\"Two measured variables make a study correlational.\"
Unfortunately, this conflates (mistakenly treats as the same)the types of claims we can make with the types of statistical testswe can use. We pick out statistical tests based on the levels ofmeasurement in our data, and while the measured/manipulateddistinction is important for interpretation (manipulated variablesallow for stronger arguments for causality), this doesn't effectour choice of test.
To make this clear, first give me an example that uses twomeasured variables but isn't tested using a correlation (adifferent test is right choice).
Second, give me an example where two variables aren't both justmeasured (at least one is manipulated) and yet a correlation is theproper test.
