In Episode 2 of Series 4 of the DIAL Podcast, we are in discussion with Professor Hans van Kippersluis from the Erasmus University in Rotterdam. Hans, Professor of Applied Economics, is the Principal Investigator on the DIAL project, Gene Environment Interplay in the Generation of Health and Education Inequalities, which has used innovative methods and data to explore the interplay between nature and nurture in generating health and education inequalities.
Transcript
Christine Garrington 0:00
Welcome to DIAL, a podcast where we tune in to evidence on inequality over the life course. In series four, we’re looking at what’s been learned from some of the DIAL projects about how and when inequality manifests in our lives, and what its longer term consequences might be. For this second episode of the series, we’re delighted to be joined by Hans van Kippersluis, Professor of Applied Economics at the Erasmus University in Rotterdam. And Principal Investigator of the DIAL project, Gene Environment Interplay in the Generation of Health and Education Inequalities – put more simply nature versus nurture. So Hans, welcome to the podcast. And I wonder if you can start by talking us through what researchers working on this project have actually been looking into.
Hans van Kippersluis 0:42
What we’ve been doing in this project is essentially incorporating the recent availability of genetic data into social science and most prominently economic analysis. And so most of our work has focused on the interplay between genes and the environment. So in the introduction, you mentioned nature versus nurture, but actually more accurately, what we’re doing is nature and nurture jointly into how they shape essentially education and health outcomes. And I think this is also the main innovation of our project, because biologists have studied nature before; social scientists have of course, extensively studied nurture, but not many have studied the interplay, the interaction between the two. And I think this was sort of the main innovation for why we got the funding some five years ago. And so what we have done is mostly studying this interplay. But along the way, we have also made some methodological contributions to a field which is very new. Then we’ve also used genetic data to test all their theories, and also, I think, enrich the framework of equality of opportunity.
Christine Garrington 1:35
Yeah, fantastic project. And as you’ve just said, you’ve made unprecedented use of genomic as well as survey data in the research, tell us a bit more about the information that you’ve been able to access? And how you’ve been able to use it?
Hans van Kippersluis 1:47
Yeah, sure. So the interesting thing is that more and more social science datasets, so data sets that have been traditionally used by social scientists, and these are mostly extensive surveys, are now collecting DNA information from their respondents. And this is often from blood or saliva. And what they did is basically, so more than 99% of DNA is the same across human beings. And so what we are using is only this remaining less than 1% of the variation. And these are called snips. And snips are points of your DNA that differ across human beings. And there’s roughly 1 million of them. And so what we do, basically also other people have done is sort of aggregating these tiny effect sizes into an index. And this is called the polygenic index. And this is telling us something about your genetic predisposition towards a certain outcome. And this is quite interesting, because this data, this new variable, essentially can be added to existing datasets. And so we have a wealth of information that has been collected in the past on surveys on existing data. And then we simply add one indicator, one new variable. This is telling us something about people’s genetic predisposition. And just to be clear, this is not like a deterministic variable. It also exhibits quite a bit of measurement error and noise. But at the group level, and that’s what we have been doing is it sort of does tell us something about your genetic predisposition, and it can help us understand how certain life outcomes like education, like health, are shaped by the interplay between your genetic predisposition and your environment.
Christine Garrington 3:07
Indeed, let’s talk a little bit now then about some of the research findings. And you know, what’s come out of this now, one piece of research we’ve spoken about this actually, in an earlier podcast episode, actually drew links between mothers smoking in pregnancy and their baby’s birth weight. I wonder if you can just sort of summarise that for you what actually came out of that what we learned
Hans van Kippersluis 3:28
this was work with with my PG students, Rita Dias Pereira and colleague Cornelius Rietveld . And for birthweight we knew that maternal smoking is one of the key environmental risk factors. And we also knew from genetic studies that genes matter in determining your birth weight. And so what we did here was essentially looking at the interaction between the two. So can higher polygenic indices protect against maternal smoking? And the answer, unfortunately, perhaps was no, in the sense that we found very, very little interaction between genes and the environmental exposure of maternal smoking. So it seems that both matter, but there doesn’t seem to be any meaningful interaction between the two. So that was, to some extent surprising, but on the other hand, also perhaps logical in the sense that maternal smoking is apparently such a devastating environmental exposure that even higher genetic predisposition cannot protect you from this.
Christine Garrington 4:16
Yeah, really interesting. And anybody who’s interested in that can listen to Rita actually discussing that in series three, Episode Seven, of our DIAL podcast called Mums Who Smoke and their Baby’s Birthweight. So do check that out if you’re interested to know a little bit more about what Rita and all of the all of your colleagues did. Now, there have been some interesting findings Hans from the project around the role of genes in a child’s education and specifically around parental investments. I wonder if you can explain a bit more about what you were looking to understand there.
Hans van Kippersluis 4:50
Yes, yeah, so this is one of my favourites studies. It’s joint work. Also with another PG student Muslimova and my colleagues Stephanie von Hinke, Cornelius Rietveld and Fleur Maddens. And the starting point there was actually a theory of human capital formation from economics. And it dates back all the way to the work of Nobel laureate Gary Becker. And one of the crucial assumptions in that model is that parental investments are complementary to your genetic endowments. And this assumption is actually very hard to test because often we do not have a good measure of endowments. And if we do, it may already be contaminated by parental investment. So many people, for example, use birth weights. But of course, well as we just learned, maternal smoking may have a large effect on your birth weight, so it’s not fully free of your parents’ behaviour. And the other thing is that your parental investments often respond to endowments. So if you have a child with specific needs, of course, parents respond to this. So the problem of testing this assumption is that endowments and investments are actually always very closely entangled. And that makes it very hard to test whether they are complementary or not. So what we did here was using one’s genetic endowment, and that is actually has a very nice property and that it’s fixed at conception, so it cannot be affected by your parental investments. And what we did was using the child’s birth order to proxy for parental investments. So what we know from earlier studies is that firstborns tend to get more parental attentions on average than later points. So this is one after all, because they have undivided attention until the arrival of later borns. And this extra parental investment is actually independent of your endowments. It simply derives from the fact that you have more time if you have one child as opposed to multiple children. So what we did in this study is looking within families comparing siblings that were first born to later borns, and then further analysing whether this firstborn advantage was stronger for firstborn siblings who randomly inherited the higher polygenic index for educatio. I think this was a nice, very unique setting to test this theoretical assumption that parental investments are complements to genetic endowment.
Christine Garrington 6:45
What did you find here? Then what do we learn about the role of genetics in affording in affording certain children advantages later on in life?
Hans van Kippersluis 6:53
So what we found was that indeed, the firstborn effect seems to be stronger for siblings who randomly inherited higher polygenic indices. And I think this is evidence in favour of this theoretical assumption of complementarity between endowments and investments. And it also means that your genetic predisposition cannot just give you a direct advantage. But it also means that this advantage may be kind of amplified by your parental or your teacher investments. And this complementarity, I think also suggests once again, that for disadvantaged children, so the other side of the coin, we need to start very, very early and follow up these early investments also with data investments to make them as productive as possible.
Christine Garrington 7:29
So Hans, some fascinating research and findings. I wonder if there’s been a standout or surprising finding for you from the project.
Hans van Kippersluis 7:36
I think methodologically, what we’ve learned is that there’s still a world to explore in terms of using genetic data in social science, because what we have seen is that polygenic indices can be a great tool to improve our understanding of the things we just talked about. But I think the way we use these polygenic indices, are shall I put this sort of a bit naive, in some sense, because what we do is we first construct a score or an index by regressing an outcome on all of these 1 million individual genetic variants. And as you can imagine, if you do these 1 million regressions, then it will be a lot of noise in these coefficients, and these estimates also come with some uncertainty. And what is surprising to me, what I’ve learned is that many researchers simply sort of seek to use this polygenic index as if it’s some kind of a transferable and deterministic index. And there’s hardly any account in the literature on the uncertainty in this index. And I think what we have done in one paper is actually showing how this uncertainty is sort of leading to different conclusions, because what we did is basically looking at the polygenic index for cardiovascular disease. And in cardiovascular disease, more and more people are using these polygenic indices, this genetic data for personalised decisions regarding, for example, the use of statins. And what we did was sort of constructing six different polygenic indices using different discovery sample using different methods of constructing this polygenic index. And what was fascinating and actually maybe astonishing to see is that only 6% of the individuals are in the top quintile of the polygenic indices, if you look across these six different ways of constructing the same polygenic index. And I think this is fascinating, because it shows that even though polygenic indices are now increasingly being used, apparently it matters a great deal about how you construct these things. And this is one thing we have shown, I think this is quite remarkable, and also an important methodological contribution.
Christine Garrington 9:19
A really important contribution to how this research might develop in the future. Right, absolutely. And then just finally, Hans, I wonder what this all of this work tells us about the interplay between genes in our environment, or, as we’ve talked about nature and nurture, not nature versus nurture, in better understanding and in tackling inequality.
Hans van Kippersluis 9:41
So it’s very hard, I think, to give sort of direct policy leads or implications, but there’s a few leads. One thing is that I think we need to start early. We knew already that inequalities arise early in life. And I think this focus on genetics gives us yet another clue that it’s very important to start early. And also because of the work I mentioned about complementarities, it’s very clear that later investments are more effective if the person has had already more investments early in their life. So that’s clearly one more general policy implication, I think. And I think our work is also showing how sort of genes and environment shaping jointly inequalities. And I think this has important implications for the discussions about equality of opportunity. I mean, if you look at politicians across the entire political spectrum, everybody seems to be agreeing that equality of opportunity is a great thing, and that your health and your income should not depend on your parental background. But let me ask two questions about this. One is, what about your genes? There’s hardly any discussion about whether inequalities that are deriving from genetic advantages or disadvantages are fair or not. And what we’ve also shown in this project is that parental background seems to reinforce genetic advantages. So even if you believe that parental background should not be leading to inequalities and your genes may, then how do you treat the interaction between the two? So I think we should have a clear discussion here a societal discussion about what is fair here. And I think that’s why our research is very important, because 30 years studies have already shown that people’s preferences for redistribution, for example, depends strongly on whether they perceive inequalities as fair or unfair. So I don’t think we are political activists here. But I do think that showing how genes and the environment jointly shape outcomes such as health, education, income, but really help people to make up their own mind as to what they regard as fair or unfair inequalities.
Christine Garrington 11:23
Hans thank you very much some some big advances here. But still some big questions to answer, I guess is the is the summary but fascinating work and thank you for taking time to share it with us. So finally, thanks to Hans van Kippersluis for discussing the findings and implications of DIAL dial project Gene Environment Interplay in the Generation of Health and Education Inequalities. You can find out more about this and other dial research on the website at www.dynamicsofinequality.org. We hope you enjoyed this episode, which is produced and presented by me Chris Garrington of Research Podcasts. And don’t forget to subscribe wherever you find your podcasts to access earlier and forthcoming episodes.