With predictive methods, such as classical density functional theory and predictive density gradient theory (pDGT), it is possible to model bulk phase properties and interfacial tensions using the same model. For nonassociating fluids, these models can be used to predict interfacial properties for systems that lack experimental data. For associating components, however, predictions often show large deviations to experiments, which is at least partially rooted in highly correlated pure component parameters. Therefore, we use interfacial properties for discriminating pure component parameters by amending the PCP-SAFT parameter estimation for water and alcohols by including surface tension data in the objective function. To obtain a comprehensive comparison between different association models, a multiobjective optimization is performed. By analyzing the resulting pareto fronts, it is shown that including a fitted dipole moment improves the results for water but not for alcohols. The result of the multiobjective optimization is inconclusive about the optimal choice of association scheme for water as the preferred model changes along the pareto front. For small alcohols, in contrast to chemical intuition, the 4C association scheme gives the best results. For longer alcohols, the pareto analysis shows the limits of the homosegmented modeling approach.