Research Philosophy & Guidelines

research philosophy

The Lab’s research is not centered on a specific scientific subdomain but rather spans a diverse array of interdisciplinary subject matter related to weather, climate, and human-environment interaction. The philosophy is that useful research will come about from asking high-level, important questions on topics in which precise knowledge is currently lacking. Due to the complex nature of the subject matter of human-environment interaction, there are great opportunities to make meaningful advances in our research area. Our strength is being able to take the tools traditionally associated with one discipline and apply them to another discipline in creative ways that allow for new insights to be made.

The seeds of an important piece of research often come about from asking simple curiosity-driven questions of the form “I wonder if X has an influence on Y?”, “I wonder how big X’s influence on Y is relative to Z’s influence on Y?”, or “does X cause Y to change or does Y cause X to change, or are both caused to change by a third factor?”. We look primarily for the first-order answers to these questions.

The point of research in our lab is not to demonstrate how complex the world is – complexity is our starting point. Rather, the goal of our research is to cut through the complexity to find the upshot. We want to turn data into readily communicable knowledge.

We are happy to guide student’s research down the paths that we feel are most promising but students are encouraged to propose their own ideas and to follow their curiosity and intuitions. We find that people are happiest and most productive when they are allowed to take ownership of their projects and feel a high level of agency.

Mental health is critical to student happiness and research productivity. Unfortunately, graduate students are at higher risk of anxiety and depression than the general population. We encourage students to take steps to stay mentally healthy. These include maintaining an appropriate work-life balance, making time to exercise, practicing mediation, and engaging with a supportive community of fellow graduate students and faculty. Cognitive-behavioral therapy applied to the stresses of graduate school can also be helpful.

General guidelines

The pillars of scientific understanding are I) observations, II) basic theory, III) experimentation and IV) mechanistic modeling. Students should strive to prominently feature creative applications of at least two of these pillars in their research.

Practical guidelines

1. As a first step in their research, students must gather an appreciation for what is known and what is unknown in their subject matter. A possible pathway to build this knowledge foundation is as follows:

a. Read textbook chapters that underlie the research topic

b. Read published “review” articles on the topic

c. Read seminal journal articles on the topic

d. Read contemporary journal articles on the topic

When students read journal articles, they should take the time and effort to truly understand what they are reading. Casually reading 10 papers for 30 minutes each, will amount to 5 hours of reading papers but true understanding will be very limited and students will forget most of what they read in a matter of days to weeks. Thus, it is better to spend those 5 hours trying to understand a single paper (or section of a paper) as completely as possible. This will allow the information to penetrate into the student’s long-term memory and add analysis tools to the student’s repertoire.

2. Students must obtain the computer skills necessary to perform data analysis. This does not require a degree in computer science but it does require an initial familiarity with a high-level programming language (i.e., Python, Matlab, R, IDL, etc.). Once they have an initial familiarity with a language, further learning can mostly come about “on the job” (in the service of their actual research goals).

3. Once the student has achieved a sufficient level of proficiency in the subject matter and is able to conduct some data analyses and produce plots, they have the tools necessary to create new knowledge and insights!

We encourage researchers to make as many curiosity-inspired plots as possible. This is the “divergent phase” of research where the researcher explores many avenues in which the research could go down. We encourage researchers to maintain a scout mindset in research - allowing the data to dictate where the research goes. At some point, they will inevitably stumble upon an interesting result that shows something that is either unknown, underappreciated or contradicts something that is widely accepted. This could easily be a ‘null’ result (e.g., expecting to see a relationship between two variables and finding none). Another “finding” might be a novel way of showing something that is widely accepted. Regardless of the type of finding, this is the seed of a section of a master’s thesis or a journal paper. After these initial results are found, one possible means of progression is as follows:

a. The researcher should write a Nature-style intro paragraph according to this guide. This helps keep the focus on a relatively narrow question instead of a general area of inquiry. This is also a check on whether the researcher knows the existing literature well enough to begin writing a paper or the thesis section.

The number one problem I observe in scientific research is unclear thinking about what scientific question is being asked and how the research attempts to answer that question. It is very common for researchers to get emersed in the details and to produce a lot of “results” but to ‘miss the forest for the trees’ and be unable to really articulate what the point of the research is. Writing the Nature-style intro paragraph and continuously iterating on it forces the researcher to keep the big picture in mind.

b. The researcher should perform additional related analyses that they are curious about or that would be useful for testing alternative explanations for the data. This entails making dozens of figures and having some system for organizing them (this system can be as simple as putting them in a ~50-slide PowerPoint document). If any of these results fundamentally undermine the initial results or if any of these results suggest unanticipated insights, the researcher is free to start over at step “a” above and revise the story of what the data is telling us. At some point, the researcher will enter the “convergent phase” of research where the focus of the research is decided upon and distilled.

The scientific method is often iterative in practice and there are many instances of looping back and revising:

c. At this point the researcher can actually write the paper or the thesis section, keeping in mind the big picture and narrow focus (from the Nature-style paragraph) and referencing the most relevant (not most convenient) results from the large set of analyses conducted.

The researcher should try to be conscious to structure the writing in a logical manner:

Researchers should also try to think like a reviewer when iterating and improving the work. Reviewers are often prompted to answer questions like the following when turning in their evaluations:

“Do the authors identify other literature on the topic and explain how the study relates to this previously published research?”

“Is the subject addressed in this article worthy of investigation?”

“Is the information presented new?”

“Are the conclusions supported by the data?”

“Is the presentation appropriate for the type of data being presented?”

“Is there enough data to draw a conclusion?”

“Do the results support the conclusions?”

“Do the conclusions overreach?”

“Do the authors discuss any limitations of the study?”

d. Finally, the researcher should make publication-quality figures (i.e., all the panels/labels neat, with consistent text-figure nomenclature, etc.). These figures should be pulled from the most relevant (not most convenient) results in the large set of analyses conducted. Some people will recommend that the main figures should be decided upon before the writing begins. This is up to the researcher but sometimes it is not clear what the main figures should be until one starts writing.

Keep the big picture in mind at all times when communicating the research. Along these lines, AMS and many other journals require “significance statements” (aka “broader context”, “key points”, “plain-language summary”, “highlights”, etc.). This formatting forces researchers to distill their research into high-level takeaways and makes us lift our heads from the weeds & revisit the question of “wait, what’s the point of doing this research?“. These concise clear statements force concise clear thinking. They should not be thought of as annoying formatting requirements for a journal but rather they should be fundamental to how we are thinking about our research throughout the process. Producing pieces of information like these is the point of doing research - to take enormous amounts of messy data and derive bits of understandable knowledge from it.

The other thing that these plain-language summaries emphasize is clear writing. A common problem I see among new researchers is a kind of misplaced pride in presenting huge amounts of results in a complicated manner. Researchers like to show way too many results because they want to show their audience just how much work they did and they like to advertise complexity because it can make the researcher feel smart. But the point of communicating science is to distill what was learned into as simple and concise of a framework as possible so that the take-home message can be communicated to other humans with limited time and attention. Researchers should strive to be in the top-left quadrant below when communicating and writing: