AI in interpretative qualitative research
Quick access
methodological paper
Hybrid Interpretation of Text-Based Data Using Dialogically Integrated LLMs (Krähnke, U., Pehl, T., & Dresing, T. (2025))
Practical guidance
Introduction to hybrid interpretation with three LLMs. A practical step-by-step guide
Webinars
See hybrid interpretation live or help shape it in a research workshop? Visit our suitable webinars/courses!
Prompts & example
The prompt templates used and full documentation of a hybrid interpretation of a text segment can be found here:
GDPR-compliant hybrid interpretation
A workshop report on local interpretation with four LLMs on a MacBook Pro with Gemma 3, Qwen 3, Mistral 3.1 and Llama 3.3
Serendipity prompting
The methodized curiosity: Using LLMs as a meeutic impulse generator instead of an answer machine in the "phase 0" of the analysis.
Hybrid interpretation is an elaborate method for integrating AI into interpretative qualitative research (Krähnke, Dresing, Pehl 2025). It enables an insight-evoking dialog between several AIs (Large Language Models / LLMs) and the researcher, with the aim of a profound and comprehensible interpretation of the text. This article describes the practical implementation of this procedure, which can be implemented largely free of charge via a browser.
Previously widespread approaches to the use of AI in qualitative social research, such as the use of ChatGPT, often only deliver superficial results (keyword “make a summary”). In addition, strategies of elaborate input to the LLM (prompt engineering) are pursued, which, however, require a high level of development expertise. Nevertheless, automatically generated interpretations by individual prompts are usually not particularly differentiated.
With the hybrid interpretation described here, we have found that we can carry out high-quality and differentiated analyses with AI. We ourselves are always surprised and delighted by the depth and sophistication of the interpretation suggestions. This awakens in us the desire to analyze and interpret again and again.
Here we describe the features of the approach and provide instructions on how anyone interested can try it out for themselves – including using tools that are available free of charge.
6 reasons that make this approach so special
The approach reduces the need for elaborate prompt design and instead enables a natural discussion language when dealing with AI. This makes the method more accessible and less technically demanding.
Three different LLMs (currently ChatGPT, Claude and Gemini) are integrated into the research process and moderated by the researcher. Studies show that the interlinking of different LLMs leads to an improvement in the quality of the output of each individual LLM. This significantly increases the overall quality of the analysis. This variance increases the diversity of perspectives due to the different “bias” of the LLMs involved. The confrontation with different points of view encourages more differentiated answers and mutual reference, which increases the diversity of perspectives.
The researcher takes on an orchestrating role as a moderator and remains actively involved in the interpretation process. She critically examines the interpretations offered, asks specific questions, gives instructions and reflects on the contributions of the AI models. This active control ensures that the analysis is targeted and in line with the research interest.
The analysis takes place in several rounds, with each round imitating the style of a lively group discussion. The AI models are confronted with the answers of the other LLMs and the assessments of the researcher. In the process, the analyses are deepened, argued in a more differentiated way and their interpretations are examined. This iterative approach results in an increasingly refined and multi-layered interpretation of the research material.
The complete documentation of the process addresses the often criticized “black box” problem of AI usage. This enables an intersubjective comprehensibility of the interpretation and makes the individual work performance of the researchers in the context of the orchestration of the work process clear. The researcher’s examination of the various interpretative approaches is documented transparently. It can be retraced in terms of plausibility and quality, which increases the validity of the research.
Hybrid interpretation allows students to gain their first practical experience of qualitative interpretation. The integrated LLMs act as a kind of sparring partner for the interpretation work.
Hybrid interpretation helps students to get to know, apply and critically evaluate different perspectives on text interpretation. You do not simply receive a finished result or automatic coding, but rather material for your own examination of the text.
By actively engaging with different approaches to interpretation, they are encouraged to develop their own, well-founded positions. This process not only promotes analytical thinking, but also the ability to synthesize and evaluate complex interpretations.
Podcast 'Methods:Suitcase' from 18.02.2025
You are currently viewing a placeholder content from Spotify. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationIn this episode, Thorsten Dresing and Anna-Barbara Heindl discuss how generative AI works in principle and clarify what impact this has on the use scenario of AI in reconstructive research. Thorsten and his colleagues suggest using Large Language Models such as ChatGPT, Claude and Gemini to get inspiration and irritation for different interpretation possibilities in parallel to their own interpretation work – analogous to human interpretation groups.
First of all, the user needs (partly) free accounts with all three LLMs and chooses a short one. A simple, open starting question is formulated for the LLMs to initiate the analysis process.
In the first round of analysis, the three LLMs describe the text from different perspectives. The researcher then reflects on the statements and derives the next steps.
This is followed by several iterative discussion rounds in which the researcher deepens the interpretations and discusses them with the LLMs. This process is moderated and controlled by interim comments and reflections from the user until the user determines the conclusion of the analysis.
The researcher identifies key aspects and develops their own interpretative perspective. Finally, a conclusion is drawn up that summarizes the interpretation developed.
To ensure traceability, the entire course of the discussion is documented and commented on in Word or an f4 project.
During our test runs, we had by far the best experience with the following three LLMs in terms of the appealing quality of the interpretation suggestions. To use these, each person needs their own account, initially free of charge. You can simply click on the links and register. (as at Dec 2024)
The full use of all three LLMs can therefore be associated with costs of up to 60 euros per month. Fortunately, this is not necessary for a test run as part of a course or for moderate use.
To work completely free of charge, use Claude and ChatGPT in the limited but free versions and Gemini in the trial month. This means that you can only use Gemini for a maximum of 4 weeks and Claude only to a limited extent during this time.
Unfortunately, the free version of Claude limits the amount of text that can be edited and output within 5 hours. This is achieved with just a few analysis runs, as Claude not only counts the output, but also the input word quantity. You must organize yourself in such a way that you carry out one or two analysis runs within 5-hour time slots. Then cancel Google’s Gemini again in good time. ChatGPT also has an upper volume limit, but this is not reached so quickly.
Shorter analysis text sequences and shorter LLM responses enable more iterations. Optimization by restricting the response length (in the start prompt) significantly reduces the total amount of text. However, longer answers are often better argued. This needs to be weighed up.
The implementation of this working proposal is not yet GDPR-compliant, as all data is transferred to the respective providers outside the scope of the GDPR. Therefore, do not use any material that is critical under data protection law and contains personal data. Instead, use extracts from publicly available data, simulated data or data for which you have explicit written consent for this use.
Here you can download the current version (11.09.2024) of the start prompts described above, which can simply be copied and pasted.
You will also find an example of hybrid interpretation in the project file. Here you can follow an entire discussion to analyze a fictitious interview passage.
Citation of this article
Dresing, T., Pehl, T., Krähnke, U. (2025). Introduction to hybrid interpretation with 3 LLMs: A practice-oriented approach Step-by-step instructions for a text segment. https://www.audiotranskription.de/hybride-interpretationsgruppe
This blog post is based on: Krähnke, U., Pehl, T., & Dresing, T. (2025). Hybrid interpretation of text-based data with dialogically integrated LLMs:
On the use of generative AI in qualitative research… https://nbn-resolving. org/urn:nbn:de:0168-ssoar-99389-7
further literature:
Lieder, F. R., & Schäffer, B. (2023). Teaching and learning reconstructive research methods with generative language models in hybrid research workshops? Journal of Psychology, 31(2), 131-154. https://doi.org/10.30820/0942-2285-2023-2-131.
(This article structures and explains specific prompts for qualitative research and mentions the idea of a “hybrid research workshop”, which we have taken up and expanded with our approach of hybrid interpretation).
Example
You are currently viewing a placeholder content from Vimeo. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationInterviewer: To start with: How long exactly have you been with move2035 here in Marburg?
Participants: The first meeting took place in September. That was in ’23. And they’ve been very active ever since.
Interviewer: And how would you describe your work?
Participants: My job is actually to coordinate this entire meeting. And the attempt to move it forward.
We have often discussed this fictional passage in our classes, initially without any specific methodology or background knowledge. At first, people usually notice details such as coordination, meetings, time, and location, as well as words like “actually” and “attempt,” which are often interpreted as a critical tone. In most cases, the general view is that “there is nothing more to be found in that section of the text.” While a content-analytical perspective might be satisfied with this, it is precisely the hermeneutic interpretation that offers nuanced insights. And this can be achieved through the approach of hybrid interpretation. Below, we summarize the insights gained from each round. For those who would like to follow along in more detail: the full transcript of the conversation is included in the f4 project below.
The LLMs identify several significant linguistic patterns: The participant’s use of the personal pronoun “she” instead of “we” is interpreted as a possible indicator of a distant attitude toward the group. The precise time given for the start of the project suggests that this moment holds special emotional or organizational significance. This self-characterization as a coordinator reflects a mediating role. The phrase “try to move it forward” could indicate existing obstacles or resistance in the process.
These linguistic observations led us to critically reflect on the participant’s use of the modal particle “eigentlich.” This could indicate a potential discrepancy between the formally assigned role and the function actually performed.
The LLMs identify a characteristic ambivalence in the participant’s positioning, which manifests itself in the oscillation between the roles of insider and outsider—a phenomenon classified as a “professional dilemma.” The use of the modal particle “actually” could indicate a latent range of tasks that have not been explicitly stated. The position of coordinator may entail limited decision-making authority. A specific LLM emphasizes the significance of the lexeme “attempt,” which could indicate shared or externalized process responsibility.
These observations lead to the analytical question of the interdependence between this complex (self-)positioning and the actor’s motivational structures.
The LLMs develop a nuanced interpretation of distancing as a potential strategy for maintaining professional objectivity. This perspective allows for a methodological duality: it simultaneously facilitates an empathetic understanding of group dynamics and the exercise of leadership functions. In this theoretical conceptualization, distance can be reinterpreted as a specific form of engagement. The persistence of commitment evident in the material, despite adverse conditions, points to intrinsic motivational structures. Within this analytical framework, the use of the lexeme “attempt” can be interpreted as a reflection on the limitations of one’s own scope for action.
Within the context of move2035, participants navigate a complex web of multiple role identities, characterized by the tension between professional detachment and personal involvement. The preference for using the personal pronoun “she” rather than “we” can be interpreted as an intentional distancing strategy to maintain professional objectivity. The modal particle “actually” indicates a pronounced process of reflection regarding one’s own conception of one’s role and may point to unexpressed aspects of the configuration of the activity. The phrase “try to move it forward” simultaneously reflects intrinsic motivational structures and a keen awareness of potential constraints in the sphere of action. The precise temporal framing of the project’s inception suggests that the undertaking holds considerable personal significance, while the detached language used points to a high level of reflective competence.
At first, we only recognized superficial details: coordination, meetings, times, and key words. The hybrid interpretation group allowed us to delve deeper and understand the participant’s complex role identity. We discovered the tension between closeness and distance, the challenges of his position, and his motivation. This iterative process allowed us to paint a multifaceted picture from a simple statement—one that would have remained hidden without this method.
We’re curious to hear about your experiences with our proposal. How do you find the discussion process, and how would you rate the quality of the responses and the value added by this approach? Any feedback helps us improve and further develop this approach. We would therefore greatly appreciate it if you could share your thoughts with us via email or phone.
This podcast was automatically generated from this blog post using Google Notebook LM to make the content accessible in a conversational format. Entertaining and surprisingly well-done, automatically generated presentation. Starting around the 12-minute mark, though, it gets ridiculous and has nothing to do with our proposal anymore 🙂
