Write competing predictions
Suppose a stale screen could come from caching or from a missed state update. A cache bypass and a state-transition trace answer different questions. State the expected result for each theory before running the check. Otherwise it is easy to reinterpret any outcome as support for the original guess.
Keep the experiment bounded
Change one relevant condition at a time and preserve the initial reproduction. Prefer a deterministic test or a narrow diagnostic observation when available. Remove temporary instrumentation after it has served its purpose. If the result is ambiguous, record why and choose the next discriminating check.
Does one passing experiment prove the cause?
It supports a prediction under the conditions tested. Check for other explanations that make the same prediction before treating it as a complete causal account.
A question for your assistant
For [failure], list the two strongest explanations supported by current evidence. For each, state a prediction and an observation that would weaken it. Propose the smallest check that distinguishes the explanations, without modifying production data.