sensPlot: Create a contour plot sumamrizing the result of the...

Description Usage Arguments Details Author(s) References See Also Examples

View source: R/sensPlot.R

Description

sensPlot creates a contour plot that summarizes the results of the sensitivity analysis obtained from GLM.sens. The plot region is defined by the coefficient on U in the outcome model (vertical axis) and that in the treatment model (horizontal axis). Each contour represents the combination of sensitivity parameters for U that lead to the same treatment effect estimate.

Usage

1
2
3
4
sensPlot(x, contour.levels = NULL, col.zero = "red", lty.zero = 1,
       col.insig = "blue", lty.insig = 1, data.line = TRUE, X.pch = NULL, 
       signif.level = 0.05, labcex = 0.75, limit.Xplot = FALSE, txtlab = FALSE, 
	  which.txtlab = NULL,...)

Arguments

x

an object of class sensitivity.

contour.levels

numeric vector of levels at which to draw contour lines. The default is NULL.

col.zero

color of the contour representing the combination of zetas that lead to the treatment effect estimate of 0. The default is "red".

lty.zero

line type of the contour representing the combination of zetas that lead to the treatment effect estimate of 0. The default is 1 (solid line).

col.insig

color of the contour representing the combination of zetas that makes the treatment effect estimate statistically insignificant at a given level. The default is "blue".

lty.insig

line type of the contour representing the combination of zetas that makes the treatment effect estimate statistically insignificant at a given level. The default is 1 (solid line)

data.line

logical. If TRUE a grey contour corresponding to the treatment effect estimate obtained with sensitivity parameters set equal to the coefficients (across all observed confounders) that are farthest from the origin (0,0). The default is TRUE.

X.pch

vector of length 2 giving plotting symbols confounders with a positive association with the outcome and a negative association with the outcome, respectively. The default is c(3,6).

signif.level

this option specifies the statistical significance level at which the significance contour is drawn. The default is 0.05.

labcex

letter size of the treatment effect estimates on the contours. The default is 0.75.

limit.Xplot

logical. If TRUE this option limits the plot region to the minimum and the maximum of the sensitivity parameters and the covariates are plotted on the left or the right end of the figure. If FALSE the contour is extended to the coefficient of the strongest confounders.

txtlab

logical. Label plotted covariates with variable name.

which.txtlab

numeric vector of covariates to include, i.e. c(1:3) shows labels for first three covariates.

...

Other arguments to be passed to all calls to contour

Details

Plots contours of treatment effect estimates under varying combinations of sensitivity parameters, with parameters associated with observed covariates for benchmarking. Options allow highlighting of sensitivity parameters where significance is lost/gained, where treatment effect is reduced to zero, or with treatment effect consistent with most extreme covariate.

Author(s)

Nicole Bohme Carnegie, Masataka Harada, Jennifer Hill

References

Carnegie NB, Hill JH and Harada M. Assessing sensitivity to unmeasured confounding using simulated potential confounders (under review)

See Also

treatSens,plot.default,plot.formula.

Examples

1
#See the manual for treatSens.

treatSens documentation built on May 29, 2017, 10:31 p.m.