problem.so Solve today’s problem

Competing Hypotheses Evidence Matrix

Keep two plausible explanations alive while evidence is reviewed. Score each observation against both hypotheses to expose conflicts and neutral facts.

What this tool preservesYour inputs, calculation, and assumptions remain visible together.
Observation 1
Observation 2
Observation 3

Your result will appear here

Complete the inputs, then use the button to calculate or organize the result.

Why use an evidence matrix?

Method: An evidence matrix reduces confirmation bias by asking how every observation fits more than one explanation. Conflicting evidence is often more informative than evidence that both hypotheses predict.

Compare evidence across explanations

  1. Write two explanations before scoring evidence.
  2. Describe observations without causal wording.
  3. Mark support, conflict, or neutral for both hypotheses.
  4. Seek the next observation with different predictions.

Use contradictions to narrow the investigation

For a recurring system slowdown, peak demand and a configuration fault may both explain long response times. Normal performance under high demand conflicts with the first hypothesis, while a slowdown at quiet times supports the second.

A score cannot measure source reliability

Record the observation before its interpretation

Write the measured event, timestamp, document, or direct observation first. Then score its relationship to each explanation. This order prevents the favored hypothesis from being smuggled into the evidence description. Preserve the source beside each row so another reader can distinguish a direct record from a recollection, summary, or inference. If an observation depends on a measurement threshold, state that threshold instead of reducing it to present or absent.

A tied score means the current evidence does not discriminate well. The next action should not be to collect more of the same. Choose an observation that one hypothesis predicts and the other does not. Give special attention to reliable contradictions because one failed necessary prediction can be more useful than several weak points of agreement. Reopen both explanations if the result changes when a disputed row is removed.

Boundary of the result

Simple scores summarize direction, not evidence quality. A weak observation should not outweigh one reliable contradiction.

Continue with the method

Use the competing hypotheses evidence log and the hypothesis testing guide and the pareto contribution analyzer to move from a quick result to a complete reasoning process.