Design of simulation scenario A.

True regression coefficients (β) and Variance Inflation Factor (VIF) for k = 20 variables explaining insulin sensitivity. They were calculated to estimate insulin sensitivity (i.e. ln[SI+1]) by OLS: a.1(fasting glucose), a.2 (120’ post load glucose), a.3 (AUC glucose*), b.1(fasting insulin*), b.2 (120’ post load insulin*), b.3 (AUC insulin*), c.1(fasting C-peptide**), c.2 (120’ post load C-peptide**), c.3 (AUC C-peptide*), d.1(fasting proinsulin*), d.2 (120’ post load proinsulin*), d.3 (AUC proinsulin*), e.1(fasting amylin***), e.2 (120’ post load amylin***), e.3 (AUC amylin*), f (age), g.1 (BMI*), g.2 (waist circumference*), g.3 (hip circumference*), h (uncorrelated noise variable)
* ln[x]
** ln[x+1]
*** sqrt[x]
Design of simulation scenario A.