Theoretical framework

Fishing and aquaculture in Colombia represent two important sectors of food production for domestic consumption and exports, and they are two multipliers of local economy that contribute to overcoming poverty in rural areas. The country has an important potential for the development of aquaculture, which is based on a great continental and marine water richness, an adequate climate for the cultivation of species, and a wide range of aquatic organisms capable of domestication (^{AUNAP, 2014}).

Fish farming in Colombia brings together multiple economic agents involved in the different activities for production and marketing of the final and intermediate goods in the value chain. These activities are related to fingerling production, fish raising and fattening, fish processing and marketing channels (^{Espinal, Martínez & González, 2005}).

In the fishing chain several problems arise, one of which has to do with production, as it is made in ponds or water mirrors for subsistence fish farming. Other issues that arise in the sub-sector are related to the use of by-products of the fish farming industry, since it is sensitive to a reduction of the purchasing power of the people. Fish farmers are faced with problems of public order, given that most of the raw materials for the production of the concentrates are imported (^{Ministerio de Agricultura y Desarrollo Rural, 2011}).

From the technological point of view, the obtaining of fish derivatives with high or medium technological level is almost nil. The technology applied in the production and commercial management of the fishing resources is deficient, and there is no transformation or handling of byproducts of the Aquaculture industry such as viscera, skins, scales, and heads, among others. These byproducts represent about 12% of the total weight of the production but there is no generation of commercial by-products, except for some sausages and charcuteries; and applications from biotechnology remain scarce (^{Ministerio de Agricultura y Desarrollo Rural, 2012}).

The mentioned problems, accompanied by a scarce or no application of technological management tools, as in the case of foresight studies, do not show an encouraging scenario or evidence of improvement alternatives when marketing products and byproducts at national level. As a matter of fact, the bets on technological innovations in by-products in the fishing sector are not clear, so that specific foresight studies using the Delphi method could help decision-makers set bets on technologies, technological innovations and technology and innovation management activities that can be adopted by stakeholders in the fishing sector.

Foresight is a science that sets its goal in the development of tools to plan knowledge of the future.

Gastón ^{Berger (1967)}, one of the founders of the discipline, defines it as "the science that studies the future to understand and be able to influence it; although sometimes the term futurology refers to other disciplines not based on the scientific method".

According to Jordi ^{Sierra (1992)}, it is defined as "the science that studies the future to understand and be able to influence it. Although in fact it is, paradoxically, a science without object that moves between the need to predict what can happen and the desire to invent the best possible future. Although becoming cannot be accurately predicted, we can imagine our preferred tomorrow".

In turn, Luke ^{Georghiou (1996)} describes foresight as "a systematic means of assessing scientific and technological developments that could have a strong impact on industrial competitiveness, wealth creation and quality of life".

With the contribution from these concepts and many more scholars it can be claimed that foresight is a science that sets its goal in the development and use of tools to know and plan the future, with high risks. It is the foresight of the future with the aim of implementing timely actions in the present, leading us to situations that are longed for and not fortuitous.

In this sense, there are groups of thinkers who have aligned themselves under this concept. Within the schools of foresight is the Anglo-Saxon approach, for which the Delphi method constitutes its core tool. This method presents its own characteristics and differentiates the Anglo-Saxon school from others, such as the French, with emphasis on scenarios, and the Australian, which is more related to social perspective.

The Delphi concept

Linstone and Turoff, quoted by ^{Landeta (2006)}, define the Delphi concept as the "method of structuring a group communication process that is effective in allowing a group of individuals, as a whole, to address a complex problem".

Valdés (1999) and Moráguez (2001), quoted by Moráguez (2006), explain that the Delphi method consists in the systematic use of the intuitive judgment of a group of experts to obtain a consensus of informed opinions. It is essential that these opinions are not manipulated or influenced by the criteria of some other experts.

This technique seeks to get specialists to agree on a problem raised. The method is based, then, on the principle of collective intelligence, which tries to achieve a consensus of opinions expressed individually by a group of people carefully selected as experts qualified in the subject, by means of successive iteration of a feedback questionnaire with the average results of the previous round, with the application of statistical calculations (^{Linstone and Turoff, 2002}; ^{Powell, 2003}).

As for Delphi features, it is a systematic and interactive method that relies on a panel of independent experts, using a number of expert opinions in anonymous communication with feedback; it is a flexible consensus-building tool in which judgments are summarized and resubmitted in order to refine the problem across a wide range of fields (^{Helmer & Rescher, 1959}). According to ^{Rowe and Wright (1999)}, Delphi has four important characteristics: anonymity, interaction, controlled feedback, and statistical aggregation of a group of responses.

Origins of the method

^{Nielsen and Thangadurai (2007)} mention that the Delphi method began in 1940. According to ^{Landeta (2006)} and ^{Dalkey and Helmer (1963)}, its origin is attributed to the RAND Corporation (Research and Development) in the late 1940's in Santa Monica, in a study that was published twelve years later. The method originated from a Cold War study to identify potential U.S. industrial targets and their vulnerability to Soviet munitions (^{Fletcher & Marchildon, 2014}; ^{Rowe & Wright, 1999}; ^{Linstone & Turoff, 1975}).

According to Castelló and Callejo (2000), Delphi is a method that has been used in national exercises of technological foresight; it began to be used by Japan in its five-year prospective exercises in order to define the country's long-term growth direction.

For ^{Abend (2002, p. 32)}, "the Delphi method proposes a question and invites the opinion or solution from a group of highly qualified experts. The process is conducted anonymously".

The Delphi method has had applications in many sectors around the world, with a marked emphasis on foresight studies in technology and innovation (^{Van der Duin, 2006}; Castelló & Callejo, 2000). However, since its beginnings in the 1940's, it has undergone changes in its methodology to such an extent that the applications presented in recent years correspond more to a "modified Delphi", characterized by anonymity, the presentation of different alternatives to consensuses and a smaller number of rounds, among other aspects (Cabero, 2013).

Several applications of Delphi studies around the world, with their purposes, country of origin and number of experts, can be observed in ^{Zartha (2014)}. There are notable applications in Spain in the pharmaceutical sector, probiotics, social sciences and information and communication technologies; in the United Kingdom, in web 2.0; in Finland, in predicting market variables and CO₂ emissions; in Canada, in health and carpal tunnel syndrome; in France, in influenza epidemics; and in the United States, in e-commerce, the Millennium project, developing leadership profiles of new products and online education.

Among the Delphi studies analyzed in Colombia we find works on engineering teaching, biodegradable packaging, nutrition and food policies, sustainable production plans in natural parks, the Colombian electricity sector, the leather sector, footwear and leather goods, the agricultural sector, and construction activities, among others. The analysis of some of the applications of the method in Colombia is presented in Table 1:

Table 1 Applications of the Delphi Method in Colombia 

Source: authors' own elaboration

Regarding the limitations of the method, these are related to the scarce application of new statistical parameters; several studies have shown little rigor in the use of these indicators, such as ^{Amanatidou (2011)}, ^{Chuang, Chia and Wong (2013)}, Castelló and Callejo (2000), ^{Cancelo, Neyro and Baquero (2014)}, ^{Chang, Huang and Lin (2000)}. Other works only use some combination of some statistical parameters, such as mode, average, median, and standard deviation (^{Wakefield & Watson, 2013}; Cabero, 2013; ^{Moore, 2011}; ^{Zeedick, 2012}).

^{Rowe and Wright (1999)} found other weaknesses, such as surface analysis of responses and poor rigor in expert selection. ^{Kauko and Palmroos (2014)} recognize the little emphasis or detail in data analysis between rounds. ^{Fletcher and Marchildon (2014)} refer to the low number of specifications for selecting the winning variables. ^{Cancelo, Neyro and Baquero (2014)} criticize the fact that the method is used only for past analysis and not for estimating future data. ^{Debin et al. (2013)} and ^{Zeedick (2012)} mention the apparent lack of use of indicators to measure the level of participants' expertise.

^{Creange and Careyron (2013)} present results with only one round. As for the importance of conducting a post-poll questioning whether the panelists took into account the responses of others and how many of them contributed comments, ^{Kauko and Palmroos (2014)} draw attention to using panels of panelists or interest groups - stakeholders (^{Hussler et al., 2011}, quoted by Kauko & Palmroos, 2014). Another type of gap is related to the number of panelists; ^{Rowe and Wright (2001}) mention that a range between 5 and 20 people is sufficient.

Identification of the need or purpose

This activity established the need or purpose of the foresight study and the contributions that could be generated to the macro-project "Alternatives for the use of by-products derived from agro-industry fishery - ALTPEZ".

Selection of the monitor team

The monitor team should be made up of people with the capacity to study and investigate the issue in question, and it is desirable to establish specializations in the functions to be performed by each of the group members. The monitor team was selected from the members of the research macro-project ALTPEZ from Universidad del Cauca. A methodological team was also chosen to be part of the monitor team, following ^{Konow and Pérez (1990)}. This methodological team consists of one or more experts in the Delphi Method and complements the group of thematic experts in the monitor team.

The functions of the monitor team are:

Expert profile

The profile of the experts was defined as follows: Professionals, academics and researchers in the field of fish by-products, preferably graduated from Engineering programs such as process, chemical, agro-industrial, and food and biological engineering, who meet the following requirements:

Academics and Researchers:

Entrepreneurs and experts from the public and private sector and interface entities (technological development centers, productivity centers, business incubators, science and technology parks, among others) that meet the following requirements:

Number of experts

A total of 147 e-mails were sent, 115 of which were effective. 37 experts answered both rounds.

Construction of the technological / thematic / innovation tree

For the construction of the first-round questionnaire, it was necessary to carry out technological surveillance on the existing topics, technologies and innovations at a national and international level. To do this, the following search equation was constructed: [TITLE-ABS- 1 (waste OR residues OR by products OR derivatives))]. This equation was monitored in databases specialized in scientific articles such as Scopus and in patents in Freepatens on line with the aim of extracting the themes, innovations and technologies that were included in the format of the first Delphi round.

Construction of the first Delphi Round

With the topics selected from the technological tree, 5 groups were defined, each with their respective themes. In total 143 items were defined.

Objective group

The invitation with the Delphi first-round questionnaire was sent by email to 147 national and international experts from companies in the productive sector, universities, government, public and private sector, among other organizations. 27 email deliveries failed and 5 people responded that they did not have experience in the field, for a total of 115 effective emails.

Answers

The first-round questionnaire was completed by 37 national and international experts who filled out the surveys and sent their responses. This allowed to have the opinion of experts from other countries, especially those with academic, business or research experience in fish by-products or in the two fish species that were the focus of the research (trout and tilapia). Table 2 shows the number of participants by nationality.

Table 2 Number of participants by nationality 

Source: authors' own elaboration

Definition of the priority groups

With the average consensus percentage per group of topics, it was then established that a topic would be considered as a priority in the first round if it had a consensus percentage higher than the average of the thematic group, and a modal value greater than or equal to 4 or 5. With respect to the ratings, the modal value is the most repeated rating within the number of responses.

Subjects with a modal value of less than or equal to 2 and with a percentage higher than the group average were classified as non-priority in the first Delphi round. The other subjects that did not fulfill the previous conditions were considered subjects in discussion.

The classification gave rise to three groups of topics:

Results of the first Delphi round

After the statistical analyses were performed according to the method mentioned in the methodology, we found:

These topics are described in the questionnaire of the second Delphi round. Table 3 presents the priority themes in each group:

Table 3 Priority themes in each group 

Source: authors’ own elaboration

Data sheet

The percentage of participation by experts’ nationality can be observed in figure 2.

NATIONALITY

Source: authors' own elaboration

Figure 2 Expert nationality 

As can be seen, the nationality of the experts who answered the first Delphi round is: Colombia, 84 %; Mexico, 8 %; U.S., 3 %; Uruguay, 3 %; and Brazil, 2 %.

Second round

Priority themes

In total, 54 priority themes corresponding to 38 % of the topics presented in the Delphi method were obtained. Table 4 shows the Delphi results in the first and second rounds.

Table 4 Delphi method results of the first and second rounds 

Source: authors own elaboration

Results analysis

The number of experts who answered the two rounds were 37. According to previous works in Colombia and the world, some Delphi studies with a smaller number of experts have been observed. Internationally there are studies with 10, 15, 7, 30, and 32 experts (^{Wakefield & Watson, 2013});10 experts (^{Kauko & Palmroos, 2014}); 24 and 20 experts (^{Liimatainen et al., 2014}); 32 experts (^{Creange & Careyron, 2013}); and 14 experts (^{Landeta, 2006}). In Colombia, the studies of ^{Velásquez and Castro (2013)} and ^{Zartha et al. (2015}a; 2015b) were performed with 26 and 30 experts, respectively.

It must be highlighted that the exercise included two rounds to avoid "experts' fatigue" and for the generation of results in times that allow to align this study with bets on planning or strategic direction. The three rounds were actually articulated and assembled into two rounds in the present study. This was achieved by allowing to attach justifications or comments to the initial 143 items in the first round (something that in previous exercises was only done in the second round and thus forced a third survey). In this way, in the second round experts could already have arguments for and against, and the change-of-item activity could be done in the same second round. In previous years, ^{Zartha (2015}a; 2015b) had done three rounds to obtain the same results; this prolonged the process, with the additional consequence of not having the same number of experts in the third round as in the first.

Highlighting the importance of issues discussed in the first round but which were considered by the same experts in the second and last round is one of the aspects that is neglected in the results analysis of the Delphi method. This helps to counter criticism of the method in terms of the insufficient round-to-round analysis mentioned by ^{Kauko and Palmroos (2014)}. On this aspect, as regards the issues that were discussed in the first and second rounds of the present study and which were prioritized, we have the following: fish silage meal, in the group ANIMAL FEEDING; physical properties, in the group PROCESSES, in the sub-group Process Parameters; fish hydrolysate and proteases, in the group FUNCTIONAL PRODUCTS; bioaccumulation, in the group BIODIVERSITY AND BIOPROCESSES; competitive intelligence, in the sub-cluster technological foresight and technological surveillance; and measurement of intellectual capital in the sub-group knowledge management. Finally, in the cluster of innovation, technology and knowledge management, it is important to take into account the themes / technologies / innovations for possible bets at an industrial level both in Colombia and in those regions that generate or want to take advantage of fish related products where investment priorities are desired.

Continuing with the analysis between rounds, the following are topics that were prioritized in the first round and discussed in the second round: pellets of fish silage, feeders for fish feeding, silage ensilage of fish, antioxidants and zootechnical parameters, in the group ANIMAL FEEDING; classification of machines by physical characteristics, washing equipment and fish silage equipment, in the machines and equipment sub-group of the group PROCESSES; oil from fish waste, from the group FUNCTIONAL PRODUCTS; hydrolyzed fish protein and bio-oils, from the group BIODIVERSITY AND BIOPROCESSES; group memory, community practice and content management, in the knowledge management subgroup, and innovation systems, R & D evaluation and foresight studies, in the management sub-group, from the group INNOVATION, TECHNOLOGY AND KNOWLEDGE MANAGEMENT. It is important to clarify that the second round gave the experts the opportunity to validate or change their ratings based on the justifications given by the experts, so the result of round two is the final result. Therefore, from the point of view of the experts surveyed, these themes / technologies / innovations do not become the bet or priority for companies or institutions that want to promote these issues.

The topics that were common in the first and second round (total number of topics) are the first bets for companies or future research and technological development centers that want to recommend and generate new productive projects in regions with a fishery orientation and with potential in value-added generation in fish by-products. These topics are detailed in Table 5 in the Decision R1 and Decision R2 columns with the term "priority".