The core of science is testing ideas with evidence. Real science does not follow a single fixed list of steps done by one person alone; it depends on a community, and the scientific community acts as fact checker and critic, evaluating evidence and ideas. The community checks ideas through publication, peer review and replication. Peer review is one of the checks, and replication is another.
Spontaneous generation was the old idea that meat could turn into flies. In 1668 Francesco Redi tested it with jars of meat, and he wrote down what happened in each jar, so others could compare. In 1747 James Lind, a naval surgeon, compared six treatments for scurvy in a controlled trial. Redi tested the idea and recorded the outcome, and Lind compared the treatments in a trial.
In the 1840s a doctor in Vienna, Ignaz Semmelweis, noticed that women whose doctors came from the autopsy room without washing their hands had very high death rates. Before germs were known, high death rates in maternity wards were blamed on forces like polar currents or radiation from the constellations. Semmelweis made a prediction and tested it: on 15 May 1847 he ordered everyone to scrub with chlorinated lime between the autopsy room and the delivery room. At Vienna's hospital the maternal death rate fell from 12% to 1% in two years, but Semmelweis met strong opposition, and his findings were resisted by hospital and medical authorities.
In 1854 John Snow, a London doctor, suspected the water from the Broad Street pump after a burst of cholera. Snow found little to see in the water, so he held back from a firm conclusion at first. His further inquiry found no other circumstance common to the area except the water of that pump. Snow wrote his reasoning down himself, and his words are a primary source, a first-hand record from the time.
In 1907 Oscar Pfungst explained Clever Hans: the horse was reacting to tiny signals from the person asking. Pfungst's test was simple. Hans could not answer if the person asking did not know the answer, or if a screen kept him from seeing the questioner's face. The lesson is over 100 years old: if you know the answer, you can cue the person or animal you are testing without meaning to, so good experiments are built so the person watching cannot cue the result.
In peer review, a journal's editors send an article to several other scientists in the same field. Peer review helps assure that published research meets minimum standards for scientific quality. Peer-reviewed work meets the standards of science but is not automatically correct. A report in which scientists describe their own research is a primary source, and primary sources tend to come first in the publication cycle, while secondary sources come later and comment on them.
Replication means another scientist repeating a study to see whether the same result appears. A result that cannot be replicated suggests the first understanding or method was not good enough, and when a study cannot be replicated, scientists suspect that their understanding or their methods are not good enough. In some fields, scientists replicate their own results before they publish, to be sure the finding was not a fluke. Scientists often share their methods so others can replicate the work and weed out fraud or sloppy work.
In 1919 astronomers used a total solar eclipse to test whether the Sun's gravity bends starlight, as Einstein's theory said it would. Einstein's theory predicted starlight near the Sun would shift by 1.75 arcseconds, and Eddington's teams measured 1.98 and 1.61, while Newton's theory asserted a smaller bend of about 0.87 arcseconds, so the numbers could tell the two ideas apart. Karl Popper said the impressive thing about the test was the risk: the theory forbade an outcome that could have happened. Falsifiability means a claim could be shown wrong by some possible observation.
A theory that no conceivable event could refute is not scientific, in Popper's view. Confirmations count only when they come from risky predictions, and adding untestable add-on ideas to protect a theory from refutation makes it no longer scientific. Confirmation bias covers how people search for, interpret and remember information so that it fits what they already believe.
Where the author cites sources, check them. If a report mentions studies without naming them, that is something to check on. Look for the author's stated purpose and viewpoint, not only their facts. Two things can rise and fall together without one causing the other: ice cream sales and drownings both rise when the weather is hot, and the weather is called a confounding variable here, a third thing that causes both. Before you believe something, ask who says it and how they know.
Words to know
- replication: repeating a study
- peer: fellow scientist
- falsifiability: can be disproved
- confounding: mixing up
- cue: hidden signal
