Graham is wrong headed or is out there lying.
According to a highly touted study by Rush, Hannity, et al, that even with 70% of the Hispanic vote, Romney would have still lost. The Hispanic vote made up only about 7% of the 2012 electorate.
So 70% of 7% is little of little.
Disagree with that analysis, just mathematically speaking. If as you say, the Hispanic vote was 7% of the electorate (I've heard 8 or 9% mentioned elsewhere), if Romney got 70% of that 7%, that would give him 4.9 percentage points of the national popular vote from Hispanics. In reality, he supposedly got something like 27% of the Hispanic vote, or about 2 percent of the national popular vote from Hispanics. That would mean that if Romney got 70% of the Hispanic vote, he'd be getting 3 more percentage points of the national popular vote, and by the same token, Obama would have gotten approximately 3 percentage points less of the total national vote. So it would be net swing of six points in Romney's favor, more than enough to push him past Obama in the national popular vote. The effect of these shifts on the electoral vote can't be determined.
Of course, we can't factor in the massive fraud and cheating in this analysis, disguised by pro-Obama commentators as "a good ground game" or a superb job in getting out the vote. The view here is that cheating and fraud, targeted to swing states, probably made the difference between Obama "winning" and losing.