CONDUCTING A SENSITIVITY ANALYSIS
AT VARIOUS points in this book, we have seen how the value of model inputs can be difficult to establish with absolute certainty. For instance, HRQL scores can be obtained from various generic preference-weighted instruments, each producing a slightly different number (Gold and Muennig, 2002). Because these differences arise due to differences in the way the studies are designed, they represent a form of nonrandom error. Most model inputs will be derived from a sample and therefore also contain random error. (For a discussion of random and non-random error, see Chapter Eleven.)
We can usually ...