以下是引用tianya1199在2011-6-7 17:15:00的发言：
我刚上GOOGLE 查了下，Which is better between mean, median and mode also depends on which type of data are we considering. There are basically 3 kinds of data: Nominal Data (qualitative data). For eg, marital status can be married, single, divorced or de facto. Ordinal data = the data are actually ranked, for eg google is the number 1 search engine and yahoo is the no 2 search engine. Interval (numerical): for example: age, height, length, breadth etc.
If we are looking at an interval(numerical) data, we can use any mean, median or mode. Mean is generally the best measure for statistical interference if there are no extreme values. When there are extreme values it is better to use median. Mode is very rarely used.
If we are looking at nominal data, we cannot calculate mean. Like in the given example, marital status can be married, single, divorced or de facto. Now look at the following table
发表于 2011-6-7 17:22
| 