1.14: We’ve got to watch out for our biases.

In 2001, the journal Behavioral Ecology changed its policy for reviewing manuscripts that were submitted for publication. Its new policy instituted a double-blind process, whereby neither the reviewers’ nor the authors’ identities were revealed. Previously, the policy had been a single-blind process in which reviewers’ identities were kept secret but the authors’ identities were known to the reviewers. In an analysis of papers published between 1997 and 2005, it turned out that after 2001, when the double-blind policy took effect, there was a significant increase in the number of published papers in which the first author was female (FIGURE 1-17). Analysis of papers published in a similar journal that maintained the single-blind process over that period revealed no such increase.

This study reveals that people, including scientists, may have biases—sometimes subconscious—that influence their behavior. It also serves as a reminder of the importance of proper controls in experiments. If knowing the sex of the author influences a reviewer’s decision on whether a paper should be published, it is also possible that researchers’ biases can creep in and influence their collection of data and analysis of results. (Even the decision about what—and what not—to study can be influenced by our biases.)

It can be hard to avoid biases. Consider a study that required precise measurement of the fingers of the left and right hands—comparing the extent to which people in different groups were physically symmetrical. The researcher noted that when she measured individuals from the group she predicted would be more symmetrical, she felt a need to re-measure if the reading on her digital ruler indicated a big asymmetry. Because she was thinking that the person’s left- and right-hand fingers should be symmetrical, she assumed she had made an error. She felt no such need to re-measure for subjects in the other group, for whom asymmetries confirmed her hypothesis. To control for this bias, the researcher connected her digital ruler to a computer and, without ever having the number displayed on the screen, transmitted the measurement directly to the computer when she pushed a button. In this way, she was able to make each measurement without introducing any regular bias.