The Ill Fated Stats Tutor

Mike was a nice guy. In the way that left-handed math geniuses are nice guys. He’s smart and he knows he’s smart. He really wants to help. Let me restate that. He really wants to get paid to do his job. When I called his employer (a nanny and tutor renting operation) they weren’t sure they had anyone who could really do tutoring for a graduate school student. Mike volunteered.

You should sympathize with Mike.  He teaches Chemistry and Calculus.  He’d just have to brush up a little, right? Right. He had no idea what he was in for. To be fair, neither did I. The questions I have about Statistics are very basic.  Without going too far into it, I’ll try to explain.

We were dealing with bell curves when I arrived. Bell curves and hypothesis testing. Zscores, Tscores, and rejection regions.  One of the questions I had was what’s under the line of the curve. Is that all people? is it answers? Is it  confidence? When I’m doing means testing and I come up with an answer, (1.684 for example) where is that answer on the curve? Is it on the line? or in the hump of the curve.

Mike would get confused about my questions. Sometimes when he got confused he’d smile and shake his head. Sometimes he’d pinch the bridge of his nose. Sometimes he’d say, “What do you mean?” Seemed pretty basic to me. Sometimes he’d answer and his answer would be wrong. Mike didn’t admit being wrong. Even when he made me miss two questions on my midterm with his wrong-ass answers.

He got confused when I asked what a certain kind of equation or theory told us about the real world. There’s this thing you can do with graphs and taking the means of a group of means.Sampling distribution of the sample means, it’s called, I think. You can do a little example of it if you want here.  If you do it enough, even a very bizarrely irregular dataset will become regularized.  I wanted to know why it was useful to take something so unusual and make it look regular. You could have a really unusual distribution, little peaks and valleys or two big peaks, whatever. If you run it through this process of equations, it would become a normal curve which you could then use to learn stuff.

How does that not use manipulation of data to tell a lie? Well, if I was describing it right, it sounded like my stats professor was maybe wrong. He couldn’t say he didn’t know. Especially when he didn’t know. He spent about 25% of my/his time/money flipping through my stats book. Shit, I can look up stuff in the book for free myself.

I gave him an example (one that jumped to mind) of a therapeutic intervention that you wanted to test. Say you have a therapy that you think will reduce the number of some hand-flapping tic in the kids in your special ed classroom. You want to test your therapy against a control group. In the control group you do nothing and measure the number of hand flaps one day and the next day. In your experimental group, let’s say you cut off all the hands of the children.  You count the hand flaps one day, cut the hands off and then count the hand flaps the day after that. Your results should be pretty drastic. It won’t look like any kind of a curve.  It will be a kind of block showing all the kids with total absence of flapping behavior. If you run that block shaped graph through this special process enough times, the block (in theory), will become normalized to look like a bell curve.  Then you can analyze the curve in all sorts of useful ways.

For whatever reason, Mike got stuck on the amputation of the hands on the kids in my experimental group and he was at a loss to explain to me the theoretical explanation. I really needed someone who knew enough about the subject that they wouldn’t get distracted by my weird examples. Mike wasn’t the guy. “Can we make it something else? like we tied their hands down?” Sure we could, but I still wanted to know how the new curve could really be useful to analyze anything in the real world. Isn’t that what statistics is supposed to do?

His fate was sealed when I changed the answer to a question because of what he told me, and got it back marked incorrect. The right answer was the one I had first. I didn’t go back, but I think I passed the class anyway.


5 thoughts on “The Ill Fated Stats Tutor

  1. You wanna do WHAT to kids hands??

  2. Not YOUR kids’ hands.

  3. emily says:

    ah. so you meant it literally when you said “it’s not me. it’s him.” i had thought maybe you were tired of the nose-bridge-pinching thing.

  4. Terry McDanel says:

    Wow, that little website is coool. I played with that for an hour. I am glad i did not have such video games in college. I would never have gotten my degree.

    Your question “I wanted to know why it was useful to take something so unusual and make it look regular.” I think this is brilliant. It is the essence of the quandary of statistics. It goes to the very heart of the subject of analysis. I hope you asked it in class. I am reading a book by Richard Feynmen right now, and … know this is going out on the line here by a pretty ignorant hack who is way far down on the Feynmen scale… but i think he would be proud of you. I will not pass judgment on your tutor because i am not qualified but i am tempted.

    Sampling is sampling of reality, and equations are idealized approximations. Your question is relevant to several areas of math and science, including the most fundamental math question “Is ‘math’ really out there or in our heads?”

    And in quantum mechanics “Yeh.. even tho a probability wave can go thru two holes at the same time a ‘real’ electron cannot!” How can light be both a particle and a wave? The standard answer to the question in quantum mechanics is not really what are you observing? but rather how? The “what” depends on how you observe it.

    The question i think you hit on is the question that math theorists have pointed to as evidence that math is a artificial construction of the human mind. It is not really “out there”.

    Your hand amputation example, i less optimistically must rejoin, does not work. You divided the experimental set into two arbitrary groups of observation, then are trying to combine them with linear equations. To use quantum mechanics as an example, if you watch one hole or the other for the electron to pass thru, you are dividing your experimental sample into two groups (hands and no hands) and measuring particles, a series of single occurrence events, we call “particles”. You will not see waves of probability when you do that, only a series of instances.

    I would suggest having the children stick their hands into meat grinders. Some will react quicker than others. This will give your experiment a linear output as a function of reaction time. When you graph it out you will see linear functions with waves, or as unfortunate cases would have it… no waves.

    Altho Ramsey County Child Protection may have some input into this.

    But seriously, your basic question is right on. This really did trouble Einstein. He refused to believe that reality had any purely statistical basis, that there are theoretically observable … what i guess one could awkwardly call “substructures” with a plethora of variables that give rise to what we observe. Its just that usually it is too complicated or difficult to observe them. Incidentally, he was proven wrong by a guy named Bell in the case of quantum mechanics.. but i think that is not relevant to statistics and human affairs. I have neighbors with easily discernible substructures.

    So it is summer vacation. There are kids running around pointlessly waving everywhere. Lets go buy a meat grinder?

  5. Terry, I wish I was as smart as you seem to imagine I am. I do have a pretty good ‘this doesn’t make sense’ detector. It doesn’t alway make my life better. I hope some day to be articulate enough to argue my case better than I can right now.

    I passed the class, tho. With an A. Still don’t believe it.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s