I am analyzing the factors that affect the time-on-market when selling a house using -stcox-.
Because my data set includes houses that were listed anytime between 2012 and 2018, I'd like to allow the baseline hazard to vary for each year. I considered stratifying the regression by the -strata()- option but as part of my research I want to observe the "time" effects.
I then constructed dummy variables for each year (2012=0), ran -stcox- but got a very low hazard ratios for the last year (2018).
My questions are:
- Is my approach to include the time dummy variables correct?
- Is there any reason why I got such a low hazard ratio for 2018? Could it be related to the fact that my data set includes right-censored observations?
Code:
stcox log_size log_price house_age i.year failure _d: isSold == 1 analysis time _t: NumOfMntsOnMarket id: HouseID Iteration 0: log likelihood = -48285.125 Iteration 1: log likelihood = -47533.344 Iteration 2: log likelihood = -47472.034 Iteration 3: log likelihood = -47466.515 Iteration 4: log likelihood = -47466.406 Iteration 5: log likelihood = -47466.406 Refining estimates: Iteration 0: log likelihood = -47466.406 Cox regression -- Breslow method for ties No. of subjects = 6925 Number of obs = 6925 No. of failures = 5921 Time at risk = 18419 LR chi2(9) = 1637.44 Log likelihood = -47466.406 Prob > chi2 = 0.0000 ------------------------------------------------------------------------------ _t | Haz. Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- log_size | .7465455 .0093898 -23.24 0.000 .7283668 .7651779 log_price | .732612 .0354403 -6.43 0.000 .6663416 .8054732 house_age | 1.001557 .0006235 2.50 0.012 1.000336 1.00278 | year | 2013 | 1.186201 .04516 4.49 0.000 1.10091 1.278099 2014 | 1.173349 .0468697 4.00 0.000 1.08499 1.268904 2015 | 1.263186 .0541346 5.45 0.000 1.161418 1.373872 2016 | 1.123802 .0608042 2.16 0.031 1.01073 1.249524 2017 | .4400602 .030792 -11.73 0.000 .3836644 .5047458 2018 | .151731 .01692 -16.91 0.000 .1219422 .1887968 ------------------------------------------------------------------------------
0 Response to Time Dummy Variables in stcox
Post a Comment