The real problem I have with the study is that the numbers are absurdly unrealistic. 10 murders prevented? I might be able to accept the idea that every two executions prevent one murder or some similarly modest number, but 10 murders for every execution not performed is absolutely absurd. That would suggest that states without the death penalty should have enormous murder rates--and that simply isn't the case; in fact, they tend to be below average.
It sounds like a pseudo-scientific study that found what it expected to find because it expected to find it.
Correlation of variables A and B can mean three things:
1) A causes B,
2) B causes A, or
3) something causes both of them.
OK, so we have a correlation between high murder rates and executions for murder. Our options:
1) high murder rate causes death penalty
2) death penalty causes high murder rate
3) something causes both the death penalty and the high murder rate
This study (once properly peer-reviewed and followed up on) basically says that option 2) is flat wrong. Indeed, if option 1) is the case then states without the death penalty won't have high murder rates (if they did, they'd have the death penalty).
That leaves option 3) as your only leg to stand on. But if the underlying cause to the death penalty is not present, then an underlying cause of high murder rates is not present. Again, you don't have a leg to stand on.