FINDING RELIABLE POLITICAL POLLS

INTRODUCTION

My quest for reliable polls began with the presidential polls used for the 2016 presidential election.  With only two exceptions, they were completely wrong.  If you knew those two exceptions in advance, you could have predicted the outcome of the election.  I did so and therefore was 75% certain that Donald J. Trump would become our next President.  It was really simple:  I looked at the polls that accurately predicted the previous (2012) presidential election.  There were only two, one of which polled daily, the Rasmussen Daily Presidential Tracking Poll.  Therefore, I only believed the Rasmussen poll.

BODY

I initially thought that the reason for the poor poll results was simply the biased media funding fake polls in order to create a bandwagon effect (“push polls”) for their politician.  However, I decided to examine polls and find out why they differed so much.  The following is what I learned:

a) The design of a poll can easily create any outcome the pollster wants.  For example, the pollster determines what percentage of Democrats, Republicans, and Independents to use in his or her polls.  These percentages largely determine the results of a poll.  Most of the recent polls used 24% of Republicans although the 2016 and 2018 election exit polls showed 33% of Republicans voted…which explains why the polls were so wrong.

b) Whether a poll is of: a) all adults, b) registered voters, or c) likely voters,  is the next most important factor in determining the results of a poll.  Only half of all adults vote, 80% of registered voters, and over 90% of likely voters vote.  The most accurate poll of Presidential elections has been the Rasmussen daily presidential tracking poll which polls “likely” voters.

c) The way in which poll questions are worded can significantly influence the results of the poll.  For example, a question can be worded: 1) do you approve of the President? OR 2) do you approve of the President’s job performance? OR 3) do you like the President? OR 4) etc., etc., etc.  Often poll questions are deliberately ambiguous and therefore the results can’t be trusted.

d) How many people you poll is very important in order to get a representative sample.  Nothing under a 1000-person-poll is very reliable.

e) How you select the people to be polled is very important.  For example, only selecting people from the inner city would result in an overwhelming number of people favoring a Democrat and would not be a representative sample.  Same is true for a poll conducted of New Yorkers only.

f) Who is sponsoring the poll is extremely important.  A poll funded by the New York Times, Washington Post, or CNN will not depict a Republican President favorably because those newspapers and that TV station have a partisan agenda and are therefore unreliable.

CONCLUSION

I’ve attempted to show how polls can be manipulated to shape public opinion rather that guage it.  I believe that most political polls are seriously and deliberately flawed and therefore I don’t believe them to be reliable.  Past performance is the best way to judge a poll.  For example, the Rasmussen Daily Presidential Tracking Poll I used to believe to be the best of all of the Presidential polls, based on its past performance in both the 2012 and 2016 presidential elections.  Recently, the Trafalgar polls have been very accurate. Looking at past performance, therefore, is the quickest and easiest way to determine how accurate or misleading a poll is.

Print Friendly, PDF & Email

Leave a Reply

Your email address will not be published. Required fields are marked *