**6. Practice implications and lessons learned**

The analysts who support the decision process usually face several practical problems concerning the development of the model and the interactions with the actors involved in the process. In the specific case, the activity was limited to the evaluation stage, which is part of a larger decision process; nonetheless, there were some issues related to the presence of several stakeholders and the need to translate the solutions' specific technical features into the criteria, which represent properties that are more general but are strictly connected with the key objectives.

consequence would follow if the alternative "650 m" were implemented, as it does not comply with the European TEN‐T's standards. Further, the layout and track circuits of this solution would increase the congestion in the port terminal. Conversely, the two alternatives "750 m" assure seamless integration of the port in the Baltic‐Adriatic and Mediterranean corridors, the removal of the current bottlenecks and higher economic efficiency. Still, the solution "750 m B" is more flexible than "750 m A" because it enables to operate trains 650 m‐ and 750 m‐long

In order to evaluate the sensitivity of the alternatives' priorities to changes in the stakeholders' weights with respect to the overall goal, some different combinations of weights were tested. The interval of variation was set between 0.2 and 0.6, as these were reputed the minimum and maximum relative levels of importance of the key stakeholders that could be acceptable in the context of the decision. Even if there were some changes in the values of priorities, the overall

As a final step, the estimated capital costs of the alternatives were taken into consideration (**Table 17**, third column). The benefit‐cost ratio, namely the ratio between an alternative's overall priority and the normalized capital cost, was used as an indicator of efficiency. **Ta‐ ble 17** (fifth column) shows the results: the alternative "750 m B" keeps the first position, while "Do minimum" gains two positions. In fact, this second alternative has a cost that is eight times lower than the capital cost of "750 m B," but B's expense is repaid by its high level of overall benefits. The value of the ratio is low for "650 m," and this seems to suggest

At the conclusion of the evaluation, the results were illustrated and discussed with the decision‐makers, who approved the proposal of further investigating and refining the technical contents of solution "750 m B," in order to seize opportunities for funding it.

**Alternatives Overall priority Estimated investment cost (M€) Normalized cost B/C ratio** Do minimum 0.064 6.5 0.046 1.38 650 m 0.136 38.5 0.275 0.49 750 m A 0.276 45.0 0.321 0.86 750 m B 0.524 50.0 0.357 1.47

The analysts who support the decision process usually face several practical problems concerning the development of the model and the interactions with the actors involved in the process. In the specific case, the activity was limited to the evaluation stage, which is part of a larger decision process; nonetheless, there were some issues related to the presence of several stakeholders and the need to translate the solutions' specific technical features into the criteria,

that it should not be taken into consideration in the future stages of the project.

without increasing the occupancy of tracks or delaying train dispatching.

136 Applications and Theory of Analytic Hierarchy Process - Decision Making for Strategic Decisions

ranking proved to be stable.

**Table 17.** Overall results.

**6. Practice implications and lessons learned**

The key stakeholders had different roles (see Section 3), were involved in the project at different stages, and their representatives had different levels of expertise and knowledge of the problem. Such situation is not infrequent in strategic transport decisions and should be carefully managed. The opportunity to have an open and informative discussion with the stakeholders, and the quality of the information obtained can be delicate aspects. The case study confirms that it is necessary not only to identify the key stakeholders but also to individuate their representatives and involve them at an early stage of the evaluation process. However, the authors of this study observed that there is a relation between the interest and role of the stakeholder in the process (in particular, if a stakeholder is the promoter of one or more solutions) and the quality of the information retrieved from their representatives.

The interview of the actors is a key activity in multi‐actor evaluation and should be carefully planned and, possibly, tailored to the respondent role (expert, policy maker etc.). In the specific, a structured interview was arranged only for the key stakeholders, while the experts were consulted in the context of group discussions within the project team. On the one hand, the interview allowed the respondents to focus their attention on the topics and their actual contribution to the evaluation; on the other, they did not always feel at ease with making comparisons between pre‐determined criteria and asked for examples that could better describe their practical meaning. Not surprisingly, the representatives of the stakeholders involved in the NAPA project had less problems in the assessment.

The last observation is also valid for the experts of the project team. Still, some methodological issues were highlighted. Firstly, even if the separation between stakeholders' weighting of criteria and experts' judgment of the alternatives' performance was considered in line with the "political" and technical contents of the evaluation, an open discussion involving both figures could have been effective in exchanging viewpoints and circulating information. Secondly, the need to subsume many technical features of the solutions under broad criteria was considered arguable. In some cases, the association of the characteristics (such as track configuration, capacity, transit times, etc.) with a specific criterion was not immediate; further, this approach seemed to negatively affect the transparency of the evaluation. The opportunity that AHP offers for using both pairwise comparisons and direct measures could have been seized, and it could have satisfied some experts who stress the difference between "objective" and "subjective" measures of performance. However, this would have implied an additional effort to select the measures of performance that were meaningful for a criterion and non‐redundant; in addition, the development stage of the project could not assure that the needed data were available. On these grounds, it was decided to employ the original methodological solution, based on pairwise comparisons. The last issue that was raised by the experts concerned the possible interdependencies among several factors (criteria or features), which were not considered in the AHP model. It was acknowledged that the model was a simplified repre‐ sentation of the problem: other methods that can deal with interdependencies, such as the analytic network process [36], should be used if a more refined evaluation was needed at the cost of a greater intellectual effort and a broader participation of experts and stakeholders.
