Abstract Data grouping is a common statistical method, which could produce misleading or false results. Some cases of real data analysis are presented and interpreted in this paper. It is argued that analysis of data could be arbitrary, but the explanation of results should be constrained by study objects and circumstances. Some techniques to avoid grouping errors are provided in this paper, and a perspective to understand Simpson paradox discussed.
|
Received: 23 January 2008
|
|
|
|