Amount of Detail

.80




Amount of Information

.62




Number of Companies

.50




Complexity

.45






11

Foreign vs. Domestic Origin

3

Origin of the Business (foreign or domestic)

.77




Languages

.71






12

Pop-up Windows

2

Pop-up Windows

.89

Web Worthiness also resembles a general evaluative dimension often found in factor analysis of consumers’ evaluations of conventional advertisements (Wells 1964a; 1964b; Leavitt 1970; Schlinger 1979; Aaker and Bruzzone 1985; Aaker and Stayman 1990; Pashupati 1994). Academic and applied studies show that this dimension — Attitude Toward the Ad – correlates with memorability, purchase intention and other indications of effectiveness (Shimp 1981; Batra and Ray 1986; Mackenzie, Lutz, and Belch 1986; Brown and Stayman 1992). Thus, Web Worthiness can be considered, at least provisionally, an overall index of Web site effectiveness. Because Web Worthiness accounted for so much of the variation among Web sites, we factor-analyzed the items that loaded highest on it (principle component analysis with Oblimin rotation and eigen value of 1.00 as the stopping criterion). This secondary analysis identified two facets (see Table 2). The first facet includes considerate, friendly, trustworthy, appealing, professionally designed, clear purpose and helps enhance company image . It emphasizes affective responses. The second facet includes organized, clear in layout, clear navigation structure and not easy for users to get lost. It stresses the cognitive aspects of the surfing experience.

Table 2. Secondary Factor Analysis of Web Worthiness(Oblimin Rotation)

Item Facet 1 Facet 2
Considerateness .89
Appearance .88
Friendly .84
Image Enhancing .84
Professional Design .81
Trustworthiness .70
Clear Purpose .40
Ease of Getting Lost
-.99
Organization
.65
Navigation
.63
Layout
.63
Eigenvalue 7.1 1.1
% of Variance 59 9
Total % of Variance 68

Web Worthiness Illustrated

To illustrate the implications of Web Worthiness, we take a site that rated high on it. The high Web Worthiness example is a site developed by Devine & Pearson Communications, an Internet advertising agency. Let’s utilize the facets in Table 2 to understand why devine-pearson.com was so highly regarded by our raters. Readers of the present article can use the facets portrayed in Table 2 to make their own evaluations. When surfers enter the Devine & Pearson Communications site (Figure 2), the first impression is that it is eye-soothing as well as eye-catching. It uses red, gray, blue, yellow and green blocks to indicate the services its sponsor offers – advertising, design, interactive, consulting and public relations. With JavaScript, these blocks have interactive rollover effects: when the user points to a service block, the block enlarges itself a little bit. (see Stage 1 in Figure 2, the mouse is pointed to the red advertising block.) By clicking the enlarged block, users are led to the section they desire. Although JavaScript is constantly under debate, it is clean and tasteful here. In the whole depth of the site, it helps emphasize the highlights. The second picture in Figure 2 shows what happens when the Advertising section is selected. Note that in this page, other blocks turn white so that the Advertising block stands out. This site demonstrates the essence of simplicity – clear illustrations and adequate white space offer a comforting zone for imagination and focus on the most important aspects of each page. The color blocks and opening introduction clarify the purpose and mission of the site. With professional and creative design, it enhances the company image and gains trust from its visitors. Surfing devine-pearson.com continues to be pleasant. Wherever the user surfs, the header of the site remains consistent and the block color highlights the relevant section, so that users are reminded where they are and can find the way back and forth easily.

Figure 2: Example-Devine&Pearson Communications


Facet 1: Site image and Attitude–First Impression Stage

Site image and Attitude--First Impression Stage

Site image and Attitude--First Impression Stage 2

 

 

Facet 2: Navigation–First Surfing Experience Stage

Navigation--First Surfing Experience Stage

 

Go to this site

The site offers quick-loading thumbnail pictures that provide initial vivid glances. It then offers options to enlarge the pictures for closer examination. Not surprisingly, devine-pearson.com also earns high praise on Facet Two. It is organized, easy to navigate, clear in layout, and makes an effort to make sure that users will not get lost while surfing.

Thus, like Evaluation in investigations of differences among constructs and Attitude Towards the Ad in investigations of differences among traditional advertisements, Web Worthiness taps cognitive and affective attributes that are likely to attract or repel potential users.

Factors Two Through Twelve

Factor One, Web Worthiness, summarizes overall evaluation. Factors Two through Twelve focus on specific attributes. Surfers’ evaluations of these attributes are likely to depend upon usage circumstances and upon the needs and purposes of individual Web users.

Factor Two: Outreach

Factor Two distinguishes sites that offer links, references, bibliographies and frequent updates from sites that do not. As the factor loadings indicate, large businesses are more likely than medium or small business to do that. Banner ads also link to out-site resources. This factor might be called Outreach. Our interviewees mentioned that they prefer extensive Outreach when they are addressing topics that require expertise to understand. However, they prefer little or no outreach when they are just dropping in to a flower e-shop for a dozen fresh roses.

Factor Three: Expertise Requirements.

The third factor separates sites that require users to have high expertise from sites that do not, and sites that are highly technical from sites that are not highly technical. This factor is labeled Expertise Requirements . Its valence is both individual and situational. For example, one of our interviewees is a computer science major. He said I could go to a site and have no problem understanding it and surfing it, while my girlfriend (an English major) does not get it at all. Of course, sometimes I go to a site not as a computer geek, but as an ordinary consumer. In those cases, I really want the site to be down to earth — hide all the technical stuff behind the interface. Just show me a virtual store that I can shop.

Factor Four: Completion.

The forth factor embodies the degree of sophistication and completion of the Web site. It is symbolized by two criteria – how well the site facilitates shopping online and whether the site itself is finished or under construction. The depth interviews suggest that this factor influences surfing experiences a lot. Interviewees often mentioned that it is important for them to know whether they can actually use a shopping cart or a shopping bag to proceed with immediate check out from the .com site. Many also expressed frustration at being led to a site that is still under construction. Because sites that permit ordering online tend to be complete, these two items loaded relatively highly on the same factor. However, these two attributes are conceptually distinct and both are important. They may need to be kept separate in certain practical situations. Here, covariance provides a conceptual guide, not a rule to be blindly followed.

Factor Five: Local Versus International.

The fifth factor distinguishes sites that focus on local markets from sites that focus on regional, national or international markets. It also distinguishes B-to-B sites from B-to-C sites and sites that target small or medium-sized market niches from sites that target the general population. This dimension identifies marketing elements that may be of particular interest to specific users. For the user we quoted earlier, who just wanted to find some casual wear for himself, being led to a B-to-B site was certainly a major turnoff. On the other hand, a manager in search of information about a particular business may want to limit the search to B-to-B sites only, or to a certain size of market. Like online ordering and completeness, the site attributes that loaded together on Factor Four — B-to-B versus B-to-C and market size — are conceptually and operationally distinct. In some applications, to be described later in this article, they may need to be kept separate, even though they loaded on the same factor. Again, covariance provides a guide, not a requirement.

Factor Six: Power Consumption.

The sixth factor categorizes Web sites based on how much computing power the site requires and how fast it downloads. Because highly interactive sites are likely to have high system requirements, Interactivity also loads on this dimension as a trade-off to download speed and system requirements. Interactive elements based on Dynamic HTML, JavaScript, and CSS. mouse rollovers, and pull-down menus, require more time to load and place relatively heavy demands on the user’s computer. In the depth interviews, Power Consumption emerged as a situational factor. Interviewees said that when they are using high-speed LAN or T1 Internet connections (at school or at work) they are less concerned about power consumption and more attracted to sites that consume more power. However, when using a PC at home with a 28.8K or 56K modem, they prefer sites that demand less system resources.

Factor Seven: Accessibility.

The seventh factor emphasizes accessibility. It identifies sites where users have to pay or register to get what they want. Like Power Consumption, Accessibility is an individual preference dimension. Some users believe that registration forms are always annoying. These users protest that registration is even more annoying when they forget their member name or password, as often happens. On the other hand, some users believe that membership gives them a sense of community, that less accessible sites confer a sense of exclusivity, and that at least some exclusive sites provide information or entertainment that is not available more conveniently or at lower prices elsewhere.

Factor Eight: Contact Options.

The eighth factor categorizes Web sites based on the contact options the site sponsor offers –email, telephone, fax, mail, 1-800 number, and online chatting. This factor identifies sites that make it relatively easy to contact the site sponsor. This factor is important to Net surfers who need direct contact to do business or to seek redress when transactions are unsatisfactory(www.internetworldness.com). However, easy contact is not always necessary or even desirable. Ebay, the largest e-auction site, has over 4 million items for sale each day, and acts only as an intermediary between individual sellers and individual bidders. For most customers, direct human contact with an Ebay representative would not be desirable or needed. Similarly, msn.com–the portal site sponsored by Microsoft – hosts more than 200 million users per month, making it the #1 destination on the Web worldwide. But since msn.com acts only as a medium to integrate information, direct personal contact would seldom be essential.

Factor Nine: Text Versus Pictures.

The ninth factor separates sites using many graphics from sites using no graphics, and sites mainly using text from sites mainly using pictures. Our depth interviews show that this dimension is preference and situation dependent. For instance, when our interviewees talked about selection of news sites, some of them said they prefer many illustrations while others said they prefer mainly text. When they talked about looking for direction from one place to another, some said they would prefer maps while others said they would prefer words. Most agreed that when the destination is very unfamiliar, they would prefer both graphic and textual assistance. Similar contingencies have been noted in other media. For instance, studies of students’ preferences in textbook design have fond that some students place very strong emphasis on the quality of writing in a textbook, while others emphasize graphic aspects and organizational elements (Besser, Stone and Nan 1999). These preferences are also situational. Graphical methods seem particularly appropriate in describing and representing abstract structures, conveying spatial relationships, and making information given in text form more memorable, while text seems more appropriate for conveying familiar, concrete information rapidly (Williams 1993). Graphic variation is helpful in reducing boredom and inattention produced by homogeneity in text, while text is better at presenting nuanced detail (Nielsen 1990). The present findings suggest that these distinctions and contingencies are also likely to be Web design considerations.

Factor Ten: Complexity.

The tenth factor includes items that separate sites with lots of information, lots of detail and much complexity from sites that seem relatively simple to the user. This dimension is frequently mentioned in advertising research (George 1998), system design (Lawler and Elliot 1996), public affairs (Al-Menayes and Sun 1993), social psychology (Satish and Streufert 1997), behavioral science (Goswami 1998, Sweller 1998), information technology (Hurt and Hibbard 1989; Jacobson 1992; Banker, Davis and Slaughter 1988), human resources (Meyer, Shinar, Leiser 1997), and advertising (Morrison and Dainoff 1972; Anderson and Jolson 1980, Chamblee, Gilmore, Thomas and Soldow 1993). Because consumers are limited information processors (De Heer 1999), there comes a conflict between limited cognitive capacity (Miller 1956) and information intensity. This conflict differs from person to person. Optimal complexity is also situational. Some researchers (Streufert, Suedfeld and Driver 1965; Sieber and Lanzetta 1964) have found a curvilinear relationship between perceived environmental complexity and information search, with the latter declining after a certain level of the former. Others (Stiles 1974, Lussier and Olshavsky 1974) found no evidence of a decline in search as the environment became more complicated. Bettman (1979) suggested that time can be a crucial mediator. A limited amount of time to make a choice forces decreases in search and level of processing. However, if the searcher can devote as much time as desired, the relationship continues to be strictly positive. Other studies show that processing capacity and effort depend on task difficulty, task purpose, and level of involvement (Kahneman 1973; see also Hahn, Lawson, and Lee 1992; Stiff 1986). Similar contingencies were evident in the depth interviews. A surfer who is visiting a site to check today’s weather may not demand or even want much detail. If this same person is visiting a site to plan a vacation, he or she may want an extended forecast, discussion of temperature and precipitation ranges, and detailed accounts of seasonal variation. Similarly, preferred level of complexity may depend on financial risk or personal involvement. An e-shopper who wants complicated specification in a laptop purchase may want simple specification when purchasing minor grocery or drug store items.

Factor Eleven: Foreign Versus Domestic Origin.

The eleventh factor distinguishes foreign Web sites from domestic Web sites and single language sites from multi-language sites. These variables go together because Web sites sponsored by foreign companies are more likely to offer bilingual or multi-lingual versions. This dimension can help users whose native language is not English to identify sites they can understand. It can help business managers identify suppliers or potential customers who will follow familiar laws and customs, and to select business situations that will be relatively free from unexpected complications.

Factor Twelve: Pop-up Windows.

The twelfth factor separates sites that have many pop-up windows from sites that have few or none. Some Web users believe that pop-up windows are always annoying, especially windows that display advertisements (Star Tribune May 8, 1999). Other users say that pop-up windows help them find information that would be hard to locate if the windows had not been available. Some site designers object to pop-up windows in general, especially daughter windows that do not give customers a back button (Rose 2000). Others believe that pop-up windows provide flexibility that surfers can find helpful (Goodrich 2000).

Summary of the Twelve Factors

The lengthy list of attributes in Figure 1 can be summarized in twelve dimensions. As in previous analyses of differences among entities, overall evaluation accounts for the largest share of variance. The cognitive and affective elements of this dimension distinguish Web sites that attract from Web sites that alienate potential users. The remaining dimensions — Outreach, Expertise Requirements, Completion, Local vs. International, Power Consumption, Accessibility, Contact Options, Text vs. Pictures, Complexity, Foreign vs. Domestic Origin, and Pop-up Windows – summarize objective properties of Web sites themselves. These properties may be desirable or undesirable depending on the background, needs and purposes of the individual Web user, and upon the nature of the usage situation. All twelve factors have managerial and theoretical implications.

Managerial and Theoretical Implications

The twelve dimensions identified here are distinctions Web navigators use intuitively to discriminate among Web sites. They are important to search engine designers because they show how to improve the search experience. They are important to site designers because they set the terms of competition. They are important to Web researchers because they identify key independent, intervening and dependent variables, and because they sketch the limits of acceptable generalization.

Managerial and Theoretical Implications of Web Worthiness

The first dimension, Web Worthiness, accounts for the largest share of variation among Web sites. It is managerially important because it determines whether a site deserves additional attention. A low rating on that dimension says that the site is not user-friendly. Using the crude but informative rating system shown in Figure 3, search engine designers can calculate Web Worthiness ratings for sites registered with them. Like restaurant, hotel movie and TV program ratings, these ratings identify choices that customers would find most worth making.

Figure 3: Web Worthiness Rating Form

The following items assess your general impression toward the web site you just visited. Circle the number that best indicates your agreement or disagreement with each statement. 

 

Definitely Disagree                  Definitely Agree

This web-site is considerate

1…………2…………3…………4…………5

This web-site is appealing

1…………2…………3…………4…………5

This web-site is open and friendly

1…………2…………3…………4…………5

This web site enhances the company’s image

1…………2…………3…………4…………5

The design of this web-site is professional

1…………2…………3…………4…………5

This web-site is trustworthy

1…………2…………3…………4…………5

The purpose of this web-site is clear 

1…………2…………3…………4…………5

It is easy to get lost in this web-site

1…………2…………3…………4…………5

This web-site is well-organized

1…………2…………3…………4…………5

The navigation structure of this site is easy to use

1…………2…………3…………4…………5

This web-site has clear layout 

1…………2…………3…………4…………5

Web Worthiness ratings would benefit the Web community in at least two ways. First, they would reduce frustration resulting from unsuccessful trips to poorly-designed sites. Second, by applying peer pressure, they would improve site design in general. Subject to this criticism, site designers would pay attention to site attributes that most matter to their audiences. The scale shown in Figure 3 can be of equal value to site designers and their sponsors. Because this scale evaluates attitude and image, organization and navigation, sites that score low on it are unlikely to provide high returns on their designers’ or their users’ investments. Finally, because Web Worthiness accounts for such a large share of the variation among Web sites, it can serve as a matching or mediating variable in research where Web sites are the units of observation. Findings based on sites that score high on Web Worthiness are unlikely to generalize validly to sites that score low on Web Worthiness, and findings based on samples of sites that differ significantly in Web Worthiness are unlikely to be otherwise comparable. Thus, the scale shown in Figure 3 can serve as an effective matcher in experiments, and as an efficient stratifier in selection of Web site samples. Because Web Worthiness resembles Attitude Toward the Ad, it seems likely to correlate with memorability, purchase intention and other indications of Web site effectiveness. It may therefore serve as a useful dependent variable. What properties of Web sites improve Web Worthiness? How do Web Worthiness scores change across topics, message types or audiences? Answers to these questions frame principles of Web communication.

Managerial and Theoretical Implications of Dimensions Two through Twelve

While the valence of Web Worthiness is unidirectional, the valences of dimensions Two through Twelve depend upon the requirements of the usage situation and the background, needs and preferences of individual users. Thus, upfront information on the attributes tapped by dimensions Two through Twelve is likely to make a Web search more efficient. The closest any current search engine comes to offering this option is the filtering feature offered by HotBot, which offers sorts by media type, page depth and pornographic content. The present analysis demonstrates that this small set of filtering categories falls far short of covering Web users’ key distinctions. The service advocated here resembles the current personalization or customization services offered by Yahoo and MSN. In Yahoo, My Yahoo allows users to select and organize topics that are most relevant to their interests. In MSN, One-Click personalization allows users to add Wall Street Journal Business News, Advice for Women, or Gaming Tips to their own MSN pages. In implementing this service, first time users would be asked to select the dimensions that are important to them. The next time the user returns to the search engine, only the selected dimensions would be shown. If the user wants to reselect a dimension that has been suppressed, a click would re-select it and the portfolio of dimensions would be updated. To further illustrate this strategy, we present a pseudo search engine with a selection crated by a typical user (see Figure 4). By offering site choices screened according to individually selected dimensions, a search engine could identify its user’s most satisfactory sites and suppress all others.

Figure 4: Pseudo Search Engine

Pseudo Search Engine

 

Managerial Implications for Web Advertisers. The screener advocated here would provide a value-added directory that would allow Net advertisers to ensure attention from the customers they most want to reach. As our depth interview respondents and other critics (e.g.www.internetworldnews.com) have noted, current search engines do not do that adequately. Sites that pass the personalized editing advocated here would be most likely to be opened and searched by their specific target audiences.

Managerial Implications for New Media Planners. The dimensions described here have managerial implications for new media planners. Before placing an advertisement on a Web site, the planner could ask whether the profile of the target audience fits the site’s purpose. For instance, if an advertisement is intended to attract consumers globally, the planner could use the site’s score on Factor 5 to make sure that the Web site itself is suitable for an international market. Similarly, the site’s score on Factor 11 would show whether it offers multi-language versions. Instead of being scattered across Web sites, advertisements would be concentrated on sites where surfers with relevant interests, purposes, and computing equipment would be most likely to select and explore them.

Theoretical Implications for Net Researchers. The twelve dimensions identified here represent at least a first rough cut at defining and describing the most noticeable differences among Web sites. They therefore represent potential independent variables, potential mediating or moderating variables and (in most cases) potential dependent variables in research where Web sites are the units of observation. Web Worthiness provides a prime example. What are the consequences of high Web Worthiness? Does it lead (as seems likely) to satisfaction with the surfing experience? Does it lead to longer stays, more thorough exploration of the site, more repeat visits? Does it lead to more favorable attitude toward the site sponsor? Does it lead to other forms of contact, such as ordering online, a request for more information, a request for a sales call or a visit to a brick-and-mortar retail center? All these questions seem answerable, and all seem worth thorough investigation.

Dimension two — Outreach — implies a similarly intensive set of research questions. On what topics, for what purposes, and among what subsets of Web users and/or Web usage situations, is high Outreach desirable? Among whom and under what circumstances is minimal Outreach better? Similar questions about Expertise Requirements, Power Consumption, Accessibility, Complexity and other dimensions seem likely to be equally productive.

The Web is so vast and Web sites differ in so many ways that the number of possible comparisons among Web sites is now virtually infinite. The twelve dimensions outlined here provide a first draft summary of the dimensions that Web users are most likely to find consequential.

Design Implications for Net Researchers. At least some of the dimensions identified here seem likely to play important roles as mediating or moderating variables. For example, Burns (2000) found that congruity between the perceived personality of a Web site and user’s personality is related to the user’s evaluation of the site (Burns 2000). But this investigation used only two Web sites. Is its outcome also true of others? How much more informative this study would have been if it had employed a larger sample of Web sites stratified along the dimensions described here. Similarly, Eighmey (2000) and others (Stafford and Stafford 2000) have applied uses and gratifications theories in studies of Web-based communications. So far, this work has been limited to small, unsystematically selected samples of Web sites, and to small, unsystematically selected samples of respondents. Before we can be sure that their findings apply to other kinds of Web sites, to other kinds of respondents, or to other site-respondent combinations, we will need more systematic selection of Web sites and of respondents along sets of relevant dimensions.

We already know that in e-commerce research, findings which hold true for one kind of Web site do not necessarily hold true for others, and that Web users are not all alike. The dimensions described above identify the intuitive categorization schema Web users employ. They provide a first cut at identifying the limits of generalization. Without partialing these differences out, or otherwise taking them into consideration, results based on one selection of Web sites and one selection of Web users cannot be applied with certainty to others.

Suggestions for Future Research

In proposing this preliminary set of twelve dimensions, we are fully aware that it represents a rough map at best. Like the crude charts of New World explorers, this map no doubt exaggerates some features and omits or underemphasizes others. The most we can claim at this point is that it represents a rough guide, a summary of features of cyberspace that deserve fuller and more systematic investigation.
One obvious limitation of the present report is that it focuses on commercial cyberspace. Which of the dimensions outlined above apply equally to .gov, .edu or .int Web sites? Some version of Web Worthiness may apply equally in these additional domains, though some of the details may be different. But dimensions Two through Twelve may fail to emerge, or may emerge in forms that would be quite different. What, for instance, would be the .gov, .edu or .int equivalent of B-to-B vs. B-to-C, or the dimension here labeled Accessibility?
Therefore, a logical extension of the present investigation would be to conduct similar probes of the .gov, .edu and .int domains. Some of the distinctions made here might not reappear. Some distinctions that did not appear here may emerge when .gov, .edu or .int sites are contrasted. Whatever the outcome of this extension, it would be likely to be of value to search engine and Web site designers, and consequently to Web navigators.
Finally, all of the constructs outlined here cry out for more thorough operational definition. With the two-facet Web Worthiness scale proposed in Figure 3, the Attitude Toward the Site scale derived in an earlier investigation (Chen and Wells 1999) and the much more general good — bad, like — dislike, effective — ineffective, bipolar scales used in other studies of Web site differences (Burns 2000, McMillan 2000), we now have at least three ways to measure general Web site evaluation. Which of these alternatives is best for what purposes? Or do they correlate so highly that which one is used doesn’t make any real difference?
Similarly, the terms used here to define factors Two through Twelve are at best crude approximations. Serious work to spell out the meanings and implications of Outreach, Accessibility, Complexity and the other dimensions outlined in the present study would require fuller and more careful operational definition. But these are all good topics for additional research. In the course of this new work, our understanding of, and ability to navigate in, cyberspace would be certain to be furthered.

References

Aaker, David A., and Donald E. Bruzzone. Viewer Perceptions of Prime-Time Television Advertising. Journal of Advertising Research 21, 5(1985): 15-23.

Aaker, David A. and Douglas M. Stayman. Measuring Audience Perceptions of Commercials and Relating Them to Ad Impact. Journal of Advertising Research, (1990) 7-17.

Barsalou, Lawrence W. (1992) Cognitive Psychology: An Overview for Cognitive Scientists, Erlbaum, Chapters 1,2.

Batra, Rajeev and Michael L. Ray. Affective Responses Mediating Acceptance of Advertising, Journal of Consumer Research, 13, September(1986): 234-249.

Besser, Donna. Gerald Stone, Luan Nan (1999) Textbooks and teaching: a lesson from students. Journalism and Mass Communication Educator, Wntr 1999 p4(1)

Brewer William F.(1994) Charles Egerton Osgood: 1916-1991. (psychologist). American Journal of Psychology, Winter, 107(4), 582-98.

Brown, Steven P. and Douglas M. Stayman. Antecedents and Consequences of Attitude toward the Ad: A Meta-analysis. Journal of Consumer Research 19, June (1992): 34-51.

Burns, Kelli S. (2000) Branding in Cyberspace: Using the Congruity of Consumer and Web Site Personality to Unravel Online User Satisfaction, in Mary A. Shaver ed. Proceedings of the 2000 Conference of the American Academy of Advertising, pp. 9-18.

Chen, Qimei and William D. Wells (1999) Attitude toward the Site Journal of Advertising Research, Vol. 39, No. 5, Sept./Oct. pp. 27-38.

Cohen, Joel B. and Basu, Kunal (1987) Alternative Models of Categorization: Toward a Contingent Processing Framework, Journal of Consumer Research, Vol. 13, March 1987.

Eighmey, John. (2000) The Relationships Among Informativeness, User Enjoyment, and Marketing Perceptions of Commercial Sites on the Internet, in Mary A. Shaver ed. Proceedings of the 2000 Conference of the American Academy of Advertising, pp. 231-232.

Goodrich, Anne (2000) Sites for sore eyes, Brandweek; New York; Jan 10.

Green, Paul E., Frank J. Carmone and Leo B. Fox (1969) Television Program Similarities: An Application of Subjective Clustering, Journal of the Market Research Society, 11(1), 70-90.

Green, Paul E., Yoram (Jerry) Wind, Abba M. Krieger, and Paul Saatsoglou (2000) Applying Qualitative Data, Marketing Research, Spring, 17-25.

Hamm-Greenawalt, Lisa (2000) Network Buys-Want a Broad Reach on the Great Big Net? Try Ad Networks, Internet World, May 1.

Kelly, George A. (1963) A Theory of Personality. New York: Norton Publishing. Leavitt, Clark. A Multidimensional Set of Rating Scales for Television Commercials. Journal of Applied Psychology 54, 5(1970): 427-429.

Lehaney, Brian, Steven Clarke, Vikki Kimberlee and Sarah Spencer-Matthews, The Human Side of Information Development: A Case of an Intervention at a British Visitor Attraction, Journal of End User Computing, pp. 33-39, 11(4), 1999.

Lidsky, David and Nancy Sirapyan (1998) Find it on the Web, PC Magazine, December 1.

MacKenzie, Scott B. Richard J. Lutz and George E. Belch. The Role of Attitude Toward the Ad as a Mediator of Advertising Effectiveness: A Test of Competing Explanations, Journal of Marketing Research, 23 (1986): 130-43.

McMillan, Sally J. (2000) Interactivity is in the Eye of the Beholder: Function, Perception, Involvement, and Attitude toward the Web Site, in Mary A. Shaver ed. Proceedings of the 2000 Conference of the American Academy of Advertising, pp. 71-78.

Murphy Gregory L. and Mary E. Lassaline (1997) Hierarchical Structure in Concepts and the Basic Level of Categorization, in Lamberts, Koen and David Shanks (eds.) Knowledge, Concepts and Categories.

Niedenthal, Paula M., Halberstadt, Jamin B. and Innes-Ker, Ase H. (1999) Emotional Response Categorization, Psychological Review, 106(2), pp. 337-361.

Nielsen, Jakob (1990) The Art of Navigating Trough Hypertext. Communications of the ACM, March 33(3) p296(15)

Osgood, C. E., Suci, G, J., & Tannenbaum, P. H. (1957). The Measurement of Meaning. Urbana: University of Illinois Press.

Pashupati, Kartik. The Dimensions of Attitude Toward the Ad: A Re-exploration. Proceedings of American Academy of Advertising.

Riggins, Frederick J. (1999) A Framework for Identifying Web-Based Electronic Commerce Opportunities, Journal of Organizational Computing and Electronic Commerce, 9(4), pp. 297-310.

Rose, G., H. Khoo, and D.W. Straub (1999) Current technological impediments to business-to-consumer electronic commerce, Communications of the Association for Information Systems, 1.

Rose, Sarah (2000) The truth about online banking, Money; New York; April. Schlinger, Mary Jane. A Profile of Responses to Commercials. Journal of Advertising Research 19, 2 (1979): 37-46.

Shimp, Terence A. Attitude toward the Ad as a Mediator of Consumer Brand Choice, Journal of Advertising 10, 2(1981): 9-15.

Stafford, Marla R. and Stafford, Thomas F. (2000) Identifying the Uses and Gratifications of Web Use, in Mary A. Shaver ed. Proceedings of the 2000 Conference of the American Academy of Advertising, p. 70.

The Virtual Mall Gets Real, Business Week, pp. 90-91, Jan. 26, 1998. Net Used to Extend EDI’s Reach, Computerworld, p.1, Feb. 23, 1998.

Wells, William D. EQ, Son of EQ, and the Reaction Profile. Journal of Marketing, 28 (1964a): 45-52.

______________ Recognition, Recall and Rating Scales. Journal of Advertising Research, 4 (1964b): 2-8.

Williams, Thomas R. (1993) Text or graphic: an information processing perspective on choosing the more effective medium. Journal of Technical Writing and Communication, Winter, 23(1) p33(20).

 

Appendix A.

Web site URL: ____________________

Rater: ____________________________

Date: _____________________________

Please evaluate the web site you are rating. Circle the choices that you think best describes this web site.

1. Amount of Detail DETAIL

2. Amount of expertise needed to understand the site EXPERTIS

3. Amount of information provided by the site INFO

4. Appearance APPEAR

5. Target Size TARSIZE

6. How technical is the site? TECHNIC

7. Type of site TYPESITE

8. Background BACKGRND

9. Probable effect on company image COMIMAGE

10. Complexity COMPLEX

11. What is this web site’s attitude toward the user? SITEATT

12. Finished or under construction? CONSTRUC

13. How much effort does this web site make to “educate” its users through explanations, demonstrations, tours, tutorials EDUCATE

14. Average income level of users of this site INCOME

15. Gratification from using this web site GRATIFY

16. How can the user contact the web site’s sponsor? CONTACT (COUNTS)

17. Web load speed LOADSPED

18. How easy is it to get lost in this web site? GETLOST

19. Graphics GRAPHICS

20. How many banner ads are on this site? BANNERAD

21. How many companies are on this site? NO_COMP

22. How is the information conveyed? INFOCONV

23. Interactivity INTERACT

24. Does the site provide company background or history? HISTORY

25. How many Languages? LANGUAGE

26. Does this site have unnecessary layers? LAYERS

27. Layout LAYOUT

28. Links LINKS

29. Navigation structure NAVIGATE

30. Does this site offer products and/or services? PRO_SERV

31. Is this site cold or friendly? COLD_FRE

32. Can the user order online from this site? EMAIL YES=2; NO=1

33. How well organized is this site? ORGANIZE

34. Origin of the business (foreign or domestic) ORIGIN

35. Pop-up windows POPUP

36. Purpose of Site PURPOSE

37. Relevant references (such as related hyperlinks or bibliography)? REFERENCE

38. Do you have to register to obtain what you want from this site? REGISTER

39. Do you have to pay to obtain what you want from this site? PAY

40. Size of the business BUS_SIZE

41. System requirement SYS_REQ

42. Target market MARKET

44. Trustworthy TRUST

45. Type of the business BUS_TYPE

46. Updated UPDATED

47. Web design WEBDESIG

II. The following items assess your general favorability toward the web site you just visited. Circle the number that best indicates your agreement or disagreement with each statement.

This web-site makes it easy for me to Definitely disagree Definitely agree build a relationship with this company. 1…………2…………3…………4…………5   EASYBLD
I would like to visit this web site again in the future. 1…………2…………3…………4…………5   VISTAGAN
I’m satisfied with the service provided by this web site. 1…………2…………3…………4…………5   SATISERV
I feel comfortable in surfing this web site. 1…………2…………3…………4…………5   COMFSURF
I feel surfing this web site is a good way for me to spend my time. 1…………2…………3…………4…………5   GOODTIME
Compared with other web sites, I would One of the Worst One of the Best rate this one as 1…………2…………3…………4…………5   RATING

 

About the Authors

William D. Wells is Professor in the School of Journalism and Mass Communication at University of Minnesota at Twin Cities. Qimei Chan is a doctoral student in the School of Journalism and Mass Communication at University of Minnesota at Twin Cities.