Monday, February 27, 2012

On Science 4: The Scientific Method

If you are like most people once upon a time you heard/learned that the scientific method includes a series of predefined, more or less rigid steps that if followed, like a recipe as shown below, will result in "science."  This view of how science is done is just way too simple.

All the images in this posting are from a fantastic web site on the process of science put together by the good folks at UC Berkeley.  You can access it at: http://undsci.berkeley.edu/article/howscienceworks_01.

To get back to the process of science...it is true that the general steps outlined above do matter and are part of what happens, but there are MANY ways that scientists work.  A more accurate representation of how science is really done is shown below:


Rather than a static set of steps, the scientific approach to asking questions, obtaining empirical evidence, analyzing those data, and developing conclusions is a dynamic process.  Most scientists start the process in the top sphere "Exploration and Discovery" where they either ask questions about new observations or see things differently through new technology, or identify a meaningful problem through curiosity and inspiration, but ideas and motivation for carrying out research can happen at any stage of the process and in any of the spheres in the figure above.

Regardless of where a scientist starts, the process of inquiry almost always includes the generation of a scientific argument.   This step is shown in the upper half of the center sphere, "Testing Ideas."  A scientific argument includes three parts - hypotheses (Null and Research Hypotheses - see the earlier posting on the Power and Strength of Science), expected observations, and actual observations.  The Research Hypothesis is the researcher's best explanation of what they thing the eventual answer will be, and it is used to generate a set of expected or predicted results.  The researcher then uses one of any number of approaches to collect empirical observations. This can be done through experimentation, observation, etc.  Once the empirical observations are in hand the researcher moves to the lower part of the central sphere and carries out an analysis of the results...that is, they almost always usually use statistical methods or other mathematical tools to see how well the observed data and the predicted data match up.  This can lead to a variety of outcomes.

The observed observations may not support the expected observations, so the expected set and the hypothesis that led to them has to be upgraded, improved, changed, or even tossed out in favor of a better hypothesis and set of expected explanations.  This lead the researcher back to the top sphere or over to the sphere labeled "Community Analysis and Feedback."  No matter what path a project follows through this overall process, the research must pass through the "Community Analysis" sphere at least once.  This is where the observations and explanations are subjected to critical peer-review.  Peer-review is a rather ruthless process where other scientists who are experts in the field look over the hypotheses, expected data, observed data, methods of analysis used to examine the data, the conclusions reached by the researcher, etc.  The basic outline of what happens during peer review is shown in the figure below:


Major benefits of subjecting the outcome of a study to peer review include: minimizing or eliminating researcher bias in the results of the paper, rejecting papers that had poor research designs, inappropriate or internally inconsistent conclusions, helping the researchers address weaknesses in writing, analysis, etc.

If the results and conclusions of a study make it through peer-review (usually with a significant amount of additional review and revision work) the study can be accepted for publication in a professional journal.  Once this happens, the conclusions can inform new research in the top sphere or inform policy or be recognized as new knowledge, etc., in the sphere labeled "Benefits and Outcomes."  This is also when the larger scientific community can access the work, assess its validity and value and, if desired, replicate the study or carry out studies similar to it that will help determine whether the conclusions of the paper hold up under additional scrutiny.

Remember that the Null Hypothesis is always the preliminary explanation that is tested.  By so doing the scientific community eliminates false explanations, and thereby moves progressively closer to the true explanation about the question of interest.

Keep in mind that the outcome of any sphere can lead to any other sphere in the chart, and this dynamic, cyclic process is a much better representation of how science is actually done than the over-simplified "recipe" shown at the top of this posting.

Once a preliminary explanation has been tested using as many available approaches and sets of data as possible, and it still holds up, the hypothesis can become a theory - the most powerful kind of explanatory statement in science.

I heartily refer you to the website indicated above for the full treatment of how science works.  It's well worth the time, and I have all of my general education students review this site as part of the course.

No comments:

Post a Comment