## I wanted to be a statistician.

Way back when, in a time so long ago, I had a love affair with numbers.  It started early and I found I had a feel for numbers.  Even in early elementary school, fractions, percentages, averages and the like came almost intuitive for me.  What could be more simple and truthful than 1+1=2?  I took great delight in the discovery of baseball cards!  What incredible fun they were!  When my dad introduced me to the World Almanac and Book of Facts, I was in heaven!  Even before I knew what R square meant, I could see trends and curves.  I had no idea what a sine wave was called, but I knew they existed.  As I grew a bit older, my fondness for numbers grew.  They were pure and truthful.  They were a representation of past, present, and future.  As I continued my education, mathematics of any sort was always my favorite subject.  With very little effort my grades were always good in mathematics.  My fondness continued to grow.  In my eyes, not only did I see truth and purity, I saw beauty!

Then, a most serendipitous, horrible, wondrous event happened.  I read a book by Bill James!  My love for sports statistics had endured and now, I had so many more avenues to explore!  No longer did I have to settle for obviously flawed statistics such as BA, OBP, SLG AVG, and the likes, I had Sabermetrics!  How divine!  I even attempted to apply some of the concepts to other sports!  (I suppose basketball could have similar applications.)  My first try at college, had me majoring in beer consumption and chasing tail, but I managed a couple of mathematic courses such as trig and calc.  But, something happened during all of this.  The more knowledge I gained, and more ability with numerical processes and abstract concepts, the more I began to view my beautiful numbers in a less subtle light.  Where I once saw purity, truth and beauty, I began to see deception, and the stark ugliness of manipulation.  While my precious numbers were still mesmerizing, at times, she became less like the beauty at a church social and more like a tantalizing lady on a pole in some dimly lit bar.

So, what does this have to do with anything?  It has to do with science and the easy manipulation of numerical processes.  Take for example a recent guest post at Climate etc.  Now, I want to be clear, I’m not criticizing the post, Paul Clark, the methods described nor the author.  I’m simply stating there is a huge potential for making errant assumptions using the methods and I wanted to illustrate some of the more simple manipulations applied to climate science.  And then I want to introduce you to some difficulties applied to science in general.  I found a link to this somewhere, I’ll be darned if I can remember where I picked it up, but its worth a read.  The title is, “Odds Are, It’s Wrong”  It has a subtitle, “Science fails to face the shortcomings of statistics“.  Apparently, this is a reprint of an earlier submission.  And I’m glad it is.  Heaven knows, I’ve known about some of these problems since the time I saw my numbers dancing on a pole.  It lists some of sciences shortcomings, and many peoples’ lack of understanding about various subjects, such as Statistical insignificance, Multiplicity of mistakes, Clinical trials and errors, and Bayesian probability.  They were more generous than I would be about Bayesian probability, but that’s probably because Keynes relied on it.

No, I’m not saying statistics are bad, they’re not.  Further, good honest statistics are required for advancement of civilization in general, and in the specific case of climate science, we are in dire need.  I gave up on being a statistician of any sorts long ago.  I did as soon as I realized, in the wrong hands, my beautiful numbers weren’t simply a seductive siren on a pole, but could manipulated as easily as a pimp manipulates a street walker in search of her next fix.  Its best to keep the numbers as straightforward as possible.  The less statistical acrobatics needed, the less times you’ll see her on a pole.  1+1 still equals 2, its just that she doesn’t seem to in the same way she used to.

This entry was posted in Uncategorized. Bookmark the permalink.

### 10 Responses to I wanted to be a statistician.

1. Yes, beautiful girls can be a problem.

• suyts says:

Some people say ‘lies, damned lies, and statistics.” I say “angels, sinners and devils.”

2. Pingback: Dividing by zero | suyts space

3. henryp says:

Stats in climatic research is rather simple.
Just stick with actual measurements. Do not deviate from the measured number, how implausible it may seem at the time. If you have a month with less than 15 daily data, just fill in the month average as calculated over time, so it does not affect the slope of your trendline.

• suyts says:

Agreed. That’s why I don’t get much into some of the more heavy statistical discussions on climate. I find, more often than not, they are academic exercises and give us little insight to reality.

• Henry P says:

My comment was written in 2011
and I have subsequently changed my stance on filling in long term averages for missing [monthly] data.
Namely, we want to to record how the weather is changing.
I changed the procedure for missing monthly data (if less than 15 days) to:
If you have a month with less than 15 daily data, just fill in the (month) average of the year before and the year after [ so we can see how the weather is changing].
If you do it all right you can get 100% correlation of what it happening on earth with the temperature.