6.6 Other Estimation Problems
In Sections 6.4 and 6.5, we assumed that data are always drawn from a representative sample of the whole population. For some researchers this state of affairs is something of a luxury. In this section, we discuss a number of common problems that need to be taken into account in practical application and the statistical methods of dealing with them.
By data contamination we mean a set of observations that do not “belong” to the sample—see Section 6.2.3. The essentials of the formal approach can be explained using a simple model based on the distribution function (in fact, we have already seen the elements of this model in a different context—see Section 6.4.2). The idea is that, instead of observing ...