BIOL 1400 -- Lecture Outline 2

"Whatever we cannot easily understand we call God; this saves much wear and tear on the brain tissues." --Edward Abbey

I. Experimental design

  1. Suppose you want to test a hypothesis. (Let's say that our hypothesis is: taking certain pills keeps you from catching the flu.)
    1. Suppose that people with the flu take your pills and then get better. This still doesn't mean that the pills are doing any good, since most of the time people recover from flu even if they take no medicine at all!
      1. The post hoc fallacy is the assumption that if X happens, and then Y happens, that X must have caused Y. (Or in Latin, post hoc ergo propter hoc, which means "after it, therefore because of it.")
      2. This is a logical error -- two things may happen together, but that's no proof at all that one of them (such as taking the pills) caused the other (such as recovering from the flu).
      3. Good experiments allow you to test whether X caused Y, rather than just assuming it blindly
    2. Many different factors, or variables could be influencing and affecting what you see.
    3. For example, whether someone gets the flu or not could be affected by:
      • the weather
      • other medical conditions (such as allergies)
      • the people you come in contact with
      • your overall state of health (nutrition, etc.)
      • . . . and so on. . .
    4. A well-designed experiment will isolate only one of these variables.
    5. That way, it can test the hypothesis that that specific variable (the experimental variable) has an effect on something observed.
  2. A good experiment uses controls.
    1. A control is a part of an experiment in which the variable being tested (the experimental variable) is left alone.
    2. Thus you can compare the control with the experimental group, in which you fiddle with the variable that you want to isolate and test (and only that variable).
      1. In a clinical trial, which is the best-known kind of experiment to the general public, we would divide a group of test subjects into two groups. In this example, all would have the flu, or be at risk for catching one. . .
      2. One group would get, say, Doctor Waggoner's Miracle Cure; the other group would get nothing.
      3. If the people who got the Miracle Cure got over their illness faster than those who did not, we would have confirmed the hypothesis.
      4. Note that we would have to match the experimental and control groups as closely as possible (same ages, genders, races, etc.)
        1. Watch for the placebo effect! In drug trials, people have been found to feel better, and get better, if they think they're getting an effective medicine (even if what they're getting is useless).
        2. To avoid this, in a clinical trial, both experimental and control groups receive a treatment, but the control group gets something completely ineffective (say, a pill made of sugar, or an injection of mild saltwater solution). This ineffective treatment is a placebo.
        3. Drug tests are usually done double-blind: Neither the patient, nor the doctor who gives the treatment, knows who is getting the placebo and who is getting the real drug.

    Go to Previous Notes | Return to Lecture Schedule | Return to Syllabus | Contact the Prof | Go to Next Notes