There are three kinds of lies; white lies, damned lies and statistics.
– Mark Twain
A recent Guardian article, deriving from an Essential Media poll around Australian’s feelings towards Christmas, is an example of how statistics can be twisted or misrepresented to prove any point one cares to prove. In this case, it’s a phenomenon we can call Proportional Misrepresentation.
The main thrust of the article is that “Greens are the most likely voters to dislike Christmas, says poll”. The writers draw on the statistical data from the poll to support their claim because numbers equals proof, 78.2% of the time.
One such datum which unveils the Green’s grinch-like nature, is the figure of 26% of Greens voters surveyed stating that they would not encourage their children to believe in Santa. The highest of all groups. Bah-hum buggers!
The claim is supported by a chart resembling this one:
There it is for all to see… Greens are winning the race in the bid to eliminate Santa Clause.
There is a problem with this representation though and the problem is the representation is not representative!
Going on 2013 Federal Election results, the general proportion of people by party preference is as follows:
Libs/Nat – 45.6%
Labour – 33.4%
Other – 12.4%
Essential’s sample of 1,764 respondents indicated a 45%, 35%, 10% and 10% distribution respectively, very much aligned with the 2013’s election figures.
Normalising the Essential poll data so that it is a proportional representation of the Australian voting public, the charts begin to tell another story.
Now we see that it is actually much more likely that some who doesn’t believe in Santa is actually a Coalition or Labour voter. The number of Greens who don’t believe in Santa is minuscule in comparison.
The point becomes more obvious if we focus only on those who said they do not encourage belief in the Santa myth.
Now it would appear that the Greens are only a minority of those who are trying to banish Santa. Which story is true? Which perspective is more reflective of reality?
As I mentioned at the beginning, statistics can be used to tell any story. Be wary of anyone who tries to pass them off as facts without clarifying the nature of the dataset.
The Guardian article headline is most certainly misleading. The article presents party support groups as equal in size when that is not the case. Granted, the journalists were not assisted by the way Essential Media presented their findings, however it is not Essential Media’s job to fool-proof their data from journalists.
This is another example of people who don’t understand numbers misrepresenting numbers. Or it is an example of people who do understand numbers misrepresenting numbers. Either way the numbers are misrepresented, the headline is, by most reasonable interpretations, misleading and the world is probably a little worse off as a result of it.
References / Notes:
 – A 100% made up statistic in case you were wondering.
 – “Total” column removed as it skewed the axis scale