May 12, 2017 Is International ME/CFS and FM Awareness Day
International ME/CFS and FM Awareness Day is May 12th, 2017. Jody Smith shares some information about upcoming events and ways you can be heard ...
Discuss the article on the Forums.

"Is withholding your data simply bad science, or should it fall under scientific misconduct?"(July3)

Discussion in 'Other Health News and Research' started by Dolphin, Jul 4, 2015.

  1. Dolphin

    Dolphin Senior Member

    Messages:
    10,557
    Likes:
    27,561
    Possibly of interest to somebody. The PACE Trial investigators have been reluctant to give out some of their data.

    http://blogs.lse.ac.uk/impactofsoci...secrecy-bad-science-or-scientific-misconduct/


     
  2. anciendaze

    anciendaze Senior Member

    Messages:
    1,773
    Likes:
    4,497
    There is always the unwritten rule that those who do the work of collecting data get the first crack at analyzing it. We are well past the point where that applies to PACE. Those authors have milked about all the positive spin from their data possible.

    Now let me try a thought experiment relevant to the alleged objective improvements in patients. With a completely random binary response, like a coin flip, you could expect 50% to appear improved and 50% to appear worse. With a better range of possibilities than a single binary experiment, you should expect something like 40% better, 40% worse and 20% with no detectable change. If you resolutely ignore adverse outcomes, and do some wishful thinking, you can convert this to 60% responders and 40% nonresponders. That is what the PACE authors have reported, without detailed supporting evidence.

    By counting those who do not contribute objective data both before and after therapy as participants in the trial you can allow the effort of participating to stratify your cohort so that most of those who decline the optional walk test after therapy also happen to be those who did not benefit. This will produce an apparent improvement in an objective measure -- which will not be replicated. What you would be measuring in that case was the effort patients put into the trial, not any benefits that came out. More rigorous exercise regimes would produce better stratification.

    I don't actually know that the PACE authors did this. I don't know what they actually did. They are however behaving exactly as if allowing anyone else to examine data, and independently decide the percentage of patients who responded, would undermine their claims.
     
    Valentijn, catly, ahmo and 1 other person like this.
  3. SOC

    SOC

    Messages:
    7,802
    Likes:
    16,454
    I've been wondering about this withholding data problem. Is this a recent development, or is it more common in certain fields? When I was researching and publishing, it was a given that you included your data in appendices so that anyone could verify your conclusions. I understand that's not possible with journal page number limitations, but in that case the data was prepared for immediate release upon request.

    IMO, there's something innately dishonest about refusing to release your data. What are you afraid of? If your work is sound, you should be happy for other researchers to see what you've done. ....Oh, I see.... :p
     
    Daisymay, Valentijn, JaimeS and 5 others like this.
  4. lansbergen

    lansbergen Senior Member

    Messages:
    2,410
    Likes:
    2,398
    Withholding data is like a constructor not informing the client what materials are used.
     
    SOC and JaimeS like this.
  5. Snow Leopard

    Snow Leopard Hibernating

    Messages:
    4,577
    Likes:
    12,155
    South Australia
    It is truly bizarre that of papers stating "data available on request", only 44% of authors actually did so on request!
     
  6. Sean

    Sean Senior Member

    Messages:
    3,185
    Likes:
    17,325
    Withholding data is like demanding everybody else sign a contract you have written without giving them the chance to read it first.
     
    Valentijn, lansbergen and SOC like this.
  7. alex3619

    alex3619 Senior Member

    Messages:
    12,264
    Likes:
    33,684
    Logan, Queensland, Australia
    I wonder if withholding data should be considered grounds for retracting the study?
     
    SOC, Ritto, Sean and 2 others like this.
  8. lansbergen

    lansbergen Senior Member

    Messages:
    2,410
    Likes:
    2,398
    I think it should.
     
    Sean and alex3619 like this.
  9. adreno

    adreno PR activist

    Messages:
    4,735
    Likes:
    9,896
    Bad science would be flawed research methodology, or interpreting the data wrong. But deliberately withholding data is misconduct.
     
    alex3619 and Sean like this.
  10. Scarecrow

    Scarecrow Annie Gsampel

    Messages:
    1,850
    Likes:
    5,029
    Scotland
  11. alex3619

    alex3619 Senior Member

    Messages:
    12,264
    Likes:
    33,684
    Logan, Queensland, Australia
    Secrecy is the enemy of good conduct.
     
    Cheshire, SOC and Scarecrow like this.

See more popular forum discussions.

Share This Page