Since I have being working with tones of statistics lately, I figured I would dust off my stats software and do some graphical representations of my current academic status. This got me thinking about why people have such a hard time with social scientists who spew out stats like there is no tomorrow, and expect you to fully comprehend what it is they are trying to say. I have a paper called, “How to Lie With Statistics, (2000)” that seems to suggest that any number associated with a label could be presented for any argument that is “spun” for the media to sway or persuade you into believing what the research is trying to claim.
Take for instance my lovely graph below. My grade for this class is 82 percent, and my standing is A-. So what does all this other garbage mean when you are looking at the bars and curve? The bar graph that you learned from high school means something totally different than what statisticians mean with this type of graphical layout. Sadly, it would take a lot of pages to lay out the basics of what I’m saying here. So I’m not even going to attempt it.
What this graph is showing is that all of the ten members of the class are above the 60 percent mark, and that the curve is suggesting that the mean (average) of the class is around 73-4 percent. The bars are actually individual observations, in other words, the 60 percent range has 5 students whose marks fell between 60 and 79 percent.
The curve is what the “normal distribution” is, and has a “standard deviation” of 9. Also this line shows that everyone is within the upper two quantile as far as marks goes. Anything at the right side of the curve-tail, that is greater than 9 percent, from the 74 percent mean, is above normal, or is the highest group in the class. I’m not one of those. Most of you are lost now—right.
The graph looks pretty, but most researchers don’t like to use them because they are difficult to paint the whole picture. Instead you will see lots of text/numbers that are divided into data sets with their corresponding labels to show accurately what the research data is saying.