sample.size: Determine the Optimal Sample Size for a Given Population

In mrdwab/mrdwabmisc: Miscellaneous R functions, mostly for data processing

Description

The sample.size function either calculates the optimum survey sample size when provided with a population size, or the confidence interval of using a certain sample size with a given population. It can be used to generate tables (data.frames) of different combinations of inputs of the following arguments, which can be useful for showing the effect of each of these in sample size calculation.

Usage

sample.size(population, samp.size = NULL, c.lev = 95, c.int = NULL,
  what = "sample", distribution = 50)

Arguments

population	The population size for which a sample size needs to be calculated.
samp.size	The sample size. This argument is only used when calculating the confidence interval, and defaults to NULL.
c.lev	The desired confidence level. Defaults to a reasonable 95%.
c.int	The confidence interval. This argument is only used when calculating the sample size. If not specified when calculating the sample size, defaults to 5% and a message is provided indicating this; this is also the default action if c.int = NULL.
what	Should the function calculate the desired sample size or the confidence interval? Accepted values are "sample" and "confidence" (quoted), and defaults to "sample".
distribution	Response distribution. Defaults to 50% (distribution = 50), which will give you the largest sample size.

Note

From a teaching perspective, the function can be used to easily make tables which demonstrate how the sample size or confidence interval change when different inputs change. See the "Advanced Usage" examples. The following formulae were used in this function:

- p))/(c^2)

1)/pop))

Author(s)

Ananda Mahto

References

• See the 2657 Productions News site for how this function progressively developed: http://news.mrdwab.com/2010/09/10/a-sample-size-calculator-function-for-r/

• The sample.size function is based on the following formulas from the Creative Research Systems web page Sample size formulas for our sample size calculator: http://www.webcitation.org/69kNjMuKe

Examples

# What should our sample size be for a population of 300?
# All defaults accepted.
sample.size(population = 300)

# What sample should we take for a population of 300
#   at a confidence level of 97%?
sample.size(population = 300, c.lev = 97)

# What about if we change our confidence interval?
sample.size(population = 300, c.int = 2.5, what = "sample")

# What about if we want to determine the confidence interval
#   of a sample of 140 from a population of 300? A confidence
#   level of 95% is assumed.
sample.size(population = 300, samp.size = 140, what = "confidence")

## ========================================= ##
## =========== ADVANCED USAGE ============== ##

# What should the sample be for populations of 300 to 500 by 50?
sample.size(population=c(300, 350, 400, 450, 500))

# How does varying confidence levels or confidence intervals
#   affect the sample size?
sample.size(population=300,
       c.lev=rep(c(95, 96, 97, 98, 99), times = 3),
       c.int=rep(c(2.5, 5, 10), each=5))

# What is are the confidence intervals for a sample of
#   150, 160, and 170 from a population of 300?
sample.size(population=300,
       samp.size = c(150, 160, 170),
       what = "confidence")

mrdwab/mrdwabmisc documentation built on May 23, 2019, 7:15 a.m.