Births by Day of the Year

I've prepared this page as a response to the End of year effect? blog entry in Kaiser's Junk Charts blog.

Kaiser cites a December 20, 2006, New York Time article, To-Do List: Wrap Gifts. Have Baby. which claims that parents routinely schedule induced or caesarian births just prior to the first of the year, to gain a tax deduction for an additional dependent child for the entire year. The article includes a very uninformed graphic (shown at right) that shows births for the narrow period from 23 December 2002 to 10 January 2003, which supposedly backs up the claim. However, if you take averages of the plotted numbers, the January numbers are greater than the December numbers.

The chart suffers from a number of problems. The greatest problem is that the chart only shows 19 days of data, not enough to account for the day-of-the-week variations which become obvious when looking at a whole year's racords. In addition, it has special colors for Mondays and Fridays, without indicating what makes these days of the week special.

I went to the web site of the National Center for Health Statistics (part of the Centers for Disease Control and Prevention) to find this data and correct the shortcomings of the chart. I could not find the specific years that were plotted in the Times article, but I found complete year data for 1994 through 2001.

I constructed a panel chart below, showing data for the period from 1 November of one year through 31 January of the following year. The actual data are shown with blue lines and labels with the initial of the day of the week(S-M-T-W-T-F-S). The red lines show the average by day of the week for the three month period, not including the Monday-through-Saturday period around Thanksgiving and not including the period from 21 December through 4 January. The day-of-the-week labels for Thanksgiving, Christmas, and New Year's Day are highlighted with a red-outlined, yellow-filled square. The panels range from 1994-1995 at the bottom to 2000-2001 at the top.

An obvious pattern for births by day of the week is seen: Weekend births (around 8000 per day) are significantly lower than weekday births (about 11,000 to 12,000 per day). This is due to scheduled births (both induced normal births and non-emergency caesarian births) taking place during the week, not on weekends. This daily pattern is consistent for most weeks of each three month period in this chart.

When a more complete record of the data is viewed, the assertion of the New York Times article can be examined more closely. It is obvious that births on 1 January are markedly lower than would be expected for the day of the week. Births on 25 December (Christmas Day) are also lower than expected for the day of the week, by an even greater amount than the New Year's Day variation. In fact, the day of the year with lowest number of births is Christmas Day. Thanksgiving also shows a pronounced decrease in births compared to typical Thursdays, and perhaps more of an effect than New Year's. The day of the year with the greatest number of births is either the Tuesday before Thanksgiving (as in 2000), the Tuesday before Christmas (1999), or the Tuesday before New Year's (1998).

It is evident that more births are induced just before these three holidays, so the patients and doctors are not tied up during the holidays. I plotted data for the entire year 2000 to see whether this was true for other holidays during the year. The blue line shows the actual data, while the red line shows the value for each day of the week averaged over the whole year. Sure enough, while the general weekend-weekday pattern held for most weeks of the year, the chart below shows that Memorial Day, Independence Day, and Labor Day also had declines in number of births compared to the expected number for the day of the week. These declines were significant but not as pronounced as those on Thanksgiving, Christmas, and New Year's Day.

Because the New Year's dip in births is not as pronounced as those at Christmas and Thanksgiving, I surmise that the holiday effect of induced births is due more for the convenience of patients and doctors than simply for the tax deduction.