The Reproducibility Crisis

Over the past decade, science has been grappling with what is widely termed a reproducibility crisis.

過去十年來，科學界一直在面對被廣泛稱為「再現性危機」的問題。

When researchers attempted to replicate influential studies, a disconcerting proportion failed to yield the same results.

當研究人員嘗試重現那些具影響力的研究時，有令人不安的比例無法得出相同的結果。

Psychology and biomedical research have been particularly scrutinised in this respect.

在這方面，心理學和生物醫學研究尤其受到嚴格檢視。

Part of the problem lies in incentives that reward novel findings over rigorous verification.

問題的一部分在於，獎勵機制偏重新穎的發現，而非嚴謹的驗證。

Journals have traditionally been reluctant to publish studies reporting inconclusive or negative outcomes.

期刊向來不願刊登報告結果不明確或為負面的研究。

Small sample sizes and flexible statistical analyses further inflate the risk of spurious conclusions.

樣本數過小與統計分析方式彈性過大，進一步提高了得出虛假結論的風險。

In response, many scientists now register their hypotheses and methods before collecting any data.

為此，如今許多科學家在蒐集任何資料之前，就會先登記自己的假設與方法。

Sharing raw data openly allows others to scrutinise and, if necessary, challenge the analysis.

公開分享原始資料，讓其他人能夠審查這些分析，並在必要時提出質疑。

Far from undermining science, this self-correction arguably demonstrates its enduring strength.

這種自我修正非但沒有削弱科學，反而可以說展現了科學歷久不衰的力量。