I know you guys have been following the debate on H-War. A couple of the posters there posed a question that I have been thinking about and haven't been able to figure out the answer to. It is this: regarding the German and Allied casualty rates for the Italian Campaign data listed by Col. Dupuy in NPW, the Germans suffered slightly higher total casualties and a higher rate of casualties than the Allies. Yet, Col. Dupuy asserts that the Germans had a higher score effectiveness than the Allies, regardless of posture and whether or not they lost the engagement. This does not seem consistent with the casualty data for the database.

I have now read Attrition, and the higher German casualty rate can probably be explained by the fact that they were losing most of the engagements, but this does not square with Dupuy's assertion about German score effectiveness.

Can you guys shed some light on this? I have been wanting to post an explanatory response to H-War (although that seems to do litle good) but haven't been able to work this out satisfactorily. So I figured it was time to turn to the experts.

Cheers,

quote:

---

Originally posted by yadernye:
It is this: regarding the German and Allied casualty rates for the Italian Campaign data listed by Col. Dupuy in NPW, the Germans suffered slightly higher total casualties and a higher rate of casualties than the Allies. Yet, Col. Dupuy asserts that the Germans had a higher score effectiveness than the Allies, regardless of posture and whether or not they lost the engagement. This does not seem consistent with the casualty data for the database.

---

If one is consistently outnumbered and outgunned, then one would expect to take more casualties and loose more battles. The model scores all the weapons (a very rough approximation of combat power) and these scores are then compared. Furthermore, the scores are modified by the conditions of the combat (terrain, fortifications, weather, etc). What generated the conclusion that the Germans had generally higher combat effectiveness is that when all these factors are taken into account, they tended to performed better than the predicted outcome of the model, whether on attack or defense (meaning the model had no implicit bias in one mode or the other). As the model is complex, and the CEV is based upon a comparison to the predicted result of the model, then some people just called it "voodoo" and dismissed it outright.

There are other ways to get there than the QJM. For example, in our EPW Report Phase I & II, we do the same test using simple aggregate averages and statistical comparisons. In particular, we compare the performance of different nationalities in low odds attacks on both the attack and defense. There is no scoring and no QJM/TNDM used. This still produces very similar answers to what the QJM produces, using a completely different methdology.

Furthermore, the British OR community has also looked into this subject, primarily David Rowland, and also has come up with similar results.

The fact that there are three different methodological tests that have produced similar results seems to be simply ignored.

Thanks for the response. I agree with everything you stated above, and I believe the QJM/TNDM methodology is fundamentally sound.

My question is a bit more esoteric: Regarding the German score effectiveness (as opposed to CEV), Col. Dupuy claimed that they inflicted greater casualty rates on their enemies regardless of whether they were attacking or defending, victorious or defeated. However, Figure 7-6 of NPW would seem to contradict this, as the Germans are listed as having sustained both higher total casualties and a higher daily loss rate, despite their higher score effectiveness.

I would agree that the German loss rates were due to being outnumbered and losing most of the engagements in the database, but this appears to contradict Col. Dupuy's assertion of a score effectiveness superiority regardless of posture and success. This specific question, I suppose, is how can the Germans have had a higher score effectiveness while suffering both higher aggregate casualties and daily loss rates?

It is this specific issue that has been brought into question by some of the H-War list members. This matter relates more to the issue of score effectiveness as opposed to CEV, but because Col. Dupuy associated the two in NPW, they tend to be synonymous in the minds of his critics. Critics seize upon such discrepancies as proof of flaws in the methodology.

My own guess is that consolidating the statistics of the originbal database resulted in a somewhat distorted picture. The QJM models tactical combat at a fairly specific level, that of the engagement in Col. Dupuy's taxonomy of combat. This lies at the upper range of the tactical realm and the lower level of the operational. As such, while it accounts splendidily for tactical factors, it does not wholly represent the operational level, where factors such as maneuver, supply, fatigue, etc. play a larger role. Many of these operational factors can be represented in the QJM, but Col. Dupuy and his colleagues haven't had the time or sufficient data to calculate them separately, and they are often aggregated together in the CEV. I also suspect this is partly due to the methodology, which looks specifically at tactical engagements, as opposed to the relationship of individual engagements to a complete operational campaign.

Thus, the German casualty rates can be explained by operational factors that really lie beyond the scope of the QJM methodology. That is, they occured due to the operational results of the engagments, i.e. encirclements, disorganized withdrawals, and such. It seems to me that you touched on this notion briefly in your exchange with John Sloan on H-War regarding KIA/WIA vs. captured.

It also seems to me that the distinction of exactly what frame of combat the QJM models is lost on most of its critics. "Combat effectiveness" is precisely what it means, and should be differentiated from operational, strategic and overall military effectiveness. I think this distinction is blurred because Col. Dupuy sometimes stretched the boundaries himself in demonstrating the use of the QJM. But this also illuminates that the boundaries between levels of combat are often blurry and ill-defined as well, and that the relationships between them are complex and non-linear.

These are just some thoughts I've been mulling over. I would be quite interested to hear your response.

Cheers,

Sorry for the very delayed response on this. As always, juggling more balls than I have the talent to handle.

quote:

---

Originally posted by yadernye:
Regarding the German score effectiveness (as opposed to CEV), Col. Dupuy claimed that they inflicted greater casualty rates on their enemies regardless of whether they were attacking or defending, victorious or defeated. However, Figure 7-6 of NPW would seem to contradict this, as the Germans are listed as having sustained both higher total casualties and a higher daily loss rate, despite their higher score effectiveness.

---

I do not think this table (table 7-6, from page 104)contradicts that assessment. First the total numbers engaged are 1.90 to 1 in favor of the allies. Even if one factors in an average defense factor of 1.3 to 1.5 for the Germans, this still leaves the allies at an advantage. The other major factor that would need to be accounted for is airpower, which almost always was to the benefit of the allies.

Therefore, we have the allies attacking at 2 to 1, and the overall casualty exchange is equal (1 to 1.02). This looks like a German performance advantage here.

The average percent losses favors the allies (1 to 1.46). This is not surprising as the allies usually outnumbered the Germans and therefore would naturally tend to have a lower percent losses. Again, even the difference in percent losses is not to the same degree as the difference in force ratios (or firepower if one adds in allied artillery and air). Again, this points to a German performance advantage.

Let us look at the data in a different light:

The following is from our Capture Rate Study, Phase I & II, page 49.

All US Attacks (37)
Average Force Ratio 2.52 to 1
Average Loss Ratio 3.01 to 1

US Low-odds attacks (3) (.72 to 1.31 to 1)
Average Force Ratio 1.04 to 1
Average Loss Ratio .26 to 1

All UK Attacks (21)
Average Force Ratio 2.34 to 1
Average Loss Ratio 1.71 to 1

UK low-odds Attacks (4)(1.17 to 1.41 to 1)
Average Force Ratio 1.30 to 1
Average Loss Ratio 2.11 to 1

All German Attacks (17)
Average Force Ratio 1.88 to 1
Average Loss Ratio .97 to 1

German Low-odds Attacks (7)(.73 to 1.48 to 1)
Average Force Ratio .89 to 1
Average Loss Ratio .68 to 1

Note the critical comparisons, when the US attacks in 37 cases at 2.52 to 1, they loose 3.01 to 1 to the Germans. The UK shows a similar pattern with 21 UK attakcs at a force ratio of 2.34 to 1 and a loss ratio of 1.71 to one. On the other hand, when the Germans attack, the picture reverses, with their 17 attacks being at a force ratio of 1.88 to 1 and a loss ratio of .97 to one.

This comparison factors out the advantage of defense in the aggregate (although it does not address a lot of other issues, like air power). It does not use the QJM/TNDM at all, as it is based solely on manpower force ratios. The pattern is clear. It does use our slightly revised, improve Italian data.

quote:

---

Critics seize upon such discrepancies as proof of flaws in the methodology.

---

The critics appear to want to seize onto anything to disprove the methodology. Note that our Capture Rate Study establishes the same performance relationships and does not use the QJM or TDNM, just simple statistical analysis.

Between your explaination and a careful re-reading of NPW and ATTRITION, I am satisified that the question of "discrepancies" has a reasonable answer.

I've been distracted as well by my PhD thesis, but I am hoping to post comprehensive replies on this subject to H-War in the not so distant future. I've thoroughly enjoyed getting reacquainted with Col. Dupuy and the QJMA, and greatly appreicate the time you and your colleagues have taken to respond to my querries.

Cheers,