Writing your first paper: Step Eight: Discussion

This post is part of a series (that is taking a long time to write) on writing your first research paper. I plan to use this information for my graduate students. So feel free to send your thoughts, comments, and hysterical rejoinders. Also, all data is completely made-up so don't quote me on anything.

The discussion section is intended to explain the significance of your data, discrepancies and limitations of experimental design, the theory/meaning of the cumulative result, and its broader impact. So for the ice cream example paper the first part of the discussion would discuss case by case data analysis.

For example:

Liquid nitrogen ice cream displayed superior texture to hand crank and electric machine-made ice cream, but equal texture to store bought as determined by viscosity testing. Additionally, liquid nitrogen ice cream was preferred to hand crank and electric machine ice cream in taste tests. Thus, liquid nitrogen can produce ice cream with superior texture and taste, comparable to store-bought, when manufactured in the home.

This part of the paragraph summarizes the data presented in results and ties it together with our message: LN2 is a viable home alternative for creating ice cream. Next, we will discuss the limitations of our data, experimental design, and hypothesis.

However, because the test group consisted of a narrow age range, additional testing may be required to determine the appeal of LN2 ice cream in a broader population size. Additionally, our sample size was not sufficient to differentiate preferences between LN2 ice cream and the store-bought positive control. While we found that LN2 ice cream was preferred to other home techniques, it is unclear how this preference will affect consumers in the market for home ice cream makers. LN2 ice cream can be made very quickly and displays superior texture to other home methods. However, some users expressed concerns as to the safety of an LN2 system. Although we demonstrated that our self-contained system can be used safely by the average consumer, this negative perception would most likely have to be addressed in the marketing campaign. Additionally, it is not clear whether the superiority of LN2 home ice cream machines would translate into an increase in the size of the consumer market or simply an increase in the market share. Ice cream consumers will most likely continue to rely on store purchase as the primary method of obtaining ice cream products.

So in this section we talk about experimental difficulties. i.e., we tested only college kids, small sample size, no market research. We also talk about potential problems with our hypothesis, i.e., consumer safety concerns, and suggest how these could be addressed (marketing campaign). It is important here to air all your dirty laundry. If you can criticize yourself effectively and suggest possible alternative explanations, experiments, etc to address your deficiencies you will leave very little for a reviewer to pick apart, and they will be impressed with your 'tight' paper.

Finally, the impact of the theory and its broader implications are discussed.

Home ice cream machines represent a large market in the US. However, when polled the majority of users say that use is limited by poor texture and long freezing times. We have designed a home ice cream system using liquid nitrogen as the freezing medium. This system can be operated safely by the average user, and produces ice cream in seconds with texture and taste equivalent to that of store bought varieties. The LN2 machine represents a significant advance in home ice cream making, with the potential to address most consumer concerns. Although additional testing is needed to confirm, it's introduction to the marketplace could substantially increase the number of home ice cream machine users, or at least the frequency of machine use with existing consumers.

This last paragraph is the most important and is your final chance to impress the reader with what you have done. Use the paragraph wisely. This is why I like to put limitations in previous paragraphs, so the last one can discuss the positive impact of your work as much as possible.

Well, more later, I think all we have left are references!!!

I want to be a window washer

My in-laws live in Florida. Needless to say it was amazing over the holiday break. I took my kids swimming, twice. They live in South Florida, in a rather affluent area, which reminds me of Eden. In some ways, I think there is too much opulence and pretention, but in other ways I cannot help but appreciate the pleasure of it all. I have often wished that we, too, could live in SoFlo, near the beach, what an idyllic life it would be. Unfortunately, I want to be an academic and the universities in the area do not accommodate my field of study.

This weekend I was sitting in my in-laws living room watching the window washer. He looked to be in his late 30's early 40's. He seemed quite happy. I was watching the back and forth of the squeegee, which is almost hypnotic thinking about what kind of life he must lead. He is self-employed. He works when he wants to. He isn't paid so much, but I'm sure that he makes a pretty decent living. He lives in a climate that is consistently warm, with the beach minutes away. I dreamed of quitting my job and becoming a window washer.

And then the cruel reality of fate hit me. I am too smart to be a window washer. I know that sounds a little elitist, but it is true. Before I went to grad school I worked at a regular (actually pretty prestigious) company with my BS. And, I was bored out of my mind. Unfortunately, I could not be happy with only the squeegee to keep my comfort. Even at my postdoc, where I am following a fairly traditional route of research, I have moments of fleeting boredom. I think sometimes, when you have a certain level of intelligence, it becomes exceedingly difficult for the brain to be satisfied. I enjoy my research because it is maddening. It is one of the only times in my life that things don't come easily for me, and I love/hate it.

I find it extremely unfortunate that my intelligence limits my choice of career in such a way that it makes it difficult for me to live in places that I would like to. I just hope that I am making the right decisions. Sometimes I feel like I am like a trapeze artist, jumping from one swing to another. For me finding happiness is making that leap and hoping that there will be a hand to catch me on the other side.

I talk too much!

I just came from a meeting with a prospective student collaborator and am so embarrassed. I come from a long line of talkers, and unfortunately let it get away from me. Earlier in my scientific life, fear kept me in check, but for better or worse, I have developed into a knowledgeable confident scientist.

I was meeting with a first year graduate student who is interested in our work. His advisor is doing some neat stuff that would solve some of the problems that we have been facing, and he would like to collaborate with us. Ideally the conversation would have been an exchange of ideas: me saying this is what we do and the problems we have, him saying this is what we do and how we can solve those problems, and then me saying great here are the next steps we should take. Oh, if only!

Instead it degenerated into a rant on my part about the project how great it is, sure we have some problems, but they're only a small limitation...and then the worst part...hey let me give you some advice on your Ph.D, unsolicited no less. I talked about the balance between classes/research and how a post-doc would be in a better position to help us (fewer commitments) and how he should be concerned with quick results over innovation (get more funding), etc. I just hope I didn't scare the guy off because he seemed good.

My enthusiasm for mentoring just seemed to run away from me. This is a problem I have. I swear give me one Ph.D. and I seem to think I know how to run the place! Any other Chatty Kathy's out there?

Writing your first paper: Step Seven: Results-General Comments and Tables

This post is part of a series (that is taking a long time to write) on writing your first research paper. I plan to use this information for my graduate students. So feel free to send your thoughts, comments, and hysterical rejoinders. Also, all data is completely made-up so don't quote me on anything.

In my enthusiasm to discuss figures in the last post, I didn't really give too much information on the results section so I'd like to do that now. The results section is sometimes combined with discussion and in this case the general format would be paragraph presenting results, paragraph discussing implications of results. Otherwise, the results section should only present data, but draw no conclusions as to its meaning. Each experiment should be separated with a header. Data should always be presented in the most unbiased manner possible. If there were inconsistencies explain why if possible, but do not attempt to make excuses for your results. Also, if there was a problem with your assay that limits its meaning, this is a place for discussing that (although this might also be found in discussion). Data should be presented by providing the observation, the standard deviation or variance of the observation, the statistical comparison of this observation to others, and the number of experiments that were used to create the observation. Good experimental design will lead to a good results section. It is important to perform positive and negative controls where possible. Using our ice cream example, a paragraph in the results section might read as follows:

Ice Cream Texture
Ice cream texture was evaluated using a viscometer to measure viscosity and also using a double blind taste test. Viscosity measures the thickness of the ice cream and therefore is a measure of the texture. In general, a thicker texture is preferred, although high viscosity can make consumption difficult (e.g., brick-like consistency).Liquid nitrogen ice cream possessed a similar viscosity to that of store-bought (1.96 +/- 0.05 cp for LN2 vs. 1.85 +/- 0.06 cp for store-bought, t-test p value > 0.05, n=5) and was more viscous to ice cream produced by hand crank (1.24 +/- 0.15 cp) and electric machine (1.01 +/- 0.12 cp).

In the double blind taste test, 100 university students age 18-22 were asked to sample each of four types of ice cream. All ice creams were vanilla flavored, made from the same recipe, only varying in the freezing technique. Students were asked to rate the ice cream for flavor, texture, and the presence of ice crystals on a scale of 1 to 10 with 10 being superb/desirable and 1 being inedible. Results were compiled and analyzed using a student t-test with alpha equal 0.05 (Figure 1). For flavor, all ice creams were rated similarly with an average value of 7.4 +/- 0.09 for liquid nitrogen and no statistical difference between it and the other choices (7.2 +/- 0.1 store-bought, 7.0 +/- 0.05 hand crank, 6.9 +/- 0.07 electric machine). For texture, freezing techniques were preferred as liquid nitrogen > store-bought > hand crank > electric machine, with electric machine ranking far below the other three. The difference between liquid nitrogen and store bought was insignificant ( p > 0.5), but both of these techniques were deemed to produce superior texture to hand crank (p < 0.05) and electric mixer (p < 0.001). Only the electric machine was found to display ice crystals (rating = 2.5 +/- 0.09), whereas the remaining three ice creams displayed ice crystal ratings averaging 8-9.

Notice that this description gives very little discussion. The data is presented, with statistics to aide the reader, and some description of the experiment is given. It is also possible to roll some of the experimental details into materials and methods, but I find that most readers skip this section, so I like to recapitulate the basics in the results section. It is very important to not be judgmental about your data. Simply report what you found.

Finally, a word about tables:
Pretty straightforward. Tables can be created in excel fairly easily. In general, tables should contain a header row, which is separated from the data with a solid line, then the data in columns, then a solid line to separate the data from the bottom of the table, and below that line footnotes as needed. Footnotes typically are used to define abbreviations in the table, describe experimental conditions, and point out outliers. Lines should NOT be used to separate columns.

Well good luck, and soon, I promise, I will post on discussion, references, and finally conclude.

Lack of Positive Feedback: The road to self-doubt

My scientific life has been plagued with fear and self-doubt. I'm not sure if this is a woman specific thing, or a general feeling, but I have been thinking a lot about it in the last few weeks. During my Ph.D. my advisors gave me very little feedback. One advisor said good job all the time, but it was so perfunctory we never believed it. The other advisor never seemed to be satisfied with results, always wanting more. In science negative criticism is abundant. We have peer-reviewed articles, grant proposals, and conference heckling, but positive affirmations are rare. In my desire to be successful, I have been emailing my former advisors for advice on starting a lab. After several back and forths, one of them said not to worry, that I would be just fine. That email meant so much to me that I kept it. I think it was the first positive affirmation that I had in 6 years of interactions, and that is pretty sad. I know several students who only see their advisors every 1-2 months. So it is no doubt that we develop questions as to our own capabilities.

My postdoc has been very affirmational. Everyone treats me and my ideas with a great deal of respect and this has boosted my confidence enormously. However, it is still easy to get down, like when I go to a conference and see a similar presentation to my research. One group is working on an alternative solution to the problem that I am addressing, and their work is much more elegant than my approach. Despite the fact that I have been at it for one year, and they have 6-7 years behind them, it is easy to feel overwhelmed. But the more I thought about it, the more I saw the good points of my approach. It is very practical and could be used by a wider selection of people. Additionally, it has a substantial body of literature backing it up, whereas the other approach is being investigated only by that one group. So I feel better.

All of this got me to thinking. One of the greatest problems in motivating minority scientists is probably this lack of feedback. If we already question our ability to be successful due to lack of role models, it is difficult to develop confidence without positive reinforcement. I never had this problem in high school or undergrad. It started in grad school, when I no longer had regularly reported grades with which to gauge my progress. When I set back and reflect, I realize that by many standards I am very successful, but without weekly A's on my homework, it can be difficult to stay focused and see the big picture. This is one thing that I hope to provide in my lab. It should be run more like a business, with regular progress reports and evaluations, and possibly 360 evaluation, where the students tell me how I'm doing as well. The goal is to learn and without feedback how will we ever get better?