Capturing the students’ mental models using ZMET
Photo by Towfiqu barbhuiya on Unsplash
By Ragnhild Nordeng Fauchald, Postdoc, Institutt for industriell økonomi og teknologiledelse, NTNU
How much insight do you usually gain from student surveys or meetings about their learning experiences? Traditional assessment methods often provide limited depth, especially in experience-based entrepreneurship education, where linking teaching approaches to learning outcomes is challenging. With ZMET, we’ve found a way to uncover deeper, richer insights into how students perceive our teaching and their learning.
- Duration:
ZMET interview: 45-90 minutes
- Focus:
Utilizing the ZMET model to gain insights into student’s learning experiences
- Activity:
Team, interview
- Keywords:
ZMET, learning experience, mental model
About the exercise
ZMET provides educators and program managers with new insights important for alteration course or program content, structures of courses and programs or teaching approaches. In this PDW, we introduce ZMET with a practice-oriented approach aiming to give you a tool from assessing students learning. This booklet serves as a user instruction to each step of the process. We will specifically pay attention to how the mental maps are created.
Learning objectives
- Understand ZMET – ZMET can uncover deep insights into student learning.
- Apply a Practice-Oriented Approach – Gain hands-on experience using ZMET to assess learning in educational settings.
- Analyze Mental Maps – Understand how students’ mental maps are created and interpreted to reveal their perceptions of teaching and learning.
- Enhance Course and Program Development – Learn how to use ZMET insights to refine course content, program structures, and teaching methods.
Background and Description
Developed by Zaltman and Coulter (1995), ZMET helps uncover deep insights into thoughts and emotions through visual metaphors. Originally used in marketing, it is based on cognitive neuroscience research suggesting that people think in images rather than words (Zaltman, 1997). In a ZMET interview, participants bring illustrations representing their thoughts on a topic, which serve as a basis for discussion. This process captures both individual mental images and their interconnections, forming a mental model that includes beliefs, emotions, values, and sensory experiences (Christensen & Olson, 2002).
ZMET is structured around means-end chain analysis (Reynolds & Gutman, 1988), where “means” are actions or experiences (e.g., having skilled educators), and “ends” are their effects (e.g., increased motivation and learning). By mapping these relationships, ZMET reveals how different elements of an educational experience contribute to learning outcomes.
ZMET in Entrepreneurship Education
ZMET is particularly useful for assessing experience-based entrepreneurship education, where learning processes are complex. It has been used to explore differences in student learning based on startup involvement (Haneberg & Aadland, 2020), learning trajectories in extracurricular activities (Haneberg & Aaboen, 2022), and perceptions of applying for seed grants (Fauchald, Aaboen & Haneberg, 2023). Insights from these studies have informed educational improvements. ZMET has also been applied in other educational contexts, such as online learning research (Shearer et al., 2020).
Practical Information
– Interview Duration: 45–90 minutes.
– Team Requirement: At least two researchers for interviews and analysis to ensure rigor.
– Process: Requires an intensive approach, analyzing data within a short period.
– Conducting the interview: Maintain professionalism and ask open-ended, non-leading questions.
Theoretical Foundations
Zaltman and Coulter (1995) and Zaltman (1997) describe the method, its purposes and values from a marketing and advertising perspective:
- Zaltman, G. and Coulter, R.H. (1995), “Seeing the voice of the consumer: metaphor-based advertising research”, Journal of Advertising Research, Vol. 35 No. 4, pp. 35-52.
- Zaltman, G. (1997), “Rethinking market research: putting people back in”, Journal of Marketing Research, Vol. 34 No. 4, pp. 424-437.
Christensen and Olson (2002) explain ZMET in detail with references to the theories and logics behind the steps of ZMET:
- Christensen, G. L., & Olson, J. C. (2002). Mapping consumers’ mental models with ZMET. Psychology & Marketing, 19(6), 477-501.
For use of ZMET in entrepreneurship education research, see:
- Fauchald, R. N., Aaboen, L., & Haneberg, D. H. (2023). Pre-seed grant as an enabler of learning. International Journal of Entrepreneurial Behavior & Research, 29(7), 1698-1719.
- Haneberg, D. H., & Aaboen, L. (2022). Entrepreneurial learning behaviour of community insiders. International Journal of Entrepreneurial Behavior & Research, 28(2), 306-324.
- Haneberg, D. H., & Aadland, T. (2020). Learning from venture creation in higher education. Industry and Higher Education, 34(3), 121-137.
Especially the Fauchald et al. (2023) article gives an in-depth explanation of ZMET with illustrations of the analysis process.
Using ZMET: A step-by-step instruction
ZMET follows inductive, grounded-theory principles (Corbin & Strauss, 1990), starting with an open-ended inquiry into a chosen objective—such as a learning activity, course, program, or specific process. The key question is: “What are your thoughts and feelings about [the objective]?”
Once you’ve chosen an objective to explore, begin preparations. Recruit 6–15 informants (Zaltman & Coulter, 1995) and instruct them to bring five illustrations to the interview, providing at least a week’s notice for reflection. Send clear, concise instructions, ideally with an example, to ensure understanding.
Here is an example on the attached instruction for choice of photos for the research question “What are your thoughts and feelings about the ice-cream parlor”:
Picture no. 1
Related thoughts and feelings to picture 1: “… Because some of my greatest memories from childhood are when me and my brother bought ice cream on a sunny, warm days…”.
Picture no. 2
Related thoughts and feelings to picture 2: “…When we grew older, we spent as many nights as possible to watch the sunset from the beach where the ice-cream parlor is located. The sunsets there are the most beautiful”.
Wrong kind of photo:
“This is an ice-cream parlor”.
As interviewer, you start the interview with the overarching research question “What are your thoughts and feelings about “the objective”?”. The informant uses the illustrations to explain their thoughts and feelings. During the process, you only ask open-ended follow-up questions as the interviewees presented their images and commented on them. This for the purpose of getting as deep insight as possible into their views. Do not ask leading questions!
Here is an example to picture 2 above:
Informant: “…When we grew older, we spent as many nights as possible to watch the sunset from the beach where the ice-cream parlor is located. The sunsets there are the most beautiful”.
Researcher: “How come?”
Informant: “I have travelled a lot in my life. But I always dream of the sunsets beside Vesterhavet”. I
get goosebumps when I speak about it.
Researcher: “Really?”
Informant: “Yes, I don’t know what it is. But, you know… we all have some childhood memories that
stuck in us, don’t we? Maybe this is one of those for me.
Researcher: “How is that feeling?”
…
The interviews must be recorded.
After the interviews, transcribe them, noting emotional cues like laughter or tone. Then, begin coding in NVivo using a grounded theory approach (Corbin & Strauss, 1990). Researchers should discuss any disagreements on code terms. Open coding is extensive—Fauchald et al. (2023) identified around 600 codes from five hours of interviews. NVivo helps manage repeated or similar responses by linking them to existing codes. Below is an example of how NVivo visualizes code references.
All information in the transcripts are coded following Grounded Theory Approach. That makes a lot of yellow marking in NVivo!
Overview in NVivo of References and in how many Files (interviews) the references are found.
Following the grounded theory approach (Corbin and Strauss, 1990) the codes are then combined into subcategories based on their similarities. This process is done manually with the use of post-it notes in order to have a visual overview of all codes when grouping them. Through this process, the researchers discuss the categorization. Researchers should categorize data simultaneously and soon after coding to remember the interviews well and retain interview context.
596 codes ready to be categorized!
Five hours later, the researchers had discussed and sorted 596 codes into 54 categories.
The next step is to sort all codes you have in NVivo into categories. Merge the existing codes into new codes representing the categories. Visually, you now have as many codes as it is categories in the transcripts, see figure 1. Open one and one interview (File). Then click on “coding stripes” and select “All”.
Now, a new window will open beside your transcript and you will find where in the interview you have the different categories. The different colors represent one category each.
This is the starting point for mapping the mental models. You now print the transcripts with the coding stripes besides, page by page to identify the paired-construct relationships (Zaltman and Coulter, 1995). See figure 2.
You sit together and use a pen to draw arrows between the constructs (categories). See figure 3. Some pages can have only a few arrows, while others are more complex. Remember that the arrows also go between the pages you have printed!
An Excel-sheet with all categories listed both vertically and horizontally is then created. You sit down together and one person reads the connections and one plot them into the Excel-sheet: “41 to 36”, means a “0” in the square of vertical 41 and horizontal 36 marked in red, see figure 4.
Remember to create one Excel sheet for each interview. You change the numbers based on how many times the connections have appeared for each of the interviews. In the orange circle, you see a connection that appeared three times.
Figure 1: How you visualize the categories in the transcripts.
Figure 2: Using Coding Stripes, you see where the different categories of codes are located in your transcripts.
Figure 3.
Figure 4.
Create a consensus map using constructs and relationships that exceed your chosen threshold, typically two or more interviews (see Zaltman & Coulter, 1995; Christensen & Olson, 2002 for threshold guidelines). Highlight strong constructs—those appearing in three or more interviews—using bold or colors.
Here is an example of a map where those who appeared in three or more interviews are in bold, see figure 5.
In the example (Figure 5), construct 41 is an originator with no incoming links, leading to construct 36, a connector, which then links to destination construct 23. Some constructs are excluded if they don’t meet the threshold.
Transferring findings from Excel to a readable map requires patience and creativity. Position constructs carefully to avoid cluttered connections. For more examples, see Haneberg & Aadland or Haneberg & Aaboen.
Figure 5.
Good luck using ZMET!
And please ask if you have any questions or inputs to this guide.
Ragnhild Nordeng Fauchald
Ragnhild.n.fauchald@ntnu.no