Skewed distribution

#1
Hello all!

I have a sample of 30 values from a production process that should be normally distributed. In fact, it shows a distribution heavy on the left. Any idea why?
 

Karabiner

TS Contributor
#2
Some possibilities:
a) your assumption that it should be distributed normally, was wrong.
b) a rare instance of chance variation; even with n=30, some samples from a normally distributed population can be skewed.
c) the influence of an unanticipated second factor.
 
#3
Thank you for the quick reply.

a) In a stable production process with millions of parts passing through, for example drilling the same hole. If I take a random sample, shouldnt it be normally distributed?
 

Karabiner

TS Contributor
#5
A sample of n=30 is never perfectly normally distributed, even if the underlying distribution is normal.
Most of the time, the samples will a show a distribution quite close to a normal distribution. But due to
chance, a certain proportion of samples will look different.
 
#6
I know that n=30 is very small and I can not expect a perfect normla distribution, but this sample looks really off. I will have the guys do deeper digging on this topic.

Thanks
 

hlsmith

Less is more. Stay pure. Stay poor.
#7
Do you have historic data to compare it too?

Can you post the data in a histogram?

What is the actual outcome variable (e.g., time, size, etc.)?

Have you put precision bounds on it to show that it is outside an expected range for a stable process?

@Miner please help direct them on how to do this or given any relevant advice!