Welcome to the hotseat. We've prepared a guide if you'd like to read more about how it works.

Can we assume that half the observed deaths aged x did not actually celebrate their x birthday in the year of investigation?

0 votes
160 views
asked May 26 in BUS 3018F - Models by 25b14c23 (400 points)
edited May 26 by 25b14c23

*with the dx defined in the question


From 2017 Tutorial 5 part 1 Question 1i attached below:


image

 

If we assume uniform and independent birthdays and deaths throughout the year, does that mean that dx will hold the number of deaths aged x last birthday at death for the 12 months prior to the census, with half of these being deaths that occurred before the life could celebrate a birthday this year and the other half being deaths where the life turned x this year and then died? Extending this further if it is true, does this mean that at the time of census, half of the deaths in dx would be aged x+1 last birthday had they not died?


Also I am having trouble with the interpretation of dx in this question. As its defined above, a death classified age x is actually classified on the day of death, correct? So if we wanted the number of deaths in the year preceeding the census aged x last birthday at the time of census, (as opposed to at the time of death) this would be different to the defined dx above? 


Thanks




1 Answer

0 votes
answered May 26 by asilmotala (2,570 points)
selected May 29 by Rowan
 
Best answer

So to answer your question, I first need to emphasise one of the most important things in exposure to risk. RATE INTERVALS. While the question does not explicitly ask the question, you should always be thinking of this and define them for your own use. Without defining them, you've basically put yourself in an impossible situation to answer the question. While I do know the rate intervals for deaths and the census (from my own definitions), I would prefer you to do it yourself and then rather compare your attempted answer, then simply reading my answer.

So to answer your questions:

1. "does that mean that dx will hold the number of deaths aged x last birthday at death for the 12 months prior to the census, with half of these being deaths that occurred before the life could celebrate a birthday this year and the other half being deaths where the life turned x this year and then died?"

No, the definition of dx does not imply this (even after making those assumptions). You've incorrectly come to this conclusion due to an incorrect definition (implicit) of the rate interval for deaths. This make's the second half of the question ("Extending this further if it is true, does this mean that at the time of census, half of the deaths in dx would be aged x+1 last birthday had they not died?") irrelevant, as it is untrue.

2. So you seem to be thoroughly confused regarding the definition of dx in the question. dx represents the number of deaths of lives classified as age x (according to the defined rate interval - try figure this out yourself first) that have died in the 12 months prior to the census date. As a result, your question "if we wanted the number of deaths in the year preceeding the census aged x last birthday at the time of census" implies a different rate interval, and it would indeed be a different dx to the one defined in the question. This brings me to another critical point of Exposure to Risk, which I mentioned in the tutorial last week, We do not manipulate death data. Death data, when given, is usually very concise and difficult to make reasonable assumptions about to calculate deaths over a different rate interval. Thus we usually concern ourselves with manipulating census and exposure data.

I hope that helps you with this question, and allows you to make a reasonable attempt at it. If you still cannot figure it out, comment or ask another question, and I can take you through a rough version of the answer.


commented May 26 by 25b14c23 (400 points)

Thanks for the response. I'll go over the basics again and use what you've said.

...