Adam Smith Institute

View Original

Dear Polly, to criticise statistics it is necessary to know statistics

Polly Toynbee is outraged:

Rattling through a lexicon of lies at prime minister’s questions is so routine that few bother to call out Boris Johnson any more. Besides, on Wednesday all attention was on the vendetta playing out elsewhere in Westminster. But when the Labour MP Gareth Thomas challenged the prime minister on the steep rise in child poverty revealed in official figures this week – 4.3 million children and heading upwards on a steep curve – Johnson boasted shamelessly: “We are seeing fewer households now with children in poverty than 10 years ago.”

Thomas protested at “Boris Johnson’s casual disregard for the truth”. But lies seem to work very well for him, and they’re eagerly echoed as fact by those on the Tory benches.

The numbers are here.

The problem with Polly’s outrage is that there’s nothing, at least in principle, to stop both being true. That there are more children in poverty and also fewer households with children in poverty. We admit that we have not checked either statement but that’s not the point we’re making. If poverty has become more concentrated into certain households then that would explain both.

We might even expect this to happen:

The universally used measure of poverty, in Britain and internationally, is relative, counting anyone living below 60% of a country’s median income.

Well, no, it’s below 60% of median household income adjusted for household size. That adjustment for size not quite, wholly, working. If the modal family is two adults working - which it probably is - then families with large numbers of children, which are more likely to be single earning families, are more likely to be in that relative poverty. Certain people, perhaps, preferring a wealth of offspring rather than storing up consumer riches.

There is a further amusement from Polly’s statistical observations as well:

For obvious reasons, the government grasps on to a different measure called “absolute poverty”. Here’s the oddity of this number: it is anchored in 2010, so it measures how many children are still living on what was the 60% below-median level 11 years ago, when the median was of course far lower, as it rises with growth.

How excellent, so, we all agree that economic growth does matter, also that it has been happening in this decade of austerity then and that average incomes have been rising as a result?