I looked into this in some detail for
Flying Guns – World War 2: Development of Aircraft Guns, Ammunition and Installations 1933-45. These are some relevant extracts:
The fighter-bomber pilots pressed home their attacks with great courage throughout the campaign despite the often ferocious light FlaK which caused loss rates far above those experienced by fighter units (one Typhoon squadron suffered 100% casualties in an eighteen-month period). They were confident that any German tank they spotted was as good as dead, and they earned a considerable reputation for tank killing, with substantial claims being accepted. However, British operational research (OR) carried out at the time (but not publicised for obvious reasons) presented a more complex picture.
As the Allies were advancing, intelligence officers were often able to examine a battlefield shortly after an air attack, and what they discovered causes controversy even today. (Much of this section is taken from Ian Gooderson's "Air Power at the Battlefront", which explores this issue in great detail).
The evidence gathered by the OR teams indicated that very few tanks were destroyed by air attack. A British War Office analysis of 223 Panther tanks destroyed in 1944 revealed that only fourteen resulted from air attack (eleven to RPs and three to aircraft cannon). During the Mortain battle of 7-10 August, the RAF and USAAF launched sustained attacks on a German armoured column over a period of six hours, claiming 252 German tanks destroyed or damaged in nearly 500 sorties. It was subsequently discovered that there had only been a total of 177 tanks or tank destroyers deployed by the Germans and just 46 of those were lost, of which only nine could be attributed to air attack (seven to RPs and two to bombs). During the German retreat from the Falaise pocket later in August, the RAF and USAAF claimed 391 armoured vehicles destroyed. Shortly afterwards, the battlefield was examined and only 133 armoured vehicles of all types were found, of which just 33 had been the victim of any sort of air attack. In the retreat to the Seine, large numbers of armoured vehicles were left behind and Typhoon pilots alone claimed 222 destroyed, but only thirteen out of 388 AFVs examined were found to have been knocked out by RP attack. In the Ardennes salient, just seven out of 101 knocked-out AFVs were definitely or possibly attributed to air attack, compared with claims for 90. It should be noted that in the prevailing circumstances of a continuing retreat, there was no question of the German Army having recovered any damaged tanks in these later actions, in fact the battlefields were often littered with undamaged tanks abandoned by their crews.
One source estimates that probably no more than about 100 tanks were lost due to hits from air weapons during the entire Normandy campaign. In contrast, the RAF's 2nd TAF (including elements of the Air Defence of Britain which took part in the campaign) and the USAAF's 9th Air Force lost over 1,700 aircraft between them.
And on the subject of the accuracy of bombs and rockets:
In contrast, one direct hit with a bomb or 60 lb RP meant certain destruction for the heaviest tank. However, their accuracy left a lot to be desired. Even under practice conditions, the hit rate for the RPs against tanks was no better than 5%. This was graphically illustrated by a demonstration put on by Typhoons against a captured Panther tank placed in the middle of an open field, helpfully painted white with large red crosses on it to make sure the pilots could see it. Of the 64 RPs fired (launched in a typical steep dive at ranges of 750-900 m), only three hit the tank. In battle, RP accuracy was considerably worse than this, with the official British calculation of hit probability against a single tank being 0.5% (in other words, 200 RPs had to be fired for each hit). Furthermore, some 20 – 30% of RP warheads failed to explode.
And:
Bombs were even less accurate than RPs. An analysis of Typhoon bombing attacks revealed that only 50% of bombs dropped landed within 120 m of the target, with an average radial error of 144 m. The absolute minimum safety distance from friendly troops for P-47s dropping bombs was about 250 m. The probability of dropping a bomb close enough to a tank to disable it was extremely small. Napalm bombs were used later to some effect, but were an even greater hazard to any friendly troops in the area.