Online College Courses for Credit

Plus or Minus What?  Explaining Margin of Error

Plus or Minus What? Explaining Margin of Error

Author: Marble Happy

Review terms and objectives

Explain interpretation of margin of error in the news media

Find examples and explain the significance of margin of error

Margin of error is thrown around in the news media when discussing polls of all types. Most of us don't really register the meaning of margin of error and might be misled. The purpose of the packet is to define, explain and give examples of uses of the margin of error in current news sources.

See More

Why do I need to know about Margin of Error?

Read through the following article.  Did you find the margin of error?  So what does it tell you about the poll on the Governor's race in Minnesota?  Margin of error is a term frequently tossed about by the media but seldom explained.  Not knowing the purpose of margin of error can lead to confusion when trying to wade through the mire of political reporting as well as polling on many other topics...


Poll: Minnesota governor race is neck and neck

Ryan JohnsonMcClatchy - Tribune Business News. Washington: May 26, 2010.

May 26--New poll results released Wednesday show the Minnesota gubernatorial race is pretty much a dead heat -- no matter which Democrat wins the party's Aug. 10 primary.

A telephone survey of 500 likely Minnesota voters, conducted May 24 by Rasmussen Reports, gives Republican endorsee and state Rep. Tom Emmer a slight edge over the three Democratic hopefuls.

Margaret Anderson Kelliher, speaker of the Minnesota House and the DFL-endorsed candidate for governor, came in just behind Emmer, 36 percent to 38 percent. But that's within the poll's 4.5 percentage point margin of error.

Independence Party endorsee Tom Horner, a Twin Cities public relations executive and former Republican, got 11 percent of respondents' support in the Emmer-Kelliher matchup; 15 percent said they were undecided.

Emmer had a two-point lead, 37 percent to 35 percent, over Democratic gubernatorial hopeful and former U.S. Sen. Mark Dayton. Horner picked up 12 percent in that hypothetical race, while 16 percent were unsure who they would vote for.

Emmer's biggest lead in the latest Rasmussen Reports poll was in a matchup with Matt Entenza, a Minnesota lawyer and former six-term member of the Minnesota House. But it was still within the margin of error, with Emmer slightly ahead of Entenza, 37 percent to 34 percent.

Horner picked up 12 percent in that lineup, while 17 percent said they were unsure -- the highest level of undecided respondents in the three hypothetical races.

Republican Gov. Tim Pawlenty, who has held the office since 2003, announced last year that he would not seek re-election to a third term.

Other analysis

The latest results go along with a somewhat common idea among elections analysts: Minnesota's gubernatorial race is, for now, neck and neck. The Cook Political Report and CQ Politics, a Congressional Quarterly publication, both rate the race as a toss-up.

And the numbers aren't too big of a change from a March 10 Rasmussen Reports poll, which showed Emmer within 2 or 3 percentage points of four possible DFL candidates.

Respondents' support for Horner and Entenza has risen by a few points since March.

But a SurveyUSA telephone survey of 588 likely Minnesota voters showed Emmer with a higher lead than what Rasmussen Reports' latest results.

Forty one percent of respondents said they would vote for Emmer, 33 percent for Kelliher and 9 percent for Hornerw, while 17 percent were undecided.

That poll, conducted May 3-5 for KSTP-TV in Minneapolis, has a margin of error of plus or minus 4.1 percentage points.

The results changed only slightly when Dayton was included in the matchup. In that lineup, Emmer picked up 42 percent of respondents' support, Dayton got 34 percent and Horner got 9 percent, while 15 percent were undecided.

Entenza had the lowest level of support among the DFL candidates, picking up 31 percent of respondents' support compared to Emmer's 42 percent. In that matchup, Horner got 10 percent of the vote and 16 percent were undecided.





Source: ientId=24448&RQT=309&VName=PQD, retrieved July 7, 2010

So what do you need to know before we start?

 Definitions:  (you might have discovered/defined the terms before but  a review never hurts)

Sample Mean - the average of a set of data points. 

Normal Distribution - a normal distribution of data means most of the examples in a data set are close to the "average", while relatively few examples tend to one extreme or the other (think Bell-shaped curve with most of the data bunched in the center).

Standard Deviation - a statistic computed to tell you how closely the various data points are clustered around the sample mean.  When the data points are packed closely together, the bell curve is steep and the standard deviation is small.  When the data points are widely spread out (lots of variation), the bell curve is relatively flat and the standard of deviation will be larger.

  Note - you won't need to compute the standard deviation in this packet (whew!).

Chebyshev's Theorem and the Empirical Rule (summarized) - if you have a normal distribution, approximately 68% of the data will fall within one standard deviation ± of the mean, approximately 95% of the data within two standard deviations ± of the mean.

Margin of Error - in statistics, margin of error is also referred to as the confidence interval.  In a journalist's world, you will usually find the margin of safety discussed at the end of the article and stated as percentage points, "The margin of error for this poll was plus or minus 4.2 percentage points".

Source: Robert Niles,, retrieved July 7, 2010

I'll show you a Bell Shaped Curve

  Graph: One SD=68 percent of the bell curve, 2 SDs=95 percent, etc.   

The area in red shows  ± one standard deviation from the mean or roughly 68%  of your data points.  Add the area in green and you are looking at ± two standard deviations from the mean or roughly 95% of your data points.  The blue and white areas represent the rest of the story or what is often referred to as the "outliers".

Outliers are are those strange events that skew your data set - things like a brainiac student getting a perfect score on a test or a student who plays dot-to-dot on a multiple-choice, color in the bubble test and gets every answer incorrect (but makes a pretty darn cute dog picture!).


Source: ©/Copyright, Robert Niles,

Before we go any further...

"Further" Versus "Farther"

The quick and dirty tip is to use “farther” for physical distance and “further” for metaphorical, or figurative, distance. It's easy to remember because “farther” has the word “far” in it, and“far” obviously relates to physical distance.


But I digress...

I am sending you to an article by Robert Niles, a journalist and website editor, as he explains, and very clearly in my opinion, the Margin of Error.  


Read through it and then come back for our last step.



Now it is your turn...

Go back to the first slide and reread the article from the Star Tribune on the Minnesota Governor's race.  Is it more understandable now that you know about margins of error?  None of the potential candidates had improvements in their standings outside the margin of error which makes commenting on who has the lead rather irrelevant.  

For practice - find at least three articles using the concept of margin of error and see if they are telling you worthwhile information or just blowing smoke in your face.

Opinion Polls: Getting the results you want is easy. Think about this the next time an opinion poll is trumpeted as a real gauge of public opinion. "Opinion polls are not designed to measure public opinion, they're designed to shape it." Peter Hitchens

Source: BBC Worldwide - Clip from Comedy show, "Yes, Prime Minister"