Question

Lets say we have a disperse quantitative multinominal distribution on a ratio scale.

I am calculating its STDV. The number I get, both analytically and in matlab, is big - about half the mean. Which seems reasonable for this population.

I am lost how to interpret the result. I understand that 95% of a normal distribution lies within 3 STDV of the mean. In my example, this does not make sense. It would simply go past both upper and lower limits. What is the meaning of STDV for a multinominal dist like this? What is a good way of describing/measuring variance in such distribution?

Answer 1

@.Sam.
Help pls! )

Answer 2

so this is the 95% and 5% confidence levels.

Answer 3

the upper and lower values, which your figure is outside of

Answer 4

i would suggest plotting on log graph and see how it looks.