Authors retract paper linking nuclear power to slow action on climate change

Do pro-nuclear energy countries act more slowly to curb the effects of climate change? That’s what a paper published in July in the journal Climate Policy claimed. But the hotly debated study was retracted last week after the authors came to understand that it included serious errors.

An August 22 press release about the original study has been retracted by the University of Sussex, and no longer appears on ScienceDaily. An archived version notes:  

While it’s difficult to show a causal link, the researchers say the study casts significant doubts on nuclear energy as the answer to combating climate change.

But there were two errors in the paper, “Nuclear energy and path dependence in Europe’s ‘Energy union’: coherence or continued divergence?” The first, co-author Andrew Stirling of the University of Sussex told Retraction Watch, was in data transcription, and was brought to the authors’ attention by a post on Nuclear Pretty Please.

That led to a correction. But, as Stirling noted, his analysis :

…illuminated the separate time series baseline mistake. This had also been picked up by another blogger, called Nick Thomson.

Thomson lists a host of concerns about the paper, including the math and methodology, and argues that the data don’t support the authors’ main conclusion that countries which embrace nuclear energy are slower to meet climate goals.

Stirling and his co-authors — Andrew Lawrence, of the Vienna School of International Studies, and Benjamin Sovacool, also at the University of Sussex — submitted a response to that blog, Stirling said, but it has yet to be posted.

What’s more, another post by Stephen Tindale and Suzanna Hinson, blogging at Weinberg Next Nuclear on August 26 noted two issues:

– The categories of pro- and anti-nuclear are too broad and do not compare like with like;

– Reduction of greenhouse gases and promotion of renewable energy are presented together as a single objective. They are not.

Then, on November 25, the paper was retracted. The notice — which is quite detailed, and was agreed upon by all three authors, according to Stirling — begins:

It is with great regret that we, the authors, have withdrawn this article from publication in Climate Policy.

The grounds for this are that it has come to our attention that two serious errors were committed by the first author (AL) on this paper. The second (BS) and third (AS) authors entirely failed to identify and correct these errors prior to publication. Of course, all three authors share joint responsibility for this paper, and apologize to readers of Climate Policy for our omission of oversight.

Together, the errors have the effect of invalidating this particular analysis concerning the relative performance of nuclear-committed European countries in wider climate change.

The notice goes on to describe, again in significant detail, the process the authors undertook, which at one point involved a correction, rather than a retraction:

As per guidance from the Committee on Publication Ethics (COPE), we detail here the roles of the different authors of this paper. AL, a mid-career academic political scientist, conducted the quantitative analysis. AL initially contacted BS and asked him to help frame the paper as a whole. BS later invited AS to contribute a focus on the specific issue of technological lock-in.

The first error in the paper was a mistake by AL in the transcription of data from source material. This detracted from, but did not invalidate, the paper’s published conclusions. On being informed of this error by a commentary from a reader, BS confirmed this, double checked other aspects of the analysis, and promptly sent a Corrigendum to Climate Policy editors on 05 October 2016.

In these exchanges, BS and AS developed concerns over the validity of AL’s initial analysis. So, at the first opportunity when returning from travel, AS reproduced the analysis in its entirety on 20 October 2016.

This full re-analysis confirmed that a second error had been committed by AL. This mistake had also been identified in a blog, but had remained unnoticed by BS and AS until that point. This error involved the mistaken calculation of percentage emissions reductions with respect to an earlier date in the time series, rather than the stated baseline. Together with the first error, this had the combined effect of invalidating key findings of this paper, concerning the relative performance of nuclear-committed European countries.

There was also an investigation at Sussex:

With this diagnosis of a second more serious error thereby confirmed, BS and AS referred themselves that same day (20 October 2016) to an internal process at the University of Sussex, to investigate the circumstances of these two errors. All three authors informed the Editors of Climate Policy on the following day, 21 October 2016. The Editors responded on 24 October 2016, and invited us to submit a formal statement, which we did within 24 hours.

Stirling told Retraction Watch that the investigation was completed last week. A statement from the university reads, in part:

The University has very clear guidelines and expectations on research integrity and having conducted an enquiry is confident that the errors in the paper were the result of a failure of process and not academic dishonesty. The authors are very clear that these were honest mistakes and have acted quickly to clarify these errors and mitigate their effects.

While corrections and retractions are a common part of academic life, we would like to apologise to any readers of the original paper or press release for these mistakes.

The authors are committed to submitting the reanalysed data to a journal at a later date.

The retraction notice concludes:

We stand ready to co-operate fully in any onward process that might be necessary to clarify the circumstances of these errors or mitigate their effects.

We can only conclude by apologizing very sincerely once again for the failures on our own parts, that have led to this very unfortunate situation.

We note the Editors and Publishers, Taylor & Francis and Climate Strategies, of Climate Policy published the article in good faith, based on warranties we made regarding originality and provenance.

The article will remain online, digitally watermarked as retracted.

Tindale, one of those who raised questions about the original paper, told Retraction Watch he hadn’t had time to read the reasons for retraction in detail, but

…the fact that the authors and journal have accepted some data and analytical mistakes and so retracted the paper does indeed meet our concerns. And, as I’ve said on Twitter, I think that they deserve some credit for accepting their mistakes.

Update, 4 p.m. Eastern, 11/28/16: First author Andrew Lawrence tells Retraction Watch:

As lead author, I take the lion’s share of the blame for the errors in Table 2. As for what happened, the first error was in incorrectly transcribing data; the second, more substantial one, was in miscalculating the data. Both errors were a consequence of excessive haste.

Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.