Scientific Ethics & Exposition

http://www.drshakyacurrentaffairs.com/personal-vs-professional-ethics/
 Ethics

noun

1. (used with a singular or plural verb) a system of moral principles:the ethics of a

culture.

2. (used with a plural verb) the rules of conduct recognized in respect to a particular
class of human actions or a particular group, culture, etc.: medical ethics; Christian
ethics.

3. (used with a plural verb) moral principles, as of an individual: His ethics forbade
betrayal of a confidence.
4. (used with a singular verb) that branch of philosophy dealing with values relating
to human conduct, with respect to the rightness and wrongness of certain actions
and to the goodness and badness of the motives and ends of such actions.

Derived from ethos (Greek), meaning character

How we (each) evaluate what is right vs wrong

2
 Code of Ethics for Researchers
http://widgets.weforum.org/coe/#code

3
 Scientific Ethics
Science is a method of thinking, not an end
Science is only as good as the last finding
Scientists need to be creative, open-minded, skeptics
Most scientific knowledge is inaccurate or wrong
(building on the shoulders of giants)
Science is complicated and methods/analysis are flawed
Making mistakes is how science works

(Most) scientists are human…

4
 Human considerations

Science is expensive, time consuming and competitive

Scientists are rewarded for productivity - publishing in good j

Not rewarded for negative results (typically)

Not rewarded for replicating results (typically)

People have many commitments (distractions)

Livelihood/career at stake (including all that training)

5
 The Problem with Scientific Research
Research is inherently unpredictable and largely
intangible. The only thing going for it is that it consistently
delivers new ideas, new ways of doing things and new
benefits to society and is, over time, self-correcting.

$ $
$
$
$ $$
$ $ $
$ $
Promises Something..
Grants
Awards maybe
in 10 years..
6
 How science works
• Scientific impact typically lags 5-20 years and often
goes through *many* hands and minds
• Most metrics lag (e.g. citations), those that don’t are
highly distorted (e.g. JIF)
• Creativity is largely not quantifiable at the time it occurs
• Science is a continuous, incremental learning process, it
learns from its genuine mistakes
• There is impact from positive and negative results and a
critical place for diversity of thought and approach

Vaccines developed in record time!!!

7
 Covid-19 and Science
Science has been responsible for mitigating the effects of the
pandemic, but..

Scientists often disagree on best ways forward:

• SARS-CoV-2 is transmitted by particulates or by aerosol
• Herd immunity vs track/trace/isolate in place
• Vaccination…

Scientists recommend an action(s), politicians choose what to

enact
Often, scientists are “blamed” for what is done to the degree
that people attack the science/scientists and confidence
plummets
The COVID-19 pandemic will cause trust in science to be
irreparably harmed: Tim Caulfield (July 2020)
https://www.theglobeandmail.com/opinion/article-the-covid-19-pandemic-will-
cause-trust-in-science-to-be-irreparably/
8
These behaviours have been associated with increased
concerns over quality and meaning of science

What can I do?

9
 Harvard and the Brigham call for more than 30
retractions of cardiac stem cell research

NEJM, Lancet, Nature

https://www.statnews.com/2018/10/14/harvard-brigham-retractions-stem-cell/
10
 The Francesca Gino case

Harvard - high profile psychologist

Questions raised by a PhD student whose committee
told her to drop her concerns. Convinced three other
scientists to re-examine the data (Data Colada team)
Dr. Gino is suing the student (now Dr.), the three
scientists and Harvard

Written up in the New Yorker. Blog post here:

https://www.theorgplumber.com/posts/statement/
Would you do what the PhD student did?
Whistleblowers have an uphill struggle, especially
those with little agency of their own
11
 Science is Terribly Inefficient, Corrupt & Fraudulent!
Inefficient: Funders can’t accurately predict where the
next advance will come from. This FRUSTRATES them.
Corrupt: Scientists get to decide who
among their fellow scientists are to be
allocated funds! How is this even a thing?

Fraud: Everyday claims of

irreproducibility, plagiarism, outright
fraud. Can scientists even be trusted?

12
 The scientific process hasn’t changed in…

There are many claims for the father (never mother…) of

science

Aristotle (384-322 BC), a Greek philosopher who made

observations in a wide range of areas, including biology
and physics

Hasan Ibn Al-Hathem (965-1040) who made fundamental

discoveries around optics and vision and required a
testable hypothesis

Modern scientific method emerged from the 17th Century

(Newton, Bacon, Descartes, Hooke, etc)
13
 How to design experiments

Hypothesis/Idea? Big Data

Publish/add experiment

Design Experiment

Data interpretation

Run Experiment

Data analysis
14
 What can go wrong?

Hypothesis/Idea? Big Data

Publish/add experiment

Design Experiment

Data interpretation

Run Experiment

Data analysis
15
 Hypothesis (the hardest/cheapest part)

Ideas can be based on erroneous assumptions

Test your assumptions (often done after the experiment)

Should be open and you should be prepared to consider
any outcome
Preparation - know the literature and understand
weaknesses

Talk (and library work) is cheap!

16
 What can go wrong?

Hypothesis/Idea? Big Data

Publish/add experiment

Design Experiment

Data interpretation

Run Experiment

Data analysis
17
 High Volume (Big Data) Data
“Hypothesis generating” experiments

Collect large amounts of data that is then processed

Often instrument or technology-specific noise and confounder

E.g. In mass spectrometry, different techniques are

associated with different artefacts (CRAPome)
Sometimes, data automatically pre-processed, affecting
assumptions (know how it was generated)

Gene names/codes may not map, software versions update

Internally validate, if possible

18
 What can go wrong?

Hypothesis/Idea? Big Data

Publish/add experiment

Design Experiment

Data interpretation

Run Experiment

Data analysis
19
 Good Experimental Design

Statistical power: will you have sufficient data to yield a reaso

Example of clinical trials: if primary question doesn’t

reach significance, often analyze subsets - but often
under-powered.

Applicable to any experiment that requires more

than 1 week to conduct:
Do you have the right controls?
Anticipate possible outcomes
Avoid false leads/artefacts
20
 What can go wrong?

Hypothesis/Idea? Big Data

Publish/add experiment

Design Experiment

Data interpretation

Run Experiment

Data analysis
21
 Avoiding Experimental Bias

Blinding experiment

Replications (difference between technical, experimental)

All data (no deletion)

Robustness of result

Wonky reagents

Inherent variation

Document everything (notebook/electronically)

22
 What can go wrong?

Hypothesis/Idea? Big Data

Publish/add experiment

Design Experiment

Data interpretation

Run Experiment

Data analysis
23
A picture is a thousand words

24
 Fig 1C
Fig1B Cell line 1 Cell line 2 Cell line 3 Cell line 4

Gene A Gene A

Gene B Gene B

Gene C Gene C

Actin Actin

Fig2A Cell line 1 Cell line 2 Fig3A Cell line 3 Cell line 4
Gene A Gene A

Gene D Gene D

Gene E Gene E

Gene F Gene F

Actin Actin

different actins
 Catching Fraudulent Science

One of the most commony detected forms of cheating is in

images but the advent of powerful image analysis AI tools
has improved detection (across publications)

Plagiarism engines are now routinely used (but often catch

repetitive prose such as Materials and Methods)

Working against the use of such detection tools is increasing

use of LLM and ChatGPT approaches that provide answers
from existing literature. Most universities have policies
against their use (but not universally)

www.retractionwatch.com pubpeer.com
26
 U of T statement on GAI in research

Unauthorized use of generative AI tools for scholarly work at the University of

Toronto may be considered an offence under the Code of Behaviour on Academic
Matters, and research misconduct as defined in the Policy on Ethical Conduct in
Research and the Framework to Address Allegations of Research Misconduct.

https://www.sgs.utoronto.ca/about/guidance-on-
the-use-of-generative-artificial-intelligence/
This guidance addresses the use of generative AI in graduate thesis work at the University of
Toronto. It outlines frequently asked questions related to the use of generative AI tools like
ChatGPT in academic research. The guidance emphasizes the importance of academic quality,
research integrity, and transparency when utilizing generative AI in scholarly activities. It highlights
the need for transparency between students and supervisors in how AI tools are used and the
requirement to provide clear descriptions and citations when using these tools in creating scholarly
work. The document also acknowledges the potential benefits and risks associated with using
generative AI in graduate thesis research and writing.

What does this mean and is it enforcible?

27
 Which statistical test should I use?

http://methods.sagepub.com/which-stats-
test?utm_source=SAGE_social&hootPostID=7c18066e719c
7df8792db8cb38d0f839
Visual Introduction to Probability and Stats
https://seeing-theory.brown.edu

28
 Statistics, Damn Statistics

p-hacking
Inappropriate tool application

Inappropriate or missing checks

Double blinding

Use data directly from instruments (no transcription)

Journals beginning to insist on access of data, proof of

appropriate powering, etc.
29
 Publishing is changing (at last!)

Goodhardt’s Law
“when a measure becomes a target, it ceases to be a good measure.” In other words, when we use a
measure to reward performance, we provide an incentive to manipulate the measure in order to receive
the reward.

https://cmte.ieee.org/futuredirections/2022/01/07/the-changing-landscape-of-scientific-publications/

30
 What’s a Preprinter Server and why should I care?
• Typically a repository of manuscripts that have
yet to be peer reviewed.
• Typically free to use and free to access
• Take 2-5 days from submission to post
• Others may comment on your preprint
• Has a DOI and can be cited in grant applications
• Can be “scoop” insurance
• Allows rapid disemmination of information
There are many, some independent, some
associated with publishers:

31
 Paper Mills and Generative AI

AI is constantly improving. Authorships for sale.

https://www.nature.com/articles/d41586-023-03464-
x
32
 Changing Standards

Error bars in experimental biology.

Cumming, G., Fidler, F., & Vaux, D. L. (2007). The Journal of Cell
Biology, 177(1), 7–11. http://doi.org/10.1083/jcb.200611141

EMBO J and eLife - transparent review

Publish reviewers comments on paper

33
 The credibility problem of science
Society demands & is dependent upon scientific progress
but we are surrounded by so much pseudoscience.

What harm is that? Why should we trust science anyway?

The view from outside:

Science is a shelf in a toy store or a Whiz Bang TV show
Scientists are elites/inaccessible/in pocket of big XYZ 👑💰
Scientists speak an obscure dialect of Klingon 👾👾👾
Science is hugely expensive 💰💰
Scientists are more interested in the sex habits of bees
than diseases 🐝
Scientists disagree with one another and cheat 💔🎭
Scientific dogma changes all the time 🌒🌓🌔🌕🌖🌗🌘
34
 astonishing
revolutionize
staggering

How science is
corrupted in the MSM
(hyperbole and
clickbait for
devastating conditions)

http://www.telegraph.co.uk/news/health/news/12190094/Breast-cancer-tumours-killed-in-11-days-with-staggering-new-therapy.html
35
36
Our hospital pharmacies are crammed with garbage
products, masquerading as effective drugs
37
38
 We are the Enemy (in part)

There is much scientific outreach in various forms but most

is driven by need for publicity - fund-raising 💰🇨🇦

As a consequence, we use language that competes for

attention (of media, which the amplify further for public)
📰🎥

Is this ethical? Does coating science in hyperbole associate

it in the publics mind with scepticism and doubt? 😱

39
 🔍 Science is Never Settled

A major problem for communication of science is that

science is a means or method, not an end. The fact that
science is a process that constantly hacks down its own
ideas is often exploited. 🔪

Despite massive volumes of evidence for a model, it can be

dissolved by a single new compelling discovery. 🔬💡

But, not all science is created equal & burden of

veracity is not quantitative, but qualitative.

Vulnerable to false balance, misinterpretation of

meaning of expert criticism; easily subverted by
pseudoscience that lacks structural rigour.
40
 Moving Forward

What can you do? What are best practices in SciComm?

• The internet has been a major factor in enabling quackery and false
information. It is also a powerful tool for neutralizing and there have never
been better tools (social media, blogs, etc).

• Empathize and try to understand where unscientific arguments are

coming from (rather than batting back ball).

• Start young & carry on. Science is inherently interesting.

• De-formularize press releases and use plain language that is sensitive to

patients and their families.

41
 Re-Normalize Science
There isn’t a time or place. Communicate constantly. If you cannot
effectively communicate what you do & why, you will likely struggle in
science (though historically there have been exceptions…).

We run an outreach program for K-12 which works with 5,000 kids a
year. Main benefit to us is providing our own trainees with
communications skills.

We also fund-raise by asking trainees to present their work to donors -

a little like Dragons Den. Winners are usually those with ability to
effectively explain their work.

42
 So what can I do?
• Be more open and accessible!
• Be more communicative, but…
• Enough with the hype! Educate, don’t exaggerate
• Encourage risk, happenstance, challenge

• Understand sources of, and correct for, bias

• Recognize value of full spectrum of research

• Public confidence in science is fragile (see persistence

of pseudoscience), never take it for granted
• Communicate at every opportunity

43