An excellent article from the ABC highlights our current challenges in interpreting and asking the right questions of data. In short, three VFL games were used to trial new rules with a view to reduce player congestion and increase scoring. The conclusion: the three teams increased their individual scores by an average 15%, thus the trial was a success.
As the article points out however, these trials were conducted with poorer performing teams who were playing each other. So, regardless of the statistical significance (or insignificance) of 3 trials, a better comparison would be to see if they scored higher than they would be expected to score against other poor performing teams.
This is where the power of Tableau comes to the fore. A visualisation can quickly reveal underlying patterns in a data set. So while we could never conclude that the new rules had an effect on scoring over the 3 game sample size (think of the other variables here…weather, venue, availability of players etc.), we can quickly visualise our trial results against similar outcomes.
For my Tableau viz, the teams are grouped as either finishing in the top 8 or finishing in the bottom 7. We can then see that in 2018 there is a difference in the average scoring for games where top teams play each other as opposed to when bottom teams play each other. When bottom teams play each other, the average score is higher, thus we might well have expectations of scores above average for the trial games regardless of the rule changes.
Have a play with the viz to further explore the data.
Data can tell many stories…the challenge is being able to ask the right questions.
