It’s apparently not all that rare.

I would listen to your professor.

It is pretty uncommon, especially as a first author, and would be great for your grad school applications.

I know you didn’t ask this but since you’re on the fence about doing the study—what are your current N and effect sizes? Have you done a formal power analysis? You can do one easily here. If you need help estimating Cohen’s d let me know, I can show you how to do it.

It’s possible. I published (not as first author) in high school and in college.

Not rare at all. I published as an undergrad, just as a lot of honours students do if the project is a good one.

Emily Rosa was 8 years old when she published in JAMA
http://jama.ama-assn.org/content/279/13/1005.full.pdf

Go for it. Even if your paper doesn’t get published, the experience you would gain would be invaluable.

It does happen, but it is rare.

Not like… OMG YOU JUST FOUND AN ORIGINAL EDITION OF SUPERMAN rare…. But like…. 1/100 rare.

@nikipedia My study was a 2×2 between-subjects design, and the Cohen’s d for one of my IVs was 0.50, and for the other IV it was 0.38. I’m pretty sure that 0.38 is a small effect size, but that 0.50 is medium (am I wrong?). Also, the partial eta squared statistic for my interaction was .102, which I believe is also medium. My n=20.

I looked at that power analysis website (thank you by the way, that’ll definitely be handy in the future), but what are the criteria for being an underpowered study? I certainly know the difference between one-tailed and two-tailed, but what does the “observed power” number actually mean or suggest?

Underpowered means that, given your effect size and your N, you don’t have enough information to reject or fail to reject the null hypothesis.

Officially, studies should shoot for 80% power, so the “observed power” number you’re looking at should be about 0.80 (if you want me to explain more about what “power” actually means I can do that, but it’s long and kind of confusing).

Unofficially, if you need a very large N to get sufficient power to get your effect, reviewers tend to think that your effect is unreliable or unimportant. So I wouldn’t invest too much in your d=0.38 effect. But I have published with 0.52; the 0.50 is a better bet.

Some other things to consider: if you’re working with human subjects, is there any chance that a subgroup of your population is either driving or washing out your effect? For instance, do you see any differences in performance based on sex, or other variables that you have information on (age, ethnicity, tiredness—whatever you think might be relevant to your DVs)?

@nikipedia I did work with human subjects, and most of them were undergraduate freshman. My study basically looked at the effects between slow or fast music on retrospective time estimation (i.e participants has to retrospectively estimate the duration of either a fast or slow tempo musical composition). I didn’t find any correlations between the participants musical experience or ability, but I will certainly run some further correlational analysis on gender, age, etc. Thanks for the tips!