jeffeaton
commented
5 years ago The reason that the exposure duration is less than 1 year and differs for each respondent in the year of survey is because the calculation uses the date of interview for each individual (in CMC time) to calculate the exposure time for each episode. So for example someone interviewed in April 2017 would contribute 3 months of exposure in 2017 and someone interviewed in July 2017 would have 6 months of exposure in 2017.
Formally, the demographic calculations should be correct I think, though I have noticed odd estimates for some demographic indicators in the periods very near the survey. My preference would be to (1) continue to allow this case since the calculations should be formally correctand rely on users to exclude if they do not want periods for the year of the survey, but (2) do some more research to provide guidance around how this for various indicators.
On Aug 25, 2018, at 2:06 AM, hesohn notifications@github.com wrote:
When the reference year (the year that the mortality is being calculated) is survey year, the exposure duration is 1) less than 1 year, and 2) different from respondent to respondent.
Code may need to explicitly account for then Survey Year = Reference Year or fail as it does when Survey Year < Reference Year.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.
When the reference year (the year that the mortality is being calculated) is survey year, the exposure duration is 1) less than 1 year, and 2) different from respondent to respondent.
Code may need to explicitly account for then Survey Year = Reference Year or fail as it does when Survey Year < Reference Year.