Statistical assumptions and default values

PoolTools is an open-source interface to two R packages.

Analysis of pooled data (estimation of prevalence) is achieved by interfacing with PoolTestR. For more information about the package read see the PoolTestR documentation on GitHub and this article in Environmental Modelling and Software.

Design of surveys (sample size calculations, power calculations, optimisation of designs) is done by interfacing with PoolPoweR.

Important

PoolTools analyses include certain statistical assumptions, which are outlined below. If these assumptions are a poor fit for a specific analysis, using the underlying R packages PoolTestR and PoolPoweR allows the user to change these settings.

Analysis assumptions

When estimating prevalence (with or without stratification), PoolTools calls the function PoolTestR::PoolPrev(). When estimating prevalence from data with a hierarchical/clustering sampling schemes, PoolTools calls the function PoolTestR::HierPoolPrev().

In both cases, PoolTools uses built-in parameters to calculate prevalence, making the following assumptions:

• Default confidence level is 95%
  • PoolTestR parameter: level sets the confidence level to be used for the confidence and credible intervals
  • PoolTools default: level = 0.95
• When all pools are negative, the point estimate is 0
  • PoolTestR parameter: all.negative.pools sets the kind of point estimate and interval to use when all pools in the study are negative.
  • PoolTools default: all.negative.pools = "zero"
  • Explanation: 0 is the point estimate and the lower bound for the interval, and level (see above) posterior quantile the upper bound for the interval.
• The point estimate of prevalence is the posterior mean
  • PoolTestR parameter: robust determines whether the point estimate of prevalence is the posterior mean (when robust = FALSE) or the posterior median (when robust = TRUE)
  • PoolTools default: robust = TRUE
  • Explanation: PoolTools uses posterior median as the point estimate of prevalence

If these assumptions are a poor fit for a particular data set, we recommend using the PoolTestR package instead so each of these parameters can be specified by the user.