Become a fan of Slashdot on Facebook


Forgot your password?

Comment: Re:Is the math not towing the groupthink? (Score 1) 208

by drooling-dog (#49495301) Attached to: Social Science Journal 'Bans' Use of p-values

The math works fine; the problem is choosing the appropriate method. My hunch is that the biggest mistake in the use of stats in the social sciences is failing to correct p-values for multiple comparisons. That is, if you're hypothesis is limited to predicting an association between two variables, then p-values are just fine. But if you sent out a questionnaire with 20 questions on it and compute all 190 pairwise correlations between them, you'll get around 9 or 10 "significant" (p 0.05) but meaningless associations just by chance. You can't (or shouldn't) cherry-pick these and write them up like they mean anything. Yet many people do just this, often not realizing how the hypotheses were selected (it can sometimes be subtle, or buried in the history of the project).

Comment: Re:Even more obligatory (Score 1) 208

by drooling-dog (#49495153) Attached to: Social Science Journal 'Bans' Use of p-values

A useful exercise (if you can use basic statistics software) that illustrates this is to generate a bunch (say, 10 or 20) of series of random numbers and then compute the matrix of correlations (or t-values, if you prefer) between all of them. You'll find that roughly 5% of the correlations are "significant" at the p.05 level, even though the series are really random and independent. It's a trivial result and just what you'd expect by chance, but it does drive the point home that you can't rely on p-values alone if you're testing multiple hypotheses. In the latter case there are corrected measures available that take this into account.

Comment: Re: Lifestyle (Score 1) 332

by drooling-dog (#49458175) Attached to: California Looks To the Sea For a Drink of Water

It appears you're comparing total water consumption (including industrial and residential use) in other countries with residential use in California. I don't know how Cali stacks up against other states statistically, but the average you quote (178 g/day) is about 2.5X my own experience (2 person household) in Michigan.

Comment: Re:If it stops them from .... (Score 1) 23

by drooling-dog (#49395109) Attached to: How to Prepare for an IT Security Disaster (Video)

It helps to let executive management outside of IT know that you're doing something. Maybe periodic reports detailing intrusion attempts, right down to failed SSH logins (there are always lots of those). The value of defense goes up if it's clear to everyone that you're actually under siege.

Comment: Re:Disagree (Score 1) 1168

Private businesses are not public accommodations.

But that's exactly what they are, if they do business with the public. When you incorporate a business, the government grants you special rights, like being able to protect your personal assets from the liabilities of your business, and having access to a stable legal system to enforce the contracts that you make. In return, you have certain responsibilities about how you conduct your business; e.g., things covered by consumer protection laws, the way you keep your accounts, and not discriminating against classes of customers and employees out of your own bigotry. You want the advantages and protections of an incorporated business without having to follow the rules of a civilized society? Tough shit.

Comment: Send a letter (Score 3, Interesting) 107

If a bunch of Republican senators could get together and write a letter outlining the details of these abuses, I'm sure there wouldn't be any consequences (to themselves) whether the spying is classified or not. Plus, it would be a great way to limit the powers of the federal government and stick it to Obama at the same time!

Comment: Re:Didn't you get the memo? (Score 1) 320

Well, you could try making a solid, robust scientific argument that accounts for existing as well as new data. But if instead you want to put your faith in PR firms that are paid to manufacture public doubt on behalf of industries with vested interests, then you're building a political controversy and not a scientific one.

Comment: Re:exactly extreme exaggeration turns some off (Score 1) 458

by drooling-dog (#48944161) Attached to: Most Americans Support Government Action On Climate Change

...examples of leading climate researchers from Stanford, UC Berkeley, and Yale making statements like "by 2010, New York City will be underwater"

What "leading climate researchers" said this? Citations needed. Did you hear this on Fox News, or are you just making it up?

Rising CO2 levels and climate change are politically controversial only because the fossil carbon industry hired a bunch of PR firms to sow public doubt. Who needs science, when industry PR is gospel?

Comment: So what's the point? (Score 1) 351

by drooling-dog (#48898269) Attached to: Americans Support Mandatory Labeling of Food That Contains DNA

I'm not sure what the point is here. Could it be:

  • Some chemicals with unfamiliar-sounding names are harmless, therefore we should assume that all are?
  • Warning labels about chemical hazards are stupid, because the public should be sufficiently educated about chemistry and toxicology to know if a compound is dangerous by it's name alone?
  • Unfamiliar substances should be assumed to be safe unless we know otherwise with certainty?

Furthermore, if you use the name "di-hydrogen monoxide" for water, I'm going to assume you've had no chemistry beyond high school. No chemist would say "monosilicon dioxide" for quartz (SiO2) or "tri-iron tetra-oxide" for Fe3O4, for example. So if you're ridiculing people for not recognizing "dihydrogen monoxide", you're also looking like an noob to people who know better.

You can't take damsel here now.