Originally Posted by asting
... Did you factor in the sales numbers for vehicles? That is, if 150 civics were sold in an area and there are 50 used cars it does not mean the same thing as 10000 cars sold new and 50 for sale. Also, if you're not going to use a random sample you should probably weight by some city size factor.
... Perhaps everyone here reading this saw those flaws, but perhaps they went unnoticed by some and deserve to be pointed out for their sake.
When I first read the original post I wondered whether differences in the volume of new sales between brands or locations might affect the results in some way, but with my limited knowledge of statistics I could not identify what the effect might be. Same for differences between brands and cities in the total number of used cars for sale.
You point out that for different volumes "it does not mean the same thing". Can you explain how? I'm not looking for a detailed tutorial on statistics, just a general idea of whether it makes a big difference or not for the purpose that kam327 did the original analysis.