# Accuracy and Reproducibility - (Mar/07/2008 )

I am not sure but hopefully someone can clarify this for me

accuracy = standard deviation

reproducibility = mean

Therefore you would normally pick somehting with a smaller standard deviation (eg. pippete with mean of say 600uL with deviation of 2uL) than something with a more consistent mean like 650uL and deviation of 15uL. this would be if you wanted to collect 650uL.

so am i right for both of these??

I would directly you to the Wikipedia article that talks about this subject.

Accuracy can be regarded as the difference between the sample mean and the true mean. aka the bias of the measurement.

reproducibility / precision is the degree of variability in the measured value... which is characterised as standard deviation.

As for the second question, it depends.

I think it actually better to have a high accuracy and low precision is result. If you repeat the experiment enough times, you wil get the answer. And there are statically tricks to remove the effects of extreme value in the sample.

A low accuracy, high precision result is only useful if you can take into account the bias of the results ( by comparisons to controls). But if you can't take in account of said bias, or not aware of it - perhaps caused by a confounding factor, the results of a low accuract but high precision experiment remains wrong. And you might not even realise the results are wrong.

If you want 650ul, pooling ten samples of 650+/- 15ul volumes and dividing the pool into ten will probably hit 650ul.

Doing the same with sample of 600+/- 2ul volume won't help reach 650ul unless it is realised that the pipette has a bias and said bias is accounted for.

i was talking to an academic at my university and they said that accuracy = standard deviation and reproducibility=mean of volumes. so yea wikipedia the magical source of scientific knowledge lol has proven wrong once again

as for the second bit i agree as well about haing high accuracy and low precision