The Setup
A doctoral dissertation on teacher burnout had a problem that many mixed-methods researchers know well: two datasets, two software tools, and no clean way to connect them.
The qualitative data was in a coding tool — 42 teacher interviews, 15 years of teaching experience, underfunded districts. The quantitative data was in Excel — student test scores, funding levels, teacher absences, out-of-pocket spending.
The dissertation committee asked for an "integrated analysis" — not parallel findings sitting next to each other, but genuine integration showing how the qualitative and quantitative inform each other.
NVivo could code the interviews. SPSS could run the correlations. Neither could do both. This is what FableSense AI revealed in a single platform.
The Data
What we had:
- 42 teacher interview transcripts (Maria, James, Robert, Sarah, and others)
teacher_student_outcomes.csv: school-level data including test scores vs. district average, teacher absences, out-of-pocket spending, turnover intent
| Theme | Avg Score vs District |
|---|---|
| Lack of Resources | -14.8% |
| Physical Environment | -16.5% |
| Emotional Exhaustion | -12.3% |
| Administrative Burden | -8.1% |
| Healthy Boundaries | +6.7% |
The Investigation
Qualitative Coding
The code framework came directly from Maslach's burnout inventory and the literature on teacher retention:
- Emotional Exhaustion — depleted energy, cannot give more
- Lack of Resources — out-of-pocket spending, missing materials, overcrowding
- Administrative Burden — paperwork overload, compliance demands
- Physical Environment — building conditions, broken equipment
- Healthy Boundaries — self-protection strategies, sustainable practices (positive deviance)
Maria's interview — 15 years in an underfunded district — was the baseline case:
"I have 32 kids in my class. Supposed to be 24 max. No aide."
"I've spent $800 of my own money this year on supplies."
"By 3pm I'm so drained I can barely help with after-school tutoring."
Maria carried three codes simultaneously: Lack of Resources, Emotional Exhaustion, and a financial metric that would matter in the quantitative analysis — $800 in out-of-pocket spending.
James's interview added the administrative dimension:
"The paperwork is insane."
"I feel like a bureaucrat, not a teacher."
James's frustration was real. But it would prove less predictive of student outcomes than the resource and exhaustion codes. That distinction only became visible in the joint display.
Robert's interview surfaced the physical environment:
"The building is falling apart. We have mould in two classrooms."
Sarah's interview was the anomaly — and ultimately the most important finding:
"I've learned to set boundaries. I don't check email after 5pm."
"I'm actually a better teacher now. My test scores have improved."
Same district. Similar funding constraints. Different outcome. The question was whether the quantitative data supported the observation.
The Joint Display: Where It Came Together
The Network View: Finding the Causal Structure
The Integration Matrix showed correlations. The dissertation committee wanted to know whether Lack of Resources was directly affecting scores — or working through emotional exhaustion as a mediating variable.
The Network Visualization provided the structure.
Key edges and their weights:
| Connection | Weight | Direction |
|---|---|---|
| Lack of Resources → Emotional Exhaustion | 0.82 | Positive |
| Emotional Exhaustion → score_vs_district | -0.64 | Negative |
| Lack of Resources → score_vs_district | -0.71 | Negative (direct) |
| Healthy Boundaries → score_vs_district | +0.54 | Positive |
The thickest edge in the entire network: Lack of Resources → Emotional Exhaustion at 0.82.
This suggested a mediation pathway: resource deprivation increases emotional exhaustion, which degrades teaching quality, which lowers test scores. But resource deprivation also has a direct path to outcomes — separate from the exhaustion pathway.
Healthy Boundaries sat apart. It had no strong connection to the burnout cluster — no edge to Lack of Resources, no edge to Emotional Exhaustion. It connected positively to test scores through a completely different mechanism.
These weren't burned-out teachers who had adapted. They were teachers who had prevented burnout through deliberate boundary-setting.
The Integration Matrix: Confirming the Rankings
Across all 42 interviews mapped against the dataset:
| Theme | score_vs_district | turnover_intent | teacher_absences |
|---|---|---|---|
| Lack of Resources | -14.8% avg | Strong positive | Moderate positive |
| Physical Environment | -16.5% avg | Strong positive | Strong positive |
| Emotional Exhaustion | -12.3% avg | Strong positive | Strong positive |
| Administrative Burden | -8.1% avg | Moderate positive | Weak positive |
| Healthy Boundaries | +6.7% avg | Strong negative | Strong negative |
Administrative Burden produced real frustration in the interviews. But its correlation with student outcomes was weaker than the other burnout themes. Paperwork is demoralising — it's not, on its own, as predictive of poor teaching outcomes as resource deprivation or physical environment.
That distinction — which qualitative coding alone couldn't establish — came from the integration.
The Recommendation
The dissertation's Chapter 4 finding challenged the individual-level framing of teacher burnout.
The problem isn't primarily about wellness programs. It's about funding.
Emotional exhaustion matters. But in most cases it's downstream of resource deprivation. Treating burnout as a psychological problem without addressing its material causes treats the symptom, not the cause.
However — and this is the nuance the integration revealed — Healthy Boundaries teachers showed that systemic constraints don't determine outcomes uniformly.
Sarah's students scored 6.7% above district average despite the same funding environment as Maria's. That gap is the size of a policy intervention, delivered through individual practice.
The both/and recommendation:
1. Address systemic resource deprivation directly Class size caps, supply budgets, maintenance funding. Physical Environment and Lack of Resources are the strongest predictors of student outcomes. These are addressable through policy, not professional development.
2. Teach boundary-setting strategies alongside systemic advocacy Identify the Sarah patterns — practices that preserve teaching effectiveness under constraint. These become a transferable intervention while systemic changes take years to implement.
3. Separate administrative burden from resource deprivation in future research They're often treated as the same problem. The integration shows they're not. Paperwork reform helps. It won't close the test score gap that resource deprivation creates.
The Outcome
The dissertation committee got their integrated analysis. Not parallel findings sitting side by side — an actual structural model showing how burnout factors connect to each other and to student outcomes.
The Integration Matrix went into Chapter 4 as the findings table. The Network Visualization became Figure 4.2 in the discussion chapter. Every correlation was backed by traceable quotes from the interview transcripts.
The qualitative explained the human cost of underfunding. The quantitative proved how consistently it predicts student outcomes. The integration showed which interventions would move which metrics.
That's a dissertation finding. It's also a policy argument.