While the well-known birthday problem assumes an equal likelihood of being born on each day of the year, there is in fact slight, albeit statistically significant, variation in which days people are born on. I wanted to see if Google searches could be used as an indicator for this variation in births. Search behavior has been used to create indicators in the past, such as those for the flu, as well as a whole variety of financial and economic metrics (paper here), so why not for birthdays?
I compared birth data from 1994, binned into weeks, with Google Insights for Search data from 2009 for search interest in the word “birthday,” also binned by week. And, unfortunately, there is not a great correspondence:
Google search data explains less than 30% of the variation in the data, assuming that birth variation is constant each year. Apparently, people do not search for “birthday” on their own birthdays. A better indicator of birthday variation is likely its appearance on Facebook (or maybe Twitter), due to people wishing each other a happy birthday. Anyone want to try that analysis?