Laddering is an advanced interview technique for performing semi-structured interviews. It is based on the personal construct theory proposed by George Kelly (1955) suggesting that people develop personal constructs about how the world works and, on this basis, make sense of their observations and experiences. Laddering allows researchers to gain insights in the values of interview participants and how such values can be explained in the form of means-end chains. The laddering technique has been successfully applied in business research, e.g. in marketing, services, and management research.

In general, performing laddering interviews is a resource intensive and time-consuming task.  Thus, in our previous work (Rietz & Maedche, 2023) we have implemented a scripted bot that is able to perform laddering interviews in a chat-based dialog. However, although the system served its purpose, participants requested a more “human-like” interviewing experience. Furthermore, laddering should be performed as part of a comprehensive process starting with the bot introducing the research project to the informant and a presentation of a stimuli list. The informant should bel able to select two stimuli she finds personally most important or valuable. Subsequently, the interview bot should probe for ideas, attributes, consequences, and values starting with the first select stimulus, and subsequently, the second selected stimulus. Once all chains are discussed, the interview bot should briefly present the obtained chains to the informant and asks to rank the chain from most to least important. Finally, the interview bot should close the interview and stores the collected dataset in a dedicated database.

Goal of the thesis

Following a design science research process, the main objective of this thesis is to design a prototypical solution that supports configuring and executing laddering interview processes. The prototype should provide the necessary core functionality to setup and configure a laddering interview process (e.g. pre-define stimuli list). Furthermore, it should leverage large language models (LLMs) to perform human-like laddering interviews with a set of informants. Finally, the results should be captured in a structured template. The prototypical solution should also be evaluated empirically (qualitatively and/or quantitatively) with students.

Cooperation Partner

The thesis project is carried out in cooperation with Prof. Tuure Tuunanen and Juuli Lumivalo, Ph.D. from the Faculty of Information Technology, University of Jyväskylä, Finland. Both are experts in the laddering interview technique from a research (Lintula et al., 2018; Tuunen & Peffers, 2018) and teaching perspective (e.g. https://studyguide.jyu.fi/2023/en/courseunit/tjts5907/). There is an opportunity for an exchange visit as part of the final thesis (to be confirmed).

Skills required

• Good programming skills and interest in large language models
• General interest in generative AI, large language models, and/or human-computer interaction
• Strong time management and communication skills
• Fluent in English

Contact

If you are interested and want to apply for this bachelor thesis, please contact Leon Hanschmann (leon.hanschmann∂kit.edu) with a short motivation statement, your CV, and a current transcript of records. Feel free to reach out beforehand if you have any questions.

References

Lintula, J., Tuunanen, T., Salo, M., & Myers, M. D. (2018). When Value Co-Creation Turns to Co-Destruction: Users' Experiences of Augmented Reality Mobile Games. In ICIS 2018 : Proceedings the 39th International Conference on Information Systems (pp. 1-17). Association for Information Systems (AIS).

Tuunanen, T. & Peffers, K. (2018). "Population targeted requirements acquisition," European Journal of Information Systems (EJIS), 27(6), pages 686-711.

Rietz, T., & Maedche, A. (2013). Ladderbot—A conversational agent for human-like online laddering interviews. International Journal of Human-Computer Studies (IJHCS), Volume 171,  https://doi.org/10.1016/j.ijhcs.2022.102969.