4.2 Bar Charts, Clustered Bars & Segmented Stacks

Structural Formats of Bar Charts in Public Sector Testing

Bar charts are among the most versatile graphical tools in civil service reporting. The hypothetical exercises below use bars to compare administrative divisions, spending categories and years. Learn the actual assessment’s displays from its familiarisation material.

Unlike line graphs, which emphasize continuous trends, bar charts display separate categorical observations across four primary configurations:

1. Simple Bar Charts (Vertical Columns or Horizontal Bars): Compare discrete entities (capital expenditure across departments or headcount by local authority). Horizontal bars are used when category labels are long or when displaying ranked league tables.
2. Clustered Bar Charts (Grouped Bars): Display multiple data series side-by-side within each category (comparing planned vs. actual expenditure across four quarters).
3. Stacked (Segmented) Bar Charts: Present aggregate quantities where each bar is partitioned into sub-components, enabling simultaneous analysis of total volume and component makeup.
4. 100% Component Bar Charts: A variation of stacked bars where all bars are normalized to an identical height (100%), highlighting proportional distribution rather than absolute magnitude.

---

The Truncated Axis & Scale Interval Verification

The first duty when evaluating any chart in a psychometric test is to verify the baseline origin and axis graduations.

The Truncated Baseline (Non-Zero Origin)

For a bar chart encoding quantity by bar length, a zero baseline preserves meaningful length comparisons. A truncated axis, such as one starting at €80 million, can distort those comparisons. Other chart types, including line charts, need not always start at zero. Truncating the vertical axis exaggerates modest differences:

• Suppose Department A spent €82 million and Department B spent €86 million.
• On a standard zero-based axis, both bars are tall, and the €4 million gap represents an increment of under 5%.
• On an axis starting at €80 million, Department A’s bar extends 2 units above the baseline, while Department B’s bar extends 6 units. Visually, Department B appears three times as large as Department A ($300\%$), even though the actual mathematical difference is merely $4.88\%$: $\text{Relative Difference} = \left(\frac{86 - 82}{82}\right) \times 100 \approx 4.88\%$

Axis Interval Interpolation

When bar tops fall between gridlines, determine the interval step:

1. Identify the difference between two labelled major gridlines (e.g., €50,000 to €100,000 = €50,000 interval).
2. Count the equal intervals between them, not just the interior ticks. If there are 5 intervals, each step represents €10,000 (€50,000 / 5).
3. If a bar terminates halfway between the second and third minor ticks, its value is €70,000 + (€10,000 / 2) = €75,000.

---

Clustered Bar Charts: Multi-Dimensional Grouping

Clustered bar charts group two or more bars under a single categorical label on the primary axis to compare:

• Baseline Year vs. Outturn Year: Evaluating performance before and after a policy intervention.
• Budgeted Allocation vs. Actual Expenditure: Monitoring fiscal discipline.
• Permanent Staff vs. Contract / Agency Personnel: Analyzing workforce composition.

Strategic Legend Navigation

Every clustered chart relies on a legend (shading, hatching, or colour coding) to differentiate series. The primary error under pressure is series transposition—reading the "Actual" bar as "Budgeted", or tracking 2024 instead of 2025.

To prevent errors:

• Anchor your focus to the legend pattern before scanning clusters.
• Confirm whether the question requires a within-category comparison (e.g., "In 2025, by how much did Department C's actual spend exceed its budget?") or a cross-category comparison (e.g., "Which department recorded the highest percentage growth in spend between 2024 and 2025?").

---

Stacked Bar Charts: Segment Extraction Mechanics

Stacked bar charts convey two layers of data simultaneously: the overall bar height represents the cumulative total, while internal partitions represent sub-components.

The Boundary Subtraction Rule

In a standard stacked bar, only the lowest segment begins at the zero baseline. All intermediate and upper segments sit on top of preceding components.

To determine the absolute value of any non-baseline segment, subtract the segment's lower boundary from its upper boundary: $\text{Segment Value} = \text{Upper Boundary Value} - \text{Lower Boundary Value}$

Consider a regional health board staffing bar where:

• Medical & Nursing occupies the base from 0 to 450 WTE.
• Allied Health Professionals occupies the middle band from 450 to 720 WTE.
• Management & Support occupies the top band from 720 to 900 WTE.

To extract Allied Health Professionals: $\text{Allied Health WTE} = 720 - 450 = 270\text{ WTE}$

---

100% Component Bar Charts: Proportions vs. Absolute Volumes

In a 100% component bar chart, every bar is scaled to exactly 100%, isolating percentage composition across categories or years.

The Volume Independence Principle

Because 100% component bars normalize all totals, a change in percentage share does not dictate the direction of absolute volume change:

• If an agency's total budget doubles from €10 million in Year 1 to €20 million in Year 2:
  • An IT allocation that drops from 25% to 20% of the bar actually increased in absolute cash from €2.5 million (25% of €10m) to €4.0 million (20% of €20m).
• Conversely, a sub-category whose percentage share expands on the chart may have shrunk in raw volume if the overall bar total contracted sharply.

For a positive component share, an absolute quantity requires a total or equivalent information from which the quantity can be derived. If neither is supplied or inferable, the quantity cannot be determined. An absent total label alone is not enough to decide this: another stated component value may supply the scale.

---

Comparative Reference: Chart Formats & Distractor Traps

The following table summarizes the four core bar chart formats, their analytical functions, and typical exam traps.

---

Worked Example: Calculating Percentage Point vs. Volume Growth

A transport authority presents a clustered bar chart of passenger journeys (millions) across two modes: Heavy Rail and Light Rail.

• In 2024: Heavy Rail = 36.0m; Light Rail = 24.0m (Total = 60.0m). Heavy Rail share = $36.0 / 60.0 = 60.0\%$.
• In 2025: Heavy Rail = 42.0m; Light Rail = 28.0m (Total = 70.0m). Heavy Rail share = $42.0 / 70.0 = 60.0\%$.

Analysis:

• Absolute growth in Heavy Rail journeys: $42.0 - 36.0 = +6.0\text{ million journeys}$.
• Relative percentage growth in Heavy Rail: $((42.0 - 36.0) / 36.0) \times 100 \approx +16.67\%$.
• Percentage point shift in Heavy Rail’s share of total journeys: $60.0\% - 60.0\% = 0.0\text{ percentage points}$.

Even though Heavy Rail volume grew by 6.0 million journeys (+16.67%), its proportional share of total transit remained unchanged because Light Rail expanded at the identical relative rate ($((28.0 - 24.0) / 24.0) \times 100 \approx +16.67\%$).