iptw_pipeline_long: Estimate the DID g-formula using...

In audreyrenson/didgformula: The difference-in-differences g-formula

Estimate the DID g-formula using inverse-probability-of-treatment-weights with exposure models pooled over time periods.

Description

Estimate the DID g-formula using inverse-probability-of-treatment-weights with exposure models pooled over time periods.

Usage

iptw_pipeline_long(
  df_obs,
  df_interv,
  den_formula,
  num_formula,
  family,
  yvar = "Y",
  ylagvar = "Y_lag",
  idvar = "uid",
  timevar = "t",
  tibble = TRUE,
  pt_link_fun = NULL,
  binomial_n = NULL,
  models = TRUE
)

Arguments

df_obs	long format data frame, ordered by person id, then time
df_interv	same data set as df_obs with exposure variable set to the intervened status
den_formula	chr or formula. glm formula for (pooled) denominator model
num_formula	chr or formula. glm formula for (pooled) numerator model
family	stats::families object for numerator and denominator glm call
yvar	chr. Name of column in df_obs corresponding to the outcome at time t
ylagvar	chr. Name of column in df_obs corresponding to the outcome at time t-1
idvar	chr. Name of column in df_obs corresponding to the person id
timevar	chr. Name of column in df_obs corresponding to times (typically starting from 1, so that t-1=0)
tibble	lgl. Return results as a tibble (TRUE) or vector (FALSE)?
pt_link_fun	function. The scale on which parallel trends is assumed (e.g., qlogis for logit scale). Default NULL for untransformed scale.
binomial_n	int length nrow(data). Group sizes for aggregate binomial data.
models	logical. Return all models as an attribute?

Value

Estimates of E(Yt(a) - Yt-1(a)), in the form of a tibble or vector (dependening on argument tibble), for times t=1,2,...,Tt (in that order).

audreyrenson/didgformula documentation built on Oct. 9, 2022, 11:45 a.m.