Going back to university I vaguely remember my sampling techniques course (1999!). You can
either
- do a census, ie count the population (both deaths and number of triathlons)
--> due to limited data availability likely associates with substantial measurement error
or
- sample a subset that you can count well and then extrapolate to the whole population
---> this is what I suggest. Less cumbersome
Which subset to choose?
- professional triathletes (you should find all deaths, the total #, and have a good estimate about miles p.a. they do)
- you could also go for region maybe?
- slowtwitchers
- deaths during triathlon races
- maybe corroborating with professional cyclists as a cross check
- combining different subsets can further improve the quality of the estimate
As death is likely a rare event you need to ensure to have enough events so its not subject to randomness. I really don't have signifcance thresholds ready but a guesstimate would be a dozen or more events needed to be able to extrapolate reasonably.
Hope that contributes somehow to start to get going.
Roberto
either
- do a census, ie count the population (both deaths and number of triathlons)
--> due to limited data availability likely associates with substantial measurement error
or
- sample a subset that you can count well and then extrapolate to the whole population
---> this is what I suggest. Less cumbersome
Which subset to choose?
- professional triathletes (you should find all deaths, the total #, and have a good estimate about miles p.a. they do)
- you could also go for region maybe?
- slowtwitchers
- deaths during triathlon races
- maybe corroborating with professional cyclists as a cross check
- combining different subsets can further improve the quality of the estimate
As death is likely a rare event you need to ensure to have enough events so its not subject to randomness. I really don't have signifcance thresholds ready but a guesstimate would be a dozen or more events needed to be able to extrapolate reasonably.
Hope that contributes somehow to start to get going.
Roberto
Last edited by:
teambernina: Aug 29, 16 0:45