The Role of Lexical Alignment in Human Understanding of Explanations by Conversational Agents

Version 20260926: This version removes the Prolific participant IDs and the Prolific session IDs (the columns prolific_id and session_id) from the data files. The random study key uuid links the files. The data are otherwise unchanged, and the analysis notebook runs on them without changes.

This dataset was collected as part of the research investigating the role of lexical alignment in understanding. The dataset consists of the responses of 179 participants recruited via Prolific (a crowdsourcing platform).

1. Dialogue with a conversational agent (CA)

2. Survey questionnaire responses from the participants.

3. Primes selected or provided by the participants.

This repository contains:

1. Cleaned data sources listed above. The data is cleaned by removing data from participants who failed attention checks, or any other factor leading to error such as experiment setup, etc.
2. A Python (Jupyter) playbook to analyse the collected data (Analysis_20230927.ipynb)
3. Experiment design (exp_design_flow.png)