As I understand it... we have two formulas. One is for the "sample variance", and the other is for the "estimation of a population variance". (Correct me if I'm wrong here)
In the first one, we divide over "n". And in the second one, we divide over "n-1".
Can someone explain how the "n-1" gives a better estimation? This is very bizarre.. i don't see why do we treat the sample any different from the population.
Thanks guys!作者: jim8z3 时间: 2011-7-11 17:42
Use the search function. Was definitely answered very well by someone sitting for the Dec exam last year.作者: Bluetick1010 时间: 2011-7-11 17:42
As the sample is drawn for population and in many cases a representative of population. So as the sample is more likely to underestimate the number of outliers by doing n-1 in calculation sample std or variance we get a higher standard deviation to account for the distortions in data not captured by the sample.
As a rule the sample std will be higher than the population std.作者: burnsy562000 时间: 2011-7-11 17:42
As we say in statistics, dividing by n-1 produces an unbiased estimation, whereas dividing by n does not.
