Shared Accommodation Services in the Sharing Economy: Understanding the Effects of Psychological Distance on Booking Behavior

Abstract

The sharing economy has risen rapidly in the past decade. The development of shared accommodation encourages more hotels and guesthouses to attract customers through online accommodation-sharing platforms, which has become a meaningful way to fight against the competition of the traditional hotel industry. In this condition, what the hosts are concerned about most is how to attract customers’ attention through the platform display to increase reservations. Based on construal level theory, this paper explores how hosts’ information displayed on online accommodation-sharing platforms determines consumers’ booking behavior by influencing their psychological distance. We use machine learning methods to mine the raw data and extract the representational factors of psychological distance. Based on the data-driven behavior decision-making approach, we collected valid large-scale fine-grained secondary actual consumption data from Airbnb, the world’s leading online accommodation-sharing platform, and scientifically and intelligently processed the data using machine learning methods, then tested the hypotheses using the regression analysis software STATA15. Our findings suggest that both social distance and temporal distance have a negative impact on booking behavior of guests. In detail, subject diversity, perspective taking, and facial attractiveness in the dimension of social distance positively influence guests’ booking behavior; instant bookable in the temporal distance dimension positively affects booking behavior, while response time has a negative effect. This study contributes to the literature by empirically examining psychological distance in the booking behavior of guests in shared accommodation through the processing and analysis of actual consumption data. The findings have important practical implications for how shared accommodation service providers and sharing economy platform managers can operate better.

1. Introduction

The sharing economy refers to a series of economic activities that use modern information technology, such as the internet, to merge massive and decentralized resources and meet diversified needs, with the sharing of usage rights as the main feature [1]. In recent years, the concept of sharing economy has been extended to various industries, giving rise to a series of sharing services, such as shared travel and shared charging, which have become an integral part of people’s lives. In particular, sharing in accommodation services has emerged, relying on the Internet and mobile terminals to closely connect hosts and guests with online platforms to provide room reservations for remote customers through real-time customization services. Shared accommodation has become an emerging economic sector that China is vigorously supporting, with the number of Chinese residents using shared accommodation services exceeding 80 million in 2021 and the scale of shared accommodation transactions increasing from 12 billion in 2017 to 15.2 billion in 2021 [2,3]. Shared accommodation refers to individuals (hosts) renting out excess occupied space they own to other individuals (guests) seeking short-term accommodation through an intermediary (online accommodation-sharing platform) [4]. Accommodation-sharing platforms are one of the most important domains within the sharing economy and are becoming increasingly popular with consumers in the tourism and travel markets [5]. Founded in 2008, Airbnb has become almost synonymous with the sharing economy [6], and is now the most successful online accommodation-sharing platform, offering short-term room rentals by more than 900 million hosts in 220 countries [7].

While celebrating the brilliant gains of the online accommodation-sharing market, its development also faces some difficulties, particularly in consumer engagement. First, the issue of trust is an important factor hindering the growth of shared accommodation. In addition to facing the same economic risks as the traditional hotel market, consumers living with strangers or sharing space also face threats to their safety [8] and perceived purchase risk can have a negative impact on online purchasing behavior (purchase intention) [9]. Negative reports about shared accommodation and information asymmetries also contribute to guests’ distrust of hosts. In addition, due to competition from traditional hotels and a lack of traditional consumer acceptance, it is increasingly vital for online accommodation-sharing platforms to improve performance and attract more room bookings [10]. Moreover, in the context of regular COVID-19 prevention and control, market information changes rapidly and bears great uncertainty [11]; the COVID-19 epidemic has negatively impacted suppliers (hosts) and accommodation-sharing platforms [12,13], so hosts should optimize the information they present on online platforms in order to attract guest bookings to increase revenue. Enticing consumers to book rooms through the online presence of hosts is an important motivation for hosts to use online accommodation-sharing platforms and is critical to the sustainable operation of shared accommodation. In shared accommodation scenarios, consumers can only make initial judgments about a room through visual impressions when they travel long distances to the place of housing, and they can only have a real experience of the accommodation service when they stay in person for a period of time. This is similar to what the online shopping literature suggests: the psychological distance created by the inability of consumers to see and touch the goods up close can weaken the psychological basis of cognition, emotion, and volition in traditional consumption patterns, and can affect consumers’ purchase decisions to some extent [14]. Thus, the inability to visit the property and talk face-to-face with the host during the online booking process can also lead to a sense of psychological distance for the guest, which can impact the booking behavior of shared accommodation. Further, there is also literature that confirms the negative impact of psychological distance on guest loyalty in shared accommodation [4]. This psychological distance is likely to exacerbate the information asymmetry between consumers and hosts [15], consumers’ distrust of service providers (hosts and listings) [8,16], the sense of discrepancy between expectations and authentic experience [17] and less host-guest interaction [4] and so on will affect consumers’ booking behavior, or even give up booking [4]. This is something that hosts and accommodation-sharing platforms do not want. During the booking process, the information displayed by the host on the platform can facilitate consumers to infer information about the host and the attributes of their home, thus reducing the psychological distance. Therefore, eliminating the psychological distance between consumers and hosts requires hosts to put effort into the content and interaction of information displayed through the medium of accommodation-sharing platforms. Moreover, as one of the social media sites, the accommodation-sharing platforms choose the right marketing content that allows it to build a good relationship with its customers [18]. This is also in line with the logical view of customer orientation, where the customer is the subject of value creation [19]. Accordingly, the platform should actively understand which information reserved by the hosts will stimulate consumers’ psychological distance and thus impact their booking behavior. Then, at the practical level, they can amplify consumers’ concerns through reasonable interface design and information display, and bring consumers and hosts as close to each other as possible to facilitate booking transactions.

Although relevant literature has examined the impact of host-related factors [20,21], guest-related factors [22,23], and platform-related factors [24,25] on consumer’s booking behavior, few studies have focused on factors related to the psychological level of the individual consumer. To date, we know very little about how the information presented on online accommodation-sharing platforms affects the perceived psychological distance between guest and host and the booking behavior of guests, leaving room for this paper to fill in the research gaps. Therefore, this paper focused on understanding the impact of the psychological distance (time distance and social distance) generated by the information displayed by the host in online accommodation-sharing platforms on the guests’ booking behavior.

The research question in this paper is “what are the effects of temporal and social distance on the booking behavior of guests in accommodation-sharing platforms?” Based on the construal level theory and psychological distance literature, the purpose of this paper is to explore the information that characterizes psychological distance in online shared accommodation platforms, and to investigate the mechanism of psychological distance generated by the information displayed by hosts on online shared accommodation platforms on the booking behavior of guests.

To solve this problem, we investigate the impact of psychological distance, including social distance and time distance, on consumers’ booking behavior on online accommodation-sharing platforms. Our research provides several contributions to the extant literature. First, this research incorporates consumer psychological distance, which has been neglected in previous literature, into studying consumer booking behavior in an online accommodation-sharing context. Second, it seeks the applicability of psychological distance and construal level theory in the sharing economy and practically responding to online accommodation-sharing platforms to encourage guests’ booking behavior. Third, this research expands on the representation of psychological distance in online accommodation-sharing [26] by mining and extracting information elements that may affect guests’ booking behavior. Finally, based on the data-driven behavior decision-making approach, this paper provides a new perspective on online accommodation-sharing research methods, extending previous research [4] by combining machine learning methods with secondary consumption data from shared accommodation online platforms and exploring both text and image information.

The remainder of this paper is organized as follows. First, we provide a literature review of shared accommodation booking behavior, followed by a presentation of the theoretical foundations. Then, the research model and hypotheses are presented in detail. Next, we offer the variables and research methodology. In Section 5, the research results are discussed. Finally, the theoretical contributions and implications, limitations, and possible future research directions are discussed at the end.

2. Literature Review

2.1. Booking Behavior in Shared Accommodation

Shared accommodation, also titled “P2P accommodation”, “home-sharing”, or “short-term rentals” [27], is a new industry born from the Internet. It refers to the business model in which residents or organizations in different tourist destinations temporarily rent out their unused houses and facilities for others through online platforms to gain economic benefits [28]. Peer-to-peer (P2P) accommodation sharing platforms; that is, platforms that mediate the flow of goods and services between suppliers and consumers, one of the many business models in the sharing economy, are characterized by a high degree of both platform intermediation (i.e., the degree to which the platform is involved in the exchange) and sociality (i.e., co-presence and interaction with others in society) [29]. With the development of shared accommodations and their platforms, shared accommodation rooms are more popular among consumers than traditional hotels [5]. Take Airbnb as a thriving case, which operates in more than 34,000 cities in 192 countries and has more than 5 million listings worldwide [30]. When booking on the Airbnb platform, guests use the online platform to find a satisfactory property and make a reservation. The complete booking process consists of the following steps [31]: (1) the guest browses the online accommodation-sharing platform (website or app) to find a target property according to their needs, sends a booking request, and then attempts to contact the host through the platform; (2) if the guest’s booking request is approved, both parties reach an agreement to stay, and the guest can enter and may meet face-to-face with the host for a period of time; and (3) finally, the guest checks out and reviews the accommodation experience.

In an online accommodation-sharing scenario, guests’ booking behavior and intention are influenced by various factors [32]. We summarized these factors in

Table 1. Summary of extant research on booking behavior in shared accommodation.

Topic	Key Influencing Factors	Reference
Guest-related factors	Trust	[34] [35] [36] [37] [38] [28]
	Attachment	[37]
	Self-brand association and sense of power	[39]
	Attitudes, subjective norms, perceived value	[40]
	Perceived value and authenticity	[23] [17]
	Perceived authenticity, e-word of mouth, price sensitivity	[41]
	Guest-to-guest interaction	[42]
	Perceived risk	[43]
	Identification with service providers	[44]
	Guest emotions and satisfaction	[22]
Host and listing related factors	Advertisement	[39]
	Art attributes	[45]
	Service quality	[46] [47]
	Host profile picture	[48]
	Convenience facilities	[49]
	Host-guest relationship	[49]
	Dynamic pricing	[50]
	Price	[51]
	Service flexibility and knowledge level	[20]
	Social interaction	[52]
	Listing picture color	[21]
Platform-related factors	Website design	[49]
	Website aesthetics	[53]
	Rhetoric of belonging and legal integration	[28]
	Platform regulation and authenticity	[54] [55]
	Reputation	[24]
	Brand, existential, and intrapersonal authenticity	[25]

. At the same time, the above research presents a wealth of findings to help people understand the key elements of shared accommodation bookings. However, there is a shortage of literature investigating the effect of psychological distance on consumer booking behavior on online accommodation-sharing platforms. The literature [4] that examines the engagement behavior of shared accommodation guests from the perspective of psychological distance has only explored the impact of social and spatial distance between hosts and guests on the dimension of guest loyalty. However, what information (text, photos, digital features, etc.) influences the psychological distance of guests when browsing shared accommodation online platforms and how psychological distance affects booking behavior have not been fully explored or explained. It has been documented that host-displayed information in shared accommodations significantly impacts consumer booking behavior [33]. It is increasingly important to explore the antecedent variables that influence consumer psychological distance [32]. In addition, most existing studies use questionnaires or experimental methods, with fewer studies focusing on the actual consumption behavior of guests of online accommodation-sharing platforms in the context of big data. Thus, based on the above review, this study examines the antecedents that may make consumers psychologically distant in online accommodation-sharing platforms and the impact of different dimensions of psychological distance (social distance, temporal distance) on consumer booking behavior.

2.2. Construal Level Theory (CLT) and Psychological Distance

According to CLT, individuals process events at a higher level of construal when they perceive a greater psychological distance between themselves and the event. Psychological distance is one of the core concepts in the construal level theory [56]. As a subjective experience of how far things are from oneself at this moment, psychological distance consists of four dimensions: temporal distance, spatial distance, social distance, and hypotheticality, which constitute the basic framework of psychological distance [57,58]. Temporal distance refers to the distance in time to the present that an individual perceives an event to have occurred. Spatial distance refers to how far apart in space an individual perceives things to be. Social distance refers to how close or far one perceives something to be about oneself. Hypotheticality refers to the possibility of an event occurring or the distance from reality. The four dimensions of psychological distance form a unified structure and form the same significance in the minds of individuals [58].

It is important to note that although the psychological distance is a multi-dimensional concept, this study only focuses on the impact of temporal distance and social distance on booking behavior in an online accommodation-sharing context, as these are factors that the host can control. This study does not focus on spatial distance and hypotheticality because the spatial distance between hosts and guests is objective and not affected by the hosts’ efforts; in addition, the hosts and properties in online accommodation-sharing platforms are actual.

The published literature, construal level theory, and psychological distance, which serve as theoretical backgrounds, have been used to study consumer behavior. Research has established that various dimensions of psychological distance (temporal, spatial, and social distance) influence consumer predictions, assessments, and behavior [26]. The literature on online retailing posits that psychological distance has a more significant impact on consumers’ online trust and purchase intentions relative to product quality, perceived risk, and firm size [59], which supports the need to study the impact of psychological distance on consumer decisions. Furthermore, increased psychological distance causes consumers to adopt high levels of interpretation to guide their judgments and behavior [56,58], which can lead to stereotypes about merchants [60]. Therefore, when the psychological distance is more significant, guests’ negative stereotypes of hosts can deter them from booking.

Similar to previous studies, psychological distance is also adaptive for understanding consumer booking behavior in shared accommodation. First, psychological distance has been applied to various online shopping scenarios, including online retail, e-commerce domains, online sales promotion and the sharing economy [1,16,58,61]. Second, psychological distance contains multiple dimensions [57,58], which provide a multidimensional perspective for understanding the underlying psychological mechanisms of consumer behavior and an integrated approach for researchers to identify the key factors influencing consumer behavior from different perspectives. Finally, combined with construal level theory, psychological distance can explain how various dimensions of psychological distance in the online booking process affect booking behavior, which is the purpose of this paper.

Among the studies of psychological distance on consumer behavior, some have discussed the interaction of temporal and social distance on consumption intentions and recommendations from others, with social distance and temporal distance leading to more incredible preference shifts when consumers make decisions about future consumption [62]. Other studies have found that important factors influencing online purchase decisions are physical location, similarity, and familiarity [63]. Consumers have high purchase intentions when they perceive a small social distance from the target product [64]. High levels of construal of a good or manufacturer produces greater psychological distance and lower purchase intentions [57]. This also shows that psychological distance is an important antecedent of consumers’ willingness to purchase online.

3. Research Model and Hypotheses

3.1. Effects of Social Distance on Booking Behavior

In this research, we suggest that the social aspect of hosts is best demonstrated by the information on their personal pages on online accommodation-sharing platforms. According to research statistics, consumers view host personal information at least once in nearly half of all travel activities, and 68% of these views occur before booking [65]. Moreover, first-time users of online accommodation-sharing platforms are 20% more likely to view a host’s profile before booking than older users [65]. In this study, we argue that the personal information displayed by the hosts, such as the hosts’ personal photographs and the text of the host’s self-introduction, can influence the social distance that the guests perceive from the hosts.

3.1.1. Subject Diversity

In online accommodation-sharing platforms, hosts will introduce themselves on their personal homepages to reduce uncertainty and the perceived risk for guests and increase guests’ trust in the hosts. The amount of information in a host’s profile directly reflects the willingness of the host to present personal information to the guest. A host willing to give a profile will provide detailed information about their hobbies, emotional state, occupation, life experience, and other information; as a result, the guest will be able to obtain valuable information and communicate better with the host. The more balanced the topics contained in the host’s self-introduction text, the greater the amount of personal information shared by the host and the more extensive the information presented, reflecting the host’s willingness to showcase himself up to the consumers, not intentionally withholding information about himself, and his open and friendly attitude. The literature on interpersonal relationships points out that sharing with others seemingly by accident can reduce the social distance between people, strengthen relationships, and bring two people closer together [66]. Simply traveling in the same direction and eating the same food can foster trust, cooperation, interpersonal attraction, and relationship satisfaction [62,67]. Sharing extraordinary experiences can also enhance social intimacy between people [63], which brings more together socially.

Therefore, if the text of the host’s self-introduction on the shared accommodation platform involves comprehensive and diverse topics, it can make the guests perceive the openness and social willingness of the host; the sharing in the host’s self-introduction can close the social distance between host and guest. The guest can also get to know the host more comprehensively and reduce the perception of uncertainty [64], thus bringing the social distance between host and guest closer. In summary, subject-rich self-introduction brings the social distance between host and guest closer and may further promote guests’ bookings. Therefore, the following hypothesis is proposed:

H1: 

The subject diversity of hosts’ self-introduction is positively associated with the booking behavior of guests.

3.1.2. Perspective Taking

Perspective taking refers to the active, cognitive effort one puts into putting oneself in the shoes of others and understanding or accepting their perspectives [33]. Studies of online reviews have found that different reviewers have different perspective tendencies, with self-focused reviewers tending to vent negative emotions and look for financial gain. In contrast, other-focused reviewers tend to care about and help other consumers [68]. Reviewers use second-person pronouns (e.g., you, your) when speaking from someone else’s perspective and first-person pronouns (e.g., I, my) when speaking from their own perspective. Using the second and third person is a sign that the speaker is socially engaged or aware of society. Personal pronouns reflect the focus of attention and the subject, and the use of personal pronouns is closely related to perspective taking [69].

Perspective taking is widely recognized as one of the important factors in enhancing prosocial behavior and promoting social functioning [70]. When an individual thinks from another individual’s perspective, there is a greater overlap between their mental representations and those of others, and overlapping mental representations can generate interpersonal intimacy [33]. Therefore, perspective taking can help establish interpersonal relationships and foster intimacy with people, especially between strangers who meet for the first time [33]. Furthermore, perspective taking can reflect a person’s caring, empathic, and helpful behaviors toward others [71]. For example, during negotiations, a more extensive perspective may lead to greater consideration of what will satisfy the other side, thereby enhancing the opportunity to reach an agreement that will result in a more creative solution [72]. Reviewers in an online review environment who take the perspective of a potential consumer are more likely to be considerate of their viewers and assist them by creating the reviews they want [73]. Similarly, in the context of online accommodation-sharing platforms, hosts who write self-introductions from the consumer’s perspective may be more willing to help their guests.

Many papers have related the use of second-person pronouns to other-focus, and first-person pronouns to self-focus [69]. Previous research has measured perspective taking via the relative frequency of second-person versus first-person pronoun use in texts [73]. Likewise, we quantify perspective taking similarly, with higher values indicating that hosts are more concerned about their guests and more likely to think from their guests’ views.

In summary, we argue that using high perspective taking in the host’s self-introduction text in online accommodation-sharing platforms is likely to generate interpersonal intimacy between guests and hosts, shorten their social distance, and facilitate guests’ bookings. Therefore, the hypothesis is formulated as follows:

H2: 

Perspective taking is positively associated with the booking behavior of guests.

3.1.3. Facial Attractiveness

According to interpersonal attraction theory, many factors contribute to interpersonal attraction, with the appearance of attractiveness being the most influential factor in causing people to be initially attracted. Appearance attractiveness is also a stereotype or ‘halo effect’, meaning that attractive people are perceived to have better social skills, intelligence, helpfulness, and good qualities such as honesty and kindness than unattractive people [74].

Some studies have suggested that people with an attractive appearance are more likely to receive preferential treatment from others or society. For example, politicians with high physical attractiveness receive more votes in elections [74]. Women with good appearance and body image receive relatively higher incomes [75]. Consumers give higher service quality ratings to attractive service workers [76]. A study found that attractive salespeople gain more customer trust [77]. Furthermore, studies have shown that facial photos of merchants on a website increase their credibility and that customers’ perceptions of their facial attractiveness increase customer trust [78]. The research on shared accommodation responds to the statement that facial appearance is an important factor affecting guests’ psychology and cognition [79].

On this basis, this study argues that the guest will perceive the appearance attractiveness of a host in an online accommodation-sharing platform as the host having better social skills, bringing the guest and host closer together socially and, thus, promoting guests’ booking behavior. Therefore, we present the following hypothesis:

H3: 

The facial attractiveness of a host’s photo is positively associated with the booking behavior of guests.

3.2. Effects of Temporal Distance on Booking Behavior

As one of the dimensions of psychological distance, temporal distance affects consumers’ decision-making behavior and is used in studies on behavioral decision-making [66]. Before making a purchase, consumers are more concerned with the ethical value of the product, while in practice, they are more concerned with the feasibility of purchasing the product, such as the ease of access, the time pressure and mood of the purchase, and the crowdedness of the shopping area [80]. Research in the field of e-commerce has found that consumers’ perceived usefulness of online reviews decreases with increasing temporal distance, has an impact on consumers’ perceptions and emotions, and may have an emotional impact on other consumers’ decisions when making purchases because of further diffusion effects [81]. Bar-Anan et al. (2006) found that the time it takes to receive an item (temporal distance), the product description via a computer (social distance), and the probability of not receiving the product (degree of hypothesis) all produce higher psychological distance and influence consumers’ online shopping behavior [67].

Temporal distance is prevalent in consumer psychology [82], while for firms, psychological distance is a controllable variable, as firm managers can take steps to increase or decrease consumer psychological distance depending on the desired response [66]. For example, Amazon Prime reduced the temporal distance by shortening delivery times, thus reducing consumer impatience in waiting. Similarly, firms may increase the temporal distance to increase consumers’ desire for the product and the perception of luxury [83]. For example, there is a three-year wait to obtain a Kelly bag from Hermes. In studies of online retailer platforms, it has been found that merchants offering timely responses or quick delivery to customers can reduce psychological distance, increasing consumers’ willingness to buy [84]. Recent studies point out that reducing delivery and shipping times or offering buy-now-pay-later features can reduce the temporal distance dimension of consumer psychological distance. For example, Darty Plus offers a two-hour deliverable service, and the Paypal platform offers a home delivery service that allows consumers to purchase immediately and then pay for it later [85].

In online accommodation-sharing platforms, response time is the time a guest waits after starting a conversation with a host until receiving the first response from the host. One study found that the waiting time for order confirmation on the Airbnb platform affects customer behavior and that timely order confirmation promotes positive interactions between hosts and guests [8]. Shorter confirmation wait times can improve the booking experience for guests and positively influences their decision-making. In addition, the host’s response time reflects the host’s level of hospitality.

Therefore, we believe that if hosts in shared accommodations respond to guests’ messages instantly or make them instantly bookable to reduce the waiting time for guests, this will reduce the perception of temporal distance for guests and promote guests’ booking behavior. The following hypothesis is thus proposed:

H4: 

The response time is negatively associated with the booking behavior of guests.

H5: 

The availability of instant bookable is positively associated with the booking behavior of guests.

In addition, considering that factors such as the reputation of the host and the characteristics of the listing also influence the booking behavior of the guest [86], we include them as control variables in the study model. The control variables include the superhost, room type, price, rating score, and accommodates.

We use the following model to demonstrate the research framework and logic of this paper. We explore the effects of the temporal distance dimension and the social distance dimension of psychological distance on the booking behavior of guests, and the research model is shown in Figure 1.

4. Method

4.1. Datasets and Preprocessing

Considering the accuracy and objectivity of secondary data, the use of secondary data as a data source for empirical analysis is gradually becoming a trend [87]. We collected listing data in Beijing and Shanghai from Inside Airbnb (http://insideairbnb.com. accessed on 5 October 2021), and the time range is from September 2020 to September 2021. We chose these two cities for the study because Beijing and Shanghai are two of the most popular tourist destinations in China and the two cities offer the highest number of rooms on online accommodation platforms [88]. The raw data include publicly available information about hosts and listings on Airbnb (i.e., response times, room types, host photo URLs, room type, etc.). The dataset does not contain photos of the hosts and houses, only the corresponding URL links; therefore, we used a Python crawler to crawl the related photos based on the URLs. We perform the following preprocessing on the collected data and weed out invalid data to improve data usability. (1) First, we removed data with empty house descriptions and host self-introductions and text data containing only punctuation marks, then removed data with empty and invalid photo URLs. (2) We used the Python language recognition package to remove house descriptions and host self-introductions as non-Chinese text. (3) We used the Face++ face recognition API to remove photos that do not detect faces and the corresponding data. The remaining 8752 data points were included in the analysis after data cleaning. We present an example of variable information on the online accommodation-sharing platform in Figure 2.

It is worth mentioning that the data in this paper were collected from September 2020 to September 2021, which coincides with the period of the COVID-19 pandemic in China. The development of the shared accommodation industry has also been impacted by the epidemic, but from the data of recent years, the transaction value of shared accommodation in China in 2020 and 2021 is 15.8 billion yuan and 15.2 billion yuan respectively not much different from 16.5 billion yuan in 2018 [2]. Looking at the transaction data for the last five years (2017–2021), excluding the 22.5 billion yuan in 2019—a period of rapid growth for the sharing economy—the transaction value is basically flat in other years [2]. The user size of China’s sharing economy has increased from 69.83 million in 2020 to 81.41 million in 2021, a 16.6% increase in user size [2,89]. Therefore, we have reason to believe that the time years of data collection in this paper are generally representative of China’s shared accommodation operations, and the conclusions of this study are of little relevance to the epidemic, so this study is meaningful.

4.2. Variables Selection

Table 2. Descriptions of variables.

Dimension	Variable	Measures
Dependent variable	Booking behavior	Quantify booking behavior in terms of the number of orders, which is expressed as the number of reviews divided by the review rate.
Social distance	Subject diversity	The higher the entropy indicates that the text contains multiple topics simultaneously.
	Facial attractiveness	It is measured by the face value score, taking the value range [0, 100]; the average of the male and female scores for that face value.
	Perspective taking	Second-person pronouns as a proportion of the sum of first and second-person pronouns.
Temporal distance	Response time	within an hour = 1, within a few hours = 2, within a day = 3, and a few days or more = 4.
	Instant bookable	Instant bookable = 1, non-instant bookable = 0.
Control variables	Superhost	A host with the superhost badge is labeled 1 and 0 otherwise.
	Room type	Shared room = 0, private room = 1, and entire home = 2.
	Price	The price of the listing.
	Rating score	Ratings of the listings.
	Accommodates	The number of guests accommodated by a given listing.

summarizes the description of the variables. For the dependent variable, as the Airbnb platform does not directly present information on the number of room bookings, according to previous related studies, researchers usually use the number of reviews to estimate the number of room bookings, specifically using a 50% review rate to translate the number of reviews [31]. Ye et al. [90] note that platforms such as Airbnb only allow room renters whose transactions are completed to post. Therefore, the number of reviews can be expressed and validated as a linear function of the number of bookings, assuming that the probability of a review is constant. This study, therefore, suggests that the number of reviews on Airbnb can be used as a proxy variable for the number of bookings to characterize the actual booking behavior of guests. It has also been found that the number of online sales has a linear relationship with the number of reviews, so the number of reviews can be used as a proxy variable for the number of online bookings [91]. Therefore, our study uses the cumulative number of reviews received by a property divided by the review rate as the number of orders.

Response time indicates the waiting time from the time the guest initiates a conversation with the host until the first response is received from the host. Response times are divided into four categories: within an hour, within a few hours, within a day, and a few days or more. The four types of response times are marked 1, 2, 3, and 4, respectively. Instant bookable, also known as flash bookings, are booked instantly without waiting for confirmation from the host. Instant bookable and non-instant bookable are marked as 1 and 0, respectively.

We used the LDA topic model in natural language processing to mine the topics contained in the host’s self-described text and the probability size of each topic in the text. The LDA model is a relatively new and mature tool that researchers widely use. There is related literature using LDA to study the emotional characteristics of Airbnb green users [92]. Entropy is applied to measure the diversity of potential topics in each self-description text. Previously, some scholars used information entropy to explore the uniqueness of information in online reviews [93]. Claude Shannon originally proposed Shannon entropy to measure the uncertainty of text strings [94]. The more balanced the probability of the occurrence of letters in a string, the greater the uncertainty and the greater the entropy. Similarly, the flatter the possibility of topics contained in a text, the higher the entropy, indicating that the text includes several of these topics simultaneously, i.e., the text description contains an increased number of topics rather than depicting a particular topic. Thus, the formula for calculating the subject diversity of a text is shown in Formula (1).

$$Entropy\left(d\right)=-{{\displaystyle \sum }}_{i=1}^n{p}_{i}log\left(p_i\right)$$

(1)

where $Entropy\left(d\right)$ denotes the topic entropy of text d, n denotes the total number of topics in each text, and p_i denotes the probability that text d belongs to the topic i.

It has been shown that people who habitually use first-person pronoun expressions focus more on themselves. In contrast, using second-person pronouns is associated with focusing on others [69]. Previous studies have used the relative frequency of second-person pronouns versus first-person pronouns to measure perspective taking [95,96]. We also used the following Formula (2) to quantify perspective taking; higher values indicate a higher level of concern for others and a greater likelihood that the host thinks from the tenant’s perspective. Referring to previous studies, we added 0.0001 to the denominator of the formula to avoid zero in the denominator. In recent years, natural language processing (NLP) combined with machine learning models has been gradually introduced into text research [97]. To better measure perspective taking, we used TextMind, developed by the Institute of Psychology of the Chinese Academy of Sciences to implement statistics for the first- and second-person pronouns in self-introduction texts, which is tailored to the linguistic characteristics of simplified Chinese concerning LIWC2007 and the orthographic Chinese C-LIWC thesaurus, to implement a solution for linguistic analysis of a text. Based on the input text, TextMind automatically analyzes 103 relative features of the text, including the number of various lexical words, the number of various period words, the number of various punctuation marks, etc.

$$PT=\frac{2nd Person Pronouns}{1st Person Pronouns+2nd Person Pronouns+0.0001}$$

(2)

For the extraction of facial attractiveness variables, we use the face value score of face analysis results in Face++, a leading global face recognition platform that provides technical services, including face recognition, face detection, and face analysis. In previous literature, Face++ has often been used for facial emotion recognition and analysis [8]. In this paper, we used the Python program and Face++ API interface to detect the face of the host in the photo and obtain the face value score, which takes the range of 0–100. A higher value represents a higher and more attractive face value.

In addition to the main variables in this study, other factors significantly affect the booking behavior of guests but are not the focus of this study; thus, we placed these factors as control variables in the research model.

For the control variables, the rating score is the average review score provided by guests. On Airbnb, guests can rate items in six areas: accuracy, location, communication, stay, cleanliness, and value. As in previous studies [8], this study does not distinguish between the six different aspects of the ratings but uses the overall rating. A superhost means the owner of the property has met a series of requirements set by Airbnb; these hosts are responsive and have high ratings, frequent reservations, and few cancellations. Hosts with superhost identification are marked as 1; otherwise, they are marked as 0. The accommodates refers to the number of guests that a given listing can accommodate. The price is the list price of the room. There are three types of room types: shared room, private room, and entire home, marked as 0, 1, and 2, respectively; private room are used as the reference variable, and two dummy variables are set: shared room and the entire home.

5. Data Analysis and Results

5.1. Descriptive Results

First, descriptive statistics were performed for all variables, and the statistical analysis results are shown in

Table 3. Statistics of the variables.

Variable	Mean	Std. Dev.	Min	Max
Booking behavior	32.87	48.292	0	248
Price (Ln)	6.221	0.779	4.7	8.536
Rating score (Ln)	3.976	1.528	0	4.615
Superhost	0.394	0.489	0	1
Accommodates	3.759	3.169	1	16
Room type	1.549	0.532	0	2
Facial attractiveness (Ln)	4.148	0.173	3.694	4.453
Subject diversity	0.887	0.370	0.140	1.557
Perspective taking	0.210	0.312	0	0.997
Response time	1.262	0.747	1	4
Instant bookable	0.615	0.487	0	1

. The mean value of booking behavior is approximately 33, and the number of rooms booked varies from 0 to 248. Then, a correlation analysis was performed for all independent variables. The results of the correlation analysis are shown in

Table 4. Correlations between the variables.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
(1) Price (Ln)	1.000
(2) Rating score (Ln)	0.029	1.000
(3) Superhost	−0.004	0.208	1.000
(4) Accommodates	0.653	0.046	−0.051	1.000
(5) Room type	0.369	0.138	−0.022	0.305	1.000
(6) Facial attractiveness (Ln)	−0.004	−0.021	−0.013	−0.043	0.001	1.000
(7) Subject diversity	0.021	0.046	0.006	0.037	−0.027	−0.012	1.000
(8) Perspective taking	−0.004	0.003	0.058	−0.035	−0.019	−0.027	0.026	1.000
(9) Response time	−0.049	−0.053	−0.210	0.022	0.027	−0.008	−0.022	−0.068	1.000
(10) Instant bookable	0.091	0.090	0.190	0.035	0.051	0.039	−0.054	0.042	−0.265	1.000
VIF	1.88	1.08	1.12	1.78	1.20	1.01	1.01	1.01	1.12	1.12

, where the correlation values between the variables are below 0.8 [98], indicating that the results of our analysis are not affected by variable covariance. The price (Ln) correlates highest with accommodates, which is 0.653. This result is reasonable since the price is higher when the room can accommodate more people. Finally, a multicollinearity analysis was performed, and the results are shown in Table 4, where the variance inflation factor (VIF) of all variables is less than 3 [1], indicating that multicollinearity does not affect the results of the analysis. Finally, based on previous studies [1], price, rating score, and facial attractiveness were log-transformed to reduce bias.

5.2. Main Results

Drawing on previous studies [31], we used Stata 15 software to empirically validate the study model using negative binomial regression with a hierarchical regression approach [99]. Negative binomial regression is used to predict data for count variables [100]. We use the negative binomial regression method because the dependent variable contains only non-negative integer values and the variance is greater than the mean [101]. The dependent variable is significantly positively skewed and kurtotic (skewness = 2.127, kurtosis = 7.415).

The regression results are shown in

Table 5. Results of the negative binomial regression model.

	DV: Booking Behavior	Model 1	Model 2	Model 3
Control variables	Price (Ln)	−0.077 ** (0.031)	−0.081 *** (0.031)	−0.119 *** (0.028)
	Rating score (Ln)	1.328 *** (0.042)	1.324 *** (0.042)	1.32 *** (0.027)
	Superhost	0.820 *** (0.031)	0.823 *** (0.03)	0.753 *** (0.03)
	Accommodates	−0.059 *** (0.008)	−0.058 *** (0.008)	−0.055 *** (0.007)
	Room type
	Private room
	Shared room	−0.171 (0.160)	−0.221 (0.146)	−0.165 (0.115)
	Entire home	0.417 *** (0.036)	0.417 *** (0.036)	0.411 *** (0.034)
Independent variables	Facial attractiveness (Ln)		0.461 *** (0.089)	0.431 *** (0.084)
	Subject diversity		0.197 *** (0.042)	0.214 *** (0.038)
	Perspective taking		0.248***(0.048)	0.198 *** (0.047)
	Response time			−0.216 *** (0.021)
	Instant bookable			0.166 *** (0.031)
	Constant	−2.453 *** (0.253)	−4.56 *** (0.452)	−4.014 *** (0.400)
	Pseudo R2	0.070	0.071	0.073
	Chi-square	1925.276	2041.894	5412.847
	Observations	8752	8752	8752

Note: Standard coefficients are shown in the table. Standard errors are shown in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1.

. Model 1 included only control variables. Model 2 analyzed the effect of social distance on booking behavior in combination with the control variables. Model 3 added the factor of temporal distance to Model 2 to study the effects of all variables on booking behavior.

The results of Model 1 show that all control variables, except the shared room dummy, have a significant effect on room bookings (p < 0.01). Of these, room price and accommodates had a significant negative effect on room bookings.

The results of Model 2 indicate that there is a significant positive relationship between host self-introduction subject diversity and room bookings (β = 0.197, p < 0.01). Controlling for other factors, a 1% increase in the value of host self-introduction subject diversity will increase room bookings by 0.197%. Thus, Hypothesis 1 is supported. This suggests that if hosts include relatively more topics in their self-introduction and listing descriptions, this indicates a willingness and attitude to share. This sharing by hosts will also bring the guest closer to the host and facilitates accessing more comprehensive information about the host, which also brings the host and the guest closer socially and increases the booking behavior of the guest. We confirm previous scholars’ findings by extracting the subject diversity variable [66].

The host self-introduction’s perspective taking significantly and positively affected room bookings (β = 0.248, p < 0.01), meaning that controlling for other variables, if the value of the host’s self-described perspective taking increased by 1%, room bookings would increase by 0.248%; thus, Hypothesis 2 was supported. The results suggest that when hosts think more from their guests’ perspective, they tend to be caring, empathetic, and helpful to others [71]. This perceived interpersonal intimacy can be better established between hosts and guests, shortening the psychological distance between hosts and guests, and promoting booking behavior.

The host’s facial attractiveness had a significant positive effect on room bookings (β = 0.461, p < 0.01), meaning that a 1% increase in the value of the host’s facial attractiveness score would increase room bookings by 0.461%, controlling for other variables. Thus, Hypothesis 3 was supported. The results suggest that attractive hosts are likely to stimulate the social interest of their guests, thereby reducing the perceived social distance. Furthermore, people with an attractive appearance are usually perceived as having good social skills and beneficial qualities [74], so guests are more likely to deal with hosts who are very attractive when booking a room, which increases guest booking behavior.

Model 3 adds a variable for the temporal distance dimension, and the results show that response time has a significant negative effect on room bookings (β = −0.216, p < 0.01), meaning that for every 1% increase in the value of response time, controlling for other factors, room bookings will decrease by 0.216%; thus, Hypothesis 4 is supported. Instant bookable significantly positively affected room bookings (β = 0.166, p < 0.01), and Hypothesis 5 was supported. The results show that the host’s response time and whether the instantly bookable feature is turned on affect the perceived temporal distance of the guest and that a faster response from the host to the guest’s request reduces the temporal distance and promotes the guest’s booking behavior through feelings of hospitality. Our results also support previous scholarly research on psychological distance in the context of online retailers [84].

5.3. Robustness Checks

To test the robustness of the regression analysis, we changed the regression method and used robust regression methods to test our results. Compared to ordinary least squares regression methods, robust regression is less likely to be affected by heteroskedasticity, outliers, and nonnormality [102]. The robustness test results are listed in Model 4 in

Table 6. Results of robustness check.

	DV: Booking Behavior	Model 4	Model 5
Control variables	Price(Ln)	−0.969 *** (0.315)	−0.032 ** (0.014)
	Rating score(Ln)	2.852 *** (0.122)	−0.003 (0.005)
	Superhost	11.318 *** (0.388)	0.351 *** (0.018)
	Accommodates	−0.374 *** (0.076)	−0.020 *** (0.003)
	Room type
	Shared room	−0.850 (1.376)	−0.088 * (0.052)
	Entire home	4.255 *** (0.399)	0.180 *** (0.018)
Independent variables	Facial attractiveness (Ln)	3.248 *** (1.040)	0.249 *** (0.046)
	Subject diversity	1.271 *** (0.487)	0.129 *** (0.023)
	Perspective taking	2.332 *** (0.577)	0.099 *** (0.026)
	Response time	−0.948 *** (0.254)	−0.037 *** (0.010)
	Instant bookable	1.177 *** (0.389)	0.075 *** (0.017)
	Constant	−10.545 ** (4.702)	−0.385 * (0.209)
	R2	0.206	0.082
	Observations	8752	8601

Note: Standard coefficients are shown in the table. Standard errors are shown in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1.

and are consistent with our regression results, indicating the reasonableness and robustness of our findings.

In addition, using the number of reviews to quantify customers’ booking decisions may lead to biased results. Therefore, we used the average number of reviews received by hosts per month (i.e., the ratio between the total number of reviews received by hosts and the number of months they are displayed on Airbnb) as a proxy variable for the dependent variable and used OLS regression to test the robustness of the model. The results are shown in Model 5 in Table 6, and the significance of most of the variables from the robustness test is consistent with the regression results, indicating the reasonableness and robustness of our research model.

6. Discussion

6.1. Key Findings

Our research suggests that a guest’s perceived social distance from the host affects the guest’s booking of the property when browsing online accommodation-sharing platforms and that increasing perceived social distance reduces the guest’s booking behavior. The results confirm the important role of social distance in influencing guest bookings and host income [4,103]. Providing guests the opportunity to socialize with their destination hosts is a key reason why online accommodation-sharing services are so popular with consumers. In addition, the results also show that temporal distance impacts the booking behavior of guests. Specifically, the greater the temporal distance a guest perceives, the less likely they are to book a room. More specifically, a guest prefers a host who can positively respond to him in their accommodation, which validates previous research findings [16].

Moreover, by crawling and mining the images and text information of a host’s information homepage in the online accommodation-sharing platform, we consider the facial attractiveness of the host’s photo, the subject diversity of the host’s self-introduction, and perspective taking as important representations of social distance between guests and hosts. In addition, this paper considers the host’s response time and whether the instantly bookable function is enabled as important representations of consumers’ perceived temporal distance. The findings suggest that the information displayed by hosts on online accommodation-sharing platforms affects consumers’ perceived psychological distance, which in turn influences their booking behavior. It is worth mentioning that the facial attractiveness of the host’s photo, the subject diversity of the self-introduction text, and perspective taking positively impact consumer booking behavior. However, the host’s response time negatively affects the guest’s booking behavior, while instant bookable has a positive impact.

6.2. Theoretical Implications

This research has several theoretical contributions. First, it contributes to the shared accommodation literature by filling the gaps in knowledge about psychological distance. A review of the literature on psychological distance and consumer behavior reveals that most existing studies have focused on online consumption environments such as the e-commerce context [58,81] and the online retail context [16], ignoring the mechanism by which psychological distance functions in the context of shared accommodation. Unlike previous literature, this paper explores how psychological distance affects the booking behavior of guests in online accommodation-sharing platforms and, specifically, explores the effects of both temporal distance and social distance on consumers’ booking behavior. Overall, our study shows that different dimensions of psychological distance jointly influence the booking behavior of guests. Our findings extend the theory of psychological distance and construal level to the field of shared accommodation and even the sharing economy. The study also provides empirical support for the theoretical link between psychological distance and the occurrence of booking behavior in the context of shared accommodation. It offers a new perspective for studying consumer behavior in sharing accommodation.

Second, previous literature on the study of psychological distance used a single variable, such as the age gap or spatial distance difference, to measure social and spatial distance in consumers’ perceived psychological distance [4]. In contrast, our research contributes to the psychological distance literature by mining and extracting information about online accommodation-sharing platforms that may influence the perceived psychological distance of guests and by enriching the representation of psychological distance in online display information. In general, the study shows that the information displayed by the host of the online accommodation-sharing platform affects the psychological distance perceived by guests. Specifically, the increase in the perceived temporal distance of the guest reduces booking behavior, but reducing social distance has a positive effect on booking behavior. The study’s results also prove the impact of temporal distance in e-commerce [58] and the findings on social distance in shared accommodation [4].

Finally, the extant literature only used questionnaires and experimental methods to explore the impact of psychological distance on individual behavioral intentions or consumer decisions [104]. Our research uses accurate consumer data crawled from online accommodation-sharing platforms to evaluate the influence of psychological distance on guest booking behavior. By collecting accurate consumption data directly from online accommodation-sharing platforms, the research methodology provides a new perspective in the literature on psychological distance and consumer behavior in the sharing economy. In addition, our study explores the mechanism of the influence of both text and picture information on the booking behavior of guests by employing actual consumption data collected by online accommodation-sharing platforms using machine learning methods. This complements previous literature that only investigates the impact of text or picture information on consumer booking behavior [4,26].

6.3. Practical Implications

The results of this study have implications for service providers and platform managers in the sharing economy, which are described in the following subsections.

For practitioners in the sharing economy, especially shared accommodation hosts, this study has several implications for presenting information in a better way to attract guests to their properties. First, we suggest that the host should pay attention to the customer’s message in time and give a timely response as much as possible to reduce waiting time for interactive communication. At the same time, the host can open multiple channels to receive customer information and multiple channels dedicated to responding to the basic information needs of customers at the same time. This is because it reduces the negative impact of temporal distance on consumers’ booking behavior and increases their willingness to book. More guest bookings can also increase a host’s income [6].

Second, the findings of this research also indicate that reducing social distance promotes the booking behavior of guests; specifically, the social distance perceived by the guest from the content of the host’s self-introduction homepage significantly affects the guest’s booking behavior. In addition, a host’s carefully edited self-introduction and listing description can reduce negative guest comments [6]. Therefore, we recommend that hosts pay special attention to editing text and displaying images on their personal homepages. When introducing themselves to guests, hosts should not only introduce themselves in as many ways as possible, but should also explain what help and services guests can expect from them. Hosts can imagine themselves as guests and what concerns they would have when they move into an unfamiliar room, so they can understand their guests’ needs differently. In addition, we recommend that hosts put some thought into their photos. Although everyone’s looks are genetically determined, hosts can still give their guests a neat and clean image by getting a neat haircut, putting on makeup, or dressing up before taking photos.

For sharing economy platforms, ensuring service providers stay is key to competing with other platforms to increase consumer engagement. Because service providers are co-producers of sharing economy platforms [105], it is also beneficial for sharing economy managers to ensure the well-being and benefits of service providers. Therefore, we recommend that administrators optimize the platform to achieve these goals. For the platform, the display function of the online platform should be optimized or designed to help hosts better display information, create a good communication and interaction environment, and enhance the attractiveness of the platform to customers and tenants. For example, it can add the function of reading and displaying messages, prompting letters from guests, and increasing the interaction channels. Additionally, the platform should have a minimum number of words required for the host’s self-introduction to avoid a bad impression caused by too little self-introduction. The interface design of the platform should also show more information about the host and the host-guest interaction, which is also a valuable inspiration for the interface designers. Finally, the platform can help hosts—and, more generally, service providers—by providing them with online tutorials on taking better photos and videos to achieve better results in showcasing their services or goods.

6.4. Limitations and Future Study

There are some limitations of this research that can be explored in future work. First, due to the unavailability of accurate booking data, we used the number of reviews received by hosts divided by the review rate to represent the number of rooms booked. This is also treated in the existing studies. Still, nevertheless, there may be differences between this approach and the actual booking data, and future studies can revalidate the model with actual booking data as much as possible. Second, this paper does not explore the impact of spatial distance on consumers due to the unavailability of spatial information. Future research could continue to explore and probe other factors that influence different dimensions of psychological distance in online accommodation-sharing platforms. Third, the study data are based on an accommodation-sharing platform in China, and data from countries with different background cultures and accommodation-sharing platforms could be obtained for future research. Fourth, future research could be devoted to topics such as the operation of shared accommodation during the COVID-19 epidemic. Finally, future research can be further refined based on visitor seasonality or other visitor attributes: gender, age, education level, etc.

7. Conclusions

Based on the data-driven behavior decision-making approach, this study uses large-scale but fine-grained secondary real-world data from the shared accommodation platform Airbnb to empirically examine temporal and social distance effects on guests’ booking behavior using textual and image analysis methods. This paper is one of the first empirical studies to explore how the information displayed by hosts in accommodation-sharing platform affects the booking behavior of guests through psychological distance (derived from social psychology). Our findings demonstrate that social distance has a negative effect on guests’ booking behavior, and that temporal distance also hurts guests’ booking behavior. In detail, subject diversity, perspective taking, and facial attractiveness in the dimension of social distance positively influence guests’ booking behavior; instant bookable in the temporal distance dimension positively affect booking behavior, while response time has a negative effect. The findings of this paper extend the impact of psychological distance in the emerging literature on shared accommodation and demonstrate the unique impact of psychological distance in shared accommodation and, by extension, the sharing economy. This makes a complementary contribution to the literature on the sharing economy and has practical implications for the betterment of the shared accommodation industry.