Levels of Measurement Summary

So now I want to take the last two lessons and merge them together and see if we can draw some insights about the types of data that we’re going to encounter. Specifically, I want to make the claim to you that the four different levels of measurement we’ve just discussed– nominal, ordinal, interval, and ratio– form a sort of hierarchy. So I want to walk through this. At the base level, we have nominal. This is really the variable with the least amount of information to it. It just has categories to it. There’s no order. It has the least amount of information. If we think about this as a sort of information hierarchy, ordinal variables just have a little bit more information.

There’s categories. But now there is an ordering to them. So if you think about which way you might want to store or collect or analyse data, clearly the ordinal variable has more information. There’s more information. There’s probably more story we can tell. Beyond that, we have that interval numeric data. This is data that’s measured in numbers, but with some sort of arbitrary zero– clearly more information. Now I’ve got some sort of scale that I’m measuring with, some sort of unit of measure. And I can use numbers for it. Last but not least, I have that ratio level of measurement where I’ve measured in numbers, and I’ve got an absolute zero.

All else being equal, if you can measure something at that ratio level, you probably should, because you’re getting the most information represented. Also I think this is just a handy graphic to have, because it contains all this information in one nice, neat little image that represents the hierarchy of information from least information to most information in the data that you might analyse.