Using a change-point model to describe temporal bitter relationships among hop-derived compounds

The bitterness of four hop-derived compounds was measured over a range of concentrations using a time–intensity protocol to determine temporal bitterness relationships. Bitter attributes were compared among compounds by applying a three parameter, change-point model to all concentration-dependent attributes whereby α represented the bitter parameter’s magnitude at very low compound concentrations up to a change-point concentration, θ, and β represented the log-linear concentration dependency of the bitter parameter above the change-point concentration, θ. Panelist reliability was determined by evaluating how well their data fit the model parameters as determined via a t-statistic. Analysis of variance of the concentration-dependent parameters, maximum intensity, duration, area-under-curve, and, decreasing area, identified differences among compounds. In the case of panelist × compound interactions, within-panelist analysis of change-point parameters yielded panel-wise relationships not initially detected.