Predictor variables for a 100-km race time in male ultra-marathoners
In 169 male 100-km ultra-marathoners, anthropometry, training and pre-race experience were related to race time. Race time correlated with age, body mass, body mass index, upper-arm circumference, percent body fat, weekly running hours and kilometres, mean training speed, marathon personal best, number of 100-km finishes and 100-km personal best.
In stepwise regression, training speed, weekly running kilometres and age were the best predictors (r2 = .43). The authors conclude that training intensity may matter more than anthropometry for a successful 100-km race.
Participants
n=169 male 100-km ultra-marathoners.
Protocol
Variables: anthropometry, training (weekly running hours and kilometres, mean training speed) and pre-race experience (personal bests, number of finished 100-km runs).
Bivariate correlations with 100-km race time, then stepwise multiple regression.
Key Results
Bivariate correlations with race time (positive r = slower; negative r = faster):
| Variable | r |
|---|---|
| Personal best 100-km time | .72 |
| Personal best marathon time | .65 |
| Mean speed in training | -.56 |
| Percent body fat | .45 |
| Mean weekly running kilometres | -.43 |
| Body Mass Index | .29 |
| Circumference of upper arm | .26 |
| Age | .24 |
| Number of finished 100-km ultra-runs | .24 |
| Mean weekly running hours | -.21 |
| Body mass | .20 |
Stepwise regression: training speed (p < .0001), mean weekly running kilometres (p < .0001) and age (p < .0001) were the best predictors.
Prediction equation (n=169, r2 = .43): 100-km race time (min) = 1085.60 − 36.26 × (training speed, km/h) − 1.43 × (training volume, km/wk) + 2.50 × (age, yr).
Practical Application
In 169 male 100-km ultramarathoners, a faster mean training speed (r = -.56) and more weekly running kilometres (r = -.43) went with faster race times and, with age, were the best predictors; percent body fat also correlated (r = .45) but was not among the best predictors in the regression. To "what in my training predicts a faster 100 km?": how fast and how much you run in training; in the regression these beat the anthropometric measures.
Reading the equation: each extra km/h of mean training speed was associated with 36.26 min less race time and each extra weekly kilometre with 1.43 min less. The equation explains less than half of the variation (r2 = .43), so use it as a rough guide, not a race-time predictor for one runner.
Weekly hours correlated more weakly (r = -.21) than weekly kilometres (r = -.43): in this sample, kilometres tracked race time more closely than hours.
Inference, not tested here: faster training speed may partly reflect ability rather than cause it; the study cannot separate the two.
Not from this paper; the base's own practice: how a road-ultra week is built is in SYNTH-65.
Limitations
Cross-sectional and correlational; no causal inference.
Men only; the model explains less than half of the variance in race time (r2 = .43).
How training was recorded (self-report or logs) is not stated in the abstract.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level
cross-sectional, n=169 (≥100)