Skip to main content
. Author manuscript; available in PMC: 2019 Jul 3.
Published in final edited form as: Int Econ Rev (Philadelphia). 2018 Mar 9;59(3):1571–1619. doi: 10.1111/iere.12314

Table A1:

Research Sample Selection based on Assignment Method for Missing Smoking Values

Assumptions about smoking behaviors
used to determine research sample
Lifetime
smoking
observed
(%of
individuals)
Smoking
values
missing
(%of
person-years)
Resulting
sample
size
Research
sample
name

Begin with original sample of males who have at least two consecutive exams (N=2274). Use reported per-exam smoking info 27.4 22.5 622 A
+ fill in missing per-exam smoking values with data from subsequent retrospective questions 32.5 19.9 739 B
+ fill in up to 4 consecutive missing values when same smoking choice observed before & after missing spell 55.1 13.2 1253 C
+ fill in all missing values when at least 50% of smoking choices are observed and that smoking behavior is the same 55.6 12.9 1265 D
+ fill in missing values before death when at least 3 consecutive exams of the same smoking behavior is observed before death 77.1 8.1 1754 E
+ fill in missing values before death with last observed smoking choice before death 85.1 4.2 1935 F