thank you for your example but this formula looks to strict for me.
I have a scale for social layer with lower, middle and upper class. If I would use this kind of formula, it would mean that all parents of lower class children wouldn't be interested in sending them into high school. That wouldn't be realistic. In your example there would be a correlation size of 1 between
I think it would be necessary to define which amount of correlation of two items should be generated inside the sample data.