Pie

Originally posted at www.moderndaypearls.com.

I love pie charts.  As well as other graphs.  And pie.  But if there's one thing I can't stand when it comes to graphs is ones that show polls with incomplete information that lead readers to believe more than one-third of Americans want to keep fighting the implementation of the Affordable Care Act.  Exhibit A:

What is the problem, you say?  This pie chart is very informative.  It has percentages displayed.  Captions.  A survey question.  A citation at the bottom.  The Kaiser Family Foundation has even chosen Chicago Bears colors for this lovely pictorial display!

Wrong.  The problem with this little pie chart is that we are not told how many people were surveyed to answer this question.  Was is 10,000?  Or 49?  Did this researcher choose a random sample to best represent the general population in a careful, scientific manner?  Or did he just survey some of the guests at a wedding reception he went to last Saturday night?  Hard to tell.  We need a sample size, people.

There are also polls like this one, asking readers:  "What was your reaction when Anderson Cooper came out?"

While it goes without saying (but I'll say it anyway) that these poll results are nowhere remotely close to being generalizable to the larger population, it mostly annoys me that it's placed on a news website.  Even if simply to increase interaction with readers, the poll produces no valid conclusions.  Respondents can only select one answer, and the answer do not necessarily have anything to do with each other.  Believe it or not, you might have shrugged when you heard the news, but maybe it also prompted you to start watching Anderson Cooper on CNN.  Who knows.

Yet by far the most misleading poll I've seen online today is this one by ABC News, which tries to pass off sub-standard survey methods for actual news (no offense).  In an article trying to inform readers that the public is still not fully supporting the Affordable Care Act, the author writes:  "The poll found that 43 percent had a favorable view of the Supreme Court's decision to uphold the law and 42 percent a negative view."  Have these people not heard of margin of error?  If, in this case, the margin of error is anything greater than 0.5 percent, then the results of the poll could be radically different, as in possibly more people have a negative view of the decision than a positive view.  I hope that's not the case, but some clarification in the data source (and - ahem - a sample size) would help readers better understand how valuable a poll might be.

What do you think?  Are your opinions influenced by flawed online polls you see online?  Do you also love pie charts?

p.s.  I am a huuuuuuge fan of the Kaiser Family Foundation, which does highly reputable research on health care topics affecting all of us.  Just because I like criticizing charts does not mean I am in any way disapproving of their very valuable work.  Same goes for you, ABC News/Washington Post and Seattle Times.

Images are from linked pages above.


More Like This

Recent Posts by moderndaypearls

Comments

In order to comment on BlogHer.com, you'll need to be logged in. You'll be given the option to log in or create an account when you publish your comment. If you do not log in or create an account, your comment will not be displayed.