Introduction

The global tourism industry has been experiencing a crisis over the past few years due to epidemics and health problems (Abdi et al., 2023; Aburumman, 2020; Ketter and Avraham, 2021). Although the movement of people between countries may be temporarily restricted due to COVID-19, the substitution effect is likely to increase demand for domestic tourism (DeMicco et al., 2021; Volgger et al., 2021). According to the Korea Culture and Tourism Institute (KCTI), 45.8% of respondents answered yes when they were asked about their intentions to participate in domestic travel despite the persistence of COVID-19 (KCTI, 2020). In addition, a result from a mobile survey conducted by the Gyeonggi Research Institute reported that 39.4% of the respondents said they had traveled in Korea after the COVID-19 outbreak. The reason for traveling was stress due to social distancing, which accounted for 50.8% (Cheung and Lee, 2020). This occurrence seems to be compensated consumption for refraining from outside activities. Although there are many studies discussing the COVID-19 crisis and exploring how this pandemic might transform society and tourism, only a few studies try to examine domestic family tourists’ behavior (or behavioral intention) based on empirical data analysis (Álvarez-Díaz et al., 2023; Wen et al., 2020). In addition, although many studies have emphasized the necessity and importance of research to better understand domestic travel behavior (Suzuki et al., 2023), there are not many studies that analyse the preferences of major factors constituting domestic factors. Accordingly, the purpose of current study is to assess the relative importance of various domestic family travel attributes by using a stated preference (SP) choice experiment (CE) and measure the marginal-willingness-to-pay (MWTP) for the attributes that make up domestic family travel in the new era of post COVID-19. Compared to other research methods, this method is appropriate to understand potential tourists’ trade-offs among a set of domestic tourism attributes and identify their opinions. The method contributes to identifying the relative importance of the proposed attributes. It was also intended to understand differentiated preference patterns embedded in groups based on their income level and the presence of children. In order for domestic travel to develop successfully in the era of COVID-19, it can be a valuable academic attempt to identify the selection attributes of tourists related to domestic travel and to determine what decisions they make.

The current study fills the research gap using a CE which considers tourist preferences as a variety of other determinants that influence the decision-making process (Álvarez-Díaz et al., 2023). Additionally, Qui et al. (2020) estimated the Chinese local residents’ willingness to pay (WTP) in order to mitigate the COVID-19 risk related with tourism. Regarding the methodology, although the study is one of the first studies that used a triple-bounded dichotomous choice contingent valuation (CV) method (CVM), CVM has a limitation that it only estimates the individual’s WTP according to the change of the single attribute of a specific resource (Bennett and Adamowicz, 2001). Unlike CVM, CE enables measuring the MWTP values of different levels of multiple attributes (Louviere et al., 2000).

Literature review

Domestic family travel decision-making process

Park et al. (2020) defined family travel as “a purposive time spent together as a family group, which may include extended family, doing activities different from normal routines that are fun, but that may involve compromise and conflict at times” (p. 2). Their definition regarding family travel is followed here, but day trips are excluded because this study also attempts to explore the preferences for the type of accommodation during COVID-19 when they stay overnight. Additionally, this study is concerned with family travel over the weekend or during a short break, which includes long weekends, rather than holidays or vacations that require a greater amount of time, money, and schedule organization in advance.

It is very important to understand the family travel decision-making process for the promotion of the tourism industry and academic development in the tourism field (Kang et al., 2003). Travel decision-making is the process of choosing and purchasing a trip in order to satisfy various needs, which can be seen as multi-faceted decision-making that occurs in the process of moving to and staying at a destination (Etgar, 1978). More specifically, it is the process of (1) desiring travel, (2) exploring and evaluating the alternatives, (3) making a final decision, (4) participating in travel activities, and (5) retuning in order to evaluate the entire travel experience (Jeng and Fesenmaier, 2002). Therefore, in order to be a series of awareness systems that occur before and after tourism actions, they can be understood as having a sequential and hierarchical structure (Mayo and Jarvis, 1981). The first step, desire to travel, is a step to perceive the wish to travel, such as I want to go on a trip or where do I want to go, what do I want to do, and for how long? The first step is a general decision step of deciding whether or not to go on a trip. Additionally, it is affected by internal factors within the family and by the external social environment (Jeng and Fesenmaier, 2002). Maditinos and Vassiliadis (2008) categorized the external events in tourism into four categories, which include (1) natural disasters, (2) terrorism, (3) political instability and war, and (4) epidemics. People tend to be reluctant to travel during epidemics when social distancing and self-isolation are encouraged. As such, COVID-19 has an impact on the travel decision making process from the very first step, and it is considered as one of the travel restrictions mentioned above in this study. According to a survey on the Korean tourism during the MERS outbreak in June 2015, 52.9% of the respondents answered there was an experience of canceling a planned trip due to MERS, which accounted for more than half of the respondents. In addition, if MERS continues, 64.8% of the respondents answered I will cancel a planned trip or avoid going-out (Choe et al., 2021). Thus, in light of past experiences, the epidemic has had a great influence on people’s decision-making process.

The examination of family behavior as a consumer entity has garnered substantial scholarly attention (Srnec et al., 2016). A plethora of studies have delved into the various sub-determinants that influence family travel decisions, including the timing and location of purchases, payment methods, travel destinations, durations, expenditures, modes of transportation, travel activities, accommodations, and the presence of children (Backhaus et al., 2023; Davis, 1976; Jia et al., 2023; Minnaert, 2021).

Martin (2014) identified price as a paramount factor influencing these decisions, with some participants expressing a WTP a premium for superior travel product quality. However, in the context of the COVID-19 pandemic, safety from contact with others has emerged as the most critical determinant of family travel (Minnaert, 2021).

The process of decision-making often depends on which family member, whether it’s the husband, wife, or children, assumes the responsibility of making these choices, or if the decision is a collective effort involving all family members (Backhaus et al., 2023). Consequently, numerous studies have focused on elucidating the roles of different family members within the family travel decision-making process. While earlier studies from the 1980s suggested that husbands played a more dominant role in this process (Ritchie and Filiatrault, 1980; Van Raaij and Francken, 1984), more recent research indicates a shift towards the dominant roles of wives and children (Backhaus et al., 2023; Pan and Shang, 2023; Schänzel and Yeoman, 2015). Schänzel and Yeoman (2015) discussed this trend in family tourism, emphasizing the increasing importance of children as a consumer target within the realm of family tourism.

Tourists’ preferences in post COVID-19

Tourists tend to consider domestic tourism safer than international tourism (Matiza, 2020). This is called “home-is-safer-than-abroad bias” (Wolff et al., 2019). It is suggested that the tourism stakeholders may need to concentrate on fostering domestic tourism as a short-term measure in order to revitalize their own tourism industry and revive the international tourism industry in the mid to long term (Matiza, 2020). Before promoting domestic tourism, tourism stakeholders should be constantly aware of tourist’s behavioral demand/preference on tourism type, travel needs, and products such as accommodation, destination, activities etc. They also need to be consistently flexible to recognize and respond to them (Matiza, 2020). The current literature expecting traveler behavior/preference in post-COVID-19 is outlined as follows.

First, it is expected that safety and hygiene will emerge as decisive factors regarding choosing a tourist destination/tourist facility. Safety and hygiene are the basic elements of tourism services, but in order to relieve or alleviate the tourists’ anxiety sentiment caused by COVID-19, their value and importance will be much greater than before COVID-19 (Chebli and Said, 2020). For example, Jiang and Wen (2020) suggested that hotel practitioners should offer and market safe accommodation services in terms of hygiene and cleanliness. Consequently, the demand for luxury tourism with guaranteed safety and hygiene will increase and interest in shared accommodations will decline.

Second, the demand for outdoor or nature-friendly travel will increase (Aslam et al., 2020; Seraphin and Dosquet, 2020). The prolonged spread of COVID-19 has increased the interest in physical and mental health, which is expected to increase the demand health-oriented tourism, such as wellness tourism, healing tourism, and recreational tourism (Castellvi et al., 2023). Seraphin and Dosquet (2020) asserted the placebo role of mountain tourism and second-home tourism during the post-COVID lockdown. They also said that this would greatly help people’s mental and physical stability.

Third, there is a trend of individualization and miniaturization regarding travel types. Due to the influence of COVID-19, the trend of preferring small-scale travel rather than group package tours where there are a lot of contact with others will be more prominent (Chebli and Said, 2020). Wen et al. (2020) reported that Chinese tourists are also more likely to travel independently or in small groups to stay safe.

Fourth, COVID-19 is expected to increase the tendency of tourists to prefer visiting relatively less-known tourist destinations (Chebli and Said, 2020; Karl et al., 2020). Also, tourists would prefer domestic tourist destinations that have sufficient capacity and the least crowdedness (Karl et al., 2020), which can be inferred from a combination of intrinsic motivations, such as anxiety, fear of contamination, and infection. (Chebli and Said, 2020).

Fifth, the demand for daily travel where destination is close to home will increase. Researchers predicted that when faced with certain problems or an unexpected situation while traveling, they would prefer a situation where they can easily return to their residence in order to reduce the psychological burden (Cheung and Lee, 2020). Furthermore, adjacent destinations can be considered less risky in the context of increasing anxiety and uncertainty (Romagosa, 2020).

Stated preference (SP) choice experiment

CE is one of the most important and widely used among the three SP methods, which includes the CE, CV, and best-worst scaling (Aizaki et al., 2014). SP is a survey method that measures the individual’s preferences for alternatives based on decision-making in hypothetical scenarios (Adamowicz et al., 1994). SP methods have been commonly used to the estimate economic value of non-market/non-use goods that are impossible or difficult to measure, such as environmental/natural resources (Adamowicz et al., 1994). CE is a method of inducing the respondents to choose a level change that is in conflict with each other by presenting two or more hypothetical alternatives that consist of various attributes and the WTP, which allows the respondents to choose the one they like most. This has a great advantage of being able to better understand the consumers’ actual trade-off behavior because it presents similar situations that resource users face in the decision-making process.

Unlike CV yields, which derive utility and MWTP from only a single potential attribute, CE has the advantage of being able to measure people’s preferences and MWTPs using multiple attributes and the different levels of each attribute (Louviere et al., 2000). In other words, CV relies on a single factor to determine the value of a good or service, while CE considers multiple factors and their different levels in order to estimate value. Therefore, CE presents a choice or a set of choices that contains one or more measurement attributes, which is opposed to asking the respondents directly for a monetary assessment of a given good. The respondents’ utility function can be inferred from the obtained response, and the monetary value of the multiple attributes of the utility function can be also estimated (Van Ommeren et al., 2000).

Methodology

Attributes and attributes level

One of the most important procedures with the CE method is to determine the attributes and the levels for each attribute that will be used in the model (Pike et al., 2022). The attributes in this study were developed through prior studies, consultation with various stakeholders in tourism industry, and the 2019 National Travel Survey Report, which is published every year by the Korean Ministry of Culture, Sports, and Tourism. First, focus group interviews were conducted by phone with a total of 30 individuals, ranging from 20 to 60 years old, including three males and three females per age group, which were based on the attributes that were selected from the 2019 National Travel Survey Report. The interview included questions about travel: whether they had traveled before and after the COVID outbreak, the motivation and the purpose of the trip, the environment setting of the destinations, if they had any travel companions, transport modes, travel durations, accommodation types, trip expenditures, crowdedness of the destinations, media/SNS exposure of the destinations, if there were any change with the preferences, and behaviors on their trip affected by COVID19. The interview results were shared with three experts in the field of tourism, which included two from a travel agency and two professors from the tourism management department, in order to determine attributes and their level. Table 1 presents the major attributes, levels, definition of each attribute, and the list of the prior studies that refer to the attributes and the attribute levels.

Table 1 Attributes and attribute levels.
Full size table

Choice sets and CE design

This study is composed of 6 attributes and 2–4 levels, so a total of 648 (3 × 3 × 3 × 4 × 2 × 3) alternatives are derived in full factorial design. However, it is impractical to present all 648 alternatives in order to examine the respondents’ preferences. Therefore, a data generating process designed in order to reveal the factors that influence the respondent’s choice is required. Instead of providing all 648 combinations, a fractional factorial design with orthogonal main effects was used to extract a controllable number of representative profiles and to alleviate the participant’s cognitive burden (Hensher et al., 2005). Louviere et al. (2000) argued that some types of fractional designs are necessarily affected by the loss of statistical information, but the orthogonal main effects model accounts for most of the described variance, which is between 70 percentages to 90 percentages. In addition, using CE with this orthogonal design is a great benefit over random utilities that are known to be highly correlated with each other (Hanley et al., 1998). However, presenting all 72 pairs of choice sets also places an excessive cognitive burden on the respondents. Therefore, by dividing the questionnaire into 8 versions using a block design technique, only nine pairs of choice sets were placed for each version. R program is used for the fractional-factorial design and to derive a total of 9 choice sets.

Meanwhile, respondents who give up choosing one of the alternatives are sometimes witnessed if they don’t like any of them. In consideration of the characteristics of these respondent, this study included the selection options, which consist of (1) I will go on an A type family travel, (2) I will go on a B type family travel, and the no-choice option of (3) not choose either in the choice set. The no-choice option is related to the alternative specific constant (ASC), and it is possible to confirm whether the respondent actually prefers to choose a product made with the two options or abandon the purchase through the model estimation results (Hensher et al., 2005).

Data collection and participants’ profile

In the tourism market, families have unique characteristics that set them apart from other market segments (Srnec et al., 2016), occupying the largest consumption segment in the consumer market (Kim, 2020). Therefore, the sample of this study is Koreans aged between 30 and 40 who are married. This age group is known to be as family travel decision makers and have financial resources (Bae and Chang, 2021). In fact, Bae and Chang (2021) found that the effect of perceived behavioral control on non-face-to-face tourism intention was stronger for married respondents than for single respondents. They attributed the results to the fact that married participants usually make decision on behalf of a family unit.

Most of the FGI interviewees said that they wanted to avoid tourist spots and crowdedness, which was quite expected. Relatively many changes in people’s preferences and behavior can be identified in consumers in their 30s and 40s with children. On the other hand, there was no significant change in tourism preference for singles in their 20s and those in their 50s/60s with grown children compared to those in their 30s and 40s. This may be due to the high levels of behavioral control and less cognitive constraints (Bae and Chang, 2021) in the 20s and 50s/60s age groups. The sample of the current study is Koreans aged between 30 and 40 who are married, which is based on the evidence above.

The online survey was conducted from February 18, 2021 to February 24, 2021 by via one of the Korea research companies that has the largest number of clients. The sample for this study was comprised of married individuals within the age range of 30 to 49, who have traveled with their family in Korea both before and after the outbreak of COVID-19. The participants who did not correspond to the screen questions were excluded from the survey. Prior to starting the online survey, screening questionnaires were asked in order to assess the suitability of the attributes and levels and to identify any errors in the questionnaire. A total of 632 surveys were collected and utilized for the data analysis. A summary of the demographic characteristics of the participants is presented in Table 2. The average age was 39.62 years, and the average monthly family income was approximately KRW 4,856,000. About 23% of the respondents had no child and having 1–2 children was the dominant response. The highest number of respondents had children aged three or older to preschoolers (30.4%).

Table 2 Summary of the sample characteristics (n = 632).
Full size table

Results

Model estimation results without covariates

In line with the random utility maximization, an individual i’s latent utility alternative j in the choice situation can be specified as follows:

$$\begin{array}{l}U_{ij} = ASC_j + \beta _1medTT_{ij} + \beta _2longTT_{ij} + \beta _3camping_{ij} \\\qquad + \beta _4hotel_{ij} + \beta _5resort_{ij} + \beta _6low\_media_{ij} \\\quad\quad + \beta _7cultural_{ij} + \beta _8entertaining_{ij} + \beta _9\cos t_{ij} + \varepsilon _{ij}\end{array}$$

where ASCj is an ASC to detect a possible preferences for one of the alternatives; medTT, longTT, camping, hotel, resort, low_media, cultural, entertaining, and cost are the dummy coded attributes’ levels; β1 to β9 are a set of parameter that must be calculated for each attribute that influence an individuals’ utility; β10 to β12 are the parameters that are estimated for each individual-specific variable multiplied by ASCj, which represents the dummy variable of the respondent that chooses alternative j from the set of choices; and εij is a Type I error term. ShortTT, pension, high_media, and relaxing are the omitted references. The conditional logit (CL) model was analyzed by the R package using both support CEs and the survival package. Table 3 shows the estimation results of the model without covariates.

Table 3 Model estimation results without covariates.
Full size table

McFadden’s Pseudo R2 is known for the goodness fit of the conditional logit model (McFadden, 1974). Louviere et al. (2000) said that when Pseudo R2 is greater than 0.1, the statistical significance of the model itself can be guaranteed and values from 0.2 to 0.4 for R2 represent an excellent fit (Hensher and Johnson, 1981). The Pseudo R2 value of this study is 0.189, so the significance of the estimation result is acceptable. All the coefficients of the indirect utility functions are statistically significant, which are at least p < 0.05, with the exception of 2–3 h of TRAVEL TIME. An effect coding technique is used to include the categorical choice attributes in the data set, which involves a base category for each attribute that is not included in the model specification, and the base category is computed by the negative sum of the other levels using binary coding (Hensher et al., 2005; Huybers, 2003). For instance, the coefficient evolving within 2 h, which is the base level, in the attribute of TRAVEL TIME is calculated by using the following equation: −(βmedTT + βlongTT) = −(−0.017 − 0.302) = 0.32. The positive coefficient sign for the ASC implies that the respondents preferred domestic family travel alternatives as opposed to the no-travel option, which is the base category. The negative sign of the TRAVEL TIME attribute indicates that the respondents attach a negative utility to a longer travel time, which is the base category: within 2 h. The positive sign for the TRAVEL DURATION attribute refers to the respondents who were more in favor of staying over two nights opposed to one night. This interpretation also applies to being more in favor of staying over three nights opposed to one night. The positive sign for the hotel and resort/condo suggests that the tourists prefer to stay in both a hotel and a resort/condo than a pension (the base category). The given coefficient for resort/condo is larger than the coefficient for hotel, so the respondents are more likely to have preference of staying in a resort/condo than in a hotel. Similarly, a negative sign of camping/glamping/caravan indicates that the respondents have a lower preference towards camping/glamping/caravan compared to pension. In regards to the MEDIA/SNS EXPOSURE attribute, it seems that the respondents favor lower exposed destinations/places than highly exposed ones through media/SNS (the base category). Concerning MAIN ACTIVITIES at destinations, natural scenery or relaxing, which is the base category, is significantly preferred over both cultural/historical and entertaining/sports. The COST attribute shows a negative sign, which represents that an increase in the travel cost per day is an impediment to family travel, that is consistent with demand theory. The −0.032 coefficients, interpreted as a 10,000 KRW increase in expenditure, reduces the tourist utility by 0.032. This COST coefficient can be used to the estimate implicit prices at the attribute levels (Huybers, 2003).

Model estimation for the segmented sample by income level and children

Table 4 presents the results for the CL models by the subdivision group. Most coefficients in the conditional CL model were statistically significant, suggesting that respondents’ family travel preferences were heterogeneous (Pike et al., 2022). Train (2009) recommended using rational subdivision criteria to divide respondents into groups in order to identify the cause of heterogeneity within the group. Accordingly, in this study, the respondents were classified into two groups based on their income level and whether they had children, and it was confirmed that their preferences for domestic travel were different within each group.

Table 4 Estimation result of CL models by segmented groups according to income level and child.
Full size table

The lower-income group was defined as the respondents whose household monthly income ranged from 1 million KRW to 5 million KRW, and the higher-income group was 5 million KRW or more. Signs of the other coefficients were in line with those from the aggregate model except for the hotel level of ACCOMODATION in the lower-income group. The hotel coefficient in the lower-income group changed to negative and lost its significance. Also, preference for resort also indicated insignificance. This result means that the lower-income group neither prefers a hotel or resort/condo to the pension (base-level), which is unlike the aggregated sample and the higher-income group. Hotels and resorts/condos, which are known as high-end accommodations as compared to pensions, can be seen as relative economic burdens for lower-income people. Additionally, they did not prefer to stay for more than three nights, which is unlike the aggregated sample and higher-income group, and this occurrence could also be interpreted as the result of the burden of expenses. The coefficient for the MEDIA/SNS EXPOSURE attribute of the higher-income group was not significant, which is unlike the aggregated sample and lower-income group. This is presumed to be due to hotels and resorts/condos, which is where the higher-income groups mostly stay, being dominantly located in tourist attractions.

In the case of subgroup depending on a presence of children, the signs of the other coefficients were in line with of the signs from the aggregate model, but the two groups showed somewhat different preferences than the aggregate sample in the two attributes, which include TRAVEL DURATION and MEDIA/SNS EXPOSURE. The coefficient of more than three nights in TRAVEL DURATION did not show significance for the group without children. This result means that groups who are married without children do not prefer long family travel that is more than three nights and most prefer two nights and three days at the 10% significant level. In addition, the coefficient of low-level MEDIA/SNS EXPOSURE also failed to show significance. This result was different from the results for both the aggregated sample and the group with children. This may suggest that the group without children in our study, who are primarily adults, their insignificance of media/SNS exposure may stem from a similar reduced risk aversion. As adults, especially those without child-rearing responsibilities, they may be less deterred by the risks associated with traveling during a pandemic, irrespective of the destination’s popularity or perceived risk as influenced by media/SNS exposure.

MWTP estimation of each attribute

Table 5 shows the MWTP values for each attribute for the aggregate sample, and the groups were classified based on the respondents’ income level and the presence of children. It is not advisable to compare the two segments through the magnitude of each coefficient of the model. In this regard, the MWTP value helps to measure the relative importance between the proposed attributes (Haab and McConnell, 2002) and the levels. In other words, MWTP provides insight into what attributes/levels are more favorable or disadvantageous to the respondents based on changes within the options. In addition, MWTP refers to the amount of money that a customer is ready to pay for a particular feature, which is the level of attribute of a product. Also, it is the amount customers are willing to pay for an upgrade/change from A-level to B-level in addition to what they currently have to pay. Mathematically, it can be calculated by adding a negative sign to the value that is obtained by dividing the coefficient (βi) of the non-price attribute by the price attribute (βFEE). For example, MWTP for hotel in ACCOMODATION would be –(0.167/−0.0323) ≒ 5171. This amount indicates that the respondents are willing to pay an additional 5171 KRW to change the ACCOMODATION from pension to hotel. A negative (-) MWTP means that the subject level is less preferred by the respondent than the baseline. Therefore, the customer price must be reduced in order to compensate for the downgrade to the inferior level (Haab and McConnell, 2002). Similarly, negative MWTPs for TRAVEL TIME imply that the respondents do not prefer longer travel time.

Table 5 MWTP values of the aggregate sample and segmented groups by income level and child.
Full size table

First of all, the aggregated sample was segmented by respondents’ income level. The two groups had similar preferences/patterns for most attributes and levels, but there were markedly different preferences for the ACCOMMODATION attribute. For example, the lower-income group showed high MWTPs for pension and resort/condo and negative MWTPs for hotels and camping/glamping/caravans. On the other hand, hotels and resorts/condominiums showed the highest MWTPs with the higher-income class, while pension and camping/glamping/caravan showed negative MWTPs. This result implies that the lower-income group rejected the hotel as their accommodation, while the higher-income group showed MWTP values of 118,468 KRW when the attribute of ACCOMMODATION was changed from pension to hotel, all other attributes remain unchanged.

Second, the segmented group with the presence of a child also showed distinctly different preferences for ACCOMMODATION as well. The group without children had a high MWTP of 12,768 for resort/condo and a negative MWTP for pension. On the other hand, the group with children had the highest MWTP for pension within the ACCOMODATION level, which shows the greatest preference. The reason for these conflicting preferences is that families with children seem to prefer pensions because it provides a relatively independent space and they will be free to make more noise. Furthermore, the group without children cited TRAVEL TIME as the third most important attribute after MAIN ACTIVIES and ACCOMMODATION. On the other hand, the group with children ranked TRAVEL TIME as the second most important attribute, which represents that the most imperative attributes in family travel are in order of (1) MAIN ACTIVITIES, (2) TRAVEL TIME, and (3) ACCOMMODATION for the respondents with children.

Discussions

This study explored the attributes that are given weight by family travelers and is designed to understand how much tourists are willing to pay for the various attributes that make up domestic family travel. In order to accomplish the study goals, a review of prior studies, FGI, and expert consultations were conducted, and six choice attributes with some levels were derived. According to the study results, respondents preferred the most shorter travel time, two nights and 3 days for travel duration, resort/condo for their accommodation, low media/SNS exposure, and natural scenery or relaxing of the destination.

First, the negative sign for TRAVEL TIME implies that respondents attached a negative utility to longer travel times, which is in line with prior studies (Romagosa, 2020). The reason people prefer to travel to a nearby destination seems to be because there is a risk of contact and infection during travel, and if COVID-19 spreads in the destination, they can return home quickly. Second, the positive sign for TRAVEL DURATION indicates that the respondents prefer more to stay a longer duration than just one night and 2 days (base-level). This result is consistent with the results from the Korean nationals’ domestic travel itineraries, which announced that the average lodging travel was 2.47days in the 2019 Korea national travel survey (Ministry of Culture, Sports and Tourism, 2020). Accordingly, we can see that COVID-19 did not affect the travel period. Similarly, 47% of United States respondents said they would be willing to drive for a weekend getaway, preferring nearby places that are 2–4 h away (Minnaert, 2021). As a result, in the case of travel time and duration, it can be confirmed that family travelers in Korea and the United States have similar preferences for some family travel attributes during the pandemic period. Third, the positive signs for hotel and resort/condo refer to the respondents’ preference to stay in both hotel and resort/condo than in pension, which is the base category. These results are similar to those found in a survey of 2365 Americans who preferred hotels, vacation rentals, and resorts as accommodation during the pandemic in 2021 (Minnaert, 2021). On the other hands, this finding conflicts with the results of Mckinsey and Company (2020), reporting that residents from multiple countries feel more comfortable with short-term housing rentals than hotel/resort lodgings. From these results we can see that preferences for accommodation may vary by country/lifestyle. Fourth, in terms of MEDIA/SNS EXPOSURE, the study finding indicates that the respondents favor lower exposed destinations than highly exposed places. This finding is in accordance with previous literature (Fan et al., 2023; Karl et al., 2020), which concluded that COVID-19 is expected to increase the tendency of tourists to prefer visiting relatively less-known tourist destinations. The tourist preference for places with low MEDIA/SNS EXPOSURE can be due to the lack of options to choose due to tourist access controls at major/famous tourist destinations, but intrinsic motivations (fear of infection) will be the greatest. Therefore, it is necessary to reassure tourists that it is a safe place for tourists by following strict quarantine rules at famous tourist destinations. Fifth, in regards to MAIN ACTIVITIES at destinations, the study results conclude that natural scenery/relaxing is significantly preferred over both cultural/historical and entertaining/sports. This result is in line with the survey of Chinese tourists who showed a tendency to prefer outdoors, natural scenery, and quiet tourist destinations (Fan et al., 2023). As the COVID-19 crisis continues, the demand for health-oriented tourism related to physical and mental health is increasing. The travel industry may use this as a niche market to launch tourism products.

In order to explain the heterogeneity of preference and the MWTP variation among individuals, monthly income and the presence of children were taken into account for the alternative mode specification and further investigated by segmenting the aggregate sample into two groups. First, as a result of the subdivision by monthly income, the signs for the other coefficients coincided with the signs of the aggregation model except that the sign for the hotel coefficient for the lower-income class was changed. For example, the coefficient for hotel became negative and lost its significance, and the coefficient for resort/condo also became insignificant. In other words, the lower-income group prefers neither hotel nor resort/condo to the base-level of pension. Additionally, the lower-income group did not prefer to stay more than three nights and the reason is considered to be the burden of travel expenses. Since hotel rates are relatively higher than other lodging facilities, it can be seen that type of ACCOMMODATION has a direct correlation with the tourist income. This result supports Cai’s (1999) findings that the higher-income group tends to spend more money on lodging. As such, accommodation facilities are highly related to tourist income, so accommodation facilities will have to be careful about their marketing targets and offer appropriate pricing. Second, the main results for segmenting by children’s presence are follows. The signs of the other coefficients were consistent with of the signs from the aggregate model, but the two groups showed somewhat different preferences than the aggregate sample for the two attributes with TRAVEL DURATION and MEDIA/SNS EXPOSURE. The coefficient of more than three nights became insignificant, which indicates that the group without children does not prefer long family travel time. Moreover, the coefficient of lower media/SNS exposure for the childless group is insignificant. This result might be due to the group without children being adults who can travel under quarantine rules regardless of the destination’s popularity/crowdedness. On the other hand, the group with children can be interpreted as preferring a low media exposure travel destination in order to safeguard relatively vulnerable children from the risk of infection. Previous studies revealed that children play an important role in the family decision-making process (ZaLatan, 1998), so it is natural that this study shows different results depending on the presence of children. These findings are also consistent with a recent study by Álvarez-Díaz et al. (2023). They demonstrated that socio-economic characteristics, such as income and marital status, show different domestic tourist behaviors after the COVID-19 outbreak. Therefore, tourism developers need to be careful in targeting their visitors and will have to create a tourist attraction that both children and adults can enjoy.

The results of this study also revealed the MWTP for multiple attributes of domestic family travel. All the respondents who participated in this study showed the highest MWTP of 29,822 KRW in the natural scenery or relaxing of the destination.

The strong preference for natural scenery or relaxing showed the highest MWTP regardless of income level and children’s presence. This finding offers important management implications for tourism professionals, such as focusing more on outdoor tourist attractions when possible and planning natural travel packages to meet the tourist demand. The MWTP values also implied that the respondents perceived that the TRAVEL TIME is another imperative criterion when they engage in family traveling. Accordingly, tourism policy makers should prepare conditions for leisure and recreational activities around the city in the future.

Theoretical implications

This study provides significant theoretical implications. First, the CE method was applied as a valuation model, and MWTP was estimated in order to determine the factors of family travel that gave the most optimal utility. MWTP is useful for providing marketing implications because the relative importance of the attributes can be converted into currency units and compared with each other (Boto-García et al., 2022). In addition, by conducting a scenario analysis that is based on the probability of each level, it is possible to predict what purchasing behavior the potential consumers will reveal under specific conditions. Therefore, reviewing the amount of WTP at a realistic level and expanding or shrinking the product may result in increased utility for both the suppliers and the consumers from an economic point of view. Second, this study bridges the gap with knowledge and information between the public opinion and practice. According to the media, camping/glamping/caravan have become very popular, so new keywords that are related to camping have emerged due to the sales of camping equipment increasing. However, the study results reported a negative MWTP for camping/glamping/caravan, which implies a sense of reluctance with using the facility. The camping/glamping/caravan categories can be seen as remarkably unfavorable compared to other accommodation categories and would seem to contradict the media’s claim of camping/glamping/caravan popularity during COVID-19. The study result provides an example where the popularity of public opinion does not necessarily have a big impact on reality, especially for families. As such, this study is meaningful in showing the real-world influence of camping/glamping/caravan through empirical data.

Managerial implications

The practical implications are as follows. First, it is common to travel with an unspecified number of people who are using the same product with conventional package tours. However, these types of tour packages do not match the importance of hygiene and safety that has emerged since COVID-19, and sales are expected to be low even when the pandemic ends at some countries (Chebli and Said, 2020). Domestic travel has contracted due to the nationwide social distancing campaigns. Under these circumstances, the demand for small package tour products in order to travel with close people, such as family and relatives, will relatively increase. Therefore, large travel agencies that mainly sell package tours need to prepare products that provide safer itineraries. This study is a survey of preferences for small-scale tourists, so it can be used as preliminary data in order to develop new products in accordance with the consumer’s demand. Second, the preference of hotel and resort for the higher-income group as an accommodation is proven from the study results. Accordingly, they may focus on premium products that ensure independence and minimum contact with other guests. Also, those in the hotel and resort/condo category should seek a way to manage the shared areas, such as, the lobbies, restaurants, and elevators. The study results also found that tourists with children preferred pension as their most preferred accommodation. It would be good for pensions to prepare a kids’ cafe style or a room with a swimming pool that targets families with children. In the case of camp sites, the study results suggest a strong preference for nature-based destinations in the post COVID-19 world. Therefore, permission or/and approval for the new camp sites in the future should be made mainly for nature-friendly places that can create various movement lines, such as seas, mountains, and streams, and reduce the population density. In addition, although interest in the camping/glamping/caravan category has dramatically increased in this pandemic situation, the study results revealed family tourists had the lowest preference among various accommodations. Therefore, it is necessary to prepare measures to entice families to the campsite, such as creating a safe and fun place for children to play or operating an experience activity program. Moreover, in order to become a safe place, it is necessary to promote safety rules to prevent the spread of COVID-19, such as safety travel guidelines for visitors. Particularly, as some public facilities such as shower facilities and toilets are used, it is necessary to devise measures to widen the social distance between users, such as setting rules between users or different check-in and check-out times.

Limitations and future studies

This study has some limitations that suggest the future research arenas despite its contribution to the literature. First, the samples in this study were collected from residents in Korea, so it is difficult to apply the results to other regions. However, we believe that this influence is not limited to a specific region as domestic tourism is thriving around the world (Volgger et al., 2021; Wen et al., 2020). Consequent to the demographic particularities of the current study, future investigations could benefit by broadening the geographic scope of the sample. This multi-country approach would not only enrich the dataset but also foster a more comprehensive understanding of the nuanced attributes pertinent to family domestic travel that may vary across different cultures and regions. Second, further research could extend this work by examining how tourist preferences diverge in correspondence with different stages of social distancing. For example, this is a study that analyses the differences in responses of respondents according to social distancing levels. This survey was conducted during the 2nd stage of social distancing in Korea. It can be hypothesized that heightened levels of social distancing, which impose stricter restrictions on individuals’ activities, may yield significant shifts in tourist preferences. The systematic study of these shifts could yield meaningful insights into the adaptive behaviors and decision-making processes of individuals amid different stages of an epidemic. Third, the most important part of the various CE’s procedures is the extraction of the appropriate attributes and the determination of the level where the characteristics of those attributes can be effectively reflected. In order to secure the validity of this study, the attributes and the levels were determined after carefully reviewing previous studies on family travel and the domestic travel survey report, which is where the Korean government annually investigates and publishes domestic travel behavior. Future research could advance on these grounds by adopting a more diversified approach in the extraction and representation of attributes. This could be achieved by conducting a comprehensive meta-analysis of prior studies, beyond the context of family travel and domestic tourism. Such a study would not only broaden the range of potential attributes to be considered but also provide a more granular insight into the various levels of these attributes across different travel behaviors.