Using Data for Efficiency and Effectiveness
Read this chapter for a discussion of how business organizations make decisions and the role information technology plays in decision making. What is the process that uses technology to make automated decisions? Consider system input, decision-making algorithms, and output. How does an organization track the decisions made by the system?
Organizational decision making
In any organization, decisions must be made every day. Let's consider a grocery store, as an example. Someone needs to decide how many employees need to be on duty, and what tasks they need to perform (who should be working the check-out lines and self-checkout lines, who should be re-stocking shelves, who should be preparing fresh baked goods, etc.). Decisions need to be made about inventory control issues (e.g. what items need to be reordered, and how many of each), pricing issues (e.g. what items to place on sale, how much to reduce their prices), and so on. In addition to the normal, daily decisions, there are many others that occur less frequently (e.g. whether or not to hire another employee, and if so, whom), and some that occur quite infrequently (e.g. whether or not to open another store in another location). While decision-making environments vary substantially, they typically have one thing in common. More specifically, ready access to good data, information and (in more complex environments) knowledge may lead to better (more efficient and effective) decision making.
There are different views of how decision making does (and should) occur within organizations. We start with a quick overview of the rational perspective, and then briefly discuss alternative views.
The rational view of decision-making describes an ideal situation that some people argue is difficult to achieve. In an ideal world, organizational members charged with making a decision will ask all the correct questions, gather all the pertinent information, discuss the situation with all interested parties, and weigh all relevant factors carefully before reaching their decision. This ideal world rarely exists, however. In reality, most decision makers are faced with time pressures, political pressures, inaccurate or insufficient information, and so on. As a result, many decisions that are made might appear irrational when viewed from the outside. Still, it is useful to employ the rational decision making model as a starting point.
Herbert Simon (a Nobel Prize winner) proposed a model of rational decision making near the middle of the 20 th century. His model includes four stages: (1) intelligence (is there a problem or opportunity?), (2) design (generate alternative solutions), (3) choice (which alternative is best?), and (4) implementation (of the selected alternative). The basic model of how rational decision-making should proceed is:
- Intelligence phase-collect data (and information) from internal and external sources, to determine if a problem or opportunity exists. To the extent possible, ensure that the data are accurate, timely, complete, and unambiguous.
- Design phase-generate possible alternative solutions. Ensure that as wide a selection of alternatives as possible are considered.
- Choice phase-select the best alternative solution. Identify relevant criteria for evaluation, as well as appropriate weighting for each criterion, and use these to objectively weigh each alternative.
- Implementation-perform whatever steps are necessary to put the selected alternative into action.
As an example, think of getting dressed in the morning-you gather intelligence, such as the weather forecast (or by looking out the window). You consider issues such as what you are doing that day, and who you are meeting. You think about alternative clothing options, while considering constraints such as what clothes you have that are clean. You might try on various combinations or outfits, and then make a decision.
Rationality assumes that the decision maker processes all information objectively, without any biases. In addition, this view of rational decision making implies that decision makers have a clear and constant purpose, and are consistent in their decisions and actions.
While many people strive for rational decision making, and intellectually acknowledge that rational decision making is a preferred goal, the realities are often far from this ideal. For example, many decision makers do not take the time to collect all relevant data, nor do they use well-known idea-generation techniques to help them generate a wide selection of alternatives.
In the above example (getting dressed in the morning), you may not go through all of the steps; instead, you might just pull on the first clean clothes that you find, and occasionally will find yourself inappropriately dressed (clothing that is too warm, or too informal for a meeting that you forgot about, etc.).
Behavioral theorists argue that the rational view of decision making is too simplistic. Although decision makers may go through the four phases of the rational model, they do not normally do so in any straightforward manner. Many decision makers, especially those with managerial responsibilities, have work that is fragmented, brief and varied. They face constant interruptions (telephone, email, impromptu meetings), and can only spend a short amount of time considering most decision situations. They are often under considerable stress, which can further degrade their ability to follow a rational approach to making all of the decisions that are needed.
As a result, decision makers often take shortcuts to reduce the burden of making decisions. They typically rely on a network of contacts (both inside and outside of their organization) to help them collect data and to come up with alternative solutions to a particular problem or opportunity. The term "satisficing" has been used to describe a common approach, which is to settle for the first alternative that seems "good enough", rather than expending additional time and effort to collect all available data and to consider all possible options.
In addition, human beings have many biases that can influence their decision making. We may be biased by our upbringing and educational background, by the opinions of persons whom we look up to, by our experiences in similar situations in the past, and so on. We also have personality traits (such as dogmatism, or a lack of creativity, or low willingness to accept risk). These biases and personality traits often keep decision makers from considering a full range of alternatives.
Also, it has been noted that decisions are often reached based on "political" motivations, rather than as a result of rational thought and consideration. For example, a purchasing agent might decide to obtain goods from a supplier whose products and services are inferior (lower quality products, higher price, etc.) than those offered by a competitor, because he knows that his boss is a good friend of an important individual who works for the supplier. While rational decision making implies putting organizational goals above departmental or individual ones, we know that this does not always happen.
In addition to situational factors (too many decisions requiring attention, inadequate data available, not enough time to consider options fully, and so on), decision makers are human beings with limitations. We can only keep so much information available in our short-term memory (which makes comparing options more difficult), we are poor at seeing trends in data, and we are slow (and often inaccurate) in accessing information stored in our longterm memory.
The reason for considering these issues, is to acknowledge they exist and then design information systems that can help us overcome individual and situational limitations, to try and help us move closer to having the ability to use a rational decision making process. For example, designing systems that provide summarized data (with access to more detailed data on demand) makes it easier for decision makers to retrieve the information they need, which increases the probability that they will do so (rather than taking shortcuts). Similarly, we can design systems that help decision makers to see trends in data, to compare multiple options simultaneously, and to provide more transparency in decision making (which reduces the probability of decisions being made for political reasons).
As an example, several large cities exist in dangerous areas-near nuclear power stations or in areas that prone to hurricanes, tornadoes, or floods. City administrators have plans to evacuate in cases of emergency-but think of the complications. The time of the emergency (which will affect where people are and what transportation is available), the amount of warning, the public transportation capacity (roads, railroads, airports), the severity of the emergency, the cost of the disruption-all these need to be considered and possibly traded-off against one another. Ideally, planners will have developed comprehensive plans in advance that will have considered these issues and developed plans to minimize loss of life and economic disruptions. One way to assist these planners is to provide them with robust information systems that help forecast the impact of different possible scenarios, and also to help them weigh trade-offs among competing criteria.
Decision environments (degree of structure)
Obviously, not all situations that require decisions are the same. While some decisions will result in actions that have a substantial impact on the organization and its future, others are much less important and play a relatively minor role.
One criterion that may be used to differentiate among decision situations is the degree of structure that is involved. Many situations are highly structured, with well-defined inputs and outputs. For example, it is relatively easy to determine how much to pay someone if we have the appropriate input data (e.g. how many hours worked and their hourly pay rate), and any relevant decision rules (e.g. if the hours worked for one week are greater than 40, then overtime pay needs to be calculated), and so on. In this type of situation, it is relatively easy to develop information systems which can be used to support (or even automate) the decision.
In contrast, some decision situations are very complex and unstructured, where no specific decision rules can be readily identified. As an example, assume that you have been assigned the following task: "Create the design for a new vehicle that has at least a four-foot long truck bed, is a convertible (with a retractable hard-top roof), gets at least 50 miles per gallon of gasoline, has a high safety rating, and is esthetically pleasing to a relatively wide audience". There is no "optimal" solution to this task; finalizing a design will involve many compromises and tradeoffs, and will require considerable knowledge and expertise.
With this brief introduction, we move to a more detailed discussion of the role of information systems in decision making.