Sarah Roberts

For this post, I interviewed Sarah Roberts.** She is an Associate Professor at the University of San Francisco in the Department of Obstetrics, Gynecology and Reproductive Sciences. The focus of her work is on how policies affect vulnerable pregnant women, including those who use alcohol and drugs, and those who seek abortion.

I interviewed Professor Roberts about her paper “Associations Between State-Level Policies Regarding AlcoholUse Among Pregnant Women, Adverse Birth Outcomes, and Prenatal Care Utilization:Results from 1972 to 2013 Vital Statistics.” Her main co-author on this paper was Dr. Mina Subbaraman.  

In this paper, the authors examined the effects of policies that target alcohol use during pregnancy. The types of policies that exist vary by state, and the authors looked at 8 policies altogether. Here I’ll focus on three of them: Mandatory Warning Signs (MWS), Priority Treatment (PT), and Child Abuse/Child Neglect (CACN). MWS states that “notices must be posted in locations where alcoholic beverages are sold. . .” PT allows pregnant women to go to the front of the line for substance use disorder treatment. . .,” and CACN “. . . in some cases, defines alcohol use during pregnancy as child abuse or neglect.”      

One might expect, and hope, that these policies would lead to reductions in adverse outcomes at birth. The authors used several measures of these adverse outcomes. Two examples were low birth weight and pre-term birth. The policies should also encourage women to take advantage of health care more regularly, in particular prenatal care. Their measure of this was called prenatal care utilization. The authors were able to obtain data (1972-2013) from birth certificates in all 50 states. For over half of this time period, they have every single birth certificate (!). For the rest of the time, they have 50-100% of birth certificates. 

The results? At best the finding was that the policies have no effect on birth outcomes and prenatal care utilization; and at worst, instead of leading to reduced adverse outcomes and increases in prenatal care utilization, some of the policies did just the opposite.  For example, living in a state with MWS was associated with 7% higher odds of low birth weight, 4% higher odds of pre-term birth, and 18% lower odds of any prenatal care utilization. The authors calculated that in 2015 there were about 7,000 excess low birth weight and pre-term birth babies due to MWS; and there were 6,000 excess low birth weight and 12,000 excess pre-term birth babies due to CACN policies.

So, what’s happening here? In the case of CACN, it is likely that women who use alcohol during pregnancy fear they will be reported to Child Protective Services by a prenatal care provider and that  they will then lose their children and go to jail. In fact, Professor Roberts previously interviewed pregnant women who use drugs, and the women revealed exactly this fear. In the case of MWS, these signs may contribute to stigma, and hence discourage women from disclosing their alcohol use, and asking for help. Also, some women may fear that they’ve already harmed their baby, and not realize that stopping drinking at any point in their pregnancy can make a difference, and that getting prenatal care can also help. These women miss out on prenatal care, which causes them to miss out on other, related services as well. 

I asked Professor Roberts about how she would design policies differently, given these results. She told me, “people don’t just start drinking once they’re pregnant. . . most of the research about predictors of drinking during pregnancy says that it’s what you were doing before you were pregnant that predicts what you do while you’re pregnant. . . if we can think of this as alcohol policy instead of pregnancy policy, we may be able to better reduce harm for pregnant women.” For example, we know that states vary in how they control access to wine sales. And the authors actually do find, in this paper, that outcomes were better for pregnant women in states with more of this control. So, this is just one example of the type of population-level policy that might lead to improvements for pregnant women. 

I asked Professor Roberts about some of her other recent work. Dr. Katie Woodruff, who works with Professor Roberts, is now interviewing pregnant women about their perspectives on MWS in relation to cannabis use, and in particular in the context of its legalization. 

   

Let’s talk! I would love to know what you think about this example of unintended consequences. Please submit comments and questions

**Thanks to Samantha Valente for getting me started on this research path. It's so interesting and important! I thoroughly enjoyed it!

Christina Gravert

 

For this post, I interviewed Christina Gravert. She is an Assistant Professor at the University of Copenhagen in Denmark. She is a behavioral and experimental economist. Her research focuses on why people don’t always do what they plan, and how to design policies that help them to do so, in their best interest. 

 I interviewed Professor Gravert about her paper “The Hidden Costs of Nudging: Experimental Evidence from Reminders in Fundraising,” which she wrote with Mette Trier Damgaard. In this paper, the authors carried out a field experiment, which is like an experiment in the lab, but it’s done out in the real world. They worked with a charity to determine the effects of email reminders about giving. These reminders are examples of what behavioral economists call nudges, small deliberate changes that are made to improve well-being. The charity, of course, hoped that their reminders nudged their members to give, and hence increased overall donations. 

There were two groups of donors in the experiment. One group got a single email asking for a donation (the control group). The other group also got a reminder a week later (the treatment group).  This reminder has what the authors call an annoyance cost. This is a very nice example of a non-monetary cost, which economists always include, in addition to monetary costs, when trying to understand decisions people make. Here the annoyance can be the guilt felt or the perceived pressure to give, and/or the time it takes to read the email.  However, Professor Gravert emphasized to me that all of the people in this experiment had donated to the charity before, and had opted in to receiving emails.

So, what were the consequences of the reminder emails? In general, the results were that people don’t donate unless they receive an email, and they tend to donate within a day or two of receiving an email. The reminders increased the number of people donating by about 66%. However, the reminders had another unintended consequence, which was to increase the rate of unsubscribing from emails from the charity, due to the annoyance cost. In the treatment group this rate was 3.7%, and in the control group it was only 2.1%. Doing the math, the difference between these is 76%. Overall, it’s hard to say in the long run whether reminders will help or hurt donations. The 66% is the effect on one-time donations, and unsubscribing is permanent (unless the person decides to sign up again with a different e-mail address). The net effect depends on several factors, including the relative value the charity places on donations now versus those in the future. In the paper, the authors give a range of estimates for the net effect.  

I asked Professor Gravert about alternative strategies that charities could use to avoid this unintended consequence. She told me that charities often target their reminders based on demographic factors of their donors. She recommends that they instead examine their donors’ behavioral patterns more carefully, and target on these. For example, some people always give towards the end of the year, to take advantage of the tax benefits. Another strategy, already used by charities, is to encourage people to opt-in to a direct debit donation scheme, eliminating the need for reminders. They could also allow donors to adjust the frequency of the emails themselves, or they could use text messages instead of emails. 

Speaking of text messages, in related work in progress, Professor Gravert and co-authors Kai Barron, Mette Trier Damgaard, and Lisa Norrgren, are conducting another field experiment with pregnant women in South Africa. Over the course of three months, they are using text messaging to remind the women to take their iron pills, with the goal of identifying more precisely which type(s) of annoyance costs might cause unsubscription. This is a very nice example of a reminder that has the best interest of the person in mind, right?

Let’s talk! I would love to know what you think about this example of unintended consequences. Please submit comments and questions.