Research process models

Image: teapot and mug, handbuilt

Oftentimes in school, we're taught how to do research (or science labs) in a linear process that passes through discrete stages in a specific order. At the end of the process, you're supposed to "write it up" in a lab report or scientific paper. However, this stage model of research is a lie. In this article, I articulate two models for doing research, and explore their implications.

Research stage model

The six stages of research are represented as a row of rainbow colored boxes.  The stages are described in the caption.

In the research stage model, first you write your research question, then you do a literature review, choose your theory, get data, and analyze the data. Finally, you write up your results. These stages are supposed to be done in order, moving from left to right on the image.

Under this model, research passes through discrete stages. The progression of the stages looks a bit like sections of a research paper, and it echoes the scientific method that's often taught in schools. Each stage in this model has defined tasks. For example, when you're doing a lit review, you gather papers, read them, and synthesize. Getting data is a separate stage that finishes before any data can be analyzed, and all data must be analyzed to generate your conclusions before you write up your papers.

It's easy to teach a stage model because it feels simple and linear: the stages are distinct and well-determined, and they progress in an orderly fashion. It feels comforting to plan a project which progresses through these stages, because you can look at your calendar and know when you are ahead or behind. When we're teaching people about doing research, we often simplify the research process so that it more closely mimics this model. Research-esque activities, like undergraduate labs (including course-based undergraduate research experiences), undergraduate "capstone" courses, and even some masters theses, can take up this model for doing research and use it to determine preliminary and intermediate deadlines: by this day, turn in your draft lit review; by this day, turn in your full data set. As instructors, we often simplify these research-esque projects in order to support this stage model of doing research.

However, the research stage model can create blocking tasks which impede progress on the project as a whole. Blocking tasks are ones where you can't do them (because you don't know how, because you are waiting for someone else, because they feel insurmountable, etc), but also you cannot avoid them and do something else. They block all work on the project until they can be resolved.

As research projects get bigger and more complex, there are a lot more opportunities for blocking tasks to emerge. For example, what if you are collecting data at three field sites, and one of them can't host your visit until three months later? Or -- this is super common -- you don't want to start writing until you've finished all your analysis, but there's always one more exciting avenue to pursue?

The blocking tasks problem is especially bad for emerging researchers, because they're more likely to only be working on one project at a time, and because there are more things they don't know how to do yet.

Additionally, because the research stages are sequential, if you need to "go back" to a prior stage, that can feel like a failure. Suppose your data analysis doesn't align well with the lit review you did ahead of time. If you need to read new papers to understand your data or bring in another theory to explain it better, you might feel like your research has a major setback or like you don't have time to complete your project in the original timeline. This is a problem for emerging researchers because it can negatively impact their self-efficacy as researchers more strongly.

Parallel processes model for research

The same research stages as in the stage model, but arranged in a rainbow-colored braid.

In the parallel processes model, working on each part of the project suggests improvements and amendments to the other parts. Throughout the project, you change your activities depending on what you're doing and what you've learned.

Under this model, research processes are overlapping and generative. Each analysis that you do suggests new data for you to collect or new literature for you to read. Each paper that you read suggests new analyses to perform or data to collect. All throughout, your research questions are living questions: they grow and change in response to what you're discovering. Instead of writing up your paper at the end, you engage in generative writing to help you record your results, generate new ideas, and document your work for your papers.

The parallel processes model for research can seem intimidating at first, but it creates fewer blocking tasks than the stage model. Because you're doing a little bit of each thread at a time, if you're stuck on one aspect of your project, you can work on another aspect to help unstick you (or to fill your time while you're waiting for feedback). As you learn more, you can do more. This allows your research progress to grow with you as you learn and develop as a researcher.

Plus, the parallel processes model for research is more honest than the research stage model. This is how research actually gets done in big and complex projects (and small exploratory ones). It's very common for experienced researchers to need to go back to the literature in the process of doing their analysis, or for inspiration for new data to strike while they're reading a paper. Generative writing is good practice for research, and while it's not required to write papers, it's certainly more productive than the stage model.

Emerging researchers (from undergrads to faculty new to DBER) tend to have one of three major responses when they learn about the parallel processes model:

  • omg, I’ve never heard it explicitly like this and I feel so seen

  • whoa, that’s allowable? mind. blown.

  • hmm, this feels less deterministic and therefore more scary.

If you see yourself in one of these groups, you're in good company.

Applying the parallel processes model

Something that is both great and terrible about the parallel processes model is that it does not prescribe where to begin the research process. You can start in any of the major strands of research and trust that your work will integrate activities from all strands as necessary and relevant.

If you're new to research, this feature can be scary.

Where should you start? Here are some options:

  • Read something! When you pick up a research paper or popular press article, ask yourself what's exciting about their study. The claims? the methods? the theory? the population? Doing a lit review can help you do this in a more structured way.

  • Write something! What are you interested in? Why is it interesting? Engage in some generative writing around a problem you see, or work on a reflective statement of research interests to help structure your thoughts and suggest a new study.

  • Notice something! Oftentimes as scientists, educators, and humans in the world, we see something interesting or unusual or problematic. Allow yourself the curiosity to ask questions of this thing you noticed: why does it do that? what would happen if it were different? The Research Process for Video-based Research article covers a formalized method for noticing and refining our noticing to generate a research project using classroom video.

You might notice that all of these articles push you to do generative writing in order to figure out new ideas and refine them. That's because generative writing is a major research tool. It can help you improve at doing research, no matter what your skill level is.

Once you're started, what comes next?


Research models and research advising

If, as a research advisor or lab instructor, you are strongly committed to the research stages model, you might find yourself simplifying the research-esque projects that your students work on, so that they are less likely to develop blocking tasks or sequencing setbacks. As long as this choice supports your learning goals, that's probably ok. It works pretty well in instructional labs where the major goal is for students to learn something science-y with equipment. It also works pretty well if your primary interaction with research is reading published papers (e.g. for a journal club), or if the research work you engage in is primarily replication or repetition (e.g. for a CURE).

Alternately, if your students are engaging in original research, you might need to reconceptualize their projects in order to take advantage of the parallel processes model. You need to figure out reasonable timelines for their work and help them integrate each strand in these processes, from collecting pilot data before they're "done" with a lit review to making space for preliminary analyses early enough to return to the literature. You also need to teach them how to turn to another part of the braid while they wait for your feedback, so that you are not a source of blocking tasks.

You will also need to work to help your students understand why you work this way, instead of in the more familiar stage model. By the time my students come to my lab, they have already engaged in many years of instructional labs (and sometimes also research-esque projects) in which the projects were simplified into fitting the research stages model. Often, they have already internalized feelings of failure when they need to return to literature during their analysis, and they have already experienced the paralysis of blocking tasks.

There are a few articles on zaposa to help you figure out how to mentor research with emerging scholars (singly and in groups) from a parallel processes model, or think more abstractly about how this works. I'm open to new suggestions: what else do you want to see?

Last update 2022 July