Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Correlation Matrix with Time Varying Covariates

    Forgive me if this seems like too much information, but I figured the details might help.

    I am conducting a survival analysis using a Cox proportional hazards model and am seeking advice on how to produce a bivariate correlation matrix for predictor variables (not the parameter estimates). The unit of analysis is city. The event is the formation of an organization within a city. I allow for repeated events. The dependent variable is reset to zero after each event and duration is calculated as days since last event. Time is measured in days between 2007 and 2013.

    The data set includes time constant and time varying covariates. The time varying covariates are all similar (e.g., "Total gas wells drilled within 5 miles") and were created after stsplit, at(failures). There are no ties meaning that each failure has a unique value for "duration." After stsplit, I used "duration" to identify the date of each "spell" and calculate "total wells" for the risk pool at time i.

    For the time constant covariates, bivariates are straightforward. But how do I handle the tvcs?

    Thank you!!
    Last edited by Amanda Maull; 16 Nov 2015, 14:54.
Working...
X