Usability factors that drive continued intention to use and loyalty of mobile travel application.

The expeditious development of mobile applications has transformed how travel e-commerce firms enhance customer relationships to implement their business strategies successfully. However, the literature has identified that a major determinant affecting customers' considerations to reject a mobile application is a deficiency of usability. Therefore, unlike the typical spotlight on general usability variables found in the prior literature, this study concentrated on investigating more specific usability factors that drive the continued intention to use and mobile application loyalty of a mobile travel e-commerce application. To achieve this, this study utilized the mobile application usability, which we adapted to form an Indonesian version. The dataset was obtained through a survey of 248 users of an Indonesian mobile travel e-commerce application, Traveloka. The collected data were analyzed using partial least square structural equation modeling (PLS-SEM) to test the hypotheses. The results indicated that application utility and the user interface structure emerge as significant drivers for both outcomes. This study could help travel e-commerce firms to develop their mobile strategies effectively, especially in terms of their usability, which may ensure that their customers continue to use the mobile application.

Introduction

Travel and tourism have effectively substantiated how e-commerce may alter the composition of an industry and, in the course of this transformation, establish a rich profusion of business opportunities (Jaiswal, 2017). Drawing on the literature (Hart and Saunders, 1997), as seen from a business process perspective, we define e-commerce as the application of information technology (IT) to automate business transactions and workflow. In the increasingly competitive tourism market (Chang 2020), the emergence of e-commerce in delivering tourism products is well acknowledged as it has improved sophistication in business operations not only in terms of effectiveness but also cost-efficiency (Golmohammadi et al., 2012). With regard to the consumer aspect, the global nature of e-commerce allows for cost-efficient operations enabling firms to access new markets, attract new customers, deliver products and services, and collaborate with business partners (Zhu and Kraemer, 2002). Despite the industry's trend towards digitalization, customer relationships are still key to the success of the increasingly commoditized travel and tourism industry (Mamaghani, 2009). However, integrating online technology into existing business strategies is a crucial challenge for firms, essentially in finding unique ways to attract and retain their customers (Hoehle and Venkatesh, 2015).

Based on the Digital 2022 report, the number of Internet user has increased, with 5.32 billion people (a penetration of 67% of the population) being mobile users (Kemp, 2022). In Indonesia, as many as 160.7 million Indonesians (60% penetration of the population) have utilized a mobile device to search for information on the Internet (APJII, 2019). Given that the Internet and mobile devices are being used more often due to global connectivity, Budd and Vorley (2013) remarked that the utilization of mobile technology by travel agent firms is aimed at reaching new target markets, minimizing distribution costs, and increasing customer satisfaction by providing seamless travel experiences. Consequently, mobile applications have gradually changed, transforming how travel agent firms reach out to their customers (Ostdick, 2016). Now more than ever, mobile applications are an effective approach for travel agent firms, not just to control their customers' experience and improving the relationship that affects customers' behavior towards the application (Tsang et al., 2004). From individual travelers' perspectives, mobile applications using technology such as intelligent travel assistants can play an important role in a tourist's life cycle (Phaosathianphan and Leelasantitham, 2020).

Thus, considering the mobile application is an effective approach, a well-designed mobile application could help users achieve their goals efficiently and increase their satisfaction with the application (Hoehle et al., 2016). Venkatesh and Ramesh (2006) suggested that the usability (an extension of ease of use) of a mobile application is specifically essential to attain an effective and remarkable user experience. However, market research demonstrated that usability deficiency is a major determinant influencing customers' considerations to reject a mobile application (Forrester Research, 2010). Furthermore, based on the literature review (Hong et al., 2004; Venkatesh and Ramesh, 2006; Wells et al., 2011b), usability has a positive association with the continued intention to use a mobile application. In addition, Forrester Research (2010) suggested that the usability of a mobile application influenced mobile application loyalty. Probing deeper into the travel and tourism context, prior research (Fang et al., 2017) has also suggested that ease of use is an important factor that influences users' intention-to-use behavior when using a mobile travel application.

Several existing related studies have focused on finding the driving factors that influence the intention to use travel e-commerce platforms (Bhatiasevi and Yoopetch, 2015; Kim et al., 2009; Renny et al., 2013; Suki and Suki, 2017). While usability is an important driver, a gap can be identified in the literature. Previous studies have mainly focused on examining usability factors in general (e.g., perceived ease of use, perceived usefulness), paying limited attention to the specific usability factors (e.g., information structure, content relevance, user interface graphics) that influence users' intention to use in the mobile context. To better understand these factors and address the gap in the literature, in this study, we believe it is important to investigate which usability factors in the mobile context affect users' intention to use and mobile application loyalty of a mobile travel e-commerce application. Thus, our study sought to enrich the scientific discourse in the usability and e-commerce literature using the specific case of an Indonesian mobile travel e-commerce application.

Literature review

This section reviews associated research as references for this study, including travel e-commerce and usability dimensions, mobile application usability, and the mobile application usability (MAU) instrument.

Travel e-commerce and usability evaluation dimensions

The expeditious development of mobile applications has transformed the ways in which travel e-commerce firms enhance their relationships with their customers to successfully implement their business strategies (Tsang et al., 2004). In responding to this challenge, our literature review found that related studies have focused on investigating the key factors that influence users' intention to use a travel e-commerce platform. However, we recognized the following limitations in the prior literature.

First, while some research (Forrester Research, 2010) indicated that the deficiency of usability is a major factor influencing customers' considerations to reject a mobile application, much of the prior literature only examined usability factors in general (e.g., perceived ease of use, perceived usefulness), paying limited attention to specific usability factors (e.g., information structure, content relevance, user interface graphics) when investigating the factors that influence users' intention to use a travel e-commerce platform. Our literature review found that much of the research used a framework posited in Davis' (1989) technology acceptance model (TAM), which emphasizes that ease of use and usefulness are two factors that determine users' acceptance of a system or technology. For instance, in Bhatiasevi and Yoopetch's (2015) study on travel e-commerce sites, Renny et al. (2013) and Kim et al.'s (2009) studies on online airline e-booking sites, as well as Suki and Suki's (2017) research on mobile airline e-booking applications, the usability variable integrated into the model was only represented by the ease-of-use variable. Similarly, other studies on the use of mobile travel e-commerce applications in Indonesia Traveloka suggest that ease-of-use has a positive effect on continued usage (Santoso, 2021), and its website design may also influence consumer purchasing decisions (Mahendratmo and Ariyanti, 2019). In addition, other studies have used concepts from psychological research, such as the stimulus–organism–response (S–O–R) model, to conduct such research (e.g., Fang et al., 2017). Second, we found that there has been limited literature devoted to probing the study of travel e-commerce in the mobile application context. Much of the literature has investigated this in the context of general websites (e.g., Bhatiasevi and Yoopetch, 2015; Kim et al., 2009; Renny et al., 2013). However, based on Similar Web (2020), most travel e-commerce users access the service specifically using a mobile application (73%) rather than using websites via deskbottom (27%). This shows that the customers of travel e-commerce come dominantly from the users of the mobile application.

Thus, we need a framework that holistically evaluates travel e-commerce applications in the mobile context to inquire into the specific usability factors that influence users’ intention to use. The next sections discuss the exploration of mobile application usability instrument in solving the issues posed earlier.

Mobile application usability

Educing from the definition of usability according to the International Standards Organization (ISO), we define mobile application usability as the extent to which a mobile application can be used by specified users to achieve specified goals with effectiveness, efficiency, and satisfaction in a specified context of use (Venkatesh and Ramesh, 2006). It is necessary to consider that the term mobile application usability is distinct from the term mobile device usability, which donates to the extent of user-friendliness of an operating system (Adipat et al., 2011). We reviewed the literature on information systems (IS) and human-computer interaction (HCI) regarding how mobile applications were conceptualized, as well as the corresponding dimension used to evaluate mobile application usability. Grounded on our study, we discovered the following four limitations in the prior literature about mobile application usability.

First, in much of the literature we examined, mobile application usability has been conceptualized and evaluated without emphasizing the context-specific factors which are pertinent for users when operating mobile applications. Many of the studies typically used the conceptualization of mobile application usability as evolved from website usability (e.g., Venkatesh and Ramesh, 2006). Second, we found that the majority of mobile usability measures only used the dimensions that integrated usability results or outcomes (e.g., effectiveness, efficiency, learnability) without explaining the specific usability factors (e.g., information structure, content relevance, user interface graphics) that rationalized users' perceptions about the associated usability outcomes established in the mobile context. In addition, the instrument does not explain the relationships among variables, i.e., how the variables affect one another (e.g., Baharuddin et al., 2013; Coursaris and Kim, 2011; Hussain, 2012; Tan et al., 2013). Third, our literature review found that much of the research that evaluated mobile applications was conducted in laboratory environments. Despite being able to identify error rates and task completion time, this scenario is unable to estimate and elucidate why a certain task requires longer or shorter periods to be completed by users (e.g., Adipat et al., 2011; Ziefle and Bay, 2006). Fourth, we discovered that mobile application usability had been conceptualized and interpreted inconsistently by different researchers. Some studies argued that effectiveness, efficiency, and satisfaction were constituent assessments of mobile application usability instead of being an outcome of overall mobile application usability. For instance, Baharuddin et al. (2013) used ten design elements, namely effectiveness, efficiency, satisfaction, usefulness, aesthetics, learnability, simplicity, intuitiveness, understandability, and attractiveness, to measure mobile application usability. Furthermore, we recognized that other studies merged concepts typically found in the TAM literature (e.g., usefulness and ease of use) (Bhatiasevi and Yoopetch, 2015; Kim et al., 2009; Renny et al., 2013; Suki and Suki, 2017) with HCI principles (e.g., user interface attractiveness) (Ramadhan et al., 2019) and concepts from marketing research (e.g., satisfaction and attitude) (Coursaris and Kim, 2011).

In summary, when associating such concepts with mobile application usability, there are notable theoretical and methodological obscurities in both conceptualizing and measuring mobile application usability found in prior studies and literature. The next section provides a further discussion to address the issues raised earlier.

The MAU instrument

To respond to the prior issues, and as informed by Hoehle and Venkatesh (2015), the MAU has emerged as an instrument that can help identify the impact arising from certain usability principles on the overall usability of a mobile application, hence allowing practitioners to determine which features are needed and what would make it easier for users to use the mobile application. The framework was developed based on Apple's user experience guidelines ascribed to the company's reputation in the mobile application market and its success often being associated with its user-friendly interface (Hoehle and Venkatesh, 2015).

The structural model of the MAU posits three levels, namely the first-order construct, which covers more specified usability factors; the second-order construct, which covers more general usability factors; and the third-order construct, which contains the outcome variables (see Figure 1). Further explanations regarding the model are as follows.

Figure 1

The MAU structural model.

Second-order construct: application design

Application design refers to the degree to which a user perceives that a supporting feature in a mobile application is effectively designed to support its main utility. Prior studies have evinced that when appraising the design of a mobile application, users are influenced by several factors, namely branding, data preservation, instant start, and orientation. First, branding defines the degree to which a user perceives that the mobile application integrates branding pertinently. Dou et al.’s (2010) research on website design has suggested that subtle branding essays have an impact on users’ satisfaction with the overall website design. Second, data preservation defines the degree to which a user perceives that the mobile application preserves data automatically. This is consistent with prior research that has suggested that it is important for mobile applications to preserve data that the user inputs (Adipat et al., 2011). When users are required to enter data more than once (e.g., when moving from one screen to another), they may become exasperated and dissatisfied with the application (Adipat et al., 2011; Kurniawan, 2008; F. B. Tan et al., 2009). Third, instant start defines the degree to which a user perceives that the mobile application starts immediately after switching it on. Mobile applications that are only gradually ready to be used after being switched on usually lead to users becoming frustrated (Devaraj et al., 2002; Galletta et al., 2006). Finally, orientation defines the degree to which a user perceives that the mobile application can be used comfortably, despite how the device is being held, horizontally or vertically. Practitioners must pay attention to the information displayed on the mobile application when the device is being operated in landscape or portrait orientation (Wobbrock et al., 2008).

Second-order construct: application utility

Application utility is defined as the extent to which a user perceives how well a mobile application serves its basic and main purpose. Prior literature has indicated that when evaluating the overall utility of a mobile application, users are affected by several determinants, namely content relevance, search, and collaboration. First, content relevance defines the degree to which a user perceives that the mobile application spotlights the most apposite information. Mobile applications must focus on displaying the most relevant content for their users (Venkatesh and Ramesh, 2006). Furthermore it is important for users to meet users' expectations and that the main objectives of the applications are emphasized (X. T. Li et al., 2009; Thong et al., 2002; Wells et al., 2011a; Wells et al., 2011b). Second, search defines the degree to which a user perceives that the mobile application helps them to explore and find information. A search feature leads to an improved user experience since it assists users in navigating applications quicker and easier (T. Hess et al., 2009; Wells et al., 2005). Last, collaboration defines the degree to which a user perceives that the mobile application allows them to connect with other individuals. Applications should facilitate collaboration to assist users in sharing information with other users (T. Hess et al., 2009; Oulasvirta et al., 2007).

Second-order construct: user interface graphics

User interface graphics refers to how effectively designed a mobile application's user interface graphics are as perceived by its users. Prior studies have suggested that when appraising the predominantly user interface graphics of a mobile application, users are affected by several factors, namely subtle animation, realism, and aesthetic graphics. First, subtle animation defines the degree to which a user perceives that the mobile application incorporates animations subtly. Animations used in the mobile application should be designed delicately and are not extensively used to prevent luridness (T. Hess et al., 2009; Lim et al., 2000). Second, realism defines the degree to which a user perceives that the mobile application makes use of realistic icons or pictures. The embodiment of realistic icons or images in mobile applications can enhance user experience (Kang, 2007). Finally, aesthetic graphics defines the degree to which a user perceives that the mobile application displays aesthetic graphics. It is important that the graphics are designed to be aesthetically pleasing (Aladwani and Palvia, 2002; Cyr et al., 2009; T. J. Hess et al., 2005; Wells et al., 2011a; Wells et al., 2011b).

Second-order construct: user interface input

User interface input is defined as the extent of how easy it is for users to input data into a mobile application. Prior literature has revealed that when evaluating the overall user interface input of a mobile application, users are influenced by several determinants, namely fingertip-size controls, control obviousness, effort minimization, and the de-emphasis of user settings. First, fingertip-sized controls define the degree to which a user perceives that the mobile application employs fingertip-size controls. An appropriate size of controls could ease users to select functions in mobile applications (Kurniawan, 2008). Second, control obviousness defines the degree to which a user perceives that the mobile application employs controls that are immediately apparent and comprehensible. It is necessary that the controls and buttons are designed to be intuitive to use (Jokela et al., 2006; Seffah et al., 2006; Sørensen and Al-Taitoon, 2008). Users will become frustrated if it takes a long time to learn how to use a mobile application; hence, the user interface should be clear when the application is used for the first time. Third, effort minimization defines the degree to which a user perceives that the mobile application minimizes the essay to input data. The effort users put in to enter data should be as little as possible. Mobile applications must provide features that support users to input easily, such as providing options in drop-down menus or supporting automatic data entry. Finally, the de-emphasis of user settings defines the degree to which a user perceives that the mobile application de-emphasizes user settings. Although settings is an essential feature, it should be designed to not frequently prompt users for adjustment when they are using a mobile application (Sørensen and Al-Taitoon, 2008; F. B. Tan et al., 2009; Tung et al., 2009).

Second-order construct: user interface output

User interface output refers to the extent to which a user perceives that contents are presented effectively on a mobile application. Prior studies have suggested that when appraising the general user interface graphics of a mobile application, users are affected by several factors, namely user-centric terminology, concise language, and standardized user interface elements. First, user-centric terminology defines the degree to which a user perceives that the mobile application makes use of user-centric terminology. Mobile applications should use terms that users can understand easily, and they should avoid technical jargon. Text that contains technical terms is often more difficult to read and understand, leading to users becoming frustrated. Second, concise language defines the degree to which a user perceives that the mobile application employs concise language. The text should be written as concisely as possible, and the use of overly lengthy descriptions should be avoided (Gebauer et al., 2010; Jokela et al., 2006; Kurniawan, 2008; Robbins and Stylianou, 2003). Last, standardized user-interface elements define the degree to which a user perceives that the mobile application makes use of standardized user interfaces that other mobile applications commonly use. The implementation of standardized user-interface elements makes users feel that they are familiar with the interface being used (T. Hess et al., 2009; Wells et al., 2011a; Wells et al., 2011b).

Second-order construct: user interface structure

User interface structure is defined as the extent to which a user perceives that a mobile application is well-structured. Prior literature has evinced that when evaluating the overall user interface structure of a mobile application, users are influenced by several determinants, namely a logical path and top-to-bottom structure. First, top-to-bottom structure defines the degree to which a user perceives that the mobile application displays frequently used content at the top of the interface. Users often intuitively scan the most important and relevant information starting from the top of the screen (T. X. Li et al., 2009; Valacich et al., 2007; Wells et al., 2011b); hence, the main information needs to be placed at the top. Second, a logical path defines the degree to which a user perceives that the mobile application presents information logically and predictably. Mobile applications need to implement logical steps, and users should be able to easily predict the direction of the process (Adipat et al., 2011; T. Hess et al., 2009; Wells et al., 2011a; Wells et al., 2011b).

In summary, MAU was developed to assess the more specific usability factors in the context of mobile applications to investigate their relationships with users' continued intention to use and mobile application loyalty. The instrument has six variables in the first-order construct, 19 in the second-order construct, and two outcome variables in the third-order construct. The measure's constructs were based on Apple's user experience guidelines and reinforced by suggestions from prior literature. Thus, the instrument is expected to be an appropriate measure for this study.

Study approach

This section elucidates the procedures conducted to finalize this study. As seen in Figure 2, the procedure consists of a total of six steps. Steps one and two have been discussed in the previous sections. Accordingly, this section spotlighted an explanation of the instrument adaptation method, data collection method, and data analysis and assessment procedures.

Figure 2

The study's methodological approach.

Instrument adaptation into an Indonesian version

Globalization enables the creation of multinational and multicultural research. However, the appropriateness of adapting measures for use in a language other than the source language needs to be considered. In this study, the measure being used was based on the original English version but was intended to be used in the research scope of the Indonesian population. Thus, following the determined scenarios of cross-cultural adaptation application (Guillemin et al., 1993), we recognized that the measure was to be used across cultures. Prior literature (Beaton et al., 2000) has suggested that a measure used in different nations, cultures, and languages should follow a particular cross-cultural adaptation procedure, thus being able to produce equivalency between the source version and target version exhaustively based on the content. Hence, the items from the measure must not only be translated appropriately in terms of their linguistics, but they should also be adapted in terms of culture to maintain their content validity at the conceptual level across different cultures.

Based on the literature on the cross-cultural adaptation approach (Beaton et al., 2000), this study developed an adaptation of the MAU instrument in Indonesian based on a five-step procedure, namely forward translation, synthesis, back-translation, expert study, and face validity (see Figure 3). The explanation of each procedure is as follows.

Figure 3

The cross-cultural adaptation procedure.

The first step of cross-cultural adaptation is forward translation, which means translating the source (English) into the target (Indonesian). Two bilingual translators whose mother tongue was Indonesian worked to produce two independent translations (T1 and T2). These translators had different profiles and backgrounds. The first translator, which was the author of this study, was well aware of the context of the questionnaire being translated. Meanwhile, the second translator was neither aware nor informed of the context being quantified and had no background in HCI. The second translator was a fluent English speaker with a background in law.

After this, the two translators sat together to discuss each of the produced translations by comparing and evaluating their translations based on the source questionnaire. They subsequently synthesized them into a single translation, namely T-12. A document was made to summarize the issues addressed and how they were resolved during the synthesis phase. Thereafter, another translator who spoke both English and Indonesian translated the T-12 back into the original language (B1) without being informed of the original items. The translator was also neither aware nor informed of the context being examined and had a background in medical healthcare.

Then, an expert reviewed all the materials, including the original questionnaire, each translation (T1, T2, T-12, and B1), and the corresponding written reports, to reach a consensus on the discrepancies. As a result, a pre-final translation of the questionnaire was produced and was subsequently tested on five target users through face validity. The face validity was conducted as a subjective gauge that assessed whether the respondents understood the questionnaire well enough. In addition, each perception of the items was explored to ensure that the adapted version retained its equivalence in the applied context. After going through two iterations of the cross-cultural adaptation, the questionnaire was ultimately translated into an appropriate final version, which can be seen in Appendix A.

Data collection

In this quantitative research, data was collected through an online survey using the MAU instrument, which had been adapted into Indonesian beforehand. The sample of this study comprised Traveloka mobile application users who have made transactions with the application. As the most-used travel e-commerce application in Indonesia (Similar Web, 2020), we chose Traveloka as the object of the evaluation. Accordingly, it was expected to be an excellent representation of mobile travel applications that have been successful in the application market. The questionnaire was created using Google Forms, with distribution carried out through Instagram, WhatsApp, and Twitter. The responses were measured using a Likert scale ranging from strongly disagree (1) to strongly agree (7). We obtained informed consent from all participants through the questionnaire, ensuring their voluntary and anonymous participation as well as data confidentiality. Research ethical consideration for this study was approved by the Manager of Research and Community Services at the Faculty of Computer Science, Universitas Indonesia.

First, we inspected the sample's data quality by omitting invalid responses, resulting in a final dataset of 230 valid cases. In sum, from 264 total responses, 34 responses were excluded from the sample, as five respondents had not used Traveloka, 18 respondents accessed it from the mobile web, and 11 respondents failed to answer the reverse-coded item question correctly. This number of respondents is regarded as sufficient based on the inverse square root method (Kock and Hadaya, 2018), which requires a minimum of 129 samples for this case. The authors observed that most of the respondents who participated in our survey were women (70.87%). In terms of age group, most of the respondents were 20–29 years (73.48%). It was also observed that the respondents came from diverse fields: A large number of respondents were students (47.39%), followed by those in the government or military field (9.56%). Table 1 summarizes the complete characteristics of the sample.

Table 1

Characteristics of the sample.

Variable	Categories	Sample size (n = 230)	Percentage (%)
Gender	Male	67	29.13
	Female	163	70.87
Age groups	<20	23	10.00
	20–29	169	73.48
	30–39	14	6.09
	40–49	8	3.48
	50–59	16	6.95
Work field	IT	18	7.83
	Banking and financial	13	5.65
	Government and military	22	9.56
	Medical healthcare	8	3.48
	Construction and engineering	8	3.48
	Education	13	5.66
	Marketing and advertising	9	3.91
	Student	109	47.39
	Other	30	13.04
Primary phone use	Android	170	73.91
	iPhone	60	26.09

Data analysis

Afterward, the data was processed using Smart-PLS 3 software. In this study, the constructs of collaboration and orientation were not included in the evaluation. The collaborative construct was removed because, fundamentally, a travel e-commerce application does not aim to collaborate or connect with other people. Further, the orientation construct was omitted because Traveloka does not support an automatic screen orientation adjustment. The structural model used in this evaluation can be seen in Figure 4.

Figure 4

The structural model used in the study.

Results and discussion

In this section, the data analysis results are explained in two separate sections, namely the measurement model and structural model assessment. The evaluation was tested using the partial least square structural equation modeling (PLS-SEM) algorithms. As Hair et al. (2011) suggested, we chose the PLS-SEM over the covariance-based structural equation modeling (CB-SEM) because our research is explanatory. Hence, the PLS-SEM was a more appropriate method to gauge the model.

Measurement model evaluation

To confirm the construct reliability and validity, a measurement model evaluation was carried out. The assessment included item evaluations, namely indicator reliability, through the inspection of the outer loading value from each item (see Appendix B). In accordance with guidelines from the literature (Hulland, 1999), some items with outer loadings below the value of 0.40 were omitted (BRAN2, EMM4, FTSC3, FTSC4). All item-to-construct loadings varied in the range of 0.619 and 0.959, and only two loadings were less than 0.70 (STAR2 and TTPS5), thus supporting indicator reliability (Hair et al., 2019). In addition, all constructs indicated composite reliability that was higher than the minimum value of 0.70, as Bagozzi and Yi (1988) suggested. Moreover, the average variance extracted (AVE) results surpassed the suggested value of 0.50 (Bagozzi and Yi, 1988), showing that the convergent validity was well established. Next, to assess the discriminant validity, we used the heterotrait-monotrait (HTMT) ratio of correlation: “The mean value of the item correlations across constructs relative to the (geometric) mean of the average correlations for the items measuring the same construct.” Henseler et al. (2015) propose a maximum threshold value of 0.90 for satisfaction. Our research showed that three of HTMT values exceeded the threshold, namely user interface input to application utility (0.905), user interface structure to user interface graphics (0.903), and user interface structure to user interface output (0.945). This may be due to the involved constructs being conceptually familiar one to another as output variables. However, when discriminant validity is assessed using Fornell and Larcker's (1981) criteria, i.e., the AVE for each construct should have greater value than the squared correlations between constructs, all items share more common variance with their respective constructs than with other constructs. Hence, the result showed that our constructs demonstrated discriminant validity when assessed using Fornell and Larcker's criteria. Moreover, collinearity was assessed using the variance inflation factor (VIF), with the suggested threshold is below the value of five (Hair et al., 2019). The results showed that all of the VIF values satisfy the criteria.

The remainder of the results regarding the measurement model evaluation and VIF can be seen in Appendix C and Appendix D.

Structural model evaluation

Having established an appropriate measurement model, we subsequently evaluated the relationships, as well as the path coefficients, among the constructs. To be considered significant, the path coefficient should exceed 0.10 and have a p-value below 0.10 for the two-tailed test, with a significance level of 10% (Hair et al., 2011). In this case, the results indicated that all path coefficients in the first-order construct, except for de-emphasized user settings and standardized user-interface elements, satisfied the criteria. These two constructs were shown to have negative and insignificant relationships with their pairs (de-emphasized user settings toward user interface input and standardized user-interface elements toward user interface output). Hence, it can be inferred that de-emphasized user settings was not a factor influencing users' perception of the overall design of Traveloka's interface input. In addition, with regards to the overall user interface output, users did not consider standardized user-interface elements as one of the factors influencing their perceptions. We noticed that this may have occurred since most of the icons used in a travel application context are distinct from the others' (e.g., transportation- and accommodation-related icons). Despite its uniqueness, Traveloka provides a description under each icon, hence mitigating any confusion that may arise when users try to understand uncommon symbols. The complete results can be seen in Appendix E.

Meanwhile, the results of the path-coefficient assessment in the second-order construct suggested that only application utility and user interface structure had meaningful association with continued intention to use and mobile application loyalty (see Table 2 and Figure 5). After delving deeper into the results, we noticed a specific pattern that occurred among these relations. Compared to application design and user interface graphics/input/output, the significant variables were two factors that fundamentally assist users to get the main task done when purchasing a product in a mobile travel application, namely application utility and user interface structure.

Table 2

The second-order construct's structural path estimates.

Second-order construct	CITU1	MAL2
Application design	-0.087	-0.065
Application utility	0.437∗∗∗	0.236∗∗
User interface graphics	-0.032	0.140
User interface input	0.193	0.080
User interface output	-0.073	0.026
User interface structure	0.237∗∗	0.254∗

∗p < 0.10; ∗∗p < 0.05.; ∗∗∗p < 0.01.

Continued intention to use.

Mobile application loyalty.

Figure 5

The second-order construct's structural model assessment results.

The application utility helps users to search and find the travel product they want to purchase and serves them with relevant information e.g., search and filter feature for all travel products. In addition, the application utility also suggests that the application can also facilitate content relevance such as displaying and sharing of appropriate travel information e.g., purchased tickets and boarding passes. Furthermore, the user interface structure assists the user to complete basic process tasks such as when purchasing a travel product through a logical booking path and top-to-bottom information structure corresponding to users’ needs and priorities.

On the other hand, other usability factors, namely application design, user interface graphics, user interface input, and user interface output, did not produce significant effect on continued intention to use and mobile application loyalty. These four factors are traditionally considered as the visual appearance aspect of a mobile application as opposed to its functional aspect. These findings seem to contradict previous research by Fang et al. (2017) who suggested that attractive appearance affects the intention of use through the psychological engagement variable. In other words, in the case of Fang et al. (2017), the relationships between interface attractiveness are not through usability variables. Thus, the existing literature cannot be said to be conceptually equivalent to this study.

Upon inspecting the Traveloka application ourselves, we suspect that the lack of support of the four factors may be due to several reasons. For overall application design, as an application that is typically not used routinely, users might not bother much about application design elements such as instant start and data preservation. Another speculation is that the Traveloka application seems to deliver a well-implemented instant start and data preservation experience as previously filled out information for returning customers are saved or preserved within the application locally, even if the users are not logged in to their accounts. Similar instance may also be applicable to the user interface graphics/input/output as the application provides convenient input mechanisms with relevant output interfaces and is supported by aesthetically pleasing and realistic graphics with minimal animation. This indicates that users may tend to prioritize functions over user interfaces and graphics aspects of a mobile application.

In terms of the predictive capacity of the model, the r-square values were assessed for all six constructs in the second-order and the two constructs in the third order. As Hair et al. (2011) suggested, the r-square values of 0.25, 0.50, and 0.75 can be described as weak, moderate, and substantial, respectively. In this study, as seen in Figure 5, the results of the PLS-SEM algorithm showed that the r-square for continued intention to use was 0.414, while for mobile application loyalty, it was 0.380. This indicates that the six variables in the second-order construct explained 41% of the variance in continued intention to use and 38% of the variance in MAU. This indicates that the model had weak to moderate predictive capacity in explaining both continued intention to use and mobile application loyalty. Meanwhile, the r-square variables in the second-order construct were 0.468 for application design, 0.417 for application utility, 0.501 for user interface graphics, 0.508 for user interface input, 0.448 for user interface output, and 0.459 for user interface structure. Hence, these results also indicate that the predictive capacity of the model was weak to moderate in explaining each of the second-order construct variables.

Based on the results, we offer suggestions for practitioners who are looking to develop a mobile travel e-commerce application that can maximize continued intention to use and loyalty. Practitioners should consider prioritizing the dimensions of application utility and user interface structure over other mobile application usability dimensions. Specifically, practitioners should place an initial focus on improving the main user experience of task completion when purchasing a travel product before concentrating on application design and/or user interface graphics/input/output. We propose that these results may potentially contribute to travel e-commerce firms maturing their mobile strategies effectively, especially in terms of their usability, which may ensure that their customers continue using the application.

Conclusion

This study used the MAU model to understand how specific usability factors could affect users’ decisions to use a mobile travel e-commerce application and stimulate mobile application loyalty. The results suggest that practitioners should pay careful attention to the dimensions of application utility and user interface structure, with an initial focus on improving the main user experience of task completion when purchasing a travel product before concentrating on application design and/or user interface graphics/input/output. To confront the lack of usability in mobile travel e-commerce applications, we hope that this study may contribute to travel e-commerce firms developing their mobile strategies effectively, especially in terms of their usability, which may ensure that their customers continue to use the mobile application.

Despite the objective of this study having been achieved, it had some methodological limitations. For example, the use of a small number of samples limited the comprehensiveness of our findings. Other than that, the analysis of the study implies that only application utility and user interface structure have a positive association with continued intention to use. Since the framing type of task done by the users on the application were not asked specifically through the survey, whether it is framed under a hedonic or utilitarian mindset (Cyr and Head, 2013), we have limitation in discovering whether the results apply equally on both of the mindsets or tend to incline to one side. The study also did not differentiate between experienced and novice users, hence the result might dominantly reflect one of the two groups of users. Therefore, a further study that includes framing the type of task and user's experience level as consideration in supporting the analysis could make the study more comprehensive. Furthermore, using a mixed-methods approach comprising both quantitative and qualitative research could benefit this study, allowing for new insights and further explanations to be found, which could enrich and extrapolate on what has been found in the previous literature.

Declarations

Author contribution statement

Panca O. Hadi Putra & Indra Budi: Conceived and designed the experiments; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Raden Ajeng Wuriandita Wahyumurti Candra Kirana Dewi: Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Funding statement

Dr. Panca O. Hadi Putra was supported by [BA-1199/UN2.RST/PPM.00.03.01/2020].

Data availability statement

Data will be made available on request.

Declaration of interest's statement

The authors declare no conflict of interest.

Additional information

No additional information is available for this paper.

Table A1

Items in Indonesian

Label	Item
AEST1	Aplikasi Traveloka menggunakan gambar (foto atau ilustrasi) yang bagus.
AEST2	Aplikasi Traveloka menggunakan grafis yang beragam, bagus, dan memikat sehingga menarik perhatian Anda pada aplikasi tersebut.
AEST3	Aplikasi Traveloka menggunakan grafis yang menarik perhatian.
AEST4	Aplikasi Traveloka diuntungkan karena memiliki grafis yang bagus dan memikat.
BRAN1	Aplikasi Traveloka menggunakan gambar atau warna khas yang lembut dan tidak mencolok.
BRAN2	Aplikasi Traveloka tidak memaksa saya menyaksikan suatu iklan.
BRAN3	Tanpa Anda sadari, Aplikasi Traveloka mengingatkan Anda pada ciri khas aplikasi tersebut.
BRAN4	Aplikasi Traveloka menyatukan desain khas aplikasinya secara konsisten (serasi).
CLAN1	Aplikasi Traveloka hemat dalam menggunakan kata tanpa mengurangi makna.
CLAN2	Aplikasi Traveloka menggunakan bahasa yang singkat, padat, dan jelas.
CLAN3	Aplikasi Traveloka menyampaikan informasi dengan ringkas.
CLAN4	Aplikasi Traveloka menggunakan teks yang tepat dan ringkas.
COLL1	Aplikasi Traveloka membantu Anda berbagi informasi dengan orang lain.
COLL2	Aplikasi Traveloka memungkinkan Anda terhubung dengan orang lain.
COLL3	Aplikasi Traveloka mendukung Anda melakukan kolaborasi dengan orang lain.
COLL4	Aplikasi Traveloka membantu Anda berinteraksi dengan orang lain.
COOB1	Aplikasi Traveloka membuat Anda memahami fungsi utama aplikasinya dengan cepat.
COOB2	Aplikasi Traveloka memberikan instruksi yang mudah dipahami.
COOB3	Aplikasi Traveloka memiliki kontrol yang fungsinya cepat dipahami.
COOB4	Aplikasi Traveloka menerapkan kontrol yang mudah digunakan dan dipahami.
CRLV1	Aplikasi Traveloka mempertegas informasi yang ingin Anda temukan.
CRLV2	Aplikasi Traveloka mempertegas informasi yang penting bagi Anda.
CRLV3	Aplikasi Traveloka mempertegas informasi yang Anda pedulikan.
CRLV4	Aplikasi Traveloka mengangkat informasi yang relevan (penting) bagi Anda.
DAPR1	Ketika ditutup, aplikasi Traveloka secara otomatis menyimpan data Anda.
DAPR2	Aplikasi Traveloka tidak mengharuskan Anda menyimpan data secara manual ketika keluar dari aplikasi.
DAPR3	Aplikasi Traveloka menyimpan data secara otomatis dan Anda bisa mulai kembali di tempat Anda mengakhiri sesi Anda sebelumnya.
DAPR4	Aplikasi Traveloka memungkinkan Anda keluar dari aplikasi dan mulai pada tahapan yang sama ketika masuk kembali.
DUS1	Aplikasi Traveloka tidak meminta Anda mengatur preferensi dalam aplikasi.
DUS2	Aplikasi Traveloka mengurangi penekanan pada pengaturan pengguna.
DUS3	Aplikasi Traveloka tidak memaksa Anda mengganti pengaturan dalam aplikasi.
DUS4	Aplikasi Traveloka tidak meminta Anda mengubah pengaturan dalam aplikasi.
EMM1	Aplikasi Traveloka mempermudah Anda dalam memasukkan pilihan Anda.
EMM2	Aplikasi Traveloka meminimalkan upaya Anda untuk mengetik informasi.
EMM3	Aplikasi Traveloka menyediakan pilihan pada kolom isian sehingga Anda tidak perlu mengetik.
EMM4	Aplikasi Traveloka memungkinkan saya menjalankan proses tanpa harus memasukkan data.
FTSC1	Aplikasi Traveloka menggunakan kontrol berukuran ujung jari.
FTSC2	Aplikasi Traveloka menggunakan tombol berukuran ujung jari.
FTSC3	Aplikasi Traveloka menggunakan kontrol berukuran besar.
FTSC4	Aplikasi Traveloka menggunakan kontrol berukuran kecil yang mengharuskan Anda memilih dengan seksama sebelum menekannya.
LP1	Aplikasi Traveloka memberikan langkah-langkah yang logis untuk diikuti pengguna.
LP2	Aplikasi Traveloka menerapkan langkah-langkah yang logis.
LP3	Aplikasi Traveloka menyediakan langkah-langkah yang logis untuk diikuti pengguna.
LP4	Aplikasi Traveloka menerapkan langkah-langkah yang bisa diprediksi.
ORIE1	Aplikasi Traveloka tidak memaksa Anda mengubah orientasi layar (memutar perangkat).
ORIE2	Aplikasi Traveloka bisa digunakan dengan nyaman terlepas dari cara Anda memegang perangkat Anda.
ORIE3	Aplikasi Traveloka memutar tampilan konten ketika Anda mengubah orientasi layar perangkat (horizontal atau vertikal).
ORIE4	Aplikasi Traveloka bisa digunakan dengan nyaman walaupun Anda memegang perangkat Anda secara horizontal atau vertikal.
REAL1	Aplikasi Traveloka menggunakan ikon (simbol) atau gambar realistis (misal, tempat sampah) untuk membantu Anda memahami fungsinya dengan lebih baik.
REAL2	Aplikasi Traveloka membantu Anda mengerti fungsinya dengan memberi label berupa ikon (simbol) atau gambar realistis (misal, tempat sampah).
REAL3	Aplikasi Traveloka menggunakan ikon (simbol) atau gambar sehari-hari untuk menggambarkan fungsinya (misal, tempat sampah untuk menghapus objek).
REAL4	Aplikasi Traveloka menggunakan ikon (simbol) atau gambar realistis (misal, tempat sampah) dalam berkomunikasi.
SANM1	Aplikasi Traveloka menggunakan animasi secara efektif dalam menyampaikan informasi.
SANM2	Aplikasi Traveloka menggunakan animasi sesuai kebutuhan.
SANM3	Aplikasi Traveloka tidak berlebihan dalam menggunakan animasi.
SANM4	Aplikasi Traveloka menggunakan animasi sederhana dalam menyampaikan informasi.
SEAR1	Ketika Anda mencari informasi, aplikasi Traveloka menampilkan hasil pencarian saat Anda mengetik.
SEAR2	Aplikasi Traveloka membantu Anda mencari informasi melalui suatu kotak pencarian.
SEAR3	Ketika Anda harus mencari informasi, aplikasi Traveloka menampilkan suatu kotak pencarian.
SEAR4	Aplikasi Traveloka mempermudah pencarian informasi.
STAR1	Aplikasi Traveloka muncul dengan cepat dan memungkinkan Anda langsung menggunakannya.
STAR2	Aplikasi Traveloka membutuhkan waktu lama untuk mulai digunakan.
STAR3	Aplikasi Traveloka tidak memerlukan waktu lama untuk mulai digunakan.
STAR4	Setelah dibuka, aplikasi Traveloka bisa langsung digunakan.
SUI1	Aplikasi Traveloka memiliki tombol dan ikon (simbol) yang mirip dengan aplikasi lain.
SUI2	Aplikasi Traveloka memiliki tombol dan ikon (simbol) yang pernah saya gunakan di aplikasi lain.
SUI3	Aplikasi Traveloka menggunakan tombol dan ikon (simbol) yang pernah Anda lihat di aplikasi lain.
SUI4	Aplikasi Traveloka menggunakan ikon (simbol) standar yang sudah Anda ketahui dari aplikasi lain.
TTPS1	Aplikasi Traveloka meletakkan informasi yang paling sering Anda gunakan di urutan atas.
TTPS2	Aplikasi Traveloka menampilkan informasi paling penting di bagian atas layar.
TTPS3	Aplikasi Traveloka mencantumkan informasi paling penting di bagian atas layar.
TTPS4	Aplikasi Traveloka mencantumkan fitur yang paling sering Anda gunakan di urutan teratas.
TTPS5	Aplikasi Traveloka menempatkan fitur yang paling jarang Anda gunakan di urutan bawah.
TTPS6	Aplikasi Traveloka menempatkan fitur yang paling sering Anda gunakan di urutan atas.
UCT1	Aplikasi Traveloka menggunakan istilah yang Anda mengerti.
UCT2	Aplikasi Traveloka menghindari penggunaan istilah teknis (khusus).
UCT3	Aplikasi Traveloka tidak menggunakan istilah teknis (khusus).
UCT4	Aplikasi Traveloka menggunakan istilah yang bisa dimengerti.
DES1	Secara keseluruhan, saya merasa aplikasi Traveloka didesain dengan baik.
DES2	Secara umum, saya merasa aplikasi Traveloka memiliki desain yang bagus.
DES3	Secara umum, aplikasi Traveloka didesain dengan baik.
DES4	Saya sangat puas dengan keseluruhan desain dari aplikasi Traveloka.
PURP1	Menurut saya, aplikasi Traveloka mengutamakan fungsi.
PUPR2	Secara keseluruhan, saya merasa aplikasi Traveloka bermanfaat.
PUPR3	Secara umum, aplikasi Traveloka berfungsi sesuai tujuannya.
PUPR4	Secara umum, aplikasi Traveloka penting bagi saya.
INTG1	Secara keseluruhan, saya merasa grafis yang ditampilkan dalam aplikasi Traveloka didesain secara efektif.
INTG2	Secara umum, tampilan grafis dalam aplikasi Traveloka didesain dengan baik.
INTG3	Secara umum, saya menyukai grafis yang ditampilkan dalam aplikasi Traveloka.
INTG4	Secara keseluruhan, aplikasi Traveloka memiliki tampilan grafis yang sangat bagus.
INP1	Secara umum, aplikasi Traveloka memungkinkan saya memasukkan data dengan mudah.
INP2	Secara keseluruhan, aplikasi Traveloka memiliki langkah-langkah pemasukan data yang didesain secara efektif.
INP3	Saya sangat puas dengan langkah-langkah pemasukan data dalam aplikasi Traveloka.
INP4	Secara umum, mudah bagi saya untuk memasukkan data ke dalam aplikasi Traveloka.
CONT1	Secara umum, informasi dalam aplikasi Traveloka disajikan secara efektif.
CONT2	Secara keseluruhan, saya merasa aplikasi Traveloka menyajikan informasi dengan baik.
CONT3	Secara keseluruhan, saya merasa aplikasi Traveloka menyajikan informasi secara efektif.
CONT4	Saya sangat puas dengan cara aplikasi Traveloka menyajikan informasi.
STRU1	Secara keseluruhan, saya merasa aplikasi Traveloka menyusun informasi secara efektif.
STRU2	Secara umum, aplikasi Traveloka disusun dengan sangat baik.
STRU3	Saya sangat puas dengan cara penyusunan aplikasi Traveloka.
STRU4	Secara umum, aplikasi Traveloka disusun dengan baik.
CITU1	Saya berniat terus menggunakan aplikasi Traveloka.
CITU2	Daripada menghentikan penggunaan, saya ingin terus menggunakan aplikasi Traveloka.
CITU3	Saya memperkirakan akan terus menggunakan aplikasi Traveloka.
CITU4	Saya berencana terus menggunakan aplikasi Traveloka.
CITU5	Saya tidak berniat menggunakan aplikasi Traveloka di masa yang akan datang.
CITU6	Saya cenderung akan terus menggunakan aplikasi Traveloka di masa yang akan datang.
MAL1	Saya mengajak teman dan kerabat saya menggunakan aplikasi Traveloka.
MAL2	Saya mengatakan hal-hal baik tentang aplikasi Traveloka kepada orang lain.
MAL3	Saya akan menggunakan lebih banyak layanan yang ditawarkan oleh aplikasi Traveloka di masa yang akan datang.
MAL4	Saya akan menyarankan aplikasi Traveloka kepada orang yang meminta pendapat saya.
MAL5	Saya menganggap aplikasi Traveloka sebagai pilihan pertama saya.

Table B1

First-order construct's indicator reliabilities

Construct and items	Loadings	Construct and items	Loadings
Aesthetic graphics		Branding
AEST1	0.847∗∗∗	BRAN1	0.729∗∗∗
AEST2	0.887∗∗∗	BRAN3	0.830∗∗∗
AEST3	0.885∗∗∗	BRAN4	0.906∗∗∗
AEST4	0.855∗∗∗
Concise language		Control obviousness
CLAN1	0.850∗∗∗	COOB1	0.885∗∗∗
CLAN2	0.919∗∗∗	COOB2	0.875∗∗∗
CLAN3	0.893∗∗∗	COOB3	0.934∗∗∗
CLAN4	0.940∗∗∗	COOB4	0.903∗∗∗
Content relevance		Data preservation
CRLV1	0.879∗∗∗	DARP1	0.719∗∗∗
CRLV2	0.921∗∗∗	DAPR2	0.723∗∗∗
CRLV3	0.831∗∗∗	DAPR3	0.815∗∗∗
CRLV4	0.830∗∗∗	DAPR4	0.774∗∗∗
De-emphasize user settings		Effort minimization
DUS1	0.774∗∗∗	EMM1	0.856∗∗∗
DUS2	0.773∗∗∗	EMM2	0.893∗∗∗
DUS3	0.852∗∗∗	EMM3	0.778∗∗∗
DUS4	0.854∗∗∗
Fingertip-size control		Logical path
FTSC1	0.937∗∗∗	LP1	0.929∗∗∗
FTSC2	0.951∗∗∗	LP2	0.860∗∗∗
		LP3	0.898∗∗∗
		LP4	0.797∗∗∗
Realism		Subtle animation
REAL1	0.902∗∗∗	SANM1	0.883∗∗∗
REAL2	0.916∗∗∗	SANM2	0.899∗∗∗
REAL3	0.912∗∗∗	SANM3	0.812∗∗∗
REAL4	0.755∗∗∗	SANM4	0.839∗∗∗
Search		Instant start
SEAR1	0.785∗∗∗	STAR1	0.781∗∗∗
SEAR2	0.852∗∗∗	STAR2	0.664∗∗∗
SEAR3	0.831∗∗∗	STAR3	0.864∗∗∗
SEAR4	0.874∗∗∗	STAR4	0.811∗∗∗
Standardized user-interface element		User-centric terminology
SUI1	0.850∗∗∗	UCT1	0.849∗∗∗
SUI2	0.894∗∗∗	UCT2	0.879∗∗∗
SUI3	0.904∗∗∗	UCT3	0.849∗∗∗
SUI4	0.923∗∗∗	UCT4	0.865∗∗∗
Information structure		Information structure (cont'd)
TTPS1	0.792∗∗∗	TTPS4	0.864∗∗∗
TTPS2	0.882∗∗∗	TTPS5	0.619∗∗∗
TTPS3	0.893∗∗∗	TTPS6	0.827∗∗∗

∗p < 0.10; ∗∗p < 0.05; ∗∗∗p < 0.01. Insignificant indicators are dropped (BRAN2, EMM4, FTSC3, and FTSC4).

Table C2

Second and third-order construct's indicator reliabilities

Construct and items	Loadings	Construct and items	Loadings
Application design		Application utility
DES1	0.912∗∗∗	PURP1	0.885∗∗∗
DES2	0.935∗∗∗	PURP2	0.896∗∗∗
DES3	0.914∗∗∗	PURP3	0.882∗∗∗
DES4	0.904∗∗∗	PURP4	0.738∗∗∗
User interface graphics		User interface input
INTG1	0.895∗∗∗	INP1	0.909∗∗∗
INTG2	0.903∗∗∗	INP2	0.912∗∗∗
INTG3	0.923∗∗∗	INP3	0.930∗∗∗
INTG4	0.899∗∗∗	INP4	0.914∗∗∗
User interface output		User interface structure
CONT1	0.935∗∗∗	STRU1	0.879∗∗∗
CONT2	0.920∗∗∗	STRU2	0.922∗∗∗
CONT3	0.917∗∗∗	STRU3	0.914∗∗∗
CONT4	0.916∗∗∗	STRU4	0.863∗∗∗
Continued intention to use		Mobile application loyalty
CITU1	0.933∗∗∗	MAL1	0.868∗∗∗
CITU2	0.915∗∗∗	MAL2	0.853∗∗∗
CITU3	0.945∗∗∗	MAL3	0.854∗∗∗
CITU4	0.959∗∗∗	MAL4	0.916∗∗∗
CITU5	0.776∗∗∗	MAL5	0.850∗∗∗
CITU6	0.895∗∗∗

∗p < 0.10; ∗∗p < 0.05; ∗∗∗p < 0.01.

Table C1

First-order construct's AVEs and composite reliabilities

Constructs	Composite reliability	AVE
Aesthetic graphics	0.925	0.755
Branding	0.864	0.680
Concise language	0.945	0.812
Control obviousness	0.944	0.809
Content relevance	0.923	0.750
Data preservation	0.844	0.576
De-emphasize user settings	0.887	0.663
Effort minimization	0.881	0.712
Fingertip-size control	0.942	0.891
Logical path	0.927	0.761
Realism	0.928	0.764
Subtle animation	0.918	0.737
Search	0.903	0.699
Instant start	0.863	0.614
Standardized user-interface element	0.940	0.798
Information structure	0.923	0.670
User-centric terminology	0.920	0.741

Table D2

Second and third-order construct's construct reliabilities

Constructs	Composite reliability	AVE
Application design	0.954	0.840
Application utility	0.914	0.727
User interface graphics	0.948	0.819
User interface input	0.954	0.840
User interface output	0.958	0.850
User interface structure	0.941	0.801
Continued intention to use	0.965	0.820
Mobile application loyalty	0.939	0.754

Table D3

The Heterotrait-Monotrait Ratio (HTMT) of correlations.

Table D4

Construct's correlations.

Table D1

First-order construct's VIFs

Construct and items	VIFs	Construct and items	VIFs
Aesthetic graphics		Branding
AEST1	2.056	BRAN1	1.375
AEST2	2.896	BRAN3	1.743
AEST3	3.109	BRAN4	2.034
AEST4	2.434
Concise language		Control obviousness
CLAN1	2.282	COOB1	2.811
CLAN2	3.636	COOB2	2.686
CLAN3	3.537	COOB3	4.997
CLAN4	4.871	COOB4	3.878
Content relevance		Data preservation
CRLV1	3.379	DARP1	1.736
CRLV2	4.283	DAPR2	1.365
CRLV3	2.058	DAPR3	2.006
CRLV4	1.928	DAPR4	1.323
De-emphasize user settings		Effort minimization
DUS1	1.989	EMM1	1.929
DUS2	1.929	EMM2	2.224
DUS3	2.204	EMM3	1.466
DUS4	2.280
Fingertip-size control		Logical path
FTSC1	2.581	LP1	4.183
FTSC2	2.581	LP2	2.357
		LP3	3.294
		LP4	1.835
Realism		Subtle animation
REAL1	3.068	SANM1	3.093
REAL2	3.708	SANM2	3.227
REAL3	3.245	SANM3	1.926
REAL4	1.718	SANM4	1.958
Search		Instant start
SEAR1	1.837	STAR1	1.491
SEAR2	2.280	STAR2	1.444
SEAR3	2.100	STAR3	2.242
SEAR4	2.116	STAR4	1.807
Standardized user-interface element		User-centric terminology
SUI1	3.780	UCT1	2.131
SUI2	4.976	UCT2	3.491
SUI3	4.861	UCT3	3.243
SUI4	2.209	UCT4	2.264
Information structure		Information structure (cont'd)
TTPS1	2.039	TTPS4	3.222
TTPS2	4.194	TTPS5	1.610
TTPS3	4.541	TTPS6	2.635

Table E1

Frist-order construct's structural path estimates

Path	Coefficient
Branding → Application design	0.377∗∗∗
Data preservation → Application design	0.128∗∗∗
Instant start → Application design	0.340∗∗∗
Content relevance → Application utility	0.261∗∗∗
Search → Application utility	0.456 ∗∗∗
Aesthetic graphics → User interface graphics	0.199∗∗∗
Realism → User interface graphics	0.316∗∗∗
Subtle animation → User interface graphics	0.315 ∗∗∗
De-emphasize user settings → User interface input	-0.059
Effort minimization → User interface input	0.322 ∗∗∗
Control obviousness → User interface input	0.352∗∗∗
Fingertip-size control → User interface input	0.250∗∗∗
Concise language → User interface output	0.405 ∗∗∗
Standardized user-interface element → User interface output	-0.001
User-centric terminology → User interface output	0.357 ∗∗∗
Logical path → User interface structure	0.536 ∗∗∗
Information structure → User interface structure	0.222∗∗∗

∗p < 0.10; ∗∗p < 0.05; ∗∗∗p < 0.01.