Wine Winners and Competition Outcomes

Evaluation of Wine Competition Outcomes and Quality Consistency

• A central inquiry in the study of complex problems is whether the outcomes of professional wine competitions truly reflect the intrinsic quality of the wines being judged.

• Data analysis suggests a significant lack of consistency in how wines are awarded across different competitive environments.

Distribution of Awards for Wines in Multiple Competitions

• When examining wines that are entered in more than $3$ separate wine competitions, the distribution of awards shows extreme variance in results for the same product.

• The legend for award distribution uses a Maximum:Minimum format to indicate the range of recognition a single wine received across different venues:

  • G:NA (Gold to No Award): This category represents wines that received a Gold medal in at least one competition but failed to receive any award in another. This occurred for $53\%$ of the wines studied.

  • G:B (Gold to Bronze): This category represents wines receiving a maximum score of Gold and a minimum score of Bronze. This occurred for $39\%$ of the wines.

  • G:S (Gold to Silver): This category represents wines receiving a maximum score of Gold and a minimum score of Silver. This occurred for $7\%$ of the wines.

  • Other: Only $1\%$ of wines fell into other award distribution categories.

• The fact that over half ($53\%$) of Gold-winning wines also received "No Award" at other competitions suggests that the attainment of a Gold medal may not be a reliable indicator of consensus quality.

Statistical Correlations Between Major Wine Competitions

• To assess the reliability of wine judging, correlations were calculated between the results of $13$ major wine competitions. A correlation of $1.00$ would indicate perfect agreement, while values closer to $0.00$ indicate a lack of relationship between the scores.

• The competitions analyzed include:

  • DA: Dallas Morning News Wine Competition

  • SC: San Francisco Chronicle Wine Competition

  • GH: Grand Harvest Awards

  • NW: Jerry D. Mead's New World International Wine Competition

  • WC: West Coast Wine Competition

  • PR: Pacific Rim International Wine Competition

  • SD: San Diego National Wine Competition

  • RI: Riverside International Wine Competition

  • LA: Los Angeles County Fair

  • IE: International Eastern Wine Competition

  • OC: Orange County Fair Wine Competition

  • SF: San Francisco International Wine Competition

  • CS: California State Fair Wine Competition

• Key Correlation Data Points:

  • The correlation between the Dallas Morning News (DA) and the San Francisco Chronicle (SC) is only $0.09$.

  • The highest correlation observed in the set is between the Grand Harvest Awards (GH) and the San Francisco Chronicle (SC) at $0.33$.

  • Many correlations are near zero, such as DA and GH ($0.02$), DA and LA ($0.02$), or SC and CS ($0.02$).

  • Negative correlations exist in rare cases, such as CS and NW ($-0.02$).

  • The generally low coefficients (most falling between $0.05$ and $0.20$) demonstrate that the results of one competition are poor predictors of the results of another.

Probability Analysis of Gold Medal Awards

• A comparison was conducted between the predicted and observed frequencies for the number of Gold medals awarded to $375$ wines, each entering $5$ competitions.

• This analysis tests whether the distribution of Gold medals differs significantly from what would be expected by chance alone.

• Frequency Comparison:

  • $0$ Gold Medals: The predicted frequency by chance is approximately $280$, while the observed frequency is slightly lower, near $240$.

  • $1$ Gold Medal: The frequency predicted by chance and the observed frequency are closely aligned, both appearing near the $100$ to $110$ range.

  • $2$ Gold Medals: The observed frequency is slightly higher than the chance prediction, which is around $25$.

  • $3, 4, \text{ and } 5$ Gold Medals: The frequencies for these categories are extremely low for both predicted and observed data, approaching zero as the number of Gold medals increases.

• The close proximity of the "Observed" bars to the "Predicted" bars in the frequency distribution suggests that the allocation of Gold medals in these competitions closely mirrors a random chance distribution rather than a systematic measurement of objective quality.