Can We Make Time for Physical Activity? Simulating Effects of Daily Physical Activity on Mortality

Table 2

Sources of variability identified by a sequence of scenario modifications.

Source identified	Reference scenario	Scenario modification
	Reference	Reallocate time from watching TV to walking
Independence^*	Watching TV ≥ walking	Respondents from bootstrap sample b, while regressions from [(b + 1) mod 40]
Sampling variability^*	Independence	Respondents drawn using full sample weights
Regression variance^*	Independence	Regressions estimated using full sample weights
Combined bootstrap variance	Independence	Combines the two previous scenarios
Mortality RR variance	Combined bootstrap	Set relative risk variance to 0.0
Randomized ages	Combined bootstrap	No perturbation of reported ages
Combined randomization	Combined bootstrap	Combines the two previous scenarios
Heterogeneous activity profiles	Combined randomization	Set all reported activity profiles equal to the mean profile

^*Required mean bootstrap variance correction.