Why “likely voter” matters

Reuters (Drudge) has a 5 day rolling average tracking poll out now showing Trump with a 2% lead, if you filter by likely voters.

Reuters poll (as filtered by Drudge)

If you turn off the likely voter filter and filter only to registered voters, it switches to +5 for Hillary (41-36%) and if you filter to everyone, including people not registered to vote, it goes to +6 for Hillary (39-33%)

Looking at respondents, 83% say they plan to vote, which is a lie. Turnout in 2012 was 58% of the Voting Eligible Population. The key thing that influences the predictability of polls is how they determine “likely voter”. Asking “do you plan to vote in November?” is worthless. “Are you a Republican or Democrat?” is almost as worthless. The private polling looks at the actual voting history of an individual.

This entry was posted in 2016 election. Bookmark the permalink.

3 Responses to Why “likely voter” matters

  1. JayMar says:

    I don’t trust polls, never have. Every single time I am polled I will give them the wrong answer. Hell, I even registered as a Democrat back in 2012 when in reality there is no one to my right. I remember telling some that I was an evangelizer for Saul Alinsky Ministries.

    • Fred Stiening says:

      It would be interesting to add marital status and parenthood to the classification information. Women in the South strongly prefer Trump. It’s the New England women who like Hillary. My hunch is there is a big disparity between single/divorced women and married mothers. How you decide what variables to use for sample weighting has a fantastically huge impact on what you actually measure. The tracking polls may be more accurate as they can fix the sample once and avoid having to weight the responses. The poll cited above has 57% of the population as women, which is well above the average for the voting age population. At 65+, women are 58% to men 42%,,but that’s only 12% of the entire population.

  2. Fred Stiening says:

    The CNN poll just released shows Trump with a +2% lead of likely voters. The spread of likely voters versus registered voters is pretty consistently 5% in favor of Trump. I’m going to apply that number to all registered voter polls.

    http://i2.cdn.turner.com/cnn/2016/images/09/05/rel13a.-.2016.post-labor.day.pdf

    Hillary’s 4 minute long coughing attack in Cleveland is a big problem, independent of her politics. The worst thing that could happen to Trump is a Hillary withdrawal. I think it is unlikely Sanders would replace her (the DNC is not required to pick Kaine), but you never know.

Leave a Reply