Apple made news recently by announcing that it will pay the costs of egg freezing for employees, thereby potentially extending fertility later in life. Brigid Schulte sees the policy as a corporate tactic to justify 90-hour work weeks, whereas Katie Benner sees value in giving workers more flexibility in their family-building options (although as Megan McArdle points out, spending your 40s and 50s chasing toddlers around is challenging, frozen eggs or no). Now that the ethics and impacts have been debated by subtler minds than my own, I step in as a bloodless behavioral scientist to say that this is a terrific opportunity to research how human beings make important decisions.
Up until now, there was no good way to study the effect of freezing eggs on child-bearing decisions because of self-selection, i.e., the women who chose to pay for the procedure were fundamentally different from those who didn’t in ways that independently affected their likelihood of becoming mothers (e.g., education, income, planfulness). But now we have an excellent instrumental variable to exploit in quasi-experimental analysis (nice explanation by Austin Frakt here): Was a woman working at Apple when the egg-freezing policy was announced or was she working at a tech company that didn’t have the same policy?
The logic here is that whether those female Apple employees freeze their eggs or not will now be determined not only by the usual self-selection factors, but also by something exogenous, namely the company’s new policy. Presuming (reasonably) that they did not know that the policy was going to be implemented before they chose to work at Apple versus a different tech company, the instrument should exert no effect on the outcome except through the policy in question.
So, what would you hypothesize? Will more egg-freezing lead to more or fewer women becoming mothers? On the one hand, one could argue that knowing that the eggs are frozen would lead more women to delay childbirth for longer, perhaps to the point where it was no longer possible. This would mean that the egg freezing option lowers the likelihood of giving birth.
But I expect the opposite, because of two countervailing forces, sunk costs and public commitment. Egg freezing involves a significant amount of work, which will be experienced as a loss if the woman never has kids. Of course, the work has been done regardless, but there is good evidence that that is not how we think about sunk costs. Rather, we want to justify them by following through. Public commitment would also work in favor of more births. We are generally more likely to make major behavioral changes when we make others aware of our plans. If you tell yourself you are dieting and leave it at that, you are less likely to lose weight than if you decline a dessert and say loudly to everyone at the table “No thanks, I am trying to lose 10 pounds”. Egg-freezing involves some public commitment to a future behavior; you don’t have to announce it to your co-workers but you will obviously tell the health care staff and perhaps people close to you as well. That in itself makes it more likely that you will follow through, even beyond the sunk costs effect.
So that’s my bet, to be evaluated empirically with an instrumental variables study: This policy will lead more, not fewer, women to become mothers.
7 thoughts on “Corporate Egg Freezing as a Research Opportunity”
My guess–small sample size(s) will nullify hopes of gaining statistical insight(s) into these behaviors.
Apple has over 15,000 female employees, some will be post-menopausal or have illness that prevent them from becoming pregnant, but even if those groups constitute a 1/3 of the workers, that's a lot of data to play with..
Given rising income inequality, this will lead to more surrogates.
I'm not sure your prior variables — education, income, planfulness — are controlled in the new situation, as I would have to assume that women who get hired by Apple represent those…pre-hoc and post-hoc….as women with more education, income, and planfulness.
I voted up your comment to help rescue you from redness. What you are raising is not a threat to internal validity as it would be in the fully self-selected case. It's possibly a limit on generalizability (depending on the population that freezes eggs, which may or may not be different), but that is a different question.
It may have a greater effect on recruitment than on births. Providing a potentially useful benefit were none had existed before may attract more people of both sexes (especially if employees' partners are included). Moreover, it would also benefit those who haven't been able to conceive naturally. That should be a control variable in your natural experiment.
Anyone who joins Apple after the policy is announced could not be in the study, as it would violate the IV assumption that the variable only affected the outcome via the route being evaluated. People who join after it was announced may have self-selected and therefore we can't conclude anything from their data.
Comments are closed.