I did some analysis in a spreadsheet for my own amusement and information and thought I should share the conclusions in case of interest or use to anyone else:
Assuming no Die Roll Modifiers (DRMs), and taking account of the distribution of results with 2d6, the average number of hits (not losses) per roll for the Combat Differentials (CD) are:
+4 = 1.667
+3 = 1.444
+2 = 1.25
+1 = 1.083
0 = 0.889
-1 = 0.694
-2 = 0.5
-3 = 0.361
-4 = 0.194
Due to the reciprocal nature between the attacker's CD and the defender's, each +1 shift in CD results in an average differential in hits increasing by approx. 0.4. Example- at 0 CD, both sides on average suffer the same number (0.889) casualties. If one side instead had an additional +1 shift, its CD would increase by 1 while the opponent's CD would decrease by 1 for the same reason. This would increase the attacker's average result to 1.083 hits inflicted on the opponent, and decrease the defender's result to 0.694. The net change is that the defender suffers 1.083 - 0.694 = 0.389 more hits than the attacker as result of the additional shift.
Looking next at DRMs, and averaging the results in all CD columns (still weighted for the distribution of results of 2d6), the average number of hits per roll are-
1.608 with +2 DRM
1.225 with +1 DRM
0.898 with no DRM
0.623 with -1 DRM
0.407 with -2 DRM
...which is roughly a change in number of hits of one third for each increment of DRM.
All of the above analysis is based on average results of 2d6, one glance at the CRT will make it apparent that the variation in results from 2d6 can have a greater effect on the number of hits than the effect from CDs and DRMs.
If you go to the website below you'll find the Excel spreadsheet I use to calculate hits, losses and kill ratios for air combat.
Good to see that my calculations appear to be consistent with that.