You are working on the first major experiment for your project. As part of the experiment, you have to draw a dose-response curve for a drug with 10 different concentrations.
The effect of that drug at each concentration point will be calculated as the average of three measurements. You collect 9 useful data points, but data point number 10 gives you trouble.
There is a large variance between the results of the three measurements, so something seems to be wrong. You cannot explain what has happened.
You show the data to one of the Ph.D. students in the lab. He agrees that the first nine data points look promising, but also notices the deviation in the tenth.
He also tells you that your supervisor is “a bit obsessed” with pretty datasets. If you would show her the data you have now, she would definitely force you to find out what went wrong and then redo the entire experiment.
“You have obviously made a mistake – if it was me I would just redo the measurement that resulted in outliers. If the result is better I would just replace the outliers with the new measurements without mention. These things often happen in the lab and there is no need to waist the time of the supervisor with every erroneus results”
Do you follow his advice?