CBMSM.fit
CBMSM.fit(
treat,
X,
id,
time,
MultiBin.fit,
twostep,
msm.variance,
time.vary,
init,
...
)
A vector of treatment assignments. For N observations over T time periods, the length of treat should be N*T.
A covariate matrix. For N observations over T time periods, X should have N*T rows.
A vector which identifies the unit associated with each row of treat and X.
A vector which identifies the time period associated with each row of treat and X.
A parameter for whether the multiple binary treatments
occur concurrently (FALSE
) or over consecutive time periods
(TRUE
) as in a marginal structural model. Setting type = "MultiBin"
when calling CBMSM
will set MultiBin.fit to TRUE
when
CBMSM.fit is called.
Set to TRUE
to use a two-step estimator, which will
run substantially faster than continuous-updating. Default is FALSE
,
which uses the continuous-updating estimator described by Imai and Ratkovic
(2014).
Default is FALSE
, which uses the low-rank
approximation of the variance described in Imai and Ratkovic (2014). Set to
TRUE
to use the full variance matrix.
Default is FALSE
, which uses the same coefficients
across time period. Set to TRUE
to fit one set per time period.
Default is "opt"
, which uses CBPS and logistic regression
starting values, and chooses the one that achieves the best balance. Other options
are "glm" and "CBPS"
Other parameters to be passed through to optim()