SECU0006 Chap.3 Describing and Visualising Data
Describing and Visualising Data
Describing and Visualising Data combines centre, spread, frequencies and graphics to communicate quantitative patterns. It contrasts mean with median, standard deviation with interquartile range, and bar charts with histograms. The central habit is to inspect missing codes, sample size, shape and denominators before presenting a compact summary.
What this chapter covers
- 01
Frequency inspection
- 02
Mean and median
- 03
Range and quartiles
- 04
Standard deviation
- 05
Histograms and bars
- 06
Honest axes and captions
Worked example · free
Compare two groups
- 1State that their centres match.
- 1Identify greater spread in the second group.
- 1Avoid a causal explanation from summaries alone.
Key terms
- Median
- The middle ordered value or midpoint of the two middle values.
- Quartile
- A cut point dividing ordered observations into four portions.
- Histogram
- A graph using adjacent intervals to show a quantitative distribution.
Describing and Visualising Data FAQ
When is the median useful?
It is useful when a skewed distribution or extreme values would pull the mean away from the typical observation. Pair it with the interquartile range and describe the skew rather than hiding the tail.
Why inspect frequencies first?
They expose missing codes, unexpected categories and sparse groups before those values distort summaries or models. This inspection also protects denominators by showing how many observations remain valid for each comparison.
What makes a chart misleading?
Unclear denominators, truncated axes, inappropriate bin widths or category graphics that imply a false numerical scale can mislead. A caption should identify the population, unit and intended comparison so the visual can stand alone.
Assessment move
Begin with frequencies and valid counts. Describe shape before selecting summaries, then pair centre with spread. Redraw the same dataset as a table, histogram and box plot to see what each reveals. Label units, denominators and sample sizes on every output. When comparing groups, separate the observed difference from explanations that require stronger evidence.
Treat description as an analytical decision rather than a preliminary formality. Inspect variable labels, types, category frequencies, missing codes and plausible ranges before producing a summary. For quantitative variables, draw a histogram and box plot, then compare mean with median and standard deviation with interquartile range. Explain which pair better represents the observed shape.
For categorical variables, report counts and percentages with their denominators and use separated bars rather than continuous bins. Practise changing bin width and axis range to see which features are stable and which are design artefacts. When groups are compared, show valid sample size, centre, spread and any important imbalance.
If survey weights are used, distinguish the raw record count from the weighted population estimate and state the analytical unit. Write captions that identify the population, encoding and main comparison without claiming a cause. Recompute one plotted percentage from its underlying count as a validation check.
Keep an error log for hidden missing codes, inappropriate averages, mismatched denominators, truncated axes, unlabeled units and captions that merely repeat titles. In the report, use each figure to advance the research question and remove redundant output. Before submission, compare the prose with the plotted values and verify that rounding does not create inconsistent totals. Add a visual audit to every descriptive exercise.
Cover the axis labels and infer what variable type the graphic implies, then reveal them and check whether the encoding matches. Rebuild a histogram with alternative sensible bins and identify conclusions that remain stable. Recalculate the mean after adding one extreme value and compare the change with the median.
Construct two small datasets with the same mean but different spread, and another pair with the same mean and standard deviation but different shapes. These contrasts prevent any one statistic from being treated as a complete description. For survey data, separate the number of records from weighted estimates and explain which quantity each graph displays. Inspect missingness by group when it could affect a comparison.
Check whether percentages total approximately one hundred after rounding and investigate any unexpected gap. Write captions as standalone messages: population, measure, comparison and main pattern. Remove colour or decoration that does not encode information. When presenting a group difference, include both absolute values and the difference so magnitude remains interpretable.
In the report, position each display near the prose that explains it and avoid repeating the entire table in sentences. Finish the editing pass by tracing every number in the narrative back to a reproducible output and every visual claim back to a visible mark on the chart. Save a ledger containing variable type, valid count, chosen centre, chosen spread and preferred graphic so every descriptive choice remains traceable.
Create a fresh descriptive comparison and ensure its centre, spread, shape, valid count and denominator can all be audited.
Working through Describing and Visualising Data in SECU0006? Sia is AskSia’s AI Statistics tutor — ask any SECU0006 Describing and Visualising Data question and get a clear, step-by-step explanation grounded in how SECU0006 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.