How to interpret the differences between visually checked range and simulated range

Hi lists,

I am working with semivariogram now.
I found that the ranges I got from either likfit or variofit are smaller 
than that from visual check.
Say the value for visual check is around 50 m, simulated ranges are 
around 35 m.
So how can I interpret the difference? Or is it possible that I did 
something wrong?
One problem with my data is that my sampling interval is a little big.

Many thanks!

Chunhua