False-positives in social science research can have particularly negative impacts when research is used to inform policy or other outcomes. These outcomes can range, from the endorsement of the idea that simply posing in a powerful stance can actually empower someone, to the suggestion that increasing austerity measures leads to economic growth.In this article, authors Joseph Simmons, Leif Nelson, and Uri Simonsohn, address the costly issues and frequency of false positives in social science research and suggest six requirements for authors that they believe are simple, low-cost, and straight-forward solutions to the problem.They first define a false positive as the “incorrect rejection of a null hypothesis”. The problem with false positives is that once they are published, they are persistent and there is little incentive to replicate findings to test their significance. The authors write “…false positives waste resources: they inspire investment in fruitless research programs and can lead to ineffective policy changes. Finally, a field known for publishing false positives risks losing its credibility.”They argue that, with current standards of disclosure, false positives are actually “vastly more likely” due to the ease with which researchers can publish “statistically significant” evidence for nearly any hypothesis.Many of these issues can be attributed to researcher degrees of freedom, or the multitude of decisions researchers can make about the design and details of their experiments. This freedom increases the likelihood that their analyses will produce false positives, and can be attributed to two factors: “ambiguity in how to best make decisions, and the researcher’s desire to find a statistically significant result”, the latter referring to the temptation and likelihood of researchers coming to conclusions that are consistent with their own desires or beliefs.The authors performed simulations to estimate the influence of researcher degrees of freedom on the probability of a false-positive result. They focused on four common degrees of freedom that increase the likelihood of a researcher falsely detecting a significant effect. These include flexibility in:
- choosing among dependent variables,
- choosing sample size,
- using covariates, and
- reporting only subsets of experimental conditions.
Want to keep
University of California, Berkeley online course,
Transparent and Open Social Science Research
- Decide the rule for terminating data collection before data collection begins and report this rule in the article. This prevents authors from adding additional observations and further testing to achieve statistical significance if initial results are non-significant.
- Collect at least 20 observations per cell or else provide a compelling cost-of-data-collection justification. Small samples are “simply not powerful enough to detect most effects” and are “more likely to reflect interim data analysis and a flexible termination rule”.
- List all variables collected in a study. This prevents researchers from only reporting convenient subsets of measurements and allows readers to identify degrees of freedom.
- Report all experimental conditions, including failed manipulations. This “prevents authors from selectively choosing only to report conditions that yield results consistent with their hypothesis.”
- Report the statistical results of eliminated observations as if those observations had been included. This requires authors to explain why they eliminated the data and encourages readers to consider the validity of the exclusion.
- Report the statistical results of the analysis without a covariate if the analysis includes a covariate. This requires authors to “justify use of the covariate,” reveals “the extent to which a finding is reliant on the presence of a covariate,” and, again, encourages readers to practice discernment about whether the covariate is warranted.
- Ensure that authors follow the requirements.
- Be more tolerant of imperfections in results (false-positive findings could be due to an “unreasonable expectation” imposed by reviewers for data to turn out as predicted).
- Require authors to demonstrate that their results do not depend on arbitrary analytic decisions.
- Require authors to conduct an exact replication if justifications of data collection or analysis are not compelling.
ReferenceSimmons, Joseph P., Leif D. Nelson, and Uri Simonsohn. 2011. “False-Positive Psychology Undisclosed Flexibility in Data Collection and Analysis Allows Presenting Anything as Significant.” Psychological Science 22 (11): 1359–66. doi:10.1177/0956797611417632.
Transparent and Open Social Science Research
Our purpose is to transform access to education.
We offer a diverse selection of courses from leading universities and cultural institutions from around the world. These are delivered one step at a time, and are accessible on mobile, tablet and desktop, so you can fit learning around your life.
We believe learning should be an enjoyable, social experience, so our courses offer the opportunity to discuss what you’re learning with others as you go, helping you make fresh discoveries and form new ideas.
You can unlock new opportunities with unlimited access to hundreds of online short courses for a year by subscribing to our Unlimited package. Build your knowledge with top universities and organisations.