Research and laboratory roles: applied workshop

Research question and evidence

Use research question and evidence to make decisions inspectable.

Turn a topic of interest into an observable or testable question.

Teaching example: A research question needs a population, variables and scope.. Predict the result, independently check the key step, and compare these assumptions with the actual role. Do not use demonstration results as real-world advice.

Protocol and variables

Use protocol and variables to make decisions inspectable.

Specify inputs, outcomes, controls and repetition.

Teaching example: Changing two conditions at once obscures attribution.. Predict the result, independently check the key step, and compare these assumptions with the actual role. Do not use demonstration results as real-world advice.

Safety and local procedures

Use safety and local procedures to make decisions inspectable.

Identify required training and local SOPs and list questions.

Teaching example: Public guidance does not replace local training and supervision.. Predict the result, independently check the key step, and compare these assumptions with the actual role. Do not use demonstration results as real-world advice.

Records and sample tracking

Use records and sample tracking to make decisions inspectable.

Record sample IDs, dates, method versions and raw results.

Teaching example: Retain anomalies and failures, not only successful results.. Predict the result, independently check the key step, and compare these assumptions with the actual role. Do not use demonstration results as real-world advice.

Analysis and uncertainty

Use analysis and uncertainty to make decisions inspectable.

Describe repeat measurements and document exclusion rules.

Teaching example: Define exclusions before viewing outcomes where possible.. Predict the result, independently check the key step, and compare these assumptions with the actual role. Do not use demonstration results as real-world advice.

Reproducibility and versioning

Use reproducibility and versioning to make decisions inspectable.

Create a data dictionary, script versions and run instructions.

Teaching example: Overwriting same-named files can break input/result traceability.. Predict the result, independently check the key step, and compare these assumptions with the actual role. Do not use demonstration results as real-world advice.

Research presentation and limitations

Use research presentation and limitations to make decisions inspectable.

Write the question, method, result, limitations and next experiment.

Teaching example: No observed difference does not prove equivalence.. Predict the result, independently check the key step, and compare these assumptions with the actual role. Do not use demonstration results as real-world advice.

Worked example · fictional

From a problem to inspectable work

Synthetic linear calibration: blank 0.2, slope 0.5 response/concentration; replicate responses 2.1 and 2.3. QC recovery is 90%, predefined acceptance 95–105%.

1. Mean response 2.2; (2.2−0.2)/0.5=4. QC fails, so this is not an accepted result.
2. Further judgment: The value looks plausible but QC fails. What next?
3. Investigate under the protocol, flag affected results and document retesting

Deliver research protocol and example log: the calculation reference is 4; the recommendation must respect Investigate under the protocol, flag affected results and document retesting. Separate numbers, inference and unknowns.

Failure case and repair

In Records and sample tracking, copy tool output without definitions, then retain the conclusion when asked: The value looks plausible but QC fails. What next?

Even a result of 4 does not explain when Investigate under the protocol, flag affected results and document retesting is supported.

Mean response 2.2; (2.2−0.2)/0.5=4. QC fails, so this is not an accepted result. Then test assumptions in Reproducibility and versioning.