This is quit an extensive write-up, but for what? Im no statistician, but this reminds me of the phrase garbage in garbage out.
Please tell me how in a country about evenly divided between Republicans and Democrats, a poll that included 374 Republicans and 300 Democrats is valid.
This is a 20% over-sampling! If this were reversed, and Democrats were over-sampled, would you take this seriously?
Okay, okay now that I actually read your analysis, I see that youre just as critical as I of the over-sampling and are adjusting the results.