SoylentNews is people
SoylentNews
U.S. Vice President Joe Biden has released another cancer progress report/wish list:
Vice President Joe Biden today released his vision for doubling progress against cancer over 5 years. It includes numerous policy recommendations and a laundry list of projects by the National Cancer Institute (NCI) and other federal agencies that would require additional funding. Biden and his wife, Jill, have met with thousands of experts and patient advocates, they explain in a 17-page strategic plan submitted to President Barack Obama, who asked Biden in January to lead the effort. "We sought to better understand and break down the silos and stovepipes that prevent sharing of information and impede advances in cancer research and treatment, while building a focused and coordinated effort at home and abroad," they explain.
The Bidens' wish list ranges from giving patients more control over their medical data to launching "a national conversation" about cancer drug pricing. They also want to see more high-risk research funding at NCI and changes to the institute's intramural research program to focus more on emerging science and major public health challenges.
An accompanying 29-page report from Biden's federal moonshot task force lists what agencies have done so far and their plans to address five strategic goals. The first goal, "catalyze new scientific breakthroughs," contains several items that "align with" the 10 research areas recommended last month by an NCI blue ribbon panel, the report says. For example, it describes Department of Defense (DOD) efforts to develop new imaging technologies for detecting early molecular changes in cells that may lead to cancer.
The Cancer Moonshot is part of the War on Cancer:
The mission of this Cancer Moonshot is not to start another war on cancer, but to win the one President Nixon declared in 1971. At that time, we didn't have the army organized, didn't have the military intelligence to know the enemy well, and therefore didn't have the comprehensive strategy needed to launch a successful attackânow we do. Because of the progress over the last 45 years we have an army of researchers and oncologists, the powerful technologies and weapons, and immense public support and commitment to action.
Related:
Biomedicine Facing a Worse Replication Crisis Than the One Plaguing Psychology
The White House Announces $121 Million Microbiome Initiative
"Cancer Moonshot" Releases Blue Ribbon Panel Report
Microsoft to "Solve the Problem of Cancer" Within Ten Years - Scientists are Skeptical
(Score: 0) by Anonymous Coward on Thursday October 20 2016, @03:52AM
Yes, because there really is a difference. The only use for "discovering" this over and over is to jump to some invalid conclusion.
(Score: 1) by khallow on Thursday October 20 2016, @02:11PM
(Score: 0) by Anonymous Coward on Thursday October 20 2016, @02:40PM
Then you have no hypothesis to test. At that point your job is to describe the situation/phenomenon as carefully as possible.
There is no legitimate point to checking if there is exactly zero correlation or zero difference between two sets of subjects unless this is predicted by someones model. This idea that there is a proper use for NHST and people are just misusing it is a common one, but it is baseless. There is no legitimate use for testing a hypothesis not predicted by anyone's model, only misuses are possible. See for example this link, where it is discussed that the founders of those methods also need to misuse the outcome of NHST when trying to put their methods into practice: http://andrewgelman.com/2016/09/10/my-talk-at-warwick-england-230pm-thurs-15-sept/ [andrewgelman.com]
Everyone misuses it because that is the only thing to do with it. At best, it is a waste of time that clutters up research reports with irrelevancies.
(Score: 1) by khallow on Thursday October 20 2016, @03:28PM
Then you have no hypothesis to test.
Hypotheses are not models. It's not at all hard to construct hypotheses of the form, "this parameter correlates with that parameter".
(Score: 0) by Anonymous Coward on Thursday October 20 2016, @03:46PM
1) That is not the hypothesis being tested by NHST. It is that there is zero correlation and the statistical model accurately describes the distribution of uncertainty. Ie it is the logical inverse with additional content.
2) I would call that a prediction, not a hypothesis, but ok.
3) Such "hypotheses" are too vague to be of any use. It amounts to saying an observation was consistent with a prediction that literally includes all possible outcomes except an infinitesimally small proportion at exactly zero. For a prediction to be useful it needs to allow us to meaningfully narrow down the possible reasons for whatever observation we are trying to explain.
This type of stuff has all been explained before by people better than me. For example, figure 3 here:
Paul Meehl. 1990. Appraising and Amending Theories: The Strategy of Lakatosian Defense and Two Principles That Warrant It. Psychological Inquiry 1990, Vol. 1, No. 2, 108-141. http://rhowell.ba.ttu.edu/meehl1.pdf [ttu.edu]
The sad truth is that so much money, time, and effort has been wasted on NHST that people have a mental block to accepting how bad it is. Just accept it: we need to redo almost everything since 1980, and a lot of stuff since 1950. Many highly intelligent people wasted their lives on pseudoscience, but I guess that's life. Try it out, the tension between logic and your worldview will disappear.
(Score: 1) by khallow on Thursday October 20 2016, @06:10PM
That is not the hypothesis being tested by NHST. It is that there is zero correlation and the statistical model accurately describes the distribution of uncertainty. Ie it is the logical inverse with additional content.
Sure, it's the negation of the hypothesis that's being tested.
I would call that a prediction, not a hypothesis
And I would call that meaningless word mincing that adds nothing to your argument.
Such "hypotheses" are too vague to be of any use.
Except of course, when you're wrong, which incidentally would be most of the time (except for really trivial correlations like X with X+constant which we would already know correlate in a known way via math). Then they aren't too vague to be of use.
It amounts to saying an observation was consistent with a prediction that literally includes all possible outcomes except an infinitesimally small proportion at exactly zero. For a prediction to be useful it needs to allow us to meaningfully narrow down the possible reasons for whatever observation we are trying to explain.
The exceptions are not infinitesimal.
The sad truth is that so much money, time, and effort has been wasted on NHST that people have a mental block to accepting how bad it is.
Just because a tool is misused doesn't mean that it doesn't work when used in a different manner (an "appropriate" manner, of course).
(Score: 0) by Anonymous Coward on Thursday October 20 2016, @06:12PM
Meh, you didn't really read any of my post. Maybe I will try again another day.
(Score: 1) by khallow on Thursday October 20 2016, @08:02PM