The test should simply consists of a series of encounters with increasingly less polite people to see whether you are sufficiently well integrated to have mastered the acceptable expressions of raising your eyebrows, quiet huffing and, at it's worst, barely audible tutting

Never mind the p-values, who cares about mediation that explains 5% or less of what is not a huge effect?

We had a flight when my son was 1, 1.5 and 3 years old. At 3, he was fine - just a bit bored. At 1 and 1.5 it was tough - overtired and uncomfortable means loud crying.

Well done, congratulations!

And lots of studies that do not need to be written, because they lend themselves to a neat design.

Fully agree, I think trying to write the final study on a topic is at least equally harmful. You see some of the consequences in applied microeconomics - lots of important questions that do not get answered, because clean causal identification is not feasible.

INTJ (I Need to Try this Journal)

In all fairness, some statistical methods are so much fun that you should be allowed to write exactly one paper whose entire motivation is "we wanted to apply this pretty cool method to some data we had lying around"

The big caveat is that our results only cover women giving birth before age 30, since we do not have data on pregnancies after age 30 (i.e., no instrument!). 8/8

This adds complexity to the results - it is possible that women giving birth later earn less because they can afford to do so, since they can rely on the income of their co-resident partner! 7/8

We also find that women giving birth later have shorter birth intervals and they are more likely to live with a partner (and more likely to live with the father of their child). 6/8

But interestingly, a later birth also reduces earnings for those who are working by roughly 6%!

This effect seems to largely come from lower working hours - a later birth increases the likelihood to work part-time at age 46. 5/8

One innovation of our paper is that we look at earnings at the extensive and intensive margin - our estimates for total earnings are not very clear. We find some suggestive evidence that a later birth improves labour force participation. 4/8

We use the occurrence of contraceptive failure as an instrumental variable for age at first birth.

I know what you want to say...😬 but it turns out that contraceptive failure is not just a strong predictor of giving birth earlier, but it is also fairly random. We checked, I promise! 3/8

We try to answer the question how the age at first birth affects labour market earnings in the long-run, using data from the 1970 BCS (=UK women born in 1970).
Earnings here are measured at age 46, mostly. 2/8