The abstract of the USAF take-off and landing ejections, 1973-85. gives an answer to two of your questions, leaving room for speculation as for the impact of 0-0 ejection seats (pun intended).
Direct quote, emphasis mine:
"...This review included only ejections clearly requiring a decision between ground egress and ejection. There were 15 aircraft with 25 crewmembers identified; 22 of them had ejected. Ejection systems performed as designed 91% of the time.
Three crewmembers were killed during ejection, yielding a survival rate of 86%. Four ejectees suffered spinal compression fractures from ejection force. Two of these also fractured other vertebrae during the parachute landing fall, for a major injury rate of 21%. In only 33% was ground egress probable. Survival and injury rates for ejection on the ground did not differ significantly from those above 500 ft (p less than 0.05). Ejection during take-off and landing phases is as safe as ejection above 500 ft. safer than other ejections below 500 ft, and does not result in excess injury rates. Ejection systems are sometimes damaged by impact or fire. In the emergencies considered, ejection offered greatly increased chances for survival over ground egress."
So, the statistics given by this study apply for ejection during take-off and landing phase and ejections above 500ft.
As for the trends after the introduction of 0-0 ejection seats, a non scientific concusion could be drawn comparing statistics produced by the study above, and the one presented by @Pilothead in his answer. In the 1973-85 study the survival rate was 86%. In th 1952-1997 study the low level ejection survival rate was only 51.2% The higher survival rate of the 1973-85 study can partially be explained by the fact that some of the ejections in the 1952-1997 study were performed with a non 0-0 seat, since they were developed during the 1960's. The 1973-85 study does, however state that "Ejection during take-off ...[is]... safer than other ejections below 500 ft.", so a fully conclusive deduction cannot be made.
Unfortunately no studies on this subject are fully accessible to general public, so answers provided by @Pilothead and me are pretty much as good as it gets. :(