Skip to main content
. Author manuscript; available in PMC: 2017 Nov 1.
Published in final edited form as: Epidemiology. 2016 Nov;27(6):835–842. doi: 10.1097/EDE.0000000000000533

Table 1.

Bias, coverage of 95% confidence intervals and root mean square error (RMSE), reported for the distributed lag linear (DLM) and non-linear (DLNM) model and for moving average (MA) models with different lag intervals (0-3, 0-7, and 0-20), in four alternative simulated scenarios. An algebraic definition of bias, coverage and RMSE is provided in eAppendix 2 (Supplemental Digital Content).

Model Linear short-lag scenario Linear long-lag scenario

Bias (x100) coverage RMSE (x100) Bias (x100) coverage RMSE (x100)

DLM 0.03 0.95 2.81 0.05 0.95 2.77
MA0-3 0.22 0.94 0.99 4.92 0.00 5.00
MA0-7 0.49 0.94 1.44 1.87 0.70 2.28
MA0-20 1.16 0.94 3.02 0.10 0.95 2.75

Non-linear short-lag scenario Non-linear long-lag scenario

Bias (x100) coverage RMSE (x100) Bias (x100) coverage RMSE (x100)

DLNM 0.43 0.94 3.42 0.62 0.94 3.38
MA0-3 0.76 0.89 1.47 7.33 0.27 7.54
MA0-7 1.68 0.84 2.38 3.34 0.50 3.73
MA0-20 3.06 0.84 4.55 1.95 0.89 3.89