エラーバーを短くするのは研究不正ですよ

      2018/04/15

論文のストーリーに合う有意差を作り出すために、エラーバーを手作業で動かして短くしてしまうのは、もちろん研究不正です。当たり前のことですが、実例をいくつか拾って紹介し確認しておきたいと思います。新入生が実験を始める時期ですし。

不正例1

Science Signaling 7:ra114, 2014 (hereafter referred to as “Paper 1”)

Chemistry & Biology 21:453-458, 2014 (hereafter referred to as “Paper 2”)

ORI found that Respondent knowingly falsified and/or fabricated data and related text by altering the experimental data to support the experimental hypothesis. Specifically:
1. ORI found falsified and/or fabricated data in Paper 1 in:
c. Figure 6D by manually manipulating the error bars to increase statistical significance of the kinase assay

2. ORI found falsified and/or fabricated data in Paper 2 in:
b. Figure 4A by manipulating data points and reducing error bars and failing to report that JM-HER3 construct had cloning tags

(引用元:https://www.federalregister.gov/documents/2015/08/31/2015-21421/findings-of-research-misconduct

不正例2

Biomed. Chromatogr. 24(4):351-357, 2010 (hereafter the “BC 2010 article”). Specifically, ORI found that the Respondent:

Falsified and fabricated Figure 4 of the BC 2010 article; (中略)In the bottom panel, the Respondent reported the measurement of DMP concentrations in liver samples obtained from three rats at 1, 30, 90, 180, 300, and 720 minutes after a single injection of 5 mg/kg DMP, requiring a total of 18 rats, while the actual data that he had plagiarized, originally generated prior to the Respondent’s arrival in the laboratory by a former postdoctoral researcher, was from plasma samples from a single rat, and the error bars for both panels were fabricated.

(引用元:https://www.federalregister.gov/documents/2012/12/28/2012-31287/findings-of-research-misconduct)

アメリカでは必ずしも論文になっていなくても、学会発表や投稿用原稿中の不正も不正と見なすようです。

不正例3

PHS finds the Respondent engaged in scientific misconduct by falsifying and fabricating data that were reported in a scientific manuscript intended for publication entitled “Increased renin transcription after inhibition of NF-YA with RNAi reveals through regulation of Ea element and Ear2” and at two professional scientific meetings.

Specifically, PHS found that:

2. Respondent falsified Figures 4, 5, 6, and 8 in the manuscript by claiming in the figure legends that 4 independent repetitions contributed to each figure’s results when the actual numbers of repetitions were n=3 for Figure 4, n=1 for Figure 5, n=3 for Figure 6, and n=2 for Figure 8; in Figure 5, error bars based on the Student’s t test further falsely claim that n was >2. He further falsified Figures 6 and 8 by reporting smaller standard errors of the mean than were obtained from the actual data, thereby giving an enhanced impression of rigor for the reported experiments.

(引用元:https://www.federalregister.gov/documents/2008/11/04/E8-26270/findings-of-scientific-misconduct

下は、実験をせずにデータポイントを捏造した例。

不正例4

Specifically, PHS found that Dr. Prasad:

(1) Fabricated data to calculate a standard error of the mean for Bcl-2 mRNA intensity values for the sham group: 16 values (four percentages for each of the four brain regions assayed), when only a single sham value of 100% was actually available, for the error bars shown in Figures 2 and 3 of a manuscript, “Regional expression of Bcl-2 MRNA and mitochondrial cytochrome c release after experimental brain injury in the rat,” submitted to Brain Research, and included in Figures 11 and 12 of NINDS grant application R01 NS41918-01, “Neurochemical mechanisms in traumatic brain injury;” and(以下略)

(引用元:https://www.federalregister.gov/documents/2002/09/05/02-22565/findings-of-scientific-misconduct

下では、SEMをSDとして報告したものも、うっかりミスといういいわけがあったのかもしれませんが、当然の注意を怠った(RECKLESSLY)不正とみなされています。

不正例5

Cell 167(6):1571-1585, 2016 (hereafter referred to as “Paper 1”)

Proceedings of the National Academy of Sciences 112(19):E2447-E2456, 2015 (hereafter referred to as “Paper 2”)

ORI found that Respondent engaged in research misconduct by intentional, knowing, or reckless falsification and/or fabrication of the research record by selectively reporting by inappropriate inclusion/omission or alteration of data points in ten (10) figures and falsely reporting the statistical significance based on falsified data in ten (10) figures across the two (2) papers and supplementary material. Specifically, ORI found that:

  • In Paper 1, Respondent falsified and/or fabricated the research record in:

—Figure 7F, by reporting that error bars represented standard deviation, when they actually represented standard error of the mean (SEM.)

  • In Paper 2, Respondent falsified and/or fabricated the research record in:

—Figure 7A (right), by reporting that error bars represented standard deviation, when they actually represented standard error of the mean (SEM.)

(引用元:https://www.federalregister.gov/documents/2018/02/23/2018-03766/findings-of-research-misconduct

下の例も論文掲載前の段階での不正認定。

不正例6

ORI found that Respondent engaged in research misconduct by intentionally, knowingly, or recklessly causing false data to be recorded, falsifying and/or fabricating data and related images by alteration and/or reuse and/or relabeling of experimental data, and reporting falsified and/or fabricated data in one (1) manuscript subsequently submitted for publication:

“Amphotericin primarily kills human cells by binding and extracting cholesterol.” Submitted for publication to the Proceedings of the National Academy of Sciences [withdrawn prior to peer review] (hereafter referred to as “Manuscript 1”)

Specifically, ORI found that:

In Manuscript 1, Respondent falsified and/or fabricated the results of Surface Plasmon Resonance data on page S7 of the Supporting Information to make the error bars smaller than the actual results of experimentation

(引用元:https://www.federalregister.gov/documents/2017/12/15/2017-26961/findings-of-research-misconduct

これくらいで十分でしょう。

 


 - 科学者の不正行為, 論文データ捏造