0

Thanks for all the help you've been over the years.

Basically I have a set of data where about 200 individuals were tracked for a number of years and I know when an event occurred during that period. The event can occur more than once for the same individual. Looking at the data I notice that events seem to be clustering together (i.e. if an individual experiences the event once, they're likely to experience it 2 or 3 times within a month), how would I formally test that?

Appreciated!

Christopher Clukay
  • 13
  • 1
  • 1
    In StackOverflow, you are expected to present data preferably in the form of a data object that can be transfered to a console session for example coding and testing. There are tests for uniform distribution of inter-event intervals, which could serve as a null hypothesis in this case prhaps? – IRTFM Mar 13 '21 at 23:38
  • My apologies. I was more looking for a general answer so I probably used the wrong forum. Thank you for clarifying! – Christopher Clukay Mar 15 '21 at 13:59

1 Answers1

0

In this multiple failure-time framework a

  • Cox`s proportional hazards regression model with
  • PWP model or AG model extension is appropriate.
  • If the risk of occurrence of the following events is affected by the previous ones then use PWP model
  • If the events that occur have the same risk of occurence then use AG model.

In this free available paper both methods a explained https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4339761/

TarJae
  • 72,363
  • 6
  • 19
  • 66