nlme: nlme
In bcallaway11/qrme: Quantile Regression with Measurement Error

function to compute nonlinear models with two sided measurement error

nlme(
  data,
  Yformla,
  Tformla,
  tau,
  tvals,
  copType = "gaussian",
  simstep = "MH",
  ndraws = 250,
  reportTmat = TRUE,
  reportSP = TRUE,
  reportUM = TRUE,
  reportPov = TRUE,
  povline = log(20000),
  reportQ = c(0.1, 0.5, 0.9),
  Ynmix = 1,
  Tnmix = 1,
  tol = 1,
  iters = 400,
  burnin = 200,
  drawsd = 4,
  messages = FALSE,
  se = FALSE,
  biters = 100,
  cl = 1
)

`data`	data.frame
`Yformla`	formula for outcome model
`Tformla`	formula for treatment model
`tau`	values of tau to estimate first step quantile regressions for
`tvals`	values of the treatment to compute conditional distribution-type parameters for
`copType`	what type of copula to use in second step. Options are "gaussian" (the default), "clayton", or "gumbel"
`simstep`	whether to use an MH algorithm ("MH") or an importance sampling algorithm ("ImpSamp")
`ndraws`	number of draws to use in MH algorithm to estimate first step quantile regressions (default 250)
`reportTmat`	whether or not to report a transition matrix
`reportSP`	whether or not to report Spearman's rho (rank-rank correlation)
`reportUM`	whether or not to report upward mobility parameters
`reportPov`	whether or not to report fraction of population below the poverty line as a function of parents' income
`povline`	value of the poverty line (default log(20000))
`reportQ`	quantiles of child's income as a function of parents' income to report (default is .1,.5,.9)
`Ynmix`	number of mixture components for outcome measurement error model
`Tnmix`	number of mixture components for treatment measurement error model
`tol`	tolerance for first step quantile regression model to converge (default is 1). Note that convergence will be sensitive to `length(tau)` and the number of mixture components included in the model
`iters`	the number of MCMC iterations (default is 400)
`burnin`	the number of MCMC iterations to drop (default is 200)
`drawsd`	starting value of standard deviations of mixture components
`messages`	whether or not to report details of computation as they occur (default is `FALSE`)
`se`	whether or not to estimate standard errors using the boostrap (default is FALSE)
`biters`	if computing standard errors, the number of bootstrap iterations to use (default is 100)
`cl`	allows for parallel processing in computing standard errors using the bootstrap (the default is 1)