plotBinconf: Function outputting a plot of confidence interval around a...

In cpsyctc/CECPfuns: Package of Utility Functions for Psychological Therapies, Mental Health and Well-being Work (Created by Chris Evans and Clara Paz)

Function outputting a plot of confidence interval around a proportion for a range of sample sizes

Description

Function outputting a plot of confidence interval around a proportion for a range of sample sizes

Usage

plotBinconf(proportion, minN, maxN, step = 1, conf = 0.95, fixYlim = FALSE)

plotCIProportion(
  proportion,
  minN,
  maxN,
  step = 1,
  conf = 0.95,
  fixYlim = FALSE
)

Arguments

proportion	numeric: the proportion sought (actual proportion will be nearest possible for each n)
minN	numeric: the smallest sample size, n, to estimate and plot
maxN	numeric: the largest n
step	numeric: the steps to use between minN and maxN, defaults to 1 but set higher if plotting a wide range of n
conf	numeric: confidence interval width, usually .95
fixYlim	logical: if FALSE, ggplot finds sensible y limits, if TRUE, y axis runs from 0 to 1

Value

a ggplot object, by default that will print but you can save it and modify it as you like

Background

This little function just plots confidence intervals (CIs) for a proportion for a range of sample sizes. I wrote it after writing classifyScoresVectorByRCI which will give CIs around observed proportions and I thought that for people not entirely familiar with and comfortable with CIs it might be useful for them to be able to see a plot of how intervals around observed proportions change with sample size. #'

History/development log

Started before 5.iv.21

Author(s)

Chris Evans

See Also

Other confidence interval functions: getCIPearson(), getCISpearman(), plotCIPearson()

Other demonstration functions: plotCIPearson()

Examples

## Not run: 

### 95% CI around proportion .5 for n from 10 to 70
plotBinconf(.5, 10, 70, conf = .95) # don't have to declare conf, defaults to .95
### notice that the observed proportion wiggles up and down as n increases as
### you can only have integer counts so functions gets nearest to the desired
### proportion, here .5, possible for that n, so for n = 10, we can have perfect .5
### but for n = 11 6/11 is .545454..

### 90% CI around proportion .5 for n from 10 to 70
plotBinconf(.5, 10, 70, conf = .90)

### 90% CI around proportion .5 for n from 100 to 200
plotBinconf(.5, 10, 70, conf = .90)

### same but fixing y limits to 0 and 1
plotBinconf(.5, 10, 70, conf = .90, fixYlim = TRUE)

### default 95% CI, exporting to tmpPlot and then changing plot
plotBinconf(.5, 10, 70) -> tmpPlot
tmpPlot +
   ggtitle("95% CI around proportion .5 for n from 10 to 70") +
   theme_bw()

### other inputs
plotBinconf(0, .95, 10, 70)
plotBinconf(1, .95, 10, 70)
plotBinconf(.7, .95, 100, 200)
plotBinconf(.3, .95, 100, 700, 5)


## End(Not run)

cpsyctc/CECPfuns documentation built on April 2, 2024, 2:03 a.m.