Articles to 2012-01-11

Zum Seitenende      Übersicht Artikel      Home & Impressum

First the link to this week’s complete list as HTML and as PDF.

There is a reason for the Flynn-effect and the need to renormalize the test results for each year cohort. In Brinch I fail to see anything more than a simple training effect.

Talk about standing things on their head. It used to be the case, that parents paid to send their children to school and many parents still do. I don’t want that back, as it excluded many, who really wanted an education but couldn’t afford it. But throwing out money to pay people for, please, please, accepting a free gift? Not surprisingly Fryer concludes: "[T]he impact of financial incentives on student achievement is statistically 0 in each city." (Mat. 7, 6)

Reshef’s result (also see Speed) may well be helpful in finding valid relationships, but I fear its main effect will be another rise in purely spurious published "results" from data mining and willful use of wrong statistic tests.

McNiven et al. seem to confirm Boyd & Richerson’s hypothesis that food production began in the early Holocene, when conditions first allowed it, everywhere all over the world. What I dislike is the pattern of single or very few very early dates with all others coming much later and nothing in between and I want to see more data before I believe the result.

And lastly Joseph Brean has another take on the basic problem of current science as evidenced by Stapel.

Science, at its most basic, is the effort to prove new ideas wrong. The more startling the idea, the stronger the urge to disprove it, as was illustrated when European physicists last month seemed to have seen particles travel faster than light, which has prompted a massive effort to replicate (or more likely debunk) such a shocking result.

Although science properly gets credit for discovery and progress, falsifiable hypotheses are its true currency, and when scientists fail to disprove a false hypothesis, they are left with a false positive. [A] false positive is “perhaps the most costly error” a scientist can make.

Both my father’s diploma and doctoral theses were of this kind, both dealing with recently found exciting effects, that were hot topics at the time. Both were only possible, because thesis advisors at the time still saw it worth their and their students’ while to reproduce already published results in their own labs.

Zum Anfang      Übersicht Artikel      Home & Impressum

Creative Commons Attribution-Share Alike 3.0 Unported License Viewable With Any Browser Valid HTML 4.01! Valid CSS!