The Drug Update

That article doesn't say anything about inconsistencies in any poll. Instead, it reports on a deal to recruit poll subjects from diverse and underrepresented segments. This is only a problem if a substantial number of those polled in the new Zogby poll were MPP-referred. Like Zogby said, in this 2002 article, 1,000 from MPP had signed up and they had 100,000 on their rolls then, which might still have some overrepresentation of conservatives. Assume that 5,000 of MPP members eventually signed up and the overall pool numbers didn't increase. Which means that if the telephone polling is truly random, 50 of the 1004 polled will have been MPP recruits i.e. 5%. This would be the maximum deviation possible. More realistically, if you replaced these 5% with a randomized group, you would expect to still get 40% of these affirming support for legalization, since that's the support among the 95% who aren't replaced, which leads to 43% overall. The general range of 40-45% is supported by the legalization ballot results in Nevada in 2002 (39.xx%) and Alaska in 2004 (44.25%), both not so liberal states.

I think you should add a cavaet to your recent post about Zogby/MPP, mainly that MPP has now, for a few years, supplied thousands of their members to Zogby Intl in exchange for new questions about marijuana legalization. It seems that the poll could be biased because its recruiting methods seem all off. It does not seem to be a representative sample of the American people. Check out the Wash Post article here: http://www.washingtonpost.com/ac2/wp-dyn/A49047-2002Jun2?language=printer