ABSTRACT
This chapter provides an isolated treatment of measurement error in binary regression, probably the most important of the nonlinear regression models. This model, described in detail in Section 6.2, has an outcome Y which takes on two values, 0 or 1, with
E(Y |x) = P (Y = 1|x) = m(x, β) = μ and V (Y |x) = μ(1 − μ). As elsewhere, with an intercept in the model we write x′∗ = (1,x′) and x′∗β = β0 + x′β1.
