Thread by @ProfLisaShu

This famous paper synonymous with my name - Shu et al. (2012) published in PNAS - is being retracted due to evidence of fraudulent data.

I should be appalled. Instead I'm delighted by this piece of investigative data journalism from the brilliant @datacolada team

Why? 1/🧵

https://twitter.com/DataColada/status/1427645912474587151

Sit back and buckle up for a tale of two papers (2012 and 2020) - and the infamous Study 3 from the 2012 paper.

2/🧵

Origins: back in 2010, then-grad student Shu (me!) together with Gino & Bazerman ran studies on the consequences of dishonesty (moral disengagement, motivated forgetting). We observed too-high levels of cheating among lab participants - expensive studies to run!

3/🧵

To avoid ceiling effects, we needed to lower the overall levels of cheating. We tried introducing a signature in order to reduce cheating behavior...and voila!

✍️ Signing first as a means to reduce cheating behavior was born in the lab

4/🧵

[Note to researchers: the original DVs of interest in the Shu-Gino-Bazerman studies were moral disengagement and motivated forgetting - NOT cheating itself. But we needed sufficient variance in cheating behavior to detect differences our DVs of interest]

5/🧵

Then in 2011: Shu-Gino-Bazerman - with 2 lab studies showing the effectiveness of signing first in reducing cheating behavior - heard through a conference presentation Ariely-Mazar's field experiment with an auto insurance company showing the same result

6/🧵

Did one team race to publish in order to scoop the other? NO: both teams thought they found the yin to their yang ☯️

A merger was born: Shu-Gino-Bazerman's Studies 1 and 2 from the lab + Ariely-Mazar's Study 3 from the field - which appeared be perfect complements

7/🧵

Did we request to scrutinize each other's datasets at the time? NO

Did Shu-Gino-Bazerman ever see or touch the Ariely-Mazar field experiment data? NO

We began our collaboration from a place of assumed trust - rather than earned trust. Lesson learned.

8/🧵

Shu et al 2012 born of this merger was soon published in PNAS, edited by a Nobel Laureate. It received great attention.

The core result that✍️ leads to greater honesty was taught across MBA and exec classrooms. Governments, corporates, startups implemented this nudge

9/🧵

Fast forward to a few years: when a research team composed of Kristal, Whillans, Bazerman (of Shu-Gino-Bazerman) tried to use signing first as a way to promote honesty in online forms - and failed to replicate the result...

10/🧵

Further replication attempts failed. Kristal and Whillans advocated for the original 2012 authors to join to run a large-scale pure replication. All agreed

11/🧵

The result: failure to replicate the original 2012 signing-first paper. Together, all seven authors published this replication failure in PNAS and in Scientific American in 2020 - explicitly rejecting the conclusion of the original 2012 paper.

12/🧵

In the process, the lead authors of the 2020 paper noted a randomization failure in the field experiment (Study 3) of the original 2012 paper. Bazerman and Shu argued for retraction of the initial 2012 paper - but the majority did not see it necessary at the time.

13/🧵

For full transparency: the lead authors of 2020 paper posted the data of their replication attempts AND the data from the original 2012 paper - and that is where the @datacolada story that covers evidence of data fabrication in Study 3 of the 2012 paper begins...

14/🧵

I am delighted that this story has come to light and the scientific record is being corrected.

Science is an evolving conversation. We must continually test and retest to confirm or disprove what we think we know and find the limits of what we don't know...

15/🧵

Thank you to the brilliant team @datacolada, especially Uri, Joe, Leif, and other anonymous researchers for continuing to do the important work of ensuring science remains...scientific.

The arc of scientific inquiry is long - but it bends towards truth.

16/16