When in doubt, speak out

could not replicate: Why didn't you ask?

(2015-07-23 06:34:46)[編輯][刪除]

轉載▼

標簽： 教育 曆史 文化

分類： biology

Years ago, a noble laureate claimed they couldn't repeat my result in a symposium; yet they never asked me.  I knew that years later; however, the doubt about my work circulated, impacting my career, so unfair. Now, it's so common (77%) as shown below.

 

*************

It’s good to talk

Help for those struggling to reproduce results could be just a phone call away.

22 July 2015

Article tools

Rights & Permissions

Survey results released last week by the American Society for Cell Biology (ASCB) included an interesting nugget. Some 72% of respondents said that they had been unable to replicate a published experimental result. Yet a higher proportion (77%) said that they had never been told that their work could not be replicated.

 

There could be many reasons for the difference. The most obvious would be that no one actually tried to replicate the research in question (or that they did not try very hard). When survey participants were asked how they responded to such problems, 55% said that they did not bother resolving the replication issue because they did not think the research was important enough to pursue. For others, the survey results suggest that if and when they did try to replicate, and failed, then they also failed to flag the problem with the original researchers. And it means that they did not ask the people who are best placed to help answer the most obvious question: what am I doing differently to you?

That is not always easy, but it should be the first response. And those on the receiving end of such enquiries should be open to them, not defensive or hostile. As this journal has pointed out before, there is often an art to science. The methods sections of papers, as rigorous as authors and journals try to make them, do not always tell the full story. They cannot pass on tacit knowledge — just as someone cannot be taught adequately from a book how to ride a bicycle.

Related stories

• Reproducibility: Stamp out shabby research conduct
• Journals unite for reproducibility
• Metascience could rescue the ‘replication crisis’

More than 800 of the ASCB’s 9,000 or so members answered the survey. They reported that the most common way to resolve problems with failed replication attempts was through collegial consultation with the lab that did the original experiments. In an era of huge competition in biomedicine — when some researchers might fear hostility or even retaliation from senior colleagues when questioning the reproducibility of their work — the survey shows that amicable collaborations, including reagent sharing and open communication, can improve science and make the work of scientists more efficient.

The ‘replication crisis’ in science, and in biological research in particular, is a serious and complex problem that will not be solved by better communication alone. This journal and others have launched initiatives that aim to address many suggested and suspected problems in reproducing results. The ASCB survey results again highlighted some of the issues: respondents rated the push to publish in high-profile journals and poor methodological training as the biggest factors.

The ASCB published a report alongside the survey results, which made some further recommendations for change (see go.nature.com/uh1wsu). These include improvements in statistics training and standardizing the way that experiments are performed.

Even if systemic problems are tackled successfully, some problems of irreproducibility will remain. Biological systems are complex and finicky, and there will always be new experiments, equipment and techniques that take time to master. That one scientist cannot repeat the work of a second does not mean that the first is unskilled or the second sloppy. Although much of the broader media attention on the replication crisis focuses on deliberate misrepresentation and research fraud, scientists and journals know that the reality is more complex, and less nefarious. Good science is often difficult science. And good scientists should not make it more difficult than it needs to be. So ask for help — pick up the phone.

Journal name:

Nature

Volume:

523,

Pages:

382

Date published:

(23 July 2015)

DOI:

doi:10.1038/523382a

Related stories and links

From nature.com

• Reproducibility: Stamp out shabby research conduct

  11 March 2015

• Journals unite for reproducibility

  05 November 2014

• Metascience could rescue the ‘replication crisis’

  04 November 2014

 

For the best commenting experience, please login or register as a user and agree to our Community Guidelines. You will be re-directed back to this page where you will see comments updating in real-time and have the ability to recommend comments to other users.