It's been recommended that I randomise samples across a 96 well plate, but not true randomisation, just don't run duplicate samples next to one another. Can someone point me in the direction of some stats to back this up? The assay has no plate or edge effects so I cannot see the value in doing this and it just introduces the potential to mis-pipette. What do people do when they split up rows or columns for some sort of randomisation do they have data to back this methodology up or is it just accepted practice without it being data driven?
I used to go through this with statisticians all the time. Your concern about mis-pipetting being a bigger problem than intra-plate variation is a real concern. You can do a few experiments to understand your intra and inter plate variation. If you are performing an ELISA then pay close attention to the skill and reproducibility of the technician's pipetting. My experience was that if human performance is maximized then where on the plate a given sample or replicate is placed will make very little difference in the outcome. This is easily tested with borderline samples across a randomized plate or a non-randomized plate. If there is very little difference in outcome than go with the easier setup to prevent mistakes that could affect results. I know the statisticians did not always agree with me.
Definitely. Randomized locations is for robots, not humans. You can try to do it a few times to test if there's a problem. But for routine use, I think you're exactly right, and I congratulate you on this self-critical insight. Human error is by far your most important problem.
Whatever supervisor told you to do it needs to get back into the lab right now and try it themselves. Getting back in touch with how stuff works on the ground will make them a more effective supervisor.
I agree.
Sometime happen in agricultural experiments, avoiding poor allocation of treatments to experimental units. In these case, a Restricted randomization applies and the experiments follow particular designs.
https://www.jstor.org/stable/2983986?origin=JSTOR-pdf&seq=1
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics