Practical and data skills, at a glance

Change a defined independent variable, measure a useful response and explain how other influences will be controlled.

• State how each variable is changed, measured or held constant.
• Independent repeats estimate variation; they do not remove systematic bias.
• A control treatment tests an alternative explanation.

Keep in mind: The independent variable must differ. Relevant other conditions should be held consistent, while a control treatment is chosen to test a particular alternative explanation.

Concentration x volume tracks the amount of solute: C1V1 = C2V2 for a simple dilution with consistent units.

• V2 is the final total volume, not the amount of solvent added.
• Successive dilution factors multiply.
• Precision of repeated readings and accuracy relative to the true value are different.

Keep in mind: Each stage acts on the concentration produced by the previous stage. The factors multiply: 10 x 10 = 100.

Choose a display that matches the variables, and keep measured values distinct from calculations and interpretation.

• Put quantities and units in table headings and axis labels.
• Keep raw repeats available alongside processed results.
• An anomaly needs investigation and a stated reason, not silent deletion.

Keep in mind: The display and fitted relationship should match the data and question. A best-fit line or smooth curve may be appropriate; categorical bars do not imply intermediate values.

A rate is change per unit time. The gradient at the start measures initial rate; a whole-run average measures something different.

• Gradient is change in y divided by change in x.
• Use tangent points for an instantaneous rate, not two arbitrary points on the curve.
• 1/time compares a common endpoint; it is a relative rate unless an amount is supplied.

Keep in mind: A horizontal section means accumulated product is approximately constant, so net product formation is approximately zero. A falling curve would indicate a decrease.

The mean describes the centre. Standard deviation describes the spread of observations around that mean, in the same units as the measurements.

• Mean and standard deviation answer different questions.
• Use n - 1 in the supplied sample standard-deviation formula.
• Standard deviation has the original measurement unit.

Keep in mind: Values can cluster tightly around a biased value. Standard deviation describes spread, while accuracy concerns closeness to the true value.

Chi-squared compares category counts with a stated expectation: add (observed - expected) squared / expected for every category.

• Use independent counts, a stated expected ratio and suitable expected category sizes.
• For these fixed-ratio one-row tests, degrees of freedom = categories - 1.
• Failing to reject does not prove the null hypothesis.

Keep in mind: Different mechanisms can produce the same ratio, and a test may have limited ability to detect a difference. Non-significance means insufficient evidence against the specified null model.

An unpaired t-test compares a difference between means with the variation and sizes of two independent samples.

• Unpaired means two independent samples, not repeated measurements of the same individuals.
• Use both sample standard deviations and sample sizes.
• Statistical significance is not proof of causation or biological importance.

Keep in mind: The decision depends on the difference relative to variation and sample size, assessed against the relevant critical value. Unequal sample means alone are not enough.

A useful evaluation names the limitation, explains its effect and proposes an improvement that addresses that effect.

• Limitation -> effect on the result -> targeted improvement.
• Repeats reduce uncertainty from random variation, not every form of bias.
• A significant association still needs a sound design for causal inference.

Keep in mind: Name the action or measurement problem. For example, a delayed start of timing changes the measured interval; a consistent zero offset creates bias. Different problems require different improvements.