Deciding Which Analytic Questions to Investigate
Most analysts, with the possible exception of those just beginning in their respective fields, have some choice regarding the questions they study and the projects on which they work. There is a skill to making these decisions — without a doubt, making wise choices regarding which problems to study makes a big difference in an analyst’s career.
Attractive questions and projects possess at least one (and hopefully, several) of the following characteristics:
They have high potential impact.The data necessary are readily available or already in hand.They are of high importance to your organization’s leadership (or they pertain to a “hot” topic in a scientific field). There is relatively little time and/or resources required to get a result.For academics, there are few competitors working on the same question, minimizing the chance you’ll be scooped.There is a relationship to an earlier project you’ve worked on that gives you a competitive edge over other analysts.Completing this project will give you a competitive edge in future attractive projects.For whatever reason, the question is of high intellectual interest to you.
Unattractive analytic questions and projects, unsurprisingly, have the opposite qualities.
Don’t feel as if you must answer every analytic question that comes your way or that you need to follow-up on every idea you have. Your analytic effort is a scarce resource and you need to optimize where you invest your time. For example, it can be an excellent decision to decline a project that would have a small impact on your organization if it will take a lot of time to complete, even if it is certain to provide a definitive result. The most important thing is for you to understand why or why not you’ve chosen to work on a particular project and to make these decisions in a rational way.
In practice, most analysts work on several projects simultaneously. This means that an analyst can put together a portfolio of projects that mixes high-risk high-reward projects with low-risk low-reward projects. Note that low-risk high-reward projects actually exist; you just have to find them! Similarly, high-risk low-reward projects should be avoided.
Once you have data in hand, the range of questions you can investigate gets considerably narrower. How do you decide what questions to ask of the data? Many analysts, when they first look at a data set, ask themselves: “What are the data trying to tell me?” This is exactly the wrong approach. The data are not trying to tell you anything. Many data sets are of poor precision, improperly collected, and more dangerous than they are useful. Sometimes the right answer is to realize you can’t learn anything from a data set, and just toss it out.
To put yourself in a better frame of mind, rephrase the way you ask your analytical questions. A more appropriate question to ask is: “What can I learn from this data set?” Or, even better: “What decision am I trying to make based on these data, and how should this new information influence my decision?”
This post is an excerpt from the recently published textbook Applying Analytics: a Practical Introduction. Find out more at www.applyinganalytics.com and at the book’s amazon page (here). Dr. Levine can be contacted at firstname.lastname@example.org.