EDITORIAL

Measuring and managing bias

See allHide authors and affiliations

Science  01 Sep 2017:
Vol. 357, Issue 6354, pp. 849
DOI: 10.1126/science.aap7679

eLetters is an online forum for ongoing peer review. Submission of eLetters are open to all. eLetters are not edited, proofread, or indexed.  Please read our Terms of Service before submitting your own eLetter.

Compose eLetter

Plain text

  • Plain text
    No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Author Information
First or given name, e.g. 'Peter'.
Your last, or family, name, e.g. 'MacMoody'.
Your email address, e.g. higgs-boson@gmail.com
Your role and/or occupation, e.g. 'Orthopedic Surgeon'.
Your organization or institution (if applicable), e.g. 'Royal Free Hospital'.
Statement of Competing Interests
CAPTCHA

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.

Image CAPTCHA
Enter the characters shown in the image.

Vertical Tabs

  • RE: Measuring and Managing Bias
    • Donald R Taves, Affiliate Associate Professor, Dept. Oral Health Sciences, Univ of Washington

    The editorial “Measuring and managing bias,”(1) advocates randomization to make sure the groups are “as alike as possible”. Unrestrained randomization is seldom used in clinical trials because of the possibility that clinical conditions can change with time, particularly with infectious diseases, where strings of assignments to one or the other groups can cause chronological bias. The restraint in the numbers assigned to groups has the undesired side effect of making some of the assignments predictable if not certain and hence vulnerable to selection bias. This sets up the horns of a dilemma. Strengthen the protection against selection bias or chronological bias and you weaken the protection against the other. A new way of controlling selection bias that avoids this dilemma has been proposed. It can be used with randomized block designs to control the type that has received almost all of the attention to date. It monitors the assignments made when the probability of the next assignment to a particular treatment group is high. Over usage of those assignments by a few investigators suggests selection bias. This is determined before the patient is assigned so it can be remedied immediately by having the patient wait until another investigator submits a patient. Control of the bias introduced by several investigators favoring a few special patients is more difficult. It requires having data for large numbers of variates for each subject. This can be accomplished b...

    Show More
    Competing Interests: None declared.
  • Adding random bias plays a key role for suppressing the effect of bias itself

    Jeremy Berg Wrote an article entitled "Measuring and managing bias" in Science Editorial (1). The GPS technology has been used for many years where a GPS receiver is embedded in our smartphone. Typical received signal power from a GPS satellite is −127.5 dBm (0.178 fW), while thermal noise floor is −111 dBm (2). In other words, the GPS received signal strength is very much smaller than noise. Why can we use the GPS positioning? The GPS technology uses the noise magic for denoising the effect of noise itself by adding random noise with zero mean. Bias is a kind of noise. The effect of individual bias can be suppressed by adding random bias in the same scheme. Adding random bias plays a key role for suppressing the effect of bias itself.

    References:
    1. Jeremy Berg, Measuring and managing bias, Science Sept. 1 2017, 357 (6354), 849
    2. Y. Takefuji, Future mobile phones, Iwanami 2003

    Competing Interests: None declared.