Source-linked AI summary
Exploring consumers response to text-based chatbots in e-commerce: The moderating role of task complexity and chatbot disclosure
Xusen Cheng, Ying Bao, Alex Zarifis, Wankun Gong, Jian Mou
TL;DR
This study addresses limited understanding of how chatbot attributes and boundary conditions shape consumer trust and responses in e-commerce. Using a survey and ordinary least squares regression within the stimulus–organism–response framework, it finds that empathy and friendliness increase trust, with effects moderated by task complexity and disclosure. Trust also increases reliance and decreases resistance, while the study remains limited to selected chatbot attributes and the e-commerce setting.
Problem
The study examines limited evidence on how chatbot attributes, task complexity, and identity disclosure shape consumer trust and responses in e-commerce.
Method
The study uses 299 usable survey responses and ordinary least squares regression within the stimulus–organism–response framework.
Results
Consumers’ perceived empathy and friendliness increase trust; task complexity and disclosure moderate these relationships, while trust increases reliance and decreases resistance.
Takeaways & Limitations
Chatbot trust and responses depend on both perceived attributes and contextual conditions such as task complexity and identity disclosure.
Takeaways & Limitations
The study examines only empathy and friendliness among chatbot attributes and only the e-commerce setting.
Abstract
from arXiv · showhide
Artificial intelligence based chatbots have brought unprecedented business potential. This study aims to explore consumers trust and response to a text-based chatbot in ecommerce, involving the moderating effects of task complexity and chatbot identity disclosure. A survey method with 299 useable responses was conducted in this research. This study adopted the ordinary least squares regression to test the hypotheses. First, the consumers perception of both the empathy and friendliness of the chatbot positively impacts their trust in it. Second, task complexity negatively moderates the relationship between friendliness and consumers trust. Third, disclosure of the text based chatbot negatively moderates the relationship between empathy and consumers trust, while it positively moderates the relationship between friendliness and consumers trust. Fourth, consumers trust in the chatbot increases their reliance on the chatbot and decreases their resistance to the chatbot in future interactions. Adopting the stimulus organism response framework, this study provides important insights on consumers perception and response to the text-based chatbot. The findings of this research also make suggestions that can increase consumers positive responses to text based chatbots. Extant studies have investigated the effects of automated bots attributes on consumers perceptions. However, the boundary conditions of these effects are largely ignored. This research is one of the first attempts to provide a deep understanding of consumers responses to a chatbot.
1. Introduction
Chatbots offer substantial e-commerce service benefits, but consumers’ perceptions can limit their effectiveness. This study examines how chatbot attributes, task complexity, and identity disclosure shape trust and subsequent responses.
- Chatbots automate customer service, handle many communications simultaneously, and can understand consumer requests effectively.
- Consumers may resist chatbots because they perceive technology-based agents as uncomfortable for personal needs and lacking empathy.
- Prior research links AI-system attributes such as anthropomorphism, humanness, empathy, and adaptivity to consumer perceptions and decisions.
- The study asks how chatbot attributes affect consumer responses, how task complexity and disclosure moderate those relationships, and how trust shapes responses.
- Using the stimulus–organism–response framework, the study tests consumers’ trust and subsequent responses to text-based e-commerce chatbots.
2.1 Stimulus–organism–response model
The stimulus–organism–response framework explains chatbot interactions as environmental stimuli that shape consumers’ internal states and behavioral outcomes.
- The SOR framework comprises stimulus, organism, and response, corresponding to environmental factors, internal conditions, and behavioral outcomes.
- In this study, chatbot attributes serve as stimuli that arouse consumers during text-based interactions.
- The organism represents consumers’ affective and cognitive processes between chatbot stimuli and responses.
- Responses include positive actions such as staying, exploring, or affiliating, and negative reactions toward the chatbot.
2.2 Stimulus: unique attributes of a text-based chatbot
The paper focuses on chatbot characteristics that make automated service interactions more human-like and shape consumers’ perceptions.
- Text-based chatbots automatically respond to consumers’ e-commerce requests, with human-likeness linked to perceived realism and usability.
- Empathy is the ability to identify, understand, and react to others’ thoughts, feelings, behavior, and experiences.
- Empathy includes cognitive and affective perceptions and represents a service-quality dimension relevant to successful consumer–robot interactions.
- Friendly service providers can improve consumers’ perceptions of service quality and encourage positive responses such as repurchase and positive word of mouth.
2.3 Organism: trust toward the text-based chatbot
Trust is the internal consumer state connecting chatbot perceptions with behavioral responses. The section frames empathy and friendliness as proposed trust antecedents.
- Consumers’ trust formation toward the text-based chatbot is treated as the organism component of the SOR framework.
- Trust is defined as an individual or group’s willingness to be vulnerable to another party.
- Prior research identifies ability, integrity, benevolence, and anthropomorphic characteristics as important trust antecedents.
- The hypotheses propose positive relationships between chatbot empathy, friendliness, and consumers’ trust.
2.4 Moderators: task complexity and disclosure of the text-based chatbot
Task complexity and chatbot identity disclosure are examined as conditions that shape how chatbot attributes relate to consumer trust. Complex customer-service tasks may weaken these relationships, while disclosure changes the effects of empathy and friendliness.
- Task complexity: Customer-service tasks vary across prepurchase, purchase, and postpurchase stages, with postpurchase requests potentially more complex.Examples include information search, service recovery, repurchase decisions, product returns, and postpurchase engagement.
- Task complexity: Task–technology fit theory motivates examining whether chatbot characteristics meet the requirements of specific customer-service tasks.When technology meets task-specific needs, the interaction is expected to be more satisfying.
- Task complexity: The study applies task complexity to understand consumer trust toward chatbots across different customer-service contexts.Complex problems may lead consumers to turn to customer service for help.
- Task complexity: Task complexity is hypothesized to weaken the positive relationships between chatbot empathy, friendliness, and consumer trust.The hypotheses specify weaker effects when the task is complex.
- Chatbot disclosure: Disclosure means revealing the chatbot’s machine identity before interaction, which is hypothesized to alter empathy’s and friendliness’s effects on trust.The study hypothesizes a stronger empathy–trust relationship and a stronger friendliness–trust relationship when chatbot identity is disclosed.
2.5 Response: avoidance/approach behavior toward the text-based chatbot
The study broadens chatbot-response analysis beyond approach intentions to include both reliance and resistance. It frames trust as related to greater reliance and lower resistance in future interactions.
- Response behaviors: The SOR framework distinguishes consumers’ approach behaviors from averting behaviors in responses to automated bots.The study operationalizes approach response as reliance and averting response as resistance.
- Response behaviors: Consumers may rely more on humans than algorithms for forecasting because algorithms cannot provide reasons or explanations for decisions.The absence of explanations is described as increasing consumers’ distrust.
- Response behaviors: Trust toward the chatbot is hypothesized to be positively related to consumers’ reliance on it.This proposition treats reliance as a future approach response.
- Response behaviors: Prior research has mainly examined acceptance, future use intention, or other approach responses to AI-based bots.Only a few studies have investigated consumer resistance, despite consumer pushback being a major concern.
- Response behaviors: Trust toward the chatbot is hypothesized to be negatively related to consumers’ resistance to it.This proposition treats resistance as a future averting response.
- Response behaviors: The conceptual model presents the proposed relationships among chatbot perceptions, trust, and response behaviors.The model is presented in Figure 1.
3. Research method
The study used a survey of e-commerce chatbot users and analyzed the proposed model with reliability, validity, common-method-variance, and regression tests.
- Sample and data collection: 299 usable participants with prior e-commerce chatbot experience were analyzed after respondents without such experience were filtered out.The survey was distributed in China through a professional survey website and social network.
- Measures: Existing multi-item scales and seven-point Likert measures captured chatbot attributes, moderators, consumer perceptions, and responses.The constructs included empathy, friendliness, task complexity, chatbot disclosure, trust, reliance, and resistance.
- Data analysis: SmartPLS 3.2.8 assessed reliability, validity, and common method variance, while SPSS 22.0 tested the hypotheses.The analysis proceeded through reliability and validity analysis, common method variance testing, and hypothesis testing.
- Measurement assessment: All measures showed acceptable reliability, convergent validity, and discriminant validity based on internal consistency, AVE, Fornell–Larcker, and cross-loading tests.Cronbach’s α exceeded the recommended threshold, AVE values were above 0.5, and square roots of AVE exceeded construct correlations.
- Common method variance: Procedural remedies and Harman’s single-factor and marker-variable tests indicated that common method variance was not a serious problem.Age was used as the marker variable, and the structural-model paths of interest remained statistically significant after its inclusion.
- Hypotheses testing: Ordinary least squares regression tested the hypotheses, including main effects, interaction effects, and trust relationships with reliance and resistance.The models examined chatbot empathy and friendliness, moderation by task complexity and disclosure, and trust’s associations with reliance and resistance.
4. Discussions and conclusions
The study examines how chatbot attributes and contextual conditions shape trust and subsequent consumer responses in e-commerce. It finds that empathy and friendliness build trust, while task complexity and identity disclosure alter these relationships.
- Findings: Empathy and friendliness are both positively associated with consumers’ trust in the text-based chatbot, with empathy having the stronger effect.Consumers are described as valuing the chatbot’s ability to understand their perspectives and specific needs.
- Boundary conditions: Task complexity weakens the positive relationship between chatbot friendliness and trust but does not significantly affect empathy’s relationship with trust.For complex tasks, consumers may prioritize professional ability and problem-solving over service manner.
- Boundary conditions: Chatbot identity disclosure weakens empathy’s relationship with trust while strengthening friendliness’s relationship with trust.
- Consumer responses: Trust is positively related to reliance on the chatbot and negatively related to resistance in future interactions.The positive relationship with reliance is reported as stronger than the negative relationship with resistance.
- Theoretical contributions: The study extends the stimulus-organism-response framework to automated text-based chatbots in e-commerce.The model links chatbot attributes with consumers’ trust and responses in this setting.
- Contributions and implications: The research incorporates task complexity and chatbot identity disclosure as boundary conditions and considers both approach and averting responses.Its practical implications address online stores and developers customizing chatbot capabilities and characteristics.
- Limitations: The study is limited to selected chatbot attributes and the e-commerce setting, while other chatbot characteristics, domains, and customer traits may yield different responses.
response to
This section presents cited prior research relevant to online retailing, artificial intelligence, trust, consumer behavior, and information systems. The supplied passages are bibliographic entries rather than substantive findings.
- Online retail and consumer experience: The section cites research on online retailing, shopping experience, store quality, and customer journeys.
- Artificial intelligence and chatbots: It includes studies addressing artificial intelligence, chatbots, social robots, avatars, and human-like system characteristics.
- Theoretical and systems perspectives: Additional citations address stimulus-organism-response models, task-technology fit, service interactions, and information-systems implementation.
Appendix
The appendix reports construct measurements used to assess chatbot perceptions, task complexity, disclosure, trust, reliance, and resistance. Table A1 organizes the measurement items and their references.
- Construct measurements: Empathy is measured through chatbot understanding of customer needs, individual attention, availability, and willingness to help.One empathy item is marked as dropped.
- Construct measurements: Friendliness is measured through kind service, friendly manner, and nice treatment during the interaction.
- Construct measurements: Task complexity is measured using items describing the customer-service task as simple or uncomplicated.
- Construct measurements: Chatbot disclosure is measured through whether the chatbot’s identity is disclosed or hidden before conversation.One disclosure item is marked as dropped.
- Construct measurements: Trust items assess honesty, capability, expectation fulfillment, and trust in the chatbot’s suggestions and decisions.One trust item is marked as dropped.
- Outcome measurements: Reliance and resistance are measured through dependence on chatbot decisions or services and preferences to avoid chatbot service.
About the authors
The authors work in information systems, e-commerce, artificial intelligence, user behavior, trust, and technology-enabled innovation. Their affiliations and research interests span institutions in China and the United Kingdom.
- Author biographies: Xusen Cheng is a Renmin University of China professor whose research focuses on information systems, e-commerce, sharing economy, AI, and behavior.
- Author biographies: Ying Bao is a doctoral candidate researching user behaviors, trust in the sharing economy, and corporate social responsibility.
- Author biographies: Alex Zarifis researches e-commerce, trust, privacy, artificial intelligence, blockchain, fintech, and online consumer perspectives.
- Author biographies: Wankun Gong studies e-government, e-service, e-commerce, human behavior, and technology-enabled innovation.