I. Data Collection

• Humans are sometimes treated in ways that are unethical when collecting data, and samples can be biased in ways that are unethical

• Examples include prisoner studies and the Milgram Experiment, in which subjects were ordered to administer progressively stronger "electrical shocks" to actors who were in another room

• Researchers can intentionally or unintentionally bias their study by using a sample that is biased. Sampling bias can seriously distort results and lead to wrong or misleading conclusions.

• Examples include:

  • cherry picking (favor specific groups who sponsor the research)

  • non-respondent bias (which occurs when those who do not respond to a survey differ from those who respond)

  • interviewer bias (the manner in which a question is asked may affect the response)

  • volunteer bias (people who volunteer to participate constitute a voluntary response sample, and they often have different characteristics when compared to participants who are selected by those conducting the survey)

II. Analysis

• Study findings are often distorted by improper analysis

• Falsified data: several surveys have attempted to determine the incidence of data falsification or fabrication among scientists

• Inappropriate statistical methods: the inappropriate use of statistical methods may lead to incorrect findings and distorted results even if the data are sound

• It is important to choose a proper significance level for a hypothesis test BEFORE collecting sample data and finding a p-value. It is bad practice to adjust the significance level after receiving results in order to make the results appear significant when they were originally not with the prior significance level

III. Ethics and Reporting

• Ghostwriting: a medical ghostwriter is someone who contributes to the writing of a research study or article but is not acknowledged in the published work

• Financial support: Organizations or companies may have business ties with the manufacturers of their products, and these relationships should be disclosed

• Reporting nonsignificant results: Only report significant results to ensure you are delivering the proper conclusions

IV. Enforcing Ethics

• The Common Rule has been established in the US as the standard of ethics for biomedical and behavioral research involving human subjects. The requirements are that:

  • People who participate as subjects in covered research are selected equitably and give their full informed, voluntary written consent

  • The proposed research be reviewed by an independent oversight group referred to as an Institutional Review Board (IRB) and approved only if risks to subjects have been minimized and are reasonable relative to anticipated benefits

• As consumers of data, it is important to have a healthy skepticism when statistics are cited in support of a finding. We should ask questions such as:

  • Who is reporting about the study?

  • Do they have any financial/personal interests or biases?

  • How were the sample data obtained?

  • Have these findings been reviewed or replicated by peers or anyone else?

  • What specifics do they provide about the methods used?

• Researchers must always adhere to the highest ethical standards in the work. We should all strive to behave ethically as follow:

  • Be complete and honest about findings, even if results are not what was expected/desired.

  • Seek advice of professionals when unsure about which statistical analyses are appropriate.

  • Always disclose financial relationships or any interests in the outcome of the research

  • Acknowledge only those who made meaningful contributions to the study.

  • Be ready and willing to share data, methods, analyses, and results.

  • Clearly identify any assumption, limitations, or outstanding questions in the research.