# Building confidence intervals using Coefficient of variation

#### jfca283

##### New Member
Hi
I have a doubt and i need to understand it.
At my work, my colleagues use the coefficient of variation for doing confidence intervals.
Not directly, obvious.
They used this formula for the lower interval
Code:
X-Z(95%)*X*Coefficient of variation/100
.
And i just don't get it.
I've played with it and tried to take it to the class familiar way
Code:
X-Z(95%)*Standard Deviation/Square Root(n)
So, can you guide me?
What am i forgetting to assimilate this new way of creating intervals?
Thanks a lot for your replies and time.

#### hlsmith

##### Omega Contributor
Yeah, I am also not quite following. Are these data coming from a regression model or not?

#### hlsmith

##### Omega Contributor
Just using data for a singe variable or is this from a more complex model?

#### hlsmith

##### Omega Contributor
Well if I am following:

X-Z(95%)*X*Coefficient of variation/100

If Coefficient was calculated (x/sd)*100, then the above would give you:

mean -(zscore*sd), which would be different from your standard approach since it would not directly address the sample size.

#### jfca283

##### New Member
That's right. Where is the "Square root(n)"?
If i am working with a population or a sample, does it make a difference in "Square root(n)"?

#### hlsmith

##### Omega Contributor
You will have to verify this, but isn't the SE of a population the SD.

Are you treating your data as a sample or population?