Multivariate permutation test to compare survival curves for matched data

Galimberti, Stefania; Valsecchi, Maria Grazia

doi:10.1186/1471-2288-13-16

BMC Medical Research Methodology

Table 2 A-B - Simulation results under H ₁ of the multivariate permutation (two-sided) Fisher (A) and Tippett (B) tests

From: Multivariate permutation test to compare survival curves for matched data

Scenario*	Strata effect (w)	N. of Strata (k)	TS_i ¹				TLS_i ¹
			4 fixed	8 fixed	perc	all	4 fixed	8 fixed	perc	all
A) Fisher test
PH	0	30	0. 734	0. 723	0.627	0.592	0. 744	0. 737	0.673	0.651
	0	50	0. 911	0. 906	0.854	0.822	0. 909	0. 906	0.871	0.847
	0	100	0. 998	0. 996	0.993	0.968	0. 997	0. 996	0.993	0.978
	0.22	30	0. 692	0. 691	0.606	0.578	0. 706	0. 720	0.658	0.638
	0.22	50	0. 882	0. 873	0.816	0.795	0. 888	0. 887	0.839	0.831
	0.22	100	0. 995	0. 995	0.983	0.978	0. 997	0. 995	0.985	0.981
ED	0	30	0.274	0.274	0.414	0.432	0.302	0.297	0. 514	0.534
	0	50	0.406	0.408	0.622	0.641	0.443	0.452	0. 682	0. 699
	0	100	0.730	0.746	0.916	0.927	0.760	0.770	0. 939	0. 937
	0.22	30	0.240	0.252	0.376	0.409	0.267	0.285	0. 470	0. 488
	0.22	50	0.387	0.252	0.616	0.632	0.423	0.285	0. 671	0.682
	0.22	100	0.710	0.727	0.894	0.902	0.739	0.753	0. 924	0. 929
LD	0	30	0. 425	0. 408	0.278	0.247	0.408	0. 400	0.298	0.266
	0	50	0. 674	0. 642	0.464	0.392	0. 648	0. 627	0.470	0.406
	0	100	0. 949	0. 937	0.802	0.749	0. 940	0. 928	0.796	0.691
	0.22	30	0. 399	0.369	0.243	0.209	0. 387	0. 373	0.273	0.245
	0.22	50	0. 626	0. 590	0.414	0.359	0. 620	0. 587	0.440	0.384
	0.22	100	0. 910	0. 890	0.740	0.712	0. 908	0. 888	0.751	0.725
CH	0	30	0.144	0.165	0.285	0.272	0.079	0.078	0.115	0.099
	0	50	0.291	0.321	0.488	0.461	0.162	0.148	0.229	0.207
	0	100	0.747	0.799	0.897	0.881	0.560	0.596	0.746	0.729
	0.22	30	0.149	0.181	0.273	0.248	0.087	0.089	0.111	0.105
	0.22	50	0.292	0.319	0.477	0.419	0.178	0.161	0.237	0.207
	0.22	100	0.736	0.794	0.878	0.844	0.575	0.589	0.744	0.668
B) Tippett test
PH	0	30	0.685	0.666	0.586	0.605	0.676	0.654	0.674	0.677
	0	50	0.874	0.866	0.828	0.807	0.878	0.869	0.856	0.875
	0	100	0.996	0.995	0.991	0.979	0.996	0.994	0.992	0.996
	0.22	30	0.646	0.632	0.567	0.572	0.657	0.663	0.659	0.684
	0.22	50	0.834	0.826	0.768	0.766	0.850	0.831	0.816	0.814
	0.22	100	0.992	0.990	0.976	0.975	0.993	0.990	0.982	0.979
ED	0	30	0.393	0.369	0.351	0.384	0.458	0.458	0. 531	0. 596
	0	50	0.615	0.571	0.579	0.597	0.661	0.646	0. 679	0. 708
	0	100	0.906	0.910	0.904	0.914	0.920	0.924	0. 937	0. 942
	0.22	30	0.369	0.343	0.347	0.365	0.429	0.422	0. 501	0. 544
	0.22	50	0.597	0.343	0.560	0.589	0.652	0.422	0.676	0. 699
	0.22	100	0.887	0.877	0.896	0.897	0.910	0.901	0.923	0. 925
LD	0	30	0. 472	0. 458	0.392	0.383	0. 438	0. 442	0.398	0.400
	0	50	0.730	0. 718	0.644	0.632	0. 688	0. 683	0.634	0.615
	0	100	0.963	0.957	0.926	0.914	0. 956	0. 950	0.920	0.918
	0.22	30	0. 425	0.397	0.342	0.318	0. 411	0. 393	0.363	0.361
	0.22	50	0. 660	0. 650	0.572	0.567	0. 658	0. 634	0.584	0.566
	0.22	100	0. 930	0. 924	0.871	0.864	0. 927	0. 920	0.874	0.858
CH	0	30	0.356	0.418	0. 475	0.521	0.183	0.155	0.191	0.163
	0	50	0.584	0.636	0. 700	0. 729	0.380	0.335	0.374	0.347
	0	100	0.920	0.931	0. 951	0. 971	0.854	0.842	0.851	0.848
	0.22	30	0.335	0.399	0. 433	0. 484	0.176	0.165	0.185	0.163
	0.22	50	0.560	0.626	0. 663	0. 722	0.408	0.358	0.385	0.348
	0.22	100	0.887	0.921	0. 953	0. 963	0.830	0.806	0.850	0.836

* PH=Proportional Hazards, ED=Early Difference, LD=Late Difference, CH=Crossing Hazards
The simulations are based on first order statistics TS_i ¹ or TLS_i ¹ under different scenarios and for different choices of time points (4 or 8 fixed or 9 points at 10-90^th percentiles (perc) or 80% of all the observed event times (all)). In the body of the table, values in bold identify the strongest results.

Back to article page

ISSN: 1471-2288

Contact us

General enquiries: journalsubmissions@springernature.com