Why did two labs get different counts?
Explain a rerun gap and identify the evidence that can resolve it.
The question
Lab A and Lab B report how many archived sensor traces fall within an agreed band. Both reports list the same 30 trace IDs. You must explain what the difference shows and what to check.
| Lab A | Lab B | |
|---|---|---|
| Trace IDs listed | 30 | 30 |
| Traces in band | 24 | 18 |
The labs report a replay of recorded data.
“What do you tell the research lead about the rerun?”
Not saved by this site. Your browser may restore text when you navigate back.
A strong answer, step by step
Open each step in turn. This is one strong answer, not the only one.
Step 1: Name the difference
What would you do here? Open the move you’d make.
Describe the second report as an independent replication.
Tempting, but it misses something. Both reports use the same recorded trace IDs. The second lab did not collect new data.
Report the difference and say that both labs used recorded data.
The strong move. The totals show a discrepancy. A replay of archived traces gives no independent acquisition evidence.
See how a strong candidate says it
“Lab A counts 24 in-band traces and Lab B counts 18. That is a gap of six in the totals for the same listed archive.”
Why it works: The totals show a discrepancy. A replay of archived traces gives no independent acquisition evidence.
Step 2: Check the inputs and rule
What would you do here? Open the move you’d make.
Compare the run records and evidence behind both counts.
The strong move. A shared ID or count leaves open whether the labs used the same files and analysis.
Assume the analysis matched because the trace IDs matched.
Tempting, but it misses something. Trace IDs identify records. They do not identify the analysis version or rule.
See how a strong candidate says it
“I’d ask for both run records. I want to compare raw inputs, analysis versions and the band rule.”
Why it works: A shared ID or count leaves open whether the labs used the same files and analysis.
Step 3: Interpret the partial fix
The run record shows Lab B used an older alignment analysis version. With the reference version on the same archive, Lab B counts 23 in-band traces. Lab A’s reference count is 24.
What would you do here? Open the move you’d make.
Report the smaller gap and compare the per-trace results.
The strong move. Changing the analysis version brings the totals closer. A one-count difference can hide more than one per-trace disagreement.
Declare the replay resolved because the two totals are close.
Tempting, but it misses something. The totals still differ, and they do not show which traces changed classification.
See how a strong candidate says it
“The in-band count gap drops from 6 to 1. The older version accounts for most of the difference in the totals. I’d compare results by trace to see which decisions still disagree.”
Why it works: Changing the analysis version brings the totals closer. A one-count difference can hide more than one per-trace disagreement.
Step 4: Close the remaining gap
What would you do here? Open the move you’d make.
Collect a new dataset as the only follow-up.
Tempting, but it misses something. New data can test a separate question. It does not explain the remaining discrepancy on this archive.
Resolve or record each disagreement before claiming the replay agrees.
The strong move. The team gets a reproducible record of the comparison and a specific unresolved question.
See how a strong candidate says it
“I’d compare the trace-level classifications, record any remaining differences and agree which run record to reuse. Independent acquisition is a separate check.”
Why it works: The team gets a reproducible record of the comparison and a specific unresolved question.
The answer, in short
- Decision
- Give the lead the version finding and compare the per-trace results.
- Because
- The same reference version cuts the count gap from six to one.
- What would change it
- Agreement on the trace-level outcomes and run records would resolve the replay comparison.
Common traps
- Equating the same IDs with the same analysis.
- Equating count agreement with trace agreement.
Check your answer
- I identified the initial gap.
- I asked for both run records.
- I explained the reduced count gap.
- I kept independent acquisition separate.
Last reviewed Nov 11, 2026