Total Pageviews

Tuesday, November 27, 2018

Science desperately needs a lot more quality, and a lot less control

An immature concept 
of science 
has captured 
the public mind.

Science has moved toward 
being a validator 
of what's socially acceptable,

Some “scientists” 
no longer understand 
the point of real science.

They see their science 
as the repository 
of ultimate truth. 

But "truth" only applies 
to junk science,
arbitrarily declared 
to be "settled science".

Real science is not final, 
but constantly evolving 
knowledge.

Unfortunately, 
the publication of 
flawed study results 
has become so common 
that it’s no longer 
even surprising.


A recently published 
Nature paper
resulted in the usual 
'climate change --
-- it’s-worse-than-we-thought'
headlines in the 
Washington Post, BBC, 
New York Times, etc.

In days 
the paper’s results 
were assessed to have 
“major problems” 
by an author of 
multiple CO2 
climate sensitivity 
papers 
(Lewis and Curry, 2015, 2018).

A glaring miscalculation
was quickly spotted, 
that undermined
the conclusion,
that estimates of 
climate sensitivity 
to a doubling 
of the CO2 level
may be too low.

And yet the paper 
was able to 
pass through 
peer review anyway.





In 2016, 
Dr. Michael Mann 
was the lead author of 
an embarrassingly 
non-scientific paper 
fraught with 
glaring methodological 
and statistical errors.

A post-publication reviewer 
(statistician Dr. William Briggs) 
wrote in his 
point-by-point critique
of the paper that 
“Mann’s errors are in no way 
unique or rare; indeed, 
they are banal and ubiquitous.”

Despite the glaring errors, 
the paper made it 
through peer-review 
and was published 
in Nature‘s 
Scientific Reports 
journal anyway.




Analyses indicate that 
as many as 7 of 10 
peer-reviewed journals 
are apt to publish 
a deliberately-written 
“hoax” paper.

“Any reviewer with more than 
a high-school knowledge 
of chemistry and the ability 
to understand a basic data plot 
should have spotted the paper’s 
shortcomings immediately. 
Its experiments are 
so hopelessly flawed 
that the results 
are meaningless." 

" … The hoax paper 
was accepted by 
a whopping 157 journals 
and rejected by only 98.

Of the 106 journals 
that did conduct peer review, 
70% accepted the paper…”  
(Murphy, 2017  
The Failure of Peer Review)





Earlier this year, 
a review paper 
(Larcombe and Ridd, 2018)
published in the journal 
Marine Pollution Bulletin 
delivered a stinging rebuke 
to the modern version 
of science’s disturbing
lack of replicability 
and verifiability.

The authors go on to detail 
a large volume of examples 
when peer-review 
failed to detect errors
in Great Barrier Reef (GBR) 
coral research.

Confirmation bias 
appears to permeate 
the peer-reviewed literature, 
slanted in the direction 
of finding evidence 
for a catastrophic decline 
in coral health.  

This isn’t the first time 
that marine research 
has been called out 
for overselling calamity 
(see Cressey, 2015, 
“Ocean ‘calamities’ 
oversold, 
say researchers 
– Team calls for 
more skepticism 
in marine research.”) 
and falling 
“into a mode 
of groupthink 
that can damage 
the credibility 
of the ocean sciences”.

As just a single example 
among the many provided, 
Larcombe and Ridd
reviewed the De’ath et al.
 (2009) study 
in which an 
“unprecedented” decline 
in Great Barrier Reef corals 
was alleged to have occurred 
between 1990-2005.

After a reanalysis 
of the measurements 
and methods used, 
Larcombe and Ridd 
used corrected data 
to show there has 
actually been 
“a small increase 
in the growth rate” 
of corals 
since the early 1900s
instead of the 
dramatic decline 
after the 1990s 
documented in the 
peer-reviewed paper.

These errors slipped past 
the reviewers’ notice too.  

This paper 
[De’ath et al. (2009): 
Declining coral calcification 
on the Great Barrier Reef.] 
studied 328 corals on the GBR, 
and indicated a 14% reduction 
in growth rates 
between 1990 and 2005. 

Subsequent reanalysis 
of the data indicated 
that the apparent 
recent reduction 
in growth rate 
was caused by: 

(a) 
Problems with the 
physical measurements 
of calcification, 
which systematically biased 
recent growth bands 
to give lower growth rates
(D’Olivio et al., 2013; 
Ridd et al., 2013), 
and

(b) 
An unjustified assumption 
that coral growth rate 
does not change 
with the age of the coral 
(Ridd et al., 2013). 

D’Olivio et al. (2013), 
working on a different set 
of GBR corals, 
showed an increase 
in coral calcification rates 
( on middle 
and outer 
shelf reefs, 
which together 
represent 99% 
of GBR corals ), 
of 10% per decade,
for the period 
~1950 to ~2005, 
... but a decrease 
of 5% per decade 
between 1930 and 2008 
on inner-shelf reefs, 
( which represent 
only 1% of GBR corals ). 

"Therefore, 
it would be hard to glean 
from these datasets 
that there is 
a documented 
decline in 
coral ‘growth’ 
parameters, 
and even harder 
to attribute change 
to a particular cause.”