Culture of science
Science is the best way we know to develop reliable knowledge. It’s a collective and cumulative process of assessing evidence that leads to increasingly accurate and trustworthy information.
Science relies on empirical evidence and testable explanations. By basing its conclusions on multiple lines of evidence drawn from experiments and observations, science seeks to build reliable knowledge and provide scientific explanations that people can use to better understand the world around them and inform their decision making.
A researcher launches a balloon carrying a radio transmitter (called a radiosonde) to measure temperature, air pressure, relative humidity, and wind speed above the Greenland ice sheet. Data gathered will help scientists better understand the atmospheric and cloud processes that affect the ice sheet, which is melting rapidly due to climate change. Credit
Right. Scientific research is both an individual and a collective activity. Scientists may act on their own in gathering data and working out ideas. But they then need to convey to others their conclusions and the methods and evidence on which those conclusions are based. These observations and conclusions can be checked and extended by other scientists.
Scientific explanations as well as evidence upon which the explanations are based should be testable. Testing an explanation requires exploring its logic and limits and making observations that either support or refute it. Once an explanation is well supported by the cumulative evidence, that explanation can be considered valid and accepted.
Scientists use observations of phenomena or objects, experiments, or computer simulations to gather information during their research. This information may be quantitative (expressed in terms of numbers) or qualitative (expressed in terms of categories or characteristics).
The kinds of data used are as varied as research itself. Data can include
• text,
• numerical information,
• images, or
• video and audio recordings.
To the extent possible, scientists are expected to make their data and the analyses that determined their results openly available.
Experts in Kenya discuss experimental plots of corn as they develop drought-tolerant hybrids, part of the Water Efficient Maize for Africa project. Credit
So that others can check and extend their work. Making data and analyses available allows honest errors to be uncovered more quickly. It also deters any temptation to fabricate or falsify results, because results can be checked against the original or new data.
Sometimes data have to be kept private—for example, to protect the privacy of patients in biomedical research. In such cases, researchers are expected to find other ways of submitting their results to the judgment of peers.
A microbiologist prepares to collect samples of a coral reef’s microbiome—all the microscopic life living in and on the coral. Coral reefs are imperiled due to many factors, including warming ocean water. Understanding the composition of the coral microbiome will help scientists determine its role in coral health and physiology. Credit
Confidence in reported results increases if those results are replicable. Other scientists using the same methods should be able to reach results similar to those reported by researchers who are studying the same scientific question. However, this isn’t always possible in practice. For example, reported results may depend on an experiment that’s difficult to re-create, an event that occurs only once, or specialized information that’s difficult to convey to others.
Researchers studying 240-million-year-old fossils are gaining new insights into how dinosaurs grew from hatchlings to adults. Credit
No. Sometimes the opposite is true. Non-replicability leads to deeper understanding and new discoveries because it highlights aspects of a scientific question that were previously overlooked. In other cases, non-replicability points to problems in the design, conduct, or communication of a study that can hamper the progress of science.
For example, several years ago two separate labs used what they thought was the same protocol in a study of breast tissue, but they got different results. Baffled, researchers from each lab conducted the experiment side by side and discovered that in the course of the experiment, one lab was stirring the cells gently while the other lab was shaking them vigorously. Both methods are commonplace, so neither thought to mention it when describing the mixing process used. However, they discovered that the mixing method affects the outcome of the experiment. Clarifying the mixing method became a way to avoid problems replicating results.
Scientists can make mistakes, just like everyone else. But if the research was conducted with rigor and the methods, data, and logic for making conclusions are clearly stated, mistakes are less likely and more easily discovered when they do occur. Strategies for improving scientific rigor include things like “blind” trials in which investigators don’t know the most likely outcome. That can reduce potential bias. Also, study of the technologies and procedures used to generate the results can reveal possible problems.
Scientific research is a human activity and therefore subject to flaws. But it’s a uniquely powerful way of learning more about the world and how it works.
Want to confirm this information is accurate? Review the details at Reproducibility and Replicability in Science.
