A Data-Driven Socio-Technical Model of Technostress in FinTech Adoption: Integrating TAM–TPB Within an SOR Framework

Muhardi Saputra, Prof. Ir. Paulus Insap Santosa, M.Sc., Ph.D., IPU , Ir. Adhistya Erna Permanasari, ST., MT., Ph.D., IPM., ASEAN Eng.

2026 | Disertasi | S3 Teknik Elektro

Pesatnya pertumbuhan layanan keuangan digital (financial technology atau FinTech) di negara berkembang, khususnya Indonesia, telah memberikan manfaat signifikan terhadap inklusi keuangan. Namun, meningkatnya kompleksitas ekosistem digital juga berpotensi memunculkan technostress yang belum sepenuhnya dapat dijelaskan oleh model adopsi teknologi konvensional seperti Technology Acceptance Model (TAM) dan Theory of Planned Behavior (TPB). Hingga saat ini, dimensi technostress berdasarkan pengalaman nyata pengguna belum teridentifikasi secara sistematis, belum terintegrasi dengan model adopsi yang sudah ada, serta belum diuji robustitasnya dalam konteks lintas negara. Oleh karena itu, penelitian ini bertujuan untuk mengembangkan dan memvalidasi model adopsi Fintech berbasis technostress dengan mengintegrasikan dimensi technostress yang dihasilkan dari analisis data pengguna ke dalam kerangka TAM, TPB, dan Stimulus–Organism–Response (SOR).

Penelitian ini menggunakan pendekatan data-driven melalui analisis lebih dari 100.000 ulasan pengguna aplikasi fintech di Indonesia menggunakan Aspect-Based Sentiment Analysis (ABSA) dan Latent Dirichlet Allocation (LDA) untuk mengidentifikasi dimensi technostress. Dimensi yang diperoleh selanjutnya diintegrasikan ke dalam model perilaku dan diuji menggunakan Partial Least Squares Structural Equation Modeling (PLS-SEM) pada 545 responden Generasi Z di Indonesia, kemudian divalidasi secara lintas negara melalui Multi-Group Analysis (MGA) dengan melibatkan 450 responden Generasi Z di Thailand.

Hasil penelitian mengidentifikasi tiga dimensi utama technostress, yaitu security, customer support, dan ease of access. Hasil PLS-SEM menunjukkan bahwa model mampu menjelaskan 62,6% variansi perceived usefulness, 52,9% variansi attitude toward FinTech, dan 42,0% variansi behavioral intention. Dari 18 hipotesis yang diajukan, tiga hipotesis terkait dimensi Usability dikeluarkan setelah evaluasi model pengukuran. Dari 15 hipotesis yang tersisa, 13 hubungan terbukti signifikan, dengan security dan customer support menjadi faktor yang paling berpengaruh terhadap persepsi dan perilaku pengguna, sedangkan ease of access menunjukkan pengaruh yang lebih terbatas. Dua jalur yang tidak signifikan — customer support terhadap perceived behavioral control dan security terhadap attitude toward fintech — menunjukkan bahwa kedua dimensi ini memengaruhi persepsi pengguna secara selektif, bukan seragam; pengaruh security terutama mengalir melalui jalur instrumental dan sosial, bukan melalui pembentukan sikap afektif — pola yang justru memperkuat, bukan melemahkan, kekhasan teoretis kedua dimensi tersebut. Selanjutnya, validasi lintas negara menegaskan stabilitas struktural model yang dikembangkan, dengan seluruh jalur hipotesis tetap konsisten arahnya dan sebagian besar tetap signifikan baik di Indonesia maupun Thailand; kemampuan eksplanatif behavioral intention mencapai 70,1% pada model gabungan, peningkatan yang sebagian besar disebabkan oleh jalur struktural terkait technostress yang jauh lebih kuat pada subsampel Thailand, bukan oleh penguatan hubungan yang seragam di kedua konteks.

Temuan penelitian ini memberikan kontribusi dalam pengembangan model adopsi fintech berbasis technostress yang dibangun dari pengalaman nyata pengguna dan tervalidasi pada dua ekosistem digital yang berbeda di negara berkembang. Dengan demikian, penelitian ini menegaskan bahwa kontribusi integrasi TAM–TPB–SOR beroperasi pada empat tingkat yang berbeda: bersifat generatif pada tingkat dimensi/konstruk, bersifat baru secara struktural pada tingkat relasi, bersifat konfirmatif yang disengaja pada tingkat relasi inti TAM–TPB sebagai pemeriksaan validitas, dan bersifat kontekstual pada tingkat kondisi batas lintas negara.

The rapid growth of digital financial services (financial technology, or FinTech) in developing countries, particularly Indonesia, has significantly enhanced financial inclusion. However, the increasing complexity of digital ecosystems may also introduce technostress, which has not been fully addressed by conventional technology adoption models such as the Technology Acceptance Model (TAM) and Theory of Planned Behavior (TPB). To date, the constructs of technostress derived from real user experiences have not been systematically identified, integrated into existing technology adoption models, or validated for robustness across different national contexts. Therefore, this study aims to develop and validate a technostress-informed fintech adoption model by integrating empirically derived technostress constructs into the TAM, TPB, and Stimulus–Organism–Response (SOR) framework.

This study employs a data-driven approach by analyzing more than 100,000 user reviews of fintech applications in Indonesia using Aspect-Based Sentiment Analysis (ABSA) and Latent Dirichlet Allocation (LDA) to identify key technostress constructs. The identified constructs were subsequently incorporated into a behavioral adoption model and empirically tested using Partial Least Squares Structural Equation Modeling (PLS-SEM) with 545 Gen Z respondents in Indonesia. The model was further validated through cross-country analysis using Multi-Group Analysis (MGA) involving 450 Gen Z respondents from Thailand.

The findings identified three major technostress constructs, namely security, customer support, and ease of access. The PLS-SEM results demonstrate that the proposed model explains 62.6% of the variance in perceived usefulness, 52.9% in attitude toward FinTech, and 42.0% in behavioral intention. Among the 15 proposed hypotheses, 13 were supported, indicating that security and customer support are significant determinants of users' cognitive and social perceptions, whereas ease of access exerts a relatively weaker effect. The two unsupported paths—Customer Support to Perceived Behavioral Control and Security to Attitude Toward Fintech—indicate that these constructs shape user perceptions selectively rather than uniformly, with Security's influence operating mainly through instrumental and social channels rather than through affective attitude formation, a pattern that reinforces rather than undermines the constructs' theoretical distinctiveness. Furthermore, cross-country validation confirms the structural stability of the proposed model, with all hypothesized paths remaining directionally consistent and predominantly significant in both Indonesia and Thailand; the explanatory power of behavioral intention reaches 70.1% in the combined model, an increase attributable in large part to substantially stronger technostress-related structural paths observed in the Thai subsample rather than to a uniform strengthening of relationships across both contexts.

This study contributes to the advancement of fintech adoption research by proposing a data-driven, technostress-informed adoption model developed from real user experiences and empirically validated across two distinct digital ecosystems in developing countries. In doing so, the study clarifies that the contribution of the TAM–TPB–SOR integration operates at four distinct levels: it is generative at the level of constructs, structurally novel at the level of relations, deliberately confirmatory at the level of core TAM–TPB relations as a validity check, and contextual at the level of cross-country boundary conditions.

Kata Kunci : Technostress, Gen Z, Fintech Adoption, SEM, machine learning, Aspect-Based Sentiment Analysis, Latent Dirichlet Allocation