Hello all,
I would like to add to the discussion about statistical consulting and authorship within an academic environment.
I provide statistical advice to researchers (graduate students, faculty and staff, and occasionally undergraduates) in the College of Agriculture and Life Sciences, North Carolina State University. I am supported by the college and researchers are not charged for the help provided. In many instances I participate as coauthor because of my involvement with the data analysis, and preparation of the manuscript, mainly Material and Methods section and presentation of results. Many times the quality of the statistical analysis enhances the presentation of results and discussion in a paper. Other times, having a statistician as coauthor gives certain confidence that analysis are done properly, specially with researchers who do not have a strong statistical background. Some of them feel that it is through analyzing their own data that they learn statistics, and I am glad to help them to gain a deeper understanding of statistical methods and inference processes. In other cases collaboration starts at early stage, and continues through the whole research process until publication of results. As part of a team applying for a grant, I will get some money, if research is funded, that will go directly to cover my salary while I dedicate a portion of my time to this particular research project. Depending in how well I estimate the time needed to complete the analysis, I may or may not need to work extra hours without pay.
Researchers are very grateful of the help provide by the statistical consulting group within the Department of Statistics at NCSU.
I agree the importance of a clear and honest communication between client and consulting statistician and will add the shared responsibility of getting the best statistical analysis within the context of the research data at hand.
I am enjoying this conversation that has derived to academic statistical consulting and will refer to some ideas discussed here to students in the statistical consulting seminar at NCSU this fall.
Thanks,
Consuelo
-------------------------------------------
Consuelo Arellano
North Carolina State University
-------------------------------------------
Original Message:
Sent: 08-12-2010 18:40
From: Manuela Huso
Subject: consulting contracts
All,
I'm really enjoying this exchange and wanted to add a little from my experience. I am employed by a single department in the College of Forestry at OSU as a consulting statistician for our faculty and students. For the most part, I do not become sufficiently involved in most projects to warrant co-authorship. Most of my work involves translation of research questions into statistical models and advising faculty and students about which statistical approach might be best to answer their research questions. Only when analyses are sufficiently complex and interesting to me do I move from consultant to collaborator.
Our Quantitative Sciences Group (QSG) has developed the attached document in a similar vein to the one shared by Lori. I attach ours to point out what I think is are two important differences that are relevant when our involvement as consulting statisticians is limited to giving advice. First, we clearly state the minimum level of mathematical and statistical background we expect of our clients in order to carry out a meaningful conversation with them, and 2) we emphasize that the ultimate responsibility for the appropriateness of the methods and the actual implementation of the procedures lies with the client.
We can talk all we want about multi-dimensional space, log transformations, etc. but if the client can't even tell you what a slope parameter means in a simple regression (real case J), then I'm not sure we are really able to communicate, nor am I sure that the client really can carry out the suggested analysis. So, minimum math/stat background is pretty critical.
If our involvement as consulting statisticians stops at the point of simply giving advice on potential approaches, it is very possible that we have not been provided all important pieces of information or perhaps we have misunderstood exact conditions that would become clear to us if we actually worked with the data ("Oh, yeah, I only measured that treatment for 3 weeks, but all the other for 5 weeks... I forgot to mention that.", etc.). Of course, with this seemingly insignificant (to the client) piece or information, our advice might be completely different. But how could we know? So, making sure that the client understands that the advice is the best it could be conditional on the supplied information is very important. The ultimate responsibility for the analysis must rest with the client.
Manuela
-------------------------------------------
Manuela Huso
Consulting Statistician
Department of Forest Ecosystems and Society
Oregon State University
-------------------------------------------
Original Message:
Sent: 08-11-2010 10:34
From: Lori Price
Subject: consulting contracts
Hi Michael,
This was developed a year or two ago and has been very useful for both us and our investigators. I know that many biostats consulting centers based in universities have similar information on their websites, because we looked at those when developing our FAQs. Unfortunately, I don't remember specifically where we looked.
I have attached our FAQ.
-------------------------------------------
Lori Price
Tufts Medical Center
-------------------------------------------