SC2 Scientific Investigation

Lesson

Roughly a fifth to a third of the scored Science questions are Scientific Investigation items. They live mostly in the Research Summaries passages — the ones that open with a paragraph of setup and then hand you Experiment 1, Experiment 2, sometimes Experiment 3, each with its own method and its own table. These questions are not really about the data. They are about the design: what the experimenters changed, what they refused to change, why they ran a second experiment at all, and what their setup can and cannot tell you.

The good news is that the answer is always printed. If a question turns on what was held constant in Experiment 2, the passage says so — usually in a flat little sentence at the end of the method paragraph that most students skim past. Read those sentences. They are the question bank.

Independent, dependent, constant

The independent variable is the column heading on the left of the table — the thing that takes a new value in each row. The dependent variable is the column on the right — the measurement. Everything named in the method paragraph with a value attached ("held at 20 °C", "5.0 g in every trial", "the same 250 mL") is a constant.

• A constant is not a variable that does not matter. It is a variable the experimenters refused to let matter, so that any change in the measurement can be blamed on the one thing they did change.
• The same quantity can be independent in one experiment and constant in the next. That switch is usually the entire point of running the second experiment, and it is a favourite question.
• Watch for constants that are held constant by construction: "four spheres of the same diameter and surface finish, made of different metals". The design there is a sentence, not a number.

What a control is actually for

A control is the trial run with the suspected cause removed. Its job is to give you a baseline: this is what the measurement looks like when the thing under test is not there. Without it, you cannot say the effect came from what you added rather than from the apparatus, the solvent, or time passing.

• The tube with no enzyme. The flask with zero nitrate. The dish with no water. The nail kept in dry air. On the ACT the control is nearly always the row with a 0 in it, or the odd trial described separately in the method paragraph.
• A control is not "the first experiment" and not "the average of the others". If a set of questions asks which trial served as the control, look for the deliberate absence.
• Sometimes there are two controls, each removing a different requirement — one nail with no water, one nail with no oxygen. Together they show that both are needed; either one alone would not.

One thing at a time — and what it costs

Change two things between trials and you cannot tell which one produced the difference. That is why every experiment in a Research Summaries passage varies one factor and freezes the rest. The cost is that the design can only answer questions about the factors it actually varied, and only over the range it covered.

This is where most wrong answers live. A choice that reads like a perfectly sensible scientific claim can still be wrong because the experiment never made that comparison. If Experiment 2 tested four surface covers on a single 10° slope, nothing in it can tell you whether grass helps more on steep slopes than on gentle ones. Nobody varied the slope while the grass was there.

Reading a multi-experiment passage

The second experiment almost never repeats the first. Ask two questions about it and write the answers in the margin:

• What changed? Usually one variable was promoted from constant to independent, and the old independent variable was demoted to a fixed value.
• Why? Because Experiment 1 raised a question it could not answer. Experiment 1 found that temperature matters; Experiment 2 asks whether pH matters too, at one convenient temperature.

A useful check: the conditions of Experiment 2 usually match one specific trial from Experiment 1, and that shared trial usually gives the same measured value in both tables. When a number appears twice across two tables, it is normally the seam where the experiments join — and it is a sign the conditions really were held consistent between them.

Which experiment answers which question

Some items give you a research question and ask which experiment addresses it. Match the question's variable to the experiment that varied that variable. A question about the effect of stirring is answered only by the experiment where stirring changed. If no experiment varied it, the honest answer is that none of them can tell you — and that choice is sometimes on the list.

Predicting a new trial

You will be handed conditions nobody tested and asked what the result would be. Work only from the trend printed in the table:

• Find the two tested trials the new one sits between, and bracket it. If 20 °C gave 13 mL and 30 °C gave 22 mL, a trial at 25 °C lands between 13 and 22 — and no closer than that, because the numbers do not have to march in a straight line.
• If two changes both push the measurement the same way, the result goes past whichever tested value was already furthest in that direction.
• Do not extend a trend by assuming it is a straight line unless the numbers say so. Doubling the string length did not double the period; five times the drop height did not give five times the crater.
• Watch for physical ceilings. Water left to cool cannot end up hotter than it started, and the number germinating cannot exceed the number of seeds sown.

A limitation is not a flaw

These two get offered against each other, and students pick the harsh one because it sounds rigorous.

• A limitation is a boundary on what the results can be applied to. Only one species was tested; only temperatures from 10 °C to 50 °C were covered; every trial ran for 14 days. Nothing was done wrong — the conclusions simply stop at the edge of the data.
• A flaw is something that undermines the comparison itself. Two variables changed at once; there was no control; the trials were not repeated; the measurement of one group was made differently from the other.

Test it like this: if the fix would be "also test X", it is a limitation. If the fix would be "the trials you already ran cannot be compared to each other until you do it again properly", it is a flaw.

Two practical notes

• No calculator on Science. Calculators are permitted on the Mathematics section only. Nothing in a Science item requires more than reading a table, comparing two numbers, or a subtraction you can do in the margin — if you find yourself setting up long arithmetic, you have misread the question.
• Every multiple-choice question has four choices and there is no penalty for guessing, so never leave an investigation item blank. Even after eliminating one design-blind choice you are guessing at one in three.