当前位置:首页 >> >> A Comparison of the Most Important Website Features in Different Domains An Empirical Study

A Comparison of the Most Important Website Features in Different Domains An Empirical Study


A Comparison of the Most Important Website Features in Different Domains: An Empirical Study of User Perceptions
Ping Zhang, School of Information Studies, Syracuse University, pzhang@syr.edu Gisela von Dran, School of Management, Syracuse University, gvondran@som.syr.edu Paul Blake, School of Information Studies, Syracuse University, pzhang@syr.edu Veerapong Pipithsuksunt, School of Information Studies, Syracuse University, pzhang@syr.edu
by focusing on users' different quality expectations. Again, the empirical data showed that the model can be used to distinguish the features that meet users' basic, performance, and excitement quality needs. Both studies imply that the specific web domain or the purpose of a website impacts what users think about the features as satisfiers/dissatisfiers or how they meet different quality needs. On the other hand, the studies did not address user perceptions on whether some features are relatively more important than others. The objective of this study is to use a bottom up approach to examine user perceptions on the relative importance of features in different domains. The results show that (1) the importance of features or families of features is dependent on the particular domain a user is working with; (2) certain features or families of features are extremely important for one domain and extremely unimportant for others; (3) there are certain features or families of features that are equally important among different domains. The current study provides designers with empirical data of the most important features to focus on when faced with design capacity limitations.

Abstract
This study uses an inductive thematic analysis approach to examine user perceptions on the importance of website design features in six different website domains: Financial, E-Commerce, Entertainment, Education, Government, and Medical. The five most important features, as well as the five most important families of features, were identified for each of the domains. The results indicate that (1) there are certain features that are equally important among different domains; (2) there are other features that are extremely important for one domain and extremely unimportant for another. The study provides empirical evidence for website designers and evaluators about which features are more important to focus on when dealing with different domains of websites. It adds value to the current literature on consumer behavior in the electronic environment and web usability studies.

Introduction
In the web environment, users are consumers. Understanding consumers' expectations and how they feel about the websites they use has recently become more important. Few current web usability studies are based on either theoretical frameworks or empirical evidence (Conger & Mason 98; Small 98; Spool et al. 99; Wilkinson et al. 97). Most studies provide some guidance for designers largely based on heuristics or rules of thumbs. These studies do not identify website design features that contribute to consumer satisfaction or dissatisfaction, address different quality expectations, nor do they provide any insight into whether some features are perceived more important than others by the users. Zhang et al. (1999, 2000) provided an emerging theoretical framework to distinguish between the website design features that satisfy users from those features that dissatisfy users. In their study, subjects were asked to classify certain features into satisfiers and dissatisfiers, which showed support for the framework. Von Dran et al. (1999, 2000) approached the issue from a different angle. They applied a marketing model to the web environment

Research Method
Few existing studies provide either theoretical frameworks or heuristics of examining the relative importance among website design features. Thus, there are few existing theories or models that can be used to guide the study. We decided to use an inductive (data driven) thematic analysis approach (Boyatzis, 1998) in this study.

Data Collection
Six domains of websites were chosen for the study. They were: ? ? ? Financial Information Websites (such as CNNfn.com, quote.yahoo.com) E-Commerce Websites (such as Amazon.com, eBay.com) Entertainment Websites (such as a cartoon or a game website)

Proceedings of Americas Conference on Information Systems (AMCIS'2000), Long Beach, CA. August 10-13, 2000. pp. 1367-1372

? ? ?

Educational Websites (such as National Geographic or a university's website) Governmental Websites (such as US Department of Labor, and the White House website), and Health or Medical Information Websites.

that in a particular domain, only one or two codes were used from a super code. Scaling of the initial code schema Before the code schema was applied to other domains, the super codes were scaled (Boyatzis 98, p134) into a more manageable list. This included the consolidation of similar super codes as a new super code with a higher level of abstraction. For example, after the scaling, the codes for s181's quotation for the entertainment domain became (1) multimedia 1, (2) interactivity 2, (3) visual design 3, (4) site responsiveness 4, and (5) links to info 5. The result of the scaling was a new code schema of 118 codes. Application of the code schema to other domains When the two raters coded the remaining domains, the original words or phrases from the subjects were either identified as belonging to an existing super code or a new super code. Consensus meetings were conducted for coding results of each of the domains and the inter-rater reliability scores were calculated both before and after the meetings. Clustering similar codes into families Clustering is defined as “…the organization of multiple themes into groups” (Boyatzis, 1998, p134). The clustering of the super codes revolved around the creation of families and placement of super codes within those families. For example, the family of “Navigation” included codes like “easy to navigate,” “navigation aids,” and “clear layout of info,” to name a few. The clustering was based on the code schema and not on any previous theories, so these families more accurately reflect the respondent’s answers. Overall scaling and clustering As an iterative process, scaling was conducted again once all domains were coded. This scaling task is coupled with the refinement of families. Several super codes with single responses (one response for the entire super code) were compressed with other super codes. Similarly, family memberships were adjusted in order to eliminate families with only one super code and to reflect stronger semantic coupling among super codes.

In a survey, subjects were asked to list, in priority order, the five most important website features for each of six different website domains. 67 graduate students at a major northeast university participated in the study. Among the subjects, 32% were male and 68% female. The average age was 33 (with a standard deviation of 8). Subjects were paid $10 upon their completion of the survey. Three of the subjects did not understand the requirement and provided unusable answers, and these sets of data were dropped during the analysis. Table 1 shows the example answers from one subject.

Coding in Thematic Analysis
Thematic analysis (Boyatzis, 1998) was conducted on subjects' answers to the questions. In this data-driven approach, two independent raters worked directly from the raw answers to extract words and phrases, which were used to generate the codes. This close relation between the codes and the raw answers helped to improve the coding consistency between the raters. The codes are measured by the magnitude of the appearance (that is, frequency). The software used for the coding was ATLAS.ti, version WIN 4.2. Developing a initial code schema The unit of the analysis (defined as a quotation in ATLAS.ti) was regarded as the whole answer a subject had for one domain. The unit of coding (a code in ATLAS.ti) was the particular features that subjects listed in their answers. The codes were developed using the original words and phrases in the quotations in one domain, the financial domain. Most responses in the survey are manifest, however, some are latent, which were interpreted by the raters (Boyatzis 98, p16). A consensus meeting with a third rater resolved the disagreements between the two independent raters. This included establishing rules of how to break quotations into meaningful units of coding while keeping the priority ranking provided by the users in the codes. Thus, for s181's answer for the entertainment domain (which is a quotation, see Table 1), five codes were developed with the priority embedded in the codes: (1) multi-media 1, (2) interaction 2, (3) display/images/graphics 3, (4) quick download time 4, and (5) links 5. In this paper, a super code is defined as a term with distinctive meaning; and a code is a super code with a suffix indicating the priority. For example, "customization" is a super code and may include five codes: customization 1, customization 2, customization 3, customization 4, and customization 5. It is thus possible

Data Analysis and Results
Some subjects mentioned that they did not use or never used websites in certain domains (see s181 in ecommerce domain in Table 1). Thus, they could not and did not provide any opinions on which features were most important. For those participants who provided their perceptions on some or all domains, the analysis was conducted at two levels: code and the cluster (or family as noted in ATLAS.ti) of codes. Sometimes designers or evaluators of websites need to focus on a small number of

Proceedings of Americas Conference on Information Systems (AMCIS'2000), Long Beach, CA. August 10-13, 2000. pp. 1367-1372

factors that affect user perceptions of websites. Thus, it may be helpful to group features into higher level of abstract units, namely clusters or families of features. These families may provide a better overview of the characteristics of website features. Since subjects were able to give a list of features with priority (order of importance), we used this information in our analysis in the form of weighted frequencies at both the code and family levels.

Weighted Rank of the Five Most Important Families for Each Domain
There is a total of 15 families/clusters of features as shown in Table 3. One of them is for the responses of "Do not use / never used the domain" and is disregarded from the analysis. The weighted score of a family is calculated by using the weighed scores of the super codes belonging to the family. Table 4 lists the five most important families for each of the six domains. Table 4 shows that. 1. 2. 3. Navigation is ranked among the top three most important families in all domains. Completeness/Comprehensiveness of Info is among the top two most important families in all but the Ecommerce and Entertainment domains. Site Technical Features (most responses are from Search Tool feature) is ranked from the 3rd to the 5th family in all but Financial and Entertainment. This implies that users take whatever is available on the first page/immediate access on websites of these domains. They don't expect to search in these websites. Currency/Timeliness/Update is among the top three for the Financial, Medical and Government domains. Accuracy is listed as the 4th or 5th family for the Financial, Medical and Government domains. Readability/Comprehension/Clarity is ranked as 4th or 5th for the Financial, Education, and E-Commerce domains.

Weighted Rank of the Most Features for Different Domains

Important

For each code in each domain, the weighted score is determined by the frequency of the code in the domain multiplied by the weight for the priority that was assigned by the subjects. That is: Score = PriorityWeight * Frequency, while PriorityWeight is designed as: First priority (most important) has a weight of 5, Second (second most important) 4, third 3, fourth 2, and fifth 1. Table 2 lists the five most important features for each of the six domains based on the weighted frequencies. The following are some observations from the table. 1. 2. Financial domain has high requirements on the nature of the information, such as up to date, accuracy, multiple sources, and timeliness. Easy to navigate is also very important as ranked as number 4 for the financial domain. For all other domains, however, easy to navigate is highly ranked as either number one or two. Thus, it is a must-have feature for all six domains. Up-to-date info is very important for the financial domain, and is true for the government, medical and entertainment domains. The feature, however, is not listed within the five most important features for the education and e-commerce domains. Entertainment domain has high demand on visual design, multimedia and site responsiveness, which are not in the list of any of the other five domains. Search tool is commonly ranked by the following four domains as important: education, government, medical, and e-commerce. Education and medical domains require comprehensiveness of information, which is not ranked within the five-most important list in the other four domains. Accuracy of information is most important for the medical domain, somewhat important for the financial, education and government domains, but is not within the five-most for the e-commerce and the entertainment domains. Security of data is ranked number one in e-commerce domain but does not appear in any other domains.

4. 5. 6.

3.

Table 4 also indicates there are three domains that require unique families. For example, the Education domain requires Info Reliability/Reputation; E-Commerce demands Security/Privacy and Product and Service Concerns; while Entertainment requires four unique families: Visual Design, Engaging, Info Representation, and Site Accessibility/Responsiveness. Figure 1 depicts the similarities or differences among the domains in terms of the composition of the most important families. Some domains share the common families. Figure 1 confirms some of the observations from Table 4. Specifically, Figure 1 shows that: 1. The Government, Education and Medical domains have similar "patterns" of the most important families. For example, they all have high F02 and F07 and low F04, F06, F09, F12, and F13. This means that designers can focus on the concerns that these domains have in common. Financial, E-Commerce and Entertainment do not seem similar to any other domains, or to each other. This implies that these domains should be designed differently from other domains focusing primarily on the particular reason that users come to the sites.

4. 5. 6.

7.

2.

8.

For E-Commerce domain, users treat products/services as website features. This implies that having impressive

Proceedings of Americas Conference on Information Systems (AMCIS'2000), Long Beach, CA. August 10-13, 2000. pp. 1367-1372

or great website features alone is not enough - users need good products and services from the website.

Conger, S.A. and R.O. Mason, Planning and Designing Effective Web Sites, Course Technology, Cambridge, MA, 1998. Small, R.V., Assessing the Motivational Quality of World Wide Websites, ERIC Clearinghouse on Information and Technology, Syracuse, NY, 1998, (ED 407 930). Spool, Jared, T. Scanlon, W. Schroeder, C. Snyder, T. DeAngelo, Web Site Usability - A Designer's Guide, Morgan Kaufmann Publishers, Inc. San Francisco, California, 1999 von Dran, Gisela, Ping Zhang, and Ruth Small (1999), Quality Websites: an Application of the Kano Model to Website Design, Proceedings of the 5th Americas Conference in Information Systems (AMCIS'99), August 13-15 1999. von Dran, Gisela and Ping Zhang (2000), A Model for Assessing the Quality of Websites, Proceedings of Annual Conference, American Association for Information Science (ASIS 2000), Chicago, November 13-16, 2000 Wilkinson, G. L. , L.T. Bennett, and K.M. Oliver, Evaluation Criteria and Indicators of Quality for Internet Resources, Educational Technology, May-June, 1997, 5259. Zhang, Ping, Gisela von Dran, Ruth Small, Silvia Barcellos (2000), A Two-Factor Theory for Website Design, Proceedings of the Hawaii International Conference on Systems Science (HICSS 33), Hawaii, January, 2000. Zhang, Ping, Gisela von Dran, Ruth Small, Silvia Barcellos (1999), Web Sites that Satisfy Users: A Theoretic Framework for Web User Interface Design and Evaluation, Proceedings of the International Conference on Systems Science (HICSS 32), Hawaii, January 5-8, 1999.

Discussion and Conclusions
The analyses of codes and the families of codes show some interesting facts about users' perceptions on importance of website features and families of features. 1. 2. The importance of features or families of features is dependent on the particular domain a user is working with. Certain features or families of features are extremely important for one domain and extremely unimportant for others. For example, Engaging is the 2nd most important for Entertainment, but almost the least important for the other five domains; Security/Privacy is the most important family for ECommerce domain but is listed not important in the rest of the domains. There are certain features or families of features that are equally important among different domains. For example, Navigation is among the top important families of all the domains.

3.

The findings provide practical suggestions to at least three types of people. For website designers, the study implies that different domains should be designed with different foci of important features. For website owners or corporate strategists of E-C websites, the study indicates that users regard the website design and company products/services as one unit. For independent website evaluators/critics, this study recommends that different domains require different sets of evaluating criteria/tools.

References
Boyatzis, Richard E. (1998). Transforming Qualitative Information: Thematic Analysis and Code Development. Thousand Oaks, CA: Sage Publications.

Table 1. Example Answers by One Subjects Subj. 1. Financial ID current information (ie. recent updates), variety of different markets, readily s181 available detailed information, other links, graphs and other supporting historic data 2. ECommerce don't really utilize ecommerce websites 3. Entertainment 4. Educational multi-media, interaction, displays, sharp images, graphics, quick download time (if applicable), links navigation to find appropriate material, good searches (advanced features), downloadable publications, so you don't have to view online (pdf files), links, references 5. Governmental organization, table of contents, current information, easy access to current regulations, good searches, down loadable regulations 6. Health or Medical references to medical associations, current information, searches, different points of interest, accessibility

Proceedings of Americas Conference on Information Systems (AMCIS'2000), Long Beach, CA. August 10-13, 2000. pp. 1367-1372

Table 2. Five Most Important Features Order 1 2 3 4 5 Order 1 2 3 4 5 Financial Up-to-date info. Accuracy of info. Multiple information sources Easy to navigate Timely info. Score 92 81 76 52 32 Educational Easy to navigate Search tool Accuracy of info. Comprehensiveness of info. Clear layout of info. Score 107 85 72 55 54 Score 87 60 53 53 52 Governmental Easy to navigate Clear layout of info. Up-to-date info. Search tool Accuracy of info. Entertainment Visual design Easy to navigate Site responsiveness Multimedia Up-to-date info. Score 100 77 66 64 62 Score 172 70 68 58 50

E-Commerce Score Health or Medical Security of data 121 Accuracy of info. Easy to navigate 97 Easy to navigate Appropriate explanatory text 59 Search tool Search tool 45 Up-to-date info. Product and service price concerns 44 Comprehensiveness of info.

Table 3. A List of Families of Super Codes and Frequency Counts Across Domains FID Family F01 Accuracy Completeness/Compr F02 ehensiveness of Info Currency/Timeliness/ F03 Update F04 F05 F06 F07 F08 F09 Definition FIN E-C ENT EDU GOV MED No errors, correct, exact, precise, right, true 19 11 2 19 14 21 Large in scope or content, containing a variety 51 13 23 53 42 51 of information or sources Information is current, up to the moment, real 46 10 16 18 32 27 time, timely Cognitive advancement, emotional connections, Engaging 5 3 42 13 2 7 personal expressions Info dependable, the condition of being held in Info Reliability/ 7 5 19 14 17 high esteem, being authoritative, high reputation 11 Reputation of info source The way information is presented, maybe in Info Representation 16 11 22 8 1 3 different format/media, customized displays Navigation Features to make navigation possible, site maps 31 65 33 55 53 35 Visual Design Visual Appearance 2 7 46 19 3 3 Features concerned with products/services Product and Service offered/sold through the website, not about the 8 64 4 5 10 12 site itself; price and availability of Concerns products/services Ability to comprehend the meaning of written or Readability/Compreh 19 11 22 20 18 printed words or symbols, to perceive or receive 17 ension/Clarity well Information that directs to the point, having to 19 1 0 12 8 13 Relevant Info do with the matter at hand Confidentiality of info, things that gives or 7 47 4 1 6 9 Security/Privacy assures safety and guarantee Site Accessibility/ Being able to access the website; responsiveness 12 19 21 10 4 8 Responsiveness of the site to user's request in terms of time. Site Technical Features such as search tools, downloadable 6 19 2 30 24 22 Features (printer friendliness), chat rooms. Do not Use / never 12 3 10 2 12 8 used Total Frequency 262 299 241 286 245 254 Totals 86 233 149 72 73 61 272 80 103

F10 F11 F12 F13 F14 F15

107 53 74 74 103 47 1587

Proceedings of Americas Conference on Information Systems (AMCIS'2000), Long Beach, CA. August 10-13, 2000. pp. 1367-1372

Table 4. The Most Important Families Order Financial 1 Currency/Timeliness/ Update 2 Completeness/ Comprehensiveness of Info 3 Navigation 4 5 Accuracy Readability/Comprehension/ Clarity Score Educational 173 Navigation 129 Completeness/ Comprehensiveness of Info 93 Site Technical Features 81 54 Info Reliability/Reputation Readability/Comprehension/ Clarity Score Governmental Score 186 Navigation 193 142 Completeness/ 114 Comprehensiveness of Info 98 Currency/Timeliness/ 96 Update 79 Site Technical Features 76 78 Accuracy 62

Order E-Commerce 1 Security/Privacy 2 3 4 5 Navigation Product and Service Concerns Readability/Comprehension/ Clarity Site Technical Features

Score Health or Medical 201 Completeness/ Comprehensiveness of Info 196 Navigation 162 Currency/Timeliness/Update 64 Accuracy 56 Site Technical Features

Score Entertainment 149 Visual Design 111 97 87 64 Engaging Navigation Info Representation Site Accessibility/ Responsiveness

Score 172 132 105 74 68

Figure 1. The importance of families for six domains: similarity and differences among domains
Governm ent
E duc ation

M edic al

F14 F13 F12

F01 200 100 0

F02 F03 F04 F05 F06
F13 F12

F14

F01 200 100 0

F02 F03 F04 F05 F06

F14 F13 F12

F01 200 100 0

F02 F03 F04 F05 F06

F11 F10 F09 F08 F07

F11 F10 F09 F08 F07

F11 F10 F09 F08 F07

Financial

E -Com m erc e

E ntertainm ent

F14 F13 F12

F01 200 100 0

F02 F03 F04 F05 F06 F13 F12

F14

F01 200 100 0

F02 F03 F04 F05 F06 F13 F12

F14

F01 200 100 0

F02 F03 F04 F05 F06

F11 F10 F09 F08 F07

F11 F10 F09 F08 F07

F11 F10 F09 F08 F07

Proceedings of Americas Conference on Information Systems (AMCIS'2000), Long Beach, CA. August 10-13, 2000. pp. 1367-1372


赞助商链接
更多相关文档:
更多相关标签:
网站地图

文档资料共享网 nexoncn.com copyright ©right 2010-2020。
文档资料共享网内容来自网络,如有侵犯请联系客服。email:zhit325@126.com