Reply #46: Surveyed = Dem primary voters. HUGE margin of error in this poll.

[spooky3]

It appears that CBS polled a TOTAL of 1125 voters, then had to separate them into Repub. primary voters, Dem. primary voters, and independents and others. Interestingly, nearly all the items reported were for the subsamples, but CBS reported MOE only for the full sample, and the biggest influence on MOE in a random sample is the size of the sample.

So, we have to guess about the MOE for the numbers they and you in the OP reported.

Even if I made the most generous assumption that everyone leaning Democratic was considered a "Dem. primary voter" (and that would include a lot of people who are Dems. but not very interested in politics) the sample could not have been more than 350-400 voters. With a sample size that small, the MOE is probably in 6 percentage point range or higher. Further, because they had to oversample African Americans in order to get a reasonable sample size, then weight it afterward, those 350-400 voters represent an undersampling of non-African Americans voters. That wouldn't be much of a problem in a larger sample, but in a small one, it could make a big difference, since those non-African Americans may not have been very representative of the larger group of non-African Americans.