One thing to consider is that Republican affiliation is a minority opinion, so it makes sense that they would be outnumbered in a random sampling, though I would expect 2:1 rather than 3:1.
In a truly random U.S. sample you would expect Republican identifiers to be nearly as numerous as Democratic identifiers.
Not sure where you got your numbers from.