Never mind the numbers, what about the social trend?

Massive social transformation!

New, sexy social stereotypes!

Bit of gender-war thrown in!

This story had the lot.

Was itwidely reported? You bet.

Here it is: 

TEN TIMES MORE STAY-AT-HOME DADS THAN TEN YEARS AGO. 

"The number of UK dads staying at home to care for young children has risen ten times in as many years, according to new research revealed today." 

Here's a general rule of statistical reporting: if it sounds amazing, don't expect it to be true. 

Well, it sure sounds amazing. So guess what? 

According to the Labour Force Survey that formed the baseline data for this supposed change, in 2000 the proportion of males who gave 'looking after family/home' as the reason for their economic inactivity was 5.9 per cent. 

In spring 2010, it was 6.2 per cent. 

Is that an increase of "ten times" - or 900 per cent?

I make it an increase of about 5 per cent (though 5 per cent of a changing total). Wrong then, by about 895 per cent? Quite possibly. Though who knows. Because, frankly, with this data we have no way of telling.

So how did Aviva, an insurance company, come up with its number - summarised in its press release quoted above? Why did reporters call on commentators to offer convincing explanations for it? And why did the media swallow it in the first place?

Basically, they tried to count the number of stay-at-home dads amongst 1 million men and compared it to a number for stay-at-home dads amongst 10 million men ten years later. Surprise, surprise, there were about ten times more. For an explanation of how they managed to make this mistake, go to the end of this piece. 

Result? Rubbish? Very likely. Checked by the media? What do you reckon? 

But what would checking mean? It would mean going to the Labour Market Review from the Office of National Statistics and following your nose to the historical inactivity rates where you can find decent stats on inactivity and the reasons for it, compared on a consistent basis. 

But, hey, that's not much fun compared to inviting all sort of interest groups and commentators to offer explanations for this dramatic social change, describing female breadwinners, new men and the like. 

The story shifted from the premise - that there has been a big change - to the explanation without pausing to think about whether the premise is true. Thus are mythical statistics born. 

There might be some such changes in society. Was this survey a good way of measuring them? No. Was it in any way a defensible piece of statistics? Decide for yourself.

Is the magnitude of any such change likely to be as great as reported? Let's say that I'd be surprised. Are we a lamentable trade, suckers for any survey that comes across the desk?

A spokesperson for Aviva said she was convinced that this was a growing trend, but would check the figures. 

Picking up on these things becomes dull after a while; there are so many. But every now and then one so rattles the cage that ... So, apologies to regular readers.

How they did it:

First, Aviva had to find out what was happening ten years ago. 

So it went to the official Labour Force Survey data, as published in Social Trends. They noted that in 2000 there were 3 million men classified as 'economically inactive'. 

Some 5.9 per cent of these - 177,000 people - gave 'looking after the family/home' as a reason for their inactivity. 

But Aviva took a much smaller figure, by looking at the wrong column: a subset of the 3 million - about a third of them - who were also looking for work but had been unavailable in the last four weeks. 

By taking account of only a third of the economically inactive men, the number 'looking after family/home' is also reduced - to a third of the real number. 

That's why Aviva says there were only 60,000 stay-at-home dads in 2000. 

So the first error was to misread the official data. The next was to compare it with its own survey. 

Aviva commissioned a survey of just over 1,000 people (the Labour Force Survey is about 100,000 people). About 1,000 people presumably means about 500 men - not a large sample. 

But are these men all 'economically inactive' as in the baseline data? No. All the survey report on the company website says is that they all have dependent children. 

So are they really even stay-at-home dads at all - or do some of them have jobs? Probably an awful lot of the latter. Whatever, 6 per cent say they are 'primary carers'. Six per cent of 500 men is about 30. 

The survey then said that there are about 10 million men with dependent children, so extrapolated its 6 per cent to a population of 10 million to produce a total of about 600,000 stay-at-home dads (albeit, with or without jobs). 

Then it compared this with the 60,000 figure for 2000 and said: 'wow!'

That is, Aviva compared a wrong proportion drawn from official figures in 2000 with an extrapolation of a small survey today, based on different criteria.