Today's live phone Quinnipiac poll (Sanders 49%; Clinton 46%). Do not be fooled by robo-call polling

When people being polled are not talking to a real person, they say they support Trump or Clinton in higher numbers than people report when they are giving a live response to a real person or being polled according to other accepted polling methods.

Here is the aggregation of all robo-call polls in Iowa (including today's):



Looks like Clinton has an 8% lead.

Here is an aggregation of all the accepted methodology Iowa polls where data was gathered by live landline and cell phone polling (e.g., Quinnipiac, Des Moines Register/Bloomberg/Selzer, NBC/WSJ/Marist, Monmouth University, ARG, Iowa State/WHO-HD, FOX, CNN, Loras College, Mason-Dixon/AARP, NBC/Marist, Suffolk/USA Today, Suffolk) UPDATED TO INCLUDE TODAY'S QUINNIPAC POLL:



Looks like a dead heat with Sanders coming into the caucus with a lot of momentum.

I'm not a huge fan of internet-based polling because it is not as reliable or as well accepted as live landline and cell phone polling, but it is much more reliable than robo-call polling. Here is an aggregation of the internet based Iowa polls (e.g., CBS/YouGov, Morning Consult):



This internet polling confirms the live landline and cell phone polling and refutes the robo-call polling.

Don't be confused by the junk polls out there, the Iowa caucuses are tied.

Also, bear in mind that attempting to poll a primary is much harder than polling a general election, and attempting to poll a caucus is much harder than polling a primary. A poll is a poll -- the caucus is the vote. The aggregate of polling in Iowa in 2008 understated Obama's support by 7% and understated Edwards support by 4% (but pretty much nailed the ceiling for third-place Clinton's support). Polls provide a general picture of the movement in the race, but getting out the vote is what provides delegates.

This is the sprint to the finish. Let's finish strong!