For a vision system measuring dimensions on extruded rubber, a basic repeatability study should be performed. Use 30 parts that span the range of variation you expect to see. measure each part twice. you should introduce the parts randomly to the vision system each time.
And of course the same method will be performed for the scale...if teh scale is analog and read by the operator you should have several operators weigh 30 parts twice. Assess each operators repeatability then assess the reproducibility of th eoperators. If hte readout is digital, you are probably safe skipping the operator component. However, if there is some technique to sample prep of the material to be weighed you should include operators even with a digital scale as their technique is part of the measurement system.