Measurement Scales (3 of 6)

next previous
The answer is an unequivocal "NO." There is really no room for argument here. What can be questioned, however, is whether it is worth knowing that the mean rating of color displays is higher than the mean rating for B & W displays.

The argument that it is not worth knowing assumes that means of ordinal data are meaningless. Supporting the notion that means of ordinal data are meaningless is the fact that examples can be made up showing that a difference between means on an ordinal scale can be in the opposite direction of what they would have been if the "true" measurement scale had been used.

If means of ordinal data are meaningless, why should anyone care whether the difference between two meaningless quantities (the two means) is due to chance or not. Naturally enough, the answer lies in challenging the proposition that means of ordinal data are meaningless. There are two counter arguments to the example showing that using an ordinal scale can reverse the direction of the difference between means. next previous