Michael Pearce is a PhD candidate in Statistics at the University of Washington, working under the supervision of Elena A. Erosheva. His primary research interests include preference learning and developing Bayesian statistical models for social science problems. In his spare time, Michael enjoys running, biking, and paddling around the Puget Sound.
Check out the full article in Significance
Episode Description
When American comedian and actor Betty White died, fans lamented the fact that she had just missed making it to her 100th birthday. They felt she’d been robbed of achieving a significant life moment. Some researchers think that this century could see more people making it to that moment and beyond. That’s the focus of this episode of Stats and Stories with guest Michael Pearce.
+Full Transcript
Rosemary Pennington
When American comedian and actor Betty White died, fans lamented the fact that she had just missed making it to her 100th birthday. They felt she had been robbed of achieving a significant life moment. Some researchers think that this century could see more people making it to that moment and perhaps beyond. That's the focus of this episode of Stats and Stories where we explore the statistics behind the stories and the stories behind the statistics. I'm Rosemary Pennington. Stats and Stories is a production of Miami University's departments of statistics and media, journalism and film, as well as the American Statistical Association. Joining me is regular panelist John Bailer, Chair of Miami statistics department. Our guest today is Michael Pearce, a fourth year PhD student in statistics at the University of Washington. His primary research interests include nonparametric theory and Bayesian models applied to the social sciences, he spent two summers as an Applied Statistics intern with the applied math research group at polling research and technology. Pearce is also the co-author of a Significance article exploring whether we could see a record number of people breaking the century mark in the coming century. So Michael, thank you so much for joining us today.
Michael Pearce
Thank you for having me.
Rosemary Pennington
I guess before we get started, if you could just talk about the study that you were writing about for significance, and sort of what you guys did, and maybe why you felt compelled to study this?
Michael Pearce
Sure, happy to. So the key idea is that people really do care about how long humans can live. This is a big question that's really central to a lot of our identity as a species. And a lot of people in the past have studied this problem, and they've studied it from many different lenses. Some of them are biologists who study various biological qualities that we may or may not have, that may indicate how long we can live, if there's a cap to the human lifespan, that sort of thing. And on the other side, there's a lot of demographers who've studied this problem. This relates to broader questions over population sizes, fertility rates, those sorts of things, really important questions to the study of demography. But when it comes to studying the maximum human lifespan, there's also a lot of problems and challenges with this sort of analysis. So one of them is that data in this context is very hard to come by high quality data, because there's all these tendencies of people who are extremely old to misrepresent their age, so it's very difficult to like to have a proper study of this sort of thing. There's also a lot of people who would like to make bold claims over the maximum human lifespan or the lack of it. And it's really hard to test hypotheses about these sorts of things. So we said, why can't we just take a statistical approach, just like the way a lot of demography, modern demography is working, and study the overall patterns of people who are living to these extreme ages and see if we can learn something about the world's increasing population, in addition to the patterns of people who live over age 100 times and see if we can combine those things, to have probabilistic forecasts of how old someone practically might live this century? So not answer the question of if there's a maximum or not, but just say, how likely is it that if these trends continue, we'll see someone to live to various extreme ages like 125, or 130.
John Bailer
I liked the Significance feasts a lot. And, you know, this issue of some of the ingredients here, I mean, one is this, this first key and component, as you noted, there was that kind of this flattening of mortality after age 110. I thought that was kind of, you know, all of a sudden, I'm thinking of the hazard function, you want to be drawing in a failure time analysis class and a survival time class. But also the idea that, that the pool of people that is sort of feeding into this probability calculation is growing. And then before we know, we'll dive into some other questions related to the types of data and where you sort of found reliable data for this. Can you talk about those two key ingredients to this calculation?
Michael Pearce
So the first ingredient is the increasing population? I think this is uncontroversial, though. The world's population has been growing pretty significantly over the past century, and it continues to grow. And that's actually a major component of this sort of analysis. It's maybe the biggest component here. So if we're talking about people to live beyond age 110, you just need people to actually reach that number. And you know, as the world population grows, I mean, if you think about the people who are age 110, now, you know, they were born in the early 20th century, the world population was much smaller back then. And over the coming several decades, just a lot more people were born and this is because of population growth and so much of the developing and developed world. So that's a big aspect here. We just have more chances to see someone, you know, break these barriers just because there's so many more people. The second component is this idea of the flattening mortality after age 110. So the exact reasons for this are unclear. And I'm certainly not the person to talk about this. But as we do as are statistical demographers we can study the patterns that we're seeing. So even though I can't answer why, I can answer that it is happening or it's not happening. And we do see this strong pattern. So people who reach age 110, we don't see strong differences in their mortality based on their their sex, which is surprising or where they live or where they grow up. I think that's also very surprising. But these are patterns that we're seeing across the[ documented cases of these super centenarians. And then the other thing that we see about them is that they tend to die at a 50% year over year rate. So if you reach age 110, you have about a coin toss, chance of living to age 111. And this sounds very, very bleak. But this is the overall pattern that we're observing, and it tends to stay flat, which is perhaps the most surprising aspect of this. So exactly why that's happening. I'm not sure if it could be that these people have excellent genetics, or that they're doing things in their life that promote their longevity. But this is the overall pattern, we're seeing off the 1000s of documented cases of the Supercentenarians.
Rosemary Pennington
Could you talk a little bit about where you found the information you looked at in your study, because you, you know, obviously you said you can't talk about like the why they're slipping so long, we can talk about where you found the data for the population to look at, to give you this information.
Michael Pearce
So the data comes from the International database on longevity. And this database has been a big effort of many different institutions to try and go through all these documented cases of people who have reached age 110. And then actually go through and make sure that there's a whole variety of supporting evidence to document their claim. So some challenges that exist in verifying someone's age are the fact that, you know, if they were born in a country that did not have excellent record keeping when they were born, I mean, remember, we're talking about people who were born, you know, at the end of 1800s, or the early 1900s, mostly, it's not too surprising to think that many of these countries would not have excellent records, or those records would still exist. So a lot of the effort of this database was to go through and try and find records as best they could, and see whether or not they met a sort of standard criteria. And it wasn't just finding a birth record, but also finding records throughout that person's life to verify that nothing funny had happened, there had been no name switching, something like that going on. So the remaining list of people, or a group of people who have really, really highly verified birth and death records.
John Bailer
And you know, but one thing that was really striking to me is in looking at kind of the countries that were included in that, that final dataset, is that you have kind of US and Canada plus European countries plus Japan, you know, so you so there's a, you know, there, there's probably the the potential for the future, we're getting a lot more information. But but you know, South America, Africa, you know, India, China, you know, this, this, this probability, I mean, so maybe I'm thinking that perhaps it's an underestimate, you know, for the calculation, given the, you know, when I think about kind of the size of the countries that are have been omitted, or the size of continents that have been omitted, that, you know, the calculations, I it makes a heck of a lot of sense what you all did in terms of being able to verify sort of the the information as part of the modeling process. But it also makes me think a little bit about what, in the future, how much more input into this, these probability calculations, there could be?
Michael Pearce
Yeah, that's absolutely true. When we conducted this study, we wanted to be as perhaps conservative as possible, given the fact that in this field, there are so many people who make really bold claims. So when we did our analysis, we wanted to use really highly verified data. And we also wanted to only use population projections from the countries that we had had this data for of actual Supercentenarians. Because we're not entirely sure that these countries are like other countries, or if this is a thing that is unique to those countries. So because of that the population forecasts we're using are certainly not the population forecasts of the entire globe over the coming century. They're they're actually quite limited. And in that way, our production may be very conservative.
Rosemary Pennington
Michael, you called yourself a statistical demographer at one point early in this conversation, and I wonder if you could just sort of explain for people who have not heard that term, what that means and sort of why you got interested in this particular area of statistics.
Michael Pearce
Sure. So demography is the study of various population trends and statistical demographers go beyond more standard rules, thinking about birth and death counts, that sort of thing, and try to use statistical modeling to understand broader, broader problems or broader patterns under a statistical lens that includes uncertainty. So the key thing here is that you're trying to not only have a good model that represents what's going on in the world, but also understanding the uncertainty behind it. I got into this field because of a professor in my department at the University of Washington. His name is Adrian Raftery. And he's one of a few really excellent statistical demographers in our department here. But he taught a class on statistical demography. And I took it as an elective for my PhD program. I didn't really know what to expect. But I knew he was an absolute expert at it, and I joined the class. And that's how I got into this.
John Bailer
Oh, that's very cool. So you've talked a little bit about the kind of the data that you were using as kind of this empirical base on which to do these projections? And [and, you know, the, the, as many of the articles noted, the one that cited your you know, the work that you, you did that there was a Bayesian strategy here. So there was some, there was some prior information as well, that perhaps you were using. So can you talk a little bit about, you know, how both what, what was believed about the systems, and then what data were available, was combined in terms of making some of these projections.
Michael Pearce
So in demography, it's becoming more and more common to use a Bayesian method, because it turns out that the demographic patterns are pretty constant, or at least slow changing over time. So if you have data on people's, you know, birth and death rates over the course of the past century, like that's a pretty good indication about what's going to happen over the coming century. And oftentimes, you can pull information from one country or another to learn about the demographic patterns of another country, maybe that you don't have a lot of information on. So that's the type of thing that people do in Bayesian statistical demography. And that's exactly what we did here. So in our model, we have this general idea over people's death rates when they are over age 110. And we use the information from these very highly verified supercentenarians over the past, I guess, 100 years, and use that as our prior information. In the Bayesian analysis.
Rosemary Pennington
You're listening to Stats and Stories. And today we're talking with the University of Washington's Michael Pierce. So the headline for the Significance article is something like could this century be a record breaker for oh, yes, let me Will this be a record breaking century for human lung longevity? John is holding up the article so I can see it for those of you who cannot see. So I wonder, is this going to be a record breaking century? What did your study find?
Michael Pearce
So I feel pretty confident that this century will be a record breaking century for longevity? I can't say exactly by how much, but I do feel as if someone will break the current record for maximum human lifespan this century.
John Bailer
So what does that mean? You gotta you gotta you know we're on the edge of our seats, Michael, you got to let us know, what's the current record? And what's the chance of breaking it?
Michael Pearce
So the current record was actually set about 25 years ago by John Carlmont of French of excuse me, John Carlmont of France. She was 122 years old, and 64 days old when she died. Currently, the record holder is Connie Tanaka of Japan, she is 119 years old currently. So I'm not exactly sure if Miss Connie Tanaka will break the record, she certainly could. But the key thing is that we are expecting so many more people to reach age 110, this coming century that we think that this record will continue to be pushed out. And it's quite likely that someone could reach the age of 125 or 126, or perhaps more this century.
John Bailer
So every person that reaches this, this 100 and 110, Mark starts flipping a coin.
Michael Pearce
Yeah, more or less.
John Bailer
Do I get to 111? Yep. Heads? Or do I get to 112? Yep, head. So if you can flip 10 heads in a row, you can get to 120?
Michael Pearce
Yeah, I know. It's funny. It sounds like an almost too neat pattern. But when you look at the data of these people, it's striking how close it is to that pattern.
Rosemary Pennington
So you got a fair amount of coverage in the news media about this study. And I wonder if you could just talk about maybe specifics, if you could just sort of talk about what your overall impressions were of the coverage of your work?
Michael Pearce
Well, so like, like most things related to statistics, and I can't blame people for this. There's certainly a lot of misunderstanding about how statistics works, and how studies like this work. So there, there was a little bit of coverage that did not really accurately represent what we were trying to say. So, you know, we didn't want to make claims that someone would surely live to certain ages, all of this has a caveat of, you know, under the assumptions of this model under the data, we have that sort of thing going on. And I think that because of those assumptions, or the fact that we limited ourselves to certain countries, you know, our claims can be too conservative or perhaps you know, not conservative enough. You know, there's all of that and that wasn't necessarily talked about in the news media that's not quite as flashy. The other thing is that people tend to focus on, you know, how can I live to be 120 years old? The focus on the individual and And this sort of work is not about individuals. It's about a broader population pattern. So I can't tell you anything about how you could live to be 120 years old. And I can't do anything myself either. I don't know the answer to that question. But I do believe that we've seen so many people reach age 100 at a time that this pattern is really becoming clear. And because of that, I think people will live to these extreme ages.
John Bailer
I think you got at least a partial answer. I mean, the the partial answers live to 190. And then you got a half a chance. And then repeat. That's inductive reasoning. Yeah, good. Yeah. I thought that, you know, your description of some of the headlines really resonates. Because some are almost breathless. You know, as I, as I looked at these, you know, it's just like, lifespans of 130 years could be possible this century. Well, that's not what you really sad. There were others that seemed to some of the headlines really just sort of sounded a little more true to me, you know, like the idea that researchers say the probability of living past 110 is on the rise. But then, here's what you can do to get there. So it was kind of like, there was a fair rendering of the conclusion, or at least a closer rendering than what I saw in some of the others. But then it was really kind of pushing, you know, kind of this this behavior role, this lifestyle choice, this seven little bit of the biology and biological understanding that was coming to play. So I thought that was kind of a, it was, it was interesting to see how your story was leveraged, you know, and multiple outlets, I mean, because it really captured imagination. I mean, it captured our imagination. I mean, heck, we wanted to talk to you, man. I mean, it was a it was a I mean, that was it was a lot of fun. But But I just was really, really found that fascinating. And I, I know that Rosemary could probably talk a little about the experience of maybe an editor, writing a headline that a journalist was, was capturing,
Rosemary Pennington
Ya know, and I wonder to sort of if, like, headlines are always designed, right to catch your attention and to be flashy. And I don't know how closely you looked at the reporting. But I wonder, John and I have just finished writing a book on stats and news. And one of the interesting things we found was that sometimes headlines can be very flashy. And then the star story sometimes was pretty solid. And I wonder if, when in the reporting, if you felt like, Were there moments where like the headlines were not doing the story justice, or maybe the reporters were closer to what you guys actually got than what was being sort of talked about?
Michael Pearce
Sure. So in some cases, I did not look too deeply into the details, because I just knew that the article was about something that I wouldn't be too happy about. But I will say that a number of people could reach out and write really informed articles that discuss these issues. And in more nuance, of course, they might still be writing stories that do try to get at, you know how people can live longer, healthier lives. And that's absolutely fine. But they weren't trying to make claims over this is how you live to age 120. And that's really the key. So I'll give credit. There were some some nice major news articles in The Washington Post, as well as in the University of Washington's News Service, that describe it in great detail, sort of the implications for society over this work, which is a really key aspect of this, as well as the fact that you know, this is a model, and that we make these projections about how old someone could live, that doesn't necessarily apply to individuals.
Rosemary Pennington
It's really interesting. You raised that point, because it did feel like a lot of the news coverage was very much like, again, as John pointed out, like here, here's how you could do to live a long life. And it feels like there are a lot of implications for us collectively, right? If people are living longer, even if they're not making it to 110 Just the fact that human beings seem to be living longer. It does seem like there is little stuff that we need to talk about that perhaps the news coverage didn't quite get in the way it could have.
Michael Pearce
Yeah, absolutely true. So you know, we're certainly still still seeing the world population increase. But we're, we are also going to see these extremely old people, their numbers as a population group to increase over the coming century. So you know, if you live to age 100, or age 110, you know, of course, like for the individual, that matters quite a bit. But as a population group, I mean, not only are we going to see a lot more people living to 110, and we're also going to see a lot more people living to age 100. And that's going to really significantly change this distribution of age in our society.
John Bailer
Yeah, I mean, I thought that was an interesting point. Because a lot of times when you see these types of stories, it's often framed in terms of the average life expectancy, you know, so it's in the talk about what's going to happen with the extreme. What's the chance that this, this large, this, this large, extreme value is actually going to be bigger than previously observed large extreme values? And that's, that's, that's a subtler story about distribution that I think is often covered. And I think that's a real challenge.
Michael Pearce
Yeah, but I'm really happy you brought that up, because this is actually something that we discussed quite a bit when we were waiting for my co author, Adrian Raftery, where we're discussing how to frame this paper. So many of the researchers who studied this problem before have actually tied the relationship of flattening average life expectancy over the past several decades to the fact that someone is not going to be able to live beyond a certain age, they tried to create a relationship there. And something that we tried to say was, you know, the mean is not the maximum. I mean, these are if you said, if you say it that way, that that sounds like an absurd claim to try and equate them together. And they're certainly not. And when you're talking about a population that's as large as ours, I mean, the mean of a group of 8 billion people or more, is definitely different from the maximum that that group could attain. So that's the type of thing that we're seeing here. And we really tried to couple those ideas when when writing this this research paper,
John Bailer
You know, one of the things that I was thinking about, as I, as I read this was, I was thinking about COVID. And, you know, not because it just doesn't seem like I could wake up any day and not think about it. But you know, but in particular, thinking about the kind of impact that has been observed in mortality. And if I think that I would, what I saw was that, if you looked at the total number of deaths in the United States, it was something like one in 400 US citizens or residents of the US, but among age, ADM, older was more like one and 100. And I'm just curious to think about if, you know, and your reaction to this, do you expect that maybe some of the impact, particularly of COVID, among the oldest individuals who are at more risk of death with this disease? Is this going to impact some of your probability calculations if you were to redo them?
Michael Pearce
So it's certainly tough to say, but at first glance, I would say probably not too much. And the reason is that when we talk about the maximum recording age of death, what we expect to happen by the end of this century, it really relies on order of magnitude changes, and the population of people reaching age 110. So one of the 100 people over the age of 80, roughly, dying of COVID is an absolutely devastating burden on our society. Absolutely. But it's not the type of thing that's going to impact this calculation, specifically, what we're really requiring to happen is that you have, like 10 100 times more individuals reaching age 110, for us to actually see these sorts of noticeable increases in the maximum lifespan,
Rosemary Pennington
You know, there's so much news coverage of this particular issue, even away from your own research, right? There's a lot of stuff about things like, you know, eating a diet inspired by Japanese culture to live to be 100. I've seen that article so many times. Right. Yeah. And so I wonder if, you know, as we're wrapping up, if you given sort of what you research what you've seen, in your experience now with media coverage, if you could sort of provide some insight into what people who are, you know, reading news or seeing news stories about this issue of longevity, right in the human lifespan, what things they should keep their eye out for to make sure what they're reading is sound and is not just sort of boosterism garbage?
John Bailer
Sure, how do you really go back Rosemary,
Rosemary Pennington
I am a journalist, I feel empowered to say that boosters?
Michael Pearce
Well, so when you're reading news coverage about anything related to your longevity, or your health, remember that it's really important to think about, you know, how was, how did this recommendation come about? Was it through a scientific study that had a large enough sample size that actually could be generalizable to you? Or was it only with a very small group of individuals who look absolutely nothing like you. So that sort of thing is really important to remember. That's more general health advice. And then when it comes to studying your longevity, it's very unclear exactly the reason why people live to ages as extreme as this, you know, if you die because of a specific disease, or because of a car accident, and that's the type of thing that one could prevent, one could hopefully, you know, cure disease, that sort of thing. But when it comes to living to this extreme of an age, it's not exactly clear how these people are surviving or dying. And there's so many untestable assumptions there. So because of that, I would say, you know, if you're trying to have a longer lifespan, just just remember that there is quite a bit of uncertainty there, it might be something that is actually totally beyond your control, that allows people or you as an individual to to live to that extreme and age. And we just don't know at this point.
John Bailer
So I'm curious what would be next, the next study you would like to do that might relate to this, if there was a sort of a follow up or a spin off on this, you know, what, what kind of exploration or is of interest?
Michael Pearce
Well I think that as we get more and more information on the Supercentenarians it's these people who live to age 110 I would really like to do a deeper exploration. And of whether or not there are differences in individuals based on where they grew up, or something like their sex, or other qualities that that we can really study. Currently, you know, we have a limited amount of data. So based on the data that we do have, we don't really see a difference there. But I do think it'd be interesting as we have more and more data on this, you know, as people reaching age 110, or coming from countries that were able to maintain really accurate records that we can verify their claims. I'd like to study that more. I also think that it'd be interesting to see how this pattern of this font mortality may or may not be changing, or may extend to younger ages, that there is some evidence right now that maybe age 110 is not the exact right start of this pattern. Perhaps it could be people who are aged 100, an age or age 105. And I'd like to see how that's happening and how that might impact these projections.
Rosemary Pennington
Well, that's all the time we have for this episode of Stats and Stories. Michael, thank you so much for joining us today. Stats and Stories is a partnership between Miami University’s Departments of Statistics, and Media, Journalism and Film, and the American Statistical Association. You can follow us on Twitter, Apple podcasts, or other places you can find podcasts. If you’d like to share your thoughts on the program send your email to statsandstories@miamioh.edu or check us out at statsandstories.net, and be sure to listen for future editions of Stats and Stories, where we discuss the statistics behind the stories and the stories behind the statistics.