[Update: Monday 23rd May 2016- There is an additional follow up article here]
[Update to the Update: and a second followup here]
538.com wrote a post about U.S. births and Friday 13th
Andrew Gelman (who did work in 2012 the 538 article was based on) wrote a response
Which Statschat mentioned
Which then prompted me to check the New Zealand situation
The New Zealand data
There is public data on New Zealand births by day at Statistics New Zealand (see the Excel spreadsheet on the right, and the sheet with the raw numbers within it)
However, this is aggregated across the period 1980 to 2014, so unlike the US data, we do not know the individual year’s births, and thus the day of the week the births fell on. This means we cannot identify a specific group of “births on Friday 13th”, so cannot apply the same methods discussed in the other articles. But it would be none the less interesting to know if New Zealander’s are afraid of Friday the 13th enough for it to noticeably affect the demand for scheduled deliveries (noting that New Zealand has a largely public health system rather than private payments which could also affect the degree to which superstitions are accommodated).
What we can do is this- If we check for how many Friday the 13ths there were in each month across the years 1980 to 2014 (inclusive) we find that most months have had 5 Friday the 13ths, except for June (with 6 Friday the 13ths) and October (with 4 Friday the 13ths). If there is some kind of casual link between fear of Friday 13th and demand for scheduled services, we would expect some kind of correlation. If there is some kind of correlation, we would expect the number of births (relative to the seasonal pattern, which in New Zealand is 9 months after New Years) born on the 13th in months with fewer Friday the 13ths to be greater than months with more Friday the 13ths contained within the data.
Unpicking the graph, the pale green background is the typical births for the month, based on births on the 4th, 5th , 6th, 7th, 8th, 18th, 19th, 20th, 21st, and 22nd of the month- dates far enough away not to be influenced by the presence of Friday the 13th, but also not on significant annual dates like Christmas or New Years day. The dashed green line is the mean of the aggregated births for these dates, and the coloured green background extends two standard deviations above and below the mean.
The black line represents births on the 13th of the month, the coloured dots are how many Friday the 13ths were in the aggregated data period.
While births on the 13th of June, with 6 Friday the 13ths, is well below the mean for June, it is about as below the mean as April which only had 5 Friday the 13ths. More importantly for a correlation, the births on the 13th of October (4 Friday the 13ths) are well below the mean, and well below the births in the neighbouring months (which both contain 5 Friday the 13ths). This is the opposite of what might be expected in a correlation (where the expected result was more births in the months with the fewest Friday the 13ths).
If all the results were the opposite way round, we might have evidence that there was a relationship and it was the opposite of what we expected. With the mixed bag of no particular pattern, the exploratory analysis suggests there is a lack of evidence that fear of Friday the 13th is connected to demand for obstetric services in the New Zealand public health system.
The R code used to generate the above:
monthnum <- 1:12 dts <- seq.Date(from=as.Date("1980-01-01"), to=as.Date("2014-12-31"), by=1) library(lubridate) thrt <- dts[weekdays(dts)== "Friday" & mday(dts) == 13] number13s <- table(month(thrt)) download.file("http://www.stats.govt.nz/~/media/Statistics/browse-categories/population/pop-birthdays-table/most-common-birthdays-19802014.xlsx", destfile="b.xlsx", mode="wb") library(readxl) bnums <- read_excel("b.xlsx", sheet = 2, skip=2) names(bnums) <- "day" thirteens <- bnums$day == 13 comparison <- bnums$day %in% c(4,5,6,7,8,18,19,20,21,22) #comparison dates 4,5,6,7,8,18,19,20,21,22 bmeans <- apply(bnums[comparison,2:13],2, mean) bsd <- apply(bnums[comparison,2:13],2, sd) bhigh <- bmeans + (2 * bsd) blow <- bmeans - (2 * bsd) upper <- max(c(bhigh,blow)) lower <- min(c(bhigh,blow)) plot(monthnum,unlist(bnums[thirteens,2:13]), type="n", ylim=c(lower-50,upper+50), frame.plot = FALSE, ylab="NZ births", main="Births 1980-2014, Statistics NZ", xlab="month") polygon(x=c(0:13,13:0), y=c(bhigh,bhigh, bhigh,blow,rev(blow),blow), col="#00FF0011", border="#00FF0022") lines(monthnum,bmeans, col="#00550055", lty=2) lines(monthnum,unlist(bnums[bnums$day==13,2:13]), type="l", ylim=c(lower-50,upper+50)) points(monthnum,unlist(bnums[bnums$day==13,2:13]), col=number13s, pch=19) legend(x=10, y=6200, legend=c("4 Fri 13ths", "5 Fri 13ths", "6 Fri 13ths"),pch=19, col=c(4,5,6), bty="n", cex=0.6, y.intersp=1.2)