Evaluation procedure

The evaluation procedure is based on the AHP (Saaty, Reference Saaty2005), which is capable of translating experienced decision makers' qualitative and quantitative assessments into a multi-criteria classification. The AHP produces weights for each evaluated criterion after decision makers perform pairwise comparisons of criteria and alternatives (Scholl et al., Reference Scholl, Manthey, Helm and Steiner2005; Alfares and Duffuaa, Reference Alfares and Duffuaa2008). Weights of criteria and alternatives are combined for ranking alternatives. AHP is easy to use and scalable, and its hierarchy structure can be easily adjusted to fit many sized problems (Velasquez and Hester, Reference Velasquez and Hester2013). Other MCDM methods (e.g., TOPSIS or Promethee) need to rely on a second tool to compute criteria weights.

Figure 1 shows a flow diagram that represents the entire process of reaching a compromise between the consistent judgments required by the AHP and the knowledge provided by those involved. In the first stage, criteria and sub-criteria to assess the dietary patterns were compared through responses to individual questionnaires answered by a group of stakeholders who represent institutions of the Valencia FPC. Their judgments were then aggregated to find the corresponding weights. In the second stage, a workshop was held in Valencia with a smaller group of experts from the same institutions. These experts were classified into three sub-groups according to the main criteria. For the main goal of identifying a sustainable diet, each sub-group proposed (by consensus) the weights and ranking of the alternatives (or diets) for each criterion or sub-criterion.

Fig. 1.

Flow diagram for acquiring a compromise solution using the Analytical Hierarchy Process (AHP).

The hierarchy of the process is based on three criteria (environmental, health and socio-economic factors) and two pairs of sub-criteria (carbon/water footprint and consumer/producer perspective). This hierarchy divides the main goal into a set of elements (Fig. 2). The three main levels of the hierarchy are the goal, the criteria (factors relating to each alternative that affect the main goal) and the alternatives for which trade-offs are made to reach the goal. As indicated, the main goal is to choose the best dietary pattern from the alternatives that could be recommended or considered as dietary guidelines. The alternatives were:

• • Mediterranean dietary pattern (MDP)

• • Pescatarian pattern (PES)

• • Vegan pattern (VEG)

• • Flexitarian pattern (FLEX).

Fig. 2.

The hierarchy structure for this study.

First stage: criteria's pairwise comparisons

The first stage of the evaluation procedure, after arranging the hierarchy structure for the main goal, was to identify the priorities of the experts (or decision makers) together with the weights to attach to the core criteria. The interviewer explained the methodology to the experts who were compelled to determine the relative importance of each criterion with respect to others.

The criteria were compared as follows: ‘What is the relative importance of each criterion compared to the other one from the decision makers’ point of view?’ This relative importance was in relation to the main goal. In the AHP, a verbal scale is used to enter different judgments for each level but the software requires numbers; therefore, and according to Saaty (Reference Saaty2005), the verbal scale is converted to an ordinal scale. For a given pair of alternatives or criteria A and B, the scale measures the intensity of preference by attaching values between 1/9 (B is extremely more important than A) and 9 (A is extremely more important than B), 1 meaning ‘equal importance’.

If each pair of elements in this row is compared, the number of comparisons is given by Equation (1):

(1)$${\bi N} = \displaystyle{{{\bi n\;}\lpar {{\bi n}-1} \rpar } \over 2}$$

where N is the number of comparisons and n is the number of elements. There were three pairs: environmental impact vs health, environmental impact vs socio-economic factors and health vs socio-economic factors. The same operation was performed for the two sub-criteria within the environmental and socio-economic criteria. This process yields a matrix of priorities or relative weights of criteria, sub-criteria and alternatives. Priorities in the AHP are expressed as numbers between 0 and 1. These numbers reflect relative preference. AHP ratings are assumed to be consistent, in other words, they assumed that decision makers are rational. Saaty (Reference Saaty, Gass and Harris2013) proposed the use of a Consistency Ratio (CR). If CR is greater than 0.1, the judgments should be modified until they reach a CR lower than 0.1. Otherwise, the set of answers must be excluded from the analysis.

A total of 52 experts from different backgrounds were surveyed in the first stage, from which 33 responses could be collected. However, eight were discarded because CR > 0.1, retaining 25 questionnaires (see first stage survey's questions as Supplementary Material). As for the composition of the group with consistent answers, 12 of them were female and 13 were male. Specifically, ten participants were members of research institutions (food technology, nutrition, agricultural sciences and policies), three participants from local public institutions, five participants from the food chain including farmers, co-ops, local catering, food manufacturing and consumers, and seven from NGOs connected to the environment, nutrition and poverty. Super Decision software (v.3.2) was used to create the AHP model.

Grouping the pairwise comparisons is an important step when a group of experts offer judgments and want to reach a consensus, in this case regarding certain dietary patterns. The chosen aggregation method for the first stage (Fig. 1) was based on aggregating individual priorities using the geometric mean method.

Description of dietary alternatives

To define the dietary alternatives considered in the multi-criteria assessment (second stage), adult dietary patterns in the Valencia metropolitan area were taken into consideration. The EAT–Lancet Commission on healthy dietary guidelines for sustainable food systems was used as a benchmark to define a healthy diet (Willett et al., Reference Willett, Rockström, Loken, Springmann, Lang, Vermeulen and Jonell2019). Using the ranges indicated in this benchmark, the four alternative patterns were further adjusted and provided to the experts. Actual dietary patterns (Generalitat Valenciana, 2013) were also provided as a complementary information (Table 1).

Table 1.

Composition and macronutrient intakes (g day⁻¹) for the Mediterranean, vegan, flexitarian and pescatarian dietary patterns

Source: Authors' elaboration based on proposed guidelines for dietary patterns (Willett et al., Reference Willett, Rockström, Loken, Springmann, Lang, Vermeulen and Jonell2019).

The Mediterranean dietary pattern (MDP) is a traditional diet in Mediterranean countries. It is characterized by a high consumption of vegetables, fruits, nuts, seeds, legumes, whole grains, bread, fish, seafood and olive oil. The MDP covers moderate consumption of protein from poultry, eggs, cheese and yogurt. It is rare however to consume red meat and highly processed foods. The vegan pattern excludes the consumption of all animal-based foods, also fish and dairy products. Popular foods amongst vegans include grains, nuts, legumes and beans. The flexitarian diet is an eating pattern that promotes crop-based foods whilst permitting the consumption of meat and other animal-based products in small quantities, thus being more flexible than vegan diets. Furthermore, the pescatarian pattern refers to a vegetarian one that also allows the consumption of fish and other seafood. It relies heavily on crop-based foods such as whole grains, nuts, pulses and healthy fats and seafood is the principal source of proteins. All dietary alternatives must ensure that people's nutritional needs are met. This means eating a set of foods that comply with the daily nutritional guidelines while taking advantage of all available food choices.

Second stage: workshop

The second stage of the fieldwork was carried out with participants from the first stage who attended a workshop to rank the alternative diets according to the criteria and sub-criteria (see second stage questions as Supplementary Material). The reciprocal pairwise comparison matrix among dietary patterns for each criterion, sub-criteria and the main goal was built by consensus. The workshop brought together 14 stakeholders (six female, eight male) who confirmed participation. They previously received some information, alongside the workshop schedule, including, first, a small presentation about the main objectives of the study and the methodology. Secondly, a presentation with the results of the first stage showed the weights of the defined criteria and sub-criteria with its initial rank. The stakeholders were classified into three sub-groups in terms of the three main criteria. The nutrition and health group had four members: two nutritionists from the Red Cross, one food technologist and the manager of a catering company. The environmental group had five members, all of whom were academics: a specialist in LCA applied to food, two specialists in animal science with an environmental focus and another two specialists (one in water management and one in agricultural economics). The socio-economic group had five members: two NGO representatives (Right to Food Observatory and CARITAS), a consultant to farming organizations, a farming organization leader and an agricultural economics PhD student. In this way, the sample reflected the possible interests and backgrounds of different institutions working in Valencia and its metropolitan area. Therefore, the workshop reflected the complexity of the food governance process. After a 40-min discussion in each sub-group, AHP priorities were generated for each criterion and sub-criterion. This was achieved by constructing the reciprocal pairwise comparison matrix, although this time by comparing pairs of alternative dietary patterns for each criterion or sub-criterion. Afterwards, each sub-group had 10 min to express its judgments to the other sub-groups. Additionally, there was a general discussion between the groups to reach consensus.

Findings and discussion

After the first stage, judgments were recorded to obtain the priority (or weight) for each criterion and sub-criterion. The global priorities (i.e., each sub-criterion's contribution to the main goal) were then calculated. Across the hierarchy, the global priorities sum to 1. The results in Table 2 present health, with 44%, as the main priority, but the other priorities are also relevant: environmental concerns represent 33%, while socio-economic factors represent 23%.

Table 2.

Final ranking of the alternatives according to the experts' judgments on sustainability criteria and sub-criteria

FLEX, flexitarian; MDP, Mediterranean dietary pattern; PESC, pescatarian; VEG, vegan.

Source: AHP analysis.

Results showed that, although socio-economic factors were part of this holistic analysis, the valuation of the producer perspective was perhaps not too high. A reason for this outcome is that many stakeholders live in the urban area and are not growers, despite Valencia's rich peri-urban agricultural area. However, some participants work with food producers or in related agricultural research fields. The importance of the producer's perspective would have definitively been different if the process had taken place in a rural area. Therefore, this AHP analysis reflects a decision-making process in the metropolitan area.

The process of comparing pairs of criteria or sub-criteria requires decisions regarding which one is most relevant from the experts' point of view with respect to selecting the most appropriate dietary pattern. This decision can be difficult, but the role of experts is to determine the most important criterion from their point of view, and they must assess the relative importance of one criterion with respect to others.

The final ranking of the dietary alternatives after the workshop (second stage of the study) is shown in Table 2.

The workshop illustrates the trade-offs that policymakers face when designing food strategies in urbanized societies where both cultural aspects and the interests of local production also play prominent roles. MDP appears, according to the process, as the most suitable pattern with respect to the criteria and sub-criteria. The MDP was ranked first in terms of the health criterion and socio-economic criterion. This result is consistent with the studies that focus on the MDP and its health value (Sofi et al., Reference Sofi, Abbate, Gensini and Casini2010; Curtis and O'Keefe, Reference Curtis and O'Keefe2002). Cultural considerations could explain the high weight of the MDP. From the water consumption perspective, the MDP was ranked lowest because of the relative importance of animal production. The experts highlighted other environmental advantages due to lower greenhouse gas emissions related to close proximity of food production to the city and the prominence of fruit and vegetables.

In terms of environment, the vegan diet was ranked first by the workshop experts. Animal products in big cities are typically from intensive farms, which require high quantities of animal feed and drinking water throughout the life cycle of the animals. Even though animals produce manure that can contribute to soil fertility, intensive animal production is a major source of harmful emissions, as indicated by Gerber et al. (Reference Gerber, Steinfeld, Henderson, Mottet, Opio, Dijkman, Falcucci and Tempio2013). Although organic livestock is less harmful to the environment, experts were asked for their judgments based on the actual production systems and technologies used for mass consumption in the city. The pescatarian dietary pattern was ranked second by the environmental experts, who considered two kinds of fish sources. The first refers to wild-caught fish. In this case, although the pescatarian diet has almost no impact on the consumptive use of water, it has a negative impact on carbon footprint because of the emissions from the fishing vessels and the transport to retail outlets. The second source is fish farming, which could also be intensive in freshwater consumption (because of fish feed). However, the experts considered that wild fish is the most consumed in Valencia.

Regarding the health sub-group's judgments, experts considered MDP the most balanced alternative with respect to health recommendations. However, they stressed the similarities between the MDP and both flexitarian and pescatarian diets, which could be considered to some extent modified forms of the MDP. In experts' minds, a balanced vegan diet needs dietary complements to meet the nutritional guidelines. However, the four proposed patterns meet the EAT–Lancet nutritional guidelines and also support the transition toward the environmental thresholds established by Springmann et al. (Reference Springmann, Clark, Mason-D'Croz, Wiebe, Bodirsky, Lassaletta and Jonell2018). Therefore, the AHP exercise suggests an ‘ideal’ model that could be obtained as a weighted combination of the four alternatives with the weights shown in Table 2.

The socio-economic sub-group considered culture, affordability and local production as the main decision elements. Their choice was in favor of products that are heavily produced in the surrounding area. Although the region of Valencia produces some fish from the Mediterranean Sea, most of the fish consumed in Valencia comes from the Atlantic Ocean. The cost of breeding some species of fish consumed heavily in Valencia is still high and the pescatarian diet was the least preferred from the producer perspective. Although fish consumption is significant in Valencia, the sub-group admitted that if consumers relied solely on fish protein, there would not be an economic benefit for the household. Furthermore, a pescatarian diet is not fully consistent with the food culture in Valencia, where people habitually combine meat and fish.