Wednesday, March 8, 2017

Misleading Statistics Blog

Misleading Statistics
They're Everywhere!

Decadal Average 131
Average of what?


source: National Oceanic and Atmospheric Administration. (n.d.). Tornado graphs. Retrieved from http://www.weather.gov/gld/tornado-tornadographs 

     Who knew that a graph about tornadoes could be so misleading?  This decadal average graph of tornadoes that was published by NOAA has a couple of issues.  First, NOAA says that the decadal average of tornadoes is 131.  What exactly do they mean by average?  Huff (1954) says that the arithmetic mean, median, and mode can all be considered when looking at statistical data (p. 28).  Depending on which one of these you use, your data could appear very skewed.  It would be beneficial if NOAA provided exactly how they determined this "average."  At first glance, most anyone would assume that it would be the arithmetic mean; however, when tested, the true arithmetic mean of the given data in the graph is 120.7.  This makes me wonder how they arrived at the average for each decade as well.  

I know the purpose of this graph was to evaluate the average number of tornadoes that occurred each decade; however, this graph leaves me wanting more information.  The 1990s and 2000s had extremely high averages of tornadoes.  Were there particular years that the number of tornadoes was extremely high?  I wouldn't know because this graph is missing so many important factors.  Also, this is not a fair representation of the 2010s because an entire decade has not passed yet.  Even though in the title it says that these are the averages for 1950-2012, I feel that this is very misleading.  There are still seven more years to consider in the decade of 2010.  Based on this graph, it appears that there has been a significant decrease in tornadoes from the 2000s to the 2010s; however, they are are only reporting three years of the last decade.


Average National Gas Cost

Or is it?

Fox News


source: Theel, S. (2012). Fox still struggling with basic chart concepts: Gas price edition. Retrieved from https://mediamatters.org/blog/2012/02/21/fox-still-struggling-with-basic-chart-concepts/185049



The above graph of the national average cost of gas has two misleading factors.  One, as described in my above post, relates to the loosely coined phrase "average."  Was the mean, median, or mode used to determine the "average" cost of gas?

This graph is also misleading in the fact that the increments of time range from, last year, last week, and current.  This graph would carry more informational value if along with the "last year" phrase, they included a date or even a month.  The same would apply to the term, "last week."  Was this Monday of last week, Tuesday of last week, etc?  A better way to represent this national cost of gas in comparison to last year would be to graph the national average of each month.  This would better show the viewer the fluctuation in the prices and better explain what seems to be a significant rise in the cost from $3.17 to $3.51.





Huff, D. (1954).  How to lie with statistics.  New York, New York: W.W. Norton and Company.