What makes a good poll?
That’s the question Chris Stirewalt and Dana Perino answered in this week’s Data Dive, on I’ll Tell You What.
Stirewalt said that polling averages are better than a single poll because they tend to negate all outliers.
“Like a Slim Jim,” he said, “All kinds of polling ‘meats’ put into one savory, smoked sausage.”
“You can’t trust [automated] “robo” polls that only poll landlines,” he said, crossing out another few of the ten polls on a chart of the most recent presidential ‘horse-race’ polls between Hillary Clinton and Donald Trump.
“One-third voters are ‘cell phone only’,” he explained
He crossed out the internet polls, like Survey Monkey, because “internet access remains difficult for poor voters and older voters.”
“And, people don’t feel comfortable with online polls.”
Stirewalt said that “only human callers to landlines and cell phones can capture” the most accurate polling figures today.
But he noted that Caller ID makes it difficult for even that system to be the most flawless, as many people won’t pick up for a number they don’t recognize or suspect to be ‘spam’.
Watch the full clip above to find out which poll Stirewalt found most accurate by his “Slim Jim” model and let us know what you think in the comment.