Nonojohn wrote:
Both are appropriate testing conditions, given proper randomization; the variability of each group would be accounted for during analysis. However, the method you proposed would be a step they could've taken to improve the power (and reduce false negatives); similar to, but not quiet equal to, increasing the sample size. I'm not sure you can claim that. We don't really know the Ventoux testing conditions. Here is the description:
Quote:
Approximately 12 days (range 10–16) after the last dose participants competitively climbed Mont Ventoux (Vaucluse département, France) in an open course via Bédoin
And:
Quote:
Weather conditions on Mont Ventoux (afternoon of June 19, 2016) in Bedoin were around 20°C and 40 km/h northern wind, and at the top were around 5°C and
85 km/h northern wind, without precipitation.
It sounds like they climbed from the south. With a strong headwind. If I'd mass-started that climb you know what I'm doing? I'm parking myself behind the leader. And then trying to come around him in the final few hundred meters. So even if I'm capable of way more time trial power, that wouldn't be apparent from a difference in finish time.
If the climb was time-trialled, say with 1:00 gaps, then differences in wind gusts could easily massively affect times. Was the wind increasing with time? If it was a time trial, were time-trial drafting rules in effect?
Throwing a large population at the problem only helps if these types of variations are mean-centered. Things like changing wind or group tactics probably aren't mean-centered. So throwing people at the problem doesn't necessarily help. You might need a longitudinal type study were climbs on lots of different days with all kinds of different weather conditions are used. And various mass start tactics are only averaged out over tons of different mass starts.