honestlyreal

I saw a statistical tweet this morning: the sort that makes me pause and think. What is it really saying, and what is the story that could be told?

The average local government public sector pension is £4200 a year. For women it is £2870. Gold-plated? Really? I think not.

Now, setting aside the comment about gold-plating (with which I heartily agree, for the record—these numbers are insultingly small) and the fact that no source is given, which is perhaps forgiveable given the space constraints of a tweet—what is the actual story here?

We don’t know if the £4200 figure is for male pensioners, or for all pensioners. It makes a big difference to the resulting headline.

This is why.

Let’s say the £4200 is the average male pension. This gives us headline comparisons that “women get 19% less than average, and 32% less than men”. 32%. Gulp. Powerful, huh?

But what if that £4200 is the average across all pensioners? To get any further, we need some information about the male:female ratio of local authority pensioners. Let’s assume 50:50 on the basis of no further information.

Now our headline comparisons are “women get 32% less than average, and 48% less than men”. Wow. This is looking much worse. (I’ve resisted the crap journalism temptation to put “over 48%”, by the way. Rounding is rounding.)

However, I’ll go out on a limb here and suggest the male:female ratio is tilted towards female pensioners. This is based on women living longer than men, having an earlier retirement age, and my gut feel that large swathes of local government employment have lots of women.

If we go for 35:65 male:female, we now get the headline comparisons: “women get 32% less than average (as before), and a whopping 57% less than men”.

Now that is a headline. (But are you getting a funny feeling about the attempt to compare one party with another, and with their collective average, in a two-party situation? Perhaps you should?)

So there you have a story about a 19% disadvantage, or a 57% one, from the same numbers, depending on how you interpret them. That’s quite a difference.

Were I writing up this story as a data journalist or campaigner, I’d want to pin down this type of population detail, and also correct for differences in the amount of time worked over a career for each sex (assuming that this contributes to calculating a pension figure).

I’d also want to show how much the underlying differences in base pay rates were making to the pension figures. Were historical (and current!) disparities in pay passing straight through to affect pension entitlements, or were there other factors (and perhaps inequalities) at work?

And is this getting better, compared with data from say 10 years ago, given some of the recent moves towards addressing long-standing gender inequality in pay?

Lots there, isn’t there?

We’re opening up a lot of data.

Let’s make sure we open up our analytical skills to match.