July 2011 Archives

• The p-value doesn't tell scientists what they want (it is the probability of the data given that H0 is true, and scientists would like the probability of H0 or H1 given the data)
• H0 is often known to be false
• P-values are widely misunderstood
• Leads to binary yes/no thinking
• Prior information is never taken into account (Bayesian argument)
• A small p-value could reflect a very large sample size rather than a meaningful difference
• Leads to publication bias, because significant results (i.e. p < 0.05) are more likely to be published

Image via Wikipedia

Aaron Swartz, a 24-year-old programmer and online political activist, has been indicted in Boston on charges that he stole more than four million documents from the Massachusetts Institute of Technology and JSTOR, an archive of scientific journals and academic papers. (Read the full indictment below.)

Mr. Swartz was indicted last Thursday by the United States Attorney for the District of Massachusetts, Carmen M. Ortiz, and the indictment was unsealed Tuesday. The charges could result in up to 35 years in prison and a $1 million fine. ("Internet activist charged in M.I.T. data theft," by Nick Bilton, The New York Times, 19 July 2011).

Image by pmarkham via Flickr

One Caribbean species, the Blue Iguana of Grand Cayman island, found nowhere else in the world, is looking like that rarest of things, a threatened species roaring back from the brink. Once down to perhaps fewer than a dozen animals, the long-tailed lizards, some growing to 5 feet and weighing 30 pounds, now number about 500, suggests a tally from a weeklong health screening that ended July 3.

As future scientists and educators, we, the undersigned students, encourage Congress and the President to make sustained investments in the nation's scientific research, education, and training programs. The extramural, competitive, peer-reviewed grant programs administered by federal agencies are critical to our nation's scientific enterprise and future.

...

Addressing the nation's current budget challenges is essential, but we must not sacrifice investments in our future. One way to grow our economy is through scientific innovation. Science, however, cannot move forward in an environment where one year brings a rapid budget increase and the next year a precipitous budget cut. Research and development require people, facilities, and equipment - recruiting, retaining, and building this infrastructure requires a sustained and predictable investment.

UNITED STATES FISH AND WILDLIFE SERVICE RESOURCE MANAGEMENT
For necessary expenses of the United States Fish and Wildlife Service, as authorized by law, and for scientific and economic studies, general administration, and the performance of other authorized functions related to such resources, $1,099,055,000, to remain available until September 30, 2013 except as otherwise provided herein: Provided, That none of the funds shall be used for implementing subsections (a), (b), (c), and (e) of section 4 of the Endangered Species Act, (except for processing petitions, developing and issuing proposed and final regulations, and taking any other steps to implement actions described in subsection (c)(2)(A), (c)(2)(B)(i), or (c)(2)(B)(ii) of such section):

(a) Gives the Secretaries of Interior and Commerce authority to determine whether species are threatened or endangered.
(b) Describes the basis for determining whether species are threatened or endangered and for designation of critical habitat.
(c) Provides for changes in the list of threatened and endangered species. The only part funded under the House Interior-Environment appropriations bill are those that would allow species to be removed from the list or to have their status changed from endangered to threatened.
(e) Provides for protection in "similarity of appearance" cases, i.e., where a threatened or endangered species might be confused with one that is not listed.

Image via Wikipedia

Over the past 50 years, we've seen a number of gigantic policies produce disappointing results -- policies to reduce poverty, homelessness, dropout rates, single-parenting and drug addiction. Many of these policies failed because they were based on an overly simplistic view of human nature. They assumed that people responded in straightforward ways to incentives. Often, they assumed that money could cure behavior problems.

Fortunately, today we are in the middle of a golden age of behavioral research. Thousands of researchers are studying the way actual behavior differs from the way we assume people behave. They are coming up with more accurate theories of who we are, and scores of real-world applications. (emphasis added)

Many readers complain about what they view as a missing comma in a sentence like this: He bought apples, pears and bananas.

Style guides for book and academic publishing generally would insist on another comma after "pears," the so-called serial comma. But news writing has traditionally omitted the serial comma -- perhaps seeking a more rapid feeling in the prose, or perhaps to save time and effort in the old days of manual typesetting.

We do use the additional comma in cases where a sentence would be awkward or confusing without it: Choices for breakfast included oatmeal, muffins, and bacon and eggs.

Whether to include the serial comma has sparked many arguments. But it's easily answered in favor of inclusion because omitting the final comma may cause ambiguity, where as including it never will.

[T]o have crafted a page-turner out of the history of statistics is an impressive feat.

With her severely injured eyes covered by dark glasses and her dreams brutally derailed, Rumana Monzur returned on Tuesday to Vancouver, where she wept as she spoke of her hopes for the future.

...

"It's so different - but it feels great to be here among all you beautiful people," Ms. Monzur said, speaking through tears as her father gripped the handles of her wheelchair. "And I want to see you beautiful people again. I really do." ("UBC student maimed in Bangladesh attack returns to Vancouver", by Wendy Stueck, The Globe and Mail, 6 July 2011)

Graphic from the New York Times

The medium variant of the 2010 revision produces a 2050 world population that is, as Pearce notes, 156 million higher than that projected in the 2008 revision. This 1.7% difference is comparable to that between earlier revisions produced during this decade.

1. Pearce and Zlotnik agree that the projection for 2050 is 156 million higher than in the 2008 revision.
2. That difference amounts to only 1.7 percent of the total expected (roughly 9 billion).

Since the projections don't diverge until after 2050, it may not make much of a difference to policy makers in any context. Even long-term projections for social insurance programs like Medicare and Medicaid in the U.S. don't go out beyond 2050. By the time we need accurate projections for 2100, we'll be a lot closer to it, and small differences in fertility rates will have far less of an impact on those projections.

Going after Mr. Mann only discourages the sort of scientific inquiry that, over time, sorts out fact from speculation, good science from bad. Academics must feel comfortable sharing research, disagreeing with colleagues and proposing conclusions -- not all of which will be correct -- without fear that those who dislike their findings will conduct invasive fishing expeditions in search of a pretext to discredit them. That give-and-take should be unhindered by how popular a professor's ideas are or whose ideological convictions might be hurt.

Scientific progress depends on transparency, the Board said, but "the sharing of research data is vastly different from unreasonable, excessive Freedom of Information Act requests for personal information and voluminous data that are then used to harass and intimidate scientists."

For teaching me that the Oxford comma resolves ambiguity, I'd like to thank my parents, Sinead O'Connor, and the Pope.

For teaching me that the Oxford comma resolves ambiguity, I'd like to thank my parents, Sinead O'Connor and the Pope.