On the Limitations of Science in the Study of Running
Posted on January 21 2011
Image via Wikipedia
As runners, we often want concrete, black and white answers to questions about things like shoe design, running form, training methods, and so on. Does running in a minimal shoe make you less prone to injury than running in a traditionally cushioned, heel-lifted shoe? Does switching form to a midfoot strike make you faster? Will wearing compression sleeves make you recover faster after a race? Unfortunately, answers to questions like these aren’t simple, often because there are too many competing variables that make clear-cut conclusions difficult. Furthermore, an answer to one of these questions might not apply to to every single person – humans are variable, and changes will affect people in different ways. A given shoe might be great for one person, but terrible for another – it’s really hard to say what the result will be with any certainty.
As an example, I ran the Manchester City Marathon last November in a fairly minimal racing flat (Saucony Grid Type A4). My legs suffered during the race, and I irritated something in my left foot which cause me some trouble for about a week afterward. The easy and most tempting conclusion was that the shoes were to blame – I wore a flat for a very long, intense race, and it was the wrong choice. However, the culprit could just as easily have been the difficult, hilly course combined with the fact that I had just run an intense, BQ effort marathon a month before. In fact, in reading my race report from the same race the previous year, I experienced similar leg issues in Saucony Fastwitch lightweight trainers, which are much more cushioned and have a standard heel lift. Furthermore, a few weeks before the Hartford Marathon in 2009 I had irritated my foot in seemingly the same way as I did in Manchester, so maybe I’m simply susceptible to irritating what I think is my peroneus longus tendon. So where does this leave me? It leaves me to conclude that I have no idea exactly how I hurt my foot, why my legs suffered, or whether the shoes were to blame or not (or how much blame I should attribute to them). Such is the difficulty of trying to pinpoint clear causative factors when it comes to running injuries and the like.
One way to get at answers to questions like these is to turn to scientific experimentation. If we could run more controlled comparisons and hold as many variables constant as possible, maybe we could come to firmer conclusions. This is great in theory, but even science has its limitations. Let me walk you through a hypothetical example. Suppose we were interested in whether a given shoe could encourage a flatter, more midfoot foot strike (a very big goal for a lot of people these days, and a big marketing claim for manufacturers). Let’s take the two shoes I mentioned above, the Saucony Fastwitch and the Saucony Grid Type A4. Both are very lightweight, but they differ in some basic structural properties. The Fastwitch has a standard heel lift, whereas the Grid Type A4 has a much lower heel (see pictures below). We could have 20 people run by a high-speed camera in each shoe, measure the angle between the foot and ground for each, and compare the results between the two shoe conditions.
It’s quite possible, perhaps even likely, that the runners would on-average exhibit a flatter-footed, midfoot style landing in the flatter shoe (the Grid Type A4). However, the key phrase here is “on-average.” I would almost guarantee that there would be individuals who would buck the average trend and perhaps not show much difference between the shoes, or even show the reverse pattern. Therein lies a problem with scientific studies of running – in order to derive meaning from messy data, we use statistical analyses to compare distributions of numbers and means for groups, and often lose sight of the individual. So, we might conclude from this study that the Grid Type A4 is a better shoe, on average for this sample, at encouraging a midfoot strike, but that does not mean that it will serve this purpose well for every single individual. Perhaps other factors associated with the design of the shoes (e.g., midsole firmness, heel shape, forefoot width, etc.) will uniquely affect how isolated individuals will land in them.
In reading scientific studies of running biomechanics I see this often – a study will demonstrate a pattern, such as that a particular shoe design reduces ground impact loading rate (how rapidly the foot impacts the ground), but when you look at the variation among subjects it falls out over a fairly wide range. In other words, the results are non-systematic, and the individual gets swamped by the general trend. However, that doesn’t mean that the outlying individual who does not exhibit reduced impact loading rate in the shoe, or who might even exhibit increased loading rate, is unimportant. Individuals that don’t follow the general trend don’t want to get hurt either and would not benefit from following the advice suggested by the data – this is why scientists are often bad at making concrete statements, and why I always hesitate to do so on this blog. We tend to hedge and qualify statements by saying that our data “suggest” a given pattern, or that a “significant effect” was found. We rarely “prove” anything. I’m not saying this is bad, it’s just reality.
As a scientist I believe strongly in the value of controlled experimentation, but I also believe that we need to constantly keep in mind its limitations. Neglect of the individual in favor of the overall trend is one of them. Humans are variable, we are not machines, and we shouldn’t be treated like them. Science can be a very useful guide, but rarely does it provide absolute, definitive answers. It’s very hard to know in a black-and-white way what the effect of any shoe design, form change, or training method will be for any given person. This is the reason why scientifically informed, yet careful and open minded self-experimentation is perhaps ultimately the only thing that can truly work on an individual level.