Kinds of Research

What kinds of research have been done in the field?

The studies on RF radiation today can be divided into two classes. Epidemiological findings and controlled research. While each type of study is distinct and important in its own sense, the reader must have a good understanding of both in order to fully assimilate all of the information available on the health effects of PCS and cellular technology.

Controlled

This type of study is the most typically conjured image when the word research is mentioned. It is based upon the scientific method, which generally consists of five steps:

1) Problem Statement: A question that the researcher is interested in solving.

2) Analyze Problem: Gather relevant data and make observations about your problem.

3) Formulate Hypothesis: Develop a hypothesis, or a specific solution to the problem in step one, based on accumulated research. The hypothesis is the central statement of the research.

4) Test Hypothesis: Devise an experiment to test the validity of your hypothesis.

5) Accept/Reject Hypothesis: Based upon the findings of your experiment, determine whether the hypothesis is true or false.

Although the details of each step may differ from scientist to scientist, this general format is always preserved.
In order to better understand this type of experiment, let us consider an illustration. Suppose that we are fascinated by the orange juice commercial where the manufacturer inserts a straw straight into an orange. We ask, "Does the size of the straw matters in the ease of orange penetration?" This then, is our problem statement. We then proceed to read some books on the properties of oranges and straws, in effect analyzing our problem. Now, suppose that after reading the text, we decide to say that the size of the straw does not make a difference in the ease of which it may be inserted into an orange. This statement, which is our best "guess" at the answer to our question becomes our hypothesis. In order to prove whether or not our claim is correct, we proceed to the next step and perform an experiment to test our hypothesis. We select oranges of similar texture, size, color, shape, etc. and then take straws of various sizes and insert them into the nearly identical oranges.
Suppose that we find in our experiment that the bigger the straw, the less easily it pierces the orange's skin. Does this mean then that our hypothesis is incorrect? No! One test case is not enough, since we might have selected oranges that are not similar internally. In this case, our finding might be a result of the orange in itself and not the straw; a coincident! So we repeat our study many more times. Now let us suppose we find that in almost every test, the smaller straw pierces the orange more easily than the bigger one. We may now safely assume that our hypothesis is incorrect and reject it. As you can see, the great advantage offered by controlled experiments is the degree to which a scientist can control the experiment. By setting up an experiment in a particular fashion, the scientist can insure that a specific variable is responsible for the effect in question. In our example above, we were able to create a situation in which only one effective variable existed -the size of the straw. By picking nearly identical oranges, we have in effect controlled for all other variables. For example, we know that the color of the orange can not be the reason why some straws go in to oranges more easily; after all, we have selected the oranges such that their color is the same.
It becomes readily apparent then, that the controlled experiment is most useful for divining a causal relationship between two things. The ability to eliminate all but a few select variables insures that any differences in the experiment are likely to be a result of those variables.

Epidemiological

In the epidemiological study, the approach to science is somewhat different. We still begin with a problem statement, but the question now is usually about association. From there on the focus differs from that of a controlled experiment. Data is still gathered and analyzed as in step two of the controlled research, but no experiment is actually carried out and no hypothesis is formulated. Instead, the epidemiologist examines the data looking for whether the association declared in the problem statement actually exists.
For example, let's consider the example we used above concerning straws and oranges. The epidemiologists problem statement would be: "Does an association exist between straw size and ease of orange penetration?" Notice that the question is rather general and is only concerned with an association.
From there, the epidemiologist collects all available data on straws poking oranges. Perhaps there is none. The epidemiologist would probably then conduct a survey, asking people who happened to have poked an orange with a straw what the size of the straw was, what the characteristics of the orange they used was, and whether or not the straw penetrated the orange. From this information they look for some for of association. Suppose it just so happened that more people with green oranges were able to pierce them with a straw then with any color. The epidemiologist would then try to determine whether or not the statistic was significant. While there is no way in an epidemiological study to be certain that the association is not a coincidence, there are ways of calculating the probability of the association. Usually, if the probability of the finding happening by chance is slight, an association can be established between the two.
It is important to note at this time that even if an association is established, an epidemiological study (unlike controlled research) in no way explains how or why that relationship exists. Using our example, there is no way to know why it is that the green orange is more easily pierced by the straw than others simply by looking at people's surveys. To understand how or why, we would need to perform a controlled experiment.
At this point, it becomes clear what the strengths and weaknesses of epidemiological studies are. While these studies prove useful in finding associations, they can not explain the association. It could very well be that the findings were purely coincidental. In order to be absolutely sure, the study would have to be followed with a controlled experiment. Hence, an epidemiological study's greatest strength is in providing a direction for controlled research to proceed.