The results from a study on ironman-distance triathletes is now available on-line: http://trisurvey.net/
Male, English-speaking non-elite, ironman-distance triathletes were surveyed using an on-line questionnaire for descriptive characteristics, race history, and training history.
Useable results were then sorted into high, mid, or low success groups by their race time or their race place by age-group (n=39, age 35.0±5.9 y).
Training patterns for a 12-month period, based on training distance or training hours in swimming, bicycling, and running, were generated for each group.
There were statistically significant differences (p<.05) in several data sets.
The most notable difference was a positive relationship between higher bike training distances and race success, particularly during the final months leading up to an ironman-distance race.
There were observed differences between group patterns, the most outstanding being the high success group showing greater distances of bike training, which is consistent with other study’s findings.
A suggested annual training volume pattern for ironman-distance triathletes was proposed based on the common training volume pattern elements used by the high success group triathletes sampled in this study.
Note that the proposed training pattern is based on results from a small sample size and may not be valid for all athletes.
Thanks for the link. Very interesting and scholarly study. I must take small bites of it, chew thoroughly and digest morsel by morsel. Only then, may I go back for more…
The study seems to show that while there was a correlation between bike distance and race time there was no correlation between time spent on the bike and race time.
So… those who are faster in training are faster in racing.
Let me preface this by saying that I am not a statistician, so I could be wrong about both these comments - would be be happy to be corrected.
1- Is multiple Spearman ranks the right way to anylse the data? It seems to me that ANOVA would be more useful here, so you could look at interactions between variables.
2- If you do this many individual tests, with your p value set at 0.05, the odds of a positive test by chance are high ie every 5th test you do has a risk of being significant by chance. There are ways of compensating for this, but from what I can remember none of them are entirely satisfactory.
re: #2 - I think you meant ever 20th test would be significant by chance.
There’s a correction factor known as the Bonferroni method, which simply takes the number of tests that a study does and multiples the P values by that number. So if a study does 10 tests, you would have to be P<0.005 to still be significant at P<0.05 after the correction.