Describe a result you could not reproduce and how you investigated.
Pick a real example where a result would not repeat. Describe what you tried to reproduce and what differed. Explain how you went back through your lab notes, methods, materials, equipment and settings, and compared them step by step. Say who you consulted, and what tests you ran. Share what you found, or what you concluded if there was no clear cause. Show that you reported the outcome honestly and kept clear records.
How would you test whether a treatment changes an outcome?
Start with a clear question and the outcome you will measure. Include a control group that does not get the treatment, so you can compare. Use enough replicates, meaning repeated samples, so results are not just chance. Assign samples to groups at random to avoid bias, and keep everything else the same. Choose reliable measurement methods and record them. Analyse the results with a suitable statistical test, and repeat the experiment to confirm.
How do you choose between a promising new experiment and completing validation of an older result?
Start by asking which choice adds more scientific value. An old result that is not validated may weaken everything built on it, so check what depends on it. Then look at the new idea. Is it time sensitive, or could someone else publish first? Compare the time, money, samples and people each needs. Often you can do a small pilot of the new work while finishing validation. Share your reasoning with your manager and agree on the plan together.
A result supports your hypothesis but a control failed. What do you do?
Do not claim the finding. A failed control means the experiment cannot be trusted, even if the result looks good. Stop and diagnose the cause, checking reagents, equipment, setup and technique. Write down what you find. Then fix the problem and repeat the experiment with working controls. Tell your supervisor openly about the issue. Only report a result once it holds up with valid controls.