Generation, iteration, reflection

Image: berries in the US Virgin Islands.

This article explores three core elements of successful research. They tie together many of the zaposa articles about doing research, and they are foundational to my research practice.

If you've read several zaposa articles and you're looking for deeper connections, you're ready to read this one. If you're just getting started with these resources, it might be more fruitful to start with generative writing, lit reviews, or making beautiful posters.

Core elements of successful research

Let's look abstractly at the parallel processes model for research. The parallel processes model is a pretty high-level look at how to do research and grow as a researcher. At its core, the parallel processes model asks you to iterate through each kind of research activity (represented as strands in the braid) in order to generate new ideas from reflecting on other activities. As you know more, you can do more.

Let's look abstractly at the paper revisions process in Writing Better Papers. The three exercises in that article are quite directive: do these things, and your paper will improve. Taken together, they ask you to reflect on your paper to generate a reverse outline and restructure the paper, then iterate through each piece of your outline to signal your argument and refine your word choices. The revisions process is, itself, an iteration of your original writing to make the first draft.

These three core elements -- generation, iteration, and reflection -- are deeply foundational to my practice of doing research. Let's look at them explicitly:

  • Generation: as we learn more, we generate new ideas to pursue. It is important to build opportunities to generate new ideas throughout the research process: though reflection, through playfulness, and through communicating with others.

  • Iteration: as we learn more, we return to old ideas to refine them. It is important to plan for iteration throughout the research process, so that that we're iterating through each kind of activity more than once. Iteration is not just doing the same thing again; to iterate, you need to make changes or improvements each time.

  • Reflection: if we want to learn more, we need to reflect on how our our ideas grow and change. Recording new ideas will facilitate later reflection; reflecting on old ideas helps with generating new ones. Reflection allows iteration to refine ideas rather than simply repeat past activities.

These three core elements appear at many different scales in my research practice, and they're so fundamental to how I think about research that it's almost impossible for me to think about doing research (or learning how to do research) without them.

Are these core elements necessary for successful research?

Yes.

However, we don't usually talk about them as if they are ordinary elements of everyday research practices.

When we talk about our research projects, we don't usually talk about how we generated our ideas, except maybe as a clever little aside in informal storytelling. We don't usually talk about how we build opportunities into our ordinary practices to encourage idea generation, and we especially don't talk about the overwhelming number of ideas we generate that don't work out. There's a certain narrative of brilliance which supports only talking about the ideas that work, and which attributes the creation of wonderful ideas to especially talented people. I think idea generation is necessary for doing research, but the mechanisms for doing it are opaque and the narratives around it are needlessly exclusionary.

In contrast, iteration is explicit in our research discourse in several ways. A common research design is to pilot an experiment or activity before going to scale -- this is iteration. Writing a draft of a paper then revising it? Also iteration. Presenting your work first to a research group, then in a conference poster, and finally in a journal article is another kind of iteration in the research communication process.

Of these three core elements, discourse about reflection varies the most widely based on what kind of research you're doing. For some kinds of qualitative research, reflection is an explicit and central part of the research process (for example, in this research process for video-based research). Not only do your reflections form part of your research record, they are supposed to shape your analyses. In contrast, for some kinds of quantitative research, reflecting among your initial data analyses to figure out how to proceed is fraught with difficulty. You have to carefully separate phases of data collection, analysis, and reflection lest you fall into p-hacking. Nonetheless, even in these quantitative research designs, researchers need to engage in reflection to learn from prior studies and shape future ones.

More broadly, the ongoing work of being a successful researcher involves a lot more than just doing one research project from ideation to publication. Maintaining a research program across multiple projects requires a constant stream of generation, iteration, and reflection to figure out new ideas and processes, improve on them, and set priorities and directions. If your primary engagement with research is through reading published papers, you probably don't see all of this work. If the scope of your research project is small, you might not engage with this aspect of research work very much.

Apply core elements to research activities

These core elements are so baked in to all of my research work that it's really hard for me to think about research without using them. However, they're not always apparent in how I've written the articles here.

The following resources and articles on zaposa take up the core elements and apply them across many different research activities. I've grouped them here based on their audience; look to the main articles page to see them grouped by kind of activity.

Articles for emerging researchers (and their advisors)

These pages are written for two audiences: established and emerging researchers. If you're a new researcher -- or an experienced researcher working in a new research field -- these are for you! Alternately, if you advise students (undergrad or grad), you can share these with your students to help ground their work.

  • Generative writing is a fundamental tool in my research toolbox. It's so important that it's the whole purple bar in the parallel processes model. Generative writing can help your students connect their emerging ideas about literature to possible analyses to pursue, or understand what's going on in their data, or collect together ideas to form a paper. In particular, the section on "How do I collect my writing into a paper" addresses the non-linearity of the parallel processes model directly. This is a great article for everyone at any level.

  • Lit reviews looks more directly at the part of the parallel processes model that intersects with literature, pushing back against the stage model's implication that a paper has a single lit review as well as giving some practical advice for how to actually engage in reading and understanding published literature.

  • Beautiful posters uses iteration, generation, and reflection to help you refine your argument for a poster and then represent it graphically. They're entirely implicit in this article, and it's appropriate to share this with undergrad researchers in any STEM field -- I know some faculty who use it in their undergraduate capstone research courses.

  • Statement of research interests leans heavily into generation and reflection to help you figure out what you're interested in and why. The primary audience for this article is grad students and postdocs who are starting to think about their future research directions; faculty seeking tenure, promotion, or a new job will also find this helpful.

  • Writing better papers has three specific activities to refine and revise a research paper, assuming that you already have a rough draft. This article (and the companion on writing as a generative process) were the result of a multi-year project to understand how to teach writing for scientific papers to graduate students whose first languages are not English.

Articles for advisors

These pages are written for advisors of emerging researchers, or for emerging researchers who are also advising students. They might not be as useful to read if you're a student, but sometimes I share them with my students so that they can think more about the work of advising or reflect on the activities we do together.

  • Research process models introduces the parallel processes model and explicitly talks about how iteration and generation are productive for the practice of doing research as well as for learning how to do research.

  • Research design takes the parallel processes article one step further to focus on how to design research projects from the perspective of aligning your research questions, theory, access to data, and research methods. There are guided exercises to iterate through these three aspects from a design perspective.

  • Research with undergrads takes up the parallel processes model and applies it to an entire research program in which undergraduates conduct research. For each student, it explicitly picks them up in the middle of the stages model, and acts to scaffold in all of the other stages in an as-needed basis.

  • Research process for video-based research is specifically focused on the activities and products of getting a student started in their first video-based research project using the more general advice from Research Process Models and Research with Undergrads.

Where do these core elements come from?

Story time!

I started writing these zaposa resources before I articulated these core elements. It turns out that collecting these resources together is generative for me, and the process of iterating through each of them eventually turned into reflections on connections between them.

In summer 2022, I had a series of conversations with PEER participants and directors, some of my students, and some farflung collaborators around principles for research. I realized that my reflections were generating connections for me, but that I hadn't shared those connections well with others. In conversations, my friends and collaborators helped to refine and articulate these ideas.

Where did the core elements come from? They came from years of iterations in teaching research skills to emerging researchers, generative writing about research, reflections on my work, and conversations with others. They came from themselves.

Did you invent them?

Ah, no. These core elements are intrinsic to many people's research practice, and other people have written about them (or related sets of core elements) extensively.

Why not revise all the zaposa articles to include the core elements explicitly?

I don't think this is a valuable goal.

If all you need is a collection of articles about key research activities, then you might not want the overhead of thinking abstractly about common elements across them. To find value in this collection, it is not necessary to foreground generation, iteration, and reflection.

Furthermore, if all you need is one or two articles to support some other work that you're doing, then the additional overhead of abstraction might be counterproductive. For example: there are huge books on how to make beautiful posters; this one article is appropriate to assign to undergraduate researchers in STEM fields so that they can think about making (and presenting!) their first poster. It's not too long, it's not reliant on any particular software, and it doesn't require any particular artistic skill. While it relies heavily on iteration, generation, and reflection, teaching those ideas explicitly as important skills would obscure the central goal.

Also, I don't want to overstate the importance of these core elements. I mean, sure, they're everywhere in my research practice. But some of these articles aren't really about the practice of doing research. Determining authorship on papers is an activity for researchers, but articulating authorship order models and their reasoning isn't. Thinking about iteration, generation, and reflection doesn't help me have a conversation with my coauthors about determining authorship order, other than in the vaguest sense of "hey, let's check in on this several times".

Last update: July 2022