Applying a series of diagnostic and calibration functions to a series of matched data sets to determine impact of matching on TE bias, variance and total error, and to select the best matching parameters.
# S3 method for mlr
summary(object, power = FALSE
, power.control = list(rnd = TRUE, d = 0.5, sig.level = 0.05
, niter = 1000, rnd = TRUE)
, max.method = c("single-covariate", "covariate-subspace"
, "absolute")
, verbose = FALSE, ...
, orsq.min = 1e-03, orsq.max = 1e0, n.orsq = 100)An object of class summary.mlr, with the following elements:
Same as input.
Matrix of aggregate bias values, one row per calibration index, and three columns: 1) single-covariate maximum, 2) covariate-subspace maximum, and 3) absolute maximum, in that order.
Matrix of biases, one row per calibration index, and one column per candidate omitted term.
Vector of normalized variances, one per each value of calibration index.
Matrix of power calculations, one row per calibration index. Each row is identical to output of mlr.power for that calibration index value.
Matrix of standardized mean differences, one row per calibration index, and one column for each included or omitted covariates.
Output of mlr.combine.bias.variance applied to bias and variances at each calibration index value.
An object of class mlr, typically the result of a call to mlr.
Boolean flag indicating whether Monte-Carlo based power analysis must be performed or not.
A list containing parameters to be passed to mlr.power for power calculation.
Which constrained bias estimation method must be used in bias-variance trade-off and other analyses?
Whether progress message must be printed.
Parameters to be passed to/from other functions.
Minimum value of omitted R-squared used for combining normalized bias and variance.
Maximum value of omitted R-squared used for combining normalized bias and variance.
Number of values for omitted R-squared to generate in the specified range.
Alireza S. Mahani, Mansour T.A. Sharabiani
Link to a draft paper, documenting the supporting mathematical framework, will be provided in the next release.