I have two groups (e.g. male/ female) and employment status (yes/no) per individual for 12 months in a row. Status may change every month (e.g. January employed, February unemployed, March employed, April employed....)

My intention is to find out whether both groups have a different pattern or not.

Any suggestions what approach might be suitable (eventually recommend a R package please).

  • 1
    $\begingroup$ There are specialised approaches to this but a simple way that I have found useful would be to model the time differences between the occurrences of a single binary response value (1 or 0). In other words, you don't model the occurrences of a Bernoulli event but the time taken between these occurrences (and hope that it doesn't come out as white noise). $\endgroup$
    – Digio
    Commented Jul 17, 2017 at 14:47
  • $\begingroup$ @Digio: Thanks for response. However I have problems with translating this approach in a mathematical system, because groups can switch their status every month from employment and unemployment and sample size is pretty large, so I need something automatized. Do you have an idea how I can deal with this issue (mathematical evaluation strategy)? $\endgroup$
    – Jens
    Commented Jul 20, 2017 at 13:52
  • $\begingroup$ You can achieve what I suggested with a simple SQL query that will transform your time series from binary to continuous within seconds before you start any kind of analysis. $\endgroup$
    – Digio
    Commented Jul 20, 2017 at 14:21
  • $\begingroup$ Since the goal is to understand the pattern of changing between groups, maybe some state model, start with a Markov model ... see for instance stats.stackexchange.com/questions/135573/… $\endgroup$ Commented Nov 2, 2021 at 17:06


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.