It's not that much different when they are shooting at you, IF YOU HAVE BEEN TRAINED RIGHT.
It's absolutely true that realistic training can make a profound difference in combat effectiveness. This is the main reason that professional armies tend to do so well in combat compared to conscript armies or untrained rabble. Anyone who doubts this should review the accounts of small unit combat in Iraq. Professional Coalition soldiers routinely kill many times as many insurgents as they lose. And this is usually in urban combat, which has historically tended to overwhelmingly favor the defender. The tremendous disparity in kill ratios exist even if you eliminate enemy casualties killed by airstrikes and artillery.
However, I haven't found anything in my studies to support the idea that training (or by extension, combat experience) can effectively
eliminate the effect of being shot at for real. The historical record shows no such phenomenon. For instance, the US Army did comprehensive studies after WWII and Korea and found that combat veterans reported being terrified in combat and that this did not decrease with additional exposure to combat. This came as a surprise, btw, as the army expected that repeated exposure to combat would make soldiers less "gun shy". The reverse seemed to happen after a point; soldiers began to lose effectiveness after a certain point.
My combat veteran friends and relatives (Korea; Vietnam) are unanimous that being shot at has a serious effect on one's behavior and accuracy in combat. (FYI--I had to study this topic extensively for my miniature wargame rules
A Fistful of TOWs, which has the thesis that troop quality--primarily training--is the single most important factor on the battlefield).
Still, 1 in 20 guys never fires in any given combat. (USMC, 2003)
S.L.A. Marshall claimed that 3/4 of troops never fired their weapons in combat in WWII. This claim was hotly disputed by the troops (including relatives of mine, who I personally found highly credible) at the time. But curiously, the Army accepted his data and used it to justify adoption of the assault rifle. After Marshall's death a number of historians have questioned Marshall's claim and noted that he produced little verifiable evidence to back his claims up. My own review of his work produced deep skepticism. FWIW, I think that the Army accepted his claims because it allowed them to argue for a new, expensive weapon system that Congress might not have otherwise funded.
So I tend to be skeptical of such claims.
That said, I'd expect that the real ratio -- particularly among professional soldiers -- is closer to the number you quoted than to Marshall's numbers.