What AI Reveals in STEM Lesson Plan Analysis

According to the Pre-service mathematics teachers’ integration of design thinking in STEM lesson plans study published in PLoS One on November 2, 2026, pre-service mathematics teachers included some design thinking phases strongly but integrated the full process unevenly. AI qualitative analysis can help education researchers compare lesson-plan content with questionnaire themes, surface missing phases, and connect frequency patterns to participants’ explanations.

• The PLoS One study analyzed 24 collaboratively developed lesson plans and questionnaire responses from 55 senior pre-service mathematics teachers during a 15-week course in the 2021 to 2022 academic year.
• The study reported that Define appeared in 95.8% of lesson plans, Prototype in 91.7%, and Ideate in 83.3%, while Empathy appeared in only 20.8% and Test in 54.2%.
• The PLoS One researchers found that only 3 of 24 lesson plans included all five design thinking phases, while 1 lesson plan included no design thinking phase.
• The study reported 90% initial inter-rater agreement for a randomly selected 25% sample of questionnaire coding and 93.1% initial agreement for independent lesson-plan evaluation.
• The study’s participants described alignment with other disciplines, learning objectives, materials, assessment, time, and curriculum coverage as major planning challenges.

The PLoS One study found that pre-service mathematics teachers treated design thinking as valuable but often implemented it as a partial, product-oriented process. The study used a qualitative case study design involving 55 senior participants at a state university in Türkiye, 24 group lesson plans, and individually completed open-ended questionnaires.

The PLoS One researchers analyzed lesson plans with a STEM integration rubric and coded the presence and quality of five design thinking phases: Empathy, Define, Ideate, Prototype, and Test. The researchers also used an inductive four-stage thematic process for questionnaire data: familiarization, initial coding, category development, and theme construction.

The results suggest that the most common lesson-plan structure moved from defining a problem to generating an idea and building a prototype. The PLoS One study interpreted the low representation of Empathy as evidence that user-centered inquiry was frequently bypassed, while the lower representation of Test suggested that prototypes were often treated as final products rather than solutions to evaluate and improve.

The participants’ own words explain the implementation gap. PMT 25 said, "The most challenging part was aligning the design with other disciplines and finding suitable learning outcomes." PMT 16 added, "I think design is necessary, but it should not take too much time because there is a program that needs to be covered." The PLoS One study attributed these comments to challenges involving interdisciplinary alignment and curriculum time.

The study also recorded positive expectations. PMT 22 said, "We expect students to think creatively, be curious, use imagination, and enhance communication and cooperation skills." The PLoS One researchers connected these responses to participants’ belief that design thinking could support engagement, creativity, mathematical modeling, and real-world learning.

Education researchers should analyze design thinking integration as both a structural and substantive phenomenon. The PLoS One study distinguished between a phase being mentioned and a phase being meaningfully implemented through mathematical reasoning, quantitative analysis, or evidence-based justification.

Teacher educators should examine relationships between document evidence and participant explanations. The PLoS One study found that participants selected topics based on real-life problems and student learning, with the codes "problems from own experience" appearing 34 times and "student learning" appearing 19 times. Those motivations did not always translate into explicit user research or iterative testing in the lesson plans.

Qualitative researchers should preserve both frequency and context. A frequency such as 20.8% for Empathy identifies an uneven pattern, but participant quotations about materials, curriculum time, and assessment help explain why the pattern occurred.

AI-enabled qualitative research is useful when the dataset contains multiple formats and analytical levels. A researcher can use AI to locate references to Empathy, Test, assessment, materials, or mathematical modeling across lesson plans and questionnaires, then manually inspect the excerpts that support each theme.

AI analysis should remain subordinate to the study design and codebook. The PLoS One researchers used two researchers, explicit coding procedures, inter-rater checks, reconciliation, and triangulation, so an AI workflow should add traceability rather than remove human review.

The November 2, 2026 PLoS One study shows why qualitative analysis of STEM lesson plans must examine both design thinking phase frequency and the depth of mathematical integration. The study’s 24 lesson plans and 55 participant responses reveal a clear pattern: defining, ideating, and prototyping were common, while empathy, testing, and sustained iteration were less consistently represented.

AI-enabled qualitative research can make this type of mixed-data analysis more efficient by connecting documents, codes, frequencies, quotations, and cross-segment comparisons. Researchers should use AI as a traceable research assistant, retain source evidence, and apply human review to interpretation and reporting.

Evidano helps research teams analyze qualitative documents and open-ended responses in one workspace. Try Evidano for free to organize your next education research project.