Gamma vs. Tau's b

#1
Hi everyone

I am studying political science, and we have during our classes of statistical research learned to use gamma for ordinal variables. Our teacher has taught us, that gamma often overestimates the correlation since it doesn't count for ties. On a note he furthermore told us that Kendall's Tau corrects for ties, but didn't really stress that it is worth using.

I have since been researching a bit, and I can't really figure out what the reason for using gamma instead of Tau is.
Is there any rational explanation on this?

This is my first post here, hoping i am posting in the right subforum and that my question isn't too stupid :)
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
If I recall, the gamma is best applied to ordinal variables with 5 or less groups. Not sure about Tau's b.