1.2. Mean Estimation in the Binary Choice Problem
Another motivating example can be found in the estimation of mean of the binary choice model.
Mean estimation in the binary classification
Our model is from the (case 2) of the previous article.
P(Y=1|Z=z)=F0(z), Z∼Q, F0∈Λ,Λ:={F:R→[0,1], F is increasing}.This time the estimand is the population mean
θ0=θF0:=∫F0(z)dQ(z)=EF0(Z)=P(Y=1)=EY.By definition, it is clear that E[Yi−F0(Zi)]=0.
We will only consider the case where Q is known. In this case, the MLE of F0 becomes
ˆθn=θˆFn=∫ˆFn(z)dQ(z),where ˆFn is the MLE of F0.
For F∈Λ, define
θF=∫F(z)dQ(z),then by the classical central limit theorem, we get
1√nn∑i=1(F(Zi)−θF)d→N(0,Var(F(Z))).Again, if we prove some form of functional CLT for all F∈Λ, then together with the result form the classical CLT, we get the asymptotic distribution of the MLE.
1√nn∑i=1(ˆFn(Zi)−ˆθn)=1√nn∑i=1(F0(Zi)−θ0)+oP(1)⟹√n(ˆθn−θ0)d→N(0,∫F0(z)(1−F0(z))dQ(z)).
We will take the fact 1n∑ni=1ˆFn(Zi)=ˉY for granted. By this fact, √n(ˆθn−θ0)=√n(ˉY−θ0)−√n(ˉY−ˆθn)=√n(ˉY−θ0)−1√nn∑i=1(ˆFn(Zi)−ˆθn)=√n(ˉY−θ0)−1√nn∑i=1(F0(Zi)−θ0)+oP(1)=1√nn∑i=1(Yi−θ0−F0(Zi)+θ0)+oP(1). By the classical CLT, 1√nn∑i=1(Yi−F0(Zi))d→N(0,Var(Y−F0(Z))). The variance can be rewritten as Var(Y−F0(Z))=E[Y−F0(Z)]2=E[Y2−2YF0(Z)+F20(Z)]=E[Y2−F20(Z)]=EY−EF20(Z)=EF0(Z)−EF20(Z). Hence the result follows from the Slutsky's theorem.
References
- van de Geer. 2000. Empirical Processes in M-estimation. Cambridge University Press.
- Theory of Statistics II (Fall, 2020) @ Seoul National University, Republic of Korea (instructor: Prof. Jaeyong Lee).