Wildlife managers in many countries estimate cripple rates by asking shooters! They hand out survey forms asking shooters how many birds they downed but didn't retrieve. This is a bit like estimating the incidence of wife bashing by asking husbands if they beat their wives.
A 1987 Canadian study by Nieman gives you some idea of how bad this method This study was done in 3 regions of Canada and dealt with cripple rates (birds downed but not retrieved).
Cripples were first estimated by questioning shooters --- this is a common approach to cripple estimation in Australia also. But, in addition, researchers also watched those same shooters from concealed hides and made their own cripple counts. The observers (Canadian Wildlife Service staff) counted between 5 and 8 cripples for every 10 birds bagged, but shooters admitted to around 2 cripples for every 10 birds bagged. Either shooters have appalling memories or they lie.
The result are shown below. The X, Y, and Z in the figure are 3 regions of Canada in which the study was done. Note also that I have, in the above description, translated the study's "cripple rate" figures into cripples per bagged bird. When they talk of a 50% cripple rate, they mean 1 bird crippled for each bird bagged. Note also that my wounding rates include birds hit but not downed so will always be greater than or equal to cripple rates.
