Saturday, June 17, 2006

Defending the Daycare in Quebec Study

The feminist blogs have been trying to attack the methodology in the Baker-Gruber-Milligan study of daycare in Quebec. (Abstract here).

The methodological criticisms don't hold up.

The first criticism, by the Human Early Learning Program, is that the study does not compare kids in daycare with kids not in daycare, but kids in Quebec during the time universal daycare was introduced with kids in the ROC at the same time. This is not a bug, it's a feature. One of the limitations of previous studies -- which involved that very comparison -- is that people who choose to put their kids in daycare will be different, in a whole number of ways, from people who make the opposite choice. These differences "confound" the results. It's nobody's fault, but it is a methodological limit of that type of comparison.

Quebec's decision to create this new program -- at a time when no other jurisdiction in Canada was doing the same thing and when high quality data in the National Longitudinal Study of Children and Youth (by StatsCan) was being taken across Canada using the same methodology -- provided an opportunity to side-step this problem. The policy change meant that the very same kind of people who were not putting their kids in daycare in the ROC were doing so in Quebec. So that avoids the particular confounding problem at issue in earlier studies.

The second criticism -- that it is odd to use 6 year-olds as a control group for per-schoolers in daycare -- is also based on a misunderstanding. The basic comparison is between the pre-schoolers in Quebec during the introduction of $5 per day daycare, and per-schoolers in other provinces. But this could pick up trends where Quebec and the ROC are diverging for other reasons. The six year olds are a control to prevent that -- to the extent that six-year-olds in Quebec are getting more throat infections or are more inclined to hyper-activity than their counterparts in Ontario, that trend would be attributed to global differences, and only the divergences of the pre-school kids would be attributed to the policy change.

One commenter on Lawyers, Guns and Money suggests that the study may be picking up declines in daycare quality as a result of the massive expansion of spaces. Baker-Gruber-Milligan do address that possibility, and cite data that -- at least in terms of training and certification -- there was no decline in quality in Quebec either relative to the past or relative to the ROC.

So the results are not really in dispute. $5 per day daycare led to a lot more kids in daycare, more labour force participation by women, increased health and behavioural problems for children and more reported unhappiness and stress by both children and parents.

As with any empirical study, there are still unresolved interpretive issues. It is possible that what is being picked up in the findings of greater stress and health and behavioural problems is a one-time "socialization shock" that happens to all kids, sooner or later, when they go from a home setting to an institutional setting (and to all parents suddenly dealing with the double shift). It could be that experiencing this earlier means that it is less of a problem later on -- and there is other empirical evidence that kids who have been to daycare are better socialized when they reach K-12 schools.

Moreover, B-G-M does not purport to be a full cost-benefit analysis of daycare. There may well be offsetting benefits to the parents and to the family of the higher income, greater work continuity and so on.

This survey doesn't change my own policy leanings, which are for pro-natalist subsidies which we let parents decide what to do with. This is the one major issue I side whole-heartedly with the Conservatives on*, and I can easily see it delivering them the election.

*I side uneasily with them about Afghanistan. I'm against them on Kyoto, Kelowna, taxes and softwood.

Update: An award for pith and substance to Christine who manages to say this better and shorter here.

No comments: