The Snapshot graphs used by Sommelier Journal in our Tasting Panels are designed to portray the group consensus while also representing the spread of opinions. This is accomplished through the use of boxplots or “box and whiskers” statistical analyses. A general way to view these plots is that the tighter the box, the more agreement there was in the group. A wider box, whiskers, and outliers show more disagreement.
The Box The “box” in a “box and whiskers” graph contains the 25th percentile to the 75th percentile of scores. The median, or 50th percentile, is indicated by a dark black line in the box. It is possible for the box to contain only the median, or for it to contain only the median and scores above or the median and scores below (see the examples below).
The Whiskers The “whiskers” are the thin lines extending outward from the box. These indicate scores within the range of statistical norms. One way of thinking about them is that while they are not in the box—the 25-75% range—they are also not “outliers”—scores outside statistical norms. Note that a particular wine may not have any whiskers, may have whiskers on only one side, or may have whiskers that are longer on one side than the other.
The Dots The dots indicate “outliers”—extreme scores that lie beyond the range of statistical norms. Note that a wine may not have any dots, or may only have them on one side. More than one dot will not appear on any side because of the relatively small number of tasters.
Examples
Scores
10.00
12.00
14.00
16.00
18.00
20.00
This is an evenly spread boxplot, with the median halfway between the 25th and 75th percentiles and whiskers extending equally on both sides. As you can see from the underlying scores, opinion had to vary widely, but evenly, to make this result possible. It is important to note that a consensus is not required for a wide range of scores to be included in either the box or the whiskers.
Scores
10.00
14.00
15.00
15.00
16.00
20.00
This example contrasts with the previous one. The spread of scores is exactly the same (from 10 to 20), but because there is more agreement here around the 14-16 range, the 10 and 20 scores are outliers.
Scores
12.00
13.00
16.00
16.00
16.00
16.00
In this plot, the median and the 75th percentile are the same, so there is only one side of the box (the left side). The underlying scores show that the consensus opinion was clustered at the top level, with a few scorers disagreeing at lower levels. Because of the cluster at the high end, a score of 12 becomes a whisker outside the box, while a score of 13 is the 25th percentile.
Scores
12.00
12.00
12.00
12.00
15.00
16.00
In this plot, the median and the 25th percentile are the same, so there is only one side of the box (the right side). Scores were clustered at the lower end, with a few at the higher end. Because of the cluster at the low end, a score of 16 falls outside the box, while a score of 15 is the 75th percentile.
Scores
15.00
15.00
15.00
15.00
15.00
15.00
In this plot, everyone gave the same score, so the median, the 25th percentile, and the 75th percentile are all the same. There is no box or whiskers.
Scores
8.00
14.00
14.00
14.00
14.00
20.00
This plot has outliers on both sides, with no whiskers. Everyone gave the same score, except for two who disagreed by a wide margin on either side.
Scores
11.00
15.00
15.00
16.00
17.00
18.00
This example shows how an outlier can occur even when there is disagreement. One score is far enough away from the cluster of the other scores to be outside statistical norms.