Posts by John Mikesell

We're raising generations of MUTANT KIDS, says Icelandic study

Blue Moon

Mutation provides raw material for selection.

Lesbians like straight men, researchers find

Statistical significance has nothing to do with one's predilictions

A sample size of 20 could be quite representative. No one can tell either way from an article in the popular press. As to statistics (complained about by an earlier commenter--who seemed more interested in sticking to beliefs rather than openly considering new information), popular press articles are always a lousy place to judge scientific merit and statistical significance. Doesn't have anything to do with the subject of the article. If an individual is really interested in the statistical basis of scientific results reported in the popular press, he or she can go to the source, in this case the "Proceedings of the National Academy of Sciences" either online or in a decent library at a research University and assess the scientific results directly. Scientific results are always a matter of public record, just depends if one wants to make the effort (which admittedly can be substantial) of analyzing the record--but unwillingness to make the effort does not invalidate scientific results. Significant sample size depends not on one's predispositions, but on the level of variation represented by the phenomena one is studying. It also depends on how representative a sample is of the population one is studying. A sample size of 7 individuals may be significant in some cases, while in other cases a sample size of 100 might generate indeterminate results.