QC Sec 4 Science & Tech common mistakes
QC Sec 4 Science & Tech · Understand the traps and how to avoid them.
Reading a value off a line graph by looking only at the plotted point's height and ignoring the axis scale, so a reading of 4.2 is written as 42 or 0.42 when the axis is labelled in thousands or with a multiplier such as ×10³.
Candidates scan for the shape of the curve and grab the nearest number on the axis without checking the axis label, unit, or multiplier. Under time pressure the eye matches the gridline, not the scale, so a mislabelled axis goes unnoticed.
What to do instead
Before reading any value, say the axis title, unit and step size out loud or write them down. Then interpolate explicitly: state the two gridlines the point sits between and the fraction of the gap. For any graph with a multiplier, convert the raw axis number to the stated unit in a written step so the multiplication is visible and checkable.
Treating a strong correlation shown in a scatter graph as proof that one variable causes the other, and writing conclusions such as 'increasing temperature caused the enzyme to denature' from observational data with no manipulated variable.
Graphs are visually persuasive and the language of trend ('as X increases, Y increases') slides into causal language. Candidates are trained to describe trends and rarely stop to ask whether the independent variable was actually controlled by the experimenter.
What to do instead
For every trend statement, write a matched pair: a descriptive claim ('Y is associated with X') and an evidence claim ('this would support causation only if X was manipulated and confounders were controlled'). Check the method description for a manipulated variable before allowing causal wording into the answer.
Confusing the independent and dependent variables, or listing a controlled variable that was not actually held constant, when describing an experimental design. A typical error is naming 'temperature' as the independent variable when the method varies substrate concentration and holds temperature fixed.
Candidates memorise the definitions but do not apply them to the specific scenario. The word 'variable' triggers a memorised phrase rather than a reading of what the experimenter changed versus what was measured.
What to do instead
Use a fixed sentence frame tied to the scenario: 'The experimenter changed ___ and measured ___; everything else, including ___ and ___, was kept the same.' Fill it only from the method text. If the method does not state a variable was controlled, say so rather than inventing one.
Designing or evaluating a control group that is not a fair comparison, for example giving the treatment group a different volume of solution, a different number of subjects, or running it at a different time of day, then concluding the treatment worked.
Candidates focus on the treatment and treat the control as an afterthought. They also tend to accept any 'control group' as valid without checking that it differs from the experimental group in exactly one respect.
What to do instead
List every difference between the two groups in a two-column table and count them. A valid design has exactly one difference, the independent variable. If more than one difference exists, name the confound and state how to remove it (equal volumes, matched subjects, randomised order, same time of day).
Reporting a measurement with more significant figures than the instrument supports, such as recording 12.345 g from a balance reading to 0.1 g, or quoting a mean to four decimals when the raw data have one.
Calculators and spreadsheets return long decimals, and candidates copy them without applying the precision of the measuring instrument. Significant figures are treated as a formatting rule rather than a statement about uncertainty.
What to do instead
Before calculating, write the resolution of each instrument (for example ±0.1 g) and the number of significant figures in each raw value. Round the final answer to the least precise input and attach the uncertainty, e.g. 12.3 ± 0.1 g. State the rule used so the marker can see the reasoning.
Adding or averaging percentage uncertainties incorrectly, for example adding the percentage uncertainties of two measurements that are multiplied, or averaging repeated readings and dividing the uncertainty by the number of readings when the readings were not independent.
Candidates learn one rule ('add uncertainties') and apply it to every operation. They do not distinguish absolute from percentage uncertainty or between random and systematic error.
What to do instead
Use a decision tree: for addition and subtraction, add absolute uncertainties; for multiplication and division, add percentage uncertainties; for a power, multiply the percentage uncertainty by the exponent. For repeats, quote the standard deviation or half the range and state whether the spread reflects random error.
Proposing a technology solution that ignores a stated constraint, such as recommending solar panels for a site with no battery storage and low winter irradiance, or scaling up a system without checking that the power, bandwidth or cost budget allows it.
Candidates pattern-match a problem to a familiar technology and defend it with general benefits. They skip the constraint list in the prompt and do not quantify whether the proposal meets it.
What to do instead
Extract every constraint from the prompt into a checklist (power, cost, environment, time, capacity) and score the proposal against each one. Any proposal that fails a hard constraint must be modified or rejected with a stated reason, and at least one quantitative check (units, budget arithmetic, capacity calculation) should appear.
Describing a diagram's structure without explaining its function, for example labelling the parts of a feedback loop or a circuit but not stating how the output feeds back to adjust the input, or how the component changes the behaviour of the system.
Diagram questions invite label-copying, and candidates equate naming parts with demonstrating understanding. Function words ('so that', 'which causes', 'thereby') are missing from their answers.
What to do instead
For every label, add a 'which does what' clause. For system diagrams, trace the path from input to output and back, writing one sentence per arrow using causal connectives. If the diagram shows feedback, state whether it is negative (stabilising) or positive (amplifying) and give the consequence.
Writing conclusions that overreach the data, such as claiming a result applies to all species, all temperatures, or all concentrations when the experiment tested one organism at three temperatures, or ignoring a data point that contradicts the stated trend.
Candidates want a decisive answer and generalise from a small sample. They also tend to smooth over anomalies because acknowledging them feels like undermining their conclusion.
What to do instead
Bound every conclusion with the tested range ('within the range 10–40 °C, for this species'). Explicitly mention any outlier and either explain it or flag it as unexplained. Use hedged verbs (suggests, is consistent with) unless the evidence is direct and replicated.
Confusing energy, power and efficiency in applied problems, for example reporting a 2 kW heater running for 3 hours as using 6 kWh of power, or calculating efficiency as output divided by input without converting to the same units first.
The words are used loosely in everyday speech, and the formulas look similar. Candidates substitute numbers into a remembered equation without tracking the units through the calculation.
What to do instead
Write the unit of every quantity next to its symbol before substituting (P in W, E in J or kWh, t in s or h). Convert all quantities to consistent units first, then calculate, then check the unit of the answer against what was asked (power vs energy). For efficiency, show input and output in the same unit and express the result as a percentage.