Asimov was wrong

[eirias]

Isaac Asimov reputedly once said, "The most exciting phrase to hear in science, the one that heralds new discoveries, is not 'Eureka!' (I found it!) but 'That's funny...'" I'm sure he's right that new discoveries are often signaled by this phrase, but those breakthroughs account for only a small proportion of the times you hear it uttered. The rest of the time, it merely heralds stupid mistakes, like, you know, having labelled your data wrong. Unexpected and bizarre correlations popping out of your item analyses? Variables commingling in most unseemly ways? Check your labels before you pronounce your results destined for either Science or the dustbin. ;)

Comments

[tiurin.livejournal.com]

Errr...what did you "find" initially?

[harleybitch.livejournal.com]

ouch sounds painful! I think one of my prof's said 'what you dont see in this paper is the 80% of the work that was complete and total crap'. I try to remind myself that when an experiment doesnt work this is as important as when it does. But it always makes it harder when you make a dumb mistake (like not calibrating videos or not accounting for the camera possibly not being in the exact same position from trial to trial.) Just remember now we are learning how to be professional scientists we are allowed a few mistakes.

[rms10.livejournal.com]

I feel your pain. Dude, I'm living your pain.

A professor told me that only about 10% of what you do ever gets published, and then another professor chimed in and said that they really freakishly good people publish maybe 15% of what they do.

[cognative.livejournal.com]

I feel you pain as well. I remember when I was and undergrad I totally thought people published like 75% of what they did. Boy was I wrong.

I just had the 'That's funny..' moment with my study (again). And nothing but confusing and inconclusive data is coming out of it.

[eirias.livejournal.com]

Thanks for the sympathy, guys. :) Actually, it's not so bad - the big problem was when I realized a few months back that my stimuli were way messier than I had noticed before, which put us in danger of having unusable results. Basically what I was trying to do in this experiment was to "train" people on a new system of music (passive exposure) and see if they could pick up on the regularities without being taught. It turns out that last spring's results were hard to interpret because of the messiness I mentioned. The odd "finding" I referred to here was a secondary analysis I did this summer, trying to see whether the regularities in the exposure system correlated with people's responses - what I initially "found" when I did this analysis a couple weeks ago was that they didn't, and that people's responses only correlated with regularities in "normal" Western music (a big bummer!); but I realized last night that I actually had mixed up two of my samples, and when I fixed it, the correlations came out *better* than they had before.

So, to answer tiurin's question, I haven't really "found" anything worth publishing yet, it's just an interim result that indicates that we're not *totally* on the wrong track and that people are in fact sensitive to the regularities in the system. Now we just need a slightly cleaner system and 100 more adults and maybe we'll be good to go. Score.

[eirias.livejournal.com]

Damn. Are these more in-depth analyses of your FYP data, or have you collected people for Phase II of this project (the one you talked about in CPS brownbag in the spring)?

[cognative.livejournal.com]

this is phase II. Similar to my FYP but different. There is some good data but not as much as we would like. I'm about ready to just go with what we have so I can have something actually published. I think my advisor may want to comb the data a little more though.