Research Data Package
=====================

Title: Human-AI Collaboration for Decision-Making in Agile Sprint Planning: A Practitioner-Informed Conceptual Model

Authors: Ungarala Sai Krishna Yashwanth, Mihir Singh Thakur

Institution: Blekinge Institute of Technology, Faculty of Computing
Programme: MSc Software Engineering
Course: PA2534
Supervisor: Henry Edison
Year: 2026

---

Contents of this package
--------------------------

1. interview_guide.pdf
   The semi-structured interview guide used for all nine interviews.
   Contains three parts: background questions, two vignette scenarios
   (Vignette A for anchoring bias, Vignette B for algorithm aversion),
   and trust calibration questions.

2. codebook.pdf
   The anonymised codebook containing all 12 codes with operational
   definitions. Organised into three groups: Group A (anchoring codes),
   Group B (algorithm aversion codes), Group C (trust calibration codes).

3. consent_information_sheet.pdf
   The information sheet shared with all participants prior to interviews,
   describing the study purpose, anonymisation commitments, and
   participant rights.

4. README.txt
   This file.

---

Note on transcripts
--------------------

Full interview transcripts are NOT included in this package due to
participant confidentiality commitments made at the time of data
collection. Transcripts are available to the thesis supervisor
(henry.edison@bth.se) and examiner upon request.

---

Study overview
--------------

This study investigated how anchoring bias and algorithm aversion
affect decision-making in AI-assisted Agile sprint planning, and how
practitioners calibrate trust in AI tools. Nine semi-structured
interviews were conducted with Agile practitioners across Scrum Master,
Tech Lead, Engineering Manager, and Team Lead roles. Two vignette
scenarios were embedded in each interview to surface bias mechanisms
without relying on self-report. Thematic analysis following Braun and
Clarke's six-phase framework was applied to all transcripts, structured
around the four meta-dimensions of the HACO taxonomy (Dubey et al., 2020).

The study produced a three-component conceptual model for Human-AI
collaboration in Agile sprint planning: role zone definitions,
interaction principles, and a risk identification checklist.

---

License: Creative Commons Attribution 4.0 International (CC BY 4.0)
