Bradley–Terry model

Bradley–Terry model A model, proposed by Bradley and Terry in 1952, that describes the probability that one treatment is preferred to another. An experiment compares t treatments, two at a time. When two treatments (j and k, say) are compared, the outcome is a statement of which of the two is preferred. The model specifies that there are constants π₁,π₂,…, π_t, with 0≤π_s≤1, for all s, and , which are such that the probability that treatment j is preferred to treatment k is