My Sasha Science For Olympic Trials and St George Marathon

[Dallen]

I took the qualifying times and trials tmes for everyone who qualified at St George. For comparison sake I did the same for all of the Chicago marathon qualifiers since I consider Chicago to be the standard for a flat fast marathon. Data is as follows.

Schuetze: 2:16:42 (07)   2:21:36      +4:54
Peterson     2:18:09 (07)   2:22:34      +4:25
Lander      2:18:25 (06)   2:20:09      +1:44
Sundwall     2:18:55 (07)   2:30:41                   +11:46
Danley      2:19:34 (07)   2:29:32      +9:58
McCombs    2:20:50 (07)   2:34:58                  +14:08
Fielding    2:21:47 (07)   



Abdirahman    2:08:56      
Sell       2:10:47      2:11:40      +0:53
Verran       2:14:23      2:17:10      +2:47
Johnson       2:15:03      2:17:58      +2:55
Morgan       2:15:11      2:16:28      +1:17
O’Brien       2:15:13
Leslie       2:15:20
Humphrey    2:15:22      2:20:34      +5:12
Ordway       2:17:13      2:18:10      +0:57
Hartman       2:15:50      2:15:57      +0:07
Aciniaga      2:16:58      2:17:08      +0:10
Rosendahl    2:17:05      2:33:58      +16:53
Seaton       2:17:32
Sutton       2:19:18      2:18:59      -0:19
Lucas       2:18:50      2:20:48      +1:58
Ernsberger    2:18:56
Pelletier       2:20:26      2:17:10      -3:16
Fellows       2:19:23      2:18:45      -0:38
Young       2:19:25      2:23:06      +3:41
Mentzer       2:19:33      2:18:23      -0:50
Ryf       2:19:37      2:32:26      +12:49
Rizzo       2:20:12      2:18:30      -1:42
Levassiur    2:20:15
Gramlich    2:20:19
Patannanan    2:20:19      2:38:36      +18:17
Williams     2:20:28      2:24:57      +4:29
Stelljes      2:20:32      2:24:19      +3:47
Arce      2:20:32      2:39:20      +18:48
Jeulund      2:20:33      2:31:31      +10:58
Dusen      2:20:35      2:29:31      +8:56
Stanko      2:20:37      2:27:23      +6:46
Mitchell      2:20:48
Franzen       2:20:49      2:31:53      +11:04
Wehrman     2:20:55      2:27:33      +6:38
Gilman      2:20:57      2:28:07      +7:10
Post      2:21:00      2:19:25      -1:35
Snyder      2:21:00      2:21:30      +0:30
Kutter      2:21:05      2:26:34      +5:29
Cox (07)      2:21:42      2:20:12      -1:30
Little      2:21:48      2:33:03      +11:15
Wadsworth   2:21:51      2:23:08      +1:17
Callinan      2:21:56
Horton      2:21:58      2:22:03      +0:05


The data is not nearly as pretty as I would have liked to have seen, but I doubt that there will ever be a better data set for comparison. You can draw your own conclusions, here is my analysis.

St George vs Trials: If you throw out all of the outliers it leaves only Schuetze, Peterson and Lander from St George, and I actually think Lander is an outlier. I conclude that St George was 4-5 minutes faster than the trials.

Trials vs Chicago: Throwing out the outliers leaves a large group fraily evenly spread between -1:30 and +1:30.  I am going to base it off of the A qualifiers who were fairly consistant. Dropping a few obvious outliers shows the trials to be about 45 seconds slower than Chicago. You can argue, but I think that this is fairly good data.

St George vs flat course (Chicago): Combining the above analysis shows St George to be about 3-4 minutes fast.

As a side note, as I looked over the qualifying data there was very consistant data showing that the Austin marathon is about 90 seconds faster than Chicago (a whle lot of double qualifiers were right at this range).

[Michelle Lowry]

Why throw out "outliers" especially since the data set is so small?  Why not average and have a standard deviation that is realistic?  I would think that there would need to be an alternate explanation for the data being skewed from the norm (like a broken leg, missing a significant turn, etc) rather than a bad race to justify throwing out data points, since we all have the potential to have bad races.  You can actually do true statistical analysis on this data, though my stats are rusty and thus I'll NOT be doing it.

[Dallen]

There are a couple reasons for throwing out the outliers.

First, the qualifying outliers were excluded because anyone who had a bad day didn't qualify. I want to compare good days to good days.

Second, the outliers would throw the data way off. Many people ran the trials more than 10 minutes slower than their qualifying times. If you use all of the data it would show St George to be about 8 minutes faster than the trials. If you throw in Logan's predicted finishing time it would be over 10 minutes. Not accurate.

[Paul Petersen]

Interesting analysis. Thanks for putting the time into it. BTW, you forgot Friscone for St. George. He would be a big-time outlier. BTW, some people do qualify on bad days. If you are a 2:14 guy and run 2:19, that is a bad day. There were actually quite few people like that. For instance, Josh Cox basically qualified with a solo tempo run. Dan Browne didn't have a marathon time to qualify with. Yuot ran is B qualifier a couple days after XC Nationals. etc.

But you have laid out way too many numbers and done too much work to qualify this as "Sasha Science". :-) This is closer to statistics. But the problem with both are the same: there are just way too many variables to consider. It is fun to speculate, but that's all it is: speculation and shooting from the hip. Here is one big variable: The St. George qualifiers (except for Lander and Friscone) only had 4 weeks to recover. Except for Cox, all the Chicago qualifiers had 13 months to focus solely on training and the Trials. I think this is HUGE, and alone skews things. I feel like I recovered from St. George quite well, but in reality 13 months of uninterrupted, focused training would have been much better. Just look at Lander. The dude had fresh legs with 5 miles left. And due to all my tapering and recovery periods, I have not gained any fitness in almost 2 months. Note to self: qualify earlier next time.

On that note, Chicago runners had 13 months to gain fitness. You can do a lot in 13 months. For instance, 13 months ago, I was only able to run 3 miles/day. My marathon PR was 2:26:30, and my Half PR was 1:12:50. And 13 months before that, my marathon PR was 2:35:25 and my half marathon PR was 1:14:00. The point is: people improve over time, sometimes by a lot. Many people that PR'ed at Trials were simply "due", and no course could stop them. It's that dreaded "human and time variable", that messes up both stats and Sasha 'Science'.

The pack running dynamic and intense competition was also something unique to Trials. Even at a big race, there is no way I would be in 90th place at the Half with 1:10:00. People run faster in a pack. I think the pack and the competition was enough to overcome the course and accommodate PR's. Great competition always brings out great performances; I have seen it on the track, grass, and roads. The course was not slow enough to stop this. I do agree that the course was not nearly as slow as some people thought it would be. The course was challenging, but not too bad, and I was able to find good rhythms the whole race. But it was also pretty dang hilly.

However, Dallen, after all my criticisms, I think your net conclusion is basically right: St. George is roughly 4 minutes faster than Chicago. Maybe more, maybe less depending on the person, but somewhere in that ballpark. It's definitely more than 2 minutes and definitely less than 10 minutes. :-)

Personally, just shooting from the hip, I think the Trials course itself may be 2 minutes slower than a flat sea level course. Perhaps more. Most commentary I've read and talked to are saying something similar.

A good question might be: How fast could have Ryan Hall ran if he were racing Geb at Berlin? I think he would have at least had an American Record, but I don't want to put limits on him. He looked fantastic on Saturday, a definite gold medal threat.

[Sasha Pachev]

Some comments.

Michelle - this is why Sasha Science succeeds where regular science fails. A runner with experience can look at a performance, splits, prior races, etc, and unscientifically but nevertheless correctly recognize that a runner had an extremely bad day or was simply more fit in one race than in the other. Intuitive wisdom gained from experience tells you which numbers are valid for comparison, and which are not.

Regarding Ryan Hall - I think that had he raced in Berlin against Haile  he would have had a shot at beating him and taking the world record. Hall is just beginning to learn how to use his half-marathon speed in the marathon, and I think in the Trials he showed that he is getting  closer to his true marathon potential, which I believe to be around 2:03. His splits of 1:06:17 - 1:02:45 show that he probably could have run around 2:07:30 on that course had he had pacers and been racing non-tactically.

Thoughts on Chicago - It's been a while since anybody ran under 2:07 there, and 2:07 often wins. Which tells me sometimes Chicago is not that ideal of a course - it may be flat, but you get winds and rain. So when the conditions are very good, you may see 2:06, and even under, but otherwise, it is probably a minute or two slower than a true ideal record-eligible course.

[Dallen]

I thought about pointing out the flaws in my analysis, but the post was already too long. I agree with the points Paul made.

I forgot to put into my analysis the fact that the 2006 Chicago marathon was a windy day, similar to the trials, and probably slowed down the good runners by about a minute. This would show the trials to be about 2 minutes slow, as Paul pointed out.

An important thing to remember about the trials is that it is set up to help the pepple in the 2:20-2:22 range. It looks like the runners who should have been aiming for a 2:25-2:30 still went out with the faster guys, and ended up running the first half too fast which inevitably results in a death march at the end. I assume this is what happened with Logan and Sean. 

As for Ryan Hall: I don't think the 2 minutes applys to him. He is clearly better than 2:07. The race was too slow and he looked like he had way too much energy at the end. But he was smart and ran for the win, not for time. I think he could do a 2:05-2:06 right now. He might have 2:03-2:04 potential someday, but he is still not in the same ballpark as Haile Geb...

What I would really love to see is a large and deep prize structure at the olympics (or any other race) that would pull all of the real talent. Then maybe we could get all the top guys to run the same race on the same day.