Christmas comes earlier every year, doesn’t it?
A number of news outlets seem to think so. They picked up on an online article published on the StatsLife site of the Royal Statistical Society.
The writer (Nathan Cunningham) analyses the volume of web searches on a variety of Christmas related terms as they relate to Ireland. The technique he uses identifies the tipping point between weeks when searches are what I’ll term ‘non-Christmassy’ or ‘Christmassy’.
He finds that in 2007, the tipping point was in the week beginning 11 November. But by 2013, the tipping point was in the week beginning 25 August. The years in between appear to confirm a trend of Christmas being in people’s thoughts earlier.
He does sound a note of caution though, writing:
“Given the sample size of just seven observations, it is difficult to confirm whether this trend reflects a true early arrival of Christmas, or is merely a chance observation resulting from random variation in the data.”
One of the things I impress on those attending my talks and workshops is – does a data-related claim “feel right”?
To me, that near mid-November finding for 2007 just didn’t feel right. If anything I would have expected closer to late August, as was found for more recent years.
This may be to do with my UK perspective.
August is when most school pupils are on holiday and have much more opportunity, perhaps, to start highlighting what they want for Christmas to their parents and others.
Adults are likely to be concluding their summer holidays and so can start to think about the next round of big expenditure – Christmas.
And, of course, like it or not the commercial focus starts to turn towards Christmas. It may be our minds turn to Christmas earlier because they are prompted to!
With any claim it’s worth sense-checking.
Google has a tool that allows search trends to be explored over time. I used this to gather data on search volumes for each year from 2004 to 2013 on the term ‘Christmas’. The data I have is of volumes expressed on a scale of 0 to 100 for each week of the year, with 100 being assigned for the peak value.
As there has been a general growth in search activity over the years it seemed right to use this sort of standardised data.
From it I used Excel to visualise the data, dividing the year into weeks in which search volumes scored 9 or less, between 10 and 24, and 25 and above.
To my eye there doesn’t seem to be a dramatic difference between 2007 and 2013, though it does seem that there might have been earlier thinking of Christmas around the time of the financial crisis.
But my analysis is based on something far less sophisticated than the constrained model-based clustering algorithm used by the article’s author.
So it’s always worth having another look at the data. The table below has the week number for when the search volumes for each year topped scores of 5, 10 and 25.
|Year||Week when volume score reached 5||Week when volume score reached 10||Week when volume score reached 25|
This does suggest that, in recent years, Christmas is entering people’s minds earlier, though with a lowish search volume. Note that week 36 is early to mid September so not far off my feeling that things start around late August.
I would stress that this is an unsophisticated analysis. It doesn’t – as Nathan Cunningham’s method does – identify a tipping point from non-Christmassy to Christmassy.
It’s always worth a bit more testing so I tried a couple more search terms. For many people, Christmas is Christmas cake – and a good Christmas cake is made some time before the festive period. And there’s always the Christmas music – so I tried a search on the Pogues (who can resist joining in the chorus of “Fairytale of New York”?!)
These are the results for when we start towards peak-Cake and peak-Pogues…
|2004||03 October 2004||28 November 2004|
|2005||25 September 2005||04 December 2005|
|2006||24 September 2006||03 December 2006|
|2007||30 September 2007||09 December 2007|
|2008||21 September 2008||07 December 2008|
|2009||13 September 2009||06 December 2009|
|2010||19 September 2010||12 December 2010|
|2011||04 September 2011||04 December 2011|
|2012||02 September 2012||09 December 2012|
|2013||01 September 2013||08 December 2013|
There does seem to be some evidence for Christmas cake searches creeping earlier. But the Pogues seem resolutely a feature of the festive season proper.
Nathan Cunningham urged us to be cautious about his result. It shouldn’t be overblown. At the same time, though, there does seem to be something to it.
But I’d suggest that this is more due to the growth in use of the internet access and in online shopping. In other words, we’ve always thought of Christmas a long way out, but now we’re more likely to use the internet to help us begin our preparations.
UPDATE (12 November 2014): The Telegraph newspaper had an article on people’s perception of Christmas coming earlier – “Christmas gets earlier every year? No it doesn’t”.