3. Marco Conceptual
3.1. La ciudadanía, un concepto dinámico
Despite the previous section discussing at length the problems associated with Delphi, authors have pointed to the advantages of Delphi. In comparison to the disadvantages, there is a lot more consensus over the advantages of Delphi.
One of the first discussed advantages of Delphi related to the costs of implementation. Helm er ( 1 966) suggested that a Delphi study is relatively inexpensive to conduct and involves much less effort than a conference. A well designed mail questionnaire can elicit information from relatively large numbers of participants who cannot physically come together.
Dalkey ( 1 969) is well known for his support of the Delphi technique. He claims that Delphi is a highly effective experimental structure:
"I can state from my own experience, and also from the experience of many other practitioners, that the results of a Delphi exercise are subject to greater acceptance on the part of the group than are the consensus arrived at by more direct forms o f interaction".
Evidence of the superiority of Delphi over alternative forms o f technological forecasting was provided in a Rand memorandum prepared for NASA - the National Aeronautical Space Administration. In this document it was stated that the Delphi technique may be more capable in the long-range planning pr�cess of NASA than any other known futuristic analytical techniques (Models, Simulation, Worth Relevance Trees, Decision Analysis, Operational Gaming, Systems Synthesis, Scenarios, Cohort Analysis).
"The systematic (and preferably structured) utilisation of experts' opinion and speculation is perhaps the principal and most promising forecasting tool in the technological-scientific-social domain with which we are concerned".
He claims that in the case where value conflicts arise in a group, "Delphi procedures can in principal be most useful for determining its nature and extent".
According to Cyphert and Gant ( 1 970) Delphi is a more socially representative tool because modifications to Delphi have meant that "expert" respondents are no longer limited to "highly educated and experienced specialists" but rather, they can include any type of person who can contribute the relevant information required.
Jolson and Rossow ( 1 97 1 ) pointed to several advantages of Delphi. The technique avoids time consuming and argumentive meetings. Delphi assists the experts in acquiring an integrated overview of the problem area and acts as a catalyst in crystallizing the reasoning process, even in the absence of a group consensus. If the Delphi process is properly managed it can be highly motivating because it's systematic procedures lend an air of objectivity to the outcomes. But Jolson and Rossow suggest that probably the major advantage of Delphi lies in compelling each judge to make explicit what elements of a situation he/she takes into consideration and clarify the concepts he/she uses.
In spite of all his criticisms in the previous section, Weaver ( 1 972) too suggests that the Delphi technique is of value. In particular, Weaver sees Delphi being useful as:
-A method for studying the process of thinking about the future.
-A pedagogical tool (teaching tool) which forces people to think about the future in a more complex way than they usually would. This was also highlighted by Fusfeld and Forster ( 1 97 1 ) who claim that the key advantage of Delphi is that it stimulates thinking and involves management in the forecasting process.
-A planning tool which may aid in probing the priorities held by members and constituencies of an organisation.
Holland ( 1 972) found that of all the intuitive tools used in technological forecasting to obtain estimates from experts, the Delphi technique proved superior to both the committee method and questionnaire method.
Linstone and Turoff ( 1 975) have suggested that one of the strengths of Delphi is its ability to make explicit the limitations on the particular design and its applications. While Delphi {like most research methodologies) has inherent problems, these problems assume greater clarity since Delphi makes the ·communication process and it's structure explicit. Therefore the researcher using Delphi has the advantage of recognising the boundaries of validity. According to Linstone and Turoff, absence of these boundaries turns research into mythology.
Tersine and Riggs ( 1 976) point to the value of anonymity in Delphi. They claim that a participant finds it easier to change his mind if he has no ego involvement in defending an original estimate. In other words, he is less subject to the halo effect and bandwagon effect discussed earlier. Tersine and Riggs claim:
"Delphi encourages individual thinking, forces a panel to get on with the business at hand, and forces respondents to move towards a consensus, unless strong convictions to the contrary are held."
They went on to say that a particular strength of Delphi is its versatility. Originally Delphi was developed to solve problems of a military nature. Since this time Delphi has had numerous applications in fields such as Government planning, education, and business.
Helmer, as quoted in Boucher ( 1 977), provides a defense for the relative inaccuracies of Delphi. He suggests that, because of the pragmatic nature of futures research, its function is primarily predictive rather than exploratory. By forecasting the future
environment and the consequences of alternative plans for coping with that environment, it attempts to improve the decision making process. While it would be good to have a deeper understanding of the underlying causes, the worth of research such as Delphi has to be measured in terms of the quality of the decisions it makes possible rather than of its explanatory force. Helmer goes on to say that the researcher using Delphi constructs an ad-hoc model, fully aware that it is imperfect and in need of later correction as more data and experience is accumulated. Helmer does not claim that such imperfect models produce perfect foresight. It is not correct to reject a Delphi research effort simply because the resulting decisions are often non-optimal, or result in the most desirable outcome. Its objective is to produce the best practically attainable decisions at that time.
Nash ( 1 978) pointed to the non-technical nature of the Delphi technique. His work suggested that it is appropriate for use with a population not familiar with more complex research techniques.
Delphi 's features of response anonymity, controlled information feedback, and statistical group response allow an equal opportunity for respondents to affect the decision-making by the group. This also reduces the likelihood of the "band-wagon" effect or the "halo effect" with highly articulate or high-status committee members (Tersine and Riggs,
1 976; Pill, 1 97 1 ; Helmer, 1 966)
Riggs ( 1 983) suggested that there is also the possibility that the Delphi process may accommodate novel and interesting feedback to respondents, thus minimising the possibility of overlooking some divergent viewpoints. This suggestion is supported by Pill ( 1 97 1 ) and Battersby ( 1 979).
Elliot ( 1 986) argues that the principal benefit of long-range forecasting techniques (such as Delphi) lies not in the product but in the process itself. The real objective of conducting a forecast is not the production of a highly accurate "snapshot" of the industry in the future, but more importantly in the identification of adaptive strategies that organisations might pursue to ensure their continued viability and progress. This
view is also held by Helmer (as quoted in Boucher, 1 977), who suggests that Delphi is the best planning tool when one is faced with imperfect information.
Of central relevance to this thesis are two studies: one by Sack ( 1 97 4) and the other by Elliot ( 1 986). These studies are related to forecasting in banking, and are discussed in Chapter IV. Their research had the central aim of examining the appropriateness of Delphi versus other forecasting techniques in the banking area. Both researchers concluded that the Delphi method was the most appropriate.
5 . 5 ALTERNATIVE APPROACHES TO DELPHI
The reason for forecasting technology is quite simple - to maximise gain or minimise loss from future conditions. Two questions need to be asked. Firstly, are there any alternatives to forecasting? If the answer is no, then one must ask if there are any alternatives to Delphi as a method of forecasting.
Alternatives to Forecasting
Martino ( 1 983) suggested six alternatives to forecasting.
1 ) No forecast: In other words, facing the future blind folded. Obviously an organisation will not survive in these circumstances. Even if the environment is unchanging, this still needs to be known.
2) Anything can happen: This attitude suggests that future changes are completely
random. Nothing can be done to influence the future so, therefore, there is no point in trying to forecast it.
Organisations which run on this philosophy will only survive in the short term as they will be overtaken by competitors who realise that it is more important to attempt a forecast than have none at all.
3) The glorious past: This attitude looks to the past and ignores the future. They assume that actions which enabled them to survive in the past will enable them to survive in the future. Unfortunately, this will not be the case if the future is different
from the past.
4) Window-blind forecasting: This attitude assumes that technology runs on a fixed track (like a roller-blind) and the only way is up. The future will be the same as the past except technology will be 'higher' , ' faster', 'better' , etc. This attitude ignores the fact that a particular technical approach may come to a halt or move sideways if another technical approach supersedes it.
5) Crisis action: This means being 'reactive' - waiting for the problem or crisis to
arrive and then taking action. This action has two assumptions:
*That there will be time to respond effectively.
· *That a forecast would not have assisted in avoiding the crisis. These assumptions may not hold.
6) Genius forecasting: This is more of an alternative to rational and explicit forecasting methods than an alternative to forecasting itself.
This approach simply involves asking a single genius about the future. While sometimes these forecasts may be correct, more often they are incorrect. Where rational and explicit forecasting methods are available, these should be used.
Recitation of the alternatives indicates clearly that, if an organisation wishes to exercise more control over its destiny, there is no real alternative to forecasting. This leads to the next question which is, given the argument posed by this thesis, which is the most appropriate technological forecasting technique? In particular, what are the
alternatives to Delphi? Alternatives to Delphi
1 ) Forecasting by Analogy: This method attempts to compare historical patterns with existing situations in order to forecast future progress and developments. However, there is nothing inevitable about the outcome of a situation. There is no guarantee that if the circumstances are repeated in detail, the outcome will be repeated also. While this method is suitable for the long term (ie longer than two years), problems arise when there is no analogy and situations are unique (Chambers et al, 1 97 1 ) .
2) Curve Fitting: Curve fitting is often done to approximate the basic trend component of a time series. Historical data is required for this method. The identified pattern of the past is used to forecast ahead. However, there is no guarantee that the past will bear a relationship to the future, and the method is less useful for forecasting the long term. In other words, it is difficult to recognise the signs or precursor events that signal a .change of pace or direction. Also, there are difficulties in determining which form of curve will best fit the available data and give an accurate forecast for the future (Makridakis and Wheelwright, 1 978).
3) Trend Extrapolation: One of the limitations of curve fitting previously discussed is that it is not possible to forecast long into the future. This is a situation in which trend extrapolation is useful. A forecast is based on a weighted sum of past observations.
The assumption however is that the aggressive actions which shaped the trend in the past will continue. An established trend may be altered by the introduction of a competitive product or a different technology; a cyclical pattern may be changed by counter-cyclical government policies; a seasonal pattern may be changed by the company's own marketing actions. If it is known that any of these circumstances may occur, another forecasting method is needed - otherwise turning points may be missed. This method is less useful for forecasting the long term.
4) Scenario Development Methods: Scenario writing takes a well-defined set of assumptions, and then develops an imaginative conception of what the future would look like if these assumptions were true. In selecting the most appropriate scenario, the decision maker must determine the viability of the assumptions. However, according to Makridakis and Wheelwright ( 1 978), scenarios should not be used to assess the likelihood of occurrence. Rather, they should be used after the forecasts have been made to draw attention to these possibilities. Chambers et al ( 1 97 1 ) suggests that this is a poor long term forecasting technique.
5) Cross Impact Matrices: With this method, a list of events likely to have an impact on the system being analysed is generated. The probabilities of each of these events happening are then estimated. The conditional probability of event 'A' happening given that event 'B' has happened, for all possible events A and B, is also estimated. From these assumptions it is possible to refine the probabilities relating to the occurrence of individual future developments and their interaction with other developments. This method is more suited for short to medium term forecasts.
The problem with this method is that the probabilities of a particular development is often sequence dependent (ie dependent on a sequence of activity). This increases the magnitude of the problem, but does not change the fact that this technique has found a number of uses in business and government (Ayres, 1 969).
The methodologies explored thus far, including Delphi, are 'exploratory' in the sense that they start with the knowledge and assessments about the past and seek to forecast the future.
The following methods are subjective methods, which are used in circumstances for which there is very little historical knowledge. Owing to the subjective nature of these methods, the reliability of their results is often questionable.
6) jury of Executive Opinion: This simply involves the executives of a corporation sitting around a table and deciding as a group what is the most likely outcome for the
future. One of the main drawbacks of this method is the potential for the 'band wagon' effect. Chambers et al ( 1 97 1 ) suggest that this technique is more suited for short term forecasting.
7) Fonnal Surveys and Market Research Based Assessments: An alternative to using
a handful of experts is to sample the population whose behaviour and actions will determine future trends and activity levels of the item in question. This method has its value, although when applied to this thesis it has less use since it is the hypothesis that the future of banking will be determined more by what bankers consider important (with little consideration for the customer). Also, surveys may tend to probe more for current attitudes than future attitudes. This method of forecasting is more suited for the short to medium term than the long term.
The next collection of forecasting methods examined are normative methods. An exploratory forecast has implicit within it the idea that the capability will be desired when it becomes available. Normative methods, however, start with the future needs and identify the technological performance required to meet these needs.
8) Relevance Trees: Relevance trees use the concepts and methodologies of
decision theory and decision trees to assess the desirability of future goals and to select those areas of development that are necessary in order to achieve the desired goals. Specific technologies can then be singled out for further development (Martino, 1 983). According to Fowles ( 1 975) this technique, especially in its quantitative form, is of limited value in futures research because it requires a good deal of very precise data. 9) Morphological Models: A morphological model is a scheme for breaking a problem down into parallel parts, as distinguished from the hierarchical breakdown of the relevance tree. The parts are then treated independently. For example, possible solutions to the problem 'engine type' may be internal combustion, external combustion, turbine or electric (Martino, 1 983). This technique is more suited to the medium to short term.
10) Mission Flow Diagrams: This involves mapping all the alternative routes or sequences by which some task can be accomplished. All the significant steps on each route must be identified and the researcher can then determine the difficulties and costs associated with each route.
The normative models discussed above have many applications. However, these methods may tend to impose rigidity on the solutions proposed, and they are in no sense a substitute for creativity or imagination.
5.6 AND VALIDITY
Some of the techniques mentioned in the previous section were qualitative in nature. There is a school which questions the validity of qualitative research techniques such as Delphi. In particular the doctrine of positivism holds that:
- science is the only true source of knowledge
- science deals only with that which is able to be observed and manipulated
- human matters should only be investigated using the methods developed for studying the physical world.
Positivism, in its modern form in market research, has come to mean those attitudes which prefer research that is seen as involving a minimum of interpretation and a maximum of facts. These nearly always mean quantitative techniques.
This positivist's view contrasts with humanism because positivists believe that studies of people, in areas such as psychology, sociology, marketing and market research, call for special people-orientated methods, and methods of the natural sciences are incomplete, or even inappropriate.
Positivists' believe that science is a set of specific methods for trying to discover facts about the real world. If there is a theory about how things work, then data can provide an objective test of that theory. Humanists, however, are relativists (ie people who believe that objective knowledge does not exist and that all knowledge is relative to the knower). Data do not provide an objective test of a theory because data are created, at least in part, by theory. All knowledge claims are equally valid and there is no basis on which to make judgements among the various contenders (Gabriel, 1 99 1 ).
It is of use to contrast the views of positivists and humanists when it comes to their views of validity and reliability. The positivists view of validity is best shown by an example. A measuring instrument is said to be reliable if one can get the same answer when using it on different occasions. For instance, instruments for measuring length