R/residQ.R

In benkeser/drinf: Computes doubly-robust estimators and doubly-robust confidence intervals

#' residQ
#' 
#' This function computes the residuals that are used as outcomes in the 
#' reduced dimension regressions for Q. 
#' 
#' @param L2 A \code{vector} outcome of interest
#' @param A0 A \code{vector} treatment delivered at baseline.
#' @param A1 A \code{vector} treatment deliver after \code{L1} is measured.
#' @param Q2n A \code{vector} of estimates of Q_{2,0}
#' @param Q1n A \code{vector} of estimates of Q_{1,0}
#' @param g1n A \code{vector} of estimates of g_{1,0}
#' @param g0n A \code{vector} of estimates of g_{0,0}
#' @param abar A \code{vector} of length 2 indicating the treatment assignment 
#' that is of interest.
#' 
#' @return A list with elements rQ1 and rQ2 that are the outcomes in the
#' reduced dimension regressions on g0n and g1n, respectively. 

residQ <- function(
    L2, A0, A1, Q2n, Q1n, g0n, g1n, abar, ...
){	
	# These two residuals are regressed on g0n
    rQ1_1 <- as.numeric(A0 == abar[1]) * (Q2n - Q1n)
    rQ1_2 <- as.numeric(A0 == abar[1] & A1 == abar[2])  / (g1n) * (L2 - Q2n)

    # This residual is regressed on g1n
    rQ2 <- as.numeric(A0 == abar[1] & A1 == abar[2])  / (g0n) * (L2 - Q2n)
    return(list(
        rQ1_1 = rQ1_1, rQ1_2 = rQ1_2, rQ2 = rQ2
    ))
}