Quote:
Originally Posted by risingstar3110
I have a question. what is the name of the test/analyse method which you take the biggest item out of the group to see whether other items significant amongst themselves?
Like let's say 20 studied groups spent 1,000,000$ for gambling each year. But Group A with highest spent already used 800,000$ a year for gambling. So while Group B spent only 50,000$, 5% of the total and right at average amongst 20 groups; they actually spent as much as 25% out of the remaining 19 groups(after take out A).
I want to say: "Based on XYZ testing method and take out the largest item (group A), group B showed a real significant contribution to gambling problem in the community". Something like that. It really sounds like sensitive test, but i'm not sure if it is one
|
^ It's been a while since I've done any intensive statistics courses or analyses .. (Finance is mainly adding and subtracting after all ..
) but from what I recall, what you're thinking of is "sample skew" which is addressed here:
http://www.itl.nist.gov/div898/handb...on3/eda35b.htm
What you're proposing seems straightforward, but please do understand that in statistics, if your sample above is done through proper sampling methods.. you can't and shouldn't just manipulate your sample that way. What you should do is determine whether you should transform your dataset or perform analysis more appropriate for the sample distribution.. Please feel free to tell me that I'm wrong
EDIT: Also wanted to point out that you should also do an analysis for skewness and kortosis as they're valuable information about your sample/data set. Also keep in mind that there are often reasons why samples are skewed, which is why I would hesitate to manipulate the samples in this regard.
Reply hazy, ask again later