If I have a grouped frequency distribution is it ok to use the standard deviation or should I stick to median and interquartile range for summary statistics? I'm not sure if the intervals mean that I have lost too much information for the sd to work.
2006-09-30 10:27:17 · 4 answers · asked by statstastic 2 in Science & Mathematics ➔ Mathematics
http://davidmlane.com/hyperstat/glossary.html
Statistics are a model. They represent data, they are not the data, but the data without interpretation may be meaningless. How valid would the standard deviation be in a nonstandard distribution? How far from known models is the distribution? Why was it used if it represents poorly? Is there a more meaningful distribution for the data, can it be refigured, and if not, why was the original data lost?
Peer review is needed, I get the impression that carelessness and/or inexperience muddied the outcome. Much can be learned from failure if it is analysed, maybe more than success, because failure's cause can usually be determined.
2006-09-30 10:37:18 · answer #1 · answered by Anonymous · 0⤊ 0⤋
It's OK to use the standard deviation for grouped data.
2006-09-30 17:33:15 · answer #2 · answered by daniel_cohadier 3 · 0⤊ 0⤋
stick to the median
2006-09-30 17:30:31 · answer #3 · answered by ♥.:ily:.♥ 1 · 0⤊ 0⤋
depends on what you are analyzing
2006-09-30 17:35:05 · answer #4 · answered by Anonymous · 0⤊ 0⤋