JLOGIC, on 2012-November-16, 14:56, said:
This logic feels intuitive, but is actually wrong. Suppose the missing pips are the 432, and on the first round lho plays the 4. The relevant holdings are lho having 43, 42, T2 initially, but from either of the first two he would have played the 4 only half the time, so those two are exactly half as likely as the Tx, and its fifty fifty.
In essence, although you are right that restricted choice doesn't apply at trick two, it does apply at trick one, as lho's play is forced from Tx but not from xx, and rho's play of the K is always forced.
Again, the fool proof way of looking at these is to look at the a-priori holdings. Given that lho has a doublton and rho the K, there are 6 relevant holdings, Tx-Kxx three times, and xx-KTx three times, the card lho plays at trick one does not affect the a priori holdings, so it remains 50-50.