In Trevor Dupuy's Attrition: Forecasting Battle Casualties & Equipment Losses in Modern War, Nova Publications 1995, the relevant quotation (pages 89-90) is as follows, the last sentence is key: "The ratio of the tank loss rate to the casualty rate appears to be a function of the density or proportion of tanks in the force. A force can be considered armor-heavy when the proportion of tanks exceeds 6 per 1,000 troops. When two armor-heavy forces engage in battle, whether for a short or long period, the ratio of tank loss rates to the casualty rates will be in the range of 5.00 to 8.00 for the successful force (with an average value of about 6.00) & much higher for the losing force. There are, of course, exceptions, as there always are in combat data, where each event is unique, but the pattern is clear. Furthermore the ratio of tank loss rates to casualty rates appears to remain constant as the proportion of tanks increases above 6 per 1,000 troops. On the other hand, that ratio decreases, apparently in more-or-less linear fashion, when the proportion of tanks decline below 6 per 1,000 troops. It appears further that the range of the ratio for an armor-heavy force is likely to be between about 5 & 7 for a winner & between about 7 and 10 (sometimes much higher) for a loser."

Does this still apply to today given significant changes/disparities in technology?

quote:

---

Originally posted by Tom Rodwell:
Does this still apply to today given significant changes/disparities in technology?

---

Keep in mind that he is talking about the ratio of tank loss rates to personnel loss rates. This is derived from WWII data, although much of Dupuy's modeling and analysis included data from the 1967 and 1973 Arab-Israeli Wars.

By the nature of the statistic, I am not sure that it can vary widely from the historical pattern. As loss of tanks will almost always entail loss of personnel then for the ratio to be higher over time, this would mean that much more highly selective targeting would have to occur that would result in the primary personnel losses being the tanks crews. While this would certainly change the ratios, to date, I don't not think this has been the norm in any war.

Only the US really has the precision capability on a large scale. The first war we extensively employed it in was the Gulf War, where we not only used precision weapons, but also a whole lot of traditional weapons. In the armor operations I've read about, pretty much all weapons (precise and less than precise) were focused on the target area. As such, I would not expect to see a change in such a ratio.

One could test this by an analysis of Iraqi tanks loss rates to personnel loss rates, but we have not done this specific test. The data on Iraqi losses is not very detailed or reliable. Furthermore, the Gulf War does not provide a very large number of division-level combat actions (less than 20 really). As such, it does not have "statistical robustness".

I realize your focus is on modern applicability of QJM, however I would strongly recommend having a look at a post graduate thesis entitled “Fitting Firepower Score Models to the Battle of Kursk Data Base” By Ramazan Gozel, Naval Postgraduate School. This thesis applies the Firepower score models to Chris’s rather extensive Kursk data base. Three different firepower score models were assessed; Including ATLAS ground attrition model, which is used in the TACWAR simulation. The second model is RAXD's SFS model, which is apparently utilized in the JICM simulation. The last model is the QJM model. The thesis gives a relatively unbiased assessment of all three firepower score models, elaborating on the ability of each model to predict casualties (including tank casualties). Comparisons are made to the extremely detailed information presented in the Kursk Data Base. There are limitations in each of the respective models and there ability to predict casualties relative to a World War II engagement. Presumably one would approach this from the perspective of determining limitations in modeling relative to the type of combat for which the models were developed (i.e. QJM is founded heavily in data collated from WWII combat) prior to attempting any sort of model extrapolation into post-WWII combat.

If you would like, email me your address again and I will pop a copy of this thesis into the mail for your review.

Best Regards

quote:

---

Originally posted by Jeff Duquette:
This thesis applies the Firepower score models to Chris’s rather extensive Kursk data base.

---

About a dozen people worked on this including Dr. Richard Harrison (author of "The Russian Way of War"), Richard Anderson ("Artillery Hell" & "Hitler's Last Gamble") and Jay Karamales ("Against the Panzers").

quote:

---

If you would like, email me your address again and I will pop a copy of this thesis into the mail for your review.

---

Actually, I wouldn't mind seeing this. I saw their breifing on trying to fit Lanchester to the Ardennes and Kursk data bases(and of course, the null result was achieved), but I have not seen this paper.

Email your snail-mail address. As I recall I obtained a copy of this Thesis from NTIS. I think I have a similar Thesis lying around here that utilized the Ardennes Campaign Database to compare QJM and Rand Models. I haven’t looked at this particular work very closely.

Having dug into the KOSAVE-II report a bit, my interest was naturally peaked when I came across Gozel’s work.

Gozel’s conclusions and the Kursk Data Base; German casualties were seemingly easier to fit into each of the respective Firepower Score Models...respectable correlations between actual and predicted. Correlation between actual Red Army casualties and modeled was relatively low.

Regards

Jeff Duquette

quote:

---

Originally posted by Jeff Duquette:
P.S. What is the current ETA on your upcoming book?

---

Well, right now, with the exception of a few of interviews, some biographies, some more statistics and some more analysis, I've pretty much put into it everything I am going to put in the book. As such, I am now doing my edit of the "first draft" and this is taking from one or two hours to several days a chapter. Right now, I am editing Chapter 9 (The Air War)...out of 28. As this chapter needs to be cross-checked and integrated with the rest of book, it will probably take the most time. After that, the rest of the editing is relatively simple.

Hopefully, the entire book will be edited by the end of June (or sooner), and then I will let Mr. Anderson take a chop at it. Then I will send it out to a half dozen other expert reviewers for their cut.

If it survives all this criticism, then I will send it to Westview along with the graphics (which we have not finished).

Regarding the Gozal Thesis it's in the mail to both yourself and Tom. I mailed it out last week.

quote:

---

Originally posted by Jeff Duquette:
Well I am looking forward to seeing the "unabridged" version.

---

Its actually a major problem. You can either fully describe the battle, which includes some detail on the actions of each division each day, or just outline the overall battle and throw it a couple of war stories highlighting the highpoints. Fully describing the battle takes a thousand pages. Most (almost all) authors, by necessity, take the second approach.

quote:

---

Regarding the Gozal Thesis it's in the mail to both yourself and Tom. I mailed it out last week.

---

Received it. Thank you very much. I am going to avoid reading through it for a couple of weeks as I have competing deadlines I need to hit (the Urban Warfare Studies and finishing the Kursk book). After I read it, I'll post up some comments.