The Numbers Don't Add Up to a Single Cause: Decision Tree

Why this matters

Sometimes every reading you take is trustworthy and they still refuse to tell one story. The pressure says undercharged, the temperature says overworked, the current says something else again, and no single fault fits all three. The wrong move is to force it: pick the reading you like, call it the cause, and ignore the ones that do not fit. A contradiction in the numbers is not noise to explain away. It is often the clearest sign you have that two independent faults are in the system. This tree turns that contradiction into a diagnosis instead of a shrug.

Start here: is the contradiction real or is it your instrument

Before you build a two-fault theory on a contradiction, rule out the boring cause: a bad reading. This is a different problem from two honest readings that disagree (see the tool-gives-a-suspect-reading card), and it is a fast check.

• Confirm the meter or gauge is in range, zeroed, and reading a known reference correctly.
• Confirm your technique and test points - a probe on the wrong terminal or a gauge on the wrong port invents contradictions that do not exist.
• Re-take the odd reading a second time, a second way if you can.

If the contradiction survives a verified instrument and clean technique, it is real. Now it is telling you something.

Step 1: ask if any single fault can produce all the numbers

Lay the readings out together and test one hypothesis: could one cause, by itself, generate every value you are seeing?

• If yes - you probably have one fault and you have been misreading its signature. Confirm that cause and move on.
• If no - if you cannot name a single fault that explains all of it without hand-waving one reading away, stop trying. The numbers are telling you there is more than one cause. Believe them.

The failure mode here is motivated reasoning: deciding the reading that does not fit "must be off" because it is inconvenient. If you already verified the instrument, that reading is data, not an error.

Step 2: sort the readings into groups

Two faults usually leave two internally-consistent sets of evidence tangled together. Separate them.

• Take each reading and ask which cause it would support.
• If they fall into two clean groups - these three fit a restriction, those two fit an overload - you have likely found your two faults, and each group is the case for one of them.
• Diagnose and confirm each group's fault on its own terms, then fix both.

Step 3: watch for the reading that is suspiciously normal

The hardest case is a compensating pair, where two faults push opposite ways and a key reading lands near normal while everything around it is wrong. A normal number surrounded by abnormal ones is not reassurance; it is a flag.

That last row matters: when the symptom is much larger than your suspected cause could produce on its own, something else is adding to it.

Step 4: fix and re-measure, expecting the second to surface

When you correct one fault in a contradictory set, the remaining numbers often snap into a clear pattern, because you removed the reading that was fighting them. Re-measure the full set after each fix. A contradiction that resolves into a clean single-fault signature confirms you found and removed one of two. If it stays contradictory, you have not found both yet.

Recap

1. Verify the instrument and technique before trusting a contradiction.
2. If no single fault explains every reading, stop forcing one - suspect two.
3. Group the readings by the cause each supports; two clean groups means two faults.
4. Treat a suspiciously normal reading among abnormal ones as a compensating pair.
5. Fix one, re-measure the whole set, and expect the second fault to clarify.

References

• Trade-standard practice on measurement-based diagnosis and reading reconciliation
• Manufacturer service documentation with expected reading ranges per condition
• See related: The Numbers From Two Tools Disagree; When Two Separate Faults Are Hiding in One Symptom