I went to the lab today to pull some data off of SPSS (Statistical Package for the Social Sciences: like Excel, but engineered for data about people). I actually have a copy, but it’s the student copy that I got for my psych stats class. And by student copy I mean worthless copy. It won’t do multivariate analyses, and doesn’t handle files that contain more than 50 variables. My study currently has about 70, so I can’t even see my data without leaving the house. I suppose I could buy a copy, but the grad pack is about $200 and, alas, I am a poor undergraduate. Anyway, I wanted to pull some numbers for a project summary that’s due this Friday, and noticed something odd. One of the participants answered “male” for every gender question I asked. Guessing that this individual wasn’t really being honest during the rest of the study either, I removed that person (participant 14) from the data set and re-ran a multivariate 2×2 ANOVA on the remaining data. I mentioned in my last post that we had gotten the p value for the speed measure down to about .12 (.051 if we ran it as a repeated-measures, which is a little dishonest). Apparently, the outlier analysis that we ran earlier (i.e., before I started presenting my findings) must have missed participant 14, because the speed measure now has a p value of .044 (with only 16 participants). T minus 5… 4… 3… 2… 1… we have significance. Ladies and gentlemen, we have a publishable result. We are happy. After removing 14, I noticed some other really interesting stuff going on, but I will wait to comment on that until I am more certain that my data analysis is reliable. Anyhow, let this be a lesson about implicitly trusting software. Computers are machines made by men and are thus fallible, so look at your data (in my case, I have someone in particular I’d like to speak with).
Which brings up a tangentially related, but very important point. The brain is not a computer (in a similar vein, the internet is [not] a series of tubes, but we can talk about that later. Maybe we’ll talk about it now, so you can think about it. Look at me pile up cheesy internet video quotes again and again in an extended parenthetical. Ok, that’s enough). Getting back to the main point, the brain is nothing like a computer. I can understand that a person may wish to use the computer/brain analogy to clarify a point, but the risk inherent in such a strategy is that people will take it at face value and actually believe that the computer and the brain work the same way. They do not. A computer uses a microprocessor to add up strings of 1s and 0s that it stores on a hard drive or solid-state memory device. A sapiens brain (i.e. yours) is a network of over 100 billion independent, plastic (the adjective, not the noun) processors, which are each connected to about 10,000 other processors both in parallel and in serial, and which modify each others’ activity via a collection of signaling molecules, many types of which can be released from a single processor at once, and each of which has a different effect that depends on both what kind of molecule it is and which individual processor it hits. Phew.
In related news, Descartes died three hundred and fifty years ago. So should have dualism. A “mind” is brain activity. It is not a spirit that rests on your pineal gland. Period.
Today’s take-home message: fund my research so I can analyze my data in pajamas.
Post a Comment