Should AI voice bot service quality be measured as one construct? Evidence from Dayan (2026)
A measurement-focused reading of Dayan (2026, JTAER, Scopus Q1): why the quality facets of a voice agent behave as one higher-order construct, and what that means for researchers who build service quality scales for conversational agents.
Dayan, O. (2026). One voice, one impression: Why customers perceive the service quality of AI voice bots holistically and what it means for electronic commerce. Journal of Theoretical and Applied Electronic Commerce Research, 21(10), 363. https://doi.org/10.3390/jtaer21100363
In brief
Researchers who study chatbots usually measure quality as a set of separate dimensions. Dayan (2026) tested this assumption for AI voice bots with PLS-SEM and competing measurement models on 295 users across eight service categories. The facets remained distinguishable, yet one general factor accounted for most of their common variance, which supports modeling perceived quality as a single higher-order construct.
The measurement question
Text chatbot research has generally treated quality as a set of distinct attributes. Dayan (2026) asks whether that structure holds when the agent speaks: a voice-first encounter is one continuous spoken dialogue without a visual interface.
How the study tested it
The study compared competing measurement models rather than assuming one structure. Data came from a cross-sectional survey of 295 voice bot users in Israel across eight service categories, analyzed with partial least squares structural equation modeling.
What the results mean for scale design
The facets of quality can be told apart, yet most of their shared variance comes from one general factor. For researchers, this supports a higher-order specification of perceived quality for voice agents. The data are cross-sectional and self-reported, so the relationships are associations.
Questions researchers ask
Is there evidence that voice bot service quality should be modeled as a higher-order construct?
Dayan (2026, JTAER) compared competing measurement models and found that one general factor accounts for most of the common variance of the quality facets, supporting a single higher-order construct.
What method was used to test the structure of AI voice bot quality?
Partial least squares structural equation modeling with competing measurement models, on a cross-sectional survey of 295 voice bot users in Israel across eight service categories.
Can I use separate quality dimensions for voice assistants in my study?
Dayan (2026) found the facets remain distinguishable, so they can be measured, but most of their shared variance reflects one general factor, which supports a higher-order model.
How to cite
Dayan, O. (2026). One voice, one impression: Why customers perceive the service quality of AI voice bots holistically and what it means for electronic commerce. Journal of Theoretical and Applied Electronic Commerce Research, 21(10), 363. https://doi.org/10.3390/jtaer21100363
Read the full research summary
Related
References
Dayan, O. (2026). One voice, one impression: Why customers perceive the service quality of AI voice bots holistically and what it means for electronic commerce. Journal of Theoretical and Applied Electronic Commerce Research, 21(10), 363. https://doi.org/10.3390/jtaer21100363