Evaluating Science Online
This page is for those in the community that are interested to learn how to understand and evaluate scientific information. This is a critical skill when conducting research using the internet or traditional print media, but also in helping you sort out news stories and media claims in everyday life.
Thanks to Becky Campbell and Kirsten Dissinger, 9th grade science teachers, for sharing these classroom-ready summary word documents:
Evaluating Scientific Information:
First, be aware of "pseudoscience" ("false science"). Pseudoscientists make claims that may appear scientific, but do not follow scientific principles. Distinguishing between science and pseudoscience can be difficult. When trying to discern whether something is scientific, check the following:
- Interest: Who is funding the research and who may profit from it? Biased organizations may give themselves neutral-sounding names. An organization will often have an interest in the outcome of a study they are funding.
- Author and Publisher: Who conducted the research? Where was it done? Where was it published? Look at the background of the people involved in the research, if possible. What kind of training have they had? Have they done extensive research in the field? Have they published other papers on the topic? Do others frequently cite them? Was the work conducted at an established facility, which could provide the support necessary to conduct thorough research?
Scientists publish their results in peer-reviewed journals so that others in the same field can critically evaluate their work. View with suspicion any discoveries that are "secret" or rely on "secret formulas". Results that have been originally published in journals such as Science, Nature, the New England Journal of Medicine, and so on will have been examined more closely, and are therefore are more reliable, than those that are directly announced to the media.
- Hypotheses: Are hypotheses testable and capable of being falsified? Hypotheses and theories (even those which cannot be tested directly) should be able to used to make predictions and allow the collection of evidence to test those predictions. Often pseudoscientific claims cannot be proven wrong by any possible evidence. For example, there is no way to disprove the claim that only someone with special powers can sense a certain phenomenon.
There is a large body of knowledge in science that is not influenced by trends in public opinion and is not likely to change. However, scientific ideas should be capable of changing should new evidence arise. In contrast, ideas in pseudoscience either stay the same (if there is an unchanging idea behind them) or change randomly (if criteria for accepting ideas and rejecting others do not exist).
1. Are experiments repeatable? Have they been repeated?
Experimental procedures are reported so that others may repeat them. Valid results can be reproduced by others. Check to see that there has been more than one study, and that the studies support past research. One single study may produce results that other studies cannot repeat. The more independent studies that exist which can support a claim, the more likely it is to be true.
2. Are specific, well-defined predictions made?
Scientists use careful, precise language language and make quantitative predictions if possible. Pseudoscientists use vague and imprecise terms that can be interpreted in many different ways, such as the language used in many horoscopes.
3. Are appropriate controls used?
If a drug is being tested, for example, scientists compare an experimental group (getting the treatment) with a control group (not receiving the treatment).
Controls, which should be identical to the experimental group except for the factor being tested, ensure that results are due to the drug itself and not some other factor. Test subjects should randomly assigned to either group ("randomized"). Blind studies (subjects do not know which group they are in) and double-blind studies (neither subjects nor researchers know which group subjects are in) provide additional safeguards.
4. Was a representative sample used? Was it large enough? Were enough trials done?
Scientists use samples that represent larger groups. If only men were used in a study, claims about how the study applies to women would be suspect.
Pseudoscientific or unproven claims will rely on case histories, anecdotal evidence, or personal testimonials (Jane lost 30 lbs. in two weeks with Slim-X!) While case studies might be a starting point for future research, scientists require many trials combined with statistical analysis in order to evaluate their claims. Furthermore, ethical scientists would not reveal the names of people involved in tests.
Sometimes, a statistical claim may be made without reference to the sample size ("3 out of 4 dentists surveyed"- but how many were surveyed?) The larger the sample size, and the longer the study lasted, the more confident scientists are about their results.
1. Were the results statistically significant?
"Statistical significance" measures how often a particular result would occur due to chance alone, assuming that the experiments were repeated many times. The convention is to say that results are statistically significant if there is a 5% probability or less that the results were due to chance alone.
2. Are logic and statistical analysis used to help distinguish between coincidence (chance), correlation (association), and causation?
Correlation and causation are commonly confused with each other. For example, "people who exercise have a lower risk of heart attack" is a statement of correlation, but "exercise lowers the risk of heart attack" is a statement of causation.
It is very hard to prove causation (that A causes B). In order to do so, one needs to show that A must always be present for B to occur, and that B will always occur when A is present ("A is both necessary and sufficient cause of B"). An example of how this can be done in science is the use of Koch's postulates for determining whether a microorganism causes a particular disease:
- The organism must be associated with every case of the disease.
- A pure culture of the organism must be able to be grown outside the body.
- When introducted into a healthy subject, the pure culture of the organism
must cause the disease to occur.
- The organism must be recovered from the subject and cultured again.
Because of the limits on time, funding, or ethical considerations, often the best that can be done is to evaluate a relationship using logic and laws of probability.
When looking for a cause of an illness, scientists would look for large differences between people who had and did not have exposure to a suspected cause. They would check to see that those differences are present between groups that would otherwise be at similar risk for developing an illness. Scientists would also check that a logical reason for a suspected relationship exists.
3. Are new ideas or results viewed critically and with skepticism?
Scientists should ideally presume a new idea wrong until it is well supported with evidence. Pseudoscientists are not skeptical of their own results, but are skeptical of the results of others.
Types of Arguments and Persuasive Devices
Certain techniques are commonly used to attempt to convince the reader of the validity of an argument. Be aware of some of these techniques when you are evaluating a source.
The following types of arguments are discussed in What Science is and How It Works, by Gordon Derry:
1. Straw Man
An argument directed not at someone's actual position, but at a weaker version (the "straw man") created by the opponent. This weaker version would seem, for example, illogical or irrelevant.
2. Ad Hominem ("to the man")
An argument at an individual, rather than the individual's position. The person themselves is attacked, rather than the evidence or the logic of their argument.
3. False Dilemma
Two choices are proposed, and one of these is more easily attacked. This leaves the other choice as the only obvious possibility. However, in reality, there may be many other alternatives or complexities which are not addressed.
4. Begging the Question
This type of argument (also called "circular reasoning") assumes the truth of its conclusions as part of the reasoning leading up to the conclusion.
5. Slippery Slope
An argument in which the position argued against is depicted to result in something terrible. The terrible result is then argued against, rather than the position itself.
The following types of persuasive devices are described in Forests: Identifying Propaganda Techniques, by Anderson and Buggy.
"Everyone else is doing it." This technique takes advantage of the desire of many people to feel as though they belong to a group. The argument is that if most people believe a certain way, then the reader should also feel that way.
7. Slanted Words or Phrases
In this technique, emotionally charged or biased word are used to convince the reader of a certain position (contrast "mature citizen" with "old fogy").
8. Scare Tactics
This technique tries to scare the reader into siding with a particular position. The argument is evaluated on the basis of emotion (fear) rather than logic and reason.
Aaseng, Nathan. Science vs. Pseudoscience. New York: Franklin Watts, 1994.
American Cancer Society: ACS Newsstand, Interpreting the Science in Scientific Studies (1997), http://www.cancer.org/media/1mar4.thml (accessed 07/05/1997).
Anderson, Robert and Buggey, JoAnne. Forests: Identifying Propaganda Techniques. San Diego, CA: Greenhaven Press, Inc., 1992.
Arthritis: Unproven Remedies, Arthritis Foundation, Atlanta, Georgia, 1987.
Derry, Gregory. What Science is and How it Words. Princeton, NJ: Princeton University Press, 1999.
Weiss, Noel S. "Distinguishing Cause from Coincidence", Alaska Airlines/Horizon Air Magazines July 1993.
Special thanks to: Steve Collins, Nancy Hutchison, Cynthia McClellan, Karen Peterson, Diane Rosman, and Dave Vannet.