BGCSE Statistics 4040 common mistakes
BGCSE Statistics 4040 · Understand the traps and how to avoid them.
When asked to describe the data-collection method used, the candidate names the instrument (for example 'a questionnaire') without stating whether the data is primary or secondary, who supplied it, or how the respondents were selected.
Textbook definitions of primary and secondary data are memorised as isolated labels, so candidates never practise linking a described study to those labels; the question stem often describes a real scenario, which invites a narrative retelling rather than a classification.
What to do instead
Train a three-line response template: (1) state 'primary' or 'secondary' and justify by who collected it and for what purpose; (2) name the instrument and the population; (3) name the selection method. Practise applying the template to five short scenarios until the classification is automatic.
The candidate draws a histogram with gaps between the bars, or plots a grouped frequency distribution with unequal class widths using bars of equal width.
Bar charts and histograms are taught close together and both use rectangles, so the visual habit of separated bars carries over; unequal class widths are rare in early practice, so frequency rather than frequency density is used for the height.
What to do instead
Before drawing, write 'continuous data = no gaps' and, where class widths differ, compute frequency density for every class and write the values in a table first. Draw the horizontal axis to a real scale with class boundaries marked, then read bar heights from the density column.
The candidate computes the mean, median and mode correctly but then answers 'the average' when asked which measure best represents the data, offering no comparison of the three.
The question is treated as a calculation task rather than a judgement task; candidates lack a rehearsed set of criteria (outliers, skew, open-ended classes, repeated values) to justify a choice.
What to do instead
Memorise four criteria as a checklist: sensitivity to extreme values, effect of open-ended classes, suitability for further calculation, and behaviour with repeated values. For each dataset, write one sentence naming the chosen measure and one sentence naming the rejected measure with the reason.
The candidate gives a probability as a fraction such as 3/5 when the question asks for the probability that at least one of two events occurs, having simply added the two individual probabilities without subtracting the overlap.
The addition rule is recalled as 'add the probabilities', and the subtraction term is forgotten because most classroom examples use mutually exclusive events where the overlap is zero.
What to do instead
Always draw a two-set Venn diagram or write the formula P(A or B) = P(A) + P(B) - P(A and B) before substituting. Check whether the events can occur together; if they cannot, state explicitly that the overlap is zero and why.
The candidate interprets a correlation coefficient of 0.85 as proof that the explanatory variable causes the change in the response variable, or claims that a strong correlation means the points lie exactly on a straight line.
The word 'strong' is read as 'perfect' and the mathematics of correlation is taught without discussion of lurking variables; candidates rarely meet a counter-example in class.
What to do instead
Learn three fixed statements to quote: correlation measures the strength and direction of a linear association only; it does not imply causation because a third variable may drive both; r = 1 would be a perfect straight line. Apply all three to any scattergraph question before writing a conclusion.
When computing a weighted index number, the candidate divides by the number of items instead of by the sum of the weights, or uses base-year quantities when the question specifies current-year quantities.
The arithmetic mean formula is over-learnt and applied to any list of numbers; the distinction between the two weighting conventions is a single phrase in the question that is easily skimmed.
What to do instead
Write the index formula with named symbols (sum of weight times price relative, divided by sum of weights) before substituting any numbers. Underline the words that specify the weights in the question and label each weight column accordingly.
The candidate computes a four-point moving average for a quarterly series and plots the result against the original time points without centring, so the trend values sit on the wrong quarters.
The centring step looks like an optional refinement and its purpose is not explained; with an even period the average falls between two periods, which is counter-intuitive.
What to do instead
Use a fixed layout: write the four values, average them, place the average in the space between the second and third values, then average adjacent pairs to centre. Check that the first trend value aligns with the middle of the first four quarters before plotting.
The candidate describes a sampling frame as 'the people in the sample' or confuses a systematic sample with a simple random sample because both appear random.
The vocabulary is dense and similar-sounding, and definitions are memorised as short phrases without a concrete example attached to each term.
What to do instead
Build a one-page glossary with a concrete school-based example for each term (population, frame, census, simple random, systematic, stratified, quota, cluster). For each method, write the exact selection procedure in two steps so the difference is visible.
When asked to interpret a confidence interval, the candidate writes that there is a 95% probability that the population mean lies in this particular interval, rather than describing the long-run performance of the procedure.
The interval is treated as a probability statement about a fixed unknown parameter; the repeated-sampling idea behind confidence levels is rarely demonstrated concretely.
What to do instead
Learn the standard wording: 'If samples were repeatedly drawn and intervals constructed in the same way, about 95% of them would contain the population mean.' Practise writing this sentence, then add the specific numerical interval, so the interpretation is always attached to a stated procedure.
The candidate concludes that a result is 'proven' or that the null hypothesis is true when the test statistic falls in the acceptance region, instead of saying there is insufficient evidence to reject it.
Everyday language treats 'not guilty' as 'innocent', and the logic of hypothesis testing is taught procedurally as a sequence of steps without the underlying asymmetry of evidence.
What to do instead
Adopt two fixed conclusion sentences and use only those: 'There is sufficient evidence at the 5% level to reject the null hypothesis' or 'There is insufficient evidence to reject the null hypothesis'. Ban the words 'prove', 'true' and 'accept' from written conclusions.