1 of 34

Ethics of Data Interpretation

MATH - BIOINF- STAT 547: Mathematics of Data

Ed Pagani, PhD

Executive Director

External Partnerships

Duke University

March 16, 2023

2 of 34

3 of 34

4 of 34

The years are 1973 and 1974 while I was a student at Columbia University completing an undergraduate degree. Between academic years, I was a summer intern working in the Psychology Department conducting experiments on learning using operant conditioning working in the lab of Dr Michael Cantor.

The top left side of the slide shows examples of behavior, innate and conditioned or learned. Pavlov (1902) started from the idea that there are some things that a animals does not need to learn. For example, dogs don't learn to salivate whenever they see food. This reflex is 'hard-wired' into the dog. In behaviorist terms, food is an unconditioned stimulus and salivation is an unconditioned response. BF Skinner was one of the first to introduce to the concept of conditioning and employed a device (shown on the left) to test hypotheses on learning. In the first studies food was used as a "reward", but to do so food was withheld from the animals in preparation for a learning session, and after some time in the session the animals would stop their activities because enough food was provided it was no longer reward. Only after another day of food withdraw could the experiment resume. Adoption of delivering a positive reward through stimulation of specific areas of the brain (medial forebrain bundle is one), could the experiments last hours, day after day.

5 of 34

Brain Stimulation Reinforcement: Implications of an Electrode Artifact

Michael B. Cantor, Richard J. Beninger, France Bellisle and Peter M. Milner

Science, New Series, Vol. 204, No. 4398 (Jun. 15, 1979), pp. 1235-1237

ignaled

nsignaled

Current

Constant voltage stimulator

Time (msec)

6 of 34

This electrode and plug assembly was used

for decades leading up to these experiments

7 of 34

Across the hallway from the operant conditioning lab at Columbia…..

Terrace arranged to have an infant chimp – named Nim Chimpsky, a poke at linguist Noam Chomsky, whose theories he challenged – cared for by a family not far from Columbia’s Morningside campus. Over three years, Terrace and a group of students and caregivers trained Nim to express himself in American Sign Language.

Columbia News, Oct 11, 2019

Carla Cantor�Director of Communications, Science and Technology

8 of 34

Terrace and his colleagues concluded that the chimpanzee did not show any meaningful sequential behavior that rivaled human grammar. Nim's use of language was strictly pragmatic, as a means of obtaining an outcome, unlike a human child's, which can serve to generate or express meanings, thoughts or ideas.

9 of 34

10 of 34

DATA

11 of 34

Falsification and fabrication

Poorly designed experimental protocols

Use of faulty research materials

Poor execution of research activities

Over interpretation of data

Incorrect application statistical analysis

Misleading information comes from…..

12 of 34

Subconscious bias or pressure to

produce results/interpret data that deliver a desired outcome

Fear of failure

Personal recognition among peers

Research support

Academic Tenure

Personal financial benefit

Job security

Loyalty to a corporation

13 of 34

14 of 34

15 of 34

MMR Vaccine Linked to Autism - fraudulent

16 of 34

17 of 34

Data for 2015 and beyond will be

updated as retractions are issued

18 of 34

INTENTIONAL

19 of 34

20 of 34

Small data set

21 of 34

A lack of access to methodological details, raw data, and research materials

Use of misidentified, cross-contaminated, or over-passaged cell lines and microorganisms

Inability to manage complex datasets

Poor research practices and experimental design

Cognitive bias

A competitive culture that rewards novel finings and under values negative results

22 of 34

23 of 34

and pressure to deliver results

24 of 34

25 of 34

26 of 34

All hope is not lost

27 of 34

28 of 34

29 of 34

Manufacturers of products and services implement internal programs, by choice or as required by state or federal laws, to closely follow design, performance and manufacturing specifications.

30 of 34

31 of 34

32 of 34

33 of 34

The research protocol for a study of a drug on hypertension requires the administration of the drug at different doses to 50 laboratory mice, with chemical and behavioral tests to determine toxic effects. Tom has almost finished the experiment for Dr. Q. He has only 5 mice left to test. However, he really wants to finish his work in time to go to Florida on spring break with his friends, who are leaving tonight. He has injected the drug in all 50 mice but has not completed all of the tests. He therefore decides to extrapolate from the 45 completed results to produce the 5 additional results.

Ethics in Research:

David B. Resnik, J.D., Ph.D.�Bioethicist

https://www.niehs.nih.gov/

Case Study 1

34 of 34

Dr. T has just discovered a mathematical error in his paper that has been accepted for publication in a journal. The error does not affect the overall results of his research, but it is potentially misleading. The journal has just gone to press, so it is too late to catch the error before it appears in print. In order to avoid embarrassment, Dr. T decides to ignore the error.

�

Ethics in Research:

David B. Resnik, J.D., Ph.D.�Bioethicist

https://www.niehs.nih.gov/

Case Study 2