Correlation
To Reduce Highway Fatalities, we should import more Lemons! (Correlation Part One)
Oftentimes, we want our research to show that one thing causes another. Unfortunately, correlation coefficients simply do not provide this information, but are sometimes erroneously interpreted in this fashion. What correlations do tell us, however, is the relationship between two variables. The statistic ranges from -1 to +1, with values closer to -1 and +1 indicating stronger relationships (0 indicates no relationship). If you can’t show causation with correlation, how can you?
(Picture from http://filipspagnoli.files.wordpress.com/2009/05/lemongraph.jpg?w=495&h=321)
====
Pirates can prevent Global Warming! (Correlation Part Two)
We have just discussed correlation coefficients and explained how they describe the relationship between two variables. We posed the question of what can show causation. Give up? The only thing that can show causation is research design that incorporates three things: correlation, time order, and nonspuriousness.
- We have already discussed correlation, and it makes sense that you would need correlation to show causation—but this is not sufficient.
- Next is time order, which just means that the cause must precede the effect in time. Again, it should make sense that the cause must come before the effect, or the relationship would be the opposite.
- Finally, we have nonspuriousness, which is just a fancy way of saying that the relationship between variables cannot be explained by some third variable. The examples in this tip of the week illustrate spurious relationships. It appears, from the correlation, that the reduction in pirates caused an increase in temperature. From above, it seems that importing lemons results in fewer traffic fatalities. What do you think could explain these relationships?
(Picture from http://www.globalwarmingisreal.com/piratesarecool4.jpg)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.