Gini ratio

jeff jefflist at whitebuddha.com
Wed May 14 06:03:46 CDT 2003


Zscoundrel wrote:

> I think that Samuel Clemmons said it best, ". . . There are Lies, Damn 
> lies, and Statistics . . ." 

That quote really annoys me.  Translation: "Statistics can be cooked. 
 Maybe stats for the Gini ratio here are cooked."  That is the question 
here.

>
>
> It would actually take days to really give you a decent analysis, but 
> as a statistics junky and an analyst I will share some of my 
> concerns.  I am bothered by MANY of the assumptions they make in the 
> process of filtering the data to arrive at their conclusions.  Also, I 
> have been following the economy since Nixon, and I don't really 
> remember hearing about this before, but I assume that is due to the 
> complexity of the data and the difficult  nature of the calculations.

Wealth inequality seems to be a very objective concept, but I do imagine 
it must be a lot of work to calculate.

>  
> One of the things I look at is how they portray the data.  The year to 
> year percentage change chart is almost useless as an indicator the way 
> they portray it.  The data is not really linear (hell, it doesn't even 
> really have any trend) but they use a line chart to represent the data 
> because that matches the other charts.  

Why is the "year to year percentage change chart" useless?  What is 
wrong with the graph not being linear?

> This tells me that the authors are more concerned with looking good 
> than clearly communicating the data. 
> It makes me wonder about what other short cuts they have taken that 
> don't show.

It seems to me the authors are showing a high level of abstraction. 
 What is wrong with that.  I would need a detailed analysis of the 
calculations to determine the accuracy of this graph but not to 
understand the point made by this graph.  I have not looked at a 
detailed analysis, but this graph does fit with my knowledge of history 
and a few other disciplines.

All you have written is your opinion.  How about some facts.  You don't 
need to spend days on this.  I think you could do a lot better spending 
10-15 minutes, if you are not too busy.




More information about the Kclug mailing list