Today some co-authors and I published a "correction" removing our names as authors on a paper

https://royalsocietypublishing.org/doi/10.1098/rspb.2020.1852

This unusual move deserves explanation despite continuing moves by a coauthor's lawyers that have had a chilling effect my expression of my scientific views

So, I will limit myself to a few statements of fact:
One of the datasets used to establish repeatability of behaviors (not originally posted on dryad) contained data on latency times for spiders to initiate a behavior (0 to 600 seconds)...

As you might expect with latency times, the integer values were exponentially distributed. Except, a handful of integers were vastly over-represented relative to (1) exponential distribution expectations and (2) adjacent integers.

I first noticed that the number 67 was >3x more common than 66 or 68. Odd, but not a clear signal of a problem. On looking for the other over-repesented integers, there was a pattern: each was a 2-digit number with adjacent values on a keyboad: 34, 45, 56, 67, 78, 89, 90.

Each of these is ~3x more common than the flanking integers (e.g., 34 is more common than 33 or 35). I see no plausible biological cause for this pattern, and the probability of this pattern from exponential or other distributions is very low.

The co-author who generated these data offered an explanation for this pattern, invoking unconscious observer bias of a few seconds here and there, and offered to redo the analyses rounding all values to the nearest 5 (e.g., 34 --> 35, etc) & provide a correction.

The journal requested such a correction. To my limited knowledge the editorial board has not ruled out a retraction *if* the submitted correction proves to be insufficient.

I would ask that the community not trash the ProcB editorial board: this is not yet the final word on this paper, and I expect they are under intense legal pressure to move slowly and cautiously.

As co-authors, we also were very cautious in coming to our decision to remove ourselves. We did this based on lengthy (weeks-long) collective examination of the dataset specifically for this paper, and was not reached lightly or gladly.

We make no claims about how the data came to be flawed, but the origins of the data is secondary to the clear biological implausibility of the numbers (which we documented with extensive statistical analysis), which suffices to lead us to end our authorship with this paper.

I post this thread, speaking as a co-author, because I believe:
1. the scientific record needs to be amended when flaws are found,
2. these changes need to be disseminated to the paper's audience,
3. that social media is an effective tool for disseminating scientific changes