Continuance in Portuguese Peer-To-Peer Accommodation Services Through the Lens of Process Virtualization Theory

A Substantial changes have taken place in the way digital technology influences hosting practices. In a quantitative study, the factors driving the digitalization of hosting practices are analyzed through the lens of process virtualization theory (PVT). The impact of the propositions set out in PVT on hosts perceptions of diagnosticity and the intention to continue using peer-to-peer (P2P) accommodation services are assessed. A survey is used to collect data from Portuguese hosts and partial least square path modelling (PLS-SEM) technique is used to test our model. Findings show support for the direct impact of process readiness (relationship and synchronism readiness) and technological capabilities (representation and reach) on hosts’ intentions to continue using P2P accommodation services. Moreover, results show support for the influence of identification and control, relationship readiness, monitoring, and representation capabilities on hosts’ perception of diagnosticity, which will influence hosts’ intention to continue using P2P accommodation services. However, sensory readiness has no significant effect on hosts’ perceptions of diagnosticity or their intention to continue using P2P accommodation services. This study extends PVT into a new research context, that of accommodation services. The Portuguese market also provides a new cultural setting for further exploration. The results offer a nuanced understanding of hosts’ intentions to continue using these services and show the importance of shaping hosts’ perceptions of diagnosticity.


Introduction
The greatest transformation of consumer society resulting from digitalization is the advent of the sharing economy (SE) (Ravenelle, 2020).In the hospitality industry, disruptive business models have engendered the development of new actors, such as P2P accommodation platforms (P2PAP) and new roles for consumers, such as individuals who exhibit both sharing and using behaviors, known as prosumers [see Hermes et al. (2020)].The processes involved in these models are mostly immaterial and lead to immediate consumption (Airoldi & Rokka, 2022).
A survey of short-term rentals conducted across the European Union (EU) of 27 and coordinated by the European Commission, has revealed a growing preference for booking accommodation services through a digital platform: 29% of EU citizens have booked a room in a private residence, an apartment or house using a platform, at least once; amongst these, 89% would recommend booking short-term rentals through a digital platform (European Commission, 2021).Results from this study also show that EU respondents find the migration of accommodation services to online environments attractive, as this type of accommodation is cheaper (63%) and often better located (43%) than traditional options.
P2PAP offer a sequence of tasks preceding the actual stay, allowing guests to find a suitable place to stay and hosts to find a suitable guest.All interaction evolves throughout the said sequence and occurs in a virtual setting mediated by a digital platform.The traditional process of finding a suitable guest involves physical interaction between hosts or their representatives and potential candidates and important persons of reference, such as neighbors or real estate agencies, and numerous visits to the property.P2PAP technology eliminates the physical interaction between hosts or their representatives and potential guests and the respective documentation.It is now a brief and dematerialized process (Airoldi & Rokka, 2022), where physical interaction happens between hosts and their computers or mobile phones to create their profiles, upload photos of their property and learn more about their potential guests (Ert & Fleischer, 2020).
The virtualization of short-term rentals using P2PAP, has automated some tasks such as matching the host with potential guests and verifying their identity.Some P2PAP even provide an automated option that skips hosts' approval, enabling instant booking.However, most hosts prefer to retain the responsibility for accepting or rejecting a booking request.Consequently, the decision-making process of hosts cannot be set apart from the ability of the short-term rental system to help them gather diagnostic information on potential guests and assess a booking request.Accordingly, an important research path is to explore the virtualization of shortterm rental services and understand how a diagnostic experience of using P2PAP can improve the virtualization of these services.
Existing Information System (IS) research has shown that diagnosticity of information is important to reduce buyers' uncertainty toward the characteristics of a product and its future performance (Dimoka et al., 2012) and how poor product diagnosticity negatively influences continued use of online recommendations systems and product purchase intention (Benlian et al., 2012).Moreover, regardless of the considerable potential of short-term rental services, through the use of P2PAP, the profitability and long-term success of these companies relies on the continuance intention of its users (Arteaga-Sánchez et al., 2020).In order to differentiate themselves from competition, P2PAP (e.g.Airbnb) are stimulating hosts' continuance intention so as to extend their engagement and loyalty (Wang et al., 2020).
Recent research has explored the factors contributing to the continuance intention of virtual sharing economy services, either relying on the expectation-confirmation theory-based model (ECT model) or trust.Studies relying on the first perspective extend ECT with variables from other models and theories, such as the SERVQUAL model, IS Success model or the technology acceptance model (TAM) (e.g.Cheng, 2021;Pang et al., 2020;Shao et al., 2020) and use them as antecedents of confirmation and satisfaction.Studies relying on trust emphasize factors related to the platform, the guest, the host or a combination of these elements (C.Y. Li & Tsai, 2022).Nevertheless, evidence provided is fragmented and fails to account for the process nature of shortterm rental services (Overby, 2008), connecting hosts, guests and a P2PAP in a sequence of tasks designed to enable digital transactions.Against this background, we aim to pinpoint the factors that shape hosts' intentions to continue using short-term rental services.For this purpose, we first build a theoretical framework grounded on PVT (Overby, 2008(Overby, , 2012)), integrating the concept of diagnosticity.PVT is an adequate theoretical lens to examine host-guest relationships mediated by P2PAP technologies because it targets communication deprived of physical interaction between people or between people and objects.Moreover, PVT has proven useful in explaining virtualizability (measured as adoption, use or continued use) in different contexts such as religious practice (Bathan & Ramos, 2022), online learning (Alarabiat et al., 2021) and digital financial services (Verbovetska, 2019).By incorporating diagnosticity into our model, we are introducing an extremely important cognitive reaction in the context of online decision support systems, as it is deemed an integral part of the online purchase process (Xu et al., 2014).
Our contribution is threefold.First, we combine research on short-term rental services with research on process virtualization, using PVT to analyze the virtualizability of short-term rental services.We believe this study is the first to provide data on the virtualizability of short-term rental services through the lens of PVT.Second, we advance the literature on diagnosticity providing new insights into the mechanisms that contribute to its formation with respect to accommodation services.We draw the reader's attention to the importance of shaping hosts' perceptions of diagnosticity and demonstrate the effect of this cognitive appraisal on hosts' continuance intention.Third, we extend research on P2PAP by analyzing hosts' intention to continue using short-term rental services to put forward suggestions to P2PAP operators for improving the performance of short-term rental services.
The study is structured as follows.Section 2 encompasses the theoretical background, in section 3 hypotheses are also developed.Research methods and results are then described.Finally, results are discussed, implications for academia and practical implications are outlined and limitations and areas for further exploration are identified.

Theoretical background
Process Virtualization Theory (PVT) Overby (2008) introduces a theory and a research model for process virtualization.PVT has been used to predict whether the migration of a physical process into a virtual environment will fail or succeed (e.g.Czarnecki et al., 2010;Hanel & Felden, 2015), but also to understand the factors underpinning success or failure of existing virtual processes (e.g.Balci & Rosenkranz, 2014).Additionally, PVT has been used to explain resistance toward virtualized processes (e.g.Barth & Veit, 2011b).Process virtualizability expresses the feasibility of a process being performed virtually, without any physical interaction connecting people or people and objects.Adoption or use of the virtual process are both forms of measuring process virtualizability (Overby, 2012).Reflecting the principle that some processes are more liable to being virtualized than others, this IS theory emphasizes the central role of IT as an enabler of virtualization.Virtualizability results from the technological capacity of the virtualization mechanism (e.g. a P2PAP) to address the characteristics of the process (sensory requirements, relationship requirements, synchronism requirements, and identification and control requirements).PVT posits that process virtualizability decreases as each of these requirements becomes relevant in the completion of the process (Overby, 2008).When process requirements are satisfied, Bose and Luo (2011) speak of process readiness.Sensory, relationship, synchronism and identification and control readiness are defined as the extent to which the respective requirement is satisfied in a virtualized process and all four are expected to have a positive influence on virtualizability (Bose & Luo, 2011;Thomas et al., 2016).The technological capacity of the virtualization mechanism is represented by three IT capabilities, which are representation, reach, and monitoring capability.Representation and reach are deemed to affect process virtualizability in a positive manner.The effect of monitoring varies in line with the empirical research context (Overby, 2012).PVT also posits that the relationship between the characteristics and virtualizability of the process is moderated by the technological capacity of the virtualization mechanism (Overby, 2012).
The complementarity between PVT and other IS theories has been established by Overby (2008Overby ( , 2012)).As the author points out, Davis' technology acceptance model (TAM) or even Venkatesh's unified theory of acceptance and use of technology (UTAUT) explain the adoption of a technological product or system based on individuals' perceptions focusing on the outcomes (e.g.ease of use or performance expectancy) while using the system in question, without explaining why.PVT can explain why system use might be perceived as complex or without advantages by users suggesting, for example, that sensory or relationship requirements of the process might compromise ease of use or limit performance expectancy.Moreover, TAM differs from PVT.While the former predicts the adoption of a system already implemented, PVT application is more comprehensive, encompassing different stages of a system's lifecycle (Barth & Veit, 2011b).It can be used to forecast whether a physical process can be successfully migrated to a virtual environment, to explain why the adoption of a virtual process was not successful, or how to mitigate individuals' resistance toward the use of a virtual process.
The role of virtualization requirements proposed by PVT has been explored by conceptual, qualitative, and quantitative studies.Conceptual applications of PVT integrating either process characteristics or the characteristics of the virtualization mechanism with constructs from other theories include cross-channel instant messaging services (M.Li et al., 2009), green IT initiatives (Bose & Luo, 2011), military training processes (Marsilio & Morris, 2015) and mobile health services (Phagdol et al., 2022).Qualitative studies have also explored extensions of PVT.For instance, in a study analyzing citizens' preference for traditional service delivery over virtual service delivery (e.g.passport application, vehicle registration, civil marriage), Barth and Veit (2011a) combine process characteristics from PVT with constructs from TAM, innovation diffusion theory and media richness theory.In a digital work environment, Hofeditz et al. (2020) combine PVT and collaboration virtualization theory to analyze the dynamics of virtual collaboration.
Studies using a quantitative approach have also produced evidence of the validity of the theory in several contexts.Exploring online banking processes, Balci et al. (2013) analyze the impact that user resistance to conduct a process virtually and perceived virtualizability both have on virtual process use.Evidence is provided for the positive effect of all process characteristics on user resistance leading to low levels of virtual process use; except for relationship requirements, all process characteristics have a negative impact on perceived process virtualizability, leading to high levels of virtual process use.Graupner and Maedche (2015) show that sensory, control and relationship requirements are the main barriers for potential use of online banking services.Furthermore, synchronism and identification requirements, which allow faster processing of transactions and authentication procedures, have a positive impact on intended use of online banking services.Finally, in an online learning environment, Alarabiat et al. (2021) examine the role of virtualization requirements on student satisfaction with the use of online learning platforms (OLP) and their intention to continue to do so.Students who perceive the learning process as highly dependent on sensory requirements seem to be less satisfied and show lower intention to continue using OLP.Similarly, students who value social interaction during the learning process will probably be less satisfied with OLP and as such, are less likely to continue using OLP.Synchronism and identification and control requirements are not relevant to student satisfaction or their intention to continue using OLP.The positive and direct impact of reach and representation on student intention to continue using OLP demonstrates that further adoption of a process is reached when IT capabilities suitably address process requirements.
PVT applications go beyond the individual level.From an organizational perspective, Thomas et al. (2016) show that process characteristics do not significantly explain why organizations undertake Green IT initiatives and that the technological context enabling virtualization plays a lesser role than the environment surrounding organizations.In a study analyzing firms' intention to adopt software as a service (SaaS), Tomás et al. (2018) demonstrate the positive influence of reach and monitoring on intention to adopt SaaS.
In summary, prior research has provided evidence for the influence of both the characteristics of the process and characteristics of the virtualization mechanism on process virtualization, thus establishing the importance of the propositions of PVT in different contexts.Following Overby's suggestion, the models usually include variables specific to the context to improve the explanatory power of PVT.
However, results are mixed and fragmented, seem to vary with the context under examination, and are also subject to the beliefs and characteristics of the participants involved (Balci & Rosenkranz, 2014).As noted by Graupner and Maedche (2015), empirical evidence is still needed for the novel theory that has not as yet been sufficiently tested.

Virtualization of the short-term rental process
Initially developed to explain the migration of processes from a physical to a virtual environment and user resistance toward virtualized processes, PVT has proven to be adequate for studying the digitalization of financial services (Verbovetska, 2019), online learning (Alarabiat et al., 2021) and religious practice (Bathan & Ramos, 2022) via technological means.In a similar manner, short-term rental services through P2PAP are a new process virtualization topic which can also be adequately explained by PVT.Undeniably, sensory, synchronism, relationship and identification and control requirements continue to be important for hosts and guests using P2PAP.For instance, with regard to main sensory perceptions, guests still value aspects of the property such as lighting, smell of the room, decorative items and room aesthetics (Lyu et al., 2019), as these contribute to a homelike feeling.Similarly, in relation to synchronism, guests expect the host to reply to their queries and confirm their reservation with minimum delay, and use hosts rate response as a measure of effectiveness (Wu, Ma, et al., 2017).Guests continue to seek direct interaction with the host, the local people, and other guests (Tussyadiah & Pesonen, 2018), confirming the importance of social interaction as a requirement for the virtual process.Safety and privacy concerns were reported to slow down the increase in the number of hosts registered in Airbnb (Chen et al., 2020), a reflection of the relevance of identification and control requirements.
Whereas PVT postulates a negative impact of each of the four process requirements on process virtualizability, we anticipate that process readiness, i.e. the capacity of the virtualized process to satisfy each one of the four process requirements, will have a positive impact on short-term rental via P2PAP.
We also anticipate that the technological capabilities of the platform will always have a positive influence on the virtualization of short-term rental services, and therefore do not explore their moderating effects.Existing research has suggested that listing popularity and ultimately, guests' booking behavior on P2PAP, can be enhanced by representative property attributes, such as the visual appeal of the property, and the respective photo verification and visual description, along with host self-description (Yan et al., 2021).Moreover, by matching consumers and services providers from all over the word, and especially of isolated or disadvantaged communities (Dillahunt et al., 2015), and underutilized or idle resources, otherwise not available (Radka & Margolis, 2011) digital platforms benefit a large network of active participants with extended reach (Sutherland & Jarrahi, 2018).Furthermore, existing research has shown that booking intentions are affected by digital features, such as star ratings and text reviews (Dann et al., 2020).These evaluative features allow hosts to choose their future guests and guests to choose a place to stay and can also provide historical data on guest and host performance, offering additional assurance to track host and guest participation.
Finally, we argue that P2PAP capacity to assist its users to accomplish their goals can lead to enhanced virtualizability.Therefore, we have integrated the concept of diagnosticity with PVT to design a model that explains the virtualization of short-term rental services.

Diagnosticity of the short-term rental system
While trying to improve the understanding of marketers and researchers on trial experiences during the purchase process, the Kempf and Smith (1998) study describes diagnosticity as consumer beliefs in the effectiveness of a trial experience to evaluate the attributes of a specific brand.Diagnosticity is a cognitive reaction relating to users' mental processes when interacting with product information presented on websites (Parboteeah et al., 2009).It has been explored in IS studies, deemed fundamental in the design of online shopping experiences (Z.Jiang & Benbasat, 2007), with impacts on users' intention to revisit a website (Xu et al., 2014) and on users' satisfaction with their decision outcomes while performing a search (Yi et al., 2017).In health information technology literature, diagnosticity of e-consultations (the system integrating patients, clinicians and technology) refers to the perceived ability of clinicians with respect to a telemedicine system enabling remote clinical assessment of patient problems (Serrano & Karahanna, 2016).Diagnosticity can also be defined as the degree of helpfulness of information.For instance, Filieri (2015) adopts a diagnosticity framework to understand what makes online reviews useful.Findings from his study suggest that when the content of online reviews is perceived to be useful to determine the performance and quality of a product, users will most likely use it in their decision-making process.Recent research on the sharing economy analyzed diagnosticity of information provided to users of sharing economy platforms.Results show that users can identify elements with greatest diagnosticity to aid their decision-making (Zloteanu et al., 2021).Results on travelers' demand for P2P ridesharing services further show that assessment relies on multiple types of information cues (intrinsic and extrinsic) and that the usefulness of some cues (i.e.reputation of the seller) is related to other types of cues (i.e.price) (Jang et al., 2021).
In a similar manner to Serrano and Karahanna (2016) and in line with our previous work (Barbeitos & Oliveira, 2022), we argue that diagnosticity is an attribute of the short-term rental system (hereafter, e-rentals), which integrates guests, hosts and the P2PAP.Since hosts are the target group of this study, the concept is defined as hosts' perceived capacity of P2PAP use to convey information about guests, helping them to assess a booking enquiry.

Research model and hypothesis
As per Overby (2008Overby ( , 2012) ) virtualizability of a process can be measured either as adoption, use or intention to use.In this study, intention to continue using P2PAP reflects the virtualizability of the shortterm rental process.Thus, we argue that the four process characteristics suggested by PVT are four precursor requirements of short-term rentals and once fulfilled, will positively impact hosts' perceived e-rental diagnosticity, as well as their intention to continue using P2PAP.In addition, we argue that the three characteristics of the virtualization mechanism, which coincide with the technological characteristics of P2PAP, will positively impact hosts' perceived e-rental diagnosticity as well as their intention to continue using P2PAP.Furthermore, hosts' perceived e-rental diagnosticity will have a positive effect on their intention to continue using P2PAP.The research model is illustrated by Figure 1.

Impact of process readiness on hosts' perceptions of diagnosticity and their continuance intention
Sensory readiness is defined as hosts' perceptions of P2PAP use to fulfil the need to allow guests to see, touch, and personally experience the characteristics of the properties offered in the market [adapted to the context of our study from Overby (2008)].High sensory requirements have been proven to be the main barriers to conducting online public services, such as civil marriage, vehicle registration (Barth & Veit, 2011a) and banking services (Graupner & Maedche, 2015).Evidence from research exploring the use of technology that stimulates sight and possibly the other senses of users during the pre-purchase stage demonstrates that this period becomes more than merely a time to collect information: The user may explore and configure their vacation (Simoni et al., 2021).These findings suggest that if hosts perceive that P2PAP have the capacity to fulfil guests' needs to see, touch, and personally experience the characteristics of their properties, the e-rental process may become more amenable for virtualization.E-rental technology enables guests to choose a property based on photos and videos (Ellison et al., 2011;Ert & Fleischer, 2020) and facilitates communication with hosts through messaging services.Therefore, the degree to which P2PAP enhance hosts' perceptions of sensory readiness will have a positive effect on their perceptions of e-rental diagnosticity, as well as their intention to continue using P2PAP.Thus, we hypothesize: H1: Sensory readiness has a positive effect on e-rental diagnosticity.
H1a: Sensory readiness has a positive effect on hosts' intention to continue using P2PAP.
Synchronism readiness is defined as hosts' perceptions of P2PAP capacity to fulfil the need to quickly complete the formalities for renting activities [adapted to the context of our study from Overby (2008)].Synchronism requirements were deemed to be the main barriers to the use of online banking services (Balci et al., 2013) and fundamental for virtual collaboration in digital work (Hofeditz et al., 2020).Synchronism requirements are better addressed by the online channel, allowing faster processing of banking transactions (Graupner & Maedche, 2015), suggesting that if hosts perceive that P2PAP have the capacity to execute renting activities quickly and with minimum delay, the e-rental process may become more amenable for virtualization.E-rental technology, such as instant messaging, enables real-time interaction between the host and multiple guests, regardless of their location.Moreover, participants avoid the "logistical hassle" and "social awkwardness" of dealing with money because transactions happen automatically and covertly (Sutherland & Jarrahi, 2018).Therefore, the degree to which P2PAP enhance hosts' perceptions of synchronism readiness will have a positive effect on their perceptions of e-rental diagnosticity, as well as their intention to continue using P2PAP.Thus, we hypothesize: H2: Synchronism readiness has a positive effect on e-rental diagnosticity.
H2a: Synchronism readiness has a positive effect on hosts' intention to continue using P2PAP.
Relationship readiness is defined as hosts' perceptions of P2PAP capacity to fulfil the need to interact with other hosts and guests in the market [adapted to the context of our study from Overby (2008)].Relationship requirements impact negatively on the use of banking services (Graupner & Maedche, 2015) and online learning services (Alarabiat et al., 2021).If hosts perceive that P2PAP have the capacity to fulfil the need to interact with other hosts and guests, the e-rental process may become more amenable for virtualization.Previous research has demonstrated that users engage in P2PAP use because they enjoy interacting with others (Bellotti et al., 2015).Users of P2PAP are driven by the expectation of social benefits, such as the chance to interact with local people and establish meaningful interaction (Dann et al., 2020;Hamari et al., 2015; Z. W. Y. Lee et al., 2018).For instance, Airbnb encourages hosts to get to know each other and participate in discussion rooms on hosting to share their experiences and ideas.Moreover, hosts are first guided to experienced host forums for help whenever facing a problem.Therefore, the degree to which P2PAP enhance hosts' perceptions of relationship readiness will have a positive effect on their perceptions of e-rental diagnosticity, as well as their intention to continue using P2PAP.Thus, we hypothesize: H3: Relationship readiness has a positive effect on e-rental diagnosticity.
H3a: Relationship readiness has a positive effect on hosts' intention to continue using P2PAP.
Identification and control readiness is defined as hosts' perceptions of the capacity of P2PAP to fulfil the need for transactions which uniquely identify guests and for features which influence their behavior [adapted to the context of our study from Overby (2008)].Identification and control requirements are better addressed by the online channel, which allows faster and safer authentication procedures during the use of online banking services (Graupner & Maedche, 2015).If hosts perceive that P2PAP have the capacity to fulfil the need for transactions which uniquely identify guests and that features exist to influence their behavior, the e-rental process may become more amenable for virtualization.As per Sutherland and Jarrahi (2018), identification and control requirements matchmake hosts and guests based on a set of attributes such as location or personal traits, offering the possibility to verify and confirm all users' IDs.Teubner and Flath (2019) conclude that P2PAP provide a technological infrastructure for keeping record of all the transactions and ensuring their security.Additionally, hosts may ask a guest to submit a photo of his/her government ID which can be compared with their profile photo.Review and rating systems allow users to carry out their screening tasks.Therefore, the degree to which P2PAP enhance hosts' perceptions of identification and control readiness will have a positive effect on their perceptions of e-rental diagnosticity, as well as their intention to continue using P2PAP.Thus, we hypothesize: H4: Identification and control readiness have a positive effect on e-rental diagnosticity.
H4a: Identification and control readiness have a positive effect on hosts' intention to continue using P2PAP.

Impact of the technological characteristics on hosts' perceptions of diagnosticity and their continuance intention
Monitoring is defined as hosts' perceptions of P2PAP capacity to authenticate participants and track their activity [adapted to the context of our study from Overby (2008)].While listing their properties, hosts disclose upfront information about themselves and their properties (Teubner & Flath, 2019) and they may develop a perception of risk with regard to informational privacy, as they cannot control their personal information.Their intention to continue using P2PAP is negatively influenced by this perception (Chen et al., 2020).If hosts perceive that P2PAP have the capacity to authenticate participants and track their activity, the e-rental process may become more amenable for virtualization.Previous IS research has established the relevance of securing and tracking transactions in the sharing economy (i.e.Sutherland & Jarrahi, 2018), as these mechanisms build up trust (Ert et al., 2016) and enforce the validity of transactions (Weber, 2014).Therefore, in e-rentals, the degree to which P2PAP enhance the ability to identify and track users' participation will have a positive effect on hosts' perceptions of e-rental diagnosticity and their intention to continue using P2PAP.Thus, we hypothesize: H5: Monitoring has a positive effect on e-rental diagnosticity.
H5a: Monitoring has a positive effect on hosts' intention to continue using P2PAP.
Reach is defined as hosts' perceptions of P2PAP capacity to allow them to rent their properties to people who they would not otherwise have access to [adapted to the context of our study from Overby (2008)].
Reach has proven to have a positive impact on process virtualizability in other contexts, such as airport check-in (Balci & Rosenkranz, 2014) and online distance learning (Alarabiat et al., 2021) because it promotes autonomy and flexibility of use, and allows the simultaneous remote participation of individuals.If hosts perceive that P2PAP have the capacity to allow them to rent their properties to people who they would not otherwise have access to, the e-rental process may become more amenable for virtualization.This technological capability gives visibility to communities which are somehow isolated (Dillahunt et al., 2015) connecting them with producers and consumers internationally and putting individuals in contact with new resources.Therefore, in e-rentals, the degree to which P2PAP broaden participation and yield new business opportunities will have a positive effect on hosts' intention to continue using P2PAP and their perceptions of e-rental diagnosticity.We hypothesize: H6: Reach has a positive effect on e-rental diagnosticity.
H6a: Reach has a positive effect on hosts' intention to continue using P2PAP.
Representation is defined as hosts' perceptions of P2PAP capacity to describe their properties in a manner relevant to e-rentals [adapted to the context of our study from Overby (2008)].In e-commerce research, this technological capability referred to as vividness or representational richness, conveys useful information on products being sold online, enabling consumers to form a clear idea of the product and its features (Z.Jiang & Benbasat, 2007).Increased perception of website diagnosticity and active learning of product knowledge have been documented as positive benefits of rich representational cues on consumer's understanding and behavior (H.Zhang et al., 2018).If hosts perceive that P2PAP have the capacity to describe their properties in a manner relevant to e-rentals, the e-rental process may become more amenable for virtualization.In P2PAP, from a host's perspective, the richness of information relates to the depth of the hosts' personal information and the information on the property itself.Therefore, in e-rentals, the degree to which P2PAP can replicate relevant on-site evaluation experiences and convey useful information on the property will have a positive effect on hosts' intention to continue using P2PAP and their perceptions of e-rental diagnosticity.We hypothesize: H7: Representation has a positive effect on e-rental diagnosticity.
H7a: Representation has a positive effect on hosts' intention to continue using P2PAP.
Previous research has shown that the assessment of the pros and cons of the use experience, will lead to increased probability of continued service use (e.g.Bhattacherjee, 2001).Following initial use, hosts form expectations regarding the performance of e-rentals on each transaction.If hosts perceive that P2PAP have the capacity to convey information on guests allowing them to evaluate a booking enquiry, the e-rental process may become more amenable for virtualization.Existing research has proven that hosts form first impressions about their guests mainly based on guests' profile photos and the information provided in P2PAP (Teubner et al., 2022), using guests' self-description, star-ratings and text-reviews.
While personal information helps create expectations of enjoyable social interaction, star-ratings help form expectations of economic value (i.e.extra income or lower ownership costs); text reviews allow expectations to be formed relating to social and economic value (Dann et al., 2020).Therefore, the degree to which P2PAP use enhances hosts' perceptions of e-rental diagnosticity will have a positive effect on their intentions to continue using P2PAP.Thus, we hypothesize: H8: E-rental diagnosticity has a positive effect on hosts' intention to continue using P2PAP.

Survey and development of measures
A survey targeting hosts registered on the Portuguese Tourism Register website (RNAL) and qualified as short-term rental service providers, was distributed through the online platform SurveyMonkey.com.The survey questions were preceded by a brief explanation of the SE to certify that the experience of respondents was valid to our research goals.With the aim of anchoring respondents to a specific platform, we asked hosts to indicate the P2PAP they most frequently use to list their properties.
Measures were obtained from previous research and the wording was modified to fit the research context to reinforce construct validity.Research constructs are operationalized in Appendix D. All items were measured using a seven-point scale ranging from "strongly disagree" to "strongly agree."Following previous research on SE and diagnosticity (J.Jiang et al., 2021;C. Li & Chau, 2018;Shao et al., 2020;Teubner et al., 2022;Wu, Ma, et al., 2017;Yan et al., 2021;H. Zhang et al., 2018), six control variables were introduced to control for the impact of host and property attributes on continuance intention and e-rental diagnosticity.Namely age, gender, education, type of host, listing experience with P2PAP and property type.The main study was carried out after a pilot study involving 30 hosts.Results from this pilot study allowed the readability of the survey to be improved and the scales to be validated, resulting in amendments to the wording of some items.

Sample and data collection
Data was collected between March 2019 and January 2020.Initially, most respondents only replied to the first question included in the e-mail without completing the rest of the survey.So, in order to increase our response rate, we decided to send additional e-mails to all these participants, explaining that the survey had more questions which required an answer.After ten days, we sent the first reminder to other hosts who had not returned a completed survey.In most cases, we had to send more than one reminder until we reached our goal.As we know that the response rate is around 30%, based on studies with similar scale (Ashrafi et al., 2022), our goal was to achieve a sample with more than 200 participants.We sent 741 e-mails to hosts randomly selected from RNAL, who provide short-term rental services.The total number of hosts registered on RNAL at the time of the survey was approximately 82 000.After several months and the screening phase, we obtained 297 valid surveys, an overall response rate of 40%, surpassing our initial goal.
Common method bias (CMB) was examined, first via Harman's one-factor to ensure the majority of the variance is not explained by a single factor (Podsakoff et al., 2003).The first factor only explains 22.99% of the variance.A theoretically unrelated marker variable was then introduced in the model (Lindell & Whitney, 2001).A maximum shared variance with other variables of 0.024 is obtained, a value considered low (Johnson et al., 2011).The method of full collinearity assessment proposed by (Kock, 2015;Kock & Lynn, 2012) was also used to detect common method bias.As per Kock (2015), variance inflation factors (VIF) scores greater

Results
The SEM variance-based technique of partial least square path modeling (PLS-SEM) was employed to assess and estimate the model.PLS is the most adequate method for this study, as our research is exploratory in nature and our complex model has never before been tested (Hair et al., 2019).Moreover the method is less strict with non-normal data (Hair et al., 2019) as is the case with our data (p < .01 based on Kolmogorov -Smirnov's test).We used SmartPLS software, version 4.0.9.3, to assist in the process (Ringle et al., 2022) and the Anderson and Gerbing (1988) two-step approach was also followed.The indicator weighting scheme was set by default as "Automatic."As our measurement model is reflexive, Mode A was used to calculate loading estimates.The algorithm runs after determining 3000 maximum iterations with a stop criterion of 10 −7 , employing the path weighting scheme.

Evaluation of the measurement model
Composite reliability (CR) values range from 0.779 to 0.964 thus surpassing the threshold of 0.7 and confirming the reliability of scales (Hair et al., 2012).Average extracted variance (AVE) values vary from 0.549 to 0.900, also surpassing the threshold of 0.5 and demonstrating the convergent validly of the model (Hair et al., 2011).Appendix B shows the abovementioned details.Except for Syn3, Ide2 and Int2 all item loadings exceed 0.70 (see Appendix A), also suggesting suitable convergent validity.Syn3, Ide2 and Int2 loadings are above the minimum (0.4) required by Hair, Hult, et al. (2014), and consequently were retained in the model.
Two criteria were used to assess discriminant validity.Appendix B shows that the square roots of AVEs (diagonal elements) are greater than the correlation between each pair of constructs (off-diagonal elements), demonstrating appropriate discriminant validity.Appendix A shows that the indicator loadings (in bold) are greater than the respective cross-loadings, thus providing additional support for discriminant validity.HTMT ratios all fall below the advocated threshold of 0.9 (see Appendix C), thus also supporting discriminant validity (Henseler et al., 2015).
Criteria relating to construct reliability, indicator reliability, convergent validity, and discriminant validity were satisfied, ensuring the adequacy of the measurement scales.

Assessment of the structural model
The initial assessment of collinearity among the exogenous constructs has shown VIF scores ranging from 1.078 to 1.670, all below the recommended threshold of 3.3 (G. Lee & Xia, 2010), demonstrating that collinearity is not a problem.After examining collinearity, Hair et al. (2019) suggest assessing the sign and statistical significance of the path coefficients in the structural model, insample predictive (R 2 ) measures, f 2 values, the Q 2 test and assess out-of-sample predictive validity using PLSpredict.The bootstrapping procedure was used to generate t-statistics and percentile confidence intervals to estimate the statistical significance of the path coefficients (Henseler et al., 2009), with 5000 bootstrap samples (Chin, 1998;Hair et al., 2014).Figure 2 presents the structural model (path coefficients, R 2 and Q 2 ) and Table 2 summarizes the results of the assessment.
Finally, the impact of e-rental diagnosticity on hosts' intention to continue using P2PAP is positive and statistically significant (β ̂ = 0.085; p = .083).Thus, H8 is supported, implying that when hosts perceive that P2PAP use conveys useful information about guests helping them to assess a booking enquiry, hosts have greater intention to continue using P2PAP.
Overall, the model explains 29.7% of the variation in e-rental diagnosticity and 38.0% of the variation in continuance intention.Comparing R 2 values with the thresholds for assessing explanatory power (0.19 = weak, 0.33 = moderate, 0.67 = substantial) proposed by Chin (1998), reveals a model with moderate explanatory  power, especially for continuance intention which has the largest coefficient of determination (R 2 ).Additionally, effect sizes (f 2 ) are small, ranging from 0.000 to 0.119, and Q 2 values are both positive, indicating medium predictive relevance for both e-rental diagnosticity and continuance intention (see Figure 2).Next, PLSpredict analysis was executed to assess the out-of-sample predictive power of the model.During the process, the guidelines proposed by Shmueli et al. (2019) were considered and the mean absolute error (MAE) values for the PLS-SEM model and the linear model (LM), as well as the Q2_predict values for the model's key endogenous constructs (continuance intention and e-rental diagnosticity) were analyzed.As can be seen in Table 3, Q2_predict values of the indicators of continuance intention and e-rental diagnosticity are positive.Results show that all the indicators of continuance intention and e-rental diagnosticity in the PLS-SEM analysis have lower MAE values compared to the LM benchmark, indicating high out-of-sample predictive power.

Discussion
We sought to examine the virtualizability of short-term rentals from the hosts' perspective.Virtualizability was measured as hosts' intention to continue using shortterm rental services through P2PAP use and our model was grounded on PVT.After testing, we found support for our model.We provide empirical evidence for the relevance of PVT and demonstrate that virtualizability results from the technological capacity of P2PAP to address the characteristics of the short-term rental process.We found that the virtualizability of short-term rentals can be maximized by addressing hosts' relationship and synchronism readiness, while also providing advanced technological capabilities, more specifically representation and reach.In other words, P2PAP which successfully address hosts' needs to interact with other hosts and guests and which offer rapid transaction support activities are likely to influence hosts' intention to continue using P2PAP services.Also, the ability of P2PAP to represent the characteristics of properties relevant to the transaction and the ability to connect hosts with other hosts and guests internationally, along with autonomy and flexibility of use are important technological capabilities likely to affect hosts' intention to continue using such services.
One possible explanation may be related to the fact that the technology incorporated in P2PAP may fulfil hosts' relationship and synchronism requirements more appropriately than traditional short-term rental processes.First, the extended reach offered by P2PAP leads to a greater network of users and resources, in turn leading to more interactions between hosts and guests at a global level, while also benefiting from preexisting community relationships on other social media platforms (Sutherland & Jarrahi, 2018).Second, and in a similar vein to that which takes place in financial services, P2PAP technology offers immediate transaction processing (Graupner & Maedche, 2015;Verbovetska, 2019).Concerning reach and representation capabilities, findings are aligned with PVT and adoption of other virtual processes (e.g.Balci & Rosenkranz, 2014;Ofoeda et al., 2018).In line with Alarabiat et al. (2021), which revealed the consistent effects of the above mentioned capabilities on students' intention to continue using online learning services, hosts are more likely to continue using P2PAP because they can show and describe the relevant characteristics of their properties to potential guests they would not otherwise have access to.Through the mass market provided by global reach and where distance ceases to be relevant, hosts can choose when and how to participate (Radka & Margolis, 2011).
Furthermore, the virtualizability of short-term rentals can be heightened by stimulating hosts' perceptions of diagnosticity while using P2PAP, by meeting their identification and control and relationship requirements and providing the advanced technological capabilities of monitoring and representation.In other words, P2PAP which successfully address hosts' needs for transactions which uniquely identify guests and features which influence their behavior, as well as the hosts' need to interact with other hosts and guests, are likely to influence their perceptions of diagnosticity during P2PAP use.Also, the ability of P2PAP to authenticate and track participants' activities and to represent the characteristics of properties relevant to the transaction are important technological capabilities likely to affect hosts' perceptions of diagnosticity during P2PAP use.
Once again, the technology incorporated in P2PAP may fulfil identification and control and relationship requirements more appropriately than traditional shortterm rental processes.On one hand, and in a similar manner to that which takes place in financial services [see Graupner and Maedche (2015)], P2PAP integrate extensive authentication features and mechanisms that are safer and more efficient when compared to traditional short-term rental alternatives (Teubner & Flath, 2019;Weber, 2014).On the other hand, while using P2PAP, hosts' become members of a large community increasing the possibility of establishing pleasant and gratifying interactions (Dann et al., 2020).When explanation of hosts' perceptions of diagnosticity is involved, the technological capability most valued by hosts is monitoring, followed by representation.These findings suggest that in order to increase hosts' perceptions of diagnosticity, P2PAP must prioritize technology that improves the ability to identify and track guests' participation and only after this has been achieved, can the replication of relevant on-site evaluation experiences be improved by conveying useful information on the property listed.Evidence on representation extends the findings of previous research (e.g.Serrano & Karahanna, 2016) to the context of P2P accommodation services.
We also demonstrate that virtualizability can be improved by maximizing hosts' perceptions of diagnosticity.As with the perceived usefulness of the ECT model, e-rental diagnosticity involves an assessment and comparison of task outcomes and task goals during P2PAP use.More specifically, an assessment of the degree to which the information conveyed by P2PAP helps hosts assess a booking enquiry.Therefore, the impact of e-rental diagnosticity on continuance is not surprising.This means greater perceptions of the capacity of P2PAP to convey diagnostic information about guests will most likely encourage hosts to continue using P2PAP.Therefore, designing for diagnosticity should be one of the imperatives for P2PAP operators (Yi et al., 2017), as we will recommend below.

Theoretical implications
Our study has significant implications for existing research.First, our study enriches PVT by showing that the interplay between the characteristics of the short-term rental process and the technological characteristics of P2PAP enhances the virtualization of shortterm rentals and the diagnosticity of P2PAP.On one hand, the study improves our understanding of how a process is successfully virtualized and identifies the influential factors and IT capabilities in the virtual context of short-term rentals.On the other hand, this study deepens our understanding of how a digital platform's diagnosticity is enhanced and identifies the IT capabilities in the virtual context of short-term rentals.We found that a combination of different aspects of process readiness and technological capabilities of P2PAP influences both hosts' perceptions of e-rental diagnosticity and their intention to continue using short-term rental services.Although previous research on accommodation sharing services has included technological enablers to explain users' continuance intention (Kong et al., 2020;Wang et al., 2020), the factors identified did not consider the specific technical characteristics of the systems nor the process nature of e-rental services.
Second, the findings from this study enrich the literature on diagnosticity, providing a new and nuanced understanding on how the factors identified can enhance diagnosticity.We found support for the role of monitoring capability as an antecedent of e-rental diagnosticity, with greater explanatory capacity than representation.This is a novel contribution meaning that diagnosticity can be propelled by a technological infrastructure that keeps records of all the transactions and allows fast and safe identification mechanisms, and to a lesser extent, by mechanisms that provide vivid representations of the properties listed and maximize the opportunities for hosts' self-disclosure.Lastly, diagnosticity is introduced as an antecedent of continuance.If hosts' intention to continue using P2PAP is heightened by platform diagnosticity, it means that it is likely that the virtualization of short-term rentals continues to be successful.

Practical implications
Additionally, practical implications should be noted.Our investigation suggests that the design of P2PAP functionalities should seek to improve all three technological capabilities identified: monitoring, representation and reach, and address identification, relationship, and synchronism readiness to increase hosts' perceptions of e-rental diagnosticity and the virtualizability of e-rental services.According to PVT, through the improvement of technological capabilities, P2PAP will also be improving process readiness as per the moderating effects suggested by this theory.Consequently, our recommendations only address technological capabilities.
Monitoring capabilities of P2PAP could be reinforced with the use of blockchain technology.Blockchain transactions allow participants to remain anonymous while third parties check their identity (Sun et al., 2016).Solutions based on this technology would avoid the disclosure of personal information before the rental contract is signed (S.Li et al., 2021).This technology can also validate user profiles and validate Internet of Things device profiles (Rahman et al., 2019).Blockchain-based short-term rental services allow the enforcement of contractual agreements without human interaction.Through the application of location information, radio frequency identification (RFID) or Near-field communication (NFC) to interact with devices and transmit guest and host data to blockchain ledgers, the obligations committed to by the parties in an agreement will be automatically executed (Rahman et al., 2019;Sun et al., 2016).P2PAP should make use of such smart contracts to facilitate check-in and check-out.By virtualizing and automating the administrative aspects of check-in and out, P2PAP would leave hosts with more time to focus exclusively on the hospitality aspects of receiving a guest.
Representation and vividness can be improved by incorporating virtual and augmented reality technologies in P2PAP.Virtual and augmented reality technologies are currently experiencing exponential growth, as a result of post-Covid 19 implications (Molina-Collado et al., 2022).Existing research shows that virtual reality (VR) is already being used by hotels (Lyu et al., 2021) and destinations (Tussyadiah et al., 2018) as a marketing tool to influence guest and tourist decision-making process.Guests may explore hotel facilities and the main attractions of a destination prior to their visit using smartphone-based virtual reality.This creates the desire to travel and make a direct visit (Subawa et al., 2021;X. Wu & Lai, 2022).P2PAP operators may wish to consider using VR advertising for the city where a property is located, complementing photos and videos of the listings with virtual tours led by a virtual host.Regardless of their significant potential use, the implementation of virtual and augmented reality technologies in hospitality involves a number of challenges, such as lack of interoperability and system feedback [see Pratisto et al. (2022)].P2PAP operators may wish to address these challenges during the design phase in order to increase the success of immersive applications in their platforms.
To extend the scale, distance and diversity of resources and participants, P2PAP need to maintain a large pool of hosts and guests to feed both demand and supply sides (Sutherland & Jarrahi, 2018).Hosts provide properties beyond what a traditional hospitality company might reasonably produce or acquire, offering guests a sizable and diversified set of properties.Size and diversity generate even greater demand.In order to better serve and increase demand from within their structural network, digital platforms are advised to transition users who play the role of service providers or consumers to prosumers (Lang et al., 2020).
Our study demonstrates that designing for diagnosticity in short-term rentals is about fitting P2PAP with technology that addresses the characteristics of the process.The task of evaluating a booking request is complex, as it requires the combination of multiple sources within the platform (e.g.ratings, reviews, profiles, photos) and from external sources (e.g.social media).This information is generated by users themselves and therefore doubts exist about its authenticity and credibility (Jang et al., 2021).Research has shown that users of digital platforms rely on their most frequently used heuristics to make a decision and the choice of the heuristics is conditioned by the context (Bae & Koo, 2018).One way of facilitating the hosts' decision process could involve design to reduce hosts' cognitive effort.Research focusing on guests shows that when choosing a place to stay, three information cues are relied on (reviews from guests or hosts, star ratings or number of reviews, where social media are not used as a reference), suggesting that this combination of cues allows the most accurate inferences (Zloteanu et al., 2021).P2PAP operators should focus on finding a similar combination for hosts which ensures that they have the necessary information about guests to make accurate inferences and that this information is immediately visible.Moreover, P2PAP should consider adjusting the type of information supplied to hosts and the type of information they search for (Smith et al., 2011).This will require taking into account the cognitive style of hosts [e.g.visualizers vs. verbalizers, see Bae and Koo (2018)] and giving them different evaluation options designed to suit different cognitive styles.

Limitations and future research
This study has limitations that should be noted.First, our work was conducted in Portugal, with mainly Portuguese hosts and was thus shaped by a specific cultural, social, and political context.Future research should analyze potential contingencies of the model in different settings or test the model in other technology-mediated contexts to assess the generalizability of findings.Second, we use a single data collection source, and as such, positive bias may have been shown by respondents when rating their perceptions.Consequently, future studies could consider collecting data from multiple sources.
Third, we drew on the process virtualization theory (Overby, 2008) and introduce the concept of diagnosticity to explore the factors that influenced hosts' intention to continue using short-term rentals.Future research could integrate a variable in the model that captures an affective reaction of hosts (e.g.satisfaction) to examine mediation effects.

Disclosure statement
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Notes on contributors
Iolanda Barbeitos is a doctoral candidate at NOVA IMS.She holds a master's degree in Management of Tourism Organizations from the Faculty of Economics of the University of Algarve, Portugal.Her research interests include technology post-adoption and human performance and interaction with technology in emerging contexts such as the sharing economy.She has 10 years professional experience as an IT consultant in the banking and financial sectors.Since 2011 she has worked as a project manager in the translation industry and is now aiming for an academic career.

Table 1 .
Descriptive statistics of respondents' characteristics.

Table 2 .
Summary of hypothesis test results.
Tiago Oliveira is a Full Professor of Information Management and President of the Scientific Council at the NOVA IMS.Aligning technology benefits to sustainability goals, his research interests include technology adoption in local energy communities, the digital divide, cybersecurity, and privacy.He has published papers in several academic journals including the European Journal of Information Management, Information & Management, Tourism Management, Decision Support Systems, Computers in Human Behavior, and the Journal of Business Research.With more than 28,500 citations to his name (https://scholar.google.com/citations?user=RXwZPpoAAAAJ), he was included in the prestigious 2021, 2022, and 2023 editions of the Highly Cited Researchers index.