HVAC Training Academy - Volume 21 | 1
HVAC TRAINING ACADEMY
Volume 21 - Evidence Reconciliation Intensive
Fact-vs-inference discipline, contradiction handling, probability updates and multi-fault reasoning
Remediation objective: This volume targets the reasoning weaknesses identified in Charlie's certification test.
Charlie must preserve facts, compare competing hypotheses, update confidence when evidence changes, and
avoid converting a plausible explanation into a confirmed root cause.
Safety: Educational reasoning only. Field electrical, refrigerant, combustion, pressure and mechanical work
requires qualified personnel, proper instruments, manufacturer procedures, PPE and applicable codes.

HVAC Training Academy - Volume 21 | 2

1. Fact, Inference, Hypothesis, Conclusion
A fact is directly supplied or reliably measured. An inference is a reasoned implication. A hypothesis is a candidate
explanation. A conclusion is justified only after the evidence sufficiently discriminates among candidates.
Never promote a hypothesis to fact because it sounds familiar.

2. Evidence Reconciliation
List every important fact and ask whether each candidate explains it, conflicts with it, or is neutral. The best
hypothesis is the one that survives the entire evidence set, not merely one symptom.
Use the complete pattern.

3. Necessary vs. Sufficient Evidence
A finding may be compatible with a diagnosis without being sufficient to prove it. Low suction can occur with
undercharge, restriction, or low evaporator load.
Compatible does not mean diagnostic.

4. Confirmed Problem vs. Root Cause
Finding a real problem proves that problem exists; it does not automatically prove it caused every symptom or that
no second fault exists.
One confirmed fault can coexist with another.

5. Evidence Updates Probability
New evidence should strengthen some hypotheses and weaken others. It does not need to prove or eliminate a
hypothesis completely.
Update confidence incrementally.

6. Contradictory Evidence
If a candidate predicts one pattern but reliable evidence shows the opposite, the contradiction must be explained
or the candidate downgraded.
Never hide a contradiction.

7. Positive and Negative Evidence
The presence of a finding can support a candidate; the verified absence of an expected finding can weaken it.
Absence matters only when the observation was capable of detecting it.

8. Measurement Reliability
Before reconciling evidence, verify that measurements correspond to the correct location, operating state,
refrigerant/equipment, and instrument method.
Bad evidence should not dominate reasoning.

9. Causal Chain

HVAC Training Academy - Volume 21 | 3
Separate upstream cause, intermediate effect, and downstream symptom. Example: restricted airflow can reduce
evaporator load, which changes coil conditions and suction behavior.
Do not confuse effect with cause.

10. Multiple Faults
HVAC systems can have more than one fault. Correcting one problem and observing a remaining symptom is
evidence that the first problem was not the complete explanation.
Do not force single-fault thinking.

11. Non-Unique Cases
If two candidates still explain all reliable facts and no discriminating evidence is supplied, the correct conclusion is
non-unique.
Uncertainty can be the correct answer.

12. Discriminating Evidence
Choose the next measurement that is expected to differ between the leading candidates.
The best next test changes the ranking.

13. Hypothesis Table
For difficult cases, build columns for candidate, supporting facts, conflicting facts, missing evidence, and next test.
Structured comparison prevents cherry-picking.

14. Reconciliation Before Recommendation
Do not recommend a repair until the evidence has been reconciled against plausible alternatives and safety
requirements.
Parts should follow diagnosis.

15. Charlie Evidence Rule
State what is confirmed, what is strongly supported, what remains possible, what is contradicted, and what
evidence is needed next.
Calibrated certainty is mandatory.

HVAC Training Academy - Volume 21 | 4
Progressive Remediation Drills

# Prompt Expected reasoning
1 Low suction is measured. Is undercharge confirmed? No. It is compatible with multiple causes.
2 A restricted filter is confirmed. Does that prove it is the only fault? No.
3 After the filter is corrected, low suction remains. What changes? The filter was a real fault but not a complete explanation; other hypotheses gain importance.
4 Two diagnoses explain all facts equally. What is the correct conclusion? The evidence is insufficient for a unique diagnosis.
5 A new fact contradicts the leading hypothesis. What should happen? Recheck the fact/method and downgrade or revise the hypothesis if reliable.
6 What is a discriminating test? A test whose possible outcomes separate the leading hypotheses.
7 Should a confirmed symptom be called a root cause? No.
8 What five labels should Charlie use? Confirmed, strongly supported, possible, contradicted/unlikely, and unknown/missing evidence.
Pass rule: A correct answer must distinguish confirmed evidence from inference, keep unresolved alternatives
alive, and state what observation would discriminate among them. Unsupported certainty is a reasoning error even
when the guessed component happens to be correct.