Tukey’s Test in statistical analysis
In: 0
Someone gives you two groups of numbers, and asks if the average for each group is possibly the same. Each individual number has random noise, so the averages of the groups will never be exactly the same, but what you can ask is “how likely is it that the averages differ by at least the amount I saw if they should have been the same?” There are several ways of doing this, what the Tukey t test does is tell you the answer when you estimate the noise in the numbers by looking at the scatter within each group of numbers, and that the noise is drawn from a bell curve. If someone told you the noise level in advance, then you’d use a Gaussian instead of a t.
Someone gives you two groups of numbers, and asks if the average for each group is possibly the same. Each individual number has random noise, so the averages of the groups will never be exactly the same, but what you can ask is “how likely is it that the averages differ by at least the amount I saw if they should have been the same?” There are several ways of doing this, what the Tukey t test does is tell you the answer when you estimate the noise in the numbers by looking at the scatter within each group of numbers, and that the noise is drawn from a bell curve. If someone told you the noise level in advance, then you’d use a Gaussian instead of a t.
Latest Answers