It is quite sad that your otherwise-excellent publication systematically fails to report error bars in your reports. Time and again I read articles and am left wondering whether the effect reported is even statistically significant. As just one example, this article said that the rate of subsequent infection from breast milk dropped from 12 percent to 8 percent. Given the numbers in each sample, it is quite likely that the difference reported is simply due to chance.

Kevin Lehmann
Princeton University
Princeton, N.J.

The results cited in the story were significant at a probability greater than 95 percent. As a general rule, for the sake of readability, Science News doesn’t include statistics on the significance of data. Readers should rest assured, however, that we don’t report results if a statistical test has failed to find them significant .–The Editors