Course description: This five-week course introduces fundamental concepts and methods for dealing with uncertainty, and for data analysis.
Completion of a self-paced online precourse is required before beginning this course. This course starts with two classes which review the basic language and concepts of probability, including the fundamental laws of probability and conditional probability, discrete and continuous random variables and their probability distributions, means and standard deviations. The third class considers joint distributions of portfolios of random variables. The final two classes provide an introduction to statistical analysis, examining sampling distributions, estimation, and hypothesis testing.
The reason a course like this comes at the beginning of every management program is simple: Whether your primary area of interest is marketing, finance, operations, accounting, human resource management, or some extension or combination of these, you constantly face the need to make decisions in the face of uncertainty. All of the courses that follow this one will use a single common language to discuss risk and uncertainty ... and our objective over the next five classes will be to become comfortable speaking that language. As well, an understanding of the environment in which your decisions will play out is absolutely essential to choosing the best decisions. Careful data analysis can provide such an understanding.
If we lived in a world of certainty, being a manager would be the easiest job in that world. Whenever a decision had to be made, we would simply consider each choice available to us and predict the consequences of that choice, and then implement whichever choice led to the most desirable predicted outcome.
Unfortunately, the world isn't so simple. (Or, perhaps it is fortunate that the world is somewhat more complex, since otherwise managers wouldn't be paid so much!) Two types of uncertainty complicate the job of a manager.
The first is called “randomness” - The outcome of a choice we make might depend on things we can't directly observe or control. Imagine that you're planning a party, and must decide how much (and what types of) food to order. Even if you know how many people you've invited, you still face uncertainties concerning how many guests actually show up, how hungry they will be, and what their food preferences are. You can't perfectly predict the outcome of your ordering decision.
Similarly, a petroleum company, trying to decide whether to drill on a tract of land, doesn't know whether there's an underground oil reservoir under the tract, and, even if there is, doesn't know precisely where it is, or how large it is.
In this drilling example, note that “Nature” has already - millions of years ago - determined the location and size of the oil reserves. Philosophers have argued for centuries over whether it's appropriate to call such things “random,” since they are actually already determinate (known to God, if you will). In all of our discussions, we'll take the view that randomness is subjective: It is in the eyes of the beholder. God might know if there's oil there. So might a competitor who has already drilled an offset well from a neighboring tract of land. But, if we don't yet have specific knowledge, then - to us - the location and quantity are indeed (subjectively) random.
The language, and tools, of probability provide us with a way to analyze decisions in the face of randomness. This is what the first core “decision analysis” class at Kellogg (DECS-430) has as its primary focus.
The randomness we'll be studying will, predominantly, be the consequence of actions taken by “Nature,” a neutral actor with no self-interest. There is another type of randomness, which arises from our uncertainty concerning the actions of other self-motivated decision-makers. This type of randomness is the focus of game theory, which we will discuss briefly in this course (and which is discussed in more detail in subsequent courses, particularly DECS-452). For example, if we were to analyze a two-person poker game, questions concerning the likelihood of hands of various strengths being dealt to one party or the other are pure “probability” questions. Issues involving how to bet, or how to interpret the betting behavior of a competitor, lie in the realm of “game theory.”
The other type of uncertainty that a manager faces is “lack of knowledge” concerning the precise nature of the relationships between actions and their consequences. A brand manager might “know” that demand for a product will be lower at higher prices, but not know the specific relationship between price and demand. This makes it impossible to predict the outcome of a pricing decision.
The field of statistics provides a tool - regression analysis - that can help us estimate the nature of the relationships we face, on the basis of relevant sample data. The second core “business analytics” course at Kellogg (DECS-431) develops and applies this tool.
To summarize, consider the proprietor of a specialty shop deciding how to price umbrellas. Ultimately, the ideal price will depend on local weather and the needs of individual customers (some of whom may have recently lost their umbrellas, or had them break), as well as on the pricing decisions made by other umbrella shops in the area. To make the best pricing decision, the proprietor must deal with probability issues (concerning the weather and consumer needs), game theory issues (concerning how other shops might price their products, or respond to his pricing decision), and statistical issues (concerning how the number of purchasers buying his umbrellas is influenced by the weather, and by his and his competitors' pricing decisions). In order to make an appropriate decision, a manager must be prepared to deal with all of these types of uncertainty.
 God forbid that I use the first paragraph of this section to sweep under the rug two important issues: Managers, in that certain world, would still face two challenges: Determining what “desirable” means, and finding the best policy when there are so many choices available that they can't simply be listed and evaluated one-by-one. We'll discuss these issues, as well, during the course.
 A recent Gallup poll asked adult Americans how they believed the human race originated. 46% of the respondents (other polls show they're mostly Republicans) chose "God created human beings pretty much in their present form at one time within the last 10,000 years or so," the statement that most closely describes biblical creationism. If you don't believe that the Earth is more than 10,000 years old, make up your own example here.