NEWS

freedom 1.1.0

Added unit tests for all functions in the package
Added functionality to the hse_finite function to allow for rounding, floor or ceiling to adjust the estimates for small fractions so that the minimum design prevalence is 1/N which is a reasonable approach in some cases. Thanks to Petter Hopp for this code.

A bug was fixed in the hse function which incorrectly selected the infinite_hse method when the sampling fraction is 1. This is now corrected and when the sampling fraction is greater than the threshold argument, the hse_finite is used as expected. To be clear, this bug probably did not result in errors very often as it should be a rare case that every member of a target population is sampled in a surveillance initiative.

Removed all tests that required values to be identical after a mathematical operation. This resulted in non-identical results when testing on R without long doubles. There were some other places in the code where tests did not fail but could under simple circumstances and this was also improved to allow a tolerance of 1e-7.