Michael Spagat
Royal Holloway University of London economics professor mainly studying war - including measurement and memory of war deaths, (possible) decline of war and fabrication in survey data from war zones
Location Living near London, originally from the Chicago area
Activity
-
Hi. Yes, I think that you can still find excess death estimates, e.g, in The Economist but for sure they are getting much less play than before and you may have put your finger on at least one of the reasons for this.
-
Michael Spagat replied to Jennifer Boag
@BobWagg Hi. Yes, you are definitely right. We will address some of the issues you raise later in the course.
-
Michael Spagat replied to Jennifer Boag
That's all great to hear!
-
Wow
-
The problem is that it requires very strong assumptions to be satisfied in order to work well or at least so that we know for sure that it will work well.
-
Michael Spagat replied to Anthony Ward
Certainly true but it's worth noting that a lot of people do want to talk about this and think it's important that someone hears them.
-
Yes, these are indeed two of the biggest problems.
-
Yes, there are efforts to record, or at least estimate, rape in war although, as you can imagine, this is harder to generate information on than war.
-
@JenniferBoag You had a busy day yesterday prowling around our discussion forum!
Yes, thanks for bringing this up and, sadly, I'm afraid we won't return to it. But, yes, re-interviews are a really good way of checking quality and making adjustment in high quality (well funded) surveys. I'm now reminded an episode from way back in the mid 1990's the...
-
@JenniferBoag Interesting! Yes, there is the theory of how a survey or census can work really well under certain conditions and then there are all sorts of unpredictable complexities that arise when you really get your hands dirty. As I've said elsewhere, I'm not advocating just throwing your hands up and saying that the everything is too complicate. But...
-
@AlysonKelman @JenniferBoag You're both making good points. I would just say that every method and every particular application of these methods has strengths and weaknesses. It's really important that these be discussed openly and honestly.
I would say that surveys often do get more respect than they deserve because they are seen as more scientific...
-
Michael Spagat replied to Pete B
I didn't know this but you drove me to look it up and I found this thread: https://twitter.com/TimHarford/status/1301417268308586497
I would observe that everyone makes mistakes and it's not realistic to aspire to never making one. The really important thing is what you do when your mistake is exposed. Do you dig in and insist that you were right? Or do...
-
@IanCampbell Yes, your second paragraph points to difficulties in determining households in a way such that every person is a member of one and only one household. Your examples are not necessarily insurmountable but require careful questioning to address, e.g., asking how much time a child spends in each of multiple households.
-
Michael Spagat replied to Pete B
@JenniferBoag Hi, yes you're right there's a whole wonderful genre there and, I can't resist saying, we get into some of this territory in my other Futurelearn course: Survival Statistics.
Interestingly, in the latest Tim Harford book you mention he expresses some regret that More or Less has done so much debunking, fearing that he might inadvertently be...
-
Hi. When trying to make a complete listing of the victims of a discrete event it makes total sense to interview people linked to people who might have been in the building to try to figure out, e.g., whether some people thought to have been inside might have been out of town or whether some people might have been living there...
-
Hi. There is another statistical method, capture-recapture, that we'll be discussion soon.
-
Michael Spagat replied to Dave Rees
Hi. Yes, the impulse to record casualties seems to be a natural human trait that arises constantly in a wide range of contexts, not just wars. There was, for example, a project to record names and basic information of all victims of the Grenfell Towers fires and, almost surely, there will be a similar project for the recent building that collapsed in Miami.
-
Hi. Yes, exactly, moving from counting to documentation.
-
Interesting. How to the criteria differ?
-
https://www.nps.gov/anti/planyourvisit/luminary.htm
Here's another approach. I just discovered this one a few days ago.
-
Yes, I think it will. Welcome.
-
Hi. Yes, indeed, this fits right in with your interest!
-
Michael Spagat made a comment
This one is impressively menacing - join up or else. The message is quite clear, unlike some of other ones I glanced at that were overloaded with words.
-
Michael Spagat replied to Tavis Reddick
Thank your large number of thoughtful comments throughout the course. It's not hard to understand why you feel exhausted. I hope that you're enjoying yourself right now in another course. Best of luck to you!
-
Michael Spagat replied to Aideen Hurley
Thank you very much for your kind words and for your many great contributions to the course!
-
Hi. I think that the US military does on occasion do on-the-ground investigations of possible civilian harm but such investigations are rare.
-
Michael Spagat replied to Erin B
Hi and good question.
When applied to war, the term "excess deaths" (with no further modifier) always (so far as I'm aware) includes both violent and non-violent deaths. And, yes, the thinking is exactly along the lines that you suggest - excess deaths are excess relative to what would have been expected without war and violent deaths should, therefore,...
-
Hi. Action on Armed Violence looked at some of these issues.
https://aoav.org.uk/2020/why-aoav-compared-us-and-uk-soldiers-deaths-in-the-war-on-terror/
I have to assume that militaries have as well but, so far as I'm aware, none has published anything on it.
-
These are all reasonable questions. The problem is that they will be hard to answer, especially if there are a large number of similar cases, each with its own nuances.
-
This is an interesting angle. These pressures must have their effects. I suppose you could try an intermediate approach according to which you might say, in effect, "we have a number of cases like this one that we exclude just to be on the conservative side even though there is a case for including them."
-
Michael Spagat replied to Erin B
Yes, very good point. Thank you.
-
Michael Spagat replied to [Learner left FutureLearn]
Looking forward to having you.
-
Michael Spagat replied to Paul Wright
Thank you for joining again! Hopefully I'll have a third course in the not too distant future.
-
Michael Spagat replied to Doug Karo
Thanks. It's been great having you and it will be great to have you again. Right now there is just one other course, Survival Statistics, but I'm contemplating another one. At the moment there isn't a concrete plan for the next run of Survival Statistics but I'd predict that it might run this time next year. Hope to see you there!
-
Michael Spagat replied to Andy Sharp
Thanks! It was great having you and all your great contributions.
-
No worries. It does sound like you did learn something from what you did on this and you were, without question, one of the most active participants in the course. I learned quite a bit from you and I think you also learned something.
-
Yes, I think you're probably right that this is one of the factors.
-
Thanks for this link!
-
Michael Spagat replied to Sue PF
I agree that as a general rule the poorer the country the less reliable the statistics. That said, there are major efforts to conduct high quality surveys all around the globe so that often decent statistics are generated even in very difficult circumstances.
-
Hi. Yes, it's true that in some sense Covid saved the lives of some people (while costing the lives of many more others). Your flu example is good. Another such example is deaths in traffic accidents which went down sharply. Of course, the families of the people who would otherwise have died in traffic accidents don't know who they are.
-
Michael Spagat replied to Sue PF
Hi. I created this particular calculation for teaching purposes but calculations like this are often done on real data. For example, the excess deaths estimates you hear for Covid are basically done like this. It's true that this takes individuals out of view in favour of looking at the big picture.
-
Yes, this is possible and this is done as we shall see in upcoming steps.
-
What terrible year this has been for you. My condolences.
-
Michael Spagat replied to Gloria H
Wow you've taken so many interesting FL courses! I hope you enjoy this one.
-
Michael Spagat replied to Pete B
So long and hope to see you again. It's been wonderful having you!
-
Michael Spagat replied to Ian Campbell
Yes, thanks all around to everyone who has made this so much fun and thank you in particular for your many great contributions.
-
Michael Spagat replied to Susan Scothern
Really glad to hear that!
-
Michael Spagat replied to Alejandro Dinas
Chévere
-
Michael Spagat replied to Corinne HG
Thank you very much. I'm particularly happy to hear that you were pleasantly surprised by the stats/maths.
You'll be happy to know that as soon as I make it through this page I will "cook" some box mix brownies with the kids and watch football.
-
Michael Spagat replied to Sue PF
Thank you so much both for your kind words and active participation. It's been wonderful having you!
Also, I love your point about the other participants. It's been really rewarding to see everyone working together and helping each other. I think we've had a pretty diverse group with lots of different perspectives but I can't recall anyone fighting. ...
-
Michael Spagat replied to Pete B
Thank for this comment and for your many great interventions throughout. Best of luck to you!
-
Michael Spagat replied to Susan Scothern
Great. That's a good result. Best of luck!
-
Michael Spagat replied to Corinne HG
It sounds like you enjoyed the course which fill me with joy.
One message I would like to impart is that there is hardly such a thing as purely quantitative work in this area. It's a mix of quantitative and qualitative although many people may not see this or be willing to recognize it. The numbers are important but just looking at the numbers isn't...
-
Michael Spagat replied to Alejandro Dinas
Thank you very much for your kind words. Enjoy your studies and the best of luck to you!
-
Michael Spagat replied to Anthony Rich
Good luck to you! I hope you have a happy and productive retirement.
-
Michael Spagat replied to Pete B
Very nice list!
-
Michael Spagat replied to Doug Karo
@SuePF and Doug.
I do think that there were substantial advantages in Kosovo, mainly stemming the fact that it's a relatively rick country, unusually rick for a country experiencing an internal war. There had been a recent census and there was good funding for all of these efforts, especially Kosovo Memory Book.
-
Thanks. I appreciate the appreciation!
-
Yes, I agree. In my eagerness to make a point about the estimates I hope that I didn't convey otherwise.
-
I think that the main pressure to inflate comes from the idea that people will take a problem more seriously if that problem is seen to be really big. So if people hear the X people have been killed in country Y they might not be interested - but if the number of people killed is 10X then they will be interested. This is a crude formulation but I think it...
-
Hi. I seem to be gradually catching up to you...just two days behind now.
Yes, it is very difficult to draw a convincing boundary between crime and war. I think that the main thing that the various conflict databases do to try to accomplish this is to look at the reason reason that non-state groups are fighting. Are they trying to overthrow the...
-
Yes, this is an issue that would undermine most the the standard approaches of the sort we're discussing now. In principle, you can develop the methods to address the issue you raise although you need to have a pretty decent idea of what is going on to be able to adapt the method in the right way.
-
Thank you very much for this link! I didn't know it and it's very interesting to me.
-
Michael Spagat replied to Aideen Hurley
Hi. These are all valid concerns although some people may welcome an opportunity to be heard.
-
Hi. Diverse sources. You can often find funders listed on the web sites of the organizations involved. But I would say mainly charitable foundations and foreign aid budgets of various countries. And, yes, it can be expensive and, depending on quality, it can be quite expensive.
-
Very interesting
-
Very nice, succinct comment.
-
Michael Spagat replied to Anthony Rich
I believe that you're referring to a comment I made that people relate more strongly to numbers of people killed, e.g., 10,000 people killed, than to rates, e.g., 2 deaths per 1,000 people per year. I didn't mean to add that giving numbers killed is necessarily preferable to giving rates and, in fact, I quite like the presentation in this step which is done...
-
Michael Spagat replied to Tavis Reddick
@CorinneHG @TavisReddick Hi. When you two do an FL course you really do it, not just going through the motions.
A few observation - If surveyors knock on the door of an (apparently) empty house they should try again later rather than interviewing the neighbours and if they still can't they should eventually give up and, possibly randomly select...
-
Michael Spagat replied to Corinne HG
Hi. Yes, there was some of that feeling. However, people are generally very reluctant to write critique papers for various reasons so the chance of this happening wasn't high. It's actually a problem for science that critiques aren't valued as much as they should be. If the objective is to discover the truth then critique should be on an equal plane with...
-
Hi. You're right that transparency shouldn't generally mean full transparency about everything. For example, surveys never disclose the names of people they interview - they are simply household 1, household 2, etc.
-
You're right, this is a problem. The notion of a household is more fluid in some places than it is in others. There might be cousins, aunts, uncles, etc. circulating through, having some meals and maybe even sleeping over sometimes. Even in the UK there are ambiguities such as university students who may, e.g., return to their parents on weekends.
-
Michael Spagat replied to Jennifer Boag
Interesting post. It's good that you remind us that there's unexploded ordinance in the UK.
-
Ah, you're right, I've just fixed the links on the "Also" list. Thanks!
-
Yes, I do believe that there are interactions between the dynamic of deaths and the dynamics of births. That said the statistics we've been working with are about the number of deaths per a fixed number of live births so these dynamics are probably of secondary importance.
-
Hi. I prefer to focus on the work itself rather than on the motives of people doing the work, especially since I can't discern what's going on in peoples' heads. But if I had to speculate I'd guess that they thought they could help to get the world to do something by finding vast numbers of deaths there although, in fact, such huge numbers are at least as...
-
Fair point. Yes, there are more than a few cases for which, on balance, the overthrow of a bad regime ends up saving lives.
The point I was trying to make was that just because excess deaths are negative doesn't mean that lives are being saved because death rates could have been heading down anyway, and would have decreased faster without war. I think...
-
Hi. Both paragraphs are interesting. On the first, it's interesting to think about why this might work better than just averaging all 10. The idea would have to be that once estimates are fairly far away from most then they are kind of crazy and unreliable.
-
Hi. Everything you say makes sense.
-
I can think of a couple of reasons that wouldn't necessarily always apply but would sometimes. First, there could be financial benefits for survivors if a death is classified as a war death. Second, a war death might be viewed as more heroic or worthy than would other kinds of deaths.
-
Yes, it is possible to have negative excess. Your examples pretty much take to from of a war saving lives (on balance) because a bad government is replaced by a better government. In such cases the excess death concept would pretty much do what we'd hope it to do - reveal that lives saved by a war exceed lives lost.
There are other scenarios for negative...
-
Michael Spagat replied to Andy Sharp
Interesting case, and not extremely far fetched. We often want to assign one unique cause of death when, in fact, their are multiple. The idea of "excess deaths" which we're just getting to tries to sidestep the question of finding case-by-case causes of deaths. But if we wish to embrace case-by-case analysis then you're idea of invoking proximate causes...
-
Michael Spagat replied to Doug Karo
Yes, these are important points. If there are seriously flaws in the way you're collecting your data then you can't compensate by collecting a lot of it - in other words, you've gotta get the art right.
-
Michael Spagat replied to Jennifer Boag
This is the first time I've heard of giving up on listing names in the UK for WW1 or WW2 because of a lack of resources. But I'm guessing that it must have happened elsewhere if it happened in one place.
-
Welcome. Yes, I think you're correct that these courses should complement each other.
-
Michael Spagat replied to [Learner left FutureLearn]
@DougKaro Hi. I think this is a fair description of the excess death idea. I would just add that even with excellent record keeping it's still complicated pinpoint causes of death, especially if you're looking for a single cause when, often, multiple factors contribute to a death. The excess death concept can help to address such amiguities.
-
@IanCampbell I think they may have provided a dataset a long time ago but not since.
-
One criticism I'd make of Syrian Human Rights Observatory is that there is not posted dataset that you can download.
-
Interesting idea. I'll try to think about this.
-
Michael Spagat replied to John Cope
Hi. I'm glad that this was a good experience for you. Judging by the date of your comment I think we can take you at your word when you say you found it a page turners although you progressed at a snails pace compared to Joanne Saltfleet (just below).
-
Michael Spagat replied to Frances Madsen
Fantastic, this makes me happy. Thank you for participating.
-
Michael Spagat replied to [Learner left FutureLearn]
Thank you for your many great contributions to this course which continue right down to step 3.26. I'm not surprised to hear that you're drained but hopefully you're glad that you did the course.
-
Michael Spagat replied to William Doyle
Quite an interesting, albeit shocking, clip.
And thank you for your active participation in the course.
-
Michael Spagat replied to Alyson Kelman
Hi. It's too bad we didn't connect better on the maths but, nevertheless, it's been great having you. I really appreciate tour many great contributions to our discussions.
-
Yes, this is a very important part of the work, hard to do and hard to automate. Data entry mistakes are really problematic.
-
Yes, indeed, it gets really complicated with many lists and, yes, when used appropriately it can offset biases of individual datasets.
-
Yes. To take an extreme case, imagine that we have two lists covering covering almost entirely different periods but with tiny overlap in time period and, consequently, only few deaths are on both lists. Then the estimate would be inappropriately high because the tiny number of deaths on both lists would be misinterpreted as telling us that each list has a...
-
Yes, this is definitely the same method and we all owe you and your fellow students huge gratitude for the way you hung in there and made sure that we aren't all covered now by an infinite number of wood lice - from your description it was a near miss.
-
Yes. All data in any statistical application should be carefully scrutinized prior to analysis yet this is often not done and too much respect goes to analysts who apply some fancy method but without necessarily understanding the data they are using. That said, I think that you're caveat applies with particular force to this method.
-
If you have more than two lists then you can relax the assumption of independence across lists although other complications grow as the number of lists increases.
