Table 2.

Simulation results: performance measures of regression parameter estimated using pseudo-observation and a model with identity link function and an independence working covariance matrix for the number of LYL due to cancer and due to other causes. The explanatory variables were age at diagnosis, sex, and year of diagnosis. Results based on 500 simulated data sets with different sample sizes (N=300,1000).

Cause	Covariate	LFP			Bias ()		empSE		ModSE		RMSE		Coverage
Cancer	Age	0.553	0.542	0.55	−10.029	−2.559	0.179	0.097	0.177	0.097	0.18	0.097	0.948	0.948
	Sex	0.749	0.8	0.777	50.949	28.345	0.569	0.318	0.573	0.319	0.572	0.319	0.954	0.95
	Year	−0.002	−0.002	−0.003	−0.133	−1.913	0.322	0.17	0.321	0.172	0.322	0.17	0.954	0.956
Other causes	Age	0.49	0.489	0.502	−0.737	12.319	0.045	0.025	0.049	0.028	0.045	0.027	0.968	0.964
	Sex	0.223	0.273	0.235	49.817	12.618	0.134	0.07	0.139	0.077	0.143	0.071	0.934	0.974
	Year	−0.021	0.011	−0.037	31.802	−15.82	0.084	0.047	0.085	0.047	0.09	0.05	0.932	0.92

LFP, least false parameter; empSE, empirical standard error; ModSE, model standard error; RMSE, root mean squared error.

Acceptable coverage range is calculated as