| from the actual paper (linked at the bottom of the coffitivity post): "A total of 188 ideas were generated, for an average of 4.48 (SD = 2.09) ideas per person. The noise level did not affect the number of solutions generated" "ideas generated by participants in the moderate- (vs. low-) noise condition were rated as more original (M = 3.87 vs. 3.66; F(1, 40) = 4.76, p < .05)." Ratings were on a 7-point scale (1 = not at all, 7 = very much)... so ... a difference in originality of (3.87-3.66)/7 = 3% (i.e. not actually particularly significant (in the usual use of the word) even though the 3% difference was statistically reliable) "... a significant effect of noise on this appropriateness index, such that ideas generated by respondents in the moderate-noise (vs. low-noise) condition were rated as more appropriate (M = 4.48 vs. 4.20; F(1, 40) = 5.34, p < .05)." again, (4.48-4.20)/7 = 4% difference... not exactly impressive, although apparently it was statistically reliable It always pains me to see scientific research reported in the media in such a way as to inflate the significance of the work ("significance" used here in the layperson sense of the word, not the statistical sense of the word). PS the statistical significance was only reported as "p < .05", which means there is a 1 in 20 chance of getting a difference that large or larger due to random sampling alone (i.e. in the absence of an actual effect). This XKCD cartoon is worth a look: http://www.xkcd.com/882/ |