Effect size

Date: 4/24/14
Distance: 4.6 miles
Location: Carrboro, NC
Catalyst: Cracked Podcast "15 Things that Secretly Influence Who You Have Sex With"

First I saw a newspaper thrown in someone's driveway. I tried to read the headline as I ran by but with the jiggling of running and the hard-to-read angle and distance of the writing I only got the words "Fines Levied" before I was past it. Those words were in my head for a long time. I assume the next word was "Against" because I can do Markov-Chain modeling in my head. But after that is the actual payload of the headline I assume but I missed it and it bugged me. Can we all be less wordy? I'd appreciate it. Thanks.

It was good to run on the Carrboro route again. I did a run in Ann Arbor recently. I wanted to do one in Chicago but failed to make the time. I wanted to do one in Great Falls, VA but just didn't do it. I did a dumb one at The Farm near Carrboro while the girls had their tennis lessons: I just ran around in circles for a mile. Dumb. I like the familiarity of a well-known route. Maybe if I learned to enjoy charting new routes more I'd be more motivated to run while traveling.

Lots of dogs on this run. One scared the crap out of me: it was running behind his owner. The owner passed me but I didn't know he had a dog with him so twenty seconds later I heard a panting, running beast sound come up from behind me I was startled enough to yelp. The guy heard me, then he took his dog and stopped on the side of the trail until I passed which was nice of him but not necessary now that I knew the dog was there.

A guy stopped me on the route and asked where was a good place to run. I gave him some options I know about but he seemed oddly skeptical about all of them. Sucks to be him, I guess.

The Cracked kids were talking about that famous psych study where male subjects cross either a rickety dangerous bridge or a sturdy safe bridge and complete a survey handed them by a pretty woman halfway across. The upshot is that the subjects who filled out the survey on the rickety bridge were more sexually or romantically attracted to the woman than the subjects on the safe bridge. It's famous empirical evidence for misattribution of emotional states: subjects attributed their excitement and/or nervousness to the woman rather than the bridge. Anyhow, I somehow never knew that the dependent measure of the study was whether the subjects called the woman 2 days later (she'd given them her phone number and said wait 2 days then call if you have questions about the study). I'd always assumed there was a survey about how they felt about the woman when the bridge task was over. The actual measure is really interesting because 2 days is a long time to possible re-evaluate your emotional attribution and/or to do other things that are more interesting than meeting a woman on a bridge and then pretty much forget the whole experiment.

So anyway: what was the effect size of that study?

By the way, I'd like to see a study done on how much exaggeration goes on when journalists recount study results. I know that a publishable social science effect is something like d=.2 or so. When that happens, there's a slight nudge in the dependent measure from the manipulation but there's a ton of noise masking that nudge. But when studies like that are recounted, I think writers tend to overshoot the results ("none of the controls showed..." or "people who saw the X all responded with more Y" or whatever). Someone should quantify that. I bet it does our science a disservice by raising the expectations that published studies are accounts of big, reliable effects. Lord knows they are not and it's a new-ish fad to earn points by pointing that out to people who don't actually know that.

Also, running on trails is hard because rocks are not evenly spaced nor equally sized.

And I should buy a sweatband for my head.