From Unlikely Voter: I’ve seen a few Republicans express serious doubts about Carly Fiorina after the latest California Senate poll from Public Policy Polling, but I think close inspection of that poll should give one pause before putting too much weight on its results.

Besides, the other new poll, from the Public Policy Institute of California, deep down is as bad for Barbara Boxer as the Republicans could ever hope for.

Public Policy Institute of California and Public Policy Polling might sound related, but they’re not, and their polls show it. Let’s look at the PPP poll first. Tossing aside the question about hairstyles, we get a key top line result of Fiorina 40, Boxer 49 (MoE 3.95). This represents a 6 point swing toward Boxer from PPP’s last poll in May. But I think the picture of the electorate it depicts is terribly unrealistic.

First there’s the issue of support for the PPACA. Previous polling by Rasmussen Reports showed a narrow split in favor of the bill, but PPP shows a wide 52-40 margin in favor. That was a key figure I saw that caused me to dig further into the PPP data.

The PPP poll is of registered voters. The firm is apparently doing nothing to weed out people who probably won’t show up, especially the Greenberg Quinlan Rosner “drop off” 2008 voters who will stay home in 2010. This is a problem because a) midterms have lower turnout than Presidential elections, and b) in a partisan wave like 2006 or the likely one in 2010, the desire to stay home will not be evenly felt by Republicans and Democrats.

This is why the PPP poll shows a split like 58% voting for Obama, 36% voting for McCain, 6% other/not voting, which almost exactly resembles the 61/37/2 split of the final 2008 Presidential election count in California. The PPP poll might be accurate if the state sent people to the homes of every 2008 voter and conducted the Senate election by hand, but that’s just not the case.

California’s next Senator will be determined by those who show up at the polls or take the time to fill out, stamp, and mail their ballots in. And I believe no serious analysis will project the 2010 electorate only to show a 2 point swing from the Democrats to the Republicans, as PPP’s poll projects. No, that swing will be higher.

That’s why I put more trust in the PPIC poll. It also shows Boxer leading, but by a 39-34 margin (MoE 2.7). This is a poll of PPIC’s model of likely voters in November. Not only is Boxer’s support remarkably weak for an incumbent, dropping below 40, let alone 50, but the 5 point edge is also a drop from before the primaries.

In May PPIC had Boxer ahead of Fiorina 48-39, so Boxer’s support has fallen badly. So has Fiorina’s but by half as much.

Any challenger would take her chances against an incumbent as weak as Barbara Boxer, I believe. Boxer leads, but Fiorina is in great position to win, especially should she make another late surge as she did in the primary.