26 March 2015 To better understand the results of this year’s vote, we have conducted some basic research into the Jury scoring pattern and would like to offer the below concise analysis for your consideration.

Post-mortem analysis
The results of the WesBank / SAGMJ Car of the Year voting always attracts some measure of analysis from OEM manufacturers, the Jury, Guild members in general, the media, and other interested parties.
All concerned give high regard to objectivity, fairness, expertise, and sufficient data.
After voting, jurors often have to deal with a fair level of scrutiny and often-unfair comment, as most commentators have little to no experience and remarkably little insight into the statistical complications that play a role in presenting a mathematically sound process. Often, the rules of the competition are also conveniently disregarded when layman analysis of the results are done.

Fairness
Within the constraints of having only 50 points to award in COTY (the same number of points as the European COTY system), the Jury has done a sterling job in terms of statistical fairness. Considering this year’s voting trend, a larger points allocation would not have altered the outcome for the top three positions, and would have served mainly to increase the number of points scored by the remainder of the field. Typically (based on analysis of the previous four COTY scoring grids) the top three cars score within 20% from each other, but in 2015 this narrowed to 4%, indicating the intensity of the competition among the top three finalists.

Transparency
Since COTY became transparent in the WesBank / SAGMJ Car of the Year competition in 2011, the Guild has worked hard to ensure that the competition is perceived to be such. Scepticism is a normal response from those who may not be familiar with the process in its entirety and, naturally, an opaque system attracts more doubt.
Looking closely at score patterns from 2011 until now, have revealed a trend for most jurors to score ever closer to mean scores, indicating that increased Jury consensus in the overall result.
Analysing the current score grid statistically, we found the highest level of juror consensus yet since 2011. This was tempered, however, by the fact that more jurors scored every finalist in 2014 than in 2015, contrary to the general trend since 2011. Scoring every finalist stabilises the score pattern for deductive interpretation, where allocated values indicate more consensus within a framework of 50 points.

2015 Grid Interpretation
Applying three different statistical tests indicated that the vast majority of the Jury scored impressively close, statistically speaking, to the mean score.
Our test applied between 0.25 and 1 point penalty for every score that differs from the average per finalist by the same value. i.e. if the Audi scored an average of 7 points, a juror who scored a 6 or an 8 for that finalist would be penalised by 1 point, and so on.
From this test, we can conclude that:
• 19 Jurors (68%) scored within 2 points from the mean score. Statistically speaking, such a small standard deviation indicates that the majority of the Jury had applied a high level of analytical reasoning when scoring the finalists.
• Fewer jurors scored randomly since 2013, indicating higher levels of experience, insight, fairness, and balance.
• 25 Jurors scored within 2.45 points from the mean, which appears to be our most accurate score pattern yet.

Alternative interpretations
Using two other tests: where a penalty is only incurred when scoring beyond 2 points from the mean score; or, alternatively, penalised when scoring more than 3 points from the mean, indicated similar conclusions to the “1 point penalty” test as above.

Although this analysis is not exhaustive, we believe it is important for the Guild and the Jury to understand the results in order to better inform the COTY workshop to be held on 4 May 2015.

More information regarding the COTY workshop will be communicated to you shortly.

Yours sincerely,
Bernard Hellberg Jr