A couple of months ago, the statistic that only 1% of men had taken up shared parental leave was splashed all over the British media. (Shared Parental Leave was introduced in the UK in 2015, and essentially allows parents to share 12 months of leave however they like. Taking it consecutively, simultaneously, alternating blocks of leave between both parents, or a mixture of the above are all permitted.)

My Family Care, the company that carried out the survey on which this statistic was based, apparetly asked Human Resources directors at 200 businesses what percentage of men in their company had taken shared parental leave in the year since it was introduced. But, as Radio's 4's excellent More or Less programme pointed out, they forgot to ask what percentage of those men were actually elligible for parental leave in the first place! Most of them won't have had children at all in the last year. Some of them won't even have any children!! Ooops.

Clearly, the fact that 1% of all men (fathers or otherwise) have taken up shared parental leave tells us next to nothing about the take up of shared parental leave. We can try to extract from this a very crude estimate of the percentage of eligible new fathers taking it up, using average birth rate figures. We definitely shouldn't be doing this, for all kinds of reasons. For one thing, applying the average birth rate to the demographic of employed men is so dubious it's almost certainly plain wrong. The birth rate amongst employed men will differ substantially from the overall birth rate per head of population - the latter figure includes retired men, for example. (This will skew the estimate far more than the points My Family Care later raised: that adoptive parents and same-sex parents are also eligible, and some fathers won't have worked long enough to qualify.) More heinous still, we're ignoring all the uncertainties in the data. With such small samples and small percentages, the uncertainties probably swamp the data and invalidate any conclusion we might try to draw. But let's blithly get out our envelope anyway, turn to the back, and see what comes out!

The UK birth rate is currently about 12 babies per 1000 population per year, or about 24 per 1000 men.^{11 1}The real figure is actually closer to 25 per 1000 men, as there are more women than men living in the UK. But we're only after a crude estimate anyway. The survey indicates that 10 out of those notional 1000 men (1%) will take parental leave. All of those must clearly have been eligible for parental leave, so those 10 must be amongst the 24 who had babies in the last year. Thus we arrive at an estimate that 10 out of every 24 men - roughly 40% - are taking shared parental leave. I wouldn't trust this figure at all. But it does demonstrate that the 1% statistic could easily be telling completely the opposite story to the headlines written about it.

How a company promoting family and childcare issues can carry out such an inept survey about parental leave is a mystery to me. There's some indication they realised their statistic was - in the words of More or Less programme producer Charlotte McDonald - "one of the worst statistics they have ever looked at". In the report on their web site, My Family Care do state that this is "just 1% of men (that is, all men, not just eligible men)". Apparently this didn't make it into the press release they put out, though.

I'd like to say that how all the main UK media outlets failed to do even a cursory fact-check and spot this before blazing inaccurate headlines all over their front pages is also a mystery to me. But there are so many examples of the media reporting complete nonsense when it comes to statistics that it doesn't surprise me in the least.

In a later post on their web site, My Family Care try to defend their survey, saying that "a few sources have questioned our stats, which is to be expected when the policy is in its infancy and where the data available is still relatively small." They also point out that "finding the true number of eligible men is not simple maths." Very true. All the more reason to find out the number of eligible men during the survey itself…by asking!

My Family Care have their hearts in the right place. Just not their statistics.

There are more robust statistics available on the older Additional Paternity Leave scheme, which started in 2011. These statistics seem to at least measure what they claim to measure (e.g. these figures from the TUC). According to this, the percentage of eligible fathers taking additional paternity leave was well under 1%.

But even that statistic is misleading in a more subtle way. Under the old additional paternity leave scheme, fathers could take up to 6 months leave if the mother returned to work earlier. Of the under-1% of fathers taking up this option, how many took 6 months, and how many took substantially less than this?

I prefer the far more informative way Sweden quotes its paternity leave statistics: 25% of all the parental leave in Sweden is taken by fathers. That's 25% of the total number of weeks of leave taken, not 25% of fathers. You'd guess from this that the percentage of men taking any paternity leave at all - as that UK 1% figure measures - would be much higher in Sweden. And, indeed, by 2005 already 90% of Swedish fathers took at least some paternity leave.

In a couple of years' time, the UK government plans to report official figures on shared parental leave take-up. When they do, watch out for them reporting it as a percentage of eligible fathers, and not also giving the percentage of total parental leave taken by fathers.