While I agree that falling standarts, increased altitudes and many more aspects may be traced down in these numbers I would like to stress that without knowledge of the conditions and targets (B-17 and P-47 may be harder to kill than Whitley and Battles?) one has to be very careful not to draw Scheinkorrelations from the database as a result of interpretation.
From my experiences with meaningful statistic analyses, an appearing statistical relationship between aspect A and B with a high degree of significancy may be related to the influence of another, yet unrevealed aspect C. The established relationship between A and B is in this case virtual and not high (or from a quantitative point of view, the relationship is less pronounced). This is a common way to mess up descriptive statistics when uncritically applied to a databse. A multidimensional factor analysis (Varimax-Kaiser rotation preferred) with screeplot may be used in order to exclude Scheinkorrelations.
Replace the 88mm FLAK by US 90mm guns -do You reasonably expect better or worser results everything else (environmental conditions, ammo, targets, altitudes, firecontroll, crew quality and logistics) beeing equal?
cant comment on your analysis techniques when we dont have good data sets of the scale that you would require. all we have are bits and pieces.
In relation to your suggestion that the target type might have a better or worse effect, as the war progressed, i would agree that there might well be variables at work that affect the outcome. you suggest that it might be harder to hit a P-47 or a B-17 compared to say a Battle or a Whitley.
I would say, that without a doubt it would, but this kind of approach still does not explain the effect. In 1941, with a genenerally less capable force attacking them (Stirlings Halifaxes, Manchesters, Wellingtons, the rounds per kill was twice that in 1942.
In 1942, the force structure attacking germany changed slightly, the main attackers became Halifaxes, Wellingtons, Lancasters, Stirlings, and a few Mosquitoes. Whilst therre was a small change in the attack force, there was a massive increase ihn the AA effectiveness. Now, as you say, there could be any number of reasons for that, and it is very easy then to try and mask the obvious by arguing that we need to look at each and every one of them. i agree that is the most thorough way of doing the analysis, but it also defeatsw the reason for using statistics in the first place. The use of statistics is primarily a crude method of finding the most likley causes, and measuring their effects.
So, using the statistics as a guide, we know that the firing accuracy of the Germans reached its peak in 1942. There was a small change to the bombing force. There was virtually no change to the defending force...the numbers did not change significantly, the crews did not change substantially, even the ordinance being used did not change all that much. in my opnion ther were two substantive changes the use of the 8 gubn battery, and the introduction of radar guided AA.
In 1943, for the first half at least, there was no wholesale Use of US bombers. The overwhelming majority of AA losses continued to be British. The force structure continued to be very similar as it was in 1942....Lancasters, halifaxes, wellingtons. The Stirlings were beginning to fade, but conversely, the Mnosquito was beginning to be a significant player, with about 10% of Bomber Command being Mosquitos. However, i would argue that the change in the force structure of the attacker, for the first half of 1943, at least, reamined only a minor adjustment to the equation.
Similarly, the hardware being used by the defenders did not yet change all that much, although I concede that some captured equipment was starting to be used. The allied countermesures (principally window) was not yet being felt. yet despite this, the effectiveness of the defences were already decreasing....what do you suggest. I believe that there is a possible twofold explanation....the aforementioned drop in the quality of the guns (ie the captured equipment), and more importantly the starting of introduction of crews of indifferent quality . However, if you have another explanation I am happy to listen.
The latter half of 1943, and 1944, saw the introduction of window (and other countermeasures) the entry of the Americans, the rapid increase in ammunition expenditures (and later in the year acute ammunition shortages) , the greater replacement of good crews with indifferent ones, the greater reliance on captured equipment, the introduction of better and higher flying attackers. So your right for 1944, so many variables are on the move that it is difficult to know which one(s) are the dominant factor in the collapse of Flak effectiveness
With regard to your last point about US 90mm guns, I would have to view the issue more carefully than I have, but the generally accepted wisdom is that allied AA was improving as the war progressed, not decreasing as was apparently the case in Germany. I have read a detailed study from the USN (but cannot find it right now), that suggests that they were expending an average of 1500 rounds per kill at the biginning of 1944, but by the end of the year, this had dropped to just 550. This could be misleading, wrong: it certainly is not comparable to the battle problem facing the german gunners