Chapter 28 Adjustment and Control
In this chapter we will (eventually) discuss adjusting data for trends or baselines, and relate that to “controlling” for variables.
This is inspired by the problem of, for example, using BART data to assess how many people took BART to a game on Wednesday. The naïve student would sum the number of riders over a period of a few hours. That’s good, but you should subtract the number of people who would be taking BART at that time anyway. And how do you determine that? Some ideas:
- Look at a nearby non-game day.
- Look at a non-game Wednesday.
- Look at a lot of non-game Wednesdays.
Adjustment will address ideas such as “adjusted for inflation” and “seasonally adjusted”— means of making fair comparisons when underlying trends change the data values.
In that section we’ll also get to address calculations like “tall for your age” or “warm for February.”
xxx slicing with filtering as a way to control variables