This Michael Barone item, via Real Clear Politics, is worth checking out. In particular, Barone kinda suggests Texas may be a better state for Obama than Ohio. The argument seems to be that downscale Southern whites are generally more receptive to Obama than downscale Rust-belt whites, and that African Americans and Latinos will account for roughly the same share of the primary electorate (which would be a net advantage for Obama, since he'll almost certainly win bigger among African Americans than Hillary will among Latinos).
Some of us have been surprised by the recent polling showing Obama doing reasonably well in Texas--and, to be fair, it's unlikely that Obama is actually up there, as this poll alleges. But Barone suggests the polling is generally on target.
And don't forget that a significant number of Texas delegates will be apportioned through a caucus (about a quarter third, I think), and that heavily Hispanic areas in Texas tend to have fewer delegates per capita because they haven't turned out heavily for Democrats in recent years. Even if Hillary wins the popular vote, she could come up on the wrong end of the delegate math there.
Update: Thanks to commenter babbfish for noting the correct percentage of delegates chosen by caucus. Looks like there are 193 pledged delegates up for grabs; 126 apportioned in the primary, 67 by caucus. (The piece I linked to in the last graf is a little off on that, but the point it makes about heavily Hispanic areas is correct, I think.)
--Noam Scheiber