Remember that time in Probability and Statistics class where you learned that as long as polls took a diverse enough and large enough sample size, they could be sure (within a certain margin of error) that the results would accurately reflect the real outcome? Me too… That’s why as I watch the results for each state unfold in the Democratic Primary season I am astonished by just how out of wack the polls have been with reality. Let’s look at the states in which data was available from RealClearPolitics:
Some of these results are in fact pretty darn close to the latest poll results but many of them are so far off you can basically disregard the poll results as complete garbage…
Look at Colorado, Minnesota, Kansas, Michigan and the latest 3 (Idaho, Utah, Alaska). The difference between the spread in the latest poll and the margin of victory for the real results was over 25 in each of these states! More than half were over 50 points off what the polls predicted!
How can it be that a poll conducted in Minnesota using a statistical analysis method which claims to have a Margin of Error of 5.7% be so far off the mark?
One big reason is the Independent Voter. As of January 7th, our nation had the highest percentage of voters registered as Independent (43%) since 1988. These voters are not being accounted for when these polls are conducted but they are coming out to vote.
States in which the primaries are open (such as Michigan) make it easy for the Independent voters to have their voices heard in the primaries. But even in states with closed primaries, we are seeing many independents register as Democrat to show their support for their candidate of choice. In Pennsylvania 52,177 have switched from Independent and other parties to Democrat in preparation for the upcoming primary.