Why Current Polls are so Inaccurate Gallup stopped doing polls because they got it wrong in the last election cycle and they felt they could no longer accurately predict elections. I have studied survey and conducted a poll for a teachers' union while I was in grad school (Ph.D. circa 2013). Respondents are typically called on a land line - this is hugely problematic since land lines are now an artifact - used by old white people with some money, less so by Latinos and poor people who use cells phones. It is illegal to call a cell phone using auto dial - but you CAN use voter data supplied by states. So a state poll can include cells but not a national poll, if that makes sense. Another problem with the current model for polling is they want "likely Democratic voters" which excludes anyone who didn't vote in the Democratic primary/general in the last 8 years. That leaves out Independents, young voters, all new voters, and a large segment of the Bernie voting population. Pollsters have not fixed that . . . yet. The good news is that any poll that shows Bernie winning is probably correct.
ANALYSIS OF WHY GROUPS DO/DO NOT VOTE FOR BERNIE SANDERS I have a theory that we are misidentifying the cause of people voting/not voting for Bernie. 1) Many people need only to learn more about Bernie's positions and life-long commitment to give him serious consideration 2) There is a media blackout on him 3) People who are connected to alternative communication (FB, Twitter, Instagram, etc.) are more likely to vote for him . . . which would explain the correlation between age and voter preference across all groups
Data Reports and Analysis
3/22/16 Survey data was published 3/21/16 Sources: