We (myself along with Nina Masters, Ramya Naraharisetti, Merlin Chowkwanyun, Sanyu Mojola and Ryan Malosh) just put out a mildly polemical paper in PLOS Computational Biology titled “There are no equal opportunity infectors: Epidemiological modelers must rethink our approach to inequality in infection risk”. The paper came out of a lot of discussion between all of the authors about why the broad majority of infectious disease transmission models have not typically treated equity – the distribution of who gets infected as a function of wealth, race/ethnicity, gender, and on – as a first-class concern alongside population-level patterns of incidence and mortality.
The reasons for this disconnect are likely at least partially historical: Transmission modeling arose out of theoretical ecology and population biology, and for a long time evolved on its own track somewhat independent of key subfields of epidemiology, and social epidemiology in particular. In the realm of HIV and STI transmission modeling, where the mapping between social and sexual networks and infection risk is more widely appreciated, inequality has long been a prominent concern. And for chronic infections like Tuberculosis, the long-established link between poverty, social disadvantage and disease has been impossible to ignore and is also reflected in the models used to look at TB transmission.
For acute infections – like SARS-CoV-2, but also influenza, RSV, and others – a tacit assumption that highly-infectious respiratory infections are essentially “equal opportunity infectors” that don’t care about lines of class, race, income and other distinctions, may have biased the modeling toolkit away from explicitly including the mechanisms that generate inequity in infection outcomes in the models that we use to represent transmission.
So, this paper is our attempt to close the gap a bit and to begin the process of building a bridge between the mechanistic inferences and whiz-bang tools of infectious disease transmission modeling and the critical insights into inequity and hierarchical approach of social epidemiology. We are not the first to highlight this gap and make suggestions about how it should be closed: Dolores Acevedo-Garcia’s excellent paper “Residential segregation and the epidemiology of infectious diseases” is a mainstay of basically all of my teaching and cited in almost all of my papers for a reason (If you want to play around with a model that puts some of the mechanisms in this paper into action, check out this Shiny app I put together awhile back). But our hope is to plot an actionable agenda for how to tackle these difficult problems going forward.
If you want to hear more but don’t want to read the paper for some reason – or if you just can’t get enough – check out this video of me giving a talk about this idea at the MIDAS Conference to Increase Diversity in Mathematical Modeling last month:
The problem is lack of data to constrain all these new parameters, ie underdetermination.
https://en.m.wikipedia.org/wiki/Underdetermination
A simple model with a few constrained parameters has a chance of being useful, but a complex one with too many will fit anything so it has no chance.
Also, I thought SIR models were developed for measles in the early 1900s. I’d be interested in more detail on how they were first used in population ecology. Maybe its in the paper.
I tend to think of the Ronald Ross malaria models as being among the earliest, and these (and the measles ones) have a broader relationship to models used in social and ecological demography. As someone who ostensibly did a PhD in social demography, I was sort of shocked when I got to my first postdoc in an ecology dept. and realized that demography was more or less anything for transmission models that were meant to capture long-term patterns.
Re: overparameterization, I think there are two counterarguments: 1. It may not necessarily imply entirely re-parameterizing the models, but at the very least thinking more carefully about their interpretation, 2. If we’re always making simplifying assumptions, why can’t we have a set of models that foreground the social-structural variability while being less focused on the ones that are interpreted as being primarily biological in nature? Ultimately our argument is that the social inequality aspect of things should be part of the initial phase of model development, but we totally agree that it is subject to the same constraints as any other model mechanism.
It would be nice to be able to use a more compartmentalized model, but it really is just not feasible given the data available.
Most of these models don’t even try to capture the effect of testing rates, let alone the issues with the test. Eg, it seems accepted by now that Ct >= 27 is rarely associated with infectious virus, yet the cutoff being used originally was Ct = 40. And it isn’t clear how this threshold has varied over space and time either since each lab/manufacturer may be doing their own thing.
Then a positive test only indicates the presence of RNA, which could be from virus stuck in the mucus (so the person never actually gets infected) or debris from an infection weeks to months earlier. So there is some distribution we need to assume to get at the actual infections we care about.
These issues are very proximal to the observations (positive test), but already we are adding all sorts of parameters that need to be constrained somehow.
But yes, the individual’s biology and environment are also crucial and ideally we could include these factors. The example I use is a seed sitting on asphalt vs in fertile soil.
The trick is designing priors that let the model fit to whatever capacity the data supports. It’s what Andrew calls the “unfolding flower” model (though for some reason, it’s not in his statistical lexicon). For those that might be interested, there’s a general framework of “penalized complexity” that stacks models and lets them incorporate whatever amount of complexity the data supports. See, for example, Simpson et al.’s Penalising model component complexity: A principled, practical approach to constructing priors.
I believe that efforts along these lines are fundamentally misguided, and here are my reasons for thinking so:
1. Disparities, contra Ibram Kendi (whom I think is wrong on most things), are a lousy way to define “racism”. If you define racism in terms of disparities, you’re talking about outcomes. But outcomes are going to be different for different people for a gigantic variety of reason that completely dwarf your causal diagram. That is, there should be a huge variety of arrows going into SES that have nothing to do with “racism”. Here I’m defining “racism” as discrimination: treating one person fundamentally worse than another because of that person’s race (and I define “race” as ethnic group, because scientifically there is only one human race). How would you deal with all those extra arrows? Talk about omitted variable bias! I’m not saying racism (as discrimination) has *nothing* to do with disparities. I’m saying, as Thomas Sowell has already said, that disparities don’t imply discrimination. Disparities don’t imply discrimination any more than the rooster crowing at sunrise causes the sun to rise: they are the same post hoc ergo propter hoc fallacy. As an aside, there are many ways of fighting racism that are completely different from the Kendi/DiAngelo approach.
2. This article illustrates other issues with disparities-first approaches: https://amac.us/the-prescription-for-modern-medicine-is-not-racial/.
The diagram is definitely simplified and I don’t think does the argument in the paper justice on its own, where we made a lot of effort to avoid being entirely reductive about the complex relationships between racism, SES and unequal outcomes. But I think there is plenty of evidence to suggest that race/ethnic inequality in health outcomes can’t be reduced to SES inequities and have to be explainable at least in part by the many ways racism is baked (one might even say structured) into social/policy/healthcare systems. Our argument isn’t about disparities or equity being first anyway, but just the modest position of ‘forward’ alongside all the other things we tend to take for granted as being important when modeling these systems.
Well, I think you have to take a step back and first ask yourself some of these questions:
1. Are disparities in health outcomes an inherently bad thing in the first place? As far as I know, the only way to achieve the same health outcome in all cases is simply kill everyone off. People are different. I am an identical twin, and my twin’s health problems are quite different from mine. So rather than focus on reducing disparities (which is not possible to begin with, other than the “Final Solution” I mentioned above), which I regard as a waste of effort, it makes more sense to focus on improving individuals’ health outcomes as best we can.
2. You are simply assuming that racism is baked into the various systems you mentioned. That is highly debated, and should definitely not be taken for granted. Even if it’s true, the exact ways that happens are also debated. I would actually agree that there is systemic racism in America. But I don’t agree with many others on exactly how that looks. Here are my examples of systemic racism: a. Planned Parenthood explicitly targeting black neighborhoods for abortion clinics for the sole purpose of wiping out blacks. b. Affirmative Action policies, particularly quotas, at colleges, universities, and businesses. c. The government welfare system perpetuating single motherhood to a catastrophic extent (see the Success Sequence literature for solutions), particularly in the black community (70% of black babies are now born to unwed mothers). So what are the ways that you see systemic racism baked into social/policy/healthcare systems?
3. You wrote, “but just the modest position of ‘forward’ alongside all the other things we tend to take for granted as being important when modeling these systems.” I wasn’t able to parse that. What exactly do you mean, there?
I’m not going to be able to reply to your reply to this, because of the hierarchical limit on reply depth on this blog, but I’ll read it.
Adrian:
“Planned Parenthood explicitly targeting black neighborhoods for abortion clinics for the sole purpose of wiping out blacks”? That’s just nuts. You can oppose abortion, that’s fine, but it’s just silly to say that the purpose (let alone the sole purpose) of abortion organizations is to “wipe people out.” I recommend that you find a way to express your political views without attributing ridiculous motives to people who disagree with you.
Reply to Andrew:
I am basing my statement on Margaret Sanger’s comments herself, as well as things her successors have said. Planned Parenthood has been extraordinarily racist from the beginning. Perhaps the word “sole” is too extreme. You can also look at where Planned Parenthood has actually built their clinics.
Adrian,
At no time did Planned Parenthood “explicitly target black neighborhoods for abortion clinics for the sole purpose of wiping out blacks.” Not as a secondary purpose, either. Margaret Sanger did have some views that many consider abhorrent now (though they were mainstream at the time), notably her belief that the “mentally defective” should not have children, but her goal with blacks as with whites was to make sure that women had control of whether or not to have a child.
In 1966, Martin Luther King Jr was awarded the Planned Parenthood “Margaret Sanger Award.” Corretta Scott King accepted the award on her husband’s behalf, and read his speech. Among other things, King said:
“For the Negro, therefore, intelligent guides of family planning are a profoundly important ingredient in his quest for security and a decent life. There are mountainous obstacles still separating Negroes from a normal existence. Yet one element in stabilizing his life would be an understanding of and easy access to the means to develop a family related in size to his community environment and to the income potential he can command.
This is not to suggest that the Negro will solve all his problems through Planned Parenthood. His problems are far more complex, encompassing economic security, education, freedom from discrimination, decent housing and access to culture. Yet if family planning is sensible it can facilitate or at least not be an obstacle to the solution of the many profound problems that plague him.
The Negro constitutes half the poor of the nation. Like all poor, Negro and white, they have many unwanted children. This is a cruel evil they urgently need to control. There is scarcely anything more tragic in human life than a child who is not wanted. That which should be a blessing becomes a curse for parent and child. There is nothing inherent in the Negro mentality which creates this condition. Their poverty causes it. When Negroes have been able to ascend economically, statistics reveal they plan their families with even greater care than whites. ”
The claim that Sanger, or Planned Parenthood, wished to wipe out blacks is simply a slander. Presumably you didn’t know this. If you do some reading — easy to do online — you will find that you’re wrong about this.
My friend used to live across the street from a stripmall that consisted of, in order:
1) Liquor store
2) Hair salon
3) Bar
4) Planned Parenthood
It was like a physical manifestation of the anticipated progression of events.
Reply to Phil:
Quote from Margaret Sanger’s letter to Dr. Clarence Gamble (of Proctor and Gamble): “We don’t want the word to go out that we want to exterminate the Negro population.” Presumably you didn’t know about this quote. If you do some reading — easy to do online — you will find that you’re wrong about this. Are you denying that Sanger wrote that? It’s pretty well-attested, and hence not slander.
The murder of unborn persons (I am using those terms very deliberately) will not solve any problems, and they will create many new ones. Adoption is the solution, not abortion or single motherhood. There are very long waiting lists for people who are very eager to adopt unwanted babies of all races, but there are huge red tape obstacles in the way.
They did not want the word to go out because it was not true. That is to say, they were wary that people would take on the exact misconception you’re currently perpetuating.
Learn to read.
Something something redlining
Adrian –
> Planned Parenthood explicitly targeting black neighborhoods for abortion clinics
So you agree that in that example, it was racism that resulted in disparate outcomes.
I thought you were arguing against that. Perhaps you’re throwing partisan red meat in there for a some reason, but your arguments is self-contradictory.
I think it’s absurd to argue that you can disentangle racism, explicitly across healthcare outcomes across racial divides. You can’t just hand wave away the legacy of 400 years of explicit racism in this country.
> But outcomes are going to be different for different people for a gigantic variety of reason that completely dwarf your causal diagram. That is, there should be a huge variety of arrows going into SES that have nothing to do with “racism”.
Of course there other factors to explain disparities across race/ethnicity – such as the interaction with SES as a predictive variable. And of course there are behavioral components which, if one is interested in doing so, could make complicated just how you’d precisely tease out the impact of historical or current day racism. But I have to say, usually when I read arguments such as yours, they seem more to be a political grievance argument than an actual scientific interrogation.
While the calculus of determining just exactly how much “racism” per se explains outcome differences across racial lines is certainly complicated, it does seem to me that your interest in trying to find the answer seems suspect. The fact that you use Kendi as an introduction to an important epidemiological question, one that predates his popularity by many years, only adds to my suspicions.
Adrian,
Ah, OK, you’re just trolling. You surely know that (a) yes Sanger said “We don’t want the word to go out that we want to exterminate the Negro population” but you have added a period rather than provide the rest of the sentence. I’m sure you also know that the gist of that part of Sanger’s letter to Gamble was that some people were falsely claiming — just as you are — that Sanger’s movement was aimed at exterminating blacks, and that she was emphasizing the importance of stopping that from getting around. Also, you say “look at where Planned Parenthood puts their clinics”, implying that they are exclusively or mostly in black neighborhoods, but that isn’t true either.
So, OK, liar liar pants on fire. I know you will continue to say this stuff even though you know it’s false, but some readers might be less well-informed so I am posting this for their benefit. If they want a place to start, https://www.factcheck.org/2011/11/cains-false-attack-on-planned-parenthood/ and https://www.politifact.com/factchecks/2011/apr/08/herman-cain/cain-claims-planned-parenthood-founded-planned-gen/ are easy starting places for this particular claim.
Reply to “somebody”:
What is it with the condescending tone on this blog? As well as anonymous comments? I am of the opinion that if you’re going to comment on a blog, you generally should tell people who you are. I see the condescension all the way from Andrew, Phil, and many others. Apparently, if you disagree with someone, that gives you the right to belittle them, call them names, etc. We call that in logic (which is NOT, by the way, a tool of “white suppression”, contra what some say) the ad hominem fallacy: attacking the person instead of the idea. Far from advancing your viewpoint, it seriously detracts from it. The only correct way to attack an argument is to show that a term is unclear (meaning, its use in the argument is not understood the same way by everyone), a premise is false, or the argument as a whole is invalid.
Your interpretation of the quote is incorrect. Sanger did not want the word to go out, because she did not want people to know about her views. Her views were what they were. The many other quotes from her writing that are quite easy to find show that she believed in forced sterilization and eugenics, particularly for Negros:
“The main objects of the Population Congress would be to apply a stern and rigid policy of sterilization and segregation to that grade of population whose progeny is tainted, or whose inheritance is such that objectionable traits may be transmitted to offspring[;] to give certain dysgenic groups in our population their choice of segregation or sterilization,…”
Who in their right mind could possibly be in favor of that? She had a contempt for human life in general, but what she regarded as “inferior races” in particular. In context, your interpretation of that quote does not fit her worldview.
Naturally, if the only newspapers you read are CNN and MSNBC, and if you think that Snopes actually fact-checks, and if you think Wikipedia is actually reliable, then your fears are going to be laid to rest. But whether your views correspond to reality is another matter entirely.
Adrian:
Nowhere did I belittle you or call you names, nor did I try to attack your argument, nor did I refer to white suppression. If you want to pick fights with people or you want to get into an argument about abortion, I suggest you go to twitter or some other such forums such as the Marginal Revolution comments page where people from the left and the right go at each other. This just isn’t the place for it. There’s a whole wide world out there and this blog is just one place. I’m not denying the importance of political argument—abortion is important!—it’s just that the comment section of this blog isn’t the best place for it, and this certainly isn’t the right place for you to criticize me for things that I haven’t done.
Jon’s post above is on social structure and the transmission of epidemics. Your take on abortion being murder, etc., is off topic and does not add anything to the discussion. But your original comment to the post did have some on-target things to say.
ACTUALLY, none of the behavior you described falls under the ad hominem fallacy. As you state the ad hominem fallacy is when you attack the person INSTEAD of the idea. It’s not fallacious to just call names and belittle someone, in addition to stating an actual counterargument.
Margaret Sanger was a eugenicist, at least a little racist, and certifiable very bad. That does not imply that planned parenthood was “explicitly targeting black neighborhoods for abortion clinics for the sole purpose of wiping out blacks”, or affect my interpretation of her quote.
Please do not quote from thin air. The quote “inferior races” does not appear in the text you supplied. You only put it in quotes to give the misleading impression that it’s a quotation from her, rather than an interpretation from you. Nor does the text you supplied refer to black people.
Again, I’m not commenting on whether or not planned parenthood is bad, or if abortion is bad, or if Margaret Sanger was a racist. I am commenting on the specific claim that it was an “explicit” goal of planned parenthood to “wipe out blacks” (See what I did there?). If it was so explicit, and you’re basing your opinion on Sanger’s writings, then you can provide somewhere where Margaret Sanger wrote that one of her goals with planned parenthood was to to exterminate black people.
I’m happy to end the thread here
Reply to Andrew:
I’m not saying you have personally belittled *me* or called *me* names. Phil certainly did the former in this thread, as did “somebody”, whoever that is. You have belittled other people, both in posts and in comments. Sentences such as ” ‘Clearly,’ huh? What an idiot. Jeez.” in your recent post on “Fools at the Hoover Institute” will come across as very condescending and belittling. I certainly wouldn’t want to be called that, nor do I see that anyone, really, deserves to be called that. And I have seen other similar comments belittling people and not just their ideas. Attacking my arguments and ideas, though, is totally fair game. I don’t regard attacking what I say or think (if done as I have outlined: questioning whether the terms are clear, the premises true, or the argument valid) as attacking myself in the slightest. That’s just good debate.
I talked about “white suppression” as a prophylactic, anticipating possible objections. It is an argument I have heard from the Far Left that logic, statistics, and even mathematics are tools of white supremacy and white suppression, a view I regard officially as “baloney.”
I don’t regard myself as a fight-picker. I don’t go looking for them: it’s not good for mental health to do that. There are some fights that come my way that I simply sidestep (most fights, actually), but others that I don’t. We need some people around willing to stand up and fight, or there would be no dialogue. In this thread, I merely mentioned Planned Parenthood in passing as an example of systemic racism. It was others: Phil, and “somebody”, who replied only to what I said about abortion and ignored the main thrust of my initial comment.
I value my mental health too much to get on Twitter – a medium that highly discourages thoughtful reflection. Certainly willing to check out Marginal Revolution, thanks for the HT.
Reply to somebody:
I’m happy to end the thread here as well. I disagree with basically everything you wrote, and I don’t see that either of us is going to budge much from our positions.
Reply to Joshua:
I was arguing that disparities do not imply discrimination. That is, that the presence of disparities does not imply that discrimination is what produced those disparities. I said nothing about whether discrimination can produce disparities (the other arrow direction for causality). Of course discrimination can produce disparities.
I don’t claim at all that racism doesn’t exist. In fact, I’ve already said quite the opposite in my threads here. Nor am I claiming that racism doesn’t produce disparities (see above).
I try to be careful with my words – not as careful as I should be, no doubt. But it is certainly not a legitimate inference to say that if I say A implies B, that therefore I was implying B implies A.
Not sure what you’re getting at with implying that my statements seem political at times. Of course they are. Often. This whole blog is highly political. That doesn’t inherently or necessarily make my statements unscientific or untrue, though. And I cheerfully admit that I am a conservative commenting on this blog – probably a nearly unheard-of phenomenon given the distinct lack of viewpoint diversity in colleges and universities – certainly Ivy Leagues such as Columbia.
Yes, you are certainly political. Just speaking for myself, I don’t appreciate your need to harp on your need to express your conservative views – and use the “distinct lack of viewpoint diversity” as a defense. I assure you I am horrified by political correctness and the movements it has spawned. But I also can appreciate the history that has led to that. On this blog, I am looking for comments free of such debates as much as possible. While you may feel that “this whole blog is highly political” I find it less so that most other places. Yes, you should check out Marginal Revolution – you will fit in perfectly there.
Why do I bother commenting on this? Your comment on Planned Parenthood used the horrible practices/intent of Sanger to tarnish the entire organization as trying to “wipe out blacks.” That same logic would condemn the entire US Constitution – the practices of our forefathers (fathers, of course) would then condemn all that follow the Constitution today.
Your point about discrimination vs disparities is a good one – and one that I believe most readers of this blog fully understand. The real modeling issues have to do with how to define and identify discrimination. There is a lot of work in the predictive modeling literature to try to identify and protect against algorithmic discrimination – and it is a particularly thorny problem. Most of the readers of this blog do not simply equate disparity with discrimination. You are correct to not equate them – but it is not helpful to pretend that readers of this blog don’t see the difference. Try contributing ideas about how to tell these apart rather than using this as a platform for a tirade against what you perceive as a lack of diverse views on this blog.
When I read the original post that you reacted to, I see it as an honest attempt to include inequities in epidemiological modeling. It is you who has interpreted this as equating inequities with discrimination. I believe the authors are saying that discrimination exists – and I agree with that. I believe they are saying that past modeling has not taken that into account – and I agree with that. I do not see them saying that if health outcomes are not proportional to the populations that this is proof of discrimination. I think you have contributed that interpretation on your own.
Adrian –
Dale’s response should suffice. But as I do somewhat differ in perspective from him:
> I was arguing that disparities do not imply discrimination.
But my question is why you’re arguing that. As Dale discusses, I don’t think anyone here assumes that all health outcome disparities are a direct function of racism. So it seems that you’re arguing against a strawman for ideological purposes. You seem to be fixated on a favored rightwing narrative, that libz all think that all discrimination is explained by racism. I suppose there might be some people who hold that view, but I’m quite sure it’s not a view held by the author of this post or anyone who studies these issues in depth. It’s like you want to throw red meat out there so you can get a response that then circles back to feeling that your misapprehension is justified.
> I don’t claim at all that racism doesn’t exist.
A strange statement – as I never suggested that you claimed that racism doesn’t exist.
> Not sure what you’re getting at with implying that my statements seem political at times. Of course they are.
I wasn’t merely pointing out that your comment seemed politically motivated. I was pointing out that you were substituting politics for a meaningful discussion of to what degree healthcare disparities are a direct or indirect function or racism. Again, introducing Kendi, whose view aren’t in any way meaningfully relevant to interrogating those questions, seems to me to be evidence of that (IMO misguided) substitution.
> Often. This whole blog is highly political. That doesn’t inherently or necessarily make my statements unscientific or untrue, though.
No, what makes your statements unscientific (I’m not in a position to determine their “truth”) is that you introduced topics like Sanger and Kendi into the discussion regarding the causal factors in play to create healthcare disparities.
> And I cheerfully admit that I am a conservative commenting on this blog – probably a nearly unheard-of phenomenon given the distinct lack of viewpoint diversity in colleges and universities – certainly Ivy Leagues such as Columbia
In my experience there are quite a few conservatives who comment here. But I have to wonder if you won’t exploit topics such as the causes of healthcare disparities to gish gallop your way into an endless list of grievance arguments. What’s next, the “war on Christmas”?
When it comes to abortion, the proper wording should be “abortion rights.” No one is in favor of abortion. The issue is: abortion rights vs. anti abortion rights. Unfortunately the forces of repression and reaction have co-opted the terminology. And, when it comes to eugenics, the very founders of the movement were Galton, Fisher and Karl Pearson. This is hardly a secret but statisticians would prefer not bringing that topic up.
Isn’t there a significant arrow into the “Death” box, from well to-do old white ppl in fancy nursing homes?
There’s only one arrow into the “death” box and it comes from a box called “life”.
All this mention of Margaret Sanger reminds me of an incident over 20 years ago at the University of St. Thomas. A student discovered a 1987 “Great Minds Meet at the Library” poster of her in “The Smoker’s Lounge [!!] in the basement of the library.” From the student it went to the local bishop and then to the head librarian by way of the University Administration. Librarians, unlike other administrators, are loath to engage in literal or figurative book burning, and she refused to remove the offending poster. News coverage included the local newspapers and the “Chronicle of Higher Education.”
While discussions were going on regarding academic freedom, the role of the Catholic Church, Sanger’s history and motivations, and so on, someone stole that poster as well as the poster about Malcolm X.
Some of the story can be found at
https://marii.hosting.nyu.edu/sanger/articles/sanger_censorship_and_catholic_church/
“The most recent attempt by the Church to quiet Sanger’s voice was no more successful than earlier ones. The President of the University of St. Thomas supported Library Director Jean Haley’s decision to leave the poster of Sanger in its place. Haley commented, “The poster is part of the Library’s collection and is protected by the Library Bill of Rights…It deserves the same protection as the rest of the collection.” (St. Paul Pioneer Press, October 7, 1993)”
At the time, my contribution to the discussion on campus regarding academic freedom and all that sort of stuff was to point out that to be consistent, the library ought to remove all statistics books because eugenics was founded by the statisticians Galton, Fisher and Karl Pearson.
look you’ve got to grasp that they’re not here to talk statistics (a tell is that they’re not, in fact, talking statistics or anything related to the topic) but rather because it’s an insufficiently moderated space where they can post Nazi content. you know – just get it out there, get people talking, get them repeating-to-rebut. you’ll get the trouble all the time right up until you finally clean out the comments.