Thursday, November 29, 2012

Lies, Damn Lies, Statistics, and Mr. Williams

Well, as I sort of expected, Mr. Williams did not take kindly to having his theories...well, proven to be total fiction.  And, as I expected, he has made a rather pitiful attempt to lie his way out of it.  There is not much else he can do....

He begins with his usual ad hominem (he refers to anyone he disagrees with as a troll) and then, well, he really blows it.

He produced a laughable graph that he attempts to claim trumps mine.  What he has actually done is resort to such a dishonest tactic that he has destroyed any hope of credibility.  You see, he is trying a bit of misdirection.  He produced a graph, that shows big values for transgender, but which, because of the way it is constructed, is actually deceptive and then he attempts to pull the wool over people's eyes (and some will reach up and help him move it into place, I am sure) by calling on people to compare to graphs that are not really comparable.

What he has done is sort of like comparing a map that has a scale of one inch to one hundred miles, that shows the Eastern Seaboard and another map, that has a scale of one inch for one thousand miles miles of the entire United States, and claiming that that they show the second shows that New York is closer to Los Angeles than it is to Washington, D.C.  You cannot make the comparison he is making.

My graph shows the relationship between three words over a longer period.  His compares several different words, over a much shorter period.  Because his graph has a different set of words, and a different time range, difference between the occurrence of the words are more exaggerated.  This makes it look like "transgender" shows up quite often.  But in truth, it just shows that it shows up more often than the other words, and in particular, he compares "transgender" with "Transgender."  That alone would render the results invalid since the program is case sensitive.

Here is a more accurate version of what he produced:

 It compares the two words from my chart, which is not what he used.  And it has smoothing of zero, which is more accurate for a smaller time period.  Simply put, there is a very tiny blip in 1974, but relatively speaking, the word "transgender" is non-existent during this period compared to "transsexual."  Oh, and yes, I could have set smoothing to 3, which would have showed an absolute flat graph for "transgender."  

It should also be noted that he took a comment out of context.  As I pointed out, later in the article, which he conveniently ignores, the word occurs in a few isolated sources.  But it was not of any significance until around 1990.

Sorry Mr. Williams, but you cannot lie your way out of this one.  You can only make yourself look more foolish.

No comments: