Writing your first paper: Step One: Am I ready?

Well, its taken a little longer than I would like, but I thought I would do a series on writing your first research paper. I am hoping to use this material for a new student handbook when I start my faculty position, so any thoughts, comments, hysterical rejoinders are most welcome.

How do I know if I have enough data for a paper?

This is a tough question and depends in part on where you plan to submit the paper. A more prestigious journal like Science or Nature will require more data, sometimes up to two years worth! The real question that you need to ask yourself is: Do I have a story to tell? Specifically, did you find an interesting effect or make something new? Did you do the experiments that explain why that effect occurs or demonstrate the functionality of the new creation?

I try to think of all of my experiments from the perspective of the story. Sometimes it is helpful to layout what you want the publication to look like well before the experiments are complete. This allows you to think about what experiments are required, what the controls should be, what the figures should look like. For example, say that I have developed a new method for making ice cream with liquid nitrogen (hehe! try this sometime its really cool). I want to write a paper about my method. What would it look like? Well, I would need to describe my technique and demonstrate that the product is actually ice cream. I would probably want to include a negative control (no liquid nitrogen or regular freezer) and a positive control (traditional ice cream maker).

However, just demonstrating the method is not sufficient. The liquid nitrogen method is cool because it is very fast and reduces the ice crystals in the mixture. The best paper would communicate this to the reader. I would need to show that the amount of time required to go from liquid to a certain viscosity (hardness) is shorter with my method, perhaps using a viscometer to get nice quantitative data. I also need to show that the method reduces ice crystals/has good texture flavor. I could do this by performing taste tests, but that is very subjective. I would want a quantitative method to measure ice crystals/smoothness of the mixture. This might be accomplished by measuring the air content (a standard measure of ice cream quality). Finally, I want to make sure that I have enough data to convince my audience that I have a statistically relevant result.

I have read several papers that are technically new, but not very exciting. For example, if 10 people have shown that liquid nitrogen is good for various types of ice cream, and if I try to write a paper saying it can also be used to make rocky road, that is not very interesting. If it can be used for chocolate, vanilla, etc. it is likely that it can be used for rocky road as well. That is not a good paper. A good paper must take the next step. For example, what if we made rocky road using liquid nitrogen and found that it maintains the texture of the marshmallows exquisitely. If we characterized marshmallow fluffiness and the distribution of marshmallows per scoop and found that they were superior to traditional methods, that would be a substantive contribution. In this sense, it is very important to place your work in context with that of others. A good couple of weeks in the library should help, and although I always seem to do this step after getting most of my data, the best researchers would do this before even starting out.

Once you have an idea of what the data will look like, you can lay out each of the figures. Then, go back and pretend that you are critiquing your paper. What additional data would you like to see? What are the most likely arguments against your work? For example, if I demonstrate my procedure with vanilla and chocolate, reviewers might wonder if it works with chunky strawberry. Would whole fruit interfere with the freezing process? Would the strawberries freeze at a different rate than the rest of the ice cream interfering with consistency? Other criticisms, might include safety and cost. Liquid nitrogen can be a hazardous substance, is not readily available to the average consumer, and might be more expensive than traditional ice cream making methods. How do the benefits of my approach outweighed these limitations? You might need to collect a few more data points, and perform a few more quick experiments. However, sometimes it might not be possible or desirable to conduct experiments to address all of these issues. In that case, be sure to address these points in the text. "We found that liquid nitrogen makes superior ice cream for smooth mixtures, chunky mixtures remain to be examined, but will likely produce equally good results because the rapid freezing technique employed should prevent ice crystal formation throughout the product..."

Tomorrow: Choosing a journal for submission