Etusivu › artikkeli › Artikkelit › 7. Delphi Analysis

7. Delphi Analysis

Kirjoittanut , • ( 0 )

Scientific analysis is a technique that allows data collected reliably to be organized in such a way that the essential meanings and connections of the phenomenon under examination are made visible in proper proportion. Typically in Delphi methods, distributions are examined quantitatively, and commentary with justifications is provided qualitatively. A particular feature of the Delphi method is that data is classified and evaluated during the research process, not only at the end.

Between Delphi rounds and at the end, results are analyzed and condensed information about the results is also shared with the panel for use. The content of the research report varies from descriptions of results to action and decision proposals and from question-specific analyses to scenarios. In the final stages of the process, two different paths branch out, the original one being the pursuit of consensus. In Consensus Delphi, the aim is often in direct decision-making preparation, where through panel work, a recommendation for action or a decision proposal is reached. In strategy preparation, two or more alternatives may also be prepared, ensuring that the discussion expands into the decision-making situation itself.

In scientific Delphi research, the consensus approach has been used, for example, in defining pharmacological or public health quality measures. In the Finnish context, consensus is less common than the use of Delphi to envision plural futures, where the starting point is the dissensus of panelists and is not sought to be homogenized during the process. Instead, the aim is to strengthen the justification of different viewpoints, i.e., argumentation. Based on well-justified futures, it is natural to write consistent future scripts or scenarios, whose probability and desirability can be tested. In the National Skills Anticipation Forum, four types of Finland are described for the year 2035. Scenario planning is a valid option when the goal is to increase readiness for unexpected futures, i.e., situations where the unlikely becomes likely or the desirable becomes undesirable. The attachment provides a more detailed description of the scenarios.

Each Delphi draws a model of what possible ”landscapes” we are moving towards regarding some aspect of the future. The Delphi process can be designed as a long-term research or monitoring program, where a map of the phenomenon’s future is consciously formed, detailed not only round by round but also time-wise like in the Finnish National Board of Education’s Future of Learning 2030 barometer.

The manager’s first and intuitive analysis occurs through the visualization of the response distribution (diagram), which reveals at a glance how the panelists’ views are distributed. In Delphi programs, such feedback is often already available in real-time for the panelists. It reflects how accumulated data is recycled and analyzed during the process itself. Disagreement is more interesting than agreement in Dissensus Delphi. On the other hand, agreement means that the research result can be obtained easily and quickly. Reading the comments (arguments) reveals what factors and perceptions lie behind the accumulating or dispersing distribution.

The view of distribution initiates a qualitative analysis. At its simplest, this involves examining how comments and the arguments they contain vary based on the scale response. Identifying differences is facilitated if the Delphi software can sort comments in descending order according to the response option. For non-scale questions, classification is different but equally separating. Each question’s important function is to create pressure to justify one’s own answer, whose wide range indicates the success of the question setting. The analysis involves observing and interpreting these differences. In exploratory research, a ”handcrafted” light analysis suffices, in which the relationships between three different variables are clarified. These variables are the classifying response (e.g., on a scale), qualitative content related to the classifying response, and the relationship between the respondent’s competence (and interest) and the responses.

A simple classification of comment data is to place comments in a four-field matrix. For example, a future claim with two criteria directly produces a four-field matrix. In this case, the question is set according to two criteria defined by the managers on an illustrative multi-tiered Likert scale. In the example of the image, the options are simplified to four: minus-minus, minus-plus, plus-minus, and plus-plus. For example, the eDelphi software (www.edelphi.org) depicts responses – both distributions and comments – two-dimensionally according to the likelihood and desirability in a four-field matrix.

The dimensions of each future thesis are likely vs unlikely and desirable vs undesirable. These are coordinates where desirable represents something sought after, undesirable something to avoid, likely a possibility, and unlikely an impossibility. The general endeavor is to navigate towards the desirable, avoid the undesirable, and if necessary, clear space for the likely from the unlikely. Future statements are selected so that each indicates a phenomenon wider than its concrete content. Panelists have weighed in on the likelihood and desirability of the claim on a Likert scale. The four-field technique schematically offers the study of four possible futures (futuribles = future possibilities).

In the web-based Delphi process, the communication environment can be updated so that voting and argumentation occur in real-time. Voting distributions can be made visible if desired, as well as ”for and against” arguments. For each argument, its significance (relevance) and validity can be assessed. Usually, the desirability (desirability) and feasibility (likelihood) of the statements are questioned.

The greater the dispersion of the statement, the more divided the panel is concerning the question and vice versa. In Consensus Delphi, the aim is to bring the dispersion value close to zero. In Dissensus Delphi, the dispersion value only matters in analyses. The manager does not need to create pressure in any direction other than that the panelists are open to revising their stance when they have good reasons for it. It is crucial to get the most relevant arguments for and against different viewpoints.

In examining the data produced by the Delphi method, it is essential to recognize that it always rests on subjective knowledge even though it concerns expert knowledge. The promise – or purpose – of Delphi based on argumentation is not to tell a certain future. Instead, it allows exploring what experts think about the future – thus uncovering experts’ perceptions and assumptions. Delphi does not settle for mere opinions but requires their substantiation. Argued knowledge is useful for decision-making background or as building blocks for scenarios, not to mention that the process can ideally shape into a learning event producing new understanding for both panelists and researchers. The hallmark of a knowledge-producing and creating panel is that dialogues are conducted between different views, and as a result of argumentation, there is also a readiness to change original perceptions during the Delphi process.

In the Future of Learning 2030 barometer, a traffic light metaphor was developed to describe how the state of different future theses varies and dynamically transforms over time. The assumed claim evolution is that the controversy phase, or polarization, triggers development, resulting in a changed paradigm or finding a third path that is not about either of the original opposites. Often, as polarization continues, it leads to commenting, argumentation, and discussion, where initially opposing views diversify and enrich into perspectives that allow listening and dialogue. As the dialogue continues, the readiness for a shared overall vision or at least an acceptable compromise increases. When consensus is sufficient, the thesis is in a resolution state.

In the dispute state, panelists’ views and arguments about future development are often polarized into emotionally charged, stark stances, which prevent hearing and listening to the opposing side’s reasoning. In the dialogue state, perceptions of future development are divided into many perspectives and possibilities, which can be examined side by side and complement each other, facilitating the transition to the discussion phase. In the resolution state, the panelists are in such agreement on the future statement that there are great opportunities to formulate a shared view and decision on future development.

In social changes, there are cultural inertial forces resulting from the defense of interests, skills, and institutional positions. Especially the near future landscape is ”mined” by these forces. Conversely, if one wants to examine the future with a focus on change, it is advisable to extend the timeframe so far that the intensity of interests diminishes and thinking opens up to even major disruptions. Our experience is that the boundary between near and distant futures runs around ten years. Change in itself is neither good nor bad. However, in societal issues, there are more advocates for status quo than for changes whose beneficiaries are not yet recognized.

Controversial issues, like all future issues, arise in this moment and space, but dealing with and resolving them is much easier in the future than in the present or even in history, although this also happens through reinterpretations of history. Anticipation theory is a way of thinking that utilizes this wisdom. Simplified, it involves taking a current issue to the future to be resolved and then bringing it back to the present. The value of this thinking lies in gaining a perspective on the natural direction of development. Without anticipation, decisions are based not only on current interests but also partly on pressure from history to act as before.

In scenario planning, it is natural to use the previously described question-specific four-field or quadrant grid. The surveys for the Competence Anticipation Forum 2035 panels were built according to a socio-dynamic multi-level model so that they cover the historically shaped situation of the phenomenon (trends, regime), pressures and changes created by the operational environment (megatrends), and reorganization enabled through technological or social innovations (niche, weak signals). A commonly used classification for structuring scenarios is PESTE, which covers the most important societal action sectors according to the initials: politics (decision-making), economy, social factors, technology, and ecology or the environment.

Ensuring the reliability of the research is important in all phases of the Delphi process. The results of the research are worthless if they cannot be trusted. The evaluation of reliability starts with the selection of panelists. They must represent different perspectives on the research subject. The survey must cover the essential features of the research subject within the scope of the research, and the questions must not be leading. Different methods can be used in collecting research data, e.g., the first round of Delphi can be conducted as an interview and the third as a workshop. The analysis also describes how the analysis has been performed.

The final resting place for analyses is the research report. Scientific report and article formats are quite standardized. Each university has its own guidelines, but the format varies minimally. In the Delphi developer community, thesis projects have been supervised at a rate of about twenty per year in the 2000s. Nearly 30 doctoral theses have been completed at various universities and faculties in the 2000s. The year 2020 was a record year for doctoral level studies when five were published from five different disciplines and universities.

During the research process, numerous decisions are made that must be described. One must select a research topic, define a suitably narrow research problem, choose an approach, key concepts, models, theoretical framework, and research methods. Then the empirical data is selected if the question concerns empirical research. After that, results are produced. That too is a kind of choice. Finally, the results’ interpretation or conclusions are processed.

The report presents the results of the research process, which can also be considered a set of justified hypotheses or claims. (Kakkuri-Knuuttila 2001) The main claim of the scientific text is its main claim or the main result of the research. The main claim is the researcher’s contribution to the scientific discussion. The task of the research report is to make the main claim credible so that readers become interested in it and are willing to believe it. The research and writing process can be understood as a gradual progression towards the main claim and the production of its justifications. The report must also describe how the research findings were reached in a way that allows the assessment of the validity and reliability of the results, preferably so that the empirical parts can be verified or falsified by repeating them. In a scientific context, justification and argumentation are referred to by the term substantiation.

Attachments

Finland Scenarios: In the Steady Finland, there is a caution against rocking the boat. Established practices and traditional institutions are favored, and risks that could arise from too rapid changes in social relations are avoided. A strong nation-state is the best instrument to maintain balanced development. In Turbo Finland, the focus is primarily on results, competitiveness, and productivity, which do not require regime-level upheavals. Instead, it requires a strong European Union, which best secures the interests of a small country in the global market, and to which Finland, even as a small country, rightfully belongs. Ecological Finland has evolved under the pressure of global environmental change. The idea is to harmonize ecology and economy in a way that benefits both. However, unlike the current order, ecology determines the economy. Diverse Finland represents the phase of disintegrating states, where the world is organized through regions and networks. Experiences of participation and sharing rejuvenate democratic institutions and energize community activities in regions.

Two of the scenarios represent continuity (business as usual) and two represent change. Another criterion separates the scenarios based on how active societal reaction is. In the Yellow scenario, there is a reactive stance towards existing political and economic institutions, in the Green scenarios, reactivity is directed at environmental issues. In the Blue and Red scenarios, the reaction is anticipatory.

Literature

‹
›

Kategoriat:Artikkelit, blogi, Tie

Avainsanat:, , , , ,

Jätä kommentti

This site uses Akismet to reduce spam. Learn how your comment data is processed.