In the paper published by Nature titled “Human Sexual Cycles are Driven by Culture and Match Collective Moods” by Ian B. Wood, Pedro L. Varela, Johan Bollen, Luis M. Rocha and Joana Gonçalves-Sá, the authors posit that;
- there is an increase in people searching for terms related to “sex” on Google over the Christmas period in Christian countries worldwide, and
- this effect can only be observed during the period of certain major religious holidays, and not on celebrations such as Thanksgiving in the USA or Easter in France
Measuring the volume of people searching for “sex”
In order to understand how the authors have misused the Google Trends data I will start with this statement from the start of their Results section;
To measure interest in sex, for each country, we retrieved the frequency by which people searched for the word “sex” using Google TrendsTM (GT)10 (Methods 1–3); henceforth referred to as “sex-searches.”
The authors have stated that they “retrieved the frequency” of people searching for the word “sex”. This is not possible using the data source of Google Trends. Google Trends data shows “how often a particular search-term is entered relative to the total search-volume” (source Wikipedia). What this means is that the frequency of “sex-searches” could in fact be lower or entirely flat over the Christmas time periods, but that due to a decrease in the frequency of search terms not related to “sex” the Google Trends index will show an increase for “sex-searches”.
The authors have themselves identified this as a potential explanation for the increase in the index here;
we could expect the holiday season to lead to a decrease in overall searches, led by school vacations for instance, originating an artificial peak for sex-related interest.
However, the authors have discounted this possibility by a series of statements which apparently negate the above possibility;
This is an extremely weak statement, that is not at all backed up by any data in this report. I find it incredulous that such an important aspect of this report hinges on the validity of this statement, and yet no evidence is provided to back up this statement. I strongly suspect that there is significant general decrease in search volumes over the holiday periods due to the large decrease in both people working in offices and students in education. However, using solely Google Trends data it is not possible to prove this.
a putative decrease in overall searches is unlikely, as a decrease in searches for school-related material can be compensated by a strong increase in searches for “presents” or “recipes”.
The authors do then attempt to account for a potential decrease in overall search volumes over the Christmas periods by looking at the Google Trends data for the terms “on”, “and”, and “the”. This is the relevant statement from the report;
This shows a distinct lack of understanding of what the data from Google Trends actually represents. Looking at the Google Trends index for terms such as “on”, “and”, and “the” does not give any information on any changes to the volumes of searches carried out over the time period. The fact that the index for these search terms does not dip or spike over the Christmas period shows that the proportion of all searches that include these terms is not materially influenced by the Christmas period. There could still be, for example, an 80% decrease in people using Google to search for stuff, but if the proportion of those searches that include the words “on”, “and”, and “the” remains the same, then the Google Trends index values will remain stable. Therefore the authors’ conclusion that this is “probably resulting in an overall neutral change” to overall search volume is completely invalid.
when we control for search-volume of very common words, such as “on”, “and”, or “the”, there is some variation around the holiday period but it is in different directions for different search terms (Fig. S2A and B), probably resulting in an overall neutral change.
On the uniqueness of Christmas
we do not observe similar increases in weekly sex-searches for other widely observed holidays, such as Thanksgiving in the USA
In the USA Christmas is a bigger celebration and people take more time off work and school than they do for Thanksgiving. Using Google Trends data, we can see that the proportion of Google searches in the USA including the word Christmas, on a weekly basis, is significantly greater than for “thanksgiving” (Chart 1). We might therefore expect that any impact on the proportion of people searching for “sex-searches” to be smaller over the Thanksgiving period than the Christmas period.
If we look at Google Trends data for “sex-searches” on a daily granularity rather than a weekly granularity it is easy to identify the fact that Thanksgiving Day does also coincide with an increase in the proportion of people carrying out “sex-searches”. In Chart 2 I have compared the Thursday and Friday of the week preceding Thanksgiving, with the Thanksgiving Thursday/Friday, and then with the Thursday/Friday of the week after Thanksgiving. This shows the clear peak in “sex-searches” for Thanksgiving. The Thanksgiving increase in “sex-searches” is just largely obscured by looking at a weekly granularity because it affects a smaller time period and is less extreme than the Christmas impact.
In Chart 3, below, I show the daily index for “sex-searches”. The obvious peaks are weekends and the troughs are weekdays.
This demonstrates that the pattern of an increase in the Google Trends search index is related to the fact that people are not at school or in the office rather than the increase being specifically related to major religious or cultural events. This insight is hidden by the authors’ presentation of Google Trends data on a weekly granularity only.
My conclusion based on the referenced article and my own lightweight research here is that sexual activity is related to holiday periods because people have more time to spend together when they are not working. The volume of people Googling “sex” may or may not increase over holiday periods and weekends and it is no possible to demonstrate this from Google Trends data in isolation.
The proportion of all Google searches that include the word “sex” does increase during holiday periods and weekends. This is much more likely to be due to the massive reduction in people sitting in offices Googling stuff that is not “sex” (unless you’re Damian Green for example), than due to an actual increase in people Googling “sex” at Christmas.
We can further enhance our understanding of this data by adding in a search term that is more strongly related to office work, such as “gmail”.
The search index for “gmail” is clearly strongly inversely correlated with the search index for “sex” and further indicates that both these terms are strongly related to office working hours. We can even take this a stage further and look at this data on an hourly basis;
Here we can see the the time of day that people are Googling for “sex” is around 1am to 3am every day. We can say that at peak “sex-search” times the volume of people Googling for “sex” is 3-4 times higher than the volume of people at the same time Googling for “gmail”, and during peak “gmail” time we can say that the volume of people Googling for “gmail” is 3 times higher than the volume of people Googling “sex”. What we can’t say from this data though is whether more people are Googling “sex” at 3am or 6pm. If the total volume of people using Google at 6pm is high enough compared with at 3am then the total number of people Googling “sex” at 6pm would be higher than the number of people Googling at 6pm.