In my earlier post about that Texas Lyceum poll, I wondered about the partisan ratio that the pollster used. I see now that Harvey Kronberg is on the case.
In response to our query, Lyceum pollster Daron R. Shaw sent us the following explanation:
"We use a likely voter screen consisting of the following:
1. self-reported registration
2. interest in the election (somewhat or extremely)
3. voting behavior over past two years (almost all or every election).
"This gets the 1,000 person sample down to 478. The party breakdown for the LV population is 42%R-42%D. Democrats are more prominent in this estimated electorate than they have been in recent years due to their relatively high levels of interest and self-reported recent voting. We could have weighted this to a 6-8 point Republican advantage, but judged it best to simply report the numbers and be clear about the underlying party breakdown. Do we believe Democrats will maintain their "enthusiasm" edge over the Republicans and make these elections close in the fall? History is against them. But given recent trends nationally and the engagement attitudes expressed in the poll, it seems imprudent to weight the data back to expected party turnout.
"Let me be clear; the internals show little to no cross-over voting. Obama (and Noriega) are not getting Republicans. They have locked down almost all Democrats, though. This is a snap-shot whose predictive value is highly contingent. Democratic candidates need to have a huge Democratic turnout, along with lukewarm Republican turnout, to get the numbers we show here on Election Day."