Variables, hypotheses and controls

A good hypothesis is a testable, falsifiable prediction with a biological justification — not just "the rate will increase". For example: "the rate of oxygen production will increase as light intensity increases, because light provides the energy for photolysis in the light-dependent reaction, until another factor becomes limiting."

• Independent variable — the one you deliberately change. State the range and the interval, and justify both.
• Dependent variable — the one you measure. State how you will measure it and with what resolution.
• Control variables — everything else that could affect the dependent variable: temperature (water bath), pH (buffer), volume and concentration of solutions, the same species, age and mass of tissue, and so on. If you cannot control one, you must at least monitor it.
• Controls: a negative control (e.g. boiled, denatured enzyme) shows the effect you see is really caused by the independent variable. A positive control confirms the method can detect the effect at all.

Reliability vs validity: a repeatable experiment gives similar results when you repeat it; a reproducible one gives similar results when someone else, with different apparatus, repeats it. Valid means it actually answers the question — a perfectly repeatable experiment with an uncontrolled variable is still invalid.

Precision, accuracy, resolution and uncertainty

These four words are constantly confused, and the marks are easy to lose.

• Accuracy — how close a measurement is to the true value.
• Precision — how close repeat measurements are to each other. A balance with a zero error can be extremely precise and consistently inaccurate.
• Resolution — the smallest change the instrument can detect (a ruler with mm markings has a resolution of 1 mm).
• Uncertainty — for a single reading from an analogue scale, it is conventionally taken as half the smallest division: ±0.5 mm on a millimetre ruler. If you measure a length using both ends of the ruler, the uncertainty doubles.

The examiner’s favourite improvement: "measure a larger volume / longer distance / greater mass", because it reduces the percentage uncertainty without changing the equipment at all. Timing 20 swings and dividing by 20 works on the same principle.

Your turn — percentage uncertainty

Your turn — a more precise balance

Serial dilutions and calibration curves

To produce a range of known concentrations you use a serial dilution. In a tenfold serial dilution, take 1 cm³ of stock and add 9 cm³ of distilled water; mix thoroughly; then take 1 cm³ of that and repeat. Each step divides the concentration by 10, giving a logarithmic range that covers several orders of magnitude with very few tubes.

• Use a fresh pipette at every step, or you will carry stock across and every subsequent concentration will be too high — a systematic error.
• A calibration curve is built by measuring a known property (e.g. colorimeter absorbance, or mass change of potato) against known concentrations. You then read the unknown off the line. Interpolation (reading within the range) is valid; extrapolation beyond the measured range is not.
• In the osmosis practical, the concentration at which there is no change in mass — where the line crosses the x-axis — is the concentration at which the solution has the same water potential as the cell contents. That is the whole point of plotting it.

Your turn — serial dilution

Means, standard deviation and significant figures

Repeat readings, discard clear anomalies (and say why), then calculate the mean. The mean alone hides everything interesting; the standard deviation tells you the spread about it.

• Plot the mean with error bars of ± 1 SD. If the error bars of two sets overlap substantially, the difference between the means may well be due to chance. If they do not overlap, a real difference is likely — but only a statistical test can tell you whether it is significant.
• Significant figures: a calculated value cannot be more precise than the least precise measurement it came from. If you measured 2.5 g and 1.4 g, do not report 1.7857.

Random vs systematic: random errors scatter results either side of the true value — repeat and take a mean. Systematic errors (a zero error, a mis-calibrated colorimeter, a thermometer reading 2 °C high) shift every result the same way, and no amount of repeating will fix them. This is why an accurate-looking mean can still be wrong.

Your turn — the mean

Your turn — standard deviation

Choosing the right test

OCR expects you to select a test and justify the choice. Every test begins with the null hypothesis: that there is no significant difference (or no correlation), and that any difference observed is due to chance.

• Student’s t-test — compares the means of two sets of continuous, roughly normally distributed data (e.g. leaf width in sun vs shade).
• Chi-squared (χ² = Σ (O − E)² ÷ E) — compares observed frequencies (counts, never percentages) with expected ones, e.g. the offspring ratio from a genetic cross, or the distribution of an organism across habitats.
• Spearman’s rank correlation — tests whether two variables are correlated when the data can be ranked (e.g. soil moisture and species abundance along a transect).

Compare your calculated value with the critical value at p = 0.05 for the appropriate degrees of freedom. The convention is that if the probability of the result arising by chance is less than 5%, the difference is significant and the null hypothesis is rejected.

Correlation is not causation. A significant Spearman correlation between two variables tells you they vary together. It does not establish that one causes the other — a third, unmeasured variable may drive both.

A systematic problem

Pick the test

What is the control for?

Which kind of variable?

A student investigates the effect of temperature on the rate of an enzyme-catalysed reaction. Tap a card, then tap the type of variable it is.

Define the term

Tap an item on the left, then its partner on the right.

The big ideas to take away

That is the whole of OCR Module 1 — Development of practical skills in biology. Press Finish to see your score.