Period analysis is a method of aggregation used in demography to capture the entirety of behavior across a human life cycle by examining people at different age groups within a time period. The typical time period used for period analysis is a year.
How it works
The alternative of cohort analysis
Let us say that we want to measure the total number of times a person is expected to do some activity A in a lifetime. A "cohort" analysis would look at a bunch of people who have completed their lives and average out the number of times people in that cohort did activity A. There are two problems with cohort analysis:
- It is heavily delayed, since we have to wait for people's lifetime to finish (or the relevant part of their lifetime where they engage in activity A to finish) to obtain estimates. For measurements that change a lot over time, therefore, the values might come in too late.
- It relies on measurements having been taken in the past; if the measurements weren't taken previously, it may be hard to do a cohort analysis.
How period analysis gets over the problem
We illustrate the period analysis approach by using a period of one year.
Let's say our goal is to estimate the expected number of times a period will do activity A in a lifetime. For a year, we observe how many times people of each age do activity A. Then, we sum up the number (the "age-specific rate" values) across all ages to get a total rate.
The period analysis essentially answers the question: "If a person were to behave, at each age, like the people who are at that age right now, what would the person's lifetime behavior look like?"
The main problem with period analysis is that it does not correspond to the actual estimate of any specific cohort. In particular, in cases where tempo effects are strong, it can be unreliable.
|What we are trying to measure||Cohort measure||Period measure||Female fertility||Completed cohort fertility||Total fertility rate|