Interesting Links for 04-11-2018

[andrewducker]

Number of vegans in UK soars to 3.5 million

(tags: vegan uk society )

Twitter and Facebook both making life easier for Nazis

(tags: nazis racism twitter facebook )

Vegan Myths: Avocados Are Not Vegan

(tags: food vegan myths bees )

Government cracking down further on tax relief for landlords

(tags: housing tax uk )

Jews Should Concern Americans More Than Russian Influence, Nigel Farage Says

(tags: Russia Jews OhForFucksSake UKIP usa )

Housing can’t both be a good investment and be affordable

(tags: economics housing )

"Eliminationism" is a term you need to become familiar with

(tags: racism usa history Nazis politics republicans )

1 in 4 Statisticians Say They Were Asked to Commit Scientific Fraud

(tags: statistics fraud science OhForFucksSake )

Theresa May secures Brexit deal that will keep UK in a customs union (rumour)

(tags: UK europe )

An error message, still found in Windows 10, is a mistake from 1974

(tags: windows operatingsystems history )

Tory DWP Secretary Esther McVey fails to declare official link to shady £20m Political Campaigning Firm

(tags: Conservatives OhForFucksSake )

Segway was supposed to change the world. Two decades later, it just might

(tags: segway transport Technology )

What Minimum-Wage Foes Got Wrong About Seattle

(tags: minimumwage economics usa )

Comments

1 in 4 Statisticians Say They Were Asked to Commit Scientific Fraud

[doug]

This is less bad than the statistic makes it look, I think. The "1 in 4" figure invites you to imagine that this is the proportion of outright fraudulent statistics in published papers, but it's nowhere near that.

This data is a sample of consulting statisticians. They will have talked to a *lot* of scientists, and the question asked was a "have you ever" question. On the other side of the equation, people requesting statistical dodginess are likely to get turned down, and will then go on ask to another statistician, and another, and another, until they mend their ways, give up in exhaustion, or find one who gives in. That will mean a small handful of dodgy scientists are likely to contact a very large number of consulting statisticians.

(I'm not a statistician, but I have taught stats, and am probably number three choice for answering stats questions in my dept, and I get *loads* of people asking me for advice. I'm also good friends with the number one and number two choices and sometimes share stories when someone does the rounds because they didn't like the answer one of the others gave because it meant more work.)

The positive way of framing this study is "three quarters of people who spend their time answering questions from scientists who don't have enough grasp of statistics to do it themselves have never even been asked to do things that would, if done, count as fraudulent." I'd be surprised if a similar study didn't reach similar conclusions among, say, professional auditors - and they are on the hook for fraud in ways that consulting statisticians wouldn't be.

I'm not saying it's not disappointing and depressing, and I'm certainly not saying that there are not serious, widespread problems with the use of quantitative data in academic publications, but it's not nearly as bad as the "1 in 4 figure" instinctively makes you think it is.

Re: 1 in 4 Statisticians Say They Were Asked to Commit Scientific Fraud

[andrewducker]

Thank you, that's reassuring.