What Do You Need to Know as a Sport Scientist Besides Sport Science | Complementary Training

This should keep me busy for the next year or two. Mladen Jovanovic is a genius.

from complemntarytraining.com

http://complementarytraining.net/what-do-you-need-to-know-as-a-sport-scientist-besides-sport-science/

What Do You Need to Know as a Sport Scientist Besides Sport Science

There are new roles emerging in contemporary sports field: sport scientist and performance manager. To be honest I have no clue how to define them – sometimes they are just a way to break out of the normal terminology (see Semantic Stretch and Power of Association in Made to Stick book, which is by the way excellent) and appear fancy and important. But sometimes they represent genuine additional roles and obligations on top of strength and conditioning / physical preparation positions. I will focus on those.

Please note that these are only my opinions, so take them with grain of salt. I will cover some of those extra skills needed and provide good resources where to find quality knowledge.

1. Basic and Advanced Statistics

As a sport scientist you will mostly deal with descriptive statistics, although if you plan publishing you will need to make generalizations or inferences to a bigger population from a sample you have. In a club settings, coaches will not give a f#ck about p values and confidence intervals – they generally have no interest in making generalizations beyond the group of athletes they are working with. Also, they are not interested in 'averages', but rather in single individuals and outliers.

This makes most of the statistical methods useless (unless you are publishing in journals). This is also a positive thing, because you will use lot simpler approaches;  approaches that are understandable by both coaches and athletes.

Here are some of the topics you will need to know

Basic

1. Effect size statistics (e.g. Cohen's D)
2. Linear regression and model fitting and model diagnostics (e.g. Cook's distance etc)
3. Concept of Reliability (Typical Error) and Smallest Worthwhile Change (SWC) when assessing an individual and performance changes

Advanced*

1. Non-Linear regression and generalized linear models (e.g. logistic regression)
2. Time-series analysis
3. Clustering, Factor Analysis and Principal Components Analysis (PCA)
4. Machine Learning, Data Mining, Predictive Modeling

* You do not need to know these in detail – only be kinda familiar with them (especially the last one) so you can hire someone to implement those with the data you have. Just make sure to know what they are and what they give to you. Besides these are definitely not going to be understandable to athletes and coaches 🙂 

I am by no means expert on those methods and I am learning them myself as well.

The good sources of information are Statistics in Kinesiology by Vincent and Weir

A New View of Statistics by Will Hopkins

Coursera courses on Statistics, Data Analysis & Machine Learning
Excellent paper on Mathematical Modeling by David Clarke and Philip Skiba

2. Data Visualization

It is very important to be able to convey data and insight to decision makers. Effective graph and table design is then a must have skill. I really love the books by Stephen Few

3. Excel

Excel is lingua franca of sport scientists. Even the richest clubs still use Excel and Dropbox combo. Here are some of the great resources for you to consider if you want to learn Excel

And excellent Excel tricks for Sport by John Lythe

4. PowerPivot

Some say this is the best addition to Excel since, well, Excel. It is very powerful add-on to Excel. I made Annual Planner for Sports v2.0 using PowerPivot.

5. R/MATLAB/Python

It is important to know at least one of those programming languages. R is getting bigger and bigger in data analysis and it is completely FREE, with a LOT of 'extensions' and libraries (see the list below). Python is also completely free and is catching up with data analysis, visualization and statistics. MATLAB is awesome engineering tool (SIMULINK is great tool), but more and more scientists are switching to R/Python.

If you want to learn R here are the couple of great books to start with

As a R user you will definitely need to learn great packages developed Hadley Wickham, like ggplot2, reshape2, plyr and stringr.

Some other 'extensions' of R are:

Shiny (to make web apps)
Slidify (to make awesome interactive presentations/slides)
Knitr (for reproducible research and dynamics documents)
rCharts and googleVis (for interactive charts)

To see excellent presentation that uses all the above features click HERE and HERE.

6. LabView

I have very little experience with LabView, but it is a important to know how it works if you plan working with lab equipment or programming your own (e.g. force mats, pressure mats). It is visual programming environment similar to SIMULINK.

7. Knowledge of some Athlete Management Software (AMS) Platforms

It is important to have at least basic experience working with some of the following AMS platforms: SMARTABASE, EDGE10, Apollo, TrainingPeaks.

8. Managing People

This is pet-peeve of mine. As Carl Valle used to say: "It's not what you know it's what you can get your athletes to do".

I have put this last on the list, but it is actually the most important.

Here are some of the titles I found helpful and some that are recommended by smarter people than me. This is by no means extensive list – there are a lot more great resources out there.

Also make sure the check Dan Ariely free course on Coursera on irrational human behavior and two books by Ray McLean from Leading Teams.