Professors Michael Spagat, Neil Ferguson & their conflict of interest in criticising Lancet report

Media Lens Message Board
Professors Michael Spagat, Neil Ferguson & their conflict of interest in criticising Lancet report
Posted by David Sketchley on October 22, 2006, 12:13 pm, in reply to "Re: exchange with Les Roberts re "main street bias""

Is it really any wonder that Professors Michael Spagat & Neil Ferguson want to discredit the Lancet report?

I think not when you look closely at their work and in particular some of their latest research papers which Spagat tells us proudly on his website "has been written up in Nature, the Economist and the Guardian":

Welcome to Michael Spagat's Home Page!"

"I am also proud to announce the launching of a new web site called Civil Conflict Analysis Resources that I hope will become a major resource for conflict researchers, especially economists and political scientists. This site is the brainchild of my infinitely energetic and entreprenuerial PhD student, Jorge Restrepo, who is about to finish his disseratiation on the Colombian conflict. Also please visit the web site of CERAC, a think tank that we have set up in Bogota specializing in the analysis of conflict and violence.

New on the research page you can find a paper on Iraq, Colombia and global terrorism that finds surprising similarities between the three. This work has been written up in Nature, the Economist and the Guardian."

The research paper is called:

""From Old Wars to New Wars and Global Terrorism," with Neil Johnson, Jorge Restrepo, Juan Camilo Bohórquez, Nicolás Suarzez, Elvira María Restrepo, and Roberto Zarama.
This paper has received press coverage in the Economist, The Guardian and Nature."

The important thing to notice however is that Professor Michael Spagat uses data from IBC for all his conflict analysis research that includes Iraq. This also applies to Professor Neil Johnson.

Obviously if all their reserach is based on IBC figures, then their findings would be virtually useless if the Lancet figures were correct.

Hence their desire to discredit it.

They are most definitely not 'honest brokers' in this affair:

"From old wars to new wars and global terrorism"N. Johnson1,7, M. Spagat2,7, J. Restrepo3,7, J. Bohórquez4, N. Suárez5,7, E. Restrepo6,7, andR. Zarama41 Department of Physics, University of Oxford, Oxford, U.K.2 Department of Economics, Royal Holloway College, University of London, Egham, U.K.3 Department of Economics, Universidad Javeriana, Bogotá, Colombia4 Department of Industrial Engineering, Universidad de los Andes, Bogotá, Colombia5 Department of Economics, Universidad Nacional de Colombia, Bogotá, Colombia6 Department of Economics, Universidad de Los Andes, Bogotá, Colombia7 CERAC, Conflict Analysis Resource Center, Bogotá, Colombia
"Our analysis uses our ownunique dataset for killings and injuries in Colombia, plus publicly available data forcivilians killed in Iraq."
"For the Iraq data we work with killings of civilians as provided by the Iraq Body Count Project."

"Universal patterns underlying ongoing wars and terrorism"Neil F. Johnson1,6, Mike Spagat2,6, Jorge A. Restrepo3,6, Oscar Becerra6, Juan Camilo Bohórquez4, Nicolas Suárez6, Elvira Maria Restrepo5,6, and Roberto Zarama41 Department of Physics, University of Oxford, Oxford, U.K.2 Department of Economics, Royal Holloway College, University of London, Egham, U.K.3 Department of Economics, Universidad Javeriana, Bogotá, Colombia4 Department of Industrial Engineering, Universidad de los Andes, Bogotá, Colombia5 Faculty of Economics, Universidad de Los Andes, Bogotá, Colombia6 CERAC, Conflict Analysis Resource Center, Bogotá, Colombia

"Our findings are backed up by extensive statistical tests on carefully prepared datasets, as discussed in the Appendices."

"PART 2: Data and methods...For Iraq we work with the CERAC Integrated Iraq Dataset (CIID). The CIID builds on the event description from three datasets that monitor violence in Iraq: Iraq Body Count (http://www.iraqbodycount.net/), iCasualties (http://iCasualties.org/oif/) and ITERATE (http://www.cba.ua.edu/~wenders/). All three sources contain event data on the Iraq war from its beginning on March 20, 2003. The first two are continually updated whereas ITERATE is updated on an annual basis so at present only extends through the end of 2004. As we discuss below, ITERATE has a very small impact on CIID so the fact that it stops early does not affect the work of this paper.The Iraq Body Count Project (IBC) monitors the reporting of more than 30 respected online news sources, recording only events covered by at least two of them. For each event IBC logs the date, time, location, target, weapon, estimates of the minimum and maximum number of civilian deaths and the sources of the information.vii IBC attaches the most confidence to their figures on the minimum number of killings in each event so the figures in the paper are based on these minimum numbers. However, figure S8 (below) shows that Figure 1 changes very little if we substitute the maximum number of killings for the minimum number of killings. The concept of civilian is broad, including, for example, policemen. The list of events, posted online, covers the full range of war activity, including suicide bombings, roadside bombings, US air strikes, car bombs, artillery strikes and individual assassinations.The IBC data has two principle drawbacks which need to be addressed in order for the reader to have confidence in our results. First, some lines in the IBC spreadsheet contain entries that are not proper events. The most important entries of this form are based on reports from morgues around Iraq. For example, entry x355a lists 26 deaths between May 1, 2003 and May 31, 2003 described as “Violent deaths recorded at the provincial morgue of Karbala”. The following hypothetical calculation illustrates how IBC handles these entries. The Karbala morgue actually reports a higher figure, say 39 violent deaths for May of 2003. However, IBC already has two events in Karbala for May of 2003, a car bomb killing 4 and a suicide bombing killing 5. It is likely that these 9 deaths are included among the 39 violent deaths recorded by the Karbala morgue so IBC subtracts them off, leaving 30. In addition, the murder rate in Karbala before the war was 4 per months so we might expect that 4 out of the remaining violent deaths would have happened even without the war. IBC subtracts off these 4 leaving the figure of 26 which is the one they enter into the database. This procedure is reasonable on its own terms, however deeply problematic for our purposes in this paper for two reasons. First, most of the deaths in entries of this form are likely to have occurred as single homicides since larger conflict events would be likely to have their own entry. Clearly we would not wish to treat 26 individual homicides as one event in which 26 people were killed. Second, most of the killings in events of this form are more tied to crime than to the conflict directly and we prefer to focus on a narrow definition of conflict killing. For these reasons we delete from the IBC database entries of this nature. We provide a list of events we deleted from IBC in Part 4 of the Appendices.The second drawback of IBC is that it measures only civilian deaths, albeit with a wide concept of civilian. Therefore, to get of fuller picture of the conflict we have added in events from iCasualties in which coalition military personnel and contractors are killed in conflict events (but not in accidents). This is a highly reliable source as the military services keep solid records on the fate of their own personnel. Finally, as a check for coverage of IBC and iCausualties we also integrated events from ITERATE, which is a global terrorism database that records terrorism events of international significance.This integration required careful matching of events between the three sources to avoid double counting. The following Venn diagrams give the results of this matching work, with event counts and numbers of killings accounted for by these events in parentheses below. They show that most of the deaths in CIID come from IBC alone but that iCasualties does make a significant contribution. The overlap between IBC and iCasualties is small, since they are measuring different things. However, there is some overlap because sometimes both military personnel and civilians are killed in a single incident. The impact of adding ITERATE into CIID is negligible, indicating that IBC and iCasualties give very full coverage of the Iraq war."


When we go to the website Spragat's protegé Restrepo has set up - the Civil Conflict Analysis Resources, we find that they too use IBC data for their research:

"Country specific datasetsIraq Body Count
Ron Francisco's civil war data (Spain and US Civil Wars)
Sutton Index of Deaths in Northern Ireland
Restrepo-Vargas-Spagat Colombia Civil War Dataset (1988-2002"



Update 03/11/06:


In fact the website which notes IBC as a country specific dataset is the Dept. of Economics at the Royal Holloway. I discovered this while I was writing the following mail to Dr. Madelyn Hsiao-Rei Hicks. :


Dear Dr, Hicks,

I have recently read an analysis by you of the new Lancet study of deaths in Iraq that was posted by Josh Dougherty on the Media Lens Message Board: A comment on the methodology described by Burnham et al. used to estimate ‘Mortality after the 2003 invasion of Iraq’ published on October 12, 2006 in the Lancet

I am interested in how you came to write this analysis. In particular I am interested to learn whether you were approached by anyone specifically asking for your comments. If so, was it , by chance, someone from IBC, Iraq Body Count, or perchance from the Royal Holloway, University of London or the Dept. of Physics at Lincoln College, University of Oxford?

I ask because I notice that you are a research associate of CERAC, and as such, you are a colleague of Prof. Neil Johnson and Prof. Mike Spagat who are also "academic visitors" to CERAC, and who also, curiously, recently criticised the latest Lancet report.

I say curiously, because the overwhelming body of opinion of scientists in the field of bio-statistics, actually supports the Lancet methodology.

I also find it interesting that Prof. Neil Johnson and Prof. Mike Spagat were the co-authors of a research paper called "Universal patterns underlying ongoing wars and terrorism"(2), an updated version of a previous research paper called "From Old Wars to New Wars and Global Terrorism"(1).
Curiously, in both research papers, calculations and conclusions were made, based on data for civilian killings provided by: IBC.
(1) "For the Iraq data we work with killings of civilians as provided by the Iraq Body Count Project."
(2) Appendices PART 2: Data and methods "For Iraq we work with the CERAC Integrated Iraq Dataset (CIID). The CIID builds on the event description from three datasets that monitor violence in Iraq: Iraq Body Count , iCasualties and ITERATE."

In fact, apart from CERAC using IBC as part of its Integrated Iraq Dataset (CIID), it transpires that the Dept. of Economics at the Royal Holloway, University of London, also uses IBC in its country specific datasets (click on 'datasets' link).

This leads me to believe that there is a conflict of interest here. If IBC figures are wrong, and the Lancet figures are correct, then the conclusions reached in these research papers and indeed any other that used the IBC figures, could be seriously compromised. Therefore, the scholars in question cannot possibly be 'honest brokers' in this affair but are merely defending their own interests.

And as a colleague of theirs at CERAC which also uses IBC, there also appears to be a conflict of interest.

I would be grateful for any comments you could find the time to make.

Yours Sincerely,

David Sketchley
Seville, Spain