Wednesday, October 14, 2009

OK, what about calling me susie?

The Numbers Guy looks at some of the latest "spanking new" research on corporal punishment. Spanking Research Needs a Time Out - WSJ.com

Statistical analysis of spanking's effects on cognition are clouded by many complicating factors. Effects can be attributed to the wrong cause, statisticians say; rather than spanking causing problems in children, it is possible that their existing cognitive problems can make spanking more likely. Moreover, any effects of spanking are difficult to measure and probably small. And unlike, say, a study on prescription drugs that removes a misleading placebo effect, no ethical study can assign some children to be spanked. Instead, parents must be trusted to remember and share their disciplinary practices.
Daniel Mundfrom, a statistician at New Mexico State University in Las Cruces, says that even without accounting for other factors, spanking at age 1 explained less than 1% of the variation in cognitive ability at age 3. In other words, maybe spanking does lower intelligence, but not by much.
....
Martin Wells, a statistician at Cornell University, re-ran the statistical test to check whether regional variations in IQ -- which is lower in Latin America and Africa -- could account for the IQ differences Prof. Straus found. After accounting for regional variations, Dr. Wells found the effect of spanking vanished. Dr. Wells plans to use the Prof. Straus's research in the classroom to demonstrate why it is important to consider alternative explanations.

No comments: