Thursday, April 12, 2018

Project Management in Academia 101: Managing your own project vs. Managing a lab

After discussing why you should care about Project Management (PM) in academia and that there are multiple PM techniques available, we can start getting down to business talking about possible applications. There is a big leap between managing an individual project and managing a lab. If you start using PM techniques early, it will make it easier to juggle managing multiple projects later, or just changing your role and transitioning into a different type of PM role in pharma or the private sector.

Managing Yourself

Let's start with managing your own project. If you have read the previous posts, you already should have some ideas on how to apply PM techniques to your daily work. If you haven't read them, click on the links above and come back in a bit.

If you are a student or postdoc, managing your project is one of the most difficult things you need to learn, often through trial and error. While PIs must have some elements of PM to hit all the deadlines necessary to submit papers, grants, and job applications for getting and keeping their jobs, most are not trained PMs and just expect you to figure stuff out like they did. Unless your boss is a total control freak that needs to double check everything, they will be ecstatic if you come up with a clear question and experimental plan (which includes deadlines for deliverables on their calendar). They will discuss ideas with you, then get off your back, tell you to ask for advice if you have any problems, and wait for the data to flow in. In some labs, especially in the US, the boss may just let you do whatever you want, and there you really need to know how to manage your time and experiments!

In the first post I outlined how to Initiate and Plan a project leading to a research paper and what are the necessary questions to ask yourself, but here I would like to delve a bit more in the mechanics of Executing and Monitoring. Because science is a recursive loop of experimenting, troubleshooting, and analyzing data leading to more experiments, the Agile PM approaches are a lot more adaptable as my guest blogger Duc Phan presented. Kanban, which is Duc's method of choice is a great place to start. Kanban meaning "card" in Japanese uses visual cues such as cards, diagrams or flowcharts to outline the project. Having tasks clearly ordered in "to be completed", "in process" and "done" is a great way to start and he discusses a great way to use the Trello software to do this here.

As you perform experiments, you need to think at two levels: 1) questions/hypotheses and 2) actual experiments. Anyone with some training can do tons of experiments, but one of the most important skills in science is to identify important questions and design the best experiments to answer them with appropriate controls. You will save yourself a lot of time and grief if you remain grounded in the big picture. So I recommend that you title cards with the question the experiments are going to answer, e.g. "What is the developmental expression profile of gene X?" "Does transcription factor Y repress activity of gene Z?". At the end you should have an image or a figure with data that answers the question, leading to other questions. Asana is another software that will allow you to organize experiments under specific headings. To get started you can also just print out an empty monthly calendar and plot out your month, or do a thorough job on Google Calendar in planning your weekly experiments. Blocking time for specific experiments and scheduling things around seminars, meeting and lectures, is one of the ways I figured out how to organize my day in grad school. What else can you do during that 2hr incubation?

This leads easily to a process we use in my lab to write papers which is also based on cards, but borrowed from screenwriting, which is storyboarding on index cards. Each piece of data/answer (graph or images) is pasted on an index card or printed out on a sheet of paper, then attached on a white board where we look at all the data to decide how to present the story. Experiments are rarely presented in chronological order in an article, and are often organized to fit a narrative. Some supporting data will end in the supplemental information and other data will need to be showcased in the first figure. Some data will be set aside because it may belong into a different paper. (Daily reminder on rigor: no data should be set aside because it contradicts the main hypothesis). By rearranging the cards in front of you, you will see how the data fits together in figures and where there are holes that need more work, leading to another round of traditional Plan/Execute/Monitor or agile Speculate/Explore/Adapt.
From Matt van dar Meer (U Waterloo)

Managing a lab

Once one moves into managing a lab, the requirements of the job often change dramatically. After the first few years time to do experiments declines, and you are balancing teaching, admin work, travel, writing grants and papers with tons of other random things you never thought could be part of your job. Deadlines come at you without interruption, and sometime without much warning. So there are two requirements now: managing your own job and managing everyone involved in the research which is people in the lab, collaborators and research administrators. I was reminded recently on Twitter that having a 5 year plan really comes in handy and I wrote about it a while ago. Briefly, plot out month by month your next 5 years: conferences, grant deadlines, desired paper submission deadlines, beginning/end of grants with progress reports, and everything else you need to get to the next level.

Then, your management style really depends on you and on your lab culture. In general, as a new investigator you will almost never be able to get postdocs that were just like you, so you will have to also budget substantial amounts of time for training and budget time for trainees to learn how to think on their own. While you may need to be patient at the beginning when you just want to tell them what to do and they ask for your opinion on everything, teaching to trust themselves and develop their own ideas, will pay you back in spades in the future when they decide to ignore you for a month because they need to get s--t done. As an advanced PM, it will be really up to you to move between traditional sequential Waterfall methods and recursive Agile methods depending on the project that needs to be completed and the stage it is at. One simple approach may be to have a separate project outline for each grant and then one for each manuscript that would fit into it.

I want to spend some time talking about Scrum. Scrum is a PM method that combines both traditional and agile concepts to get a project done in small parcels. The advantage is that there are both a long term goal, but also short term goals that give the group continuous sense of accomplishment. Each task is completed within a 2-4 week sprint which is managed sequentially like a traditional project. I find this particularly suitable when you rapidly need to generate preliminary data or when finishing a paper, because it adds a sense of urgency and focus. Responsibilities are divided between the Project Owner, Scrum Master, and Team. The Project Owner inititates and plans the project, but the Scrum Master is in charge of the sprints and of getting the Team going. You can easily imaging your Scrum Master to be a senior grad student or postdoc. I'm still working on really understanding how to adapt Scrum to the lab, but here is a nice little guide that can get you thinking.

Overall, having a plan and a general sense of whether the "lab machine" is humming or stalled will hopefully reduce your stress level and free time for other things you need to do as a leader, including going on vacation.

2 comments:

  1. Very nice! Thanks for writing and sharing this!
    I think that the Scrum-way can be very interesting in academia/research, but I have one big reservation; which, however, will probably affect *every* PM model.

    -> How to provision delays into this?

    Especially in research, where we constantly venture into the unknown, delays are inevitable.
    Sometimes, the delays are due to things that simply take longer, e.g., a computational task is time consuming and having multiple samples basically multiplies the delay.
    Other times, the delay is due to some block in the thinking/understanding what is actually going on!
    These are things that are extremely hard to "accelerate" and maybe even a larger team will be challenged by this.

    Maybe this is one of the biggest challenges in PM and I am asking for the "impossible" but, in any case, your thoughts on this are greatly appreciated!

    Best,

    Cedric


    ReplyDelete
    Replies
    1. Addendum: By "These are things that are extremely hard to "accelerate"", I meant the *blockades*, not the long-running-but-largely-straightforward-tasks.

      Delete