Survival Analysis

To begin with Survival analysis, it helps researchers assess why certain individuals face a higher risk of events like death, machine failure or drug relapse. Researchers also refer to this event history analysis.

Furthermore, Survival analysis consists of a wide variety of techniques that help the researcher analyze time-to-event models. Researchers can use survival analysis to study various topics and better understand the causes of births and deaths. Researchers can also use it to study the causes of marriages, divorces, wars and revolutions.

Questions Answered:

What proportion of a given population will survive past a set date?

Of the population that survived, what is the rate they will die/fail?

Do certain factors (e.g., age and gender) effect the probability of survival?

The first step in survival analysis is for researchers to establish the event that will analyze.  For example, they might identify four states: ‘never married,’ ‘married,’ ‘divorced,’ and ‘widowed.’  The term ‘event’ refers to a transition between states. Here, ‘married’ represents a transition from the origin of ‘never married’ to ‘married.’

The next step is to ensure the data includes a longitudinal record of when the event occurred for individuals or groups.  For example, this may include ages and marriage dates, as well as other explanatory variables (e.g., gender, income, etc.).

The dependent variable can be survival time or transition rate, and researchers conduct analyses to describe the varying proportions of surviving cases at different times or to assess the relationship between survival time and a set of covariates/predictors.

The most common event analyzed in survival analysis is death.  Other common events are unemployment, graduation from school, machine failure, etc.  The survival time data in survival analysis has two important special characteristics.  The survival time data in survival analysis is always non-negative and usually shows a positive skew.

Researchers should differentiate the survival time, the object of study in survival analysis, from calendar time. They must always measure survival time relative to some appropriate time origin.

Need help with your analysis?

Schedule a time to speak with an expert using the link below.