Sound scientific studies have the following characteristics:

• Random. A study is randomized when the participants are separated into control and test groups in a random manner, such as by a pre-determined formula or software. By randomly assigning study participants, scientists decrease the possibility for biased results.
• Double-blind. In double-blind studies, both the study participants and the scientists are unaware of whether the participant is in the control or test group. For example, in some clinical trials, neither the researchers giving the treatment nor the study participants receiving it know if they are receiving a placebo (the control group) or the drug or vaccine (the test group). Double-blind studies are the most reliable because they eliminate potential for bias on the part of both the researchers and the participants.
  
  Sometimes, however, it is impossible to perform a double-blinded study. An example would be a study evaluating the best way to provide a patient with verbal instructions for taking a medication. In this case, the researcher will know which version of text was used, but the patients will not know whether they are in the test or control group. When only the study participants are unaware of the group they’ve been assigned to, it is called a single-blinded study. Sometimes, it’s unethical to do a double-blind, placebo-controlled study, such as the evaluation of Ebola vaccines during the 2014 outbreak of Ebola in West Africa.
  
  In rare instances both the researcher and the participant know the group to which the patient has been assigned, such as when testing a new cancer treatment in someone who has no other options for treatment.
  
  While the double-blind study design is considered the gold standard, this format may not always be an ethical or technically possible option. In these instances a single-blinded or un-blinded study format may be employed. As you read media reports about studies, it is important to determine the study method used.
• Large sample size. Large sample sizes allow researchers to account for individual differences such as genetics, income, race and environmental or lifestyle choices.
• Multiple studies. Study results must be repeatable in order to be widely accepted. If a researcher tries to replicate a study’s findings and fails, it is possible that an intentional or unintentional difference was introduced that caused the different findings. Many researchers will look at similar questions in different ways; only when a finding has been reproduced many times in a variety of populations is it widely accepted.

However, not all scientific studies will meet every one of these criteria. That doesn’t mean they should simply be dismissed because they still may contribute important information. We also know that science is often interpreted and reported by third parties, such as journalists or others sharing it, so there are other ways to evaluate studies. For example, in this article, Twenty Tips for Evaluating Scientific Claims, the authors point out that a basic understanding of the “imperfect nature of science” would be an important skill for sorting through third-party interpretations of science. To help, the authors set out to “suggest 20 concepts that should be part of the education of civil servants, politicians, policy advisers and journalists — and anyone else who may have to interact with science or scientists.” The result is a great introduction to evaluating scientific claims.

The concepts included in the list include paragraph-sized explanations related to statistical concepts such as variation and bias, scientific method such as sample size and controls and interpretive error.

The best way to determine the strength of a study is to read the original paper. However, because most of us do not have the time or expertise to evaluate all scientific studies that are published each week, we rely on others, such as news outlets, to share accurate assessments with us. Therefore, these organizations should be held to high standards, and as consumers, we should assess each statement made in reports of scientific topics based on the criteria mentioned above.

Check out more about how science works and may be misinterpreted or find other resources for evaluating information.