top of page
  • John R. Harry, PhD, CSCS

Asymmetrical Force Output and Countermovement Jumps: Are We Studying It All Wrong?

There's quite a bit of literature out there on the potential influence of asymmetry on the performance of sport-specific activities, notably the two-footed countermovement jump (CMJ). Chris Bishop and colleagues published a nice and thorough review for scientists and practitioners to better understand the plethora of ways to calculate asymmetry, and they suggest the following symmetry index equation when exploring asymmetry during bilateral tests like the CMJ:

(Dominant Limb - Non Dominant Limb)/(Dominant Limb + Non Dominant Limb) x 100

For instance, an athlete displaying dominant and non dominant force outputs of 2 and 1.75 multiples of bodyweight, respectively, at a given instant in time would have the following asymmetry score:

(2-1.75)/(2+1.75)*100 = 6.67% in favor of the dominant limb

From my perspective, there's two main issues with the way this method for calculating asymmetry is used in contemporary studies and in practice. First, there is typically a reliance on peak force magnitudes or force magnitudes at one specific instant in time. This only investigates asymmetry over a 0.001 (i.e., 1 millisecond) time period (assuming the force platform is recording at 1000 Hz). Even when using variables such as average force or impulse during a specific period of time, hundreds (and even thousands) of data points are reduced to a single value. Thus, there is the potential to mask a great deal of information that could be explored when using discrete or average force magnitudes. Second, the criterion for what makes asymmetry "meaningful" is largely subjective, as the majority of work I've come across uses a 15% threshold for what constitutes a worthwhile asymmetry. In my view, it's very unlikely that 15% is equally meaningful at all instants of time or throughout the various time periods or phases of the CMJ. Another issue I have, albeit not specifically related to the symmetry index or it's focused use on discrete or average force magnitudes, is the group-based analytical approach that is used. I've covered why I think we need to expand our focus to supplement group analyses with replicated single subject approaches in a previous blog post, so I won't get on that soapbox again here. Instead, I'll present some data (unpublished - I'm working on that...) will hopefully further explain or support the statements I've made thus far and in my previous blog post. We'll start with my reasoning for the need to incorporate replicated single-subject approaches alongside the traditional group-based approach.

The table below provides discrete ground reaction force outputs at the end of the eccentric braking phase (sometimes called 'amortization force' or 'force at zero velocity') in addition to the average force outputs throughout the eccentric braking phase (i.e., the time between the minimum vertical velocity and when vertical velocity crosses zero). Data were collected from 19 NCAA Division 1 men's basketball players, with each athlete performing 3 maximum effort bilateral CMJs. Asymmetry was examined at the group level using paired-samples t-tests and Cohen's d effect sizes, with the mean and standard deviation across the athletes' symmetry indices also calculated. At the replicated single-subject level, we used the Model Statistic to determine whether the asymmetry was due to chance (essentially a single-subject t-test) along with Cohen's d effect sizes and the symmetry index.

The result I want to point out is that the group-level result indicates the average athlete from the group was not meaningfully asymmetrical for either variable, because the statistical probabilities were greater than 0.05, the effect sizes were all in the small range, and the symmetry indices were well below the often-used 15% threshold. However, the majority of athletes displayed statistically significant and super-mondo-sized differences between the limbs (74% of the sample for both variables - but not the exact same athletes for each variable) but only 1 and 2 athletes demonstrated asymmetries above the 15% threshold for the two variables. Even though it's quite clear that the replicated single-subject approach is superior for revealing the most information from which actionable decisions could be made, the question that remains is whether the Model Statistic plus effect sizes, symmetry index, or the collective use of each is the ideal approach? At this point in time, I would lean toward the latter, because more information is never a bad thing (as long as it's not erroneous information) and we can "pick and choose" so-to-speak what we feel is the best information for answering our own specific questions.

The second question or issue I have is, what does the above information really tell us about an athlete's asymmetry due to it's focus on a single instant in time and an average across a duration of time? Is there a way to get around this and still get an idea of an athlete's or group of athletes' force output asymmetry? The figure below presents the percentage of asymmetrical force output throughout each CMJ phase, although we'll focus on the eccentric braking phase (white bars) to keep things linked to the above table. To get these data, we time-normalized the CMJ phases to 100% (101 data points) and statistically tested (using the Model Statistic) each data point to identify the number of differences not due to chance (p < 0.05). As can be seen in the figure, the vast majority of athletes demonstrated "significant" asymmetry throughout a large portion of the braking phase, with 42% of the athletes displaying asymmetrical force output for more than 90% of the phase. Clearly, this method provides a greater amount of information from which actionable decisions can be made. The question that remains is, what type of actionable decisions can we make from this?

Although the example above used the Model Statistic to test for asymmetry throughout the CMJ phases, it's more than reasonable to also provide the percentage of "large" differences based on effect sizes or the percentage of "meaningful" asymmetries using the symmetry index equation and an objective threshold for what constitutes meaningful. In my view (Bernie Sanders voice - this is not a statement of political affiliation so don't jump on my back if you don't "feel the burn"), researchers and practitioners should begin combining these approaches and work diligently to uncover more objective thresholds or criteria for what constitutes a meaningful asymmetry.

The last things I'll mention before ending this post is that I've not even begun to explore how limb dominance changes throughout the phases when using the point-to-point approach referenced in the figure above. Anecdotally, I am seeing a lot of athletes demonstrating asymmetries dominated by either limb, depending on where we look within a phase or throughout the complete CMJ. In addition, I've run regression analyses on these data to see if CMJ performance (both jump height and RSImod) is predicted by asymmetrical force outputs at specific time points within the phases, when averaged across the phases, and by the percentage of asymmetrical force output throughout the phases. What we've learned from this sample is that, no, none of those methods of exploring asymmetry predicts a significant increase or decrease in either CMJ performance metric. So, we've probably got a lot of work to do within the human performance research and practitioner settings to really understand the impact of force application asymmetry on CMJ performance.

That's all for now, folks. See you next week. I'm out.

215 views0 comments
bottom of page