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

plotCIcorrelation(
  corr,
  minN,
  maxN,
  step = 1,
  conf = 0.95,
  minY = NULL,
  maxY = NULL
)

plotCIPearson(
  corr,
  minN,
  maxN,
  step = 1,
  conf = 0.95,
  minY = NULL,
  maxY = NULL
)

Arguments

corr: numeric: observed correlation (assumes Pearson correlation, approximately OK for Spearman)
minN: numeric: the smallest n for which you want the CI
maxN: numeric: the largest n for which you want the CI
step: numeric: the step by which you want n to increment (default is 1), with minN and maxN should give no more than 200 values of n
conf: numeric: the confidence interval you want, defaults to .95
minY: numeric: if you want to set the y axis limits input a number greater than -1 here
maxY: numeric: if you want to set the y axis limits input a number smaller than 1 here

Value

a ggplot object so if you don't capture it it will plot to your default plot device

Background

This is another function like plotBinconf which plots the width of confidence intervals (CIs) for a range of sample/dataset sizes.

History/development log

Version 1: 5.iv.21

Author

Chris Evans

Examples

if (FALSE) {
### simple example for observed 95% confidence interval (CI) around observed correlation of .7
### for sample sizes between 10 and 100
plotCIcorrelation(.7, 10, 100)

### similar but for dataset sizes between 10 and 500 by steps of 5
plotCIcorrelation(.7, 10, 500, 5)

### similar but asking for 99% CI
plotCIcorrelation(.7, 10, 500, 5, conf = .99)

### similar but setting lower y axis limit to -1
plotCIcorrelation(.7, 10, 500, 5, conf = .99, minY = -1)

### and you can set the upper y limit, observed R = 0 for a change
plotCIcorrelation(0, 10, 500, 5, conf = .99, minY = -1, maxY = 1)

### same but exporting to tmpPlot and then changing plot
plotCIcorrelation(0, 10, 500, 5, conf = .99, minY = -1, maxY = 1) -> tmpPlot
tmpPlot +
   ggtitle("95% CI around observed Pearson correlation of zero for n from 10 to 500") +
   theme_bw()

}