Data Analysis Using SQL and Excel

Chapter 3. How Different Is Different?

The previous two chapters show how to do various calculations and visualizations using SQL and Excel. This chapter moves from calculating results to understanding the significance of the resulting measurements. When are two values so close that they are essentially the same? When are two values far enough apart that we are confident in their being different?

The study of measurements and how to interpret them falls under the applied science of statistics. Although theoretical aspects of statistics can be daunting, the focus here is on applying the results, using tools borrowed from statistics to learn about customers through data. As long as we follow common sense and a few rules, the results can be applied without diving into theoretical mathematics or arcane jargon.

The word “statistics” itself is often misunderstood. It is the plural of “statistic,” and a statistic is just a measurement, such as the averages, medians, and modes calculated in the previous chapter. A big challenge in statistics is generalizing from results on a small group to a larger group. For instance, when a poll reports that 50% of likely voters support a particular political candidate, the pollsters typically also report a margin of error, such as 2.5%. This margin of error, called the sampling margin of error, means that the poll asked a certain number of people (the sample) a question and the goal is to generalize the results from the sample to the entire population. ...