Analysis in DAS serves decisions, not vanity metrics. Formatively, while data is still coming in, you’ll track a simple task-tag heatmap, watch emotion trends, and spot friction early. Summatively, you’ll build four crisp views: volume by action, top tags, emotion distribution, and the heatmap. Then you’ll read clusters of quotes (sorted by action, tag, or feeling), extracting 3-5 representative snippets that make patterns tangible. The synthesis culminates in shared design principles phrased as “In context X, do Y to achieve Z,”. We provide a workshop format: silent reading → small groups → plenary alignment → silent closing reflection. This converts data into commitment. You’ll learn light-touch validity moves: triangulate, replicate across teams, and look for convergence across sources. Finally, we cover feedback to contributors.
In DAS, analysis is not something postponed until the end. It is woven into the process from the very beginning of data collection. First through early sense-making while reflections are still arriving, then through systematic compilation that mixes numbers with words, and finally through collegial workshops where all participants are invited to explore the findings together. Thinking of analysis in these three steps makes it easier to see how lived experience can be turned into patterns, and how patterns can grow into shared insights that guide future practice.
Step 1: Formative analysis
As reflections come in, study leaders do not wait passively for the dataset to be complete. Instead, they scan the stream of short reflections, emotion ratings and tag choices for early signals. This early reading carries an ethnographic feel. When study leaders immerse themselves in the participants’ lived accounts, they gain a sense of the everyday textures, moods, and struggles that numbers alone cannot convey. These formative impressions are not definitive findings, but represent a kind of researcher’s diary of observing emerging patterns, noticing surprising turns, and documenting tentative insights through comments to participants. Such sense-making helps prevent the “collect now, think later” trap. It also keeps the inquiry alive, allowing adjustments to tasks or asking participants timely questions while the process is still unfolding.
Step 2: Compiling and mixing
When a cycle of action ends, the dataset is compiled. Each reflection is already linked to a task, emotion rating, tags, and feedback, which makes preparation and export into a spreadsheet or QDA software straightforward. Analysis starts with a “satellite view” – descriptive statistics showing how many reflections each task generated, which tags dominated, and how emotions varied. To then grasp the “street view”, clusters of reflections are read to uncover the lived experiences behind the counts, often sorted by task, tag or emotion, or through multi-key sorting and filtering. A central tool is the task–tag matrix, providing a heat map of where actions consistently triggered certain outcomes, such as confidence, frustration, or collaboration. Hot spots in this matrix guide attention to the most interesting reflections. Other visualisations such as spider diagrams, tag tables, emotion timelines and AI-generated text summaries can also highlight key patterns.
The strength of this step lies in combining big-and-thick data (cf. Weltevrede, 2016, p.17). Enough volume for pattern detection in big data, enough depth for explanation in thick data. Since tasks act as independent variables and tags, reflections, and emotions as dependent ones, DAS is especially well suited for causal mechanism-hunting, that is, exploring how and why certain actions produce particular effects in education or workplace settings (cf Ylikoski, 2019).
Step 3: Collective sense-making
Another defining feature of DAS analysis is that it is not reserved for a single researcher. The study leader prepares summaries, often a slide deck, large A3 printouts or a two-page report, and then invites participants to a collective workshop. There, everyone engages with the material in stages. First individually, then in small groups, then together in plenary. This prevents premature closure and ensures multiple perspectives are voiced. The session ends with a final round of reflections, written during 15 minutes of silence, capturing fresh insights sparked by the discussion.
These collective elements are more than symbolism. They deepen trust, democratise interpretation, and ensure that findings are not imposed from above but co-created. A collegial dimension also helps realise the Humboldtian vision of teachers and students as partners in knowledge creation (cf Shumar and Robinson, 2018). Finally, collective sense-making makes participants see a bigger meaning in reflective practice. It helps them see how their own everyday actions connect to shared challenges, emerging patterns, and new opportunities for improvement.
Examples of data analysis
Examples of formative analysis include studies in social work by Tjulin and Klockmo (2023, p.105) and by Tjulin et al. (2024, p.6.), where they describe formative analysis as a “reciprocal iterative process”, a “back-and-forth dialogue” between empirics and theory, performed “longitudinally and simultaneously with data collection”. Examples of tag based analysis include a study on AI-based conversational agents where tags were seen to represent participants’ “most prominent experiences” (Ericsson et al., 2024, p.759), as well as a study on coffee shop owners where tags were seen as a “means of self-expression” (Morland and Lever, 2024, p.9). Examples of task-tag matrices can be seen in the study by Viebke (2020) and in many of my own studies (Lackéus, 2020a, p.216; Lackéus and Sävetun, 2025, p.226; Lackéus, 2025, p.202). The use of AI for qualitative data analysis has just begun, but has strong potential as illustrated in a recent conference paper (Lackéus, 2025). Many of these analytic steps are supported by the research tool Loopme, where task-tag matrix, spider diagram and AI summaries have been built into the system. Loopme configurations are also shown visually in some articles (Boström et al., 2025, p.272; Ericsson et al., 2024, p.758; Lackéus, 2020a, p.210).
Challenges in data analysis
Analysing DAS data is rich but demanding (Lackéus, 2025). A first challenge is the time needed to compile and interpret hundreds of reflections, tags, and emotion ratings. This can overwhelm study leaders, especially those without research training. A second issue is the complexity of mixed data, requiring a delicate balance between descriptive statistics and lived experiences. Third, since DAS relies on self-reports, concerns about data quality have been raised. Insights may be superficially described, impacted by social desirability bias or hidden due to reluctance to share failures, raising questions about validity and reliability. Fourth, collective analysis brings both benefits and challenges. Workshops can democratize interpretation but also risk uneven contributions if not well structured. Finally there is the challenge of translating into practice. Practitioners struggle with transforming uncovered patterns into actionable improvements, and scholars struggle with how to publish insights generated with a novel methodology.
The end result of analysed DAS studies
The outcomes of DAS analysis extend beyond reports. Studies yield revised design principles and mechanism-based explanations, clarifying how specific actions led to particular outcomes. Formative analysis supports real-time improvements, with tasks adapted during the data collection phase. Collective analysis turns individual accounts into shared insight, with a final round of reflections producing meta-reflections – fresh data sparked by the analytic dialogue itself. AI-generated summaries provide quick overviews for practitioners. Seeing their words visualised also brings validation and ownership. At a scholarly level, DAS studies contribute to theory-building and research methodology, enriching debates on causal mechanisms and clinical action research.