Consulting Resources

This page lists resources for our student consultants to help them lead consulting sessions.

Consulting: Modalities

• Registration: Students in the Department are required to sign-up and enroll for consulting for at least one or two quarters (or more if desired), depending on their program. To ensure adequate enrolment during each quarter, PhD students will have to indicate and rank their preferred quarters for consulting at the beginning of the academic year.
  Requirements: 2 quarters for PhD year 1-3, 1 quarter for PhD year 4, 1 quarter for MS (year 1 or 2).
• Participation: Registered students are required to attend the consulting seminar meetings in the quarter (Wednesdays at noon).
  All other statistics graduate students are welcome to every consulting seminar meeting, as well as participating in any consulting projects.
• Consulting: In a quarter, consulting projects will be assigned to students registered for 44100 (no tuition involved).
  Registered students are assigned to time slots as teams, based on availability and determined at the beginning of the quarter in the first consulting seminar meeting.
  For example, a team of four students may be responsible for taking care of all consulting appointments for a given time slot (eg, Wednesday 2-4 pm.) Researcher clients are asked to sign up to a time slot when submitting a consulting request. The student team responsible for the time slot will meet the clients at the time (in-person or over zoom). The teams will discuss their consulting projects at the consulting seminar meeting on Wednesdays.
  More in-depth collaborations with clients can be arranged for projects deemed adequate after the first initial meeting.

Q &A: How to lead a consulting session

What is expected of me during the session?

As a student consultant, you are expected to (a) lead and contribute to the conversation with the client and (b) participate in the discussion during the Wednesday seminars.

• During a consulting session: During a typical clinic consulting session, a client will come to you, typically with a question relating to their own research. To give proper advice to the client, you should start by asking questions to understand the goal of the analysis, as well as the nature of the data at hand. The client will then typically proceed by showing or explaining what kind of analysis they have done so far, why they are dissatisfied, and why they came to the consulting program. Based on the client’s explanations, you should try to give advice and input into what should be the next step. It is not required nor advised that you take the client’s data and perform the analysis for them: one of the goals of the sessions is for the client to have learned something about statistics at the end of the session. However, if the project that your client is describing is involved, and requires a substantial adaptation or improvement of off-the-shelf statisitcal methods, do bring it up at the Wednesday seminar, and we can arrange for a group of volunteer consultants to collaborate with the client.

• After the consulting session: You and your team should write up a small summary of the session. These notes should be detailed enough to be able to accurately describe to the class the problem and what you have suggested during the Wednesday session. You should also send these notes to the client, copying Prof. Wang and Prof.Donnat in your email thread.

• During the Wednesday seminar You are expected to contribute to the consulting discussion that week. That is, you should have prepared This will be as well an excellent occasion to filter out projectsthat could be fit for additional more involved collaborations with student consultants. It is also the time and place to discuss cases that were particularly difficult, and where you would like to get the class’ feedback. Based on the outcome of this session, if you describe the case well, you should be able to get back to the client with better recommandations.

How do I lead the session?

To lead a session, you should simply gear the conversation to gain as much information as possible from the client in order to help him:

• 1- Briefly introduce yourself (10s, the goal is really to break the ice), e.g,: “we are PhD and Masters students in Statistics.”
• 2- Make the client introduce themselves, the context of their experiment/project, and the type of data that they are analyzing
• 3- Make the client explain what the purpose of their analysis is
• 4- Make sure to understand what the client has already done, and why they need help.

When you ask questions, remember that this is an exchange between the client and you, and you are trying to teach the client the way you think. So, rather than asking a set of questions such as “How many data points do you have? Are the data split into batches?”, try explaining as much as possible the rationale behind your questions: “How many data points do you have? Knowing the number of data points is important in determining the test that you will be able to use, so for instance, if you have too few (let’s say, less than 30), you might have to use a non-parametruc test. Are the data split into batches? This is important for us to know if we have to add a “batch effect” in our analysis. A big component here is your ability to explain clearly to the client the rationales and subtleties of the analysis that you are suggesting.

What if I do not know what to suggest to the client?

This is bound to happen, and you should not worry!!

• 1- Take notes during the session and bring it to the class the next Wednesday, so that we can solve it all together. If you do not know what to suggest, be clear with the client — they know you are a student and doing this for free. You can say for instance: “I apologize, this really falls beyond the scope of my expertise, this is a non-traditional problem. But let me bring it to our consulting weekly discussion session, and I’ll get back to you with further recommendations.”
• 2- Explain to the client why you are stuck, and give the client something to do in the meantime. For instance, you could explain how the client’s setting differs from traditional settings for reasons X, Y,Z. It will always be reassuring for the client to at least understand why you are stuck. You can also recommend doing further Explanatory Data Aanlysis, drawing plots, making visualizations, etc, to have a better idea of what the data looks like. With a little bit of luck, the client can rapidly send you the plot, and we can use it in the Wednesday discussion.

What is the student consulting cup?

The consulting cup is a friendly competition amongst consulting teams to be elected “best consultants” of the quarter. The winning team will get a certificate during the final consulting seminar of the quarter. Individual consultants will also gain points, that will count towards the “best consultant of the year” award handed out by the department each year. Individual consultants and teams can gain points by:

• Leading consulting sessions well (client satisfaction survey)
• Participating in the Wednesday Discussion
• Helping out with the consulting website.

Task	Points for the Team	Points for the Consultant
Client Satisfaction Score	+ 10 \(\times\) score	+ 10 \(\times\) score
Client Special mention of a consultant		+ 10 \(\times\) score
Discussion of teams’ cases during the Wed. seminar	between -20 and 20
Participation to the Wed. seminar		+ 5 for each intervention
Help with the website		+ 10
Hands-on collaboration	+ 50	+ 50