Specify a log-normal prior

Used for strictly positive multiplicative factors (the multiplicative_covariate effect multiplier). The prior is placed on the natural logarithm of the parameter, so the parameter itself is log-normally distributed with median exp(meanlog) and strictly positive support. Unlike a normal prior truncated at zero, a log-normal has no probability mass piling up against a hard boundary and is the conventional choice for a ratio-scale quantity.

Usage

lognormal(meanlog, sdlog)

Arguments

meanlog

Mean on the log scale. meanlog = 0 gives a median of 1.

sdlog

Standard deviation on the log scale (must be positive). Larger values allow the factor to depart further from 1 in either direction.

Value

A did_prior object.