Abstract
effect size, from a single study, and then combines effect sizes between studies on the same topic (along with estimates of sample size and variance) to allow detection of the overall magnitude of effect. In this way, even if studies give somewhat conflicting answers to the same treatment, an overall effect among studies can be calculated to achieve a more conclusive answer. While meta-analysis is a powerful tool to overcome the variation among studies and arrive at an answer to a particular scientific question (e.g., does a particular intervention alleviate the symptoms of a disease?), it is less powerful in its ability to detect publication bias and the selective presentation of analyses. In the biomedical sciences, such biases not only slow the progression of science, but they could also result in bringing ineffective or harmful substances to clinical trial, creating considerable financial and health costs. Thus, it is important to understand just how rampant these biases are. In the current issue of PLOS Biology, Tsilidis and colleagues take the bold step of examining bias by employing a relatively new type of approach—a sort of meta-analysis of meta-analyses. This allowed them to assess whether the numbers of studies finding statistically significant effects of a biomedical intervention were higher than what would be expected if there were no bias. Specifically, they analyzed 160 separate meta-analyses comprising more than 1,000 studies that used animal models to evaluate the efficacy of interventions of six major neurological disorders—Alzheimer disease, multiple sclerosis, two types of stroke, Parkinson disease, and spinal cord injury—4,445 comparisons in all. A large proportion of these meta-analyses (nearly 70%) reported an overall positive effect of the tested interventions on the affliction. However, most of these meta-analyses also reported
Highlights
Recent philosophical dissection of the scientific method can be caricatured as a polar debate between Karl Popper’s sober view of objective development and falsification of hypotheses and Thomas Kuhn’s more glamorous espousal of a role for ideology and subjectivity; ‘‘real science’’ as performed by the authors of this and other journals is probably a rich mix of the two
For example, that results in support of a given hypothesis are more likely to be published in a ‘‘higher impact’’ journal than are negative results, leading to what’s known as ‘‘publication bias.’’ Even within a study, bias can emerge from the choice of experimental design and/or the presentation and analysis of the results
Meta-analysis takes the magnitude of difference between treatments, known as an Selected PLOS Biology research articles are accompanied by a synopsis written for a general audience to provide non-experts with insight into the significance of the published work
Summary
Recent philosophical dissection of the scientific method can be caricatured as a polar debate between Karl Popper’s sober view of objective development and falsification of hypotheses and Thomas Kuhn’s more glamorous espousal of a role for ideology and subjectivity; ‘‘real science’’ as performed by the authors of this and other journals is probably a rich mix of the two. Animal models are essential to triage possible therapeutic interventions for human diseases prior to possible clinical trials. Dozens of studies are often published that examine the effect of a particular intervention on animal models.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call