
Estimation of Probability Measures Using Aggregate Population Data 187
Because of the assumed ability to track individuals, the NPML method is
not directly applicable to aggregate data problems treated in this chapter. It
is also not clear if the restrictive assumptions on the error random variables
could be relaxed with least squares estimation us ed in the place of the maxi-
mum likelihood e stimation method. It is common to establish the equivalence
of least s quares and maximum likelihood estimators by demonstrating that
they solve the same set of estimating equations (see, e.g., [29, Chapter 3] and
Chapter 4 above on model s election). However, ...