Lesson 1.2 · Unit 1 · NGSS practice 3
Variables, controls, and the shape of a fair test
An experiment is an argument with a hole in it, and the hole is every difference between your groups that you did not intend. If the fertilized seedlings also sat closer to the window, the experiment cannot separate fertilizer from light, and no amount of careful measuring afterwards repairs it. Controlling variables is not tidiness. It is the entire logic of the method.
- The independent variable is the one thing you deliberately change. There is exactly one per experiment.
- The dependent variable is what you measure to see whether the change mattered. It depends on the independent variable, which is where the name comes from.
- Controlled variables are everything else you deliberately hold the same: temperature, light, water, soil, the age of the seeds. These are not "the control".
- The control group is the group that does not receive the treatment, and it exists to tell you what would have happened anyway. Without it you cannot tell a real effect from normal growth.
- A confounding variable is one that changes along with your independent variable. It makes the result uninterpretable rather than merely imprecise, because two explanations fit the data equally well.
- Blind measurement protects the dependent variable. If the person measuring knows which group is which, expectation quietly shifts borderline readings.
Where students lose marks: confusing "the control" with "the controlled variables". The control is a group; controlled variables are conditions. Naming the constants when asked for the control group loses the point outright.
The investigation. Does temperature affect the germination rate of radish seeds? The figures below are constructed so the arithmetic is checkable.
Step one: name the independent variable. Temperature. Two levels: 20 degrees Celsius and 30 degrees Celsius. One variable, deliberately changed.
Step two: name the dependent variable. The percentage of seeds that germinate within five days. Percentage rather than count, so that dishes with different seed numbers could still be compared.
Step three: list the controlled variables. Same seed batch and age, 50 seeds per dish, same volume of water (10 mL), same dish and paper, same darkness, same five day window. Each is a thing that could change germination on its own.
Step four: identify the control group. Here the 20 degree dish is the comparison condition, representing ordinary room temperature. It answers "what would have happened without the warming".
Step five: read the results. At 20 degrees, 44 of 50 seeds germinate, which is 88 percent. At 30 degrees, 47 of 50 germinate, which is 94 percent. The difference is 6 percentage points.
Step six: introduce a confound and watch the argument collapse. Suppose the 30 degree dish had been given 15 mL of water instead of 10 mL. The warmer dish is now also the wetter dish. Higher germination could come from either, and the data cannot separate them. The experiment has not become less accurate; it has become unable to answer its own question.
Step seven: state the conclusion at the right strength. With the confound removed, the fair statement is that germination was 6 percentage points higher at 30 degrees in this trial. Whether that difference is larger than ordinary variation between dishes is a separate question, and it is the subject of lesson 1.3.
Questions 1 to 5 are recall. Questions 6 to 10 ask for reasoning.
- How many independent variables should an experiment have?
Show the full solution
Exactly one
- What is the difference between the control group and the controlled variables?
Show the full solution
The control group is a group that goes untreated; controlled variables are conditions held the same across all groups
- Define a confounding variable.
Show the full solution
A variable that changes along with the independent variable, so two explanations fit the results equally well
- In the worked example, name the dependent variable and its units.
Show the full solution
Percentage of seeds germinating within five days, in percent
- Why is measurement sometimes done without knowing which group is which?
Show the full solution
Because expectation shifts borderline readings when the measurer knows what result is hoped for
- A student waters one tray daily and another every three days, and also places
the daily tray on a warmer shelf. Explain precisely what is wrong.
Show the full solution
Two things differ between the trays: watering frequency and temperature. If the daily tray grows better, the cause could be the extra water, the extra warmth, or both acting together, and the data cannot distinguish them. This is a confound rather than sloppiness, because collecting more measurements from the same setup will not help. The fix is to put both trays on the same shelf so that watering frequency is the only deliberate difference. Two variables changed together, so the result has two equally good explanations
- Why does an experiment need a control group at all, if the treated group is
measured carefully?
Show the full solution
Because a measurement on its own has nothing to be compared against. Radish seedlings grow whether or not you add fertilizer, so observing that the treated plants reached 18 cm tells you nothing until you know what untreated plants reached in the same two weeks. The control group supplies that baseline, and it must experience everything the treated group experiences except the treatment itself, so that the difference between them can be attributed to one cause. It supplies the baseline showing what would have happened anyway
- A drug trial gives the treatment group a tablet and the control group nothing
at all. Explain the flaw and how to fix it.
Show the full solution
The groups differ in two ways: the drug, and the experience of being given and taking a tablet. The second can change reported outcomes on its own through expectation, so improvement cannot be assigned to the drug. The fix is to give the control group an identical tablet containing no active ingredient, so that the only difference between the groups is the substance being tested, and ideally to keep both patients and assessors unaware of who received which. Taking a tablet is itself a difference; give the control an inactive one
- In the germination example, why report a percentage rather than the number of
seeds?
Show the full solution
A count depends on how many seeds were placed in the dish, so 44 germinating means something quite different out of 50 than out of 200. Converting to a percentage removes that dependence and makes dishes with different seed numbers directly comparable, both within this experiment and against anyone else's results. It also makes the comparison easier to state: 88 percent against 94 percent is immediately readable in a way that 44 against 47 is not, once the denominators differ. Percentages stay comparable when the number of seeds differs
- Design a fair test of whether soil pH affects bean seedling height. Name each
variable type.
Show the full solution
Independent variable: soil pH, set at several levels such as 5, 6, 7 and 8 using buffered solutions. Dependent variable: mean stem height in centimeters after twenty one days. Controlled variables: seed variety and batch, number of seeds per pot, soil type and volume, water volume and frequency, light and temperature, and pot position rotated to even out any bench effects. The pH 7 pots serve as the comparison condition since that is ordinary soil, and every pH level needs several pots so that variation between pots can be judged. pH is independent, height is dependent, everything else is held constant across several replicate pots