Table 3

Coefficient estimates for each of our linear models

ParameterCoefficientEffect sizeP value (one tailed)P value (two tailed)
Model 1: event+natal_sex+(age−40)+(age−40)2+nb_predictor
sex=natal female0.1222113.00%***0.00000.0000
(Age−40)0.003750.38 %/y***0.00000.0000
(Age−40)20.000120.012 %/y2***0.00000.0000
nb_predictor−0.00424−0.42 %0.40710.8079
Model 2: event+gender_id+(age−40)+(age-40)2+nb_predictor
gender_id=’female’0.1222213.00%***0.00000.0000
gender_id=’non-binary’0.093879.84%***0.00000.0000
(Age−40)0.003750.376 %/y***0.00000.0000
(Age−40)20.000120.012 %/y2***0.00000.0000
nb_predictor−0.06803−6.58%***0.00010.0001
Model 3: event+gender_id+(age−40)+(age−40)2+is_nbm+is_nbf
sex=natal female0.1222213.00%***0.00000.0000
(Age−40)0.003750.376 %/y***0.00000.0000
(Age−40)20.000120.012 %/y2 ***0.00000.0000
is_nbm0.025842.62%0.13240.2681
is_nbf0.039694.05%0.05800.1152
Model 4: event+natal_sex+(age−40)+(age−40)2+isNB
isNB0.032253.278% (.)0.02620.0528
natal_sex=natal female0.1222413.002 %/y***0.00000.0000
(Age−40)0.003750.376 %/y2***0.00000.0000
(Age−40)20.000120.012% ***0.00000.0000
  • Coefficients for different events are ignored. The coefficients are given as a percentage increase in marathon time in the ‘effect size’ column for ease of interpretation. The final two columns contain Monte Carlo estimates for the p values of the coefficients estimated using 100 000 samples.

  • The symbols (.), *, **, *** indicate statistical significance at the 0.10, 0.05, 0.01 and 0.001 levels using a two-tailed test.