error.bars.by: Plot means and confidence intervals for multiple groups

Description

One of the many functions in R to plot means and confidence intervals. Meant mainly for demonstration purposes for showing the probabilty of replication from multiple samples. Can also be combined with such functions as boxplot to summarize distributions. Means and standard errors for each group are calculated using describe.by.

Usage

error.bars.by(x,group,by.var=FALSE,x.cat=TRUE,ylab =NULL,xlab=NULL,main=NULL,ylim= NULL, alpha=.05,sd=FALSE, labels=NULL,pos=NULL,arrow.len=.05,add=FALSE,bars=FALSE,within=FALSE,colors=c("black","blue","red"),lty=NULL,...)

Arguments

A data frame or matrix

group

A grouping variable

by.var

A different line for each group (default) or each variable

x.cat

Is the grouping variable categorical (TRUE) or continuous (FALSE

ylab

y label

xlab

x label

main

title for figure

ylim

if specified, the limits for the plot, otherwise based upon the data

alpha

alpha level of confidence interval. Default is 1- alpha =95% confidence interval

sd=TRUE will plot Standard Deviations instead of standard errors

labels

X axis label

pos

where to place text: below, left, above, right

arrow.len

How long should the top of the error bars be?

add

add=FALSE, new plot, add=TRUE, just points and error bars

bars

Draw a barplot with error bars rather than a simple plot of the means

within

Should the s.e. be corrected by the correlation with the other variables?

colors

groups will be plotted in different colors (mod n.groups)

lty

line type may be specified in the case of not plotting by variables

...

other parameters to pass to the plot function, e.g., typ="b" to draw lines, lty="dashed" to draw dashed lines

Value

Graphic output showing the means + x% confidence intervals for each group. For ci=1.96, and normal data, this will be the 95% confidence region. For ci=1, the 68% confidence region.
These confidence regions are based upon normal theory and do not take into account any skew in the variables. More accurate confidence intervals could be found by resampling.

Details

Drawing the mean +/- a confidence interval is a frequently used function when reporting experimental results. By default, the confidence interval is 1.96 standard errors.

This function was originally just a wrapper for error.bars but has been written to allow groups to be organized either as the x axis or as separate lines.

If desired, a barplot with error bars can be shown. Many find this type of plot to be uninformative (e.g., http://biostat.mc.vanderbilt.edu/DynamitePlots )

Examples

Run this code

data(sat.act)
error.bars.by(sat.act[1:4],sat.act$gender)
error.bars.by(sat.act[5:6],sat.act$gender,bars=TRUE,labels=c("male","female"),main="SAT V and SAT Q by gender")  #draw a barplot

error.bars.by(sat.act[5:6],sat.act$education,bars=TRUE,xlab="Education",main="95 percent confidence limits of Sat V and Sat Q") 

error.bars.by(sat.act[5:6],sat.act$education,TRUE, xlab="Education")  #plot SAT V and SAT Q by education