Lesson 2.1 · 8 min
Asking a testable question and testing it fairly
Objective: Turn a vague curiosity into a testable question, design a fair comparison, and explain why one result is not a pattern.
The scientific method is a habit of thinking in a specific order. First you ask a clear question. Then you make a guess about the answer, called a hypothesis, that you can actually test. Then you design a way to test it. Then you run the test, record what happens, and compare the result to your guess. The key word is testable. A question like does this compound do something is too vague to test. A question like does adding X change measurement Y under condition Z is something you can design a test around.
A fair test needs a comparison, called a control. Without a control, almost any result can be explained by almost any story, because you have nothing to rule anything out with. A control is simply a version of your test where you change nothing, so you have a baseline to measure against. The goal is always the same: isolate one variable at a time, so that if something changes, you know why.
A single observation, even a striking one, is a data point, not a conclusion. Things vary. Measurement equipment has error built into it. A result that shows up once might be real, or it might be noise. Careful researchers repeat their tests before trusting a result. Repetition tells you whether a pattern is consistent or whether your first result was a fluke.
Bias is the normal human tendency to see what we expect. Once you hope a result is real, you notice supporting observations and discount the ones that disagree. The defenses are mechanical: decide in advance what you will measure and when you will stop, record every session including the ones that go badly, and ask what result would change your mind before you begin. If nothing could change your mind, you are not running a test. You are collecting reasons.
- Hypothesis:
- A specific, testable guess about what will happen, made before you run a test.
- Control:
- A comparison sample or condition, kept the same as the test condition except for the one thing being studied.
- Bias:
- Any influence that pushes observations or interpretation systematically toward an expected answer.