COSIG has a new landing page! Check it out at cosig.net.

(Files for COSIG are still hosted on OSF!)

What is a "calculation chain" file? How can I easily identify duplicated values in a table?

COSIG's 31st guide is all about data forensics using Microsoft Excel!

osf.io/bz725

As always, all guides are available at cosig.net.

Adjusting the black/white levels in an image can reveal inconsistencies barely perceptible to the naked eye.

A 2018 article in Scientific Reports was just retracted as a result of several such inconsistencies.

Check out COSIG's guide on software for image forensics here: osf.io/g23pf

Here we are!

Our entry on reporting publication integrity issues to publishers has contact information to reach the publication integrity and publication ethics departments at most major scholarly publishers. Read here: osf.io/4edk2

All of COSIG is available at cosig.net!

A 2019 Cochrane systematic review on using lithium to treat acute mania confuses standard error (SE) with standard deviation (SD) for two studies included in a meta-analysis, resulting in an overestimation of overall effect size in that meta-analysis.

See COSIG's guide on the topic: osf.io/hp4yd

"Data leakage" occurs when a statistical or machine learning model has access to data it shouldn't during the training process. This can lead to severe overestimation of a model's predictive performance.

Data leakage is the subject of COSIG's 30th guide! Read here: osf.io/d8mzb

A recent article on preventing solar cells from overheating using nanoparticles presents this TEM image. However, this image contains apparent duplications. Can you spot them? Answers in PubPeer thread below.

See COSIG's guide on image duplication: osf.io/547re

Do you want to take part in post-publication peer review but have no idea where to start? COSIG's newest guide covers exactly this, going over some basics and suggestions on how to get started with PPPR/forensic metascience/sleuthing!

cosig.net

"Getting started" guide: osf.io/8m2tz

The vertical line test is a simple method for gauging if a trace/curve is a mathematical function or not. Certain types of data, including nearly all spectra, should pass the vertical line test.

One answer: this spectrum fails the vertical line test. There are several regions where the curve appears to double back on itself such that a single value on the x axis corresponds to multiple values on the y axis.

Can you spot any issues with this published FTIR spectrum?

One answer below...🧵

If you asked four people their age in years, the average of their ages could only ever end in .0, .25, .50 or .75.

This is the principle underlying GRIM (Granularity-Related Inconsistency of Means), now covered in a new guide for COSIG's first official update!

cosig.net

GRIM guide: osf.io/67bp4

Energy dispersive X-ray spectroscopy (EDS/EDX/EDAX) is a popular technique in materials science for analyzing the elemental composition of samples. Check out COSIG's EDX guide here: osf.io/shfjy

Can you spot any issues with this EDX spectrum? Answers in the alt text.

Anyone can do post-publication peer review.
Anyone can be a steward of the scientific literature.
Anyone can do forensic metascience.
Anyone can sleuth.

That's why we are launching COSIG: the Collection of Open Science Integrity Guides, an open source resource for all of the above.

cosig.net