robmed (version 0.5.0)

fit_mediation: (Robustly) fit a mediation model

Description

(Robustly) estimate the effects in a mediation model.

Usage

fit_mediation(data, x, y, m, covariates = NULL,
  method = c("regression", "covariance"), robust = TRUE,
  median = FALSE, control, ...)

Arguments

data

a data frame containing the variables.

x

a character string, an integer or a logical vector specifying the column of data containing the independent variable.

y

a character string, an integer or a logical vector specifying the column of data containing the dependent variable.

m

a character, integer or logical vector specifying the columns of data containing the hypothesized mediator variables.

covariates

optional; a character, integer or logical vector specifying the columns of data containing additional covariates to be used as control variables.

method

a character string specifying the method of estimation. Possible values are "regression" (the default) to estimate the effects via regressions, or "covariance" to estimate the effects via the covariance matrix. Note that the effects are always estimated via regressions if more than one hypothesized mediator is supplied in m, or if control variables are specified via covariates.

robust

a logical indicating whether to robustly estimate the effects (defaults to TRUE).

median

a logical indicating if the effects should be estimated via median regression (defaults to FALSE). This is ignored unless method is "regression" and robust is TRUE.

control

a list of tuning parameters for the corresponding robust method. For robust regression (method = "regression", robust = TRUE and median = FALSE), a list of tuning parameters for lmrob as generated by reg_control. For Huberized covariance matrix estimation (method = "covariance" and robust = TRUE), a list of tuning parameters for cov_Huber as generated by cov_control. No tuning parameters are necessary for median regression (method = "regression", robust = TRUE and median = TRUE).

…

additional arguments can be used to specify tuning parameters directly instead of via control.

Value

An object inheriting from class "fit_mediation" (class "reg_fit_mediation" if method is "regression" or "cov_fit_mediation" if method is "covariance") with the following components:

a

a numeric vector containing the point estimates of the effect of the independent variable on the proposed mediator variables.

b

a numeric vector containing the point estimates of the direct effect of the proposed mediator variables on the dependent variable.

direct

numeric; the point estimate of the direct effect of the independent variable on the dependent variable.

total

numeric; the point estimate of the total effect of the independent variable on the dependent variable.

fit_mx

an object of class "lmrob" or "lm" containing the estimation results from the regression of the proposed mediator variable on the independent variable, or a list of such objects in case of more than one hypothesized mediator (only "reg_fit_mediation").

fit_ymx

an object of class "lmrob" or "lm" containing the estimation results from the regression of the dependent variable on the proposed mediator and independent variables (only "reg_fit_mediation").

fit_yx

an object of class "lm" containing the estimation results from the regression of the dependent variable on the independent variable (only "reg_fit_mediation" and if robust is FALSE).

cov

an object of class "cov_Huber" or "cov_ML" containing the covariance matrix estimates (only "cov_fit_mediation").

x, y, m, covariates

character vectors specifying the respective variables used.

data

a data frame containing the independent, dependent and proposed mediator variables, as well as covariates.

robust

a logical indicating whether the effects were estimated robustly.

median

a logical indicating whether the effects were estimated via median regression (only "reg_fit_mediation").

control

a list of tuning parameters used (only if robust is TRUE).

Details

If method is "regression", robust is TRUE and median is FALSE (the defaults), the effects are estimated via robust regressions with lmrob.

If method is "regression", robust is TRUE and median is TRUE, the effects are estimated via median regressions with rq. Unlike the robust regressions above, median regressions are not robust against outliers in the explanatory variables.

If method is "covariance" and robust is TRUE, the effects are estimated based on a Huber M-estimator of location and scatter. Note that this covariance-based approach is less robust than the approach based on robust regressions described above.

References

Alfons, A., Ates, N.Y. and Groenen, P.J.F. (2018) A robust bootstrap test for mediation analysis. ERIM Report Series in Management, Erasmus Research Institute of Management. URL https://hdl.handle.net/1765/109594.

Yuan, Y. and MacKinnon, D.P. (2014) Robust mediation analysis based on median regression. Psychological Methods, 19(1), 1--20.

Zu, J. and Yuan, K.-H. (2010) Local influence and robust procedures for mediation analysis. Multivariate Behavioral Research, 45(1), 1--44.

See Also

test_mediation

lmrob, lm, cov_Huber, cov_ML

Examples

# NOT RUN {
data("BSG2014")
fit <- fit_mediation(BSG2014,
                     x = "ValueDiversity",
                     y = "TeamCommitment",
                     m = "TaskConflict")
test <- test_mediation(fit)
summary(test)

# }