The Poisson trick introduces . With an exactly flat log-rate prior, the baseline prior measure is . For ,
Thus flat log-rate marginalization gives the multinomial likelihood:
Alternatively, independent Poisson counts conditional on their total are exactly multinomial distributions with probabilities . Locally uniform coefficient priors make the coefficient posterior proportional to this likelihood within their flat region.
The printed BUGS prior is a proper normal distribution on , with variance , since BUGS uses precision as its second argument. It is broad but not exactly flat. Substitution shows that, apart from a coefficient-independent constant, the integrated likelihood is the multinomial kernel times the Gaussian log-rate correction to the Poisson trick
It depends on the coefficients through . Dominated convergence gives as , so the finite-variance code yields an approximation to the multinomial posterior, not exact equality. Large log rates can make prior sensitivity relevant.
The cell-wise factors form a standard Poisson regression, convenient for BUGS. They avoid an explicitly constrained count vector and permit scalar log-concave updates for intercepts and coefficients; an exactly flat-log-rate implementation also gives gamma conditional baseline rates. This can be computationally efficient, although the actual log-normal baseline prior is not gamma-conjugate and introduces nuisance intercepts. Efficiency depends on the update scheme.
Poisson trick Created 2026-09-24 Updated 2026-11-05
The Poisson trick obtains a fixed-total multinomial likelihood by fitting independent Poisson distribution cells with a free baseline rate for each group. Write with and . For an observed positive group total , maximizing its Poisson likelihood over gives .
Substituting this into the likelihood leaves
This is the conditional multinomial likelihood in . The omitted factor depends on the observed counts, not on , so profile maximum-likelihood estimators and likelihood-ratio tests agree when they are otherwise well-defined. The group baseline must be free; constraints or penalties on it can break the equivalence.
For a group with counts and saturated cell predictors, both approaches fit proportions . If a group total is zero, profiling over strictly positive has its supremum at the excluded boundary ; the conditional likelihood from that group is constant and supplies no information about its proportions. Flat log-rate marginalization gives the multinomial likelihood describes a different operation, integration over the baseline rather than maximization.