Statistical Consultancy

I spend a lot of time doing statistical consultancy.  When I talk to people about consultancy (and I do frequently – I’m incredibly unpopular at parties) I tell them that a consultation in my field (medical research) is like getting a mini-documentary on a topic I typically know nothing about, by a passionate and intelligent researcher, and at the end of it I get to help. It is a thoroughly gratifying experience.

I have been a statistical consultant for about fourteen years in various guises. My first exposure to consultancy was an undergraduate course where statistics students got course credit for helping other researchers (mostly within the university) with their data analysis. From the start I was hooked. I loved the thought that I was actually applying the actual things I had recently learned to help actual people (actually). At the start of that course we were handed a single A4 page of consultancy tips and then thrown in at the deep end. Tips that I recall included advice to wash and making sure your breath didn’t smell. Nothing I am about to say is more important than those two tips. Nevertheless here are the additional things that I think are important to remember about consultancy.


This guy gets it (

David Hand (ex-president of the Royal Statistical Society) has said

What you need to be a successful consultant: Affability Availability Ability in decreasing order of importance!

It is difficult to overstate how important niceness is. A typical client is a conscientious, hardworking and enthusiastic individual who has been forced to seek help and support outside of their own organisation and  perhaps feels isolated.

They likely want to be able to analyse their own data but feel out of their depth and unsure of what is required of them. The first thing the consultant should do is try to set them at ease and make them feel like they are in the right place for support (this is all predicated on the assumption that you are going to help them and are not a serial fantasist with delusions of statistical competence). This serves a practical purpose, as well as an altruistic one, which leads onto the next point: getting the client talking. Before that, here is what Boen and Zahn (me neither) have to say about it.

The client wants the consultant to forgive them for the many things they know they have done wrong. They want the consultant to overlook the fact that they haven’t seen you before beginning the study and that they are not as good at math and statistics as they want to be. They want you to be tolerant and to not be curt with them for not understanding your explanations. They want you not to be angry when they ask you to suggest a sample size even though they can‘t decide on what a meaningful alternative hypotheses is or what power they want. They are not asking that you agree on all their decisions, only that you be enthusiastic about getting the job done.                                               Boen and Zahn (1982)

Silence is your friend


A common client misunderstanding of the consultant’s role is that they think/hope that the consultant will instantly know what a project is about, what is required for the application to be successful and immediately know the solutions to all of the client’s problems. The client will usually be very familiar with their project and their area of expertise. They may assume that you are equally familiar with them.

Getting the client talking is important as they need to communicate to you what the question is and how you can help. Indeed, the sole purpose of the first consultation is likely to be eliciting a well-specified hypothesis or research question. The easiest way to do this it to say very little. People hate silence and will do anything to fill it. Use this to your advantage. In any case, until they tell you something about their problem you genuinely won’t have much to say.

Needless to say, when the client is talking you should be listening. I mean actually listening. It sounds simple but this is not a natural skill for most. Take note of questions you want to ask, but try not to interrupt. Clients will often answer your questions if you give them enough time but interrupting may throw them off their track and they may forget things. It is important not to rush this part, within the constraints of the meeting time (it is rare that a consultation is productive beyond an hour and a half). The client should feel they have the time to fully describe their situation and area of interest. It also helps prevent the consultation degenerating into an unproductive conversation such as “I am interested in studying cancer; how many patients do I need?” (this is barely satire). If not enough time is spent on specifying the problem it will result in much more time wasted later on.

Assessing the client

Clear language
Retaining a manifest lexicon is paramount

As a consultant you should be tailoring the advice you give to the needs of the client. Listening to the client will help you assess how much support and advice the client requires and in what areas. You will be able to pick up on the terminology they use, and use it back to them. This will help avoid jargon the client doesn’t understand and improve the communication between you both. In this situation, you are the professional so you need to meet them more than half-way.

I remember one consultation where it took me 45 minutes to realise that I was using the term “mental health measure” to mean a tool for measuring mental health, where everyone else in the room was using it to mean “mental health intervention” (as in a measure you would take to improve mental health). I think it is fair to say that my contributions to that point were less than pertinent.

You should also change the solution based on the needs of the client. It may be that a complicated statistical model or a multi centre trial is what is required to answer the question, but that may not be feasible for your client. Alternatives may need to be explored. Your suggestions and solutions should reflect what is suitable for them and their needs.

Write notes

I would recommend a smaller pencil for note-taking
I would recommend a smaller pencil for note-taking

In a consultancy situation there should be lots of new information for the consultant to take in. Taking notes will help structure the conversation but also help remind you what was discussed. In addition it will help you to remember the terminology you should use. Also, it is hard to talk and write at the same time and remember, you’re not supposed to be talking.

You should record what was discussed as well as any actions that both parties have. You should check that the client agrees and understands those actions. It is good practice to summarise this in an email to the client shortly after the meeting.


You are not obliged to work on anything that you disagree with morally or ethically. It is extremely unlikely to happen but if you are uncomfortable with the subject matter then say so. Perhaps a bigger danger is that you are concerned with the scientific conduct of a proposed study. If you feel that the scientific integrity of a study is compromised you should point this out to the client.

The trolley problem: a classic ethical dilemma. This specific situation is unlikely to come up in a consultation
The trolley problem: a classic ethical dilemma. This specific situation is unlikely to come up in a consultation

There are ethical considerations for the consultant also. As a consultant you are in a position of authority and the client may think that you will be able to solve all of their problems. It can be easy to feel that you are the expert who must be able to answer every question the client has. If the client does ask you something you are not sure the answer to, don’t bluff. It will always come back to bite you if you say you can do something you can’t. You will be stressed and they will be disappointed. There is no need. A good consultant will let the client know that they simply don’t know. This can happen as often as it needs to. What you may be able to do is tell them that you can find out (and then do find out!). You don’t have to have all of the answers to be a good and effective consultant.


2 thoughts on “Statistical Consultancy

    1. Hi Andrew,

      I learned my statistical consultancy on the streets (!) so haven’t actually read anything about it. Will definitely check this out though.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s