oneStepATT function

Calculate Additive treatment effect among the treated (oneStepATT)

An internal function called by the tmle function to calculate the additive treatment effect among the treated (ATT) using a universal least favorable submodel (on the transformed scale if outcomes are continuous). The function is called a second time with updated arguments to calculate the additive treatment effect among the controls (ATC). Missingness in the outcome data is allowed.

oneStepATT(Y, A, Delta, Q, g1W, pDelta1, depsilon, max_iter, gbounds, Qbounds, obsWeights)

Arguments

• Y: continuous or binary outcome variable
• A: binary treatment indicator, 1 - treatment, 0 - control
• Delta: indicator of missing outcome. 1 - observed, 0 - missing
• Q: a 3-column matrix (Q(A,W), Q(1,W), Q(0,W))
• g1W: treatment mechanism estimates, $P(A=1|W)$
• pDelta1: censoring mechanism estimates, a 2-column matrix [$P(Delta=1|A=0,W)$, $P(Delta=1|A=1,W)$]
• depsilon: step size for delta moves, set to 0.001
• max_iter: maximum number of iterations before terminating without convergence
• gbounds: bounds on the propensity score for untreated subjects
• Qbounds: alpha bounds on the logit scale
• obsWeights: sampling weights

Returns

• psi: effect estimate (on the transformed scale for continuous outcomes)

• IC: influence function

• conv: TRUE if procedure converged, FALSE otherwise

Author(s)

Susan Gruber

See Also

tmle,