v0.2.0

Features and functions

• New option to skip calculation of Bayesian estimates in PoolPrev(), substantially speeding up this function.
• New default function values for PoolPrev() and HierPoolPrev() to improve point estimates.
  • New robust parameter has default value robust = TRUE, which means the point estimate of prevalence is the posterior median.
  • Default value for all.negative.pools parameter isall.negative.pools = 'zero', meaning when all pools are negative, the point estimate and the lower bound for the interval will be 0.
• Removed one source of bias from prevalence estimates returned for any hierarchical models, impacting HierPoolPrev() and getPrevalence() output.
• Random effects are marginalised out when calculating population-level prevalence.
  • We no longer support specifying nested surveys using ~(1|Layer1/Layer2)
  • We recommend using the format ~(1|Layer1) + (1|Layer2), which should be equivalent as long as each level in Layer2 is unique
• HierPoolPrev() now returns estimate of intracluster correlation coefficients (ICC) at one or more levels of clustering.
• See the NEWS.md file for full details on what's changed

Development

• Updating github actions by @njtierney in #12
• Update deprecated syntax for future rstan compatibility by @andrjohns in #14

New Contributors

• @njtierney made their first contribution in #12
• @andrjohns made their first contribution in #14
• @caitlinch made their first contribution in #22

Full Changelog

v0.1.3...v0.2.0