Course Outline

list College / Advanced Statistics and Data Science (ABCD)

Book
  • High School / Advanced Statistics and Data Science I (ABC)
  • High School / Statistics and Data Science I (AB)
  • High School / Statistics and Data Science II (XCD)
  • College / Statistics and Data Science (ABC)
  • College / Advanced Statistics and Data Science (ABCD)

4.7 Sources of Variation

This is a good time to think a little more about where variation in data comes from. We already have talked about the DGP—the process that generates variation in the population from which we collected our sample of data. But the DGP includes a lot of different components that, working together, produce the variation we see in an outcome variable. What are these sources of variation?

There are three important points we want to make about sources of variation. First, variation can be either explained or unexplained.

Flowchart of Variation in data with Explained and Unexplained sources

In the word equation we presented before, THUMB = SEX + OTHER STUFF, explained variation is the portion of the total variation we were able to attribute to sex. Unexplained variation is everything included in the “other stuff” part of the equation. It’s useful to think of total variation as the sum of explained + unexplained variation.

Second, unexplained variation can be a real characteristic of the system we are studying, or it can be variation that is induced by our data collection procedures.

The same flowchart of Variation in data as above, extended with addition of Real variation and Induced variation contributing to Unexplained variation

If the variation is real, that means we can probably figure out how to explain it if we measure the right explanatory variables; this variation could be thought of as not explained yet.

Variation induced by data collection comes in three buckets: measurement error (e.g., the small random variation that creeps into our measures); sampling error (i.e., the variation that occurs from sample to sample due to the fact that no individual sample is a perfect representation of the population); and mistakes (e.g., that some students had measured their thumbs in centimeters instead of millimeters).

Same flowchart of Variation in data as above, extended with addition of measurement error, sampling error, and mistakes contributing to Induced variation

All sources of variation in data are represented in the diagram above (which we have adapted from Wild, 2006).

The third and final point we want to make is this: even though unexplained variation could be explained if we knew enough, statisticians tend to model unexplained variation, whether real or induced by data collection, as though it were generated by a random process. Let’s delve into this idea more in the next section.

Responses