Description
Deliver a standard questionnaire as a branching chatbot conversation with button or free-text answers, optionally capturing sentiment, while watching for mode effects against web forms.
A conversational survey turns each questionnaire item into a turn in a chat flow, with question, answer and follow-up branches. Respondents answer by buttons or free text, with entity recognition for dates and numbers, and optional open conversations collect context a form would miss. The ALAMEDA project modelled patient-reported-outcome questionnaires for stroke, multiple sclerosis and Parkinson's patients this way, choosing with clinicians and patient groups which questionnaires suited conversation, and fused text sentiment with camera-based facial emotion recognition into a session sentiment that could steer the dialogue. Because a validated instrument changes when its mode changes, the conversational version must be checked for mode effects (item wording, response scale presentation, social presence) against the form it replaces. The ALAMEDA evaluation was preliminary: text sentiment F1 of 0.56 to 0.76 and no ground truth for the fused estimate.
Attributes
- Discipline
- Survey & Psychometrics
- Loss type
- Siloed
- Stage
- Discovery, Post-launch
- Data
- Mixed
- Effort
- Weeks
Steps
- Gather requirements with representative respondents and domain experts, and choose which questionnaires suit a conversational format.
- Model each questionnaire as a conversation flow with question, answer and follow-up branches, keeping validated item wording and response options intact.
- Offer button and free-text answers, with entity recognition for dates and numbers and confirmation of parsed values.
- Optionally add open conversational prompts for context, and sentiment capture only with explicit consent.
- Pilot against the original form with the same respondents or randomised groups to check for mode effects on scores and completion.
- Store responses and transcripts in a structured form that analysts can link back to the original instrument scoring.
Sources
- Chatbot Research and Design: CONVERSATIONS 2022 (Følstad et al., 2023)
Also known as
- Conversational questionnaire administration
- Chatbot-delivered patient-reported outcomes
- Chatbot survey
- Chatbot-administered questionnaires with multimodal sentiment capture
Categories
- Analytics & Behaviour
- Survey & Psychometrics