Data underlying the publication: Virtual Coaching for Smoking Cessation: What are Users Preference in Ethical Principles for Human Feedback Allocation

This is the analysis code underlying the paper "Virtual Coaching for Smoking Cessation: What are Users Preference in Ethical Principles for Human Feedback Allocation" by Glebs Labunskis, Nele Albers, and Willem-Paul Brinkman. In this paper, we conduct a mixed-methods analysis of people's preferences of ethical principles that a virtual assistant for smoking cessation should follow for deciding how to allocate human feedback.<strong>Data:</strong>Our analysis is based on the data collected in an online experiment in which more than 500 daily smokers interacted with the text-based virtual coach (i.e., a conversational agent) in up to 5 sessions. In each session, the virtual assistant proposed a new preparatory activity for quitting smoking or becoming more physically active, with the latter possibly aiding the former. After the 5 sessions, participants filled in a post-questionnaire in which they answered a set of questions. Our paper focuses on people's free-text responses to the question "When a human coach cannot give feedback to everybody after each session due to time constraints, which principles/rules do you think the virtual coach should follow to decide when a human coach should give feedback to people who are preparing to quit smoking?". The complete dataset can be found here: https://doi.org/10.17605/OSF.IO/78CNR.

本分析代码对应Glebs Labunskis、Nele Albers与Willem-Paul Brinkman发表的论文《戒烟虚拟辅导：用户在人类反馈分配伦理原则中的偏好》（*Virtual Coaching for Smoking Cessation: What are Users’ Preferences in Ethical Principles for Human Feedback Allocation*）。本文针对戒烟虚拟助手在决策如何分配人类反馈（human feedback）时应遵循的伦理原则偏好，开展混合方法分析。**数据说明：** 本分析基于一项线上实验收集的数据集，共有超过500名每日吸烟者参与，他们与基于文本的虚拟教练（即对话智能体（conversational agent））开展了最多5轮交互。每一轮会话中，虚拟助手会为参与者提供一项全新的戒烟或提升身体活动量的准备活动，其中后者可辅助前者达成戒烟目标。5轮会话结束后，参与者需填写事后问卷并回答一系列问题。本文聚焦于参与者对以下问题的自由文本回答：“当人类教练因时间限制无法在每轮会话后为所有人提供反馈时，你认为虚拟教练应遵循哪些原则/规则，来决策何时为正在准备戒烟的人群提供人类反馈？”完整数据集可通过以下链接获取：https://doi.org/10.17605/OSF.IO/78CNR。