Re: normal distribution doubts

Allen McIntosh wrote:
s99999999s2003@xxxxxxxxx wrote:
If, after calculation i find that there is skewness value other than
0, then i can say that the data doesn't follow normal distribution.
That means also i cannot use empirical rule but have to use chebyshev
inequality for approximating the no. of data values within some
values of std deviation from the mean, is that correct?

Not necessarily. Statistical variability will also cause the observed
skewness to be non-zero. The question you need to answer is whether
the difference you observe could reasonably happen by chance.

.... and even if the normal distribution isn't appropriate, the "chebyshev inequality for approximating the no. of data values within some mutiple of the std deviation from the mean" wouldn't be the alternative of choice. Rather the OP should try to find a reasonably appropriate distribution to use instead of the normal. The chebyshev bounds can be rather poor approximations.

David Jones
.

Loading