Decision Making Processes Research and Analysis
The process of decision making can be split into three different concepts which are intelligence, design and choice. The intelligence is a collection of information of all types that is required within the decision making process. Examples include the modelling details and the parametric details. The design is the design of the alternatives to the decision making. Finally choice is the choice of a particular decision alternative after the evaluation has taken place.
It is to be noticed by figure 1 that there are feedback arrows which shows that it is possible to go back to a particular stage depending on the type of decision made at that particular point. An example of this is that it is possible to go back from the choice phase to the design phase if there are no options available at that time. (Zurn, 2013)
Structured and unstructured decisions
There are two main methods in which a decision can be differentiated, which are programmed and un-programmed decisions. The programmed decisions are structured decisions that occur very frequently, where the structure of a decision allows parameters to be controlled. The non-programmed decisions on the other hand are unstructured decisions which occur a lot less frequently, where it is not possible to fix any parametric values. Therefore the non-programmed decisions are more difficult to implement. An example of this is when entering parametric values into a computer to find the best decision making where the real world can be simulated mathematically. As it is more difficult for the unstructured programming to do this, the management must participate in a more involved manner.
Decision making under certainty, risk and uncertainty
Structured and un-structured methods both come under three types of considerations of decision making conditions. The decision maker faces conditions of certainty, risk and uncertainty as shown in figure 2. It can be seen that the ambiguity and chances of making a bad decision increases from certainty to uncertainty. (Storch, 2001)
The decision making under certainty is more of a structured method in comparison with uncertainty which is more of an unstructured method. Decision making under certainty is where the alternatives and outcomes are known hence decision making is more straightforward as there is no ambiguity. A good example of this is material requirement planning (MRP) where the bill of material is used when purchasing materials. Many of the materials are used to produce specific classes of products which in turn require a certain amount of materials guided by the bill of materials. Therefore a plan is in place to make a certain number of products in a given period of time. The decision to purchase raw material is dependant on the time of purchase as well as the quantity to purchase, and can be connected to the production schedule. Decisions made under certainty would also mean that the demand is known with low variability, hence the inventory costs would be kept to a minimum.
The decision making under risk is where the alternatives and outcomes are known, however the alternatives to the outcomes are known only with the probability associated with them. An example is where there are different alternatives to the decisions where different options are available for a business that would want to progress. The first option is to do nothing and the second decision is to make minor adjustments. The final decision may be to make major changes using the structured method. The three mentioned decision making processes under risk will depend on the natural scenarios that will be present. The first scenario is where the business situation remains how it is thus there is no change within the business environment. The second scenario is there may be a positive change within the business environment. The final scenario is that competition may arise at any point in the future, for example within the next few months. It is these very circumstances that complicate the situation for a decision making process as the probabilities must be assigned to a particular scenario i.e. which alternatives to use for an outcome. The decision tree (figure 3) can be used to analyse the situation a lot more clearly where the nature of the problem is defined within the tree. This is known as the nature node and the probabilities can be assigned to the states of nature. For possible alternatives and states of nature, payoffs can then be estimated. Finally for each state of nature node, the problem can be solved by computing the expected monetary values.
The final decision making is under uncertainty where the alternatives and outcomes are not known with certainty. The number of alternatives may increase and as a result, the number of outcomes will also be higher due to the increase in the possible number of choices. The probabilities associated with alternatives to outcomes are also not known.
Models of Decision Making
There are many models of decision making when faced with a particular decision situation. The first model is rational or normative decision making. When faced with a decision making situation, rational or normative decision makers obtain complex and perfect information. Furthermore certainty is eliminated and everything is evaluated rationally and logically. A decision would then be chosen that serves the interest of an organisation. An example of this is problems with the signalling on railway tracks where this problem can be avoided by collecting and analysing data and looking at the various decision alternatives until the most effective or optimal decision has been initiated. Figure 4 shows the steps to a rational decision making process. (Wu, 2010)
The steps as shown in figure 4 may not be true for the administrative model decision making process. When faced with a decision situation, the administrative or descriptive decision makers make use of incomplete and imperfect information and thus are bounded by rationality, which tend to satisfiy immediate concerns. The administrative model of the decision maker is used to find satisfactory solutions and alternatives rather than the optimal one. An example of this is when listing a car for £1500 and having 10 offers. Unlike the rational model which determines which offer has the highest value in terms of both condition and price, the administrative model would accept the first offer that meets the lowest acceptable price. Although the administrative model has a reduced decision quality, time and effort is saved (Long, 2013).
Simulation Modelling
To mimic what happens in reality, a computer simulation is used where every mathematical model is a simulation. Uncertainty can be modelled directly when there is randomness about a certain aspect of a system. The simulation modelling process involves the following steps:
- Develop the spreadsheet model
- Determine the probability distributions to use for the random inputs to the model
- Modify the model by incorporating the random inputs using probability distributions
- Recalculate the model multiple times to generate possible values of the model outputs
- Analyse output values by considering probability distribution and statistics of the outputs
The simulation itself is very straightforward and flexible for the decision making process. Furthermore, large complex problems can be solved relatively easily, which does not interfere with the real world system and allows the study of relationships. The simulation itself however may be quite expensive and relatively time consuming where the results cannot be generalised to other situations. When running the actual simulations, a good managerial input is required.
Random number simulation
During the simulation experiments, the simulation model requires a system which can generate variable values from a range of probability distributions. A sample is taken from the random variable in the form of a series of generated values that can be formed via a random number generator where the random numbers within the simulation models enables irregular behaviour to occur.
Monte Carlo Simulation
An example of statistical simulation is the Monte Carlo simulation, which is one of the first computer programming applications. This simulation is used for the generation of input data, which are then analysed by means of various regression methods, providing estimations of regression parameters of these data. The problems solved by this type of simulation in turn can be split into three sections, the first of which is when problem solving can be expensive and very difficult. An example of this is when a solution in the form of a mathematical expression cannot be found. The second case is when little is known about the complex problems. The final type is no analytical solutions to the statistical problems. Random numbers and variables are generated for the problem solving.
This theory makes use of the mathematical models in the form of input parameters, and quantitative parameters are used to describe the performance of the system. Heavy assumptions are used within this theory meaning there may be many systems which cannot be solved. Despite this, the system is important because of the various costs for providing service and the costs associated with waiting for the service. Costs are kept to a minimum for customers within the queue, however fast services at a high quality are expensive. On the other hand however, queues can be expensive as customers can leave and so the queuing theory is applied to solve this situation to find the optimum system arrangement. The queue has two important properties; the maximum size i.e. the maximum number of customers who can wait in the queue and the discipline of the queue or the way the queue is organised.
- FIFO (First In First Out) also called FCFS (First Come First Serve) – orderly queue
- LIFO (Last In First Out) also called LCFS (Last Come First Serve) – stack
- SIRO (Serve In Random Order)
- Priority Queue, that may be viewed as a number of queues for various priorities
The service configuration is another aspect of waiting line management where there are four main types of configurations. This is shown in figure 6.
Decision Making Forms
There are three main decision making forms which are the interacting groups, Delphi and the nominal group. The interacting groups consist of an existing group or a new team interacting to make the decision. The Delphi method is used for developing consensus opinions obtained from anonymous experts. There are rounds of questionnaires with feedback of past responses so that a consensus can be achieved. Finally the nominal group is a structured technique that is designed to generate creative and innovative ideas.
References
Long, B., 2013. Decision time and the battle between ‘gut feel’ and data. Grains Research and Development Corporation, Issue 103.
Storch, R., 2001. Decision Making In Organisations. [Online]
Available at: http://courses.washington.edu/inde495/lecc.htm
[Accessed 17 April 2014].
Wu, S., 2010. Models of Organizational Decision Making. [Online]
Available at: http://tx.liberal.ntu.edu.tw/~PurpleWoo/Literature/!Theory/MODELS%20OF%20ORGANIZATIONAL%20DECISION%20MAKING.htm
[Accessed 15 April 2014].
Zurn, J., 2013. Using IS for decision making. [Online]
Available at: http://eternalsunshineoftheismind.wordpress.com/2013/03/05/using-is-for-decision-making-intelligence-and-design-stages/
[Accessed 15 April 2014].