<s>
Statcheck	B-Algorithm
is	O
an	O
R	B-Language
package	I-Language
designed	O
to	O
detect	O
statistical	O
errors	O
in	O
peer-reviewed	O
psychology	O
articles	O
by	O
searching	O
papers	O
for	O
statistical	O
results	O
,	O
redoing	O
the	O
calculations	O
described	O
in	O
each	O
paper	O
,	O
and	O
comparing	O
the	O
two	O
values	O
to	O
see	O
if	O
they	O
match	O
.	O
</s>
<s>
Another	O
limitation	O
is	O
that	O
Statcheck	B-Algorithm
cannot	O
deal	O
with	O
statistical	O
corrections	O
to	O
test	O
statistics	O
,	O
like	O
Greenhouse	O
–	O
Geisser	O
or	O
Bonferroni	O
corrections	O
,	O
which	O
actually	O
make	O
tests	O
more	O
conservative	O
.	O
</s>
<s>
Some	O
journals	O
have	O
begun	O
piloting	O
Statcheck	B-Algorithm
as	O
part	O
of	O
their	O
peer	O
review	O
process	O
.	O
</s>
<s>
Statcheck	B-Algorithm
is	O
free	B-Application
software	I-Application
published	O
under	O
the	O
GNU	B-License
GPL	I-License
v3	O
.	O
</s>
<s>
In	O
2017	O
,	O
Statcheck	B-Algorithm
's	O
developers	O
published	O
a	O
preprint	O
paper	O
concluding	O
that	O
the	O
program	O
accurately	O
identified	O
statistical	O
errors	O
over	O
95%	O
of	O
the	O
time	O
.	O
</s>
<s>
The	O
study	O
found	O
that	O
Statcheck	B-Algorithm
recognized	O
60%	O
of	O
all	O
statistical	O
tests	O
.	O
</s>
<s>
Reversely	O
,	O
if	O
a	O
test	O
was	O
truly	O
inconsistent	O
,	O
Statcheck	B-Algorithm
flagged	O
it	O
in	O
an	O
estimated	O
51.8	O
%	O
of	O
cases	O
(	O
this	O
estimate	O
included	O
the	O
undetected	O
tests	O
and	O
assumed	O
that	O
they	O
had	O
the	O
same	O
rate	O
of	O
inconsistencies	O
as	O
the	O
detected	O
tests	O
)	O
.	O
</s>
<s>
Overall	O
,	O
Statcheck	B-Algorithm
's	O
accuracy	O
was	O
95.9	O
%	O
,	O
half	O
a	O
percentage	O
point	O
higher	O
than	O
the	O
chance	O
level	O
of	O
95.4	O
%	O
expected	O
when	O
all	O
tests	O
are	O
simply	O
taken	O
at	O
face	O
value	O
.	O
</s>
<s>
Statcheck	B-Algorithm
was	O
conservatively	O
biased	O
(	O
by	O
about	O
one	O
standard	O
deviation	O
)	O
against	O
flagging	O
tests	O
.	O
</s>
<s>
More	O
recent	O
research	O
has	O
used	O
Statcheck	B-Algorithm
on	O
papers	O
published	O
in	O
Canadian	O
psychology	O
journals	O
,	O
finding	O
similar	O
rates	O
of	O
statistical	O
reporting	O
errors	O
as	O
the	O
original	O
authors	O
based	O
on	O
a	O
30-year	O
sample	O
of	O
such	O
articles	O
.	O
</s>
<s>
Statcheck	B-Algorithm
was	O
first	O
developed	O
in	O
2015	O
by	O
Michele	O
Nuijten	O
of	O
Tilburg	O
University	O
and	O
Sacha	O
Epskamp	O
of	O
the	O
University	O
of	O
Amsterdam	O
.	O
</s>
<s>
Later	O
that	O
year	O
,	O
Nuijten	O
and	O
her	O
colleagues	O
published	O
a	O
paper	O
using	O
Statcheck	B-Algorithm
on	O
over	O
30,000	O
psychology	O
papers	O
and	O
reported	O
that	O
"	O
half	O
of	O
all	O
published	O
psychology	O
papers	O
 [ ... ] 	O
contained	O
at	O
least	O
one	O
p-value	O
that	O
was	O
inconsistent	O
with	O
its	O
test	O
"	O
.	O
</s>
<s>
In	O
2016	O
,	O
Nuijten	O
and	O
Epskamp	O
both	O
received	O
the	O
Leamer-Rosenthal	O
Prize	O
for	O
Open	O
Social	O
Science	O
from	O
the	O
Berkeley	O
Initiative	O
for	O
Transparency	O
in	O
the	O
Social	O
Sciences	O
for	O
creating	O
Statcheck	B-Algorithm
.	O
</s>
<s>
In	O
2016	O
,	O
Tilburg	O
University	O
researcher	O
Chris	O
Hartgerink	O
used	O
Statcheck	B-Algorithm
to	O
scan	O
over	O
50,000	O
psychology	O
papers	O
and	O
posted	O
the	O
results	O
to	O
PubPeer	O
;	O
he	O
subsequently	O
published	O
the	O
data	O
he	O
extracted	O
from	O
these	O
papers	O
in	O
an	O
article	O
in	O
the	O
journal	O
Data	O
.	O
</s>
<s>
He	O
also	O
told	O
Vox	O
that	O
he	O
intended	O
to	O
use	O
Statcheck	B-Algorithm
to	O
perform	O
a	O
function	O
similar	O
to	O
a	O
spell	B-Application
checker	I-Application
software	O
program	O
.	O
</s>
<s>
These	O
flaggings	O
,	O
and	O
their	O
posting	O
on	O
a	O
public	O
forum	O
,	O
proved	O
controversial	O
,	O
prompting	O
the	O
German	O
Psychological	O
Society	O
to	O
issue	O
a	O
statement	O
condemning	O
this	O
use	O
of	O
Statcheck	B-Algorithm
.	O
</s>
<s>
Bishop	O
,	O
who	O
had	O
two	O
of	O
her	O
own	O
papers	O
flagged	O
by	O
Statcheck	B-Algorithm
,	O
criticized	O
the	O
program	O
for	O
publicly	O
flagging	O
many	O
papers	O
(	O
including	O
one	O
of	O
her	O
own	O
)	O
despite	O
not	O
having	O
found	O
any	O
statistical	O
errors	O
in	O
it	O
.	O
</s>
<s>
Other	O
critics	O
alleged	O
that	O
Statcheck	B-Algorithm
had	O
reported	O
the	O
presence	O
of	O
errors	O
in	O
papers	O
that	O
did	O
not	O
actually	O
contain	O
them	O
,	O
due	O
to	O
the	O
tool	O
's	O
failure	O
to	O
correctly	O
read	O
statistics	O
from	O
certain	O
papers	O
.	O
</s>
<s>
Journals	O
that	O
have	O
begun	O
piloting	O
the	O
use	O
of	O
Statcheck	B-Algorithm
as	O
part	O
of	O
their	O
peer	O
review	O
process	O
include	O
Psychological	O
Science	O
,	O
the	O
Canadian	O
Journal	O
of	O
Human	O
Sexuality	O
,	O
and	O
the	O
Journal	O
of	O
Experimental	O
Social	O
Psychology	O
.	O
</s>
