Internet-based evaluation of tourism Web site effectiveness: Methodological issues and survey results
Journal of Travel Research; Boulder; Nov 2000; Patrick Tierney;

Abstract:
There is a growing reliance on the Internet and Web sites for promotion of tourism. Evaluation of Web site effectiveness is necessary because of significant costs for setup, advertising, and maintenance. Research suggests there is a great need for evaluations to go beyond hits and page viewings. The objectives of a study were to develop and apply a low-cost, automated, Internet-based survey methodology to investigate effectiveness of a tourism promotional Web site. Both online and email surveys were used to gather data from visitors to a state Web site. Although responses were received from 833 persons, response rates were low. A check for nonresponsive bias indicated that respondents were different from nonrespondents. Respondents primarily learned about the Web site through two sources and reported that the Web site significantly influenced their travel plans.

[Headnote]
There is growing reliance on the Internet and Web sites for promotion of tourism. Evaluation of Web site effectiveness is necessary because of the significant costs for setup, advertising, and maintenance. Research suggests there is a great need for evaluations to go beyond hits and page viewings. The objectives of this study were to develop and apply a low-cost, automated, Internet-based survey methodology to investigate effectiveness of a tourism promotional Web site. Both online and email surveys were used to gather data from visitors to a state Web site. Although responses were received from 833 persons, response rates were low. A check for nonresponse bias indicated that respondents were different from nonrespondents. Respondents primarily learned about the Web site through two sources and reported that the Web site significantly influenced their travel plans. However, study findings suggest that there are substantial methodological challenges in conducting Web-based surveys. Suggestions to improve evaluation methods are provided.

The trend is clear; there is growing reliance on the Internet for the promotion and sales of tourism and most products and services. Almost all companies and tourism promotional organizations now have Web sites. Sales over the Internet doubled from 1998 to $7 billion during the 1999 Christmas Holiday period (Jupiter Communications, 2000). Online revenues from travel sales are projected to increase 440% between 1997 and 2000, with travel being one of the more frequent services purchased on the Internet (Tourism Industries of America 1999). A 1999 survey of state tourism offices found that in the United States, all of the states had a tourism promotional Web site (Tourism Industries of America 1999).

Evaluation of Web site effectiveness is necessary because there are significant costs involved in setup and maintenance of Web sites ($180,000/year according to the Association of National Advertisers 1999), intense competition for Web viewers, and increasing costs for advertising to direct potential buyers to a Web site. Although direct online e-commerce sales are one good indicator of effectiveness, many tourism Web sites do not directly sell services and cannot use this as a gauge. Tourism companies that sell complex products, such as tours, have noted that visitors often only gather information from a Web site and then call or write to request a brochure or make a reservation. In this case, the Web site may have had a powerful role in the purchase decision, but the sale itself did not take place over the Internet. Therefore, additional research is needed to fully determine Web site effectiveness in promoting tourism services.

The shift to promotion of tourist products and services through the Internet provides new opportunities and challenges for researchers desiring to evaluate the effectiveness of Web sites, compared to traditional media. The nature of the Internet allows Web site managers to quickly learn the number of hits, visitors to the Web site, and advertisements. However, knowing the number of viewers of a Web page tells little about visitor characteristics, motives for visiting the site, satisfaction with the site, and actions he or she will take as a result of viewing the Web site. This urgent need to go beyond hits or page viewings in Web site evaluation is echoed by Forrester Research (1999).

In addition to traditional phone and written surveys, online or real-time surveys have been applied to tourism analysis. But as Schonland and Williams (1996) found in the 1995 Net Travelers Survey, there are unique issues and concerns with online surveys, compared to traditional approaches.

STUDY PURPOSE

The objectives of this study were to develop and apply a low-cost, automated, Internet-based survey methodology and to use it to investigate the effectiveness of a tourism promotional Web site. The official Web site of the California Division of Tourism (CalTour) located at http://www gocalif.ca.gov was used in this study because it is a comprehensive site, heavily visited, and the state was interested in documenting the effectiveness of this form of promotion. The Web site has content and images about activities and things to do, travel regions and cities, sightseeing, maps, insider tips, lodging, restaurants, and shopping, but at this time, it did not sell products online.

BACKGROUND

Published research into the effectiveness of tourism Web sites is still very limited. Tourism Industries of America (1997) conducted a telephone survey of Internet users and found that in 1997, American adults most frequently used the Internet to gather information on things to do (31%), where to go (27%), and for schedules (27%), but only 17% actually made a reservation via the Web.

Schonland and Williams (1996) were one of the first studies to employ Web-based survey techniques to evaluate use of the Internet for travel services. The survey used sponsor incentives and contained questions on travel patterns, destinations, airlines used, Internet use, and online buying. Researchers were able to acquire a large sample size (17,700) over a 12-month period, but they found a profound response bias in their data based on incentives used.

The Tourism Industry of America (1999) conducted a mail survey to assess technology utilization by state tourism agencies in the United States. In 1999, all states had a tourism Web site, and the average state Web site received more than 190,000 hits and slightly fewer than 19,000 user sessions per week. The most common approaches to measure effectiveness of state Web sites were number of hits (71 %), number of user sessions (69%), and online feedback from site visitors (67%). Findings showed the heavy reliance on state Web site effectiveness determination by simple monitoring of hits and user-session statistics.

A recent study on the effectiveness of the state tourism Web site was conducted by the Arkansas Department of Parks and Tourism in 1999 (McLemore and Mitchell 2000). Determining if persons inquiring for information via the Web actually visited the destination was a primary goal of this research. It employed an online conversion survey asking respondents if they visited the state after requesting information (68% had visited), how they learned about the state Web site (58.4% via a link from another Web site), trip duration and expenditures, and future plans to visit the state (90% had plans in next 12 months). The sample consisted of persons who had inquired for information via the state Web site. No information was gathered on visitors to the site who did not request information. Inquirers were emailed an invitation to complete the online survey and, therefore, needed to click on to the provided link or visit the survey site later. Using this approach, the study achieved a 32.2% response rate. Authors found that this online method was very cost effective in providing current market information in a very quick turnaround time, compared to conventional survey methods. Limitations of this approach for measuring effectiveness were no identification of the impact of nonresponse bias on findings and the sample being restricted to persons who inquired for information.

As the 1999 Tourism Industries of America study showed, the most common statistic used to measure Web site effectiveness is the number of hits to the site. The term hits is commonly defined as the number of files that are downloaded from a Web server (host computer). Tracking hits as a way of measuring traffic can be misleading. The number of hits a site receives is usually much greater than the number of actual visitors because a Web page usually contains more than one file. More reliable is the statistic called "user sessions," defined as "a session of activity (all hits) for one user of a Web site" (WebTrends, 2000). Web traffic software, such as WebTrends (see http://www.webtrends.com), can provide information on site visitors' domain name and type, region or country of request, organization type (com, net, org, etc.) ad viewings (clicks), and time of view. However, this type of software cannot give critical information about the user, such as income, reason for visiting the site, satisfaction with the site, or actions taken because of viewing the Web site. Clearly, to fully understand the impact of a Web site, more information than just the number of hits or user sessions is needed.

It is possible for host computers to use coded messages sent to and stored in the viewers browser, called "cookies," to identify if a browser has visited a Web site previously and track respondent preferences and page viewings. But cookies cannot identify users, and sophisticated users simply turn off the cookies option in their browser.

The ways in which detailed data are collected about the promotional effectiveness of Web sites can be quite different from traditional approaches. One approach is to email a survey to respondents and they reply by emailing it back. Email surveys require that respondent addresses are known in advance, and this may not be possible.

Online or Web-based data collection surveys are growing in popularity. These surveys are placed on a Web page, and respondents react interactively with the survey. It is common to program automated skips to make it much quicker for the respondent to read and respond to a complex multipage survey. Color images and the full color and format options of the Web are available. Respondents usually access the online survey by a link to it in an email message, or on a Web page.

Recent studies have documented a large decline in clickthrough rates from ads on a Web site to an advertiser's site, dropping from 2% to 0.5% in a few years (Neilson 1999). This suggests that it may be increasingly difficult to get potential Web "surfers" to become survey respondents. Since the respondent must click on the link to find an online survey, an incentive is often necessary to get the user to do so. Incentives are also needed with email surveys because of the difficulty in getting the attention and cooperation of persons who receive numerous emails each day, many of which are unsolicited and quickly deleted. Use of incentives with Internet surveys has been shown to cause at least three methodological concerns: response bias, multiple entry, and unwanted entries. Schonland and Williams (1996) found that when they changed incentives, survey responses and types of respondents changed. It is also possible for one respondent to enter incentive contests more than once to increase their chances of winning. Schonland and Williams found that approximately 8% of respondents to the Net Traveler Survey entered more than once. A final concern is unwanted entries caused by contest sites on the Web. Contest sites search the Web for contests and inform their visitors about the prizes and provide a link to them.

Probably one of the most serious concerns about Internetbased surveys is nonresponse bias. Email and online surveys, like traditional survey methods, can only gather data about respondents who choose to participate. But what is unique is the potentially very low response rates of Internet-based surveys and the difficulty in knowing any information about nonrespondents (Schonland and Williams 1996). This is a concem if nonrespondents differed significantly from respondents.

Nonresponse bias in tourism effectiveness studies has also been shown to be related to the tendency for respondents to be more likely to visit the destination, compared to nonrespondents (Ellerbrock 1981; Burke and Gitelson 1990). Low response rates in Internet-based surveys may exacerbate this problem, compared to traditional survey methods.

Online surveys have a potential advantage in the ability to acquire large sample sizes at a very low cost and in a short time. This is due to the automation of data gathering, and entry and compilation of data into descriptive statistics. Internet survey software is available and takes the written survey and converts it into email or Web-compatible formats, emails or administers online surveys, and then automatically collects responses, enters them into a database, and calculates descriptive statistics. Using this software on a busy Web site, it is possible to collect hundreds of responses in a short period.

A key question for Internet-based survey methodology is if problems of sample validity, due to low response rates, are mitigated by the large number of completed surveys that are possible to acquire. It appears that no published tourism studies have been able to answer this question for Internet-based surveys, but it seems very important to do so.

Two traditional approaches have been developed for the evaluation of tourism media promotion effectiveness. Table 1 describes how the goals, indicators, data collection, and sample techniques of traditional tracking and conversions studies are distinct. It is important to note if the goals of evaluation research are to measure changes in awareness (tracking study) or determine visitation by persons inquiring for information (conversion study).

Considerable tracking and conversion study research has been done on the evaluation of the effectiveness of tourism promotions (Ballman et al. 1982; Burke and Lindblom 1989; Tierney 1992; Van Hoof and Hobson 1997). Particularly insightful about the problems with these types of research was a series of articles prepared by the U.S. Department of Commerce Task Force on Tourism Research Accountability (Wynegar 1989). Findings suggested that serious methodological problems were numerous, including a lack of information about motive for seeking tourist information (many are not visit related), if the promotional material had any impact on their decision (to visit or buy), difficulty in quantifying the link between advertising and tourist spending, and sample size and precision problems (Davidson and Wiethaupt 1989; Burke and Gitelson 1990). Research evaluating the effectiveness of tourism Web sites must also incorporate these concerns into their study design.

METHOD

This research had an overall goal of developing and testing an Internet-based methodology for measuring the effectiveness of a tourism promotional Web site. It incorporated more of a tracking-study approach, wherein an attempt was made to gather data about intention to visit from a sample of all the persons who viewed the Web site, versus a conversion approach that was limited to only persons who inquired for information via the Web site.

The overall research design incorporated three phases (see Table 2). In the initial prephase survey, data were gathered while the respondent was visiting the Web site. In the postphase, a follow-up email survey was sent several months after visiting the Web site during which consumer decisions may have been made. Finally, a second prephase survey used a revised questionnaire and garnered more information on the usefulness of the Web site.

The first prequestionnaire was delivered to respondents through a link on the CalTour Web site home page during the period September 5, 1997, through January 6, 1998. The link was an animation that appeared to every third visitor to the state Web site. It flashed the words "Win," then "A Special California Jacket," and then "Enter By Clicking Here." After clicking on the animation, the visitor was linked to a Web page that contained the online survey. This alternating sampling was done to provide a systematic sample and to produce a manageable number of survey responses.

Survey incentive was an entry into a drawing to win either a California sesquicentennial embroidered jacket or a California Fun Pack of discount coupons. These particular incentives were selected because they have general appeal to persons interested in visiting California, were related to the ongoing statewide celebration, but were not too valuable as to draw contest site and unwanted viewers.

The study used Decisive Survey Web surveying software produced by Decisive Technology (see http://www.messagemedia.com/) to write the survey questionnaire and make it into a file that was later placed on the CalTour server. This software also received responses submitted online, coded them, and automatically emailed responses to an account at San Francisco State University (SFSU). The same software on a desktop computer at SFSU then read the email and automatically entered the response into a database file. Researchers at SFSU could then access the database, view quickly generated tables and charts of results, and email these to the study sponsors keeping them informed as the research progressed.

The two prephase surveys were interactive questionnaires written to glean data on the viewers reasons for visiting the Web site, their demographic characteristics, how they learned about the Web site, satisfaction with it, actions taken while on the Web site, and if they requested a brochure online. The motivation questions asked respondents if they had already decided to visit the state and wanted Web-based information to help them plan their trip, if they were considering visiting in the next 2 years, or if they had no plans to visit the state soon. The survey asked about actions they had already taken or planned to take immediately, such as if they had viewed more than five pages, linked to another Web site via the state site, or printed a page. The presurvey also asked respondents if they were willing to respond to a follow-up survey and be eligible for additional prizes. If they agreed, they were requested to provide their email address and other contact information.

The postsurvey was designed to gather information from respondents of the first presurvey to learn if they visited the state and what effect their viewing of the state Web site had on their visit. Each was asked if they made a trip to or through California that was at least partially for leisure purposes since viewing the CalTour Web site. Also asked was if they changed their travel as a result of viewing the state Web site.

Respondents were divided into residents and nonresidents, and each set was provided a separate, but similar, survey. Nonresidents were asked if they visited the state. Residents were asked if they visited a new city or area in California that was at least 50 or more miles from their home. If a resident did not visit a new area of the state more than 50 miles from home, then they were not asked further impact questions.

Respondents who visited new areas were asked if their decision to visit was made before or after they viewed the state Web site. Those who had already decided to visit were asked if viewing the Web site influenced them to stay longer and how many days longer. All visitors were asked if they changed travel plans as a result of viewing the Web site, if they visited a destination featured or advertised on the Web site, and if the Web site influenced this visit. Similar questions asked if they dined in a featured restaurant, attended an event, visited a store, or stayed at lodging seen on the Web site. Respondents were also asked their likelihood of visiting in the next 2 years and if viewing the Web site influenced their probability of visiting.

The postsurvey was sent from SFSU via email using the Decisive software. Replies were sent directly from the respondent to an email account at SFSU. The survey was emailed on April 3, 1998, approximately 4 months after data collection stopped on the first prephase. This lapse in time was designed to allow viewers to have made some decisions about future visits but not so long as to create recall problems. The postsurvey offered an incentive of a drawing to win a gift certificate for airline travel within the state. This incentive was selected so it would appeal to both nonresident and resident respondents. Nonrespondents were emailed a reminder message and another survey twice before data collection was ended.

In the second version of the presurvey, most questions remained the same to ensure comparability, but several questions were added to measure usefulness of the Web site for finding information about various types of state tourist facilities and services. Data from this survey were collected from September 4 to October 22, 1998. The second presurvey was accessed by clicking on a simple "Win Prize" image that appeared to every 10th person's viewing of the CalTour Web site home page. The incentive was a chance to win a "California Souvenir Pack," consisting of pins, posters, and pens.

RESULTS

In the 5-month period in which the first prephase survey was online, 762 persons clicked on the survey animation and completed the questionnaire. To estimate response rates for the survey Ca]Tour Web site, traffic statistics were compiled for the month of October 1998 alone. In this 1 month, 844,024 hits and 33,943 user sessions were recorded. The survey animation appeared 4,555 times, and 410 persons clicked on the animation. This is a click-through rate of 9.0% of persons who could have seen the animation. A total of 90 persons actually submitted a survey response. Respondents represented 2.0% of total animation appearances and 21.9% of persons who viewed the survey. Approximately 4.0% of initially completed surveys were entered twice, and these duplicate entries were removed before statistical compilation.

These response statistics have both positive and negative connotations. Click-through rates for Web-based ads during this same period were between 1.0% and 2.0%; therefore, a 9.0% rate for the animation was, by this measure, very high. But the 2.0% rate of survey completion based on total views of the animation is low. The potential for nonresponse seems very real. Counterbalancing this concern is that the total number of respondents was relatively large (762) compared to many tourism studies.

A total of 88.8% of the persons who completed the first prephase survey agreed to participate in the postphase survey and provided their email address. These individuals were emailed the postphase survey, and 153, or 37.3% of the total sent, returned the instrument.

Data from the second prephase survey were collected for 47 days. A total of 71 responses to this revised survey were received.

Table 3 shows that respondents tended to be widely distributed among age groups, more likely male, highly educated, and from a diverse range of incomes. Presurvey and postsurvey respondent ages were similar.

Respondent residence in the first presurvey was dominated by California (36.2%), followed by the Midwest (15.7%), Canada (13.4%), and other international countries (11.0%). More than 24.4% of respondents were from other countries, illustrating the reach of the Web site for global promotion. Table 4 shows that the percentage of nonresident respondents increased in the second presurvey.

Visitors overwhelmingly learned about the state Web site via investigations on the Internet (61.2%), using search engines, such as Yahoo and Alta Vista. No other discovery method was nearly as popular. Table 5 shows that a link in another Web site accounted for 16.4% of respondents, followed by the address in a magazine or newspaper (4.7%) and in state visitors guidebook (4.4%). These data provide a clear indication of the most important locations where consumers learned about the site.

Almost half (49.6%) of the first presurvey respondents had already decided to visit the state and were visiting the Web site to help them plan their trip. Approximately 11.9% were considering visiting the state and wanted the Web site information to help them decide. Nearly two-thirds of respondents were using it for trip decision making. About 8.6% were visiting the Web site for nontourist reasons, such as children working on school projects, business, or relocation (see Table 6). Other reasons were cited by 26.0% of respondents. Findings suggest that most of the respondents were the intended target of the Web site sponsors.

Table 7 shows specific respondent behavior and actions taken while on the Web site or planned immediately afterward. More than 77% of respondents reported that they viewed five or more pages on the Web site, 50.8% downloaded or printed Web site information, 45.9% used a link on the Web site to go to another site, 45.0% ordered a state vacation guidebook online, 24.7% called a business found in the Web site, 18.8% wrote for additional information, and 12.2% forwarded (emailed) information on the Web site to a friend. Findings suggest that respondents were active while on the Web site, and more than half took immediate actions.

A series of questions asked respondents of the second presurvey about the usefulness of the Web site in providing nine categories of information and if they wanted that type of information. Table 8 shows that 100.0% of respondents were interested in information about activities and things to do, 97.0% in travel regions and cities, 89.7% in sightseeing opportunities, 85.7% in maps, 72.1 % in information on lodging, and 65.0% in information about making reservations. Results suggest that a wide range of information was wanted by respondents but also the importance of content on activities and things to do.

The postsurvey asked nonresident respondents if they traveled to the state on a trip that was at least partially for leisure purposes, after viewing the Web site. A total of 53.3% stated that they took at least one such trip (see Table 9). Respondents who were residents were asked if they took a trip to a new area in the state, which was at least partially for leisure pursuits, and/or which was more than 50 miles from their home since viewing the Web site. A total of 78.2% of residents took at least one such trip. Findings show that of those nonresidents who traveled to the state, 79.6% took a trip to an area they had not previously visited. More than 55.8% of residents who made a leisure trip visited a new location.

Residents who visited a new location in the state and nonresidents who went to California were asked if their decision to take that trip was made before or after viewing the state Web site. A total of 75.0% of nonresidents and 58.3% of residents reported that the decision was made before they accessed the state Web site. Of those who decided before viewing the Web site, 27.8% of nonresidents and 50.0% of residents recorded that they stayed longer than originally planned due to the Web site visit. These data suggest that even though the majority of respondents had already decided to visit the state, the Web site did influence their trip plans and spending.

The next questions asked nonresidents who visited the state and residents who went to a new area if they changed their travel plans as a result of viewing the state Web site. A total of 53.1% of nonresidents and 40.0% of residents reported that they did change their travel plans in California as a result of viewing the Web site (see Table 9).

A final series of questions queried residents who visited new areas and nonresidents who came to the state if they visited a facility or service that was advertised or featured in the state Web site, and if the Web site influenced their visit to this location. Table 10 shows that 85.4% of nonresidents and 68.0% of residents visited a featured destination and that 60.4% of nonresidents and 52.0% of residents replied that the Web site influenced this decision. The next greatest influence was on visiting an advertised event or attraction, wherein 39.6% of nonresidents and 40.0% of residents reported that the Web site influenced their visit. Results show that the Web site influenced many aspects of trip planning for both residents and nonresidents.

A final indicator of the influence that the state Web site has on travel behavior was a question asking all respondents about the Web site's potential impact on visiting California in the near future. Table 11 shows that 84.9% of residents reported that the Web site increased their likelihood of visiting new areas in the state, while 80.3% of nonresidents recorded that it would increase their likelihood of visiting California in the next 2 years. Besides showing an immediate Web site impact, results suggest that the Web site will continue to influence future visitation.

DISCUSSION

This research attempted to develop an efficient online method for evaluating tourism promotional Web sites and applied this method to a well-known state tourism Web site. Findings showed that the methodology used in this study is efficient, is relatively easy to use, and can provide a low-cost means of surveying. Results suggest that this method has the potential to revolutionize some types of survey research and allow businesses to more frequently contact clients. However, this study also clearly showed the challenges facing online surveying and the evaluation of tourism Web sites.

The largest problem encountered with this research method was the low 2.0% response rate. This was confounded further by a 37.0% response rate to the postphase survey. Such a low response certainly makes one suspect the validity of the sample and how well it accurately represents the actual population of visitors to the tourism Web site. However, offsetting to some degree the nonresponse problem is the relatively large number of responses that can be economically gathered with online methods.

Does the large sample size overcome the low response rate in this study? Findings of this research suggest that they do not. It is unlikely that survey respondents are representative of the average visitor to the Web site. This is based on the discrepancy between the percentage of respondents who ordered a state visitors guide (45.9%) and the Web site traffic statistics on the number of visitors to the state Web site who clicked on the brochure order form. To confirm this, the Web master for the state site was contacted and reported that 1.5% was the average visitor guide requests, as a percentage of user sessions, between fall 1998 and winter 1999 (Wantanabe, personal communications, January 1, 2000). Therefore, respondents were much more likely to request a brochure (45.9%) compared to the average site visitor (1.5%). They were probably much more likely to visit the state as well, based on research by Ellerbrock (1981) and Burke and Gitelson (1990), which showed that respondents who complete mail and telephone surveys are significantly more likely to visit the destination than those who do not respond. This suggests that it does not matter how large a sample size you acquire in an Internet-based survey if respondents differ significantly from nonrespondents.

There appears to be few easy ways to get information on Web survey nonrespondents in tracking studies to determine if they are different from respondents. The nature of the Internet does not allow one to gather much information about nonrespondents without their providing it. Web traffic software can only provide data such as domain name and type, region or country of request, and organization type. An alternative to check for nonresponse bias is comparing key sales, coupon offer, or brochure order click rates for respondents and nonrespondents. But this approach still gives no data on key nonrespondent demographic data, such as income, or reasons for visiting the site.

Results of this study collaborate with research by Schonland and Williams (1996), which suggest that incentives used to lure Web site viewers to an online survey can influence the type of respondents and subsequent responses. This affects reliability and validity. Different basic appeals for an incentive may be more successful than others. For example, a "please help" or a "contest" appeal may be more effective than "win." CalTour has changed to a "contest" button on their home page for access to their new evaluation survey, and this appears to have increased response over a "win" appeal used in this study. Study findings also suggest that even the best incentives have only limited ability to lure a majority of Web site visitors. So what else can be done to provide better information about Web site visitors?

Placing the entry link for the survey in a more visible place on the Web page could help viewers find the survey easier. This is a revision that CalTour made based on the findings of this research. Placement of a half page "pop-up" screen to a sample of viewers' asking them to fill out a survey could boost response.

Registration of visitors, wherein they must provide a minimum of data about themselves before they are able to proceed to the rest of the Web site, is one approach being used to gather demographic data. However, Web users are so concerned about privacy of their personal information that registration has been shown to discourage sites visitation (Neilson 1999).

The methods used in this study may be much more effective for types of research in which the email address of the Web site visitor is known, such as in conversion studies, questionnaires to existing clients, and product evaluations. Much of the challenge with the method used in this research was how to get a sample of all Web site viewers, not just those who inquired for information. If the goal of the investigation is more akin to a "tracking" study and learning about influences on all viewers, then considerable advancements in sampling are needed. But where the research purpose is strictly a conversion type of goal of determining how many inquirers converted to visitors, then an Internet-based approach, such as employed in this study and that used in Arkansas, may be effective.

Despite the limitations and nonresponse concerns of the Internet-based survey method applied in this study, it did provide important new insights into a unique, but not necessarily representative, subset of the population viewing the CalTour Web site. Respondents to the Web site survey were from a broad range of income and age groups, and a high percentage were international residents. The majority of respondents had already decided to visit California and wanted the Web site to assist in trip planning.

Findings clearly demonstrate that the Web site influenced respondent travel plans to California, for both residents and nonresidents. In addition to the Web site's immediate impact, it also increased the likelihood of nonresidents visiting California and of residents going to new areas of the state in the next 2 years.

Results also suggest that a state Web site can influence large numbers of residents to visit new destinations in the state. Tourism Web sites may be able to keep more residents in state and to distribute their spending to new, often rural areas.

FUTURE RESEARCH

Future research into tourism Web site evaluation should check for and identify the influence of nonresponse bias. Findings suggest that respondents and nonrespondents may be very different demographically and undertake different actions as a result of visiting the site. Findings also suggest that large sample sizes do not correct for differences between respondents and nonrespondents. This should also be confirmed.

An important area for future research is the identification of the impact of different incentives on response rates and respondent characteristics. New incentives, placement of incentives, or new technological innovations are required to improve response and research validity.

A topic that may be useful for future Web site assessment is an evaluation of the attributes of the site itself for ease of navigation, finding, and requesting information. Such data could help improve design aspects of the site.

Finally, it is essential to verify Web site performance standards. What is the normal range for percentage of respondents who view a site but do not request information online? What is the typical percentage of viewers who request information online or who book online?

[Reference]
REFERENCES

[Reference]
Association of National Advertisers (1999). "Third Annual ANA Web Site Study Indicates Emphasis on Web Marketing: Increases in Online Advertising Expected." Press Release [Online]. Available http://www.ana.net (May 4)
Ballman, G., J. Burke, D. Korte, and U. Blank (1982). "Real Conversion Rates of Regional Advertising Programs: Working from Gross to Net Rates." Proceedings of the Thirteenth Annual Conference of the Travel Tourism Research Association, Miami, FL.
Burke, J., and R. Gitelson (1990). "Conversion Studies: Assumptions, Applications, Accuracy and Abuse." Journal of Travel Research, 28 (3): 46-51.
Burke, J., and L. Lindblom (1989). "Strategies for Evaluating Direct Response Tourism Marketing." Journal of Travel Research, 28 (2): 33-7. Davidson, T. L., and W. B. Weithaupt (1989). "Accountability Marketing
Research: An Increasingly Vital Tool for Travel Marketers." Journal of Travel Research, 27 (4): 42-4.
Ellerbrock, M. (1981). "Improving Conversion Studies: A Comment." Journal of Travel Research, 27 (1): 35-40.
Forrester Research (1999). "Web Site Success Transcends Hits and Page Views, Predicts Forrester Research." Press Release [Online]. Available: http://www.forrester.com

[Reference]
Jupiter Communications (2000). "Online Holiday Sales Hit $7 Billion, Consumer Satisfaction Rising." Press Release [Online]. Available: http://www.jup.com (January 14)
McLemore, C., and N. Mitchell (2000). "The 1999 Arkansas Internet Conversion Study." Newsletter of the Travel and Tourism Research Association, Spring, 4-5.
Neilson, J. (1999). "Web Research: Believe the Data." Alertbox [Online]. Available: http://www.useit.com (July 11)
Schonland, A., and P. W. Williams (1996). "Using the Internet for Travel and Tourism Survey Research: Experiences from the Net Traveler Survey." Journal of Travel Research, 35 (2): 81-87.
Tierney, P. T. (1992). "Developing and Testing of a Model and Method for Assessing Effectiveness of Tourism Promotional Literature." Visions in Leisure and Business, 10 (3): 4-14.
Tourism Industries of America (1997). "Travelers' Use of the Internet; 1997." Executive Report, October, Washington, DC.
(1999). "TIA Report Says Online Travel Revenue Triples in 1997: Sales to Increase 440% to $4.7 Billion in Year 2000." Press Release, February 28, Washington, DC.
Van Hoof, H. B., and J. S. Hobson (1997). "Tourism Research Inquires: The Response of the State Office of Tourism." Journal of Travel Research, 35 (4): 77-81.
WebTrends (2000). "Definition of Terms" [Online]. Available: http://www.webtrends.com (January 14)
Wynegar, D. (1989). "U.S. Department of Commerce Task Force on Accountability Research." Journal of Travel Research, 27 (4): 41.

[Author note]
Patrick Tierney is a professor in the Department of Recreation and Leisure Studies at San Francisco State University. He thanks Eileen Hook for her assistance in preparing this article.