Full disclosure: While I admit I am leaning Bernie, I am very supportive of Clinton as well.
Anyway, there is a diary up implying that another person's diary was dishonest. They suggest that the original diary's poll interpretations were dishonest because they included Democrats, Independents, and Republicans and showed a near tie with Clinton at 33% and Sanders at 30%. They then used a likely voter screen to suggest a much larger Clinton lead.
However, this was wrong. The original diary does not include Republicans. As you can see, the data is exactly the same if Republicans are filtered out. This is likely because Republicans were not polled for the Democratic Primary for obvious reasons.
This means that the Clinton 33% to Sanders 30% includes all Dems or Indies. Independent voters are eligable to vote in the Democratic primaries in many states, but not in others. Regardless, it is not dishonest to include them as it could be argued that it is a more accurate reflection of the voting populace.
When the likely voter screen is applied, the number of respondents drops from 84% to 22%. Nearly 75% of respondents are excluded using this method. On top of that, 5% of likely voters say they wouldn't vote. This is why many polls do not use likely voter screens this far from the election. All voters or registered voters may be a more valid reflection of the voting populace.
If we only look at registered Dems and Indies, Clinton is at 35.5% and Sanders is at 27.4%. Regardless, they all show a sharp increase in Sander's numbers and a decline in Clinton's since 9/17. It should be noted this is after Sander's speech at Liberty University.