Working with scientific collaborators as a grad student

Eric F. Lock(with notes from Will Thomas)

RA setting• Does applied biostatistics under supervision

of Biostat faculty and external investigator(s).• Level of interaction with investigator(s)

depends on faculty supervisor and your performance

• Later in PhD program, perhaps find RA position in another department: work directly with investigator(s)

Phases of a collaborative project• 1. Understanding the data, finding and

resolving its problems– May take 50–75% of time for “analysis”

• 2. Performing the analysis, reporting results

• 3. Writing the paper• 4. Responding to referee reports about the

paper; resubmitting the paper

Be prepared

• Read the study protocol (if available): know how the data was collected and processed

• If the investigator has published on this subject, read the paper

• Try to learn some of the science

Listen carefully and critically• Understand issues motivating the research,

and specific questions for your analysis.

• Ask questions.

• Investigators may present the issue as a particular statistical problem– Check assumptions carefully

• Or, they may have little / no understanding of statistical concepts.

The data may have problems!

Normal individuals

Diseased individuals

Me

Carefully present results• Write a report to detail data processing and

analysis• List quantitative steps in enough detail to

replicate analysis • Interpret results, don’t just list them

Writing the paper• Learn the structure of papers in medical

journals.• Statistician is responsible for figures and

sections on analysis methods, results– Altman, et. al., The revised CONSORT statement

for reporting randomized trials: explanation and elaboration. Ann Intern Med. 2001; 134:663-694.

– http://www.consort-statement.org/

• • Learn to make publication-quality plots in R

Communication challenges

• They assume you know more about their science than you actually do.

• You assume they know more about statistics than they actually do.

• Try to avoid interactions like this:– http://www.youtube.com/watch?v

=PbODigCZqL8