I'm having some trouble in wrapping my mind about the prior concentration parameter when calculating a Bayes factor for contingency tables.

Specifically, the parameter prior concentration for package bf_contingency_tab in r refers to the "a" parameter in "Gunel & Dickey, 1974" which I do not have access too.

The only text I can find regarding the prior concentration is from Jamil, Ly, Morey, Love, Marsman & Wagenmakers (2017) and state:

For the matrix of prior parameters a ∗∗ (i.e., the gamma shape parameters of the Poisson rates for the cell counts, see below), a default value is obtained when each a r c =a=1 – in the multinomial case, this indicates that every combination of parameter values is equally likely a priori. Higher values of a bring the predictions of H1 closer to those of H0 ; the prior distribution under a=10, for instance, may be thought of as an uninformative a=1 prior distribution that has been updated using 9 hypothetical observations in each cell of the table. For the data in Table 1, y ..=34, y ∗.=(18,16) a vector of row totals, and y.∗=(11,23) a vector of column totals. When a=1 then a ∗.=(2,2) and a.∗=(2,2). Consequently, ξ ∗. is a vector of ones of length R, the number of rows, ξ.∗ is a vector of ones of length C, the number of columns, and ξ ..=3 . Finally, D() is a Dirichlet function defined in Eq. 5j (Albert 2007; Gunel and Dickey 1974).

What I'm getting from this is that a larger prior concentration lowers the prior belief that h1 is different than h0, is that a correct interpretation?

Further, how would I apply this prior concentration with real data?

For example: If I were to have a contingency table of car accident fatalities when it is and is not raining, and I also had prior proof that fatal accidents are 5 times more likely to occur when it is raining (OR = 5). How would I apply this info to the prior concentration parameter?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.