The Sub 2 Hour marathon debate

The sub 2-hour marathon:  Science fact and projecting performance

I hate repeats.  Well, that is, I usually hate repeats, unless there's a very specific reason to watch the same thing twice.  And I figured that with the London and Boston Marathons only a few days away, there is no better time for a repeat than today.

The BBC yesterday did a radio show and carried this article on the possibilities of a sub-2 hour marathon.  The article is interesting and it touches on most of the points that we have looked at in the last few years, because this is by no means a new debate.  In fact, it seems to cycle around once every six months on the marathon-carousel.  Every time we have a big marathon on the horizon, people ask this question.

Personally, it seems a bit premature to me to be talking about a sub-2 hour marathon now, when the first human performance under 2:04 was as recent as 2008.  And then, by only 1 second!  To knock off another 240 seconds and break that barrier is getting a little carried away.  Let's get under 2:03 first, and then see how the future looks! That could easily take two generations of runner, another 10 years.  The big names are weighing in though - Gebrselassie and Radcliffe (who know a thing or two about marathon records) both suggest it will happen within 20 years.  A scientist predicts 28 years.  I suspect they're all a little optimistic...

The scientific merit - digging deeper

However, it is a question that has a great deal of scientific merit.  Both the mathematical modeling of performance evolution and the physiology of breaking 2 hours in the marathon are extremely interesting, and that's where the BBC article doesn't quite do justice to this question.

That is, the BBC touch on all the aspects - they mention the history, the evolution, the physiology that may be required, and the conditions that have to come together, but they varnish the surface rather than really getting stuck into it.  And so with that, I thought I'd take the opportunity to re-post an article I wrote about 7 months ago.  It asks "What is the physiology required for the 2 hour marathon?"  That article is below.

That article is below.  Also, the question is discussed from a historical point of view and using a slightly different approach to what you'll normally read in the media, in the following posts:

• Who will break 2 hours for the marathon?  And when?  Lessons from the shorter distances
• Why the marathon record was destined to drop, and why it may be about to level off again

Hopefully those three can keep you intrigued while I get my head around 10,000 hours and the next in the series on training and talent development!

Ross
_________________________________________________________________________________

The sub-2 hour marathon: What physiology will it take? (originally published September 4, 2010)

The other day, I did a post looking at the likelihood of a sub-2 hour marathon, inspired by an interesting article published in the Journal of Applied Physiology.  That was part 1, focusing on the historical evolution of world records, not only in the marathon, but over shorter 10km and 21km distances, because I'm of the opinion that a 2-hour marathon will only be possible when the 10,000m times and half-marathon times improve by at least 30 seconds and 90 seconds respectively.  So, contrary to the JAP paper, which suggested that the barrier would be broken in between 15 and 25 years, I think it will take many, many more years.

Today sees the promised Part 2, looking more at the physiology, and specifically the running economy that will be required in order to reel off 42,195 m at an average pace of 2:50/km.  It's a conceptual post, so bear with the length and the "unanswerable" questions.

And I must emphasize the point, picked up by some of you, that physiologically, our ability to separate a 2:03:59 from a 1:59:59 is limited.  Both performances are close to the "limit", and given that elite athletes are rarely tested and measured in the laboratory, and that the physiological variability is larger than the difference in performance we're looking for, you won't reach the end of this kind of discussion with a concrete answer.  However, it's an interesting process, nonetheless.

Running economy re-introduced

Way back in 2007, we did a short series on Running Economy, the first post of which can be read here.  Running economy is a measure of how much oxygen you use when running at a given (sub-maximal) speed.  The more economical the runner, the less oxygen is used, and this is crucial, because it is a characteristic of great runners that they are much more economical.

The graph below, which is redrawn from a paper by Foster and Lucia (2006), shows the running economy of a few different groups of runners.

Here, running economy has been measured as the volume of oxygen used per kilogram to run one kilometer (think of fuel economy in a car - one gallon takes you x-kilometers).  So, the Europeans shown by the pink line in the above graph are using approximately 210 ml/kg to run 1 kilometer.  You can convert this to the actual VO2 quite easily, if you know the running speed.  For example, at 20km/hour, 1 kilometer takes 3 minutes, and so these European athletes are using 70ml/kg/min at that pace.

The Africans, shown in red, are considerably more economical (this is one of the more interesting debates in physiology - there are obviously overlaps, but generally, African runners appear more economical than Europeans).  The VO2 of the Africans at 20km/hour is equal to 190 ml/kg/km, or around 63 ml/kg/min, about 10% lower than the Europeans.

Can economy and capacity predict performance?

So, knowing this, we can begin to project what kind of physiology is required in order to run a sub-2 hour marathon.  The method used is very similar to that used for cycling, when we looked at the power outputs achievable during the Tour de France in our recent Tour coverage.

There are of course assumptions that must be made, but as I tried to explain for cycling, if you make the "conservative" assumptions, you still produce an interesting picture of what is physiologically plausible.  This is an exercise in theory, not proof.  It is the first word in a debate, not the last, so bear with some assumptions and let's see what the picture reveals.

The assumptions

The first assumption that you have to make is what relative intensity an athlete can sustain for a given period, two hours in this case.

This is not too much of a guess for elite athletes, but it does vary considerably with training - an elite athlete is usually able to sustain running speeds that require about 85% - 90% of VO2max for about one hour.  Marathons are usually run at an intensity corresponding to approximately 80% of VO2max, while 10,000m is run at around 95% of VO2max.

So for a two-hour marathon, we can make the assumption that the intensity will be equal to between 80% and 85% of maximum.

Next one has to assume VO2max, the "capacity".  This is more of a lottery, because the range, even in elite athletes, can be quite wide.  Zersenay Tadese of Eritrea was measured at 83 ml/kg/min, but some elite runners have been measured as low as 70 ml/kg/min (the reason they are still competitive, incidentally, is likely due to exceptionally good running economy). 

And then finally, you must assume running economy.  This is the key assumption, because as the graph above shows, it varies quite considerably.  The east Africans have been measured as having running economies in the range of 180 to 190 ml/kg/km.  Zersenay Tadese was reported to be the most economical runner in history, using only 150 ml/kg/km.  This is so low that I'm actually skeptical about the value.  (For more on this, check out this post from 2007, and see later in this post)

Nevertheless, using three assumptions, we can create quite an interesting picture of what the sub-2 hour marathon runner will look like, at least in terms of his 'engine'.

The graph below is the first step towards bringing all three assumptions together.  It converts running economy into a VO2 (in ml/kg/min) at 2-hour marathon pace (± 2:50/km).

Obviously, this picture is only partly complete.  The next question to ask is whether an average VO2 of say 67 ml/kg/min for a marathon (at the running economy of 190 ml/kg/min on the graph) is reasonable or not?  Can an athlete sustain that intensity for the required 2 hours?  Obviously, this question is not answerable in theory, only through measurement, but we can get an idea of whether it is reasonable by asking what kind of maximal capacity is required for each of the above VO2 values to be reasonable.

The graph below attempts to establish this.  It estimates the VO2max of the runner for two conditions:  One is that the athlete is running at 80% of maximum intensity (shown in red) or 85% of maximum intensity (shown in green).

So, for a given running economy, you can now see what maximal capacity might be required, or vice-versa, given a VO2max, you can estimate the running economy required for two different conditions.

Two examples:  High economy vs lower economy

The white lines show 2 examples.  Example A is based on the measured economy of African athletes (including some Olympic medalists) who have a measured economy of around 190 ml/kg/min.  At 2-hour marathon pace, they use 67 ml/kg/min.  That athlete, running at 80% of maximum, must have a VO2 max of 84 ml/kg/min.  If they were able to sustain 85% (which I don't think is possible), they would have an estimated VO2max of 79 ml/kg/min.

Example B shows the European athletes plotted in the very first graph of this post.  Here, a running economy of 210 ml/kg/km requires a much higher maximal capacity; in this case, either 92 ml/kg/min (at 80%) or 87 ml/kg/min (at 85%).  The point is clearly that in order to run a sub-2 hour marathon, the athlete requires BOTH an exceptionally good economy, and a high maximal capacity.  I guess the third option is that they could get away with worse economy if they were able to sustain 90% of maximum for that length of time, but I don't see this as very likely (but not impossible).

The Zersenay Tadese example:  Something missing in the picture

So where does this leave us on the question of the sub-2 hour marathon?  On paper, or in theory, a two-hour marathon is possible because we have seen running economies of 180 to 190ml/kg/km, and we also see runners with VO2max values of 80 ml/kg/min.  According to the graph, a runner with these "characteristics" can run the 2-hour marathon.  However, I've yet to see the two together, with the exception of  Zersenay Tadese, with his measured economy of 150 ml/kg/min and VO2max of 83 ml/kg/min.  However, as I said earlier, there's something not quite right there...

Here's the problem:  With that kind of economy, Tadese would be able to run at a speed of 2:50/km while using only 53 ml/kg/min.  Think about that for a moment:  he is running at only 64% of his maximal capacity, at 2-hour marathon pace.  He would be jogging to the world record if that was true.

Even his 10km pace of ± 2:40/km represents an underperformance, because at that pace, he would only be using 56 ml/kg/min, or 68% of his maximum.  We know that elite athletes run at at least 90% of maximum for 10km, which means Tadese should be sustaining a VO2 of 75 ml/kg/min, which, given his economy, predicts a 10 km time of just outside 20 minutes!

Clearly, something is wrong with that picture.  I actually emailed Carl Foster about this a while back, to ask about it, and he graciously replied, but then my computer got stolen, along with that email.  So Carl, if you are reading this, feel free to chime in!  I seem to recall that it had something to do with the timing of measurements - Tadese's VO2max and running economy were not measured at the same time, and therefore using them together created a problem.  And this is crucial - when we make this kind of estimation, we're assuming that VO2max and economy are measured together.  And if they are, then the more economical a runner, the lower their VO2max is likely to be.

The VO2max-economy relationship

For this reason, the combination of high VO2max and exceptional running economy is likely to be very rare.  So, let's get back to that graph above - does the athlete shown by "Example A" exist?  Is there an athlete with such exception economy AND such a high VO2max?  Or is the athlete with that economy likely to have a VO2 max closer to 70 ml/kg/min?

That is a question I don't have the answer to.  What is clear though is that even now, the current world record represents physiology that is close to this theoretical picture presented above.  That is, the current record equates to a runner with an economy between 180 and 200 ml/kg/km and a VO2max between 75 and 85 ml/kg/min.  Both of these are reasonable assumptions, which would lead one to conclude that a sub-2 hour marathon is physiologically plausible.

However, as I emphasized in my previous post, it's not as simple as this.  First, there are other "limiting" factors - thermoregulation is one that has not been mentioned yet.  To run a sub-2 hour marathon requires significant heat loss to offset the heat production, and so the limit may well have little to do with oxygen use, and everything to do with the avoidance of hyperthermia.  This factor certainly narrows down the opportunities to break the world record, because only a cool day will suffice - Dubai, Berlin and Chicago have all produced days that are too warm in recent years.  As the 2-hour mark is neared, that becomes even more crucial, and that's a big reason why I don't think we'll see more than one, small incremental improvement every five or so years.

And then finally, I must stress that this 'pre-occupation' with VO2max and oxygen as the determinants of performance is only part of the picture.  Performance is the sum of far more physiology than just VO2max and economy, and that's why the best indication of whether that sub-2 hour performance will be possible is, in my opinion, to ask what the implications are for performance over other distances.  If the shorter distances don't improve, then the longer one is unlikely to, and you can go through this same exercise for 10km and 21km events and you realise that we're right on the limit of physiological performance.  Doping would shift that limit, yes, and perhaps training.  But I still don't see it happening within the next 20 years.

Perhaps I will be wrong...as always, your thoughts are welcome!

Ross

Chrissie Wellington on the charge and other marathon stats

Chrissie Wellington charges in SA Ironman and some other interesting marathon stats

I know I'm in the process of building a series on Talent Development and specialization, but today is a frantically full work day, and there are some really interesting sports stories from around the world, so I thought I would deviate to a (not totally) unrelated topic and look at two of them in a quicker post.

First, Chrissie Wellington was at it again yesterday, this time in Ironman South Africa.  She won the race in Port Elizabeth in a new women's record for Ironman events, an extra-ordinary 8:33.56.  That broke her own Ironman record of 8:36.13, set last year in Arizona.  That time was good enough for eight overall, and perhaps most remarkably, her marathon split of 2:52.54 was the fastest of the day, men or women (just a note, her fastest ever time is from Roth, an 8:19.13, but a number of people have suggested that this is a short course.  Others say it's not - I can't verify who is right on this one!  The men's time there is also super quick, so whether it's short or just fast, comparisons of Roth to other events seem a little tenuous...)

Challenging the men - gender differences in sport

It's yet another jaw-dropping performance from a woman who only came to the sport late.  Wellington is, if we stick to the theme of the last few posts, a classic "late specializer", and in my opinion, a great example that talent goes a long way!  Not that she was idle in her youth - she describes herself as sporty, but her focus on triathlon, following a sabbatical in Nepal, is as recent as 2007.  Within years, she's not merely winning, she's rewriting the sport's record books, and moving it into a new era.  The message she gives is not "you can be anything you want to be", but rather "It's never too late to be great!"...if you have the ability, that is..! 

And with her latest performance, the inevitable question is again whether Wellington is in the process of closing the gender gap, and how long will it take for a woman to win a men's race outright?

This was something I looked at last year, when we named Wellington as our Sportswoman of the Year for 2010.  Wellington is well on the way to 'defending' that title, but what I have described below is a similar discussion and analysis of her times relative to men's performance, because I've been asked about it a few times this morning already, and it's a pretty intriguing question in sports sciences.

Wellington in the open race - placing in the overall race

Wellington has rarely been outside the top 10 overall of her races - 7th in Roth, 7th in Korea, 8th in Arizona and now 8th in Port Elizabeth.  Her marathon times in some of her wins a regularly bettered by only a few men in those races.

The problem is that these direct comparisons present some misleading possibilities.  For example, if you took Paula Radcliffe's marathon WR of 2:15.25, she would place 13th in yesterday's Rotterdam Marathon, and 10th in London in 2010.

Yet, when you take that time and look at it against a larger group, you see that it ranks outside the top 400 PER YEAR (I recall that it was 473rd on the world lists for men last year), and outside the top 3000 in history! A once-off record can stack up well in a given race, but the same doesn't apply against the collection of races.

That's partly because a race like Ironman SA brings a good elite field, but it lacks depth - the best men are spread across dozens of races, and gather together only in Hawaii.   The way the competitive racing circuit is set up, it facilitates a "spread" of ability, not a concentration.

Similarly, the London Marathon has an amazing elite field, but it's only 6 or 7 deep, with the rest choosing Rotterdam, Paris and Boston to race in.  So once you get below 8th or 9th, you move away from the cream of the men's sport, yet you're comparing them to the very best woman in history.  And that's why the best comparison will come from either overall comparisons or at the pinnacle event, Kona in Hawaii.  And incidentally, there, Wellington finishes in the 20s - 22nd in 2009, for example, which is still exceptional.

By time - Wellington's times relative to men

The other way to compare performances is to look at time.  This also brings some problems, and it's particularly difficult to benchmark female performances in a sport like triathlon.  Three disciplines, differing conditions from one race to the next and within each race (thanks to the length), and race tactics make direct time comparisons less reliable than for track where records are often structured and paced.

You can do this for track and field, however.  The graph below shows the comparison between the men's world record and the women's world record for the track events.  You'll note that all women's records are between 9% and 13% slower than the men's times.  Given the long history and the relatively standardized conditions (for world records, which are almost always set in ideal conditions), these numbers show pretty clearly that men outperform women by around 10 to 11%.  Radcliffe's marathon record, ranked outside the top 3000 all-time, is 9.2% off the men's WR.

Now, let's look at Wellington's performances in the Ironman, as shown by red symbols below.  First, what is her place in history compared to other winners?  Below is a graph that compares the winning performances at Kona for women to the men's winning performances since 1995 (three generations of athlete).  This is done so that the varied conditions (and Kona can be very different, with heat and wind) are somewhat controlled for.

The difference, you can see, is pretty much the same as for the track and field events, with a few exceptions, where women are well down on the men (upward of 13%).  Wellington arrives in 2007, and goes from 10.8% to 9.7% to an amazing 6.7% slower than the men in her three wins.  In 2010, Mirinda Carfrae was only 9.8% off the men's winner.

Clearly, Wellington has been to the Men's Kona winner what most female track and field world record holders are to their male counterparts.  She is, from that point of view, in exactly the right place compared to the best men, and edging the event forward.

There are also other exceptional performances - the blue triangle above in 2002 is Natascha Badmann's 9:07 performance, which was only 7.5% slower than the men's winner that year (Timothy de Boom in a relatively slow 8:29:56).  This Badmann performance also shows up the flaw in this analysis - year by year is too variable - was that men's performance just very weak that year?  Did race tactics affect it?  More than likely...The fact that two years later, Badmann wins the race a full 15% slower than the men's winner confirms this - year on year comparisons of winners leaves some variability, and that's why a more "rigid" benchmark is needed.

All-time comparisons from Kona

So let's use the men's course record as the benchmark. There are problems with this comparison, too, of course.  For one thing, variable weather from one year to the next can blow out the differences when you compare isolated performances with a 'best ever' performance, but I think the trend will be revealing when combined with what we discussed above. You could do this same exercise with the average of the men's times, incidentally, and subtract about 3% off the difference.

So, in 1996, Luc van Lierde of Belgium won the race in 8:04:08, which still stands as the record today.  Comparing all the women's winning times since 1995 to that performance, Wellington's impact on the sport stands out a little more, as you can see in the graph below.

This graph suggests to me that Wellington has moved the sport into a new era for women - where the gap was previously 12%, she's moved it down to 10%.  That may seem a small improvement, but bear in mind that if the world record in the marathon were improved by 1%, it would be 1 minute 14 seconds faster.  So Wellington really has pushed the event forward.

Her Roth performance is 6.1% slower than the men's record, and yesterday in South Africa, she was only 5.8% behind the winner.  Those are remarkable performances, but again, the nature of single races does sometimes allow the very best (Wellington) closer to a 'diluted' men's field than would be the case against the very best men.

What does seem true, however, is that Wellington is getting better.  Therefore, she may well be primed to break the Ironman record in Kona this year - certainly having moved the overall Ironman record faster by many minutes, her 8:54.02 seems ready for revision.  If that happens, she may well bring the women's record down to within 9%, maybe even 8% of the men's time, and that would confirm what an extra-ordinary athlete she is.

Quite apart from all this is her media-friendly personality, her desire to engage and support the growth of the sport - she's a champion that the sport needs, and perhaps every sport wishes it could have.  Let's see what she produces next!

Marathon stats from around the world

This past weekend also saw the start of the Spring marathon season, with some quick racing in Rotterdam and Paris.  Next week are the big two races of the spring, London and Boston, and we'll be covering those with our usual analysis.

But until then, some interesting analysis came out yesterday, courtesy Ken Nakamura.  He looked at the winning times from all the major marathons and determined that Rotterdam is now the fastest marathon in the world, with Berlin and London in second and third.

Perhaps even more interesting, the sub-2:06 barrier is now the standard for elite marathon running - 24 men have cracked this barrier, the latest being Wilson Chebet in Rotterdam (2:05.27).  Compare this to 2:07, which has been broken by 75 men, and you see that the gap between good and great now lies around 2:06!

2:08, which used to be a world-class performance, is now comparably mediocre, such has been the explosion in performance in the last decade.  The rest of the world need to pay attention, because when the standard shifts, so too much expectation, or you're in danger of being left behind.  This is partly the case in SA, where we haven't improved nearly to the same degree, yet celebrate 2:10 marathons as world class.  2:06 gets you to the banquet, not 2:08...

The series on training, talent and champions resumes later this week (I hope!).

Ross

Specialization, training volume and talent development

Specialization, training volume and talent development

Yesterday I started what I hope is an interesting and thought-provoking series that addresses the issue of how we view and manage the process of sports talent development in young children.  I looked at a recent study of Danish elite and near-elite athletes where the authors concluded that "There is no delay in the athletic development that cannot be made up later with late specialization". 

Practice trajectory leads to performance?

That conclusion was inspired largely by their finding that elite athletes did LESS training than near-elite athletes up to the age of about 15, and then increased training time between 15 and 21.  The near-elite group, on the other hand, trained substantially more when very young, but peaked between 12 and 15, before actually dropping off.  The graph below, which I redrew using the data from the research study, shows what I called the "practice trajectory", and sums up the argument pretty nicely by showing how practice time (in hours per week) changes in the two groups

So there is a clear difference in practice trajectory, and it's tempting to say that this is linked to a difference in performance trajectory (in that they end at different levels).  However, that may not necessarily be the case, as I'll discuss below.

So today, I want to give my thoughts on the above graph, and give some possible reasons why the graph looks like it does.  I believe these potential solutions have some pretty significant implications for how we view "talent" and "hard work", and whether champions are born or made - it's unlikely to be as simple as we think!

Not specialization but training time

First, however, I have to clear up the definition of specialization and suggest a new name for the Danish study, which made a conclusion that I don't believe is supported by their data.  That study was called Late specialization: the key to success in centimeters, grams, or seconds (cgs) sports.

In my opinion, what the data suggest is that the paper should have been called "Delaying high training volume: the key to success in cgs sports.  That's because they actually found NO DIFFERENCE between elites and near-elites with respects to the training time spent on other sports - the elites did 63 months, the near-elites 62 months.

So the conclusion that delayed specialization predicts success doesn't hold up - it has little to do with specialization and a lot to do with training volume. This is important, because it's quite conceivable that a 12-year old who plays three sports (and is thus diversified) spends MORE time doing his main sport (for example, tennis) than another 12-year old who only plays tennis. Who is more specialized then? 

So, for the rest of this discussion, I'm going to leave "specialization" behind and talk rather about what the data showed, and that is that delaying higher training volumes in the selected sports predicts eventual placement in the elite group.  We need to discuss why that might be, because there are a few possible explanations.

Individual cases vs general principles

The other very important point to make is that all the data is an average (in this case, of 99 elite and 75 near-elite athletes) and within that data set, there'll be large variations.  So within the elite group represented by the beige line in the above graph, you may well have an individual who was doing 12 hours a week of training at 12, and you may also have near-elites doing less.  So we forge ahead with a discussion recognizing that we generalize, but also realizing that the question is that if you took 100 children, what would you recommend in order to maximize the chances of producing best possible performances?

Two models to explain the practice trajectory

With that, let's look at two models that might explain the practice trajectory.  To spare you scrolling up and down, here is the graph again:

 
I'd say there are three interesting questions about the changes over time in the two groups, which I've highlighted.

1. Why do some children do more training from the age of 9 to 15?
2. What happens in that period from 15 to 18, where there is a clear change in what the elite athletes do compared to those who will go on to become near-elite?
3. Why do the elite athletes continue to maintain training volume while the near-elites decline further from 18 to 21?

At the end of this post, I'll come back to these three questions and summarize my answers.  So if you feel like a "short-cut", jump to the bottom and get the quick version!  For the rest, to see the arguments, read on.

The motivation vs talent model

Generally speaking, I would propose that there are two "models" that could answer all three questions at once.  There's a hybrid of the two somewhere in the middle, of course, but I'll present the extremes in the interest of clarity.  The two extremes are summarized below.

I don't have the definitive answer of which model BEST explains what was observed, but I will say upfront that I believe the answer to be a combination, but I'm leaning more towards the talent model, on the right, and I'll explain why below.

The motivational/psychological model, and the role of parents and coaches

This model holds that young children who do high training volumes early will reach a point of burnout as a result of disillusionment with the sport, and a loss of desire which is amplified by a desire to try new things.

Normally (but not always) the high training volumes early will be the result of over-enthusiastic parents, who do not necessarily intend to be this way, but who create pressure for a child to train and perform in the sport from very young.  This may take the form of overt instructions ("you will train") or more subtle cues which steer a compliant child down a focus that they might not have chosen themselves (this balance, incidentally, is enormously tricky, and it would be presumptuous of me to give advice on it, so I won't...).  There are without doubt parents who adopt the more driven approach (again, think Woods and Agassi), and it's a problem here in SA, as I imagine it is wherever in the world you read this.

Competition becomes the focus, the parent's desire to win is transposed onto a child very early, and the relative importance of success is amplified.  A coach then comes on board, perhaps with the same desire to achieve success (for the team, school or club) and this further amplifies the pressure for the young athlete.

One comment from Michael on yesterday's post put this best when he described these as "stage parents", who "are the motivator behind the child's progress.  The child is a compliant learner combined with some motivation but not much...As much as the parents provided the impetus to get [the athlete] to near-elite [levels], they are not self-driven to make it into elites" (courtesy Michael on yesterday's post, thanks!)

The consequence of all this is that at some point, the athlete turns away from the time required to train, perhaps finding other sports or stopping altogether.  This explains point 2 and point 3 on the diagram above.  I received an email from a well known SA champion runner yesterday echoing these thoughts, that the child may never have truly enjoyed the sport, but did it as a result of EXTERNAL motivation (normally parents, but also peers and coaches), which was not sustainable.  It is certainly possible, as the examples some young athletes show us, but it's the general explanation for why there may be such a high attrition rate in those who train high volumes at young ages.

On the other hand, the elite athletes are those who are self-motivated, who have intrinsic desire to train, and this is motivation that normally develops later in life.  These athletes sample sports when younger, play more and compete less, and restrain the degree of "structure" in their sport until later.  Hence, training volumes are lower initially (1), increase later when the child makes the decision to commit and train hard (2) and maintains it as they achieve success (3).  The parent, meanwhile, is probably active, but facilitate the child's motivation, allowing them to develop at their own pace (again, thanks to Michael for the description!).

In this regard, the psychology, or the loss of motivation, is the main driver for what is eventually seen as a physiological change - the athlete trains less, achieves relatively poorer results and thus seems, physiologically, to have worsened.  So psychology drives behaviour (to train or not to train), which in turn drives the physiology (and performance).

The talent-physiology model

In contrast, the talent-physiology model says the exact opposite, that the physiology drives the psychology and the resultant behaviour.  Performance is thus the ultimate consequence of physiology, not motivation.

So according to this model:

• Some children do train harder when very young - this may be due to parental influence as described previously, or it may be because the child enjoys the sport at a young age (between 9 and 15, that is - you can go younger, but then there are other issues I won't go into here).  Whether it is parental or self-motivation does not matter too much at this stage
• The children who train LESS at this age, and who the data suggest will go on to be elite athletes, are not necessarily the ones who are more "balanced" or diversified, but rather the ones who have a better aptitude for sport in general and thus don't NEED TO TRAIN as hard in a specific sport in order to achieve their goals at these young ages.  They often get drawn into many other sports, precisely because of some innate ability, and the result is less training in what will ultimately become their selected sport
• In other words, their natural talent allows less training time per sport, and perhaps more 'play', whereas other children train more in a specific sport
• An additional and absolutely crucial factor is physical maturity relative to chronological age.  Call it biological maturity, but there is no doubt that children develop very differently, and so the differences between two 15 year olds may be very large.  This has performance implications, because now you have another possibility explaining training differences up to 15 years of age, also linked to physiology:
• Children who are early developers achieve more success at younger ages.  The faster, taller, stronger child at 14 is almost always dominant against their peers.  This encourages further training, both because parents encourage success and because children aged 14 or 15 recognize the value of success and so train more (as an aside, in SA, this is a particular problem, because a 14 year old who has developed early is big, fast and strong, and can easily be the school "hero" because of their rugby ability.  It doesn't do much for later developers, and nor does this early developer need to learn skills.  Later in life, when this advantage is eroded, he's left a little 'naked' and shown up for a lack of skill.  That's the theory, anyway...)
• Children who are late developers remain out of competition, may train less per sport but develop skills during this time
  
  
•  However, at around 15 or 16, these biological "gaps" start to narrow.  Late developers start to catch up, independent of training and perhaps at this stage, the 'end-result' begins to take shape.  The athletes who achieved success early on (thanks largely to early biological development) now find that their advantage is slowly being eroded.  
• Meanwhile, the athletes who possess some genetic advantage (to use a "swear word" to those who advocate the hard-work principle for success) for the sport, such as body size, metabolic adaptations, muscular differences begin to emerge as the best athletes
• The fact that this 'transition' in training volume happens at the age of around 15 to 17 is a very strong factor suggesting that training time may be linked to physiological development

 The ceiling comes into view

In other words, what is happening is that somewhere between the ages of 15 and 18, the "ceiling" comes into view.  You may recall yesterday that I wrote that genetics likely determines the ceiling in performance.  Training helps the athlete get there.  It's only at around 17 or 18 that it becomes more and more apparent to the athlete (and their coach and parents) that they have what it takes to succeed, or that they don't quite have it.

They may be good, and have achieved up to that point, but "great" lies beyond their capacity.  As a result, their training volume declines (2).  A loss of motivation?  Absolutely, but it's not the way the previous model defines it - here, the loss of motivation is simply because the athlete has recognized that there may not be a future in the sport.  They love it, and still do it (which is why they still train), but why do 2 hours a day to finish 20th?

At this point, the training history is irrelevant.  Those who trained less when younger did so either because they had more natural ability and so trained less or for more sport, or because they developed later and remained out of competition.  It doesn't matter, because the decision from 18 onwards is made by a athlete who is responding to their performance ability.  That accounts for Point 3 in the graph above.

Also, it's at this age that competition becomes fiercer, and so in order to remain competitive, an athlete must respond by increasing training volume, regardless of innate ability.   Their natural talent takes them to that point with less training than their peers, and at the age of 16 or 17, when most children have matured physically, they can begin to see that they have a future in the sport and must increase training volume.

Take a hypothetical case - a 15 year old rower, great as a junior because he is physically mature for his age.  But by 16, his peers are catching him as they mature.  They've been training less, but now their physiological development closes the gap on him.  Then he starts to lose, not badly, but enough that he realizes that he doesn't have what it takes.  At this point, his motivation levels decline and he trains less, the gap between him and those later developers getting larger and larger.  They have the added bonus that they have more "talent" to begin with, and pretty soon, you have this separation into the elite and near elite groups.

I think this is a crucial part of the answer - the children who train more younger probably do so partly because of parental pressure, but also because of success early on.  They do what they love, but they love it because they're good at it!  However, at some point, for various reasons, the status changes,  and the later developers, who trained less early, start to excel and then choose to train more to find further success.  So now you have a double-effect - not only do the most talented kids train more as they get older, but the less talented ones drop out and train less as they realize the ceiling is lower than initially thought.

Ultimately, in this model, it's the genetics, not the motivation, that "lets them down", because they've reached their "ceiling", which is set physiologically!  In this regard, the psychology is actually "pre-ordained" by the physiology.

Differentiating between the models - impossible, but intriguing

So which is it?  As I said upfront, I'm sure it's both.  In fact, there is evidence of reduced intrinsic motivation, and higher dropout rates, in athletes who specialize early with highly structured training.  And there are without doubt young athletes who stop training because they lose motivation and become disillusioned.  However, that doesn't control for the possibility that it's the physiology that actually determines the change in motivation over time, as I've suggested above.

And so if I have to pick an extreme, at least given the data of the Danish paper and my own insights and experiences, I'd go with the Talent-Physiology model to explain the changes in training and ultimate performance levels reached in elite athletes.

Why?  A few reasons.  First the training times in the Danish study up to the age of 15 are not really that high - an hour a day.  It's difficult to see that producing burnout, because it's actually rather low.  Three hours a day, that's another story.  So a "burnout" argument doesn't convince me.  

Secondly, and more importantly, the decision to back off and train less happens too early for it to simply be burnout from training and loss of motivation and disillusionment.  That would happen later, in my opinion, perhaps at school leaving age.  I do think that motivation levels may decline, but the main reason I would propose for a loss of motivation in someone at that age is that they stop being as successful, and decide that the investment (training time) is not worth the return (coming third or fourth or tenth)

And third, I don't think that a loss of motivation solely due to too much structured training could account for such severe reductions in training time in athletes who are still succeeding (and these athletes are).  I can think of very few athletes who retired from the sport while winning, and who did not return to it at some later stage.  I can think of many athletes who were burned out only once they started to lose - not winning is a very powerful force that strips away motivation!  So to see those changes at that young age (younger than 18), it would surprise me if it was a loss of intrinsic motivation as a result of too much structured training.  As a result of losing (the talent-physiology model), yes, but not simply because too much practice was bad for them earlier on.

Conclusion

So, three questions, and my three suggested explanations.  Again, let me emphasize this as strongly as I can - both models are plausible.  People lose motivation, and there's evidence showing reduced intrinsic motivation.  But that alone is not the driver of behaviour and less training, as I've argued.  Doesn't mean it doesn't happen, it's just that I would ascribe more cases to a talent-physiology model.

So here is the summary for those who jumped ahead!

1. Why do some children do more training from the age of 9 to 15?
   
   They are either pushed by parents into training, or they choose to train in response to success in a given sport.  That success might be driven by early physical development.  Meanwhile, some children train and compete less because they're late developers, or because they simply don't need the same training volume in order to compete.
   
   
2. What happens in that period from 15 to 18, where there is a clear change in what the elite athletes do compared to those who will go on to become near-elite?
   
   The ceiling comes into view.  That ceiling is determined by the genes, and some teenagers have an innate ability for a particular sport that tells them that they can "make it".  Some cannot, and they drop off in training volume, going on to become near-elites, or finding new sports altogether.  If you simply measured "intrinsic motivation" at this stage, you'd find that some teenagers have lost it, but this is not because practice is harmful, but because they aren't as successful anymore - a 16 year old is pretty savvy at picking this up!
   
   The fact those who did less training when younger go on to do more training now suggests that they may be developing physically to the point where they are now competitive and successful, or that they had superior ability to begin with and now respond to higher levels of competition by training more
   
   
3. Why do the elite athletes continue to maintain training volume while the near-elites decline further from 18 to 21?
   
   They respond to success - they recognize that the ceiling is high, and the fact that they're now in a much more competitive environment necessitates more training.  But that is worth it, because they have the capacity to succeed at the very high level.  Their behaviour (more training) is the consequence of physiology.

Obviously, there will be debate to this point.  It contradicts what I perceive to be a general culture of saying to people that they can do whatever they wish to do, provided they train hard enough.  Log the hours, you'll earn the reward.

Unfortunately, I don't believe that sport works this way, particularly the cgs sports, because they are what I would call physiologically affected sports.  All sports are physiologically affected, of course, but these are particularly determined by physiology.  I think that the example of sprint athletes is the best I can think of to illustrate that speed is not as trainable as people sometimes suggest.

For skill based sports, maybe a different debate.  What's missing in this debate (and I'm aware of this) are some illustrations and examples of how elite athletes do progress - how do top sprinters move through childhood into adulthood?  What was Usain Bolt's rise to the top, and is it typical?  Can we find any evidence for innate vs 'earned' sporting ability, especially in these cgs sports? (Because I guarantee that there are no longitudinal studies of it!)

But that's for next time, when I'll look at 10,000 hours, and the theory of training creating expertise. 

As always, this is meant as the first word in a discussion not the last, so feel free to comment!

Ross

Early vs Late Specialization: When should children specialize in sport?

Early vs Late Specialization: When should children specialize in sport?

There is no single pathway to success in sport.  If there were, we wouldn't be able to compare the stories of Chrissie Wellington, who discovered her remarkable talent late in life but went on to dominate IronMan Triathlon within a few years, to that of another endurance athlete, say Floyd Landis, who began cycling at school, with a single minded focus that took him to the professional level many years later.

There are countless cases of both examples, not only in endurance sport, but in skill-based sports - cricket or rugby players who "arrive" in their 20s, compared to the "prodigies" who are ear-marked for success from their early teenage years, or even earlier.  I am sure that in your own country, you can instantly think of one example of each.

If it took starting at the age of 4, with a parent driving a child to train for hours a day (think Agassi, Woods), then we wouldn't have cases like Roger Federer, who showed exceptional tennis ability very young, but did play other sports (on this note, Federer is reported to have begun at 6, but played football and tennis until he focused on tennis at 12 - this is still young, as we'll see later, but it's not nearly as early as other cases of tennis players.  Compare Agassi, who spent hours a day practicing at 6 years old, and who even played a match for money at the age of 9, at his father's "request")

But is there an optimal time to begin specialization in a particular sport?  This is such a loaded question that I can't possibly answer it, or even begin to cover it in one post.  So with that question begins a series of articles where I'll look at some of the evidence for whether young athletes who specialize very early on are more or less likely to succeed than athletes who delay high training volumes, competition and specialization in sport.

Without wanting to be too prescriptive, I think the following sub-headings needs to be addressed in a series:

1. What do elite athletes do?  Is there evidence to say whether early or late specialization is better?
   
2. What does science (that is, me...!) make of the 10,000 hour concept that that it takes 10,000 hours of deliberate practice to become an 'expert'? 
   
   Really, what this gets at is whether there is such a thing as "talent" or whether hard work and practice allows anyone to succeed.  It's the Coyle, Syed and Gladwell argument in Talent Code, Bounce and Outliers.  But what does physiology make of it?
   
3. What is the concept of Long-Term Athlete Development (LTAD)?  Where are its strengths, and where are its shortcomings, both practically and physiologically?
   
4. What are the implications of all this for coaches, parents, and young athletes?

So with that in mind, let's get started towards trying to answer that question.

Early and Late specialization:  Introducing the concepts

So we begin by looking at some evidence for what elite athletes do, and for that, I'll focus on a specific research paper called Late specialization: the key to success in centimeters, grams, or seconds (cgs) sports. (I linked to this paper on our Twitter feed on Monday, and I'll do a similar "article of the week" every Monday, so if you haven't yet followed us on Twitter, you can do so now!)

The title of the paper kind of gives away its conclusion, but don't worry, there is a lot more to it, including my conclusion that the paper does NOT in fact make this discovery, and there's something much more complex going on.

Here is a systematic breakdown of the paper, looking at the main research question, the rationale behind the research, its findings and how they might be interpreted.

The research question and rationale

The paper aimed to sort out which of two models for developing elite athletes was most effective in producing elite performances.  Those two models are summarized in the diagram below.  But before we begin, we have to define specialization.  In the paper, it has a rather clumsy definition, where it's a hybrid of being defined as a focus on a specific sport, as well as being measured in hours of practice in that sport.

In many cases, specialization and training volume will be related - the more you focus/specialize, the more time you have for that sport.  For example, a 15-year old with 2 hours a day to train will train more for Sport A if they are specialized than if they split the 2 hours between Sport A and Sport B.  However, this is not always the case - the same 15-year old can be "diversified" and do both sports, but still do more training in each than the specialist if they sum their time - A + B might equal four hours a day, not necessarily two.

To me, specialization should be measured as the number of hours spent training for Sport A relative to the time spent practicing for Sports B, C and D.  Someone who only practices tennis for one hour a day is more specialized than someone who trains tennis for two hours, but also plays football for an hour a day.  In the literature, however, there seems to be confusion around this, and specialization is not only "single focus", but also training time.  In other words, it's "specialization plus time practicing", and in the paper, I feel this is confused, and impacts on the conclusion.  Bear this in mind, because it will come up later...

So on your left is this model of "early specialization", where an early focus on a sport is recommended.  This is motivated largely by the framework that it takes so many practice hours to become proficient, and so you have to start young, and focus young, in order to accumulate them.  This is the Ericsson argument, and if you've read Bounce or Outliers, you'll know of Ericsson - he did a study on violinists in Berlin and found that the outstanding violinists had practiced for almost 10,000 hours, compared to only 8,000 hours for the "good" and 4,000 hours for the "normal" violinists (the ability of the violinists was assessed by the professors and teachers, in case you were wondering).

That study led to this 10,000 hours concept, which has since been applied to all kinds of skills, including sport.  There is an inherent problem with this, because sport is not the same as playing a violin in that there are without doubt physical attributes that training cannot change but which determine one's "ceiling of ability" in most sports.

The most obvious (bordering on ridiculously, in fact) example is that if you stand 1.50m tall, you'll never be a basketball star, even if you accumulate 20,000 hours of practice.  Your genetic make-up eliminates some of your options, it determines your ceiling especially when physiological characteristics are so significant to success, and then training helps to optimize how close you get to reaching your ceiling.  That's why no one succeeds without some training, but without question, some have more "talent" for a specific sport than others....  Whether the same is true of a skill-sport like tennis is debatable.  This is a great topic, but not one for today - that's why I'll set it aside for a future post as part of this series.  Let's leave this as saying that this kind of thinking drives the early specialization model.

On the negative side, there is also evidence of higher attrition rates with early specialization, and also potential negative health outcomes.  The issue is whether a young athlete who specializes at 9 or 10 is likely to continue with the sport beyond say 18, and there is some evidence that the answer is no.

On the right, the contrasting model is Early Diversification.  Here, children play a number of sports, the theory being that they develop a diverse range of skills, which are transferred across sports.  The proposed upside is that it promotes intrinsic motivation (let the child choose for themselves) and balance through increased exposure, and also ensures longevity.  The downside is that it may be too late, and by the time the person reaches adulthood, they may never overcome a potential late focus on training for a specific sport.

The only way to differentiate is to ask the question of elites, and that's exactly what the study did.

The method and findings

This kind of study is usually done by looking backwards in what is called a retrospective design.  Athletes are given questionnaires asking them to recall how much time they spent training each year.  Your alarm bells might be ringing, and rightly so, because this is a fundamental problem of this kind of research - it's reliant on memory and we all know that this is not infallible - can you remember how many hours a week you spent training in 2003?  The authors of the paper acknowledge this and they use some methods to confirm the memory of the athletes they interview, and conclude that the recall is reasonably good, given the limitation.  What will really help is a 20-year prospective longitudinal study, which I'm sure is on the way at some stage in the future.

The athletes interviewed in the study were high-level Danish athletes who were split into two groups, Elite and Near-Elite.  Elite athletes were those who had achieved Top 10 placings in World and Olympic competition or podium finish in European competitions, which is pretty impressive.  I'm looking at repeating this study here in South Africa, and if we set the standard at Top 10 globally, we'd be lucky to get 20 athletes!  The Danish got 148.  In the Near-Elite group were 95 athletes who hadn't met those criteria, but who were still on the Danish sports programme.

The athletes were then given a questionnaire looking mainly at how many hours a week they practiced, from the age of 9 up to the age of 21.  They also had to report what other sports they did and when they reached certain "milestones" in the sport, such as first international competitions, when they began intense training and when they reached the elite level.

The other very important thing to point out is that they only sampled athletes in what are called CGS sports - these are the sports measured in Centimeters, Grams and Seconds.  Think Rowing, swimming, athletics, kayaking, weightlifting, sailing, triathlon, cycling.  This is vital, because this study is NOT going to allow us to answer whether a tennis player or a golfer should start younger.  It also looks at sports that are typically more favoured by later specialization, because in general, peak performance age in these sports is in the mid to late 20s.  Sports like diving and gymnastics, on the other hand, are characterized by a peak in the late teens, early 20s, and that's a significant point to make.

Below is a summary of the main findings regarding cumulative training time, with a short explanation beneath it (it's fairly self-explanatory)

So four major findings.

• The first is that the Near-Elite group had actually gotten an earlier start than those who would go on to be elite - by the age of 9, they're 160 hours of practice time AHEAD.  
• That difference persists up to the age of 18, by which time there is no difference between the Elite and Near-Elite athletes.  
• Then, at the age of 21, the Elite athletes have pulled well clear, with about 1,100 hours MORE training than the Near-elites at that age.  
• And finally, there was no difference in the number of months spent on other sports - 63 months for the elites, 62 for the Near-Elites.  In other words, during the 12 year period of sampling, both groups spent just over 5 years in total practicing in other sports.  What is not reported is how those sports were spread out - were they done predominantly from 12 to 15, were they done for 7 months a year or all at once?

The practice trajectory:  When should the training volume be ramped up?

The above table (and the main table of results in the paper) are however incomplete.  What is really of interest is to track the practice time per year in these athletes.  For some reason that wasn't presented in the paper, but it's easy enough to do given the results, and so below is my re-analysis of their data, looking at what I would call a "practice trajectory":

So this plots the average number of practice hours PER week over time in the two groups.  Elites are shown in the beige, the Near-Elites in blue.  Quite clearly, they follow different trajectories.

• Up to the age of 9, as we said, the athletes who will go on to be Near-Elites do more than twice as much practice - it's an artificially low number, of course, because there's probably zero training up to maybe 5 or 6 on average, but it had to start somewhere!
• From 9 to 12, Near-elites do 2 hours a week more than athletes who will go on to be Elite
• From 12 to 15, Near-elites remain ahead, and the result is that by 15, they'd accumulated about 850 hours more practice than the Elite group had done
• Then from 15 to 18, it changes.  Here, the Near-Elites began to practice less, while the Elite group continued, increasing to almost 2 hours a day on average.  That shift is what causes the cumulative practice time at 18 to be equal between the groups, as we showed in the previous figure
• From 18 to 21, more of the same - the Elite group continues to spend 14 hours a week in practice, while the athletes who will go on to be Near-Elite drop down to just under 7 hours a week

So quite clearly, they follow very different pathways, and end up in different locations - one group goes on to be in the Top 10 globally or Top 3 in Europe, the other doesn't quite make that level.  They may yet, of course, the study was simply a retrospective look of a sample.

The authors then made five major conclusions. 

1. Elite athletes specialize later in their career.
2. Near-elite athletes pass through “milestones” sooner than elite athletes (I didn't go into this data, but the summary is that the athletes who go on to be Near-elite begin sport younger, train hard sooner and enter international competition around 2 years earlier than athletes going on to be Elite)
3. Elite athletes enter international competition older (as for above)
4. There is no difference in the time spent on other sports (remember, 62 months vs 63 months over the recall period)
5. "There is no delay in the athletic development that cannot be made up later with late specialization"

Some arguable conclusions, and what the study REALLY shows

Read that last conclusion again, for it is perhaps the most important one in the paper:  "There is no delay in the athletic development that cannot be made up later with late specialization".  I actually disagree subtly with this conclusion.

To me, the primary finding of the study is that success and performance in these CGS sports is NOT determined by how much TIME is spent training as a child, and that increasing the training volume later (after 15) is more than able to make up for time NOT spent training when even younger.   

As for the issue of specialization, that's a conclusion not supported by the results!  The time spent on other sports was the same - it may have followed a different pattern, but this wasn't reported.  All we know is that both groups did around 62 - 63 months of training in other sports over this period.  Specialization, defined as a single-focus on a sport, has nothing to do with ultimate performance, then.  It's more about time spent at different ages, and that "practice trajectory" I showed in the above graph.

However, in the interests of your time, and mine, I'm going to leave it overnight, and pick up this discussion again. Let's just say that this study was much more focused on training time in the final chosen sport, and it makes quite a nice case for delaying high training volumes until the mid to late teenage years.  Effectively, you need to replace "specialization" with "high training volumes" and then you have the real finding of the study!

And that's where the discussion will resume tomorrow!  That, and also we'll look at some of the reasons why that graph of practice time would look like it does - it may not be what you think!

I'm sure there will be comments and feedback, and as always, it's most welcome!  Join us tomorrow for more on this paper, and the issue of specialization vs training volume.

Ross