Return to Biostatistics Ethics Case Studies

(PLEASE NOTE: On the bottom of this page, you will find the case study we are discussing and the name of the contributor.)

TITLE Case Study:Three against one - conflict over flawed analysis


PaulWakim - 08 Feb 2013 - 06:39

The first thing I would ask myself is: how serious of a flaw are we talking about? Is it a different option in the statistical software that gives a slightly different answer? Is it a difference of opinion about the assumed distribution with no serious consequence? Or is it a much more serious flaw? One way to find out is to see whether the results of the claimed \x93flawed\x94 and \x93correct\x94 analyses are consistent? It they are, then we\x92re probably (not surely) talking about a less serious flaw. If they\x92re not, then (almost) by definition, one of the two methods has a serious flaw. About a month ago, we ran into a very similar situation where a trial PI was insisting that his method was correct (p-value around 0.04), and our coordinating center\x92s statistician thought it was not the correct way. According to the statistician, the correct method leads to a different result (p-value around 0.15). The PI was trying to convince us that both ways could be correct. My response was that if both approaches are indeed correct, then the field of Statistics is a useless field. After a little bit of research, it became clear that the PI\x92s approach was indeed flawed, and the PI was convinced. Back to the posted case: the text says that the statistician\x92s \x93statistical analysis yields a result contrary to the scientist\x92s result\x94, which means that we are talking about a serious flaw. In such case, the statistician should ask her supervisor to sign off since he believes that it should be signed off. I would think that the signature of the statistician\x92s supervisor has at least as much weight as the statistician\x92s signature. If not, then something more important is not quite right in that organization.

BiostatEthicsCaseStudyForm edit

Title_Biostatistics_Ethics_Case_Study Three against one - conflict over flawed analysis
Name_Case_Contributor Shelley Hurwitz
Deidentified_Material Yes
No_Copyright_Restrictions Yes
Case_Presentation This is a real-life dilemma contributed by a statistician in industry:

A scientist analyzed some data and the results will be sent to a governmental agency in support of product submission/approval. His supervisor suggests that he contact a statistician to make sure that the data have been analyzed appropriately. The scientist visits with the statistician and asks that the statistician look over his results and send an email message that the analysis is fine. The statistician consults with the scientist to understand the experiment and the data, asks for the raw data, and does the statistical analysis. Unfortunately, the scientist has analyzed the data incorrectly and the statistical analysis yields a result contrary to the scientist\x92s result. The scientist is unhappy and wants to submit his results. The statistician tells him that she cannot support his statistical analysis and based upon the scientist\x92s comments, knows that the issue is not closed.
The statistician reports to a scientist who is not an advocate of statistics and a few days later her supervisor invites her to a meeting in his office and although he neglects to specify the agenda for the meeting she knows the topic. She goes to the meeting and the attendees are her supervisor, the scientist, and his supervisor. When her supervisor tells her that they want to discuss the statistical analysis, she says, \x93I didn\x92t know we were going to discuss this and I need to go back to my office and get my file.\x94 She goes back to her office and sits for about 10 minutes to gather her thoughts since she is upset that her supervisor has not talked to her before scheduling the meeting.
Back at the meeting her supervisor tells her that he wants her to sign off on the scientist\x92s analysis so that it can be submitted. She is now really upset since he does not support or respect her professional judgment. With three against one what would you do? Since your supervisor wants you to accept the flawed analysis should you put the responsibility on him and walk away? Would you stand up for your ethical principles knowing your job and/or raises may be in jeopardy?

The statistician says she cannot do that and proceeds to explain why the scientist\x92s statistical analysis is flawed. During the hour-long discussion her supervisor continues to be non-supportive and pressures her into accepting the scientist\x92s analysis. After much discussion the scientist\x92s supervisor says he understands why the scientist\x92s statistical analysis is flawed and the matter is resolved. Although the incident ended favorably what would you have done if the scientist\x92s supervisor had not accepted your explanation?
See Also
Disclaimer The views expressed within CTSpedia are those of the author and must not be taken to represent policy or guidance on the behalf of any organization or institution with which the author is affiliated.
This topic: CTSpedia > WebHome > EthicsBiostat > StatEthicsCaseStudies > BiostatEthicsCaseStudy005
Topic revision: 08 Feb 2013, PaulWakim
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback