The First-Stage in an Instrumental Variable Approach to Control for Measurement

The first-stage in an instrumental variable approach to control for measurement error in an independent variable is

A) regressing the mis-measured variable on only the exogenous variables in the original regression equation and using those estimates to calculate predicted values of the mis-measured variable.
B) regressing the mis-measured variable on only the alternative value of the mis-measured variable.
C) estimating the original population regression model with the predicted values of the mis-measured variable substituted for the observed values of the endogenous variable.
D) regressing the mis-measured variable on both the alternative measure of the mis-measured variable and all of the remaining exogenous variables and using those estimates to calculate predicted values of the mis-measured variable.