Recruiting PCC students: and my experiences running one and a half experiments.


We made several trips to pcc during the course of our project. We did not try putting up posters, as we were told they tend not to be effective.


Charlie contacted several proffessors requesting permission to announce experiments in their classes. We only intruded on ecconomics and computer science classes.


Rafi:

I went to a saturday afternoon computer science class. The professor and students were quite friendly (I think the students didn't mind my delaying their test a few minutes). There were about 15 people in class that day, most seemed interested in my announcement. I should note that unlike the other recruiting expiditions, I only gave out information, I did not ask for information to contact the students. I gave them the url to sign up to ec experiment list (I told them to do that first). I also gave them the url for the experiment we ran on 11/20/2000. I told them simply that it was a simulated market, which we were running for a class. Several students wrote down the urls I provided (I left them on the board when I left), and a few asked questions. I answered honestly and frankly. I wasn't sure what exactly we were going to be testing for. I did tell them it would be an auction and they'd make some money, how much I had no idea. I also mentioned I hoped it would be fun.


The hour before the experiment started we did see several register with the system, I hope at least one was one of the students I talked to (I did not tell them to register immediately, any time from 8pm till 9pm would be fine). By the end of the experiment we had a total of five outside participants 2 turned out to be caltech students who simply check the reg system daily in the hopes of catching experiments they might have missed otherwise. 3 were pcc students, there was a fourth that was logged in for a few seconds, then we lost him and he failed to get in again. Travis watched as a few other pcc students failed to log in, one kept trying to use a social security number as a password (hopefully the new log in page will offer a choice, if for no other reason than to show them that the password is not the social secuirty number). I mostly watched the web logs and the data/reg/log to see who was actually logged in. A bit of advice, if you want to know where they are loggin in from (watching /var/log/httpd/access.log) you might want to keep a window or screen open with nslookup running (type nslookup, it will give you a prompt, just enter an ip address to see what the name is (you can also enter the address on the command line, and it will just check the one address). Whois might give more information, but if you can't tell anything from the name, chances are slim that its worth the effort to find out more information.


I don't think we had any calls for help, though considering the problems people were having, we wish they had. Unfortunately I did not give out the lab phone number when I went to pcc. I strongly recommend doing so.


Once the experiment was running, Charlie watch some of the participants by logging in as them, and looking at the histories, etc. He found some strange behavior on the part of one subject, but they eventually caught on without any assistance, I guess that's part of what ec experiments are for.

At the end of the day, Travis helped us get the winnings out of the computer. I'm still not sure what the easy way to do that is. Keep bugging Travis until he writes something to simplify that part.