3 ms·
You need an infinity of samples to be sure. But moreover, in most situations, the other variables cannot be absolutely controlled but rather are only roughly c
by joe_the_user 17y ago
You need an infinity of samples to be sure.
But moreover, in most situations, the other variables cannot be absolutely controlled but rather are only roughly controlled. How roughly depends on the field and is often the rub.
- alex_stoddard 17y agoAbsolutely true, but this is more a criticism of experiment design than statistical methodology. Also it is bias in uncontrolled variables that is most concerning. (More than uncontrolled noise although that will reduce the power to detect real differences).