Importance Ranking of Usability Indicators for Second-Hand Trading Applications Based on Exploratory Factor Analysis—Analytic Hierarchy Process toward Sustainable Development

Abstract

In contemporary society, excessive energy consumption and rapid product development cycles are seen as significant barriers to sustainable development. Second-hand trading apps have emerged as a solution to this issue that allow the extension of the lifecycle of products to reduce resource consumption and waste generation. This study aims to support the design and development of second-hand trading apps by establishing and prioritizing usability indicators. To this end, we assembled a team of 20 experts and initially identified usability indicators using the Delphi method. Then, data from surveys of 412 users were analyzed using Exploratory Factor Analysis (EFA) to reduce dimensions, and the Analytic Hierarchy Process (AHP) was used to finally rank the relative importance and priority of these indicators. The results show that of the Level 1 indicators, safety and privacy, learning and operation, and information architecture are relatively highly prioritized. For the Level 2 indicators, property protection, page information architecture, privacy protection, ease of use, and efficiency ranked highly in terms of composite weight priority. The findings of this study will aid in the usability evaluation and enhancement of future second-hand trading apps. By optimizing app usability, increasing user engagement, and thereby boosting the rate of second-hand product purchases, this research contributes positively to the sustainable development of the environment by reducing waste generation.

1. Introduction

In contemporary society, the pursuit of sustainable consumption and the imperative of environmental protection stand as pivotal concerns. The excessive consumption of energy coupled with the rapid, iterative development of products poses formidable challenges to the goal of achieving sustainable development [1,2]. In particular, consumer culture is driven by the quest for immediate gratification, which engenders resource wastage and environmental deterioration in addition to enduring adverse consequences [3,4]. For example, rapidly changing fashions and the fleeting pursuit of particular design styles have led to the trend of disposable product use. This trend not only results in the continuous growth of solid waste but also has numerous negative impacts on the environment and climate change [5]. According to an analysis by Business Insider, fashion production accounts for 10% of global carbon emissions, and 85% of textiles are thrown away each year [6]. Over the past 15 years, the number of times a piece of clothing is worn has decreased by more than one-third, to an average of seven times, leading to the generation of 92 million tons of textile waste globally each year [7]. In this social context, second-hand trading apps are playing a crucial role in providing consumers with new consumption experiences while simultaneously fostering a culture of sustainable consumption [8]. From the perspective of sustainability, second-hand goods trading not only breathes new life into products by allowing unneeded or idle items to be passed on to those who can use them, effectively extending product lifecycles; it also significantly reduces resource consumption and waste output [9]. According to a report by Schibsted, the buying and selling of second-hand items through their platform resulted in approximately 290,992 tons of steel saved in 2021, equivalent to the amount used in 11.5 million bicycles, and a further 8.2 billion plastic bags were saved simply through use of the second-hand trading app [10]. With the ubiquity of smartphones, the methods of second-hand trading have become more diversified, and second-hand trading apps have emerged as important platforms for users to conveniently buy and sell used items [11]. However, according to one report, more than 56% of users uninstall apps within the first seven days of installation [12]. Therefore, optimizing the app’s usability becomes especially crucial to attract and retain more users for second-hand trading apps [13]. However, assessing and improving the sustainable usability of second-hand trading apps remains a challenge, and the development of usability evaluation metrics wherein both user convenience and environmental impact are considered is an important research topic [14,15].

In order to achieve the above objectives, this study addresses the following research questions:

RQ1: Which indicators are closely related to the usability evaluation of a sustainable second-hand trading app?

RQ2: What is the importance ranking of indicators at different levels in the usability evaluation of a sustainable second-hand trading app?

RQ3: What is the overall importance ranking of the indicators in the usability evaluation of a sustainable second-hand trading app?

In summary, the purpose of this study is to rank the importance of usability indicators for second-hand trading apps using the Delphi method, Exploratory Factor Analysis (EFA), and Analytic Hierarchy Process (AHP) to identify the key indicators that enhance usability. The findings will not only aid in developing applications that better meet user needs, improve user experience and satisfaction, and reduce the uninstall rate but also have profound implications for encouraging users to adopt eco-friendly living habits, mitigating environmental pollution, and reducing resource wastage.

2. Literature Review

2.1. Application Usability

The concept of usability was first introduced by Nielsen in the 1970s. The definition of usability varies among researchers and fields of study [16]. Application usability refers to the ease with which users can interact with and navigate an application to complete tasks effectively and efficiently [17]. Therefore, highly usable applications are intuitive and user-friendly, minimizing user frustration while maximizing productivity.

The research by Baharuddin, R., Singh, D., and others proposes a set of usability dimensions for designing and evaluating mobile applications [18]. Hoehle, H. and Venkatesh, V. conducted a study to propose and validate a conceptual framework and survey instrument for mobile application usability to better understand the impact of mobile application usability and to provide guidance for designing effective mobile applications [14]. Sibarani, A. J. evaluated the satisfaction and usage levels of e-learning applications and their functionalities using the five usability metrics established by Nielsen [19]. Mazaheri Habibi, M. R., Moghbeli, F., and others conducted a study on the usability of applications used by pregnant women, employing Nielsen’s five usability principles [20].

These studies have made significant contributions to the evaluation of mobile application usability. However, most research has focused on broad categories of mobile applications or has conducted usability tests based on existing general mobile application metrics without considering the specific characteristics of particular mobile applications. Notably, almost all usability tests have used Nielsen’s usability principles, with little consideration of the unique attributes of specific mobile applications. Therefore, this study builds upon the metrics proposed by Nielsen and others, incorporating the specific features of particular software for a more detailed and focused investigation.

2.2. Second-Hand Trading Applications

Second-hand goods trading platforms provide a marketplace for exchange and communication between the owners of idle items and buyers. This not only extends the lifespan of products but also reduces the need for manufacturing new products [21]. Given their impact on sustainability and economic benefits for consumers, there is currently extensive research on second-hand trading apps. W. Qi and P. Yang’s study focuses on the second-hand trading apps Xianyu and Hongbulin, analyzing the differences in interaction modes and their impact on user experience to enhance the overall user experience of second-hand trading apps [22]. Jiu, Y. and Hong, Q. developed an electronic trading platform for second-hand goods based on the Android platform. Utilizing the MVP architecture pattern during platform construction, they achieved a highly maintainable, API-oriented second-hand trading platform software [23]. Jin W. explored the relationship model between changes in user needs and dimensions of user flow experience in the design of internet shopping platforms, subsequently designing an internet shopping platform based on user flow experience [24]. Barbara Borusiak et al. employed the extended Theory of Planned Behavior (TPB) to study consumers’ willingness to purchase second-hand products and visit online and offline second-hand shops. Guiot, D. and Roux, D. proposed a second-hand shoppers’ motivation scale to evaluate the motivations of second-hand shoppers [25].

In summary, research on second-hand trading platforms primarily focuses on user experience, platform design and implementation, and consumer purchase intentions. Although these studies provide effective solutions for optimizing the use of second-hand trading apps, research on the importance of usability metrics specific to the characteristics of second-hand trading apps remains insufficient. Therefore, this paper aims to develop usability metrics from a sustainability perspective, considering the unique characteristics of second-hand trading apps.

2.3. Sustainable Consumption and Second-Hand Purchases

Sustainable consumption refers to using products and services in ways that minimize environmental impact [26]. Paavola’s research indicates that people can reduce environmental harm by altering their consumption patterns [27]. According to Wikipedia, sustainable consumption relies on the efficient use of resources, reduction of waste and pollution, utilization within the regenerative capacity of renewable resources, and the reuse and upcycling of products throughout their lifecycle to maximize their utility; it also emphasizes intergenerational and intragenerational equity [28]. The concept of sustainable consumption has also influenced consumer needs and values [29], protesting against wasteful consumption and fast fashion, which has led to the emergence of an alternative model of circular fashion systems [30]. As environmental awareness and the importance of sustainability continue to increase, a new trend of second-hand purchasing is emerging [11]. The number of second-hand retailers is continuously increasing [31], growing at a rate 11 times faster than the overall retail sector [11]. By 2029, the second-hand market is expected to account for over 17% of traditional retail [11]. The rise of e-commerce has also led to a surge in online second-hand platforms [32]. To provide a good online environment, studies by Ganguly et al. and Haryanti and Subriadi indicate that a unified and aesthetically pleasing user interface [11], combined with a comfortable and satisfactory user experience design, plays a crucial role in creating an effective and emotionally appealing online platform [28]. Therefore, this paper will also study aspects of user experience and effectiveness to propose usability evaluation metrics for second-hand trading platforms, providing an evaluation framework for improving the usability of second-hand trading applications.

3. Methodology

3.1. Research Framework

In this study, the Delphi technique, Exploratory Factor Analysis (EFA), and Analytic Hierarchy Process (AHP) are used to identify usability evaluation items for second-hand trading apps and analyze their importance. This research primarily revolves around deriving evaluation items and their weights, for which the methodology of this study is divided into three 3 parts. Part I: Through extensive literature searching, a series of preliminary usability metrics were collected, and the metrics were then confirmed through the distribution of Delphi questionnaires to domain experts. Part II: EFA was used to reduce these metrics into five dimensions. Part III: Weights were calculated using AHP, and the most critical usability metrics for second-hand trading apps were identified by assessing and ranking their priorities. An intuitive overview of the research steps is provided to the reader, with the methodological flowchart illustrated in Figure 1.

3.2. Research Methods and Survey Techniques

3.2.1. Delphi Method

The Delphi method, developed by the RAND Corporation (Santa Monica, CA, USA) in 1953, is a forecasting technique based on expert opinions [33]. This method utilizes iterative rounds of anonymous surveys to deepen the understanding of uncertain issues through achieving a consensus among experts. The Delphi method particularly emphasizes a feedback mechanism, allowing experts to adjust their views after considering the opinions of their peers, thereby enhancing the accuracy of judgments and facilitating the formation of consensus. It is widely applied in exploratory, normative, and predictive analyses [34]. Typically, this process involves the careful selection of anonymous experts to form a chosen group who undergo two to three rounds of structured surveys. After each round, the questionnaire is adjusted based on expert feedback and distributed for the next round, with this process repeated until a consensus is reached among the experts [35,36,37].

Scientific Validity of the Questionnaire

Considering the scientific validity and quality control of the questionnaire, the content of this study’s questionnaire references the one developed by Ye and Jiang [38]. The questionnaire includes (1) basic information about the experts, as shown in Table 1; (2) descriptions of the indicators: each indicator’s corresponding meaning and the supplementary explanations revised by the experts; (3) a 5-point Likert scale is used for rating the indicators. To ensure the rigorous use of the evaluation tool, other Delphi questionnaire rating tools were also investigated. According to the study by Remus, A., Smith, V., et al., using different rating scales in Delphi surveys does appear to affect the results; however, the 5-point scale was found to be most consistent with the final outcomes [39]. Furthermore, a literature review on the Delphi method revealed that previous researchers have commonly employed the 5-point Likert scale [40,41,42]. Therefore, this study also adopts the 5-point Likert scale for the questionnaire; (4) the survey also assesses the experts’ familiarity with the guidelines and judgment criteria.

Scientific Validity of the Experts

To ensure the authority, representativeness, and reliability of the research results, this study follows the criteria outlined by Skulmoski, G. J. et al. [43], which suggest that experts should meet four requirements: (1) acquire knowledge and experience through investigation, (2) participate voluntarily, (3) have sufficient time to participate, and (4) possess effective communication skills [43]. Therefore, this study adopted the following criteria for including experts: 15 experts with over 8 years of experience in design-related work or research, 2 experts with more than 5 years of experience in environmental science, and 3 software engineers with over 5 years of work experience, for a total of 20 experts. All the experts participated voluntarily, and they had sufficient time to engage in the study.

Questionnaire Distribution

In this study, a questionnaire was developed based on the collected usability metrics, and a Delphi survey was conducted from 5 March 2024 to 12 March 2024. The questionnaire was distributed via email. In the first round of the Delphi survey, statistical feedback information was collected from the questionnaires, and we removed some indicators based on the experts’ suggestions. The first round of revisions was incorporated into the second round of the consultation questionnaire, where experts then rated the revised indicators. Experts were required to rate each item using a 5-point Likert scale, with 1 being “very unimportant”, 3 being “neutral”, and 5 being “very important”. Throughout the survey process, experts were also encouraged to propose new measurement indicators to ensure a comprehensive evaluation.

Table 1. Demographic information of the expert Delphi panel.

N = 20	Item	Number of Participants	Percentage (%)
Age Range (years)	30–34	7	35
	35–39	9	45
	40–44	3	15
	Over 45	1	5
Gender	M	8	40
	F	16	60
Occupation	Designer	10	50
	Design Professor	5	25
	Environmental Science Professor	2	10
	Software Engineer	3	15
Education Level	Bachelor’s Degree	7	35
	Master’s Degree	8	40
	Ph.D.	5	25
Work Experience (years)	5–9	5	25
	10–14	14	70
	More than 15	1	5

Table 1. Demographic information of the expert Delphi panel.

N = 20	Item	Number of Participants	Percentage (%)
Age Range (years)	30–34	7	35
	35–39	9	45
	40–44	3	15
	Over 45	1	5
Gender	M	8	40
	F	16	60
Occupation	Designer	10	50
	Design Professor	5	25
	Environmental Science Professor	2	10
	Software Engineer	3	15
Education Level	Bachelor’s Degree	7	35
	Master’s Degree	8	40
	Ph.D.	5	25
Work Experience (years)	5–9	5	25
	10–14	14	70
	More than 15	1	5

3.2.2. Exploratory Factor Analysis

Exploratory Factor Analysis (EFA) was initially proposed by Charles Spearman in 1904 as a multivariate statistical technique aimed at reducing data into a smaller set of summary variables and exploring the underlying theoretical structure of phenomena [44,45].

Questionnaire Design

Based on the usability indicators obtained through the Delphi survey method, an online questionnaire was created on the Questionnaire Star platform (Wenjuanxing: http://www.wjx.cn 16 March 2024) to collect the data required for Exploratory Factor Analysis (EFA). The Questionnaire Star platform has been widely used in academic research, as demonstrated by studies conducted by Zhang Yan, Chen Yongping, Wu Wei, Zhang Yan, Ma Ling, Liu Hong, and others, all of whom utilized the Questionnaire Star platform for their online surveys [40,41,42]. The online questionnaire provides a convenient option for reaching a large audience. Reverse questions were included in the survey to reduce response bias and enhance the reliability and validity of the questionnaire [46].

Responses to the reverse questionnaire that were incorrect were considered invalid. The questionnaire was conducted from 14 March 2024 to 17 March 2024. Based on a review of the literature on survey sample sizes [47,48,49], a total of 412 users aged 18 and above who had actually used second-hand trading software were surveyed. The participant information is detailed in

Table 2. Demographic information of second-hand trading application users.

N = 400	Item	Number of Participants	Percentage (%)
Gender	M	170	42.5
	F	230	57.5
Age	18–28	203	50.75
	29–39	125	31.25
	Over 40	72	18
Years of Using Second-Hand Trading Applications	Less than 1 year	102	25.5
	1–3 years	115	28.75
	More than 3 years	183	45.75

. A total of 412 questionnaires were distributed, of which 400 were valid and 12 were invalid. Among the invalid questionnaires, 8 had identical answers throughout, and 4 had incorrect responses to the reverse questions.

Research Process

Before performing an Exploratory Factor Analysis (EFA) of the data collected from online questionnaires, it is crucial to evaluate the validity and reliability of the data, determining their fitness for EFA. When the Kaiser–Meyer–Olkin (KMO) value is higher than 0.7, and the significance level in the Bartlett’s Test of Sphericity is less than 0.05, it indicates strong data validity. Additionally, a Cronbach’s alpha coefficient in the range of 0.7 to 0.95 indicates strong scale reliability [10].

After confirming the reliability and validity of the questionnaire, the data were processed using Exploratory Factor Analysis (EFA). Common factors with eigenvalues (λ) greater than 1 were retained [50]. These common factors and their included usability indicators were then grouped and named accordingly.

3.2.3. Analytic Hierarchy Process

The Analytic Hierarchy Process (AHP), developed by Thomas L. Saaty in the 1970s, is a method for accurately quantifying the weights of decision criteria [51,52]. With this method, the decision-making process becomes more scientific and objective, aiding in tackling complex decision issues to ensure the most suitable options are selected.

Questionnaire Design

The AHP questionnaire involves pairwise comparisons of indicators at each level to generate ratio scales for judgment. The AHP employs Saaty’s 1–9 scale allocation method [53] to score the indicators in the matrix. Each judgment is made by comparing the indicators in the left column with those in the top row to determine which of the two components is relatively more important. According to the literature [54], we ultimately decided to invite 20 industry experts with over 10 years of experience to participate in this questionnaire, including 3 product managers, 7 interaction designers, and 10 product designers. The demographic information of the 20 experts participating in the questionnaire is shown in

Table 3. AHP demographic information of the expert panel.

N = 20	Item	Number of Participants	Percentage (%)
Age Range (years)	30–34	1	5
	35–39	8	40
	40–44	10	50
	Over 45	1	5
Gender	M	7	30
	F	13	65
Occupation	Product Managers	3	15
	Interaction Designers	7	30
	Product Designers	10	50
Education Level	Bachelor’s Degree	3	15
	Master’s Degree	5	25
	Ph.D.	12	60
Work Experience (years)	10–14	8	40
	15–20	10	55
	Over 20 years	2	10

.

Research Process

In this study, we first established a hierarchical framework model for the indicators. Considering that YAAHP 10.1 software (https://www.metadecsn.com/yaahp/ 21 May 2024) has been used in many studies [55,56,57] and has proven its applicability in the AHP method, we input the hierarchical framework model of the indicators into YAAHP 10.1 software for verification. After successful verification, we created a pairwise comparison matrix questionnaire. Finally, the numerical comparisons in the pairwise matrices were calculated to obtain the weight values at different levels.

4. Research Execution and Analysis

4.1. Extraction of Usability Indicators

In Usability Engineering, Nielsen defines usability as the measurement of whether a system or product meets consumer requirements, categorizing it into five aspects: Learnability, Efficiency, Memorability, Errors, and Satisfaction [58]. The ISO 9241 standard [59] defines usability as the extent to which specified users can achieve specified goals with Effectiveness, Efficiency, and Satisfaction in a particular context of use [60]. Brian Shackel’s definition emphasizes the necessity of system evaluation throughout the development lifecycle, highlighting four criteria of a usability system: effectiveness, learnability, flexibility, and user attitude [61]. Dumas and Redish (1994) described usability in terms of enhancing the speed and ease for users interacting with a product to achieve their objectives. Usability can be encapsulated as emphasizing user centricity, effectiveness, efficiency, and ease of use alongside the aspects of effectiveness, efficiency, and simplicity in use [62].

The above is based on the literature and compilation of usability indicators. However, considering the specific research subjects, further literature research on the usability of applications is needed. So, we considered the variability in usability across different researchers or fields not only by referencing the indicators of usability evaluation provided by scholars but also through utilizing widely recognized platforms such as Scopus, Web of Science, and Google Scholar [63,64]. Keywords including “app usability research”, “interface design”, and “online second-hand trading app” were used to search the literature. The aim was to comprehensively select usability indicators for second-hand trading apps from multiple perspectives. Liu, X. (2023) classifies the usability of app software interfaces into several categories: timely feedback on software operations, terminology understandable by users, clarity of the interface, consistency with public cognition, realization of value through application use, error feedback that notifies users of operational mistakes, assistance that offers solutions, privacy protection safeguarding user privacy, and protection of property [65]. In their study on the factors influencing satisfaction with app interface design, Han, X., Tao, X., and Zhou, Z. (2021) detailed that interface clarity, simplicity, and the layout of graphics and text significantly impact the usability of app interfaces [66]. Richardson, B., Campbell-Yeo, M., and Smit, M. (2021) described the usability of mobile applications through seven indicators in their article: “usable”, “usefulness”, “desirable”, “findability”, “accessibility”, “credibility”, and “valuable” [67]. Padmavathy, C., Swapna, M., and Paul, J. (2019) demonstrated in their study that when using second-hand trading software, the software convenience, the pleasure derived from purchasing products, the social aspect of communicating with sellers, and the security of the payment system are aspects that capture consumer attention [68]. Qi, W. and Yang, P. (2021) argue that in mobile apps for trading second-hand luxury goods, a user-centric approach should be emphasized, focusing on “interactive design” between the platform and users, “page information architecture” design, and “navigation” design [22].

Based on the indicators from the literature review, a total of 29 usability indicators were compiled after removing duplicates and similar indicators, as shown in

Table 4. Summary of usability metrics for second-hand trading applications.

Indicator	Content	Indicator	Content
Learnability	The ability of users to quickly learn how to use the app	User-Centricity	Design and development processes centered around user needs and experiences
Property Protection	Protecting the safety of users’ property	Convenience	Ease and speed of use during the product or service experience
Efficiency	The time required to complete tasks	Simplicity in Use	Simple and easy-to-operate design
Graphic Layout	The organization and layout of visual elements and text	Pleasure	Satisfaction felt during product use
Memorability	The ability of users to remember how to operate the app after periods of non-use	Immediate Feedback	The system’s ability to provide instant responses after user actions
Interface Simplicity	A clear interface design without distracting elements	Social	Encourages interaction between users and between users and sellers
Error Prevention	Design features that reduce the likelihood of user errors	Consistency	The degree to which interface elements and operational logic are unified
Findability	The ease with which users can find the information they need	System Security	Protects the system from threats such as viruses
Satisfaction	The overall satisfaction level of users with the product or service	Clarity	The degree to which information is expressed clearly and is easy to understand
Accessibility	Ensuring that everyone, including individuals with disabilities, can use the product without barriers	Page Information Architecture	The organization and hierarchical structure of page information
Flexibility	The capacity for user customization and personalization	Terminology	Uses vocabulary that users can understand
Credibility	Users’ perception of the product’s or service’s reliability and trustworthiness	Navigation	Helps users quickly find the information they need during product use
Attitude	The impression or stance users hold toward the product, service, or brand	Help	Provides guidance and solutions for user operations and problems
Value Realization	The value obtained from using the product	Error Feedback	Offers guidance and feedback to users when errors are made
Findability	The ease with which users can find the information they need

.

4.2. Delphi Consultation

4.2.1. Expert Authority Level

The expert authority level (Cr) is primarily influenced by two factors: the basis of the expert’s judgment when evaluating the questionnaire (Ca), and the expert’s familiarity with the questionnaire (Cs). The formula for calculating the expert authority level is as follows:

Cr = (Ca + Cs)/2

(1)

The value of Cr ranges from 0 to 1. A Cr value greater than 0.7 indicates that the reliability of the value is acceptable; the higher the Cr value, the greater the expert’s authority level [38].

In this study, the calculation of the expert authority level is mainly derived from the self-assessment data provided by the experts in the questionnaire. The evaluation criteria for the experts’ self-assessment in the questionnaire are as follows:

Familiarity is divided into five levels: very familiar, fairly familiar, generally familiar, somewhat unfamiliar, and very unfamiliar. The corresponding values are 1.0, 0.8, 0.6, 0.4, and 0.2, respectively [38].

The basis of judgment is categorized into four options: theoretical analysis, practical experience, reference to relevant domestic and international materials, and intuition. These are valued sequentially at 0.8, 0.6, 0.4, and 0.2, respectively [38], as shown in

Table 5. Quantitative table of expert familiarity and judgment basis.

Familiarity Level (Cs)	Value	Judgment Basis (Ca)	Value
Very familiar	1.0	Theoretical analysis	0.8
Fairly familiar	0.8	Practical experience	0.6
Generally familiar	0.6	Reference to relevant domestic and international materials	0.4
Somewhat unfamiliar	0.4
Very unfamiliar	0.2	Intuition	0.2

.

Based on the experts’ self-assessment results regarding their familiarity with the questionnaire and their judgment basis, the average values for the judgment basis and familiarity were calculated for each round of the questionnaire. Finally, using Formula (1), the authority level of 20 experts over the two rounds of the questionnaire was calculated. The specific results are shown in

Table 6. Results of authority levels from expert consultation questionnaires.

Delphi Rounds	Cs	Ca	Cr
1	0.65	0.65	0.65
2	0.68	0.68	0.68

below.

The authority levels for the two rounds of the questionnaire were 0.71 and 0.755, respectively. Since both authority values are greater than 0.7, this indicates that the experts’ authority level is relatively high, thereby confirming the reliability of the results from this expert consultation.

4.2.2. Delphi Expert Questionnaire Results

As part of the Delphi questionnaire method, SPSS 27.0 was utilized to analyze the questionnaire data in assessing the authority and scientific nature of the indicators. Indicators were considered to have high consensus among experts and were retained when the progressive significance (P) value was less than 0.05, the mean value (M) was greater than 3.5, and the coefficient of variation (CV) was lower than 0.3 [69]. After completing the second round of questionnaires, the opinions of the experts were unanimous; therefore, a third round of questionnaire testing was deemed unnecessary. The 100% response rate for both rounds of the survey indicates that all experts attach great importance to this questionnaire. The Delphi questionnaire data are shown in

Table 7. Delphi survey results.

N = 20	First Round Delphi Data			Second Round Delphi Data
Indicator	M	SD	CV	M	SD	CV
Learnability	4.50	0.61	0.13	4.45	0.61	0.14
Efficiency	4.70	0.47	0.10	4.60	0.50	0.11
Memorability	4.40	0.75	0.17	4.30	0.80	0.19
Error Prevention	4.40	0.60	0.14	3.80	0.77	0.20
Satisfaction	4.15	0.67	0.16	3.90	0.72	0.18
Flexibility	4.75	0.44	0.09	4.60	0.50	0.11
Attitude	1.65	0.67	0.41	—	—	—
User-Centricity	2.70	1.46	0.54	—	—	—
Simplicity in Use	4.75	0.44	0.09	4.55	0.51	0.11
Immediate Feedback	4.75	0.44	0.09	4.65	0.49	0.11
Consistency	4.60	0.50	0.11	3.90	0.64	0.16
Clarity	1.60	0.75	0.47	—	—	—
Terminology	1.60	0.82	0.51	—	—	—
Help	3.80	0.70	0.18	3.30	0.66	0.20
Privacy Protection	4.60	0.50	0.11	4.55	0.51	0.11
Property Protection	4.85	0.37	0.08	4.95	0.22	0.05
Graphic Layout	3.85	0.81	0.21	1.85	0.93	0.50
Interface Simplicity	4.45	0.51	0.11	4.30	0.57	0.13
Findability	4.65	0.49	0.11	4.10	0.64	0.16
Accessibility	4.35	0.59	0.13	4.20	0.70	0.17
Credibility	1.90	1.07	0.56	—	—	—
Value Realization	3.95	0.76	0.19	3.95	0.76	0.19
Convenience	4.75	0.44	0.09	4.40	0.68	0.15
Pleasure	4.10	0.91	0.22	3.80	0.89	0.24
Social	3.90	0.85	0.22	4.00	0.73	0.18
System Security	4.85	0.37	0.08	4.85	0.37	0.08
Page Information Architecture	4.80	0.52	0.11	4.65	0.49	0.11
Navigation	4.40	0.50	0.11	3.90	0.79	0.20
Error Feedback	4.20	0.95	0.23	3.95	0.89	0.22

.

Based on the survey data from Table 2, the p values for both rounds of the Delphi questionnaire were less than 0.05. Based on the scores from the questionnaire in the first round, five indicators were removed: Attitude, User-Centricity, Clarity, Terminology, and Credibility. In the second round, Graphic Layout was removed. As a result of the two rounds of expert consultation questionnaires, 23 indicators were finally identified, as shown in

Table 8. Indicators after Delphi questionnaire.

Second-Hand Trading App Usability Indicators
Learnability	Efficiency	Memorability	Error Prevention	Satisfaction	Flexibility
Simplicity in Use	Immediate Feedback	Consistency	Help	Privacy Protection	Property Protection
Interface Simplicity	Findability	Accessibility	Value Realization	Convenience	Pleasure
Social	System Security	Page Information Architecture	Navigation	Error Feedback

below.

4.3. EFA Indicator Classification

To obtain data more conveniently and accurately, we utilized SPSS 27.0 to compute the questionnaire metrics of KMO, Bartlett’s test of sphericity, and Cronbach’s alpha coefficient. The outcomes, as delineated in

Table 9. Results of KMO and Bartlett’s test.

Test		Test Value
KMO		0.923
Bartlett’s	Approximate Chi-Square	5029.248
	Degrees of Freedom (df)	253
	Significance	0.00
Cronbach’s Alpha		0.923

, are a KMO measure of 0.923, a Bartlett’s test significance below 0.05, and a Cronbach’s alpha of 0.923. These findings collectively affirm the high reliability and validity of the questionnaire data, which thereby qualify for subsequent EFA investigation.

As part of this factor analysis, Exploratory Factor Analysis (EFA) was employed to distill the factor structure from the data. During the factor extraction process, only factors with eigenvalues λ greater than 1 were retained, eliminating any factors constituted by a single item. When the loading difference between two items on the same factor was less than 0.05, one of the items was removed, and the analysis was then re-conducted. Additionally, any item with a communality less than 0.4 or a maximum loading less than 0.35 was also excluded. After several rounds of screening and rotation, five common factors with eigenvalues λ greater than 1 were ultimately identified, retaining 23 metrics related to the usability of second-hand trading software.

Utilizing the described analytical techniques, we ascertained the contribution of each factor to the total variance, as meticulously outlined in

Table 10. Total variance explained.

Total Variance Explained
Element	Initial Eigenvalues			Sum of Squared Loadings for Extraction			Rotated Sum of Squared Loadings
	Total	Percentage of Variance Explained	Cumulative Percentage of Variance Explained	Total	Percentage of Variance Explained	Cumulative Percentage of Variance Explained	Total	Percentage of Variance Explained	Cumulative Percentage of Variance Explained
1	8.616	37.463	37.463	8.616	37.463	37.463	4.034	17.538	17.538
2	2.269	9.866	47.329	2.269	9.866	47.329	3.799	16.519	34.056
3	1.905	8.283	55.612	1.905	8.283	55.612	3.366	14.636	48.693
4	1.698	7.381	62.993	1.698	7.381	62.993	2.232	9.704	58.397
5	1.017	4.422	67.415	1.017	4.422	67.415	2.074	9.017	67.415
6	0.623	2.707	70.122
7	0.569	2.474	72.595
8	0.557	2.422	75.018
9	0.542	2.356	77.373
10	0.518	2.253	79.627
11	0.487	2.117	81.744
12	0.477	2.076	83.820
13	0.449	1.950	85.770
14	0.442	1.923	87.692
15	0.425	1.846	89.539
16	0.392	1.703	91.241
17	0.376	1.635	92.876
18	0.359	1.563	94.439
19	0.323	1.405	95.844
20	0.280	1.216	97.060
21	0.263	1.142	98.202
22	0.213	0.927	99.129
23	0.200	0.871	100.000

. Further insights are provided by the scree plot in Figure 2, which elaborates on each factor’s contribution toward the total variance. Moreover, the correlations between various usability metrics of second-hand trading software (such as user satisfaction, efficiency in operation, and ease of recall) and the identified factors (e.g., B1, B2, etc.) are detailed in

Table 11. Rotated component matrix.

	Element
	B1	B2	B3	B4	B5
Satisfaction C1	0.873
Flexibility C2	0.765
Accessibility C3	0.745
Social C4	0.736
Pleasure C5	0.729
Value Realization C6	0.694
Efficiency C7		0.865
Memorability C8		0.748
Convenience C9		0.741
Simplicity in Use C10		0.732
Learnability C11		0.716
Error Prevention C12		0.662
Immediate Feedback C13			0.862
Interface Simplicity C14			0.773
Help C15			0.745
Consistency C16			0.712
Error Feedback C17			0.703
Findability C18				0.871
Page Information Architecture C19				0.775
Navigation C20				0.732
System Security C21					0.735
Property Protection C22					0.745
Privacy Protection C23					0.769
Extraction Method: Principal Component Analysis. Rotation Method: Kaiser Normalization Varimax Method a.

^a. Rotation converged in 6 iterations.

.

Based on the results from the rotated component matrix in Table 11 and the characteristics of indicators for each dimension, we extracted and named the following common factors.

The first group (B1) of factors includes indicators such as Satisfaction, Flexibility, Value Realization, Pleasure, Accessibility, and Social Interaction. These are indicators of fulfilling user needs and enhancing loyalty as primarily achieved through emotional design, thus; the group is named “Emotional Design”.

The second group (B2) of factors encompasses indicators such as Learnability, Efficiency, Memorability, Error Prevention, Usability, and Convenience, reflecting the fundamental operations while learning and using the product. Hence, it is named “Learning and Operation”.

The third group (B3) of factors consists of indicators such as Timely Feedback, Consistency, Interface Simplicity, Help, and Error Feedback, showcasing the user experience during interaction with the product, including interface design, operational feedback, and help systems. Therefore, it is termed “Interaction Experience”.

The fourth group (B4) of factors focuses on indicators such as Searchability, Page Information Architecture, and Navigation, emphasizing how the product structure is organized to facilitate easy access and information retrieval for users, and is hence named “Information Architecture”.

The fifth group (B5) of factors includes Privacy Protection, Property Protection, and System Security, centering on the safety of user’s personal information, property, and system security, and is thus named “Security and Privacy”.

4.4. AHP Weight Calculation

4.4.1. Construction of the Judgment Matrix

The main components and corresponding indicators obtained through EFA dimensionality reduction are incorporated into the AHP decision framework, which is divided into three levels: the target objective layer, the Level 1 indicator layer, and the Level 2 indicator layer, as shown in

Table 12. AHP decision framework.

Target	Level 1	Level 2
Second-hand trading app usability indicators A	B1	C1
		C2
		C3
		C4
		C5
		C6
	B2	C7
		C8
		C9
		C10
		C11
		C12
	B3	C13
		C14
		C15
		C16
		C17
	B4	C18
		C19
		C20
	B5	C21
		C22
		C23

.

This hierarchical model was entered into the YAAHP 10.1 software for validation and weight computation in alignment with the AHP framework. Upon the validation of the hierarchical structure, an evidence evaluation questionnaire was formulated. Utilizing the scoring criteria set forth in

Table 13. Relative importance scale.

Saaty’s 1–9 Scale Assignment Method
Scale	Meaning
1	$\mathrm{Indicators} i$$\mathrm{and} j$ are of equal importance.
3	$\mathrm{Indicator} i$$\mathrm{is} \mathrm{slightly} \mathrm{more} \mathrm{important} \mathrm{than} \mathrm{indicator} j$.
5	$\mathrm{Indicator} i$$\mathrm{is} \mathrm{moderately} \mathrm{more} \mathrm{important} \mathrm{than} \mathrm{indicator} j$.
7	$\mathrm{Indicator} i$$\mathrm{is} \mathrm{strongly} \mathrm{more} \mathrm{important} \mathrm{than} \mathrm{indicator} j$.
9	$\mathrm{Indicator} i$$\mathrm{is} \mathrm{absolutely} \mathrm{more} \mathrm{important} \mathrm{than} \mathrm{indicator} j$.
2, 4, 6, 8	The importance of the indicators falls between the above scales.
Reciprocal	$\mathrm{If} \mathrm{the} \mathrm{comparison} \mathrm{between} \mathrm{factors} i$$\mathrm{and} j$$\mathrm{results} \mathrm{in} \mathrm{the} \mathrm{judgment} \mathrm{matrix} \mathrm{entry} C_{ij}$$, \mathrm{then} \mathrm{the} \mathrm{comparison} \mathrm{of} \mathrm{factor} j$$\mathrm{to} i$$\mathrm{is} \mathrm{given} \mathrm{as} C_{ij}$$=1/C_{\mathrm{j}\mathrm{i}}$.

, the responses from 20 experts were transformed into a judgment matrix, enabling systematic pairwise comparisons among indicators across various tiers. The constructed judgment matrix is as follows:

$$A=(a_{ij})n\times n=\left[\begin{array}{ccc}\begin{array}{cc}{a}_{11}& a_{12}\\ a_{21}& a_{22}\end{array}& \begin{array}{c}\dots \\ \dots \end{array}& \begin{array}{c}{a}_{1n}\\ a_{2n}\end{array}\\ \begin{array}{cc}⋮& ⋮\end{array}& \ddots & ⋮\\ \begin{array}{cc}{a}_{n1}& a_{n2}\end{array}& \cdots & a_{nn}\end{array}\right]$$

$a_{ij}$ represents the result of comparing the importance of indicators $i$ and $j$ within the same group, where $a_{ij}$> 0, $a_{ij}$= 1/$a_{\mathrm{j}\mathrm{i}}$, $a_{\mathrm{j}\mathrm{j}}$= 1, for $i$, $j$ = 1,2,3……$n$, with $n$ being the number of subsets contained in $B$.

4.4.2. Calculation of Weights for Individual Levels

By normalizing the judgment matrix based on calculating its eigenvector, we can incrementally ascertain the relative weights of the matrix. The aggregated weight values at each hierarchical level are gauged relative to the overall goal and are derived sequentially from the higher to the lower levels. In the context of a multi-tiered decision-making model, the weight assigned to an individual level serves as the basis for evaluating the relative significance of the elements within that tier [70].

The steps for calculating the individual levels sorting method are as follows:

(1)

Calculate the product of each row’s indicators in the judgment matrix $M_i$.

$$M_i={\prod }_{j=1}^m{a}_{ij}(j=\mathrm{1,2},\cdots ,m)$$

(2)

(2)

Calculate the nth root of $M_i$.

$$W_i=\sqrt[n]{M_i} (i=\mathrm{1,2},\cdots ,n)$$

(3)

(3)

Normalize $W_i$ to obtain the eigenvector ${\omega }_i$.

$${\omega }_i=\frac{W_i}{{\sum }_{j=1}^m{W}_j}(j=\mathrm{1,2},\cdots ,m)$$

(4)

(4)

The formula for the maximum value of the judgment matrix is as follows:

$${\lambda }_{max}={\sum }_{i=1}^n\frac{{\left[A\omega \right]}_i}{{n\omega }_i}$$

(5)

We determine the maximum eigenvalue ${\lambda }_{max}$ of the judgment matrix $A$ by considering $n$ as the matrix’s dimension, where $A$ denotes the judgment matrix itself and ${\omega }_i$ signifies the weight attributed to the $i$th indicator.

(5)

Consistency test:

The ratio of the Consistency Index (CI) of the judgment matrix to the Random Index (RI) is known as the Random Consistency Ratio (CR). When CR is less than 0.1, the judgment matrix is considered consistent. If CR is greater than 0.1, the judgment matrix needs to be reconstructed [71].

The formula for calculating CR is as follows:

$$CI=\frac{{\lambda }_{max}-n}{n-1}$$

(6)

$$CR=\frac{CI}{RI}$$

(7)

The $RI$ value in the formula is determined according to

Table 14. Random index (RI) value.

N	3	4	5	6	7	8	9	10
RI Value	0.52	0.89	1.12	1.26	1.36	1.41	1.46	1.49

.

First, we organize the data collected from 20 experts who rated according to the Saaty rating scale and then summarize it into

Table 15. Level 1 indicator weight.

Level 1	B1	B2	B3	B4	B5	ωi	λmax	CR
B1	1	1/8	1/5	1/7	1/6	0.0357	5.4433	0.0990
B2	8	1	5	2	1/3	0.2674
B3	5	1/5	1	1/4	1/6	0.0823
B5	7	1/2	4	1	1/2	0.2052
B6	6	3	6	2	1	0.4094

,

Table 16. Level 2 B1 indicator weight.

Level 2/B1	C1	C2	C3	C4	C5	C6	ωi	λmax	CR
C1	1	1/2	1/2	1/3	2	3	0.1366	6.4056	0.0644
C2	2	1	1/2	1/4	2	3	0.1607
C3	2	2	1	1/2	2	2	0.1977
C4	3	4	2	1	3	2	0.3336
C5	1/2	1/2	1/2	1/3	1	1	0.0842
C6	1/3	1/3	1/2	1/2	1	1	0.0872

,

Table 17. Level 2 B2 indicator weight.

Level 2/B2	C7	C6	C9	C10	C11	C12	ωi	λmax	CR
C7	1	3	7	1	5	5	0.3271	6.5349	0.0849
C8	1/3	1	2	1/4	2	2	0.1077
C9	1/7	1/2	1	1/4	1/5	2	0.0586
C10	1	4	4	1	6	7	0.3464
C11	1/5	1/2	5	1/6	1	4	0.1163
C12	1/5	1/2	2	1/7	1/4	1	0.0439

,

Table 18. Level 2 B3 indicator weight.

Level 2/B3	C13	C14	C15	C16	C17	ωi	λmax	CR
C13	1	4	8	1/3	1	0.2314	5.1747	0.0390
C14	1/4	1	2	1/7	1/3	0.0669
C15	1/8	1/2	1	1/6	1/5	0.0443
C16	3	7	6	1	3	0.4661
C17	1	3	5	1/3	1	0.1913

,

Table 19. Level 2 B4 indicator weight.

Level 2/B4	C18	C19	C20	ωi	λmax	CR
C18	1	1/5	1/3	0.1096	3.0037	0.0036
C19	5	1	2	0.5813
C20	3	1/2	1	0.3092

and

Table 20. Level 2 B5 indicator weight.

Level 2/B5	C21	C22	C23	ωi	λmax	CR
C21	1	1/2	1	0.2106	3.0183	0.0176
C22	3	1	2	0.5485
C23	1	1/3	1	0.2409

. The evaluation method can be exemplified using the first-level indicators B1 and B2 in Table 15. The ratio of the B1 indicator on the vertical axis to the B1 indicator on the horizontal axis is 1, because they are the same indicator. The ratio of the B1 indicator on the vertical axis to the B2 indicator on the horizontal axis is 1/8, indicating that when comparing B1 and B2, the B2 indicator is significantly more important. Similarly, the ratio of the B2 indicator on the vertical axis to the B1 indicator on the horizontal axis is 8, also indicating that the B2 indicator is significantly more important. Then, based on the numerical values of the expert ratings, we apply Formulas (2) to (7) to determine the weights at each level, and these weight values are marked in Table 15, Table 16, Table 17, Table 18, Table 19 and Table 20.

4.4.3. Calculation of Weights for Individual Levels

In the hierarchical analysis process, merely assessing the relative value of weights within a single tier falls short, given that each criterion operates within a more expansive framework. Consequently, the computation of overall weights becomes imperative. Aligning with the methodology adopted by the majority of researchers, we proceed by multiplying the weights of secondary indicators by the weights of their respective primary indicators [72,73,74].

The overall weight ${OW}_i$ for each individual indicator can be calculated using the following formula:

$${OW}_i=W_{primary }\times W_{secondary}$$

where $W_{primary }$ denotes the weight of the individual indicator within the parent indicator layer, and $W_{secondary}$ represents the weight of the indicator within its respective parent indicator.

Based on this formula, the total weights of all Level 2 indicators were calculated and combined with all single-layer weight values and overall weight values, as shown in

Table 21. Second-hand trading app usability indicator weight.

Level 1	Weights	Rank	Level 2	Weights	Rank	Overall Weights	Rank
B1	0.0357	5	C1	0.1366	4	0.0049	20
			C2	0.1607	3	0.0057	18
			C3	0.1977	2	0.0071	17
			C4	0.3336	1	0.0119	15
			C5	0.0842	6	0.0030	23
			C6	0.0872	5	0.0031	22
B2	0.2674	2	C7	0.3271	2	0.0875	5
			C8	0.1077	4	0.0288	10
			C9	0.0586	5	0.0157	14
			C10	0.3464	1	0.0926	4
			C11	0.1163	3	0.0311	9
			C12	0.0439	6	0.0117	16
B3	0.0823	4	C13	0.2314	2	0.0191	12
			C14	0.0669	4	0.0055	19
			C15	0.0443	5	0.0036	21
			C16	0.4661	1	0.0384	8
			C17	0.1913	3	0.0157	13
B4	0.2052	3	C18	0.1096	3	0.0225	11
			C19	0.5813	1	0.1193	2
			C20	0.3092	2	0.0634	7
B5	0.4094	1	C21	0.2106	3	0.0862	6
			C22	0.5485	1	0.2245	1
			C23	0.2409	2	0.0986	3

.

5. Results and Discussion

5.1. Importance Analysis and Results of Level 1 Indicators

As illustrated in Figure 3, Security and Privacy (0.4094) emerges as the most critical indicator within the level, followed by Learning and Operation (0.2674), Information Architecture (0.2052), Interaction Experience (0.0823), and finally, Emotional Design (0.0357). This reflects the highest level of concern users have for the protection of personal information and financial data security when using second-hand transaction apps. Furthermore, the ability to simply and conveniently use the software, easily find the information users seek, and the interface attractiveness are all important criteria for evaluating second-hand transaction apps. Although Emotional Design holds a lower weight, it plays a subtle yet significant role in enhancing user loyalty and establishing brand image. Therefore, in the design and development of second-hand transaction apps, comparatively more focus should be placed on protecting user information and assets as well as facilitating a convenient and comfortable software usage experience than on enhancing user loyalty and building a brand image.

5.2. Analysis and Results of the Importance of Level 2 Indicators

The analysis of the secondary indicators for each primary indicator was conducted in descending order of their weights, from the highest to the lowest. The specific details of the analysis are as follows:

(1)

Security and Privacy (B5) Indicator Layer.

As depicted in Figure 4, in the distribution of weights for the secondary indicators under Security and Privacy, Property Protection (0.5485) carries the highest weight, followed by Privacy Protection (0.2409), with System Security (0.2106) ranking last. This ranking reflects that protection of their financial assets and personal information is prioritized by app users engaged in online transactions over the ability to defend against viruses. Therefore, when optimizing the security and privacy of an app, priority should be given to strengthening asset and privacy protection.

(2)

Learning and Operation (B2) Indicator Layer.

As depicted in Figure 5, in the weight distribution of secondary indicators for Learning and Operation, Simplicity in Use (0.3464) emerges as the highest weighted indicator, followed closely by Efficiency (0.3271), then Learnability (0.1163), Memorability (0.1077), and Convenience (0.0586), with Error Prevention (0.0439) at the bottom. This order reveals a strong user preference for apps that boast intuitive interfaces, simplicity in mastery, the capability to efficiently accomplish tasks, and the ease of quickly learning and remembering operational procedures. Therefore, in the process of optimizing the learning and operation aspects of an app, placing a premium on Usability and Efficiency should be considered a priority.

(3)

Information Architecture (B4) Indicator Layer.

As shown in Figure 6, in the weight distribution of secondary indicators for Information Architecture, Page Information Architecture (0.5813) holds the highest weight, followed by Navigation (0.3092), and then Findability (0.1096). This indicates that users place particular importance on the app’s page layout and content organization as well as on how navigation clearly guides users to different functional areas. Therefore, when optimizing an app’s information architecture, it is recommended that Page Information Architecture and Navigation be prioritized.

(4)

Interaction Experience (B3) Indicator Layer.

As illustrated in Figure 7, in the distribution of weights for secondary indicators of Interaction Experience, Consistency (0.4661) has the highest weight, followed by Immediate Feedback (0.2314), Error Feedback (0.1913), Interface Simplicity (0.0669), and finally, Help (0.0443). This underscores the importance users place on consistency in interface text and images with common sense as well as the need for prompt responses after actions. Therefore, when designing and optimizing an app’s interaction experience, it is advisable to prioritize consistency and immediate feedback.

(5)

Emotional Design (B1) Indicator Layer.

As depicted in Figure 8, within the distribution of weights for secondary indicators of Emotional Design, Social (0.3336) is the highest weighted, followed by Accessibility (0.1977), Flexibility (0.1607), Satisfaction (0.1366), Value Realization (0.0872), and Pleasure (0.0842). This highlights that while users seek a personalized experience in app usage, they also value the app’s capability to facilitate communication and establish connections as well as its accessibility features. Therefore, in the design and optimization of an app’s emotional design, it is advisable to prioritize Social Interaction and Flexibility.

5.3. Relative Importance of Composite Weights

As illustrated in Figure 9, after conducting an analysis on the relative importance of composite weights by multiplying the weights of 23 secondary indicators with their respective higher-level indicator weights, several key findings emerged. The most critical components identified are as follows: under the Security and Privacy category, Asset Property Protection (0.2245), focusing on the safeguarding of user financial security; within the Information Architecture category, Page Information Architecture (0.1193), ensuring users’ clear comprehension of page layouts and content; and, again in Security and Privacy, Privacy Protection (0.0986) aimed at preventing the leakage of users’ personal data. Indicators with relatively lower importance predominantly fall within the Emotional Design category, including Pleasure (0.0030), where the aim is to elicit users’ emotional values through textual and visual content; Value Realization (0.0031), which is achieved through the software fulfilling anticipated goals; and in the Interaction Experience category, Help (0.004) offers detailed instructions for product use and answers to common questions. These findings offer valuable insights for future assessments of the usability of second-hand trading platform apps.

6. Conclusions

This study aimed to propose usability evaluation indicators and their weights for second-hand trading applications, enhance user stickiness, and reduce app uninstall rates. Through a literature review, 29 usability indicators were summarized. Using the Delphi method with 20 experts, these 29 indicators were refined by removing duplicates and merging similar ones, resulting in 23 usability indicators: Satisfaction (C1), Flexibility (C2), Accessibility (C3), Social (C4), Pleasure (C5), Value Realization (C6), Efficiency (C7), Memorability (C8), Convenience (C9), Simplicity in Use (C10), Learnability (C11), Error Prevention (C12), Immediate Feedback (C13), Interface Simplicity (C14), Help (C15), Consistency (C16), Error Feedback (C17), Findability (C18), Page Information Architecture (C19), Navigation (C20), System Security (C21), Property Protection (C22), and Privacy Protection (C23) (addressing RQ1).

Next, these 23 indicators were grouped and named through Exploratory Factor Analysis (EFA). The indicators were classified into five groups: Emotional Design (B1), Learning and Operation (B2), Interaction Experience (B3), Information Architecture (B4), and Security and Privacy (B5). Finally, the Analytic Hierarchy Process (AHP) was used for hierarchical structuring. The primary indicator Emotional Design (B1) includes the following secondary indicators: Satisfaction (C1), Flexibility (C2), Accessibility (C3), Social (C4), Pleasure (C5), and Value Realization (C6). The primary indicator Learning and Operation (B2) includes Efficiency (C7), Memorability (C8), Convenience (C9), Simplicity in Use (C10), Learnability (C11), and Error Prevention (C12). The primary indicator Interaction Experience (B3) includes Immediate Feedback (C13), Interface Simplicity (C14), Help (C15), Consistency (C16), and Error Feedback (C17). The primary indicator Information Architecture (B4) includes Findability (C18), Page Information Architecture (C19), and Navigation (C20). The primary indicator Security and Privacy (B5) includes System Security (C21), Property Protection (C22), and Privacy Protection (C23) (addressing RQ2).

Lastly, based on the expert ratings, the weights for different levels of indicators were obtained, including their overall weights. Among the primary indicators, the highest weight was for Security and Privacy (B5, 0.4094). Within the Emotional Design (B1) indicator, the highest-weighted secondary indicator was Social (C4, 0.3336). Within the Learning and Operation (B2) indicator, the highest-weighted secondary indicator was Simplicity in Use (C10, 0.3464). Within the Interaction Experience (B3) indicator, the highest-weighted secondary indicator was Consistency (C16, 0.4661). Within the Information Architecture (B4) indicator, the highest-weighted secondary indicator was Page Information Architecture (C19, 0.5813). Within the Security and Privacy (B5) indicator, the highest-weighted secondary indicator was Property Protection (C22, 0.5485), (addressing RQ3). The findings of this study can serve as a comprehensive guide for enhancing the usability of second-hand trading apps through design, which is crucial for assessing and improving app usability. By enhancing the usability of these apps, we can increase user engagement, encourage the purchase of second-hand products, and reduce waste generation, thereby positively impacting environmental conservation.

However, a limitation of this study is the small number of experts involved in the surveys, which did not allow for a sufficiently diverse range of opinions. Additionally, the selection of usability assessment indicators for mobile second-hand trading apps was solely based on a literature review. In consideration of these factors, future research plans should aim to expand the pool of experts participating in the surveys and the range of selected indicators toward more systematically gathering and integrating a broader spectrum of opinions and indicators.