Facet-Based Analysis of Vacation Planning Processes

Abstract

This article documents the design and results of a study on vacation planning processes with a particular focus on aggregate relationships between the probability that a certain facet of the vacation decision has been decided at a particular point in time, as a function of lead time to the actual trip, life cycle characteristics, income, travel experience, and any other facet already being decided. A binary mixed logit panel model was formulated and estimated to examine the assumed relationships. The Dutch Continuous Vacation panel was used to collect the data. Results indicate that the closer to the actual date of the trip, the higher the probability of a facet being planned. Moreover, vacation facets are planned at different points in time. We also found differences in the level of planning for different life cycle groups, levels of income, and travel experience. A discussion of the limitations of the study and possible future research directions completes the article.

Keywords

vacation planning process life cycle stages binary mixed logit model panel data

The vacation planning process involves multiple facets, which are developed over time: destination choice, choice of transport mode, timing and duration choice, choice of travel party, accommodation, and choice of activities (e.g., Dellaert, Ettema, and Lindh 1998). In principle, each of these facets may trigger a vacation planning process. The desire to use a cruise ship or to fly the A380 may be the start of a search process that allows a traveler to realize such desires, and then other facets are decided later on. Similarly, the nature of activities may limit the possible destinations and thus may be a natural start of the decision-making process. The wish to stay at a particular hotel may imply a specific destination. In most cases, however, the first facet considered will be the destination (Um and Crompton 1990; 1992; Crompton 1992; Fesenmaier and Jeng 2000; Hyde 2004), which may be a particular city or region or even country or continent. In other cases, the destination of the vacation does not matter too much, as long as the vacation type (coastal destination, city trip, winter sports, etc.) fulfills one’s vacation needs.

Any temporary choice made for a particular facet will in general reduce the feasible or realistic choice options for other choice facets. For example, travel by bus or car from Europe to the United States is infeasible. Although it is possible to use the car from London to Beijing, that option for transport mode choice does not seem very realistic for most travelers. Thus, choices made with regard to a particular choice facet either rule out options related to other facets or influence the probabilities that certain subsequent choices related to other facets will be made. In the Netherlands, with its dense train and air networks, basically all transport mode options are likely to be available at a short distance. With increasing distance, the slower modes will consequently be faded out.

The facets of the vacation planning process are likely to be influenced by the sociodemographic characteristics of the travelers. Changes in these characteristics over time may influence the vacation planning process. For instance, Davison and Ryley (2009) found that key life stages, such as having children and entering retirement, influence air travel behavior.

The purpose of the present study therefore is to contribute to the knowledge about the vacation planning process by assuming that different choice facets are planned at different times. Moreover, it aims to contribute to the literature about the role of the family life cycle in terms of its influence on the vacation decision-making process. More specifically, this article documents the design and main results of a research project that sought to understand temporal aspects of the vacation planning process. Therefore, the research questions of the present study are the following:

Is there a difference between choice facets in terms of when they are planned?

What are the effects of life cycle stages, income, and travel experience on vacation planning?

Literature Review

Vacation Planning Process

Understanding the vacation planning process is of paramount importance to travel marketing and management. Vacation choices are recognized as multifaceted and interrelated decisions, which are developed over time (Um and Crompton 1992; Woodside and MacDonald 1994; Dellaert, Ettema, and Lindh 1998). Thus, the facets of vacation planning (destination, travel parties, accommodation, etc.) are interrelated as a network. Therefore, they also constrain each other and differ given individual preferences (Fesenmaier and Jeng 2000; Jeng and Fesenmaier 2002; Hyde 2004).

There are many studies exploring the vacation decision-making process from different perspectives, such as marketing and psychology. For instance, Hyde (2008) developed a model comprising three different activities of prevacation decision making: information search, vacation plans, and vacation bookings. In that context, vacation plans were related to the planning of specific locations and activities to be performed once a destination is chosen as a function of the characteristics of both vacation and vacationer and varied with the amount of information searched. Similarly, all preparatory activities preceding the choice of a destination were included in the vacation planning, as shown in the study by Decrop and Snelders (2004).

Some studies on vacation planning have focused on information use during the various phases of the process (Fodness and Murray 1998; Jeng and Fesenmaier 2002; Choi et al. 2012). For example, Choi et al. (2012) found that tourists sought information more intensively before the actual purchase than during the other three phases, following the authors’ definition of vacation planning (at the time of purchase, after purchase, and after arriving at the destination). More specifically, they found that departure date, travel budget, length of the trip, and travel mode were mostly decided before the purchase, whereas in case of accommodation, 40.3% of respondents decided before the purchase and 38.7% at the time of purchase.

Using a mail survey, Fesenmaier and Jeng (2000) examined the structure of trip decision-making processes, exploring how travel decisions are developed during the trip planning phase. Respondents were asked to plan a hypothetical summer trip in the Midwestern United States. Using a hierarchical cluster analysis to a joint planning matrix that included 14 travel decisions, they found that trip planning seems to include five core decisions: date and length of the trip, primary destination, travel route, accommodation, and travel party. Moreover, decisions made at an earlier stage seem to condition decisions made at later stages.

Regarding the timing between first travel plans and trip departure, Choi et al. (2012) found that Chinese visitors to Macau started planning their trips about 4 weeks prior to departure. In comparison, Dellaert, Ettema, and Lindh (1998) found that Swedish travelers take about 6 to 7 weeks to plan their holidays, whereas Zalatan (1996) found that Canadian travelers spend about 15 weeks to plan an overseas trip.

Modeling the time spent on planning as a function of attributes of the trip, socioeconomic characteristics of the tourist, knowledge of the destination, and involvement of a travel agent, Zalatan (1996) found that highly educated, older people, and distance are positively correlated with the amount of planning time, whereas familiarity and use of a travel agent were negatively correlated with planning time. The purpose of travel (Johns and Gyimothy 2002) and whether the destination had been visited before (Lo, Cheung, and Law 2002) were also found to be correlated with the amount of time spent on planning.

To elaborate on these general studies, others focused specifically on the use of online search engines in the vacation planning process. Pan and Fesenmaier (2006) used microphones, cameras, records of screen activities, and visited websites to capture the online vacation planning process, based on an one-hour task in which respondents planned a vacation to San Diego. Results showed that tourists’ online planning follows a hierarchical structure, based on each particular episode and chapter of the process, and that users presented different semantic mental models compared to those found on websites. Fesenmaier et al. (2011) developed a general framework comprising three different phases of travel planning, linking together the online search process itself with the presearch conditions that drive the search process (the query formulations) and the evaluation of the overall search process that culminates in attitude formation toward search engines. They found that those respondents using search engines extensively tend to be very active and involved travel planners. Moreover, respondents differ substantially in terms of their satisfaction with the search engines’ results.

Dellaert, Ettema, and Lindh (1998) developed a conceptual framework to represent the temporal sequence of multifaceted tourist travel decisions. They asked 300 Swedish respondents, between March and May, to indicate their vacation plans for the coming June to December. Although this study was one of the first steps toward the analysis of the multifaceted character and the timing aspects of vacation planning, it has remained rather limited in the sense that the time frame was short (from June to December). Our study attempts to extend this time frame by asking respondents to indicate their real and intended vacation plans, at any time in the future. Moreover, it also addresses the role of sociodemographic characteristics in affecting those decisions. Specific focus is given to the impact of family life cycle stages in the vacation planning process.

Influence of Sociodemographic Characteristics in the Vacation Planning Process

Vacation decisions are likely to be influenced by the sociodemographic characteristics of the travelers and trip variables (e.g., Woodside and Lysonski 1989; Um and Crompton 1990; Heung, Qu, and Chu 2001). Vacation portfolios may vary between people given such factors as age, income, household composition, and education level (Weaver et al. 1994; Zimmer, Brayley, and Searle 1995; Melenberg and Soest 1996; Stemerding, Oppewal, and Timmermans 1999; Fleischer and Seiler 2002; Mergoupis and Steuer 2003; Nicolau and Mas 2005, 2006; Hellstrom 2006; Alegre, Mateo, and Pou 2010). In addition, it seems that experienced travelers shift toward a more limited vacation planning process because going on vacation has become routine in their lives (Woodside and Lysonski 1989; Fodness and Murray 1998, 1999; Gursoy and McCleary 2004; Lehto, O’Leary, and Morrison 2004; Bargeman and van der Poel 2006).

Moreover, the individual’s travel behavior may change in response to different life cycle stages, which appear to be composed of both age and household composition. The life cycle concept in tourism and travel research has been developed since the 1970s (Cosenza and Davis 1981), but in marketing research and psychology this is an even older issue, dating back to the 1960s and 1930s, respectively (Wells and Gubar 1966). In travel research in general, the use of the life cycle concept for understanding individual and household travel behavior was stressed by Zimmerman (1982). Kitamura and Kostyniuk (1986) also suggested that life cycle stage accounts for as much or more variation in travel than variables such as household size, income, number of workers, and number of cars. Ortúzar and Willumsen (2011) also identified life cycle as an important variable explaining trip making behavior.

The relationship between tourism choices and life cycle has been examined by several authors. For example, differences between life cycle travel patterns according to both gender and the purpose of travel were examined by Collins and Tisdell (2002), whereas Gibson and Yiannakis (2002) investigated gender differences in tourist preferences for respondents in the adult life phase. Cosenza and Davis (1981) and Fodness (1992) found that differences across groups in the family life cycle are reflected in information search and final vacation decisions. Differences were also found in tourist behavior in relation to the type of vacation taken and expenditures over the stages of the family life cycle (Lawson 1991).

Oppermann (1995, 1998) focused on destination choice across the life cycle. Using a retrospective questionnaire, he investigated the changes in tourism patterns along three time horizons (the previous three decades, across the life cycle, and between successive generations) and found that younger generations gained different experiences when compared to previous generations and, moreover, tended to have different tourist decisions in later life stages. In relation to transport mode choice, Davison and Ryley (2009) found that key life stages, such as having children and entering retirement, influence air travel behavior.

Other studies have focused on a specific life cycle stage. For instance, Peercy and McCleary (2011) found changes in the family vacation decision process resulting from the implementation of a year-round school calendar in a Midwestern school in the United States. Focusing on the student cohort, mostly composed of young single people, Carr (1999, 2002) discussed gender differences in leisure activities and behavior on international and domestic trips. Ross (1993) investigated destination evaluations, revisitation intentions, vacation preferences, and sociodemographic characteristics of budget travelers. Sung (2004) found that students usually prefer to arrange trips by themselves and to travel with friends. The role of low-fare airlines in influencing students’ vacation decisions is addressed by Grigolon, Kemperman, and Timmermans (2012).

Thus, these existing literatures suggest differences in the vacation planning process, which in part are associated with different life cycle stages. In turn, life cycle stage influences the nature of vacation decisions. Previous work has primarily focused on small-scale, in-depth analyses of the actual stages of the vacation decision process. This study examines the topic from a different perspective: if the findings of the in-depth studies are valid, they should be reflected in aggregate patterns of the percentage of a sample having made certain decisions at different points in time. This study was developed from this perspective and examines aggregate timing aspects of vacation planning processes for different life cycle stages. In addition, income and traveler experience were also selected to be included in the model as sociodemographic characteristics.

Data and Sample

The analysis reported in this article is based on the Continu Vakantie Onderzoek (CVO) database, which contains information about annual vacation behaviors of Dutch people. More specifically, panelists are invited to report for four quarters their sociodemographic characteristics and holiday-related variables. These data are collected by NBTC-NIPO Research, a joint venture between the Netherlands Board of Tourism & Conventions (NBTC) and TNS NIPO (part of the TNS Group, focusing on market research), specialized in research in the fields of holidays, business travel, and leisure.

On March 11, 2011, 839 respondents who participated in the CVO data collection from 2002 to 2009 were invited to complete an online survey with a set of additional questions, especially phrased for the present study. They were asked to list up to 10 vacation plans for the future and indicate which choices they had already made.

A total of 462 respondents (response rate of 55%) indicated that they had (partially) prepared plans for one or more vacations, totaling 978 plans (about 2 plans per respondent). Some key sociodemographics of the panel are shown in Table 1. It shows that males were slightly overrepresented. Most respondents were older than 35 years, had medium levels of education and income, and lived in a household composed of two people.

Table 1.

Sample Characteristics

Level	Frequency (%)
Gender
Female	47.4
Male	52.6
Age
18-34 (young)	6.1
35-54 (middle-aged)	39.8
55 or older (mature)	54.1
Income
Low (less than 28.500 €/year)	28.8
Medium (28.500 to 68.000 €/year )	47.8
High (more than 68.000 €/year)	23.4
Education
Low	7.1
Medium	57.8
High	25.1
Number of household members
1	14.3
2	53.0
3	10.0
4 or more	22.7
Life cycle stages
Young single (single respondent, aged 18 to 34)	0.6
Young couple (married respondent, aged 18 to 34)	0.0
Full nest I (household with children aged 0 to 5)	6.9
Full nest II (household with children aged 6 to 17)	18.6
Full nest III (household with all members aged over 18)	7.8
Middle-aged single (single respondent, aged 35 to 54)	12.3
Middle-aged couple (married respondent, aged 35 and 54)	6.3
Mature single (single respondent, aged over 55)	7.1
Mature couple (married respondent, aged over 55)	40.3
Travel frequency
Low (up to 2 trips/year)	38.3
Medium (from 2 to 4 trips/year)	43.5
High (more than 4 trips/year)	18.2
Number of plans informed
1	41.8
2	28.3
3	17.8
4 or more	12.1

Age and household composition were used to create the life cycle variable. In the CVO data, age was originally a continuous variable, and household composition had many different categorical levels. To simplify the classification, age and household composition were categorized into fewer classes. From these, nine life cycle variables were created based on the literature (Wells and Gubar 1966; Lawson 1991; Fodness 1992; Huntsinger and Rouphail 2012). The young (between 18 and 34 years of age), middle-aged (between 35 and 54 years of age), and mature (aged over 55 years of age) groups were divided into singles (leaving alone) and couples (living with partner). However, the young couple group, originally present in the panel, did not participate in the present data collection. Full nest I, II, and III are households with children aged younger than 5 years, households with children between 6 and 17 years, and households with all household members older than 18 years, respectively. In the present sample, the mature couples are most represented, followed by full nest II.

From the historical data, collected between 2002 and 2009, we extracted information on whether respondents were less or more frequent travelers. In more than 40% of the cases, respondents were classified as medium-frequency travelers, that is, they made on average between two and four vacation trips per year. Low-frequency travelers (up to two trips per year) made up 38% of the cases, whereas less than 20% are high-frequency travelers (more than four trips per year).

In relation to the number of vacation plans made, in almost 42% of the cases only one plan was documented, even though respondents had space to specify up to 10 plans. Almost 30% planned two trips, whereas almost 20% informed three trips. Only 12% of the respondents provided details for four or more vacation plans.

Approach and Model Specification

Respondents were asked to indicate the future vacations they would like to make, and for each choice facet whether they already had made (mentally) a decision. Details about the destination could be provided by writing down in a text box the name of the destination. If the destination was not planned yet, they could give any other label that better defined their vacation, for instance, “whale watching” or “summer vacation,” or simply leaving the space blank, indicating this facet was not yet planned. What followed was the selection of the other facets, by means of drop-down lists, which included the following:

Season: summer, autumn, winter, spring, not planned yet

Vacation purpose: city trip, visit family and/or friends, beach/water, active vacation (sports, winter sports), nature, visit to event or attraction park, not planned yet

Transport mode: car, airplane, train, bus, boat, motor home, bike, other, not planned yet

Travel party: alone, with partner, with family, with friends, other, not planned yet

Accommodation: hotel, hostel, camping, rented house/apartment, own or from a family member or friend’s house/apartment, boat, motor home, other, not planned yet

Length of stay: short (1 to 3 nights), medium (4 to 9 nights), long (more than 10 nights), not planned yet

Time frame: within 1 year, within 1 and 2 years, within 2 and 5 years, within 5 and 10 years, in more than 10 years, not planned yet

Given these data, the aim of the model estimation is to predict whether a choice facet had been decided by an individual at the present time as a function of lead time to the actual trip and a set of sociodemographic variables. The model can be expressed as follows,

P_{i j t} (1 < t < T) = \frac{\exp (μ_{i j} + α_{i j} \ln (T - t + 1) + γ_{k} L_{i k} + θ_{n} I_{i n} + β_{f} F_{i f})}{1 + \exp (μ_{i j} + α_{i j} \ln (T - t + 1) + γ_{k} L_{i k} + θ_{n} I_{i n} + β_{f} F_{i f})}

where P_ijt (1 < t < T) is the probability that individual i (i = 1, 2, . . ., 462) chooses choice facet j (j = destination, season, vacation purpose, transport mode, travel party, accommodation, length of stay), which has been decided at present time t as a function of lead time to the actual trip T.

The term µ_ij is a random alternative-specific constant, representing the average difference in utility between planning a vacation and the base alternative of not planning a vacation. The second component of this specification picks up the effect of lead time. Because this effect may be nonlinear, the logarithm of time is used, although the effect of lead time as a linear function was also tested. We allow this effect of lead time to vary between choice facets as some facets may be planned earlier than others.

The other components represent vectors of random parameters of individual and alternative-specific attributes, allowing for the possibility that planning may vary across travelers in terms of their sociodemographic profiles. The coefficients that are estimated are life cycle stages (γ_k) and income levels (θ_n), and finally, we examine whether the vacation planning process differs between frequent and nonfrequent travelers, estimated by the coefficient β_f. These parameters were effect coded. This means, for example, that for every three-level attribute, two indicator variables were constructed. The first of these, coded as (1, 0), is associated with the first attribute level. The second indicator variable, coded as (0, 1), is associated with the second attribute level. The third attribute level is coded (−1, −1) on these two indicator variables. Consequently, the estimated utilities for each attribute sum to zero across the levels of that attribute. The t-statistics of each part-worth utility indicate any significant differences against the mean utility of that attribute.

Technically, this model specification is an example of a binary logit model. Rather than estimating the standard model, we estimated a mixed-logit panel model. The mixed-logit model allows for any heterogeneity in the estimated parameters in addition to the heterogeneity observed in the explanatory variables by replacing point estimates of the parameters with some assumed distribution for the estimated parameters. In many cases, a normal distribution was assumed. This seemed a good decision for those choice problems where heterogeneity was caused by multiple underlying random processes. However, because the estimates of a normal distribution may be either positive or negative, it should be realized that consequently theoretically unlikely or undesirable estimates may be obtained. Moreover, the assumption of a normal distribution may not be realistic if the underlying dominant behavioral process would result in a skewed distribution. A second difference from the standard binary logit model is that because subjects were invited to express all their vacation plans, error terms may be correlated. We therefore estimated a panel model. Choice probabilities were calculated using Halton sequences (Train 1999). As suggested by Bhat (2001), these sequences allow drawing from a distribution of random numbers that are more uniformly spread over the unit interval.

Descriptive Analysis

Before discussing the results of the estimated model, it is interesting to descriptively analyze the observed data. Figure 1 shows the observed relationships between the probability that a certain facet has already been planned or mentally decided and the time until the trip will be made. It shows that vacation type, transport mode, travel party, and accommodation were planned in at least 90% of the cases. For trips within one year, the season was planned in almost all the cases, but for later trips it dropped to around 85% and to less than 70% for trips planned more than 10 years ahead. Destination was planned between 75% and 80% of all time frames. This lower percentage may be a result of the fact that other facets such as timing and season represent routines (Bargeman and van der Poel 2006) or reflect institutionalized conditions affecting vacation choices (e.g., summer holidays), but we also cannot rule out that the extra burden of having to type in the name of the destination may have influenced this finding. It should be noted that most of the plans (almost 80%) were made for vacations within one year, whereas 12% were for trips planned with a lead time of one to two years and 8% for trips planned with a lead time of two to five years, and only a few respondents planned trips for more than five years from the present.

Figure 1.

Planned facets as a function of time

Figure 2 also presents evidence that planning differs between travelers of different life cycle stages. It shows that full nest III and young singles had made fewer plans for destination than the other groups. Type of vacation was least planned by the young single group. In general, the differences for the other vacation facets are not substantial.

Figure 2.

Planned facets by life cycle stages

Model Results

Maximum likelihood was used to estimate the model. NLOGIT 4.0 (Greene 2007) was the econometric software used to estimate the model. The estimated parameters are shown in Table 2. Because the distribution of time may not be uniform, we included lead time as a linear and logarithm function and compared the results. Overall, the model based on the logarithm of lead time performed better than the linear model, as indicated by a Rho-square of 0.302 versus a rho-square of 0.277. According to Louviere, Hensher, and Swait (2000, . 54), Rho-square values between 0.2 and 0.4 are indicative of good model fit. We focus the interpretation on the predicted coefficients based on the logarithm function of lead time.

Table 2.

Model Results

Variable	Coefficient		SD
Constant (µ_ij)	4.835	(23.19)	0.051	(0.61)
Lead time (α_ij)
Destination	−1.248	(−17.05)	2.171	(18.72)
Season	−0.656	(−9.46)	0.411	(5.83)
Vacation purpose	−0.240	(−2.06)	0.211	(1.65)
Transport mode	−0.354	(−3.74)	0.137	(1.18)
Travel party	0.303	(1.43)	1.140	(5.46)
Accommodation	−0.474	(−5.62)	0.037	(0.36)
Length of stay	−0.455	(−5.8)	0.003	(0.04)
Life cycle (γ_k)
Young single (base level)	—		—
Full nest I	0.275	(0.84)	0.103	(0.24)
Full nest II	−0.526	(−2.84)	0.986	(5.09)
Full nest III	−0.553	(−2.4)	0.317	(1.13)
Middle-aged single	0.876	(2.91)	1.807	(4.97)
Middle-aged couple	0.122	(0.56)	2.276	(9.01)
Mature single	−0.179	(−0.72)	1.028	(3.34)
Mature couple	−0.133	(−0.82)	1.571	(10.93)
Income (θ_n)
Low	−0.264	(−2.23)	0.537	(3.54)
Medium	−0.112	(−1.22)	0.342	(3.6)
High (base level)	—		—
Frequent traveler (β_f)
Low	0.111	(1.18)	0.811	(6.9)
Medium	−0.184	(−2.2)	0.181	(1.79)
High (base level)	—		—

In addition, we tested different distributions (normal, triangular, lognormal, and uniform). Because the model based on the normal distribution gave the best results, in the remainder of this section we discuss the estimated parameters of the mixed-logit model, based on the normal distribution. One thousand Halton sequences were used to estimate the parameters and the standard deviations of the binary mixed logit model. Correlations between the independent variables had a maximum value of .393, which indicates that there are not high correlations among them.

Estimated coefficients show that the probability of a facet being planned increases as time to the vacation trip decreases, as indicated by the high and positive value of the constant. This means that the closer to the actual date of the trip, the higher the probability of a facet being planned, which is an expected finding.

Examining the estimated coefficients of lead time for each of the vacation facets, model predictions also confirm the observed high probabilities of these facets being planned when the date to the actual trip gets closer. The negative signs of the coefficients are expected. Figure 3 illustrates the logarithm curves for each of the facets. Because 80% of the plans were made for vacations within one year, 12% from one to two years, 8% from two to five years, and only a few over 5 years, the association with the observed data, shown previously in Figure 1, is stronger for lead time within one year. It shows, for example, that within one year a destination is planned in almost 80% of the cases, which is consistent given the curves shown in Figure 1. Please specify which table you are referring to with “Figures 4.1 and 4.3”; there is no Figure 4 in your article.

Figure 3.

Estimated probabilities for planned facets over time

It is important to reinforce that in this experiment destination was considered as not planned if respondents did not clearly indicate the name of the destination, that is, when their intended vacation was labeled (e.g., “summer vacation”) or for missing values. The estimated probability for season indicates that even for trips with a time frame of 10 years from now, the probability of planning this facet was high (almost 90% of the trips). The other facets also exhibit very high probabilities of being planned. The positive sign of the travel party parameter, although not significant, indicates that the probability of this facet being planned tends to 100% considering all time frames. Indeed, as previously shown in Figure 2, respondents planned travel party in almost all their future plans. In summary, the model predictions answer the research question 1, showing that there are differences according to when different vacation facets are planned. This main finding is consistent with the results of Dellaert, Ettema, and Lindh (1998), who also found these differences in their study, which was based on a completely different methodology.

Estimated parameters for life cycle stages are statistically significant for full nest II, full nest III, and middle-aged singles only. For the first two groups, composed of families with children between 6 and 17 years old and families with adults only, results suggest that there is a tendency to plan vacations when the actual date of the trip is closer, shown by the negative sign of the coefficients. This may be the result of imposed constraints of one or more members of the family, as, for instance, the couple would have to combine their vacation days from work with the children’s school holidays, making the planning of certain facets much ahead impossible. However, the positive sign of the parameter for middle-aged single people shows that they tend to plan their trip in advance. In this case, a single person might have fewer constraints to plan a vacation, as no other members of the family are involved. Therefore, the results indicate that there is a difference in the level of vacation planning for different life cycle stages. These differences are in general reflected by flexibility/constraints aspects of each particular life cycle group.

The estimated coefficients for income show that as income level increases, the level of planning also increases. This suggests that planning ahead facets of a vacation is more important for respondents with higher incomes. However, it seems that the higher the travel frequency (i.e., the more experienced travelers) the less planning is done, as shown by the frequent traveler variable. Bargeman and van der Poel (2006) found similar results. Coefficients in this case decrease as the level of planning increases. These results imply that there are indeed differences between the level of planning for different income levels and traveler experience.

As to unobserved heterogeneity, the higher the standard deviation, the higher the heterogeneity in responses, that is, the higher the variation in the vacation planning of respondents. Destination is the vacation facet with the highest standard deviation, indicating that respondents are more diverse when planning a future destination. As for life cycle stages, among the three groups with significant coefficients, for only two (full nest II and middle-aged singles) is the standard deviation significant. Low income level and medium traveler groups have significant parameters with significant standard deviations. Consequently, those respondents exhibit more variability in their vacation plans.

Conclusions and Discussion

The study of vacation planning processes is highly relevant in terms of tourism marketing and planning. A focus on individual choices and on the formation of vacation agendas contributes toward an understanding of which activities are performed by which classes of individuals under which conditions.

Vacation planning processes are potentially very complex. Travelers develop their vacation agenda and elaborate details in parallel. The process involves information search and updating expectations, preferences, and plans. At the stage of booking, preferred options may no longer be available, triggering a process of adaptation and perhaps another cycle in the planning process.

Although initial planning processes may thus include several idiosyncrasies, to the extent that travelers of the same sociodemographic profile and travel experience share similar decision mechanisms, at the aggregate level structural relationships among the probability that certain facets have been planned at a certain point in time, sociodemographics, and travel experience will exist. The focus of this study is on such aggregate relationships.

To that effect, a binary mixed logit panel model was estimated. It assumed that the probability that a particular facet being planned at a certain point in time is a function of the logarithm of lead time, the nature of the facet, life cycle stages, income, and travel experience. The results provided evidence of the assumed relationship, as indicated by a satisfactory goodness of fit, signs of the estimated parameters in anticipated directions, and significant effects for most variables.

The design of the study involved some operational decisions, which at the same time can be viewed as limitations of the present study. First, it was assumed that sociodemographics and travel experience have a proportional effect on predicted probabilities, as these variables do not vary with time. A more sophisticated specification would allow for the possibility that sociodemographics also affect the form of the temporal relationship itself. Second, the data used for the present analysis concern the issue of whether a particular facet has been planned or not at some fixed point in time. The data do not reflect the actual timing of such decisions. Although in principle such information might be collected by posing retrospective questions, we felt that the Dutch vacation panel was not the right setting to ask such questions. We felt that the reliability of such retrospective questions in the present context might be at stake. If, however, a more dedicated panel could be organized in which the timing of such decisions would be recorded more continuously, such a line of research could be feasible. This would open up the possibility of estimating hazard models and multispell competing risk models. Alternatively, Bayesian network models could be estimated in which a decision with respect to a particular facet is viewed as evidence in the network and all conditional probabilities would be updated using backward reasoning.

Third, the present study has implicitly assumed that vacation decisions are made independently of any other life trajectory decisions such as marriage, child birth, retirement, and so on. However, it may well be that travelers do keep such longer time perspectives in mind when developing their vacation agenda. Additional data should be collected to further study such interdependencies.

Finally, the present focus on aggregate relationships means that this study is not concerned with underlying behavioral mechanisms. Nevertheless, some interesting research questions immediately come to one mind. If the preferred option is not available, will travelers choose less preferred, suboptimal alternatives, or do they postpone this vacation and choose another vacation of lower priority? Because most attributes influencing the vacation choice, and especially price, vary with time, how do travelers decide in such inherently uncertain decision contexts? A better understanding of such behavioral mechanisms requires the design of dedicated stated choice experiments in which the researcher systematically varies and controls the factors considered to be of interest. Travelers’ responses to these experimental conditions are then observed, and these responses can be analyzed using an appropriate statistical method. Such detailed analyses would complement the current study on emerging aggregate relationships.

Footnotes

Acknowledgements

The study was conducted as part of the project “The value of recreation: Now, and in a completely different future”, which is part of the DBR (Duurzame Bereikbaarheid van de Randstad - Sustainable Accessibility of the Randstad) programme. It was financially supported by the Netherlands Organization for Scientific Research (NWO).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Bios

Anna Grigolon is a PhD researcher at the Eindhoven University of Technology since 2009. Her research is focused on modeling dynamics aspects of recreation behavior and agenda formation, with a lifecycle perspective. She has research interests in modeling choice behavior and spatial decisions support systems.

Astrid Kemperman is an assistant professor of urban planning at the Eindhoven University of Technology, the Netherlands. Her research focuses on the impacts of urban environments on individuals’ leisure choices and tourist route and activity choice behavior. She has a specific interest in modeling dynamic aspects of tourist choice and shopping behavior.

Harry Timmermans is chaired professor of urban planning and director of European Institute of Retailing and Services Studies (EIRASS) at the Eindhoven University of Technology. He holds research interests in modeling spatial choice behavior and spatial decision support systems in a variety of application context, including retailing, tourism, recreation, transportation and housing.

References

Alegre

Mateo

Pou

(2010). “An Analysis of Households’ Appraisal of Their Budget Constraints for Potential Participation in Tourism.” Tourism Management, 31 (1): 45-56.

Bargeman

van der Poel

(2006). “The Role of Routines in the Vacation Decision-Making Process of Dutch Vacationers.” Tourism Management, 27 (4): 707-20.

Bhat

C. R.

(2001). “Quasi-Random Maximum Simulated Likelihood Estimation of the Mixed Multinomial Logit Model.” Transportation Research B, 35 (7): 677-93.

Carr

(1999). “A Study of Gender Differences: Young Tourist Behaviour in a UK Coastal Resort.” Tourism Management, 20 (2): 223-28.

Carr

(2002). “A Comparative Analysis of the Behaviour of Domestic and International Young Tourists.” Tourism Management, 23 (3): 321-25.

Choi

Lehto

X. Y.

Morrison

A. M.

Jang

(2012). “Structure of Travel Planning Processes and Information Use Patterns.” Journal of Travel Research, 51 (1): 26-40.

Collins

Tisdell

(2002). “Gender and Differences in Travel Life Cycles.” Journal of Travel Research, 41 (2): 133-43.

Cosenza

R. M.

Davis

D. L.

(1981). “Family Vacation Decision Making over the Family Life Cycle: A Decision and Influence Structure Analysis.” Journal of Travel Research, 20 (2): 17-23.

Crompton

J. L.

(1992). “Structure of Vacation Destination Choice Sets.” Annals of Tourism Research, 19 (3): 420-34.

10.

Davison

Ryley

(2009). “The Relationship between Air Travel Behaviour and the Key Life Stages of Having Children and Entering Retirement.” Paper presented at the 12th International Conference on Travel Behaviour Research, Jaipur, India.

11.

Decrop

Snelders

(2004). “Planning the Summer Vacation: An Adaptable Process.” Annals of Tourism Research, 31 (4): 1008-30.

12.

Dellaert

B. G. C.

Ettema

D. F.

Lindh

(1998). “Multifaceted Tourist Travel Decisions: A Constraint-Based Conceptual Framework to Describe Tourist’s Sequential Choices of Travel Components.” Tourism Management, 19 (4): 313-20.

13.

Fesenmaier

D. R.

Jeng

(2000). “Assessing Structure in the Pleasure Trip Planning Process.” Tourism Analysis, 5 (1): 13-27.

14.

Fesenmaier

D. R.

Xiang

Pan

Law

(2011). “A Framework of Search Engine Use for Travel Planning.” Journal of Travel Research, 50 (6): 587-601.

15.

Fleischer

Seiler

(2002). “Determinants of Vacation Travel among Israeli Seniors: Theory and Evidence.” Applied Economics, 34 (4): 421-30.

16.

Fodness

(1992). “The Impact of Family Life Cycle on the Vacation Decision-Making Process.” Journal of Travel Research, 31 (2): 8-13.

17.

Fodness

Murray

(1998). “A Typology of Tourist Information Search Strategies.” Journal of Travel Research, 37 (2): 108-19.

18.

Fodness

Murray

(1999). “A Model of Tourist Information Search Behavior.” Journal of Travel Research, 37 (3): 220-30.

19.

Gibson

Yiannakis

(2002). “Tourist Roles: Needs and the Lifecourse.” Annals of Tourism Research, 29 (2): 358-83.

20.

Greene

W. H.

(2007). NLOGIT Version 4.0: Reference Guide. Plainview, NY: Econometric Software.

21.

Grigolon

A. B.

Kemperman

A. D. A. M.

Timmermans

H. J. P.

(2012). “The Influence of Low-Fare Airlines on Vacation Choices of Students: Results of a Stated Portfolio Choice Experiment.” Tourism Management, 33 (5): 1174-84.

22.

Gursoy

McCleary

K. W.

(2004). “An Integrative Model of Tourists’ Information Search Behavior.” Annals of Tourism Research, 31 (2): 353-73.

23.

Hellstrom

(2006). “A Bivariate Count Data Model for Household Tourism Demand.” Journal of Applied Econometrics, 21 (2): 213-26.

24.

Heung

V. C. S.

Chu

(2001). “The Relationship between Vacation Factors and Socio-Demographic Characteristics: The Case of Japanese Leisure Travelers.” Tourism Management, 22 (3): 259-69.

25.

Huntsinger

L. F.

Rouphail

N. M.

(2012). “Value of Life Cycle in Explaining Trip Making Behaviour and Improving Temporal Stability of Trip Generation Models.” Paper presented in the 91st annual meeting of the Transportation Research Board, Washington, DC.

26.

Hyde

K. F.

(2004). “A Duality in the Vacation Decision-Making.” In Consumer Psychology of Tourism, Hospitality and Leisure, vol. 3, edited by Crouch

G. I.

Perdue

R. R.

Timmermans

H. J. P.

Uysal

Cambridge, MA: CABI, pp. 161-68.

27.

Hyde

K. F.

(2008). “Information Processing and Touring Planning Theory.” Annals of Tourism Research, 35 (3): 712-31.

28.

Jeng

Fesenmaier

D. R.

(2002). “Conceptualizing the Travel Decision-Making Hierarchy: A Review of Recent Developments.” Tourism Analysis, 7 (1): 15-32.

29.

Johns

Gyimothy

(2002). “Market Segmentation and the Prediction of Tourist Behavior: The Case of Bornholm, Denmark.” Journal of Travel Research, 40 (3): 316-27.

30.

Kitamura

Kostyniuk

L. P.

(1986). “Maturing Motorization and Household Travel: The Case of Nuclear-Family Households.” Transportation Research Part A, 20 (3): 245-60.

31.

Lawson

(1991). “Patterns of Tourist Expenditures and Types of Vacation across the Family Life Cycle.” Journal of Travel Research, 29 (2): 12-18.

32.

Lehto

X. Y.

O’Leary

J. T.

Morrison

A. M.

(2004). “The Effects of Prior Experience on Vacation Behavior.” Annals of Tourism Research, 31 (4): 801-18.

33.

Cheung

Law

(2002). “Information Search Behavior of Hong Kong’s Inbound Travelers: A Comparison of Business and Leisure Travelers.” Journal of Travel & Tourism Marketing, 13 (3): 61-80.

34.

Louviere

J. J.

Hensher

D. A.

Swait

J. D.

(2000). Stated Choice Methods: Analysis and Application. Cambridge, UK: Cambridge University Press.

35.

Melenberg

Soest

A. V.

(1996). “Parametric and Semi-Parametric Modeling of Vacation Expenditures.” Journal of Applied Econometrics, 11 (1): 59-76.

36.

Mergoupis

Steuer

(2003). “Holiday Taking and Income.” Applied Economics, 35 (3): 269-84.

37.

Nicolau

J. L.

Mas

F. J.

(2005). “Stochastic Modeling: A Three-Stage Tourist Choice Process.” Annals of Tourism Research, 32 (1): 49-69.

38.

Nicolau

J. L.

Mas

F. J.

(2006). “The Influence of Distance and Prices on the Choice of Tourist Destinations: The Moderating Role of Motivations.” Tourism Management, 27 (5): 982-96.

39.

Oppermann

(1995). “Travel Life Cycle.” Annals of Tourism Research, 22 (3): 535-52.

40.

Oppermann

(1998). “Travel Horizon: A Valuable Analysis Tool?” Tourism Management, 19 (4): 321-29.

41.

Ortúzar

J. D.

Willumsen

(2011). Modelling Transport. 4th ed. Chichester, UK: Wiley.

42.

Pan

Fesenmaier

D. R.

(2006). “Online Information Search: Vacation Planning Process.” Annals of Tourism Research, 33 (3): 809-32.

43.

Peercy

M. A.

McCleary

K. W.

(2011). “The Impact of the Year-Round School Calendar on the Family Vacation: An Exploratory Case Study.” Journal of Hospitality & Tourism Research, 35 (2): 147-70.

44.

Ross

G. F.

(1993). “Destination Evaluation and Vacation Preferences.” Annals of Tourism Research, 20 (3): 477-89.

45.

Stemerding

Oppewal

Timmermans

H. J. P.

(1999). “A Constraints-Induced Model of Park Choice.” Leisure Sciences, 21 (2): 145-58.

46.

Sung

H. H.

(2004). “Classification of Adventure Travelers: Behaviour, Decision Making, and Target Markets.” Journal of Travel Research, 42 (4): 343-56.

47.

Train

K. E.

(1999). “Halton Sequences for Mixed Logit.” Working paper, University of California, Berkeley.

48.

Crompton

J. L.

(1990). “Attitude Determinants in Tourism Destination Choice.” Annals of Tourism Research, 17 (3): 432-48.

49.

Crompton

J. L.

(1992). “The Roles of Perceived Inhibitors and Facilitators in Pleasure Travel Destination Decision.” Journal of Travel Research, 30 (3): 18-25.

50.

Weaver

P. A.

McCleary

K. W.

Lepisto

Damonte

L. T.

(1994). “The Relationship of Destination Selection Attributes, to Psychological, Behavioral and Demographic Variables.” Journal of Hospitality & Leisure Marketing, 2 (2): 93-109.

51.

Wells

Gubar

(1966). “Life Cycle Concept in Marketing Research.” Journal of Marketing Research, 3 (4): 355-63.

52.

Woodside

A. G.

Lysonski

(1989). “A General Model of Traveler Destination Choice.” Journal of Travel Research, 27 (4): 8-14.

53.

Woodside

A. G.

MacDonald

(1994). “General System Framework of Customer Choice Processes of Tourism Services.” In Spoilt for Choice. Decision Making Processes and Preference Change of Tourists: Intertemporal and Intercountry Perspectives, edited by Gasser

Weiermair

Vienna: Kultuur Verlag, pp. 30-59.

54.

Zalatan

(1996). “The Determinants of Planning Time in Vacation Travel.” Tourism Management, 17 (2): 123-31.

55.

Zimmer

Brayley

Searle

M. S.

(1995). “Whether to Go and Where to Go: Identification of Important Influences on Seniors’ Decisions to Travel.” Journal of Travel Research, 33 (3): 3-10.

56.

Zimmerman

C. A.

(1982). “The Life Cycle Concept as a Tool for Travel Research.” Transportation, 11 (1): 51-69.