Garbage in, garbage out

Data can help coaches and players in the decision making process. Bad data however will lead to bad decisions, with the price being paid on the field.

Compare the 2 images below. They show a player performing the same running route but with 2 different GPS units.

The GPS unit used on the right is clearly superior than the left. But imagine you were using the units on the left with your team, but didn’t frequently check the reliability and validity of its data like shown above. You could be making some big errors in decisions that are typically informed with GPS data, such as how hard the team should practice tomorrow or which players may need to play limited minutes in the next game. Issues of data reliability and validity do not just affect GPS data, but also the data that is collected from video analysis.

Frequently checking the reliability and validity of your data is an important process, but unfortunately one that is often overlooked. Here are some simple strategies to improve your data:

• Plot your data: data can have the same mean, variance and correlation but look vastly different - see the datasaurus dozen for a fun example of this

• Automate where possible: repetitive tasks should be automated to reduce human error

• Make you/your analysts perform regular test-retests: analyze a game and then analyze the same game a few days later, how differently do the counts of events vary?

• Clear operational definitions: if a ball passes through the circle but is not touched by a player does this count as a circle entry? Regardless of your opinion, everyone must share the same definition.

• Data validation: in Excel you can create cells where only specific values/words can be entered (to reduce typing/spelling mistakes)