The Sustainable Development Goals (SDGs), a global and national commitment to help the world's population achieve 17 global goals and targets by 2030, were announced by both developed and developing nations at the Sidang Umum PBB in September 2015 (Bappenas, 2025). The following goals are pertinent to this study: Goal 9 (Industry, Innovation, and Infrastructure), Goal 12 (Responsible Consumption and Production), Goal 13 (Climate Action), and Goal 2 (Zero Hunger).

Davis created the groundbreaking Technology Acceptance Method (TAM) in 1986 to provide guidance on the adoption and usage of information technology. TAM is better than other models like the Theory of Reasoned Action (TRA) and the Theory of Planned Behavior (TPB), according to almost 80% of the data and user behavior (Marikyan, D. & Papagiannidis, S., 2023). A framework for learning and comprehending how people use and accept technology is offered by the TAM model.

Using the Technology Acceptance Model (TAM) and individual behavior toward an information system, which is based on system usage and behavioral intents, this study aims to examine the factors impacting individual purchase intentions on substandard food applications. Perceived utility and attitude have an impact on behavioral intentions. The term attitude describes a person's assessment of an information system, specifically a favorable or unfavorable response. Perceived utility and perceived ease of use are the two beliefs that determine attitude (May & Immanuel, 2022).

A favorable user experience is mostly shaped by perceived utility and perceived simplicity of use (Zhou et al., 2022; Sihite, 2023). User engagement will be boosted by perceived benefits and ease of use, which will lead to both continuing use and referrals to other online users.

Perceived usefulness, as defined by Davis (1989) in studies by Yuan et al. (2021); Fitri & Wulandari (2020); Sihite (2023), is the degree to which an individual thinks that utilizing the system will boost their productivity. As stated by Davis et al. (1989) in Ashghar & Nurlatifah (2020) and Chandra & Sijabat (2022), the following are indicators of perceived utility: Work more efficiently, job performance, increased productivity, effectiveness, ease of use, and usefulness in finishing tasks.

Perceived ease of use, as defined by Davis et al. (1989) in research by Himel et al. (2021), is the capacity to utilize a system with little effort. Purchase intention and potential customers' propensity to transact online are positively correlated with a platform's degree of ease of use (Fitri & Wulandari, 2020). According to Davis (1989), the following items are used in study (Muliadi & Japarianto, 2021; Chandra & Sijabat, 2022) to measure perceived ease of use: Ease of Use, Easy to Learn, Controllable, Clear and Understandable, Flexible, and Easy to Become Skillful.

According to Trivedi et al. (2022), in a study by Valle et al. (2024), social influence examines the impact of social networks, friends' opinions, and the people around them on shaping a person's attitude toward technology. Social influence can take the form of friend recommendations, celebrity testimonials, or the influence of specific social groups (Prasetya, 2024). According to Engel, Blackwell, and Miniard (1995) in Nurochim et al. (2022), indicators of social influence include information or knowledge, experience, and credibility.

Attitude is a person's positive or negative feelings about how they feel about performing a particular behavior (Himel et al., 2021). A favorable or unfavorable attitude toward technology can influence a person's willingness and efficacy to utilize technology (Valle et al., 2024). A more favorable attitude toward a product increases a consumer's purchase intention (Utama & Hanton, 2023). According to Ajzen (1991) in a study by Fazriah et al. (2023), attitude indicators include cognitive, affective, and conative components.

According to Pavlou (2003) and Peña-García et al. (2020) in a study by May & Immanuel (2022), purchase intention is the extent to which a customer is willing to purchase a product through technology. According to Ferdinand (2006) in a study by Wendy (2024), purchase intention consists of transactional intention, referential intention, and exploratory intention.

Conceptual Framework Drawing

H1: Attitude is positively and significantly impacted by perceived usefulness.

H2: Attitude is positively and significantly impacted by perceived convenience.

H3: Social influence significantly and favorably affects attitude

H4: Purchase intention is positively and significantly impacted by perceived usefulness.

H5: Purchase intention is positively and significantly impacted by perceived convenience.

H6: Purchase intention is positively and significantly impacted by social influence.

H7: Purchase intention is positively and significantly impacted by attitude.

H8: Attitude mediates perceived usefulness on purchase intention in a positive and significant way.

H9: Attitude mediates perceived convenience on purchase intention in a positive and significant way.

H10: Attitude mediates social influence on purchase intention in a favorable and significant way.

Descriptive analysis of variables provides an overview of respondents' responses, as seen from the high and low scores of each questionnaire indicator.

Perceived Usefulness Variable

Based on the table above, it is known that the indicator with the highest mean value for the perceived usefulness variable is PU6 at 3.876, while the lowest is PU2 at 3.495. This indicates that respondents agree that the Surplus app is generally beneficial for online surplus food purchases. However, improvements are needed in app performance to enhance time efficiency for users.

Based on the table above, the highest mean value for the perceived ease of use variable is PEOU4 at 3.857, and the lowest is PEOU5 at 3.629. This indicates that respondents perceive the Surplus app as quite flexible to use. However, providing more guides or tutorials may help users become more skilled in using the app.

Based on the table above, the indicator with the highest mean value for social influence is SI3 at 3.743, and the lowest is SI2 at 3.600. This indicates that respondents are quite motivated to use the Surplus app if their friends or family are using it. However, more active references such as from social media may be considered to further motivate users to recommend the app to others.

Based on the table above, the highest mean value for attitude is ATT3 at 3.810 and the lowest is ATT2 at 3.705. This shows that respondents rely more on behavior-based attitudes rather than emotionally-driven ones.

Based on the table above, the highest mean value for purchase intention is PI2 at 3.781 and the lowest is PI3 at 3.638. This shows that respondents are willing to buy and recommend the app to others, but they are not yet highly motivated to make a purchase due to environmental concern alone. This presents a strategic opportunity to provide more targeted sustainability education and awareness about the environmental benefits of using the app.

Structural Equation Modeling (SEM) with the Partial Least Square (PLS) methodology, processed with SEMPLS version 4.0, is the data analysis method employed in this study. Two criteria are used in the testing: the structural model (inner model) and the measurement model (outer model) (Hair, Hult, Ringle, & Sarstedt, 2016).

Convergent Validity

Outer Loading Test Results

Based on the table above, all indicators have outer loading values above 0.7. Thus, the results of the convergent validity test – outer loading – indicate that all items or indicators are valid and acceptable.

Based on the AVE test results table above, all variables have AVE values greater than 0.5. Therefore, all indicators are declared valid and meet the criteria for the convergent validity test.

1. Discriminant Validity

a. Discriminant Validity – Cross Loading

Test Results Discriminant Validity - Cross loading

Source: Author's Processed Data Results, 2025

Based on the table above, the results of the discriminant validity test through cross loading show that all statement items have loading values that exceed those of other constructs. Therefore, it can be concluded that each construct has good discriminant validity, as the indicators in each construct block load better on their respective constructs than on others.

Source: Processed Data by Author, 2025

It is clear from the above-mentioned Fornell-Larcker discriminant validity test results that each construct's square root of the AVE (diagonal values) is higher than its correlation with other constructs. This demonstrates that the conceptions meet the Fornell-Larcker criterion for excellent discriminant validity.

Source: Processed Data by Author, 2025

As can be seen from the above table, all variables have good reliability according to the results of the Cronbach's Alpha and Composite Reliability tests, with each value exceeding the suggested cutoff point of 0.7. This demonstrates that every indicator in the design is trustworthy and satisfies the requirements of the reliability test.

Source: Processed Data by Author, 2025

The R-square value for the Attitude variable is 0.737, with an adjusted R- square of 0.729, according to the R-square table above. This suggests that 73.7% of the variation in attitude can be explained by the independent variables of perceived utility, perceived ease of use, and social influence taken together. Other factors not covered in this study account for the remaining 26.3%. All independent constructions are thought to have a considerable influence on attitude because the corrected R-square is close to 75%.

In contrast, the purchase intention variable's R-square value is 0.798, with an adjusted R-square of 0.790. This indicates that the independent factors account for 79.8% of the variance in purchase intention. Other factors not included in this study account for the remaining 20.2%. Higher R-square values suggest that independent variables can more successfully explain the dependent variable, which demonstrates the strength of the structural model.

Source: Processed Data by Author, 2025

According to the results of the F-square test, perceived utility has a medium impact on attitude but a low impact on purchase intention. Perceived ease of use and social influence have a high impact on attitude but a modest impact on purchase intention. buy intention is strongly influenced by attitude, indicating that attitude is a key factor in determining buy intention.

Source: Processed Data by Author, 2025

Based on the Q-square test results, all Q² values are greater than 0, which indicates that the model has predictive relevance for both attitude and purchase intention.

Source: Processed Data by Author, 2025

Based on the table above, the SRMR value of 0.064 is less than 0.09, indicating a good model fit. Additionally, the NFI value is 0.798, which is higher than the 0.5 threshold, confirming that the model meets the criteria for model fit.

Source: Author's Processed Data Results, 2025

Source: Processed Data by Author, 2025

Based on the table above, the path coefficients (Original Sample column) range from 0.014 to 0.646. It can be concluded that 8 hypotheses have a relationship with a T-statistic > 1.96 and a p-value < 0.05, indicating a positive and significant effect. Meanwhile, 2 hypotheses show a T-statistic < 1.96 and a p-value > 0.05, indicating a positive but not significant effect.

The results of the hypothesis test show that the route coefficient value is 0.305, the p-value is 0.001 < 0.05, and the T-statistic is 3.285 > 1.96. Thus, the hypothesis is accepted, indicating that perceived utility has a positive and significant impact on attitude.

This is in line with past research by Himel et al. (2021), Agustriyani et al. (2021), and Ma & Chang (2024), which showed that attitude was positively and significantly impacted by perceived usefulness. If users think the program has practical attributes like speed, efficacy, and worth, they will be more inclined to utilize it.

The investigation found a route coefficient of 0.333, a p-value of 0.000 < 0.05, and a T-statistic of 3.973 > 1.96. As a result, the hypothesis is accepted, showing that perceived ease of use has a positive and significant impact on attitude.

This finding is in line with studies by Yuan et al. (2021), Ya & Ca (2023), and Pertami & Sukaatmadja (2021), which demonstrate that perceived ease of use has a significant impact on attitude. If the Surplus app is easy to use, understand, and navigate, users are more likely to have a positive attitude toward utilizing it.

As a result, how users feel about a program is greatly influenced by its user- friendliness.

The investigation found a route coefficient of 0.333, a p-value of 0.000 < 0.05, and a T-statistic of 3.973 > 1.96. As a result, the hypothesis is accepted, showing that perceived ease of use has a positive and significant impact on attitude.

This finding is in line with studies by Yuan et al. (2021), Ya & Ca (2023), and Pertami & Sukaatmadja (2021), which demonstrate that perceived ease of use has a significant impact on attitude. If the Surplus app is easy to use, understand, and navigate, users are more likely to have a positive attitude toward utilizing it. As a result, how users feel about a program is greatly influenced by its user- friendliness.

The path coefficient value is 0.075, the p-value is 0.371 > 0.05, and the T- statistic is 0.895 < 1.96, according to the findings of the hypothesis test. The hypothesis is thus disproved, suggesting that purchase intention is positively but not significantly impacted by perceived usefulness.

This finding aligns with studies by Chandra & Sijabat (2022), Supriyatna & Zakaria (2024), and Alfando et al. (2023), which showed that perceived usefulness has a positive but insignificant effect on purchase intention. Some users may feel that the usefulness of the application does not directly motivate them to have the intention to purchase via the Surplus app.

The test results show a path coefficient of 0.014, with a p-value of 0.817 > 0.05, and a T-statistic of 0.232 < 1.96. Thus, the hypothesis is rejected, meaning that perceived ease of use has a positive but not significant effect on purchase intention.

This finding is in line with research by Lestari (2022) and Suryani & Ramdhani (2022), which state that perceived ease of use is positively but insignificantly related to purchase intention. Although users find the app easy to use, straightforward, and functional without guidance, these aspects do not directly affect their desire to make a purchase.

The analysis indicates a path coefficient of 0.218, with a p-value of 0.014 < 0.05, and a T-statistic of 2.456 > 1.96. Therefore, the hypothesis is accepted, meaning social influence has a positive and significant effect on purchase intention.

This is consistent with findings by Dwisuardinata & Darma (2022), Vahdat et al. (2021), and Bramantyo & Utami (2022), which show that social influence significantly and positively affects purchase intention. Motivation from one's surroundings, such as friends or family, can lead users to become interested in purchasing surplus food through the platform.

The results show a path coefficient of 0.646, with a p-value of 0.000 < 0.05, and a T-statistic of 5.283 > 1.96. Thus, the hypothesis is accepted, indicating that attitude has a positive and significant effect on purchase intention.

This supports the findings of Agustriyani et al. (2021), Hafiz & Permana (2021), and Fragolia & Wibowo (2025), which demonstrate that the more positive the consumer’s attitude toward the product, the stronger their purchase intention will be.

The hypothesis test shows a path coefficient of 0.197, a p-value of 0.003 < 0.05, and a T-statistic of 2.991 > 1.96. Thus, the hypothesis is accepted, indicating that attitude can mediate the relationship between perceived usefulness and purchase intention.

This finding is supported by Purwianti et al. (2024), who demonstrated that attitude mediates the effect of perceived usefulness on purchase intention. When users perceive that the Surplus app aligns with their expectations (e.g., helping to reduce food waste, providing edible surplus food at a discount), this increases their positive attitude, which in turn strengthens their purchase intention. Similar findings were also revealed by Lestari (2022) and Pradana & Aksari (2022).

The analysis shows a path coefficient of 0.215, with a p-value of 0.001 < 0.05, and a T-statistic of 3.315 > 1.96. Hence, the hypothesis is accepted, indicating that attitude can mediate the relationship between perceived ease of use and purchase intention.

This is in accordance with Pradana & Aksari (2022), who found that attitude mediates the effect of perceived ease of use on purchase intention. Although the app is considered user-friendly, attractive, flexible, and easy to understand, these factors alone are not sufficient to directly drive users to purchase. However, they can indirectly influence purchase intention by shaping a positive attitude, as also noted by Lestari (2022) and Pertami & Sukaatmadja (2021).

A p-value of 0.003 < 0.05, a T-statistic of 2.967 > 1.96, and a path coefficient of 0.216 are the results of the investigation. The hypothesis is thus accepted, suggesting that the impact of social influence on purchase intention is mediated by attitude.

Purwianti et al. (2024) found that users may be influenced to buy excess food both directly (by friends, family, communities, influencers, and celebrity endorsements) and indirectly (by the formation of a positive attitude shaped by social norms). Recommendations from their social network may have encouraged some originally disinterested people to use the app.