The Olympic Games have been occupying a great deal of most people’s attention over the last few weeks. Were you disappointed, or pleasantly surprised, by your country’s final medal tally?
South Africa ended up with 6 medals – 3 golds, 2 silvers and a bronze – about what was predicted by most realistic commentators before the games began.
Based on form, swimmer Cameron van den Burgh and javelin thrower Sunette Viljoen were strong candidates for gold, while mountain biker Burry Stander, the men’s 4*400m relay team and a few others were realistic medal hopefuls.
In practise, however, very little ran to form. Van den Burgh won gold, smashing the world record, but Viljoen and Stander ended in 4th and 5th place in their respective events Viljoen struggled on the night, while Stander Was delayed by a crash in front of him at the start and could not recover from the setback.
Instead, our golds came from swimming sensation Chad le Clos, who famously pipped Olynpic legend Michael Phelps to the line in what must have ranked as a major upset to the US medal expectations, and from the unfancied men’s lightweight fours rowers, who beat the highly rated British team.
At the end of the day, the data wasn’t wrong – it was simply incomplete. The British were the fancied team going into the race – but struggled with the crosswind on the day. Phelps was the expected winner, but got his timing wrong at the end and coasted for a few hundredths of a second too long.
Poor data quality is always a challenge for decision makers.The most obvious example of this in the games was in the Women’s Hammer Throw, where Betty Heidler’s bronze medal winning throw, of 77.12 metres was lost in the system, in spite of the clear evidence on television that she had practically equalled the silver throw. Fortunately, the missing data was retrieved and the correct results were delivered.
Missing data is not always so obvious – if you don’t know what is missing you may be making bad decisions without even realising it. One approach is to assess data quality regularly to ensure poor quality data is identified and addressed before it has a negative business impact.