Published on August 10th, 20120
Survival Analysis – The Forgotten Tool of the Social Scientist
By Tor G. Jakobsen
Survival analysis is used when the researcher is interested in whether or not an event happens, and also when it happens. A common tool for the medicine, the social scientist community is starting to see the opportunities provided by this tool.
Many social-science phenomena are about causal relationships, that is, that they have a progress over time. The best non-experimental technique to study these processes of causality is by the use of survival analysis. The strength of survival analysis is that the observations over time enable us to estimate the chain of causality with a large degree of confidence.
Survival analysis is used widely in many sciences. In medicine researchers perform studies of the effect of treatment on patients. In sociology one can mention studies of unemployment, careers, marriage, divorce, and child birth.
The common denominator for the above mentioned examples is that they possess the same logical structure. A sample of subjects is considered a group at risk where events, like getting married, giving birth, change job, or die from an illness, can happen in a period of risk. For each subject one records whether or not the event took place within the time scope of the study, and also how long it took from when that subject entered into the risk phase. This is what defines the dependent variable in the survival analysis.
One complication associated with survival analysis is that of so-called censored observations, that is, in many cases there is no event in the period of the study: the patient survives; some women do not have children etc. This calls for the use of especially developed statistical techniques. The most common technique used for the analysis of survival data is Cox regression.