The British late war tank casualty survey wo 205/1165 a data dump

[Christianmunich]

Hi,

This is exactly what their data says tho. The tank was hit, not penned and abandoned. So the tank was ko'ed without pen. The data certainly has a definition of "non-pen" that I disagree with but my classification is 100% correct. Hit-->"no pen"-->Crew baled = Ko'ed tank without pen.

No, I am afraid you are mistaken or are misconstruing what you are reading.

(a) crew got out after the first hit had damaged the top of the gear box, but does not appear to have gone back into the turret. No fire in the front compartment;

This is clearly a penetration, they are saying that it didn't "penetrate the crew compartment".

And even clearer example is A27 which states that:

Position of Hit: a) through front of gearbox. b ) 82 x 91 mm. diameter penetration through right side of 75 mm. mantlet. c) hit on 0.5” Browning mounting on roof of turret.

Course & Effect of Projectiles: Shot a) went first through a bogie which was being carried on the front of the hull, and then into the
gearbox. It did not harm the crew who started to bale out; b ) hit the tank after the Commander had got out, while the Gunner and Operator were still in the turret, the latter under the 75 mm. gun climbing across to the Commander’s hatch. It hurt no one and appeared to have done minor internal damage. Its course could not be traced; c) hit the Gunner in the head, killing him as he was getting out. It also threw off fragments which wounded the Commander who was standing beside the tank.

So clearly, hit (a) penetrated the gearbox but didn't enter the crew compartment. Hence they classed this as "didn't penetrate the crew compartment", you fail to make that distinction.

This is another clear example in which your interpretation would be fine if you added "crew compartment" to your definition of "did not pen".

And again A32 is another good example of where you are mistaken:

The survey states that:

Position of Hit: a) penetration into left final drive; b ) penetration through lower edge of extra armour welded in front of Driver; c) penetration through centre of front of hull just above gear box; d) penetration through lower edge of 75 mm. gun mantlet to left of gun;

e), f) and g) penetrations into left final drive

; h) hit on front of left sprocket.

Course & Effect of Projectiles:

Remarks:
Shot a) went into left final drive. This shot and/or e) knocked loose the plate separating the final drive and Driver’s compartment; b ) damaged exterior of forward left sponson ammunition bin, but did not explode any of the contained ammunition. Paths of individual shots were not traced, but an unexploded Kw. K. 40 APCBC. round was found on the floor of the turret. The engine of this tank was not burnt. Some 75 mm. rounds on the floor of the turret had burnt; but rounds in all three sponson ammunition bins had not burnt. Tank burned quietly for half an hour and then went out. Shot a) or e) was the first to hit, and injured the Driver and caused the crew to bale out. This shot did not penetrate into crew compartment, and the Driver must have been injured by secondary fragments.

Note multiple use of the word "penetration" and that, in this case, the first hits injured the driver albeit by "secondary fragments".

You might want to have another look at all these:

A27, A32, A35, A42, A72, A65, A79 and B73 - all of which I would suggest you might want to reclassify.

I hope that helps.

Edited to add: that's all from me tonight, I'm not disappearing though, just got more important things to do.

Regards

Tom

In the case of A.32 I actually marked the hit incorrectly, the survey has this hit as penetration and not as scoop. I wrongly marked the hit as non pen in the other column. The last sentences of the description said it didn't penetrate so I marked the knock out hit as non pen but it in the columns of hits I already marked it as pens.

This one is certainly an odd case because they emphasize that the hit did not penetrate the crew compartment:

This shot did not penetrate into crew compartment, and the Driver must have been injured by secondary fragments.

In A.27 the pictures of the survey have the hits mixed up, but this one as well should be marked as pen.

In case of A.35 the entry cleary states the crew baled before penetration.

A.42 is also correctly compiled and correctly classified from what I can tell.

A.65 correctly compiled, but was already stated by me as an odd case.

A.72 also wrongly compiled by me.

A.79 is correct from what I can tell.

B.73 also correct. No penetration at all occured.

A.27 A.35 A.72 are without a doubt incorrectly marked by me. A.32 I mixed up the last sentence because compiling the hits and "knockout before pens" happened at different dates, so I wrongly interpreted the last sentence in the descriptions. A.72 is completely false no idea what happened there. A.27 also incorrectly marked but there are also mistakes in the hit markings on pictures, which still are pens like in hit columsn so also no idea where I got this wrong. The rest is correct I assume.

Want to point out that the mistakes lay with me here, the survey cataloged those hits properly, I compiled them wrongly. Did you see others?

[Mori]

This is not a rough sample of casualties it is amateur level research.

Well, no: it's way better than what you usually get. And same applies to what you're doing, by the way.

So, when do you share you excel spreadsheet, so that we can look what's really there and play with it?

[Christianmunich]

This is not a rough sample of casualties it is amateur level research.

Well, no: it's way better than what you usually get. And same applies to what you're doing, by the way.

So, when do you share you excel spreadsheet, so that we can look what's really there and play with it?

Tom has shown some mistakes, which I was afraid of because I did no double checking, I will make some corrections, check if I did similar mistakes and then upload it.

Kenny's data sample is useless which I will prove soon. This will be helpful to show the problems with incorrect sampling et cetera.

[Mori]

Still my question is WHEN will you share your spreadsheet. I was expecting a commitment ("I will share it") and a date of release (e.g., "in ten minutes").

It's extremely difficult to understand why you dedicate so much time to detailed answers on this and other threads but are so reluctant to share your tool. It looks like plain fear. What are afraid of?
=> That others find mistakes? Well, it would just make the thing wiser and more relevant. What's the point?
=> That others post long, agressive criticism of the tool itself? Well, check what's been written so far on this forum: you already got a lot of that, so you wouldn't be hurt by another drop.
=> That others leverage your research for their own publications? Well, how much did the contributors to this thread publish lately? Not much, right. So what's the risk?

[Christianmunich]

Still my question is WHEN will you share your spreadsheet. I was expecting a commitment ("I will share it") and a date of release (e.g., "in ten minutes").

Are you kidding me? Lmao

[Mori]

Still my question is WHEN will you share your spreadsheet. I was expecting a commitment ("I will share it") and a date of release (e.g., "in ten minutes").

Are you kidding me? Lmao

Oh, I could wait for 12 minutes instead of 10.

[Christianmunich]

Since this thread is about casualty surveys and WW2 casualty samples for tanks in general, I would like to show the problems with improper sampling and how important it is to make sure to actually evaluate data sets that can be compared.

User Kenny posted the following data set:

5th RTR Cromwells

June 20 KIA 25 WIA 4 MIA 13 Tanks

July 11 KIA 54 WIA 2 MIA 8 Tanks

Aug 10 KIA 29 WIA 8 MIA 8 Tanks


To Aug 31: 41 KIA 108 WIA 14 MIA 29 Tanks



12thSS, SS Pz Reg 12 Pz IV's

June 6-10 31 KIA 36 WIA 8 MIA 18 Tanks



June 17 2 KIA 13 WIA 5 MIA 2 Tanks


June 26-27th: 15 KIA, 30 WIA, 3MIA 11 Pz IV

June 48 KIA 79 WIA 16 MIA 31 Tanks

Kenny labels this sample with "casualty-per-destroyed-tank" or "losses-per-tank ratio".

On first glance, this already throws up questions in regards to proper sampling. What does casualty per destroyed tank mean? Every casualty in a tank while it was destroyed or casualty of the entire unit while a tank was destroyed? Did the soldier have to become WIA/KIA while actually operating in a tank or soldiers counted which were counted outside the battle? None of those questions are answered and the data likely includes every casualty happening during a time frame compared to the fully destroyed tanks in the same time frame. What about knocked out tanks?

We see the data is interesting because the British loss numbers are more than 5 times greater than the actually destroyed tanks. But a tank had only a crew of 5 maximum. How is that possible? Well because the sample does sample casualties in destroyed tanks or while retreating from a tank, or from tanks that were not actually destroyed, it mixes together numbers that are connected in a way we don't know. This is best illustrated by using another sample.

I took the numbers of the 3rd RTR in a comparable time frame to the 5th RTR and checked the ratios there. It is important to note that I did not check daily entries of the diary nor did I check how they defined a casualty because it is irrelevant to my post here. If the two units reported their numbers differently we already have proof that such numbers are of no value without proper documentation.

Here are the 3rd RTR numbers and those of the 5th RTR, the numbers of the 5th are from user Kenny I did not double check them:

wompwomp.jpg

As you see the numbers are vastly different. The difference is quite remarkable, nearly double. How is that possible? Well because taking unit casualties and comparing them to tank casualties and claiming this are "survivability" numbers is awful scientific methodology. It does not work. Which we have proven here.

You might notice how both numbers are higher than generally accepted casualty per ko numbers, which is also explained by the samples not controlling for actual tank knockouts and also including outside of tank combat casualties. The 3rd RTR had an incident with a lot of casualties that apparently included no tanks but scout cars. By Kennies metric those would be included to infer casualty ratios.....

Kenny used such data sets to make conclusions on tank survivability. We see here how this is, nicely put, hogwash. It is difficult to conduct proper scientific surveys, you have to eliminate outside factors or control for them and the get a proper random sample, neither was done by Kenny.

One "sample" varies massively form another both units fought over the same time frame in the same rough area. If the methodology is weak the results will offer no value to researchers.

I believe this here highlights why we should focus on properly conducted research and try to increase our understanding based on properly collected data. It is far to easy for forum posts to choose badly collected data and say "he look here, don't believe the other data, believe mine".

[Michael Kenny]

The 3rd RTR had an incident with a lot of casualties that apparently included no tanks but scout cars.

Half right, half wrong.
The 'incident' should be well-known to anyone who is acquainted with the Normandy Campaign so I wont insult you by repeating the details. If we make an adjustment for this very well known 'incident' the averages converge. It is so anomalous that no serious student is unaware of it and conflate these casualties with tank crew combat casualties. For instance I would never use 3RTR/2nd FFY Casualty figures without mentioning how it would distort things.

Do you not agree?

Oh by the way you mixed the headers up on your 'secret' spreadsheet.

[Christianmunich]

The 3rd RTR had an incident with a lot of casualties that apparently included no tanks but scout cars.

Half right, half wrong.
The 'incident' should be well-known to anyone who is acquainted with the Normandy Campaign so I wont insult you by repeating the details. If we make an adjustment for this very well known 'incident' the averages converge. It is so anomalous that no serious student is unaware of it and conflate these casualties with tank crew combat casualties. For instance I would never use 3RTR/2nd FFY Casualty figures without mentioning how it would distort things.

Do you not agree?

Oh by the way you mixed the headers up on your 'secret' spreadsheet.

Oh my.... If we account for this incident we actually don't "converge" we drift further apart, the casualty gap becomes even bigger. If you remove human casualties your casualty per tank actually decreases.

Not only have we shown the issues with basic math, we have shown that "time frame casualty data needs adjusting". Yeah no shit. That is what proper researchers do, they control the data and to make sure to have viable samples. Just randomly plucking casualty data for tanks and crewmen like you did and hoping it equals "casualty per tank" is just pointless.

I guess this settles it then. Thanks for supplying your "data" to show how sampling is not done.

[Michael Kenny]

Oh my.... If we account for this incident we actually don't "converge" we drift further apart, the casualty gap becomes even bigger. If you remove human casualties your casualty per tank actually decreases.

Not only have we shown the issues with basic math, we have shown that "time frame casualty data needs adjusting". Yeah no shit. That is what proper researchers do, they control the data and to make sure to have viable samples. Just randomly plucking casualty data for tanks and crewmen like you did and hoping it equals "casualty per tank" is just pointless.

I guess this settles it then. Thanks for supplying your "data" to show how sampling is not done.

It settles the issue that you have no idea what happened .

[Christianmunich]

Oh my.... If we account for this incident we actually don't "converge" we drift further apart, the casualty gap becomes even bigger. If you remove human casualties your casualty per tank actually decreases.

Not only have we shown the issues with basic math, we have shown that "time frame casualty data needs adjusting". Yeah no shit. That is what proper researchers do, they control the data and to make sure to have viable samples. Just randomly plucking casualty data for tanks and crewmen like you did and hoping it equals "casualty per tank" is just pointless.

I guess this settles it then. Thanks for supplying your "data" to show how sampling is not done.

It settles the issue that you have no idea what happened .

I know that if you decrease the dividend you get a smaller quotient.

[Michael Kenny]

I posted the original 5 RTR/12th SS table It to show there was not that much difference in German and Allied tank Unit losses.
I compared 2 more or less equal types of Unit. The fact is as a like-for-like comparison it works pretty well. Both units are roughly equal. That is there was no real difference in the supposed puny Allied tanks to an Uber-Panzer Unit. It appears the Pz IV 75mm was not that much better of a tank/man killer that it made a significant difference to the body count.
Note that in your desperate attempts to try and find fault (any fault) you ended up highlighting the fact that all the excuses you used about tank crew casualties being conflated with whole-Unit casualties apply more to the Allied Unit you picked as your killer fact than to 12th SS!

My point was made and still stands. The Allied tanks do not seem to have been unduly disadvantaged v a comparable German Unit. Despite you bluster about 'Meyer's book' that was sitting on your head and your threat to 'expose' me by posting the 'correct' 12th SS casualties you have achieved nothing but to confirm my argument.
Thank you very much for that

When do you expect you to post the numbers from Meyer? Perhaps the same time as you upload your Top Secret spreadsheet?


And 75% of all hits on a Panther penetrated.

[Christianmunich]

Sorry Kenny your data sample was shot down and proven to be hogwash. Leaving the basic math issues aside, we might want to start focusing on the Briitish sample again.

[Richard Anderson]

The British survey was the most professional collected sample of Sherman casualties, no other sample exists that was better conducted.

No Mr. Troll, I just gave you "professionally collected" samples. Or are you now saying that when Wright and Harkness collected data for their postwar survey they were being professional, but during their wartime survey work they, and the others of No. 2 ORS were not professional?

The British survey is the single best-conducted survey period. nobody said it was the only one. If the studies were as well conducted you might want to show us the photographs with all the tanks et cetera. It also gives the best data. Go ahead Mr Anderson share the other surveys that "rival" this one in data. We will wait.

How do you know it is the "single best-conducted survey" period or comma? You've already proved that in places it is ambiguous enough to confuse you on salient points...not a huge accomplishment mind you. So now you need pretty pictures for it to be the "single best-conducted survey"? Why? You are compiling data that are words and numbers, not pictures, so what do they prove? And yet, again, are you saying that when Harkness and Wright conducted this postwar survey they were at their best and brightest, but not when they and others were doing wartime survey work?

Oh, wait, I thought you've been going on about how postwar surveys aren't as good as wartime ones?

Let me formulate my request differently: Prove it Mr Anderson.

Let me format my response succinctly; Get Stuffed Mr Troll. Meantime though, since you are the one that has asserted the work by Wright and Harkness is the "single best-conducted survey" and I questioned your evaluation, then the burden is on you to prove your statement, I've already demonstrated there were other surveys, some of which the principles, Wright and Harkness, participated in. It is up to you to show why your claim is in fact true, but I hope the proof is something other than "it's got purty pitures in it".

[Christianmunich]

Mr Anderson you again made outlandish claims and when pressed for evidence you responded with an empty post.

Show us the survey that is as detailed as the British sample. A simple request for proof. Prove your claim Mr Anderons, for once just prove your claim. I knew your post like many others was not based in evidence so I directly asked you to prove it and here we are. No proof.