« Relax on Peace | Main | Class and Creativity »
October 23, 2005
Numbers
I guess what's good about having a real job is that you do eight hours of data entry and then can come home and do something else. Being a last semester student means I do something else all day and then come home to do eight hours of data entry. This number-crunching will all pay off shortly, however, when I start up the regression analysis! (I don't know what this means, but it will somehow scientifically prove my claims - as long as the r-square is large enough!) Or maybe I'm thinking t-value.
To be honest, I'm a data entry speed demon, but no good at statistics. This is a great family shame, seeing that my dad is a full-out professor of math-stats. So most of what I know about multivariate statistical analysis is what I've overheard in class - in Japanese. I learned "回帰分析" first and the English term later. Whatever the case, my professor introduces me to marketing professors from other schools as a "解釈主義" guy ("interpretivist"), because my whole research strategy does not involve processing surveys to match up with explanatory path diagrams. I'm kind of the black sheep of my grad student clan.
But to make matters worse, I'm taking another class from a rival professor who believes that inductive statistical analysis will not lead us all to the ultimate academic goal - theory! He wants us to go deductive (演繹法), which sounds great on paper, but has little to do with the pseudo-field of marketing. Reading Hayek or Hempel is hard enough but this whole "philosophy of science" genre has almost no relation to the practical business field at hand. It's like we're trying to learn whale industry management from reading Moby Dick. No, that makes it sound exciting.
Posted by marxy at October 23, 2005 2:01 AM
Comments
aw man I'd be real dissapointed if you end up with some soft science type stuff. I for one like to know where my statistics come from.
Posted by: Chris_B at October 23, 2005 4:04 AM
Talking of day jobs, I thought you might like to see the photos I took on June 17th, 2002, when I wandered into the Tokion magazine office on Clinton Street, LES, NYC to find an earnest, soft-spoken, friendly, slightly chubby-faced and very tall young man working behind the counter...
Posted by: Momus at October 23, 2005 5:48 AM
Man, I look younger or different or something in those pictures.
aw man I'd be real dissapointed if you end up with some soft science type stuff
Statistical analysis can be great and all, but there's just so much (mis)use of it in the academic social science world to make things "scientific" automatically.
Posted by: marxy at October 23, 2005 9:57 AM
What's cool about use of statistics is that it does mean you're actually looking at facts instead of just voicing an opinion. (I do that in my day job for companies, mostly with people who couldn't care less what my opinion is, but who are eager to hear facts.) Talk is cheap, but actual research is hard to come by.
Of course, when people don't believe it, they have to attack the way the numbers were generated (if it was an opinion they would just attack you personally... I could give examples of that but writing them here would be like explaining ice cubes to an eskimo.)
The methodology is often lousy in the social sciences, and usually about two thirds into the paper there's a switch from interpretation of data to straight out speculation (usually on what the writer wanted to write about before s/he discovered that the data didn't back up any of the claims.) - not that that stuff isn't interesting too, it's just mostly opinion and less fact.
Posted by: Dave at October 23, 2005 11:16 AM
If I may preempt the inevitable Momsus post here: it's all opinion anyway. There are no impartial "facts". All data entry is guided by the way you look at the problem. Etc. pp., [add further sweeping generalisations and trivialisations of results from philosopy of science ad libitum.]
Posted by: der at October 23, 2005 8:10 PM
There were these five statisticians gathered around an elephant, see...
Posted by: Momus at October 24, 2005 1:28 AM
Sure, but with five statisticians around an elephant they're still not going to think it's a fluffy bunny, a lion or a plate of spaghetti.
If you strip back the interpretation far enough you're still left with 'facts' of one type or another. (E.g. even after Momus' intepretations and elaborations, Hisae must have said something...)
I agree that the way that you look at the problem guides/influences how you see data/facts, but without the facts that's all you have left!
Posted by: Dave at October 29, 2005 12:05 PM