Surviving the hazard (yes, that is what passes for a demography joke…)

Hill Kulu gave a seminar on housing and fertility at Southampton last week.  He did use the words “linear spline” on a couple of occasions, but otherwise it was extremely accessible.  He had used a hazard model to examine housing type, house moves and conception resulting in a birth with data from the Finnish Longitudinal Fertility Register (those Scandinavians really know how to collect data…).  The ‘hazard’ in this case is becoming pregnant.  Essentially, he was trying to pick apart ‘having a baby for which you have prepared by moving house’ and ‘having a baby because you already have plenty of space in your house and so it seems like a good idea’.  (Moving house in anticipation of having a baby is an unobserved factor influencing both housing choice and conception.)  A house-move increases the risk that an individual will experience the ‘hazard’ of a pregnancy.

Kulu looked at the time between a house move and a conception: he suggested that a short gap might imply a move planned for child-bearing purposes while a long gap might imply that the decision to have a child was more closely connected to the house than to the move.  It’s hard to draw more than tentative conclusions, of course (and one of our most quantitative colleagues was even moved to suggest qualitative methods).  On that note, Kulu pointed out that some of the ‘long lag’ births (say, a conception four years after a move) could have been preceded by a brother or sister who was born much closer in time to the house move (this would mean a child conceived in the first year after a house move who is followed by a sibling some two years later).  I’d like to add that the conclusions also assume good fertility control – that women have babies when and exactly when they want them.

Aside: the analysis obviously needed a number of control variables – income, for example.  In the Nordic countries, apparently, it is standard practice to use the woman’s income alone as a stand-in or proxy for family income when you do this kind of analysis.  For this method to work, women’s work/pay patterns must be similar to men’s.  I bet you couldn’t do it here…

This entry was posted in Academic research, Demography and tagged , , , . Bookmark the permalink.