2.5.2006 | You can't run the stats too many ways or too many times

 

Only one more weekend between me and work-free weeknights. It's been two years of working full-time days and the second job at night and on weekends. These are lost years. I have few memories of what I've done and where the time has gone. But now, with evenings free, I can rekindle friendships and return to classes at Urasenke. I'm going to take cello lessons. I hope I'll stop driving my car and start riding my bike everywhere.

This weekend I worked Friday night till 2:00 a.m. and then got up early for a 9:00 a.m. massage. I was soo tired that I decided to bail on yoga. I went back to sleep till 2:00 p.m., brunched, and enjoyed a pastoral series of errands. I think that's how weekends should be: tending to household needs and indulging yourself in unstructured time. I played video games through the evening and slept long. It was Super Bowl Sunday and the city had retreated to television screens. Outside had a bit of a post-apocalyptic feel: the streets were empty and silent. I went for a run and wondered about the few people I saw who were not watching the game. At Green Lake, the storm had toppled trees and no one had come along to right them or chop them up.

Tonight there is work to do. I'm rerunning regression analyses on a two-year-old paper that should be submitted already. Each time we hand it off to the PI, some instruction to redo or do something differently comes back, sans explanation. The current problem is a reversal of one of the first decisions made once the data were reviewed. The inclination would be to analyze it as I am about to do it now, but, as was argued then, doing so would generate meaningless results due to the lack of power. I believe this to be true and remember the numbers we got when we ran this exact analysis to illustrate why it shouldn't be done. The entire set of predictions in the paper is predicated on this reasoning. Two years later, I wonder what has changed. The numbers certainly haven't. I'll bring up the issue, but I can't imagine doing so and hearing, "Oh! You're right! Of course!" It never goes like that.

 

BACK | INDEX | NEXT