Monday, December 22, 2008

Science 101 For Journalists

It appears to me that some newspapers do an excellent job of presenting self-produced stories on the hard sciences (eg Chemistry, Physics, Biology.). Although such offerings often require more than one careful reading (And often require me to "strain my brain"), the facts are often clear and the presentation appropriate for the serious reader. An excellent example of such is the December 14th offering by the Milwaukee Journal Sentinel on stem cell research!

The real problems in your science reporting lies in three areas: As to articles regarding the "soft sciences" (eg Psychology, Sociology, Political Science); Articles reprinted from wire services or other newspapers and their less careful editors or science reporters; And, printed or reprinted without included evaluation as to the existence/lack of good scientific-research standards for each such report.

Although all such "problem" articles should be put to the tests of sound science, it might be well for your paper (In a Sunday editorial section) to educate/reeducate your readers (And some reporters and editors) as to those principles of good research VS. what is, too often, inflicted on the public as such. Newspapers not underestimate their serious (The people who still read real newspapers) readers as they do carefully read stories as to real or meaningless research.

Below I have included a (Probably partial) list of subjects for any such presentation or series.

1. What is "research design"?
2. Experimental VS. control group studies; Selection and exclusion of
3. Multi-factor matching research.
4. Levels-Of-Confidence (With/without explaining T-scores).
5. The basics of "averages", "means" and other statistical terms.
6. Defining the terms and limits of any research project; And, to what extent
can any research project lead to reliable-and-valid conclusions.
7.The scientific meaning of "reliable-and-valid".
8. The scientific meaning of "bias" as to research.
9. Sources of "bias" including, but not limited to: The motivations of those funding
research; Biases in selecting subjects or data bases; "Politically correct" avoidance
of "offending" those who are easily offended; Use of non-standardized or
non-structured interviewing techniques where personal interviews are used;
Exclusion, by the experimenter or in the pool of subjects, of persons whose
involvement would be necessary to draw reliable-and-valid general conclusions;

1 comment:

Tamara said...

To go with #5... correlation does NOT equal causation.

So when scientists find a link between breastfeeding and IQ... or obesity and cancer... or whatever other scare story of the week there is, hopefully one day people will realize that one simple thing is NOT determinant of the future of your child or your health or whatever else they're trying to scare you about.