Sociology 229A: Event History Analysis

Short Assignment #2: Event History Concepts and Plots

1. Event history analysis is often used to study employment spells. Consider data on an individual’s first full-time job. The event of interest is job termination (for any reason). The hazard function will look quite different depending on the time-clock used to measure start and end times. Consider the following time-clocks: a) An individual’s chronological age, b) duration since start of employment, and c) historical time. For each, write a sentence or two explaining what you think the hazard curve will look like, and make a quick sketch of the hazard rate over time.

2. Using the same example (first job), let us think of the necessary variables to describe one hypothetical spell (start-time, end time, start state, end state, plus one constant variable such as gender and one time-varying variable such as marital status). Write down some hypothetical (i.e., made-up) data to describe one spell in single record format. (How did you decide on a correct value for the time-varying variable?) Write down data to describe another hypothetical spell using multiple-record-per-spell format – e.g., monthly or yearly spell data.

3. Suppose that your interest went beyond the simple question of job termination. Job spells may end in a variety of ways: being fired and becoming unemployed; being promoted to a better job; being demoted to a worse job; quitting to take a better job; quitting to go back to school; quitting to care for children or other family members; etc. You can define events and the overall state space in any way that you wish – ranging from broad (any job termination) to highly specific, or anything in between. The choice is driven by your research question.

Let’s look at one type of event: quitting to stay home and care for children. What would the hazard rate look like (using chronological age as the time-clock)? Make a quick sketch. What variables might have an effect on the hazard rate? Try to list several (3-4), indicating in a few words why you expect an effect. For one of those hypothetical X variables, make a quick sketch of the hazard rate for two different values of that variable (e.g., for education you could contrast people with graduate degrees vs. people with less than high-school education).

Turn in everything – your sketches, hypothetical spell data, and answers (hand-written is OK).