Download presentation

Presentation is loading. Please wait.

Published byJenna Sutterfield Modified over 2 years ago

1
What is Event History Analysis? Fiona Steele Centre for Multilevel Modelling University of Bristol

2
Overview Example applications of EHA Event history data and possible sources Methods of analysis with application to timing of 1 st partnership –Descriptive analysis –Event history modelling Further topics 2

3
3 What is Event History Analysis? Methods for analysis of length of time until the occurrence of some event. The dependent variable is the duration until event occurrence. EHA also known as: Survival analysis (particularly in biostatistics and when event is not repeatable) Duration analysis Hazard modelling

4
4 Examples of Applications Education – time to leaving full-time education (from end of compulsory education); time to exit from teaching profession Economics – duration of an episode of unemployment or employment Demography – time to first birth (from when?); time to first marriage; time to divorce Psychology – duration to response to some stimulus

5
5 Types of Event History Data Dates of start of exposure period and events, e.g. dates of start and end of an employment spell –Usually collected retrospectively –UK sources include BHPS and cohort studies (partnership, birth, employment, and housing histories) Current status data from panel study, e.g. current employment status each year –Collected prospectively

6
6 Special Features of Event History Data Durations are always positive and their distribution is often skewed Censoring – there are usually people who have not yet experienced the event when we observe them Time-varying covariates – the values of some covariates may change over time

7
7 Censoring Right-censoring is the most common form of censoring. Durations are right-censored if the event has not occurred by the end of the observation period. –E.g. in a study of divorce, most respondents will still be married when last observed Excluding right-censored observations leads to bias and may drastically reduce sample size Usually assume censoring is non-informative

8
8 Event Times and Censoring Times

9
9 Key Quantities in EHA In EHA, interest is usually focused on the hazard function h(t) and the survivor function S(t) h(t) is the probability of having at event at time t, given that the event has not occurred before t S(t) is the probability that an event has not occurred before time t

10
10 Life Table Estimation of h(t) Group durations into intervals t=1,2,3,… (often already in this form) Record no. at risk at start of interval r(t), no. events during interval d(t), and no. censored during interval w(t) An estimate of the hazard is d(t)/r(t). Sometimes there is a correction for censoring

11
11 Estimation of S(t) Estimator of survivor function for interval t is E.g. probability of surviving to the start of 3 rd interval = probability no event in 1 st interval and no event in 2 nd interval

12
12 Example: Time to 1 st Partnership Source: Subsample from the National Child Development Study

13
13 Example of Interpretation h(16)=0.02 so 2% partnered at age 16 h(20)=0.13 so of those who were unpartnered at their 20 th birthday, 13% partnered before age 21 S(20)=0.77 so 77% had not partnered by age 20

14
14 Hazard of 1 st Partnership

15
15 Survivor Function: Probability of Remaining Unpartnered

16
16 Introducing Covariates: Event History Modelling Assumptions about the shape of the hazard function Whether time is treated as continuous or discrete Whether the effects of covariates can be assumed constant over time (proportional hazards) There are many different types of event history model, which vary according to:

17
17 The Cox Proportional Hazards Model Makes no assumptions about the shape of the hazard function Treats time as a continuous variable Assumes that the effects of covariates are constant over time (although this can be modified) The most commonly applied model which:

18
18 The Cox Proportional Hazards Model h i (t) is hazard for individual i at time t x i is a covariate with coefficient β h 0 (t) is the baseline hazard, i.e. hazard when x i =0 The Cox model can be written h i (t) = h 0 (t) exp(βx i ) or sometimes as log h i (t) = log h 0 (t) + βx i Note: x could be time-varying, i.e. x i (t)

19
19 Cox Model: Interpretation exp(β) - also written as e β - is called the relative risk For each 1-unit increase in x the hazard is multiplied by exp(β) exp(β)>1 implies a positive effect on hazard, i.e. higher values of x associated with shorter durations exp(β)<1 implies a negative effect on hazard, i.e. higher values of x associated with longer durations

20
20 Cox Model: Gender Differences in Age at 1 st Partnership The hazard of partnering at age t is 1.5 times higher for women than for men. So women partner at an earlier age than men. We assume that the gender difference in the hazard is the same for all ages.

21
21 Discrete-time Event History Analysis Event times are often measured in discrete units of time, e.g. months or years, especially when collected retrospectively Before fitting a discrete-time model we must restructure the data so that we have a record for each time interval

22
22 Discrete-time Data Structure

23
23 Discrete-time Model The response variable for a discrete-time model is the binary indicator of event occurrence y i (t). The hazard function is the probability that y i (t)=1. Fit a logistic regression model of the form:

24
24 Discrete-time Analysis of Age at 1 st Partnership FEMALERespondents sex (1=female, 0=male) FULLTIME(t)Whether in full-time education at age t (1=yes, 0=no) We need to choose the form of α(t). Try: step function (dummy variable for each year age 16-33) quadratic function by including t and t 2 Choice will be guided by plot of the hazard function.

25
25 Duration Effects Fitted as a Step Function Reference category for t =age 16

26
26 Duration Effects Fitted as a Quadratic Exp(B) are (multiplicative) effects on the odds of partnering at age t Women partner more quickly than men. Enrolment in full-time education is associated with a ( ) 100=68% reduction in the odds of partnering, i.e. a delay

27
27 Non-proportional Hazards So far we have assumed that the effects of x are the same for all values of t It is straightforward to relax this assumption in a discrete- time model by including interactions between x and t in the model The following graphs show the predicted log-odds of partnering from 2 different models: 1) the main effects model on the previous slide, 2) a model with interactions t*female and t2*female added.

28
28 Proportional Gender Effects

29
29 Non-proportional Gender Effects

30
30 Further Topics Repeated events, e.g. multiple marriages or births –Multilevel modelling Competing risks, e.g. different reasons for leaving a job (switch to another job, redundancy, sacked) –Fit set of logistic regression models or a single multinomial model Multiple states, e.g. may wish to model transitions between unpartnered, marriage and cohabitation states –Include dummies for state, and interact with duration and covariates Multiple processes, e.g. joint modelling of partnership and education histories

31
31 Some References Singer, J.D. and Willet, J.B. (1993) Its about time: Using discrete-time survival analysis to study duration and the timing of events. Journal of Educational Statistics, 18: Blossfeld, H.-P. and Rohwer, G. (2007) Event History Analysis with Stata. Mahwah (NJ): Lawrence Erlbaum. Steele, F. (2005) Event History Analysis. NCRM Review Paper NCRM/004. Downloadable from Steele, F., Goldstein, H. and Browne, W. (2004) A general multistate competing risks model for event history data, with an application to a study of contraceptive use dynamics. Statistical Modelling, 4:

Similar presentations

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google