Travel Planning: Searching for and Booking Airline Seats on the Internet

Abstract

This article is based on a survey of 374 airline passengers sampled at an airport. We examine how travelers search for and then buy airline seats, determine which distribution channel dominates, calculate what percentages of online searches convert to actual bookings on the same website, and report why people switch. We also examine which travelers book, and then rebook airline seats, and why. We then measure the differences in how travelers view several Internet and website issues, the differences between Internet users and nonusers, between business and pleasure travelers, and between men and women. We update the most important reasons for airline choice, in an era when online bookings predominate. Finally, we report our findings (some counterintuitive) and their implications for airlines and third-party websites (online travel agents).

In 2009 domestic scheduled airlines in the United States carried about 600 million passengers on nearly 9 million flights, generating about 540 billion passenger miles (Bureau of Transport Statistics 2010). Traditionally, airlines sold their tickets via phone or through airline travel agents. But in 1994, the Web had arrived with the promise of becoming a mainstream medium of communications, and a distribution channel for travel services. At first, progress was slow, and in 1998, the Internet accounted for less than 1 percent of airline ticket sales (Brunger and Perelli 2008). However, in March 2002, Delta Airlines, suffering from the effects of the September 11, 2001, attacks, announced that it would eliminate the 5 percent travel-agent commission. The commission curtail was soon followed by the other airlines (travel agents currently typically charge travelers a $30 booking fee). This act of disintermediation became the main impetus for the growth of online reservations, bypassing travel agents. Soon, Werthner and Ricci (2005) reported that tourism had become the number one industry in terms of online transaction volume. Ramsey (2007) estimated that 61 percent of adult users of the Internet do travel research on the Web, while Mamaghani (2009) reported that 95 percent of users have done travel research on the Web. Today, approximately 6.5 percent of Web enquiries are travel-related searching (Jansen, Ciamacca, and Spink 2009). Thus, currently, most airline tickets are bought directly online on airline websites, through third-party websites (TPWs) such as Expedia, on auction websites such as Priceline, and on generic Internet search engines such as Google.

A number of reasons contribute to the increased use of the Web for airline bookings and the diminished role of travel agents for such purposes. First, the Internet is suitable for the purchase of intangible goods such as services in general and airline seats in particular. These intangible goods are considered “confidence goods” where purchase and consumption are separated by time and distance, and as such they can be described in detail for informed purchase. Second, customers expect that products purchased through the Internet will generally be cheaper (O’Connor 2003) because of lower distribution costs. In fact, Brunger (2009) found that the Internet had indeed reduced airline fares, mainly because of more efficient searches. Third, the Internet permits quick searches, decreased search costs (Sahay 2007), and provides a range of choices (Zhang et al. 2006). Fourth, TPWs permit one-stop shopping where airline seats, hotel rooms, and rental cars can all be purchased at the same time and place. Fifth, unlike hotels, airlines allow people who book on TPWs to collect loyalty rewards.

Having outlined the growing importance of the Internet for searching and buying airline seats, the rest of the paper will proceed as follows. We first state the primary objectives of our study, followed by our survey design. Next, we will report our survey results, and then discuss the implications of our findings. Finally, we will conclude by stating the limitations of our study, and then make recommendations for further research.

Study Objectives

There are six primary objectives of this study. First, we want to examine how travelers search for and then buy airline seats, specifically their preferred methods (Internet, telephone, travel agents, or walking up to a ticket counter). We also want to know which distribution channel dominates in the United States among options of airline websites, non-opaque TPWs, opaque auction sites, and generic Internet search engines. Second, we want to discover what percentage of airline searches on one website actually convert to bookings on that website, and how often travelers search on one website and then switch to another, and why. Third, we want to find out what kind of travelers book and then rebook airline seats, mostly for lower rates. Fourth, we want to measure the differences in how travelers view several Internet and website issues, the differences between Internet users and nonusers, between business and pleasure travelers, and between men and women. Fifth, the most important factors affecting airline choice had been reported more than twenty years ago (Toh and Hu 1988), but we wanted an update, with recent Internet and Web factors included. Sixth, and finally, we want to know what factors affect travel frequency today.

Survey Design

To achieve the study objectives above, we designed a preliminary survey of airline passengers based on our prior research experience in the field, and interviews with two airline officials and one executive of a TPW. This survey was pretested among seventeen graduate students to uncover areas of confusion or misunderstanding. A final survey instrument consisting of twenty-nine questions was then constructed and distributed at Seattle-Tacoma International Airport (SeaTac). Compared to a similar survey done twenty years ago also at Sea-Tac (Rivers, Toh, and Alaoui 1991), gaining access to the passengers for the current survey was considerably more difficult. In the post-9/11 era, we had to get permission first from the Port of Seattle, and then from the Transportation Security Administration (TSA), to comply with strict security policies. Then we had to get permission from the airlines to distribute and collect the surveys at their departure gates, which they control. Our research assistant was given a Port of Seattle security vest with a security badge, and was given strict instructions from us to stick to a carefully scripted survey protocol that would lead to a representative sample. Note that he was able to collect the surveys only on those days and at departure gates for which he had permission from the port, TSA, and the airlines. Given all the procedural limitations, we adopted an opportunistic approach to get the largest possible sample size to minimize sampling error, maximize the reliability of estimates, and increase the power of statistical tests.

The surveys were conducted in June, July, and early August of 2010, only on those days when there was double permission from the Port of Seattle and the TSA to get into the secure areas, and permission from the airlines to distribute the surveys at their departure gates. Our research assistant collected the surveys over the seven days of the week (Monday through Sunday) to minimize the effects of daily variations in airline traffic and passenger profiles, at different times of the day (morning, afternoon, evening, and night), and from as many airlines flying on domestic routes within the United States as possible (Alaska, American, Continental, Delta, Frontier, Hawaii, Horizon, US Air, and United). Given that our study encompasses travelers’ use of airline websites as well as TPWs, he avoided Southwest Airlines and Virgin America, for both do not allow TPWs to sell their tickets. Their tickets can be purchased only through their own airline websites. He approached individuals traveling alone as well as families of all races and age groups at the departure gates of as many airlines as possible. For practical reasons, he avoided approaching adults supervising young children or caring for babies.

A six-page professionally printed survey containing twenty-three questions for our study, plus another six courtesy questions for the Port of Seattle was distributed. Altogether, we managed to get 374 useable survey responses over a three-month period. Between 85 percent and 90 percent of those approached agreed to fill out the survey. Thus, nonresponse bias is not deemed to be an issue. While there were some nonresponse items, they are not rampant, except for questions about high-tech services provided by the Port of Seattle. We coded and entered the data onto Excel spreadsheets and used SPSS 1.70 to run appropriate statistical tests and analyses. The results and our findings are reported below.

Survey Results and Findings

We had 374 usable survey responses from the departing passengers (51% male/49% female). Sixty-five percent of the respondents had college degrees, with an average age of forty-one years old, and an average annual household income of $91,000. Among the respondents, 29 percent declared themselves primarily as business travelers and 71 percent as primarily pleasure travelers. This distribution is almost similar to a previous survey of hotel guests (Toh, DeKay, and Raven 2008). Note that we allowed the respondents to identify their own travel status, because behavior and attitudes are more related to self-perception than an arbitrary definition, such as the number of trips they made a year for business purposes. Together, they made an average of seven round-trips per year, and 56 percent of the respondents were frequent flier members, belonging to an average of two programs. Thus, our convenience sample appears to be balanced and representative of the general population of travelers. The sample is not dominated by “road warriors”—defined as people who travel frequently, often at full fares in the premium classes—who make up two-thirds of the passenger revenues earned by our domestic airlines (Toh, Fleenor, and Arnesen 1993).

The most remarkable change in the way travelers search for and book airline seats in the last fifteen years or so is the widespread use of the Internet. Results show that 82 percent of the respondents had used the Internet to either search for or book their airline seats for the flight they were taking at the time of the survey. For their particular flights, there were 833 Internet searches, 68 percent of which occurred on TPWs, also called online travel agents, listed in order of popularity as the following: Expedia, Travelocity, Orbitz, Kayak, Cheapsites, Sidestep, Yahoo!, Allegiant, and Travel Axe. The other 15 percent occurred on websites of airlines on which they flew, 8 percent on generic Internet search engines (Google and Bing in that order of popularity), 5 percent on auction websites (Hotwire and Priceline in that order of popularity), and the remaining 4 percent on other airlines’ websites. Our results are not surprising, as found in a study by Morosan and Jeong (2008) that people have a favorable attitude toward TPWs. Some reasons for the popularity of TPWs, notably Expedia, are that they provide price as well as package bundling (Carroll, Kwoknik, and Rose 2007), and make it easier for buyers to compare airlines and fares 24/7 (Cho and Agrusa 2006).

Among those respondents who had used multiple sites, when asked which one they spent the most time searching, they reported, in descending order: Expedia (37%), Travelocity (18%), Orbitz (17%), websites of airlines on which they flew (9%), Kayak (5%), auction sites (3%), Cheap Flights (2%), and less than 2% for all the others combined. Thus, we can see that TPWs are the most popular search sites, followed by websites of airlines on which they flew, while auction sites and generic search engines are much less popular in airline searches. When asked where they booked their airline seats, they reported in descending order: Expedia (33%), their own airline’s website (27%), Travelocity (17%), Orbitz (16%), while only three people booked on auction sites. Thus again, the TPWs dominate.

When asked which channel they used most often to search for a flight, the results were, in descending order: Internet (83%), travel agent and corporate travel planner (14%), telephone (3%), and walk-in counter (less than 1%). When asked which method they used most often to book, the results were somewhat similar, in descending order: Internet (79%), travel agent and corporate travel planner (16%), telephone (3%), and walk-in counter (2%). Note that only a few (2%) who had searched on the Internet had switched to brick-and-mortar travel agents. Although these traditional travel agents play a decreasing role in domestic airlines ticket sales, they still book a large number of international flights (Pearce and Schott 2005) and trips for the cruise industry (Toh, Rivers, and Ling 2005).

Also note that there were 833 reported sites searched on the Internet, but only 264 reported bookings on the Internet, a ratio of 3.2 site searches per booking, indicating that there was a lot of search-then-switch activity. For their flights, 231 out of 264 who booked on the Internet (88%) admitted that they had searched on one or more websites, but switched to another when booking. Notably, our results also show that only 13 percent of those who had searched on airlines’ sites switched to TPWs when they were booking, and only 10 percent of those who had searched on TPWs switched to airlines’ sites when booking. Furthermore, results of a chi-square test for independence was significant (p < 0.0001), indicating that within the four types of search engines (airlines, TPWs, auction sites, and generic sites), travelers who switched tended to remain within the same type of website when they were booking. Also, 95 percent of those who had searched on the Internet also booked on the Internet. When those who had switched were asked why they had switched, the reasons given, in descending order, were: found lower fare (66%), registered on that website (17%), convenience (10%), avoid fees (3%), more secure site (2%), bonus miles (1%), and others (1%). Among those who had changed their reservations, the reasons given, in descending order, were: lower fare (75%), more convenience (13%), registered on the website (12%), and others (0%). Thus again, lower fares are the most compelling reason to change flight reservations.

We next compared the demographic characteristics between those who book on the Internet and those who do not. Among those who used the Internet most often to either search or book their flights, 20 percent declared themselves as primarily business travelers and 80 percent as primarily pleasure travelers. Because among all travelers 29 percent had declared themselves primarily as business travelers and 71 percent primarily as pleasure travelers, this result suggests that pleasure travelers are more likely to use the Internet than business travelers. In fact, whereas only 56 percent of business travelers used the Internet, the corresponding figure for pleasure travelers was 93 percent, a significant difference (p < 0.0001). This result is consistent with the fact that pleasure travelers pay for their own seats and the perception that the Internet provides the widest selection of flights and fares (Brunger 2009). There was no significant difference between the average ages of those who used and did not use the Internet, albeit the average annual household income of the former was $89,000 versus $100,000 for the latter, a marginally significant directional difference (p = 0.07). The difference in annual household incomes of the two groups further attests to the attraction of the Internet as a source of cheap fares for those to whom airline travel is less affordable.

The average degree of agreement score (from a low of 1 to a high of 5) with the statement that the Internet allows for quick flight and fare comparisons was 4.73 versus 4.29 for those who did and did not use the Internet, respectively, a significant directional difference (p = 0.001). With respect to the statement that the Internet offers the lowest air fares, the respective average scores were 4.29 versus 3.86 (p = 0.02). As to the importance of the travel agent or corporate travel planner, among those who used the Internet the average score was 2.80 versus 3.32 for those who did not, again a significant directional difference (p = 0.001). Finally, among those who made their own reservations, 94 percent used the Internet, while among those whose reservations were made by others, the figure was only 54 percent, a significant directional difference (p < 0.0001). Note that all four results are significant in the intuitively expected direction.

A counterintuitive finding was that among those who used the Internet, the average number of round trips made over the last twelve months was 6.5, compared to 10.7 made by those who did not use the Internet, and the difference was significant (p = 0). One would argue that those who travel often have more incentive to learn how to make online bookings to get cheaper flights. A possible explanation to this seemingly counterintuitive finding is that those who travel often are typically business travelers whose corporations make the reservations. In fact, among those who declared themselves as business travelers, 36 percent said they used a travel agent or corporate travel planner, compared to only 4 percent of those who declared themselves as pleasure travelers.

More interestingly, while gender division among those who used the Internet was nearly equal (49% male/51% female), of those who did not use the Internet, 62 percent were male. Considering that our sample was almost a 50/50 split by gender, the gender directional difference among those who did not use the Internet is impressive. Measuring directly, among females 86 percent used the Internet, while the corresponding figure among males was only 78 percent, a significant directional difference (p = 0.04). This difference is partly due to the fact that percentage-wise fewer women are business travelers compared to men. Among women, 14 percent returned to the Internet to find a lower fare, whereas only 12 percent of males reported doing so. In other words, the result suggests that women are more aggressive price shoppers than men when doing travel research on the Web. Among those who used auction sites, the gender split was 52/48 percent, again favoring women. Measuring directly, we found that among females, 8 percent used auction sites, whereas among males, the corresponding figure was 7 percent, an insignificant difference, but in the intuitive direction. Even more interesting is the fact that 66 percent of the women switched when searching for a flight and subsequently booking for the flight, but the corresponding figure for men was only 55 percent, which was directionally significant (p = 0.02). All of these figures point to the fact that women are more frequent and intensive users of the Internet, and are savvier at finding lower fares. In fact, when asked for the reason for switching, the majority of women (62%) mentioned lower fares.

We now report the differences between those who use the Internet for searching/booking travel and those who do not, and between respondents who considered themselves as business and pleasure travelers. Results of the comparative analysis are reported in table 1 with scores in a Likert-type scale, ranging from 1 (strongly disagree) to 5 (strongly agree). Likert-type scale is the most frequently used scale in survey questionnaire research (Cook, Hepworth, and Warr 1981), and is considered the most useful in behavioral research (Kerlinger 1986).

Table 1.

Significant Directional Difference in Degree of Agreement toward Beliefs in Internet Usage in Travel Planning (Increasing scale of agreement from 1 to 5)

We next report the differences between those who use the Internet for traveling and those who do not, between respondents who considered themselves as business and pleasure travelers, and between males and females. Results of the comparative analysis are reported in table 2 with scores, also in a Likert-type scale, ranging from 1 (strongly disagree) to 5 (strongly agree). Results of our investigation into traveler beliefs are reported in table 3.

Table 2.

Significant Directional Difference in Degree of Importance Placed on Factors Considered in Choosing Airlines and Flights (Increasing scale of agreement from 1 to 5)

Table 3.

Measuring Beliefs (Increasing scale of agreement from 1 to 5)

Table 4.

Measuring Importance of Factors Used to Choose Airlines and Flights (Increasing scale from 1 to 5)

A few significant correlations among traveler behaviors are worth noting. Frequently checking for lower fares later and rebooking if necessary is correlated with the behavior of contacting the travel agent to see if lower fares are available (r = 0.55, p < 0.0001). Likewise, frequently checking for lower fares later and rebooking if necessary is also correlated with the behavior of waiting and checking for lower fares over time (r = 0.46, p < 0.0001), and searching on auction sites (r = 0.43, p < 0.0001). In other words, those who are fare sensitive exhibit this characteristic consistently. Conversely, those who tended to book early to get lower airfares also booked early to get good flights (r = 0.74, p < 0.0001). Results of our investigation into factors used to choose airlines and flights are reported in table 4.

Noteworthy correlations among travelers in this regard are the fact that those who placed a lot of importance on service quality also placed a lot of importance on convenient departure and arrival times (r = 0.40, p < 0.0001), on seat selection (r = 0.50, p < 0.0001), and on the quality of the website (r = 0.44, p < 0.0001). Also, those who placed a great deal of importance on their flights departing and arriving on time also placed great importance on nonstop flights. Thus, the quest for comfort and quality is manifested consistently.

Let us look at the factors that affect the frequency of travel. The correlation matrix shows that the number of round-trip flights taken annually is consistently and positively correlated with age (r = 0.03), education (r = 0.15), and income (r = 0.32). It appears that only income is a reasonably good predictor of the frequency of travel. We then conducted independent sample t-tests on the means of the number of round-trips taken in a year. The average for males was 8.14 and for females 6.19, a significant difference (p = 0.006). The corresponding figures for those who do not use the Internet and those who do were 10.59 versus 6.30, respectively, again a significant difference (p < 0.0001). However, the biggest difference was found between business and pleasure travelers at 12.94 and 4.81, respectively, a significant difference (p < 0.0001). Why did all of these occur? Upon further inquiry, we found that among males the percentage of business travelers was 42 percent, whereas for females it was only 18 percent. Among those who do not use the Internet the percentage of business travelers was 73 percent, whereas for those who do use the Internet, the percentage of business travelers was only 20 percent. Thus, the best predictor of air travel frequency, relatively to all other observations, is the type of traveler. In this survey, business travelers who are mostly male and do not use the Internet fly much more often than pleasure travelers.

Implications

Our survey results show that 82 percent of airline travelers use the Internet to either search for or book their airline seats. Among the searches, 68 percent occurred on TPWs and only 19 percent on airlines’ websites, with generic search engines and auction sites being unimportant. More interestingly, 88 percent of our respondents said that they searched on one or several websites and then switched to another when booking. However, switching activities occurred mainly within the same types of travel search site, and mainly for reasons of price. Furthermore, 95 percent of those who searched on the Internet also booked on the Internet.

In relating the foregoing results into managerial implications, we underscore two important notions. First, whereas the commissions paid by hotels to the TPWs in the hotel industry range between 15 and 30 percent, those in the airline industry are well below 5 percent, and can be null in many instances. This difference in commissions can be attributed to the different industry structure in that the hotel industry is said to be “fragmented,” while the airline industry is a well-established oligopoly. Second, TPWs accept very low commissions from the airlines because an airline seat is considered to be the “bottle of milk at the back of the store,” in that it must be sold before travelers will buy hotel rooms and rental cars in a price or product package.

Airlines want to sell through their own websites because they want to build customer loyalty through their frequent-flier programs. In addition, airlines want to manage their customers through the entire flight process by giving them reminders and email alerts of schedule changes. They also want to get rid of excess inventory by sending email alerts to their frequent-flier members on special deals in order to avoid lowering their fares on the Internet and cheapening their brand. On the other hand, airlines also realize that the bulk of airline tickets are sold through TPWs and must, therefore, work with them. Commissions must be reasonable, because TPWs provide a very important service of one-stop shopping. TPWs want the flying public to buy through them not because of the meager commissions they earn on the sale of airline tickets, but because of the increased possibility of selling higher-margin items such as hotel rooms. How can both the airlines and TPWs get travelers to buy from their own websites?

One possibility for the airlines is to offer bonus frequent-flier miles when booking through the airlines’ own websites and to take care of their customers via email from the beginning to the end (Alaska Airlines excels in this practice). TPWs, on the other hand, should not charge for bookings (a practice that ended recently). They should offer best-rate guarantees by insisting on fare parity (same price on all websites by individual airlines), and make their websites more attractive by offering price bundling (one-stop shopping) and creative product bundling (discounted integrated packages). They should also inform the traveling public that, whereas they cannot offer hotel loyalty points, airline seats sold through TPWs are eligible for frequent-flier miles.

We also discovered that auction websites are a dismal failure in the airline industry. In our survey, only 5 percent of the searches occurred on auction sites, with only three seats sold (less than 1%), compared to 13 percent of all hotel rooms sold on auction sites (Toh, DeKay, and Raven 2011). Note that on hotel auction sites, one can specify the approximate location of the hotel as well as the number of stars to ensure quality. In contrast, in the case of airline auctions, one is leery of leaving at midnight from Seattle to Miami with long stopovers in Denver and Atlanta. Thus, online channels are better off developing auction sites for hotels than for airlines. Generic search engines such as Google and Bing are not major players currently, but may become popular in the future. Results from the Port of Seattle survey suggest that airlines and TPWs must pay attention to the growing practice of travelers printing their boarding passes online (31%) and the willingness of passengers to use their mobile phones as boarding passes (71%).

In regard to the demographic characteristics of air travelers, websites should be designed to be more attractive, especially to the more educated but less wealthy people. Those who travel most often are business people, 36 percent of whom rely on their corporate travel planners or travel agents for seat reservations. Thus, both the airlines and TPWs cannot totally ignore corporate travel planners and travel agents, who together account for 16 percent of domestic airline ticket sales (higher in the case of international air travel).

Results reported in table 1 suggest that Internet users, compared to their counterparts, tend to consider the Internet as the most convenient for comparing flights and fares and tend to agree that it offers the lowest fares. This result suggests that TPWs should continue their practice of listing flights in ascending order of price, and not allow vendors to get higher page positions by paying for them (a common practice in selling hotel rooms). Internet users are also wise not to book late to get lower airfares. A separate study (Raven, Toh, and DeKay, 2011), involving time-series observations of fares on TPWs in the spring of 2010, conclusively shows that airfares tend to almost uniformly rise over time. The more fare-sensitive pleasure travelers, compared to business travelers, tend to check on the Internet, then contact their travel agent for lower fares, wait and check for lower fares over time, and check for lower fares and rebook if necessary.

An examination of table 2 suggests that business travelers, compared to pleasure travelers, place greater importance on seat selection, frequent-flier miles, recommendation of the travel agent or corporate travel agent, nonstop flights, and on-time performance. Business travelers fly more often, purchase more expensive seats, and are more reliant on travel planners and agents. The implication for the airlines is that the fare-insensitive business travelers must be reached through superior service rather than lower fares. Superior service can be achieved by offering priority seat selection, more frequent-flier miles, priority boarding, and VIP lounges for business class and even full fare passengers.

Examining table 3 on travel behavior, it appears that using the Internet to compare flights and fares, checking multiple sites to get lower fares, and using the Internet to get lower rates are considered most important for Internet users. Results reported in table 4 also indicate that lower fees and fares and service factors are considered the most important factors in airline selection. From table 4 it appears that there is little importance placed on preference for a particular airline, because frequent-flier programs have been shown to be less important in the choice of an airline than service factors and price (Toh and Hu 1988). All of this behavior relates to the fact that in the airline industry, price is paramount, and that for most travelers an airline seat is considered a commodity. The implication for the airlines in this respect is that yield management (Toh and Raven 2003) is very important. Airline seats are perishable and excess capacity exists except during peak holiday periods. Hence, coupled with the low variable cost of carriage (meals are no longer provided), fares must be low enough to fill otherwise empty seats with discretionary pleasure travelers, without unnecessarily sacrificing yield. For the TPWs, they must recruit as many airlines as possible, even though the commissions are very small or nil. This is because although there is supposed to be general fare parity, Raven, Toh, and DeKay (2011) have shown that convergence of the lowest fare available occurs only 18 percent of the time among TPWs. This figure suggests that 82 percent of the time, the lowest fare is dictated by the non-uniform presence of a deep discount carrier. The operational imperative for the TPWs is clear—recruit as many airlines as possible, and put the lowest-cost carrier on top of the search screen.

Our next observation is perhaps the most interesting. It is clear from our results that women, compared to men, are not only more frequent and savvy users of the Internet in travel research, but also more aggressive price seekers. Specifically, women are more likely to use the Internet, more likely to switch, more likely to return to the Internet to find a lower fare, and slightly more likely to use auction sites. Women also place greater importance on getting lower airline fares and fees, and place less importance on preference for a particular airline. Why did these confounding results occur? Bimber (2000) contended that women are less likely to use the Internet, and Ono and Zavodny (2003) report that they use the Web less often. The Pew Internet and American Life Project (2004) reported that 78 percent of men like to do business online, compared to only 71 percent for women. Despite these contradictory results, we do not find our findings surprising. In a recent piece on gender, Kim, Lehto, and Morrison (2007) showed that in the case of travel planning, women ascribed higher perceived importance to travel websites, searched more on websites, and do so more frequently. In a separate parallel study based on a survey of 249 hotel guests (Toh, DeKay, and Raven 2011), we were also able to confirm their counterintuitive findings.

The implications are that airlines and TPWs should pay greater attention to women in travel website design. A recent study by Zhang et al. (2009) suggests that people are relying heavily on recommendations for trip planning, so perhaps TPWs should sponsor links to reviews of hotels by women for women, with a lot of visual aids and safety features. After all, TPWs earn the bulk of their revenues from the sale of hotel rooms, not airline seats.

Our survey results show that only 13 percent of airline travelers return to websites to find a lower fare, suggesting that once a reservation is made, travelers are largely content with their choices and will stop searching. It follows that airlines are free to lower their fares to get rid of distressed seat inventory as the day of departure approaches, if they so wish. As we have noted, airlines prefer to do this via email alerts to their loyal customers. This strategy is made more compelling by the observation in table 3 that travelers did not agree with the wisdom of booking late to get the lowest fares.

Finally, we note in table 4 that service quality (convenient departure and arrival times, on-time performance, and nonstop flights) is paramount in the minds of travelers. All these service factors are within the airlines’ control, and information on these factors should, therefore, be made available to airline passengers, whenever possible. The recommendation of the travel agent and corporate travel planners was rated as the least important consideration. Thus, airlines should promote directly to the flying public via mass advertisements, and target their frequent-flier program members. However, airlines should have strong relations with corporate travel planners to capture the lucrative business travel market. Less attention could be paid to brick-and-mortar travel agents, who are on the decline, except for international air travel and cruises.

Limitations and Recommendations for Further Research

There are several limitations of our study. First, our survey protocol of approaching passengers about to embark on a flight led to a convenience survey by necessity. However, we are not aware of a probabilistic sample done on a study of this nature, because it would require having a list of the population (only a theoretical possibility) and the power to enforce compliance (even the Census Bureau cannot achieve this). We took all the precautions to get a representative sample by surveying all seven days of the week at different times of the day, by surveying passengers from as many airlines as possible, and by targeting both sexes and all racial and age groups. The demographic breakdown of our sample indicates that the sample consists of a reasonably diverse group of air travelers. We acknowledge, however, that because we approached people in the act of traveling, frequent fliers are likely to be overrepresented in the sample, compared to the population of fliers.

Second, because of the length of the survey (twenty-nine questions), we realize that there will inevitably be nonresponse items. In this respect, we expect that most of nonresponses will pertain to the difficult questions on hypothetical high-tech boarding and baggage issues posted by the Port of Seattle. An example of such questions is “Would you be willing to use your mobile telephone as a boarding pass if this service was offered by your airline?” Conceding that difficult questions can lead to nonresponse items, which in turn can lead to abandonment (Lee, Hu, and Toh 2004), we took the precaution of attaching the six high-tech questions at the end of our twenty-nine-item questionnaire. Structuring the questionnaire as such also has an additional advantage in that should participation fatigue occurs (Toh and Hu 1996), it is less likely to affect the earlier part of the survey that pertains primarily to our study on Internet usage.

Third, we acknowledge that all survey results are idiosyncratic to the data collected, the survey results themselves can be distorted (see Lee, Hu, and Toh 2000), and accuracy can depend on travel frequency (Hu, Toh, and Lee 2000). This limitation raises the question as to whether data collected in Seattle can be generalized toward the whole country. Fortunately, Seattle is a very racially diverse medium-sized cosmopolitan city with heavy industries (such as Boeing), light industries (such as Microsoft), and service industries (such as the Port of Seattle). It is also a thriving tourism destination with a growing cruise center, and is a gateway to Asia by ship and air. Given its diversity and representativeness, Seattle is often used as a test market city for new products and advertising campaigns. Moreover, it can be reasonably assumed that about half of the 374 airline passengers who responded to our survey were out-of-towners returning home from a visit to Seattle. Thus, our sample is actually quite geographically dispersed.

However, there is still the unresolved issue of the surveys being administered only in the summer of 2010. Recall that the surveys could be collected only on those days when the Port of Seattle and the TSA gave us permission to operate within the secured areas of the airport, and when the airlines also gave us permission to distribute and collect the questionnaires at their departure gates. It took several months to get permission from the Port of Seattle, and not all airlines responded in a timely fashion to our requests. Thus, we collected our surveys as quickly as possible when everything lined up and, even so, it took us nearly three months. We accordingly acknowledge that there could be some seasonality bias in our results. However, this bias is mitigated by the fact that whereas the frequency of travel is affected by the season, searching and booking behavior are largely invariant to seasons.

In summary, in spite of the previously discussed limitations, the study findings were drawn from a reasonably balanced and representative sample of respectable size. We are grateful to the Port of Seattle, the TSA, and the cooperating airlines, as well as our obliging respondents, for allowing us to collect data from real travelers in a real-life setting, just prior to airplane embarkation. Our recommendation for further research is a replicated study that is conducted during other seasons of the year, and involves different airlines.