. 2014 Feb 13;3:86. doi: 10.1186/2193-1801-3-86

Table 2.

Comparison of linear and non-linear regression analysis of changes in swim time across years in the annual fastest and annual three fastest finishers to determine which model is the best

Swim time	Kind of regression	Sum of squares	DOF	AICC	Best regression AIC-test	Best regression F-test	Delta	Probability	Likelihood
Annual fastest men	Polynomial	3120.08	42	432.8	Linear	Linear	7.5	0.02	97.8%
Annual fastest men	Linear	3237.72	46	425.2	Linear	Linear	7.5	0.02	97.8%
Annual fastest women	Polynomial	1863.47	35	357.0	Linear	Polynomial	3.1	0.1	82.5%
Annual fastest women	Linear	2183.96	39	353.9	Linear	Polynomial	3.1	0.1	82.5%
Annual three fastest men	Polynomial	3254.07	44	419.8	Linear	Linear	2.8	0.2	80.6%
Annual three fastest men	Linear	3208.71	45	417.0	Linear	Linear	2.8	0.2	80.6%
Annual three fastest women	Polynomial	1482.07	26	262.0	Linear	Linear	7.8	0.02	98.1%
Annual three fastest women	Linear	1337.63	28	254.2	Linear	Linear	7.8	0.02	98.1%

For all investigated regressions, linear was the best model. DOF = degrees of freedom, AIC = Akaike information criterion.