10. Measures of Central Tendency

🎯 Measures of Central Tendency

In engineering data analysis, a single representative value is essential to capture the central location or typical magnitude of a dataset. Discover the mathematical principles, properties, outlier sensitivity, and Pearson BTEC application standards of the Mean, Median, and Mode.

Arithmetic Mean (x̄) Median & Ranking Mode & Multimodality Linear Interpolation Skewness Analysis SPC Tolerances

3Core Averages

2Data Types

12Quiz Tasks

8Exam Qs

📊Mean (x̄)Ungrouped & weighted

⚖️Median (Q₂)50th percentile rank

🔝ModeHighest frequency

📐InterpolationGrouped distributions

⚙️EngineeringCNC & QA tolerance

🎮Self-Check12-question challenge

Summary

Comparative Overview of the Three Fundamental Averages

Selecting the correct measure depends on the nature of the data, the measurement scale (ratio/interval/nominal), and whether extreme outliers or distribution skewness are present.

Measure	Definition & Core Formula	Sensitivity to Outliers	Best Engineering Use Cases
Arithmetic Mean (x̄)	Sum of all values divided by count: x̄ = Σx / n	High — pulled strongly toward extreme values	Thermal heat balance, total electrical power consumption, mass centroid, SPC nominal centering.
Median	Middle value when arranged in order: (n+1)/2 position	Low (Robust) — unaffected by extreme magnitude	Fatigue life cycles, component reliability testing (MTTF), tensile strength with defect specimens.
Mode	Most frequently occurring observation: argmax f(x)	None — dependent purely on count peak	Tooling inventory selection, standard fastener stock (M6 vs M8), resonance frequency peak analysis.

Distribution Geometry

Skewness and Relative Positions

The relationship between Mean, Median, and Mode reveals the symmetry or asymmetry of an engineering process:

Symmetrical (Normal): Mean = Median = Mode

Positive (Right) Skew: Mode < Median < Mean

Negative (Left) Skew: Mean < Median < Mode

Pearson's Empirical Rule: For unimodal moderately skewed distributions: Mean - Mode ≈ 3 × (Mean - Median).

Mathematical Property

The Zero-Sum Algebraic Identity

The arithmetic mean is the unique center of gravity of a dataset. The sum of all individual deviations from the mean is identically zero:

Σ(x - x̄) = Σx - n·x̄ = 0

Furthermore, the sum of squared deviations Σ(x - a)² is minimized strictly when a = x̄, making the mean the optimal least-squares estimator in engineering instrumentation.

📊

1 · Ungrouped Mean, Median & Mode

Raw inspection records, sensor time-series, and quality batch samples evaluated directly from discrete observations.

⚡ Core Mathematical Formulas (Ungrouped Data)

x̄ = (Σ x) / n

Sample Mean: sum of all values divided by total count n

x̄_w = (Σ w·x) / (Σ w)

Weighted Mean: each value weighted by factor w (e.g. mass, volume)

Pos = (n + 1) / 2

Median Rank: exact middle (odd n) or mean of two central values (even n)

Mode = max frequency

Modal value: observation occurring with the highest frequency

🧮 Interactive Central Tendency Calculator (Ungrouped) Instant statistical breakdown

Enter Values (comma or space separated):

Presets: CNC Pins (mm) Tensile Yield with Outlier Resistor 100Ω Check

Click "Calculate" or select a preset above to inspect the step-by-step breakdown.

📚 Essential Terminology & Definitions

Arithmetic Mean (x̄): The mathematical average; incorporates every point; sensitive to outliers.

Median (Q₂): The 50th percentile rank; splits ordered dataset into two halves of equal size.

Unimodal: A distribution containing exactly one pronounced peak or single highest frequency.

Bimodal: Two distinct values share the highest frequency; often signals two blended machine processes.

Outlier: An observation lying abnormally far from other values, often caused by sensor failure or void defect.

Sample vs Population: Sample mean x̄ estimates the true population parameter μ.

Deep Dive

The Arithmetic Mean & Weighted Mean

For sample data x₁, x₂, ..., xₙ, the sample mean is:

x̄ = (x₁ + x₂ + ... + xₙ) / n = (Σ xᵢ) / n

When components contribute with different relative importance (weights wᵢ), such as alloy percentages or mass elements:

x̄_w = (Σ wᵢ · xᵢ) / (Σ wᵢ)

Caution: Even a single corrupted sensor reading (e.g. 0V or 9999) heavily shifts x̄. Always audit raw inspection records before computing the mean.

Robust Ranking

The Median Algorithm

To determine the median of an ungrouped dataset:

Sort the dataset in ascending order: x₍₁₎ ≤ x₍₂₎ ≤ ... ≤ x₍ₙ₎.
If n is odd: Median is the value at index (n + 1)/2.
If n is even: Median is the arithmetic mean of the two middle observations at n/2 and (n/2) + 1:
Median = [ x_(n/2) + x_(n/2 + 1) ] / 2

Engineering Advantage: If 1 out of 20 test bars snaps prematurely due to fixture slippage, the median is unaffected, providing the true characteristic performance.

📝 Worked Example 1: Precision Turned Pin Diameters CNC Quality Assurance

Six sample pins machined on a CNC lathe are measured with a digital micrometer (in mm):
25.02, 25.05, 25.01, 25.08, 25.05, 25.03.
Calculate the mean, median, mode, and verify that the sum of deviations is zero.

Step 1: Sum & Mean

Σx = 25.02 + 25.05 + 25.01 + 25.08 + 25.05 + 25.03 = 150.24 mm
x̄ = 150.24 / 6 = 25.040 mm

Step 2: Ordered ArrayAscending order: 25.01, 25.02, 25.03, 25.05, 25.05, 25.08 mm (n = 6, even)

Step 3: Median Rank

Positions: n/2 = 3rd (25.03), (n/2)+1 = 4th (25.05)
Median = (25.03 + 25.05) / 2 = 25.040 mm

Step 4: Modal Value

25.05 mm appears twice (frequency = 2); all other values appear once.
Mode = 25.050 mm (Unimodal)

Step 5: DeviationsΣ(x - x̄) = (-0.03) + (-0.02) + (-0.01) + (+0.01) + (+0.01) + (+0.04) = 0.000 mm (Verified)

📝 Worked Example 2: Outlier Impact on Tensile Yield Testing Material Science

Seven structural aluminum test coupons yield the following strength values (MPa):
445, 452, 448, 180, 450, 456, 449.
Specimen 4 (180 MPa) failed prematurely due to a macro-porosity defect. Evaluate the distortion of the mean vs the stability of the median.

Mean (with Outlier)

Σx = 180 + 445 + 448 + 449 + 450 + 452 + 456 = 2880 MPa
x̄ = 2880 / 7 = 411.43 MPa

Median

Ordered: 180, 445, 448, 449, 450, 452, 456. Position (7+1)/2 = 4th.
Median = 449.00 MPa

Mean (Clean Data)Excluding the defect: x̄_clean = (445 + 448 + 449 + 450 + 452 + 456) / 6 = 2700 / 6 = 450.00 MPa

Engineering Finding

The mean dropped by 38.57 MPa (-8.6%) due to a single defective coupon, while the median (449 MPa) remained within 1 MPa of the clean process mean!

📐

2 · Grouped Data & Frequency Interpolation

Continuous measurement ranges tabulated into class intervals. Calculating estimated mean, linear median interpolation, and modal estimation.

⚡ Grouped Estimation Formulas (BTEC Standards)

x̄ = (Σ f·x) / (Σ f)

Grouped Mean: f = class frequency, x = class midpoint (L_lower + L_upper)/2

Median = L + [ (N/2 - CF) / f_m ] · w

Linear Interpolation: L = lower class boundary, CF = preceding cumulative freq, f_m = median class freq, w = class width

Mode = L + [ Δ₁ / (Δ₁ + Δ₂) ] · w

Modal Formula: Δ₁ = f_m - f_(m-1), Δ₂ = f_m - f_(m+1), w = modal class width

N = Σ f

Total sample size across all grouped classes

⚙️ Interactive Grouped Data Interpolator Test class distributions

Adjust the frequencies for a 5-class distribution with width w = 10 (e.g. 50-60, 60-70, 70-80, 80-90, 90-100 MPa):

50–60 (f₁):

60–70 (f₂):

70–80 (f₃):

80–90 (f₄):

90–100 (f₅):

Click "Interpolate" to compute grouped mean, linear median interpolation, and modal estimate.

Interpolation Theory

Derivation of the Grouped Median Formula

When data is grouped, raw observations are lost. We assume observations inside the median class are uniformly distributed across the class interval w:

Distance into class = (Target Rank - CF) / f_m

Multiplying by class width w and adding to the lower class boundary L yields:

Median = L + [ (N/2 - CF) / f_m ] · w

Where CF is the cumulative frequency up to (but not including) the median class.

Histogram Geometry

Estimating the Grouped Mode

The modal class is the interval possessing the highest frequency f_m. The mode within this class is pulled toward whichever neighbouring class has higher frequency:

Δ₁ = f_m - f_(m-1) | Δ₂ = f_m - f_(m+1)

Mode = L + [ Δ₁ / (Δ₁ + Δ₂) ] · w

This formula is geometrically identical to drawing diagonal lines across the modal bar on a frequency histogram.

📝 Worked Example 3: Fatigue Test Cycles of 60 Aircraft Rivets Full Grouped Calculation

Sixty aerospace rivets were subjected to cyclic fatigue until shear crack initiation. Cycles to failure (in thousands, 10³):

Class Interval (10³ cycles)	Frequency (f)	Midpoint (x)	f · x	Cumulative Freq (CF)
20 ≤ c < 25	6	22.5	135.0	6
25 ≤ c < 30	14	27.5	385.0	20
30 ≤ c < 35	22	32.5	715.0	42
35 ≤ c < 40	12	37.5	450.0	54
40 ≤ c < 45	6	42.5	255.0	60
Total	N = 60	—	Σfx = 1940.0	—

Grouped Meanx̄ = (Σ f·x) / N = 1940.0 / 60 = 32.333 × 10³ cycles (32,333 cycles)

Median Class

Target rank N/2 = 60 / 2 = 30th item. Class '30 ≤ c < 35' contains items 21 through 42.
Median class: 30 to 35. (L = 30, w = 5, CF = 20, f_m = 22)

Median CalcMedian = 30 + [ (30 - 20) / 22 ] × 5 = 30 + (10 / 22) × 5 = 30 + 2.273 = 32.273 × 10³ cycles

Modal Class

Highest frequency is 22 in class 30 to 35. L = 30, w = 5.
Δ₁ = 22 - 14 = 8
Δ₂ = 22 - 12 = 10
Mode = 30 + [ 8 / (8 + 10) ] × 5 = 30 + (8 / 18) × 5 = 30 + 2.222 = 32.222 × 10³ cycles

ConclusionMean (32.33k) ≈ Median (32.27k) ≈ Mode (32.22k) indicating a highly symmetrical fatigue distribution!

⚙️

Engineering Applications & Quality Assurance

Industrial deployment in Statistical Process Control (SPC), CNC lathe calibration, material specification compliance, and reliability engineering.

🛠️ Application 1: Precision CNC Machining & Tool Wear Drift

In high-volume precision turning, cutting tool inserts wear gradually over thousands of cycles, shifting the process mean (x̄) away from the nominal target diameter. Quality inspectors sample 10 shafts every hour to detect mean drift before scrap is produced.

Nominal Specification: 20.000 mm ± 0.025 mm (USL = 20.025 mm, LSL = 19.975 mm)

🎛️ Live Tool Offset Compensation Tool

Nominal Target (mm):

Sample Mean x̄ (mm):

Enter nominal target and measured sample mean to calculate tool compensation.

🔬 Application 2: Tensile Yield Verification & Outlier Filtering

Aerospace specifications require batch compliance testing of cast titanium brackets. Porosity pockets can trigger non-representative specimen breakage. While British Standards (BS EN ISO 6892-1) dictate test protocol, comparing the median to the arithmetic mean immediately flags uncharacteristic outliers.

Criterion: If |Mean - Median| > 0.05 × Median, an outlier investigation is triggered.

⚖️ Batch Robustness Evaluation

Batch Mean (MPa):

Batch Median (MPa):

Assess batch validity against outlier disparity threshold.

⚡ Application 3: Reliability Engineering (MTTF vs Median Operating Life)

In electronic power supply design, component lifetimes follow a skewed Weibull or log-normal failure curve. The Mean Time To Failure (MTTF) is heavily inflated by the few components that survive for extended durations, whereas the Median Life indicates the exact timestamp at which 50% of units will have failed.

Right-skewed Reliability: Mode (infant mortality) < Median (50% failure) < MTTF (Mean)

🎮 Central Tendency Quick-Fire Challenge

Question: 1/12 Score: 0 Streak: 0 🔥

Press "Start Challenge" below to test your mastery of measures of central tendency!

📝

Examination Practice Questions

BTEC Higher National Unit 4002 exam-style questions with detailed step-by-step worked solutions and marking guides.

Question 1 · Ungrouped Mean, Median & Mode4 Marks

An automated robotic arm dispenses solder paste onto printed circuit boards. The mass of paste (in mg) on eight consecutive boards was measured as:
14.2, 14.6, 14.3, 14.8, 14.6, 14.1, 14.6, 14.4.
(a) Calculate the arithmetic mean mass.
(b) Determine the median and the mode.

[1 Mark] Sum Σx = 14.2 + 14.6 + 14.3 + 14.8 + 14.6 + 14.1 + 14.6 + 14.4 = 115.6 mg.

[1 Mark] Mean x̄ = 115.6 / 8 = 14.450 mg.

[1 Mark] Ordered array: 14.1, 14.2, 14.3, 14.4, 14.6, 14.6, 14.6, 14.8 (n = 8).
Middle values at 4th (14.4) and 5th (14.6) position: Median = (14.4 + 14.6) / 2 = 14.500 mg.

[1 Mark] Mode: 14.6 mg appears three times (frequency 3). Mode = 14.600 mg.

Question 2 · Outlier Impact Analysis3 Marks

A hydraulic pressure sensor logs five pressure readings (bar): 120, 122, 121, 119, 210. The reading of 210 bar was caused by an electrical spike.
(a) Calculate the mean and median with the faulty reading.
(b) Calculate the clean mean with the faulty reading excluded and state which measure was more robust.

[1 Mark] With spike: Σx = 692 bar. Mean = 692 / 5 = 138.4 bar. Ordered: 119, 120, 121, 122, 210. Median = 121.0 bar.

[1 Mark] Clean data (119, 120, 121, 122): Σx = 482 bar. Clean Mean = 482 / 4 = 120.5 bar.

[1 Mark] The Median (121.0 bar) is far more robust; it differed from the true clean mean by only 0.5 bar, whereas the contaminated mean was distorted by 17.9 bar.

Question 3 · Weighted Mean of an Engineering Composite4 Marks

A lightweight structural panel consists of three layers bonded together:
• Carbon-fiber skin: thickness 1.5 mm, density 1.60 g/cm³
• Aluminum honeycomb core: thickness 12.0 mm, density 0.08 g/cm³
• Glass-fiber backing: thickness 2.5 mm, density 1.90 g/cm³
Calculate the weighted mean density of the composite panel per unit area.

[1 Mark] Weighting factors are thicknesses: w₁ = 1.5 mm, w₂ = 12.0 mm, w₃ = 2.5 mm. Total thickness Σw = 1.5 + 12.0 + 2.5 = 16.0 mm.

[2 Marks] Σ(w · ρ) = (1.5 × 1.60) + (12.0 × 0.08) + (2.5 × 1.90) = 2.40 + 0.96 + 4.75 = 8.11 mm·g/cm³.

[1 Mark] Weighted Mean Density = 8.11 / 16.0 = 0.507 g/cm³ (or 506.9 kg/m³).

Question 4 · Grouped Mean of Insulation Resistances5 Marks

The insulation resistance (in MΩ) of 80 electric motors was tested after assembly:
• 20 to <30 MΩ: 8 motors
• 30 to <40 MΩ: 24 motors
• 40 to <50 MΩ: 32 motors
• 50 to <60 MΩ: 16 motors
Calculate the estimated mean insulation resistance.

[1 Mark] Identify class midpoints: x₁ = 25, x₂ = 35, x₃ = 45, x₄ = 55 MΩ. Total count N = 80.

[3 Marks] Compute products f · x:
8 × 25 = 200
24 × 35 = 840
32 × 45 = 1440
16 × 55 = 880
Sum Σ(f·x) = 200 + 840 + 1440 + 880 = 3360 MΩ.

[1 Mark] Grouped Mean x̄ = 3360 / 80 = 42.000 MΩ.

Question 5 · Linear Interpolation for Grouped Median5 Marks

Using the electric motor insulation resistance distribution from Question 4:
[20–30: 8, 30–40: 24, 40–50: 32, 50–60: 16; Total N = 80].
Calculate the median resistance by linear interpolation.

[1 Mark] Compute Cumulative Frequencies (CF):
Class 20–30: CF = 8
Class 30–40: CF = 32
Class 40–50: CF = 64
Class 50–60: CF = 80.

[1 Mark] Target rank: N/2 = 80 / 2 = 40th item.
Since CF reaches 32 at class 30–40 and 64 at class 40–50, the 40th item lies in the 40 to 50 MΩ class.

[1 Mark] Identify formula parameters: Lower boundary L = 40, class width w = 10, preceding CF = 32, median class freq f_m = 32.

[2 Marks] Median = L + [ (N/2 - CF) / f_m ] × w
Median = 40 + [ (40 - 32) / 32 ] × 10 = 40 + (8 / 32) × 10 = 40 + 2.5 = 42.500 MΩ.

Question 6 · Grouped Mode Estimation4 Marks

Using the same distribution from Question 4 [20–30: 8, 30–40: 24, 40–50: 32, 50–60: 16]:
Calculate the modal insulation resistance using the modal interpolation formula.

[1 Mark] Modal class is 40 to 50 MΩ (highest frequency f_m = 32). Lower limit L = 40, width w = 10.

[1 Mark] Δ₁ = f_m - f_(preceding) = 32 - 24 = 8.

[1 Mark] Δ₂ = f_m - f_(succeeding) = 32 - 16 = 16.

[1 Mark] Mode = L + [ Δ₁ / (Δ₁ + Δ₂) ] × w = 40 + [ 8 / (8 + 16) ] × 10 = 40 + (8 / 24) × 10 = 40 + 3.333 = 43.333 MΩ.

Question 7 · CNC Lathe Calibration & Drift5 Marks

A CNC machine is producing shafts with target diameter 50.000 mm. Five sample parts measured:
50.024, 50.028, 50.022, 50.030, 50.026 mm.
(a) Calculate the sample mean diameter.
(b) Calculate the tool offset correction required to re-center the machine onto 50.000 mm.
(c) Verify that Σ(x - x̄) = 0 for this sample.

[2 Marks] (a) Σx = 50.024 + 50.028 + 50.022 + 50.030 + 50.026 = 250.130 mm.
Mean x̄ = 250.130 / 5 = 50.026 mm.

[1 Mark] (b) Offset = Nominal - Mean = 50.000 - 50.026 = -0.026 mm (tool must advance by 0.026 mm toward workpiece center).

[2 Marks] (c) Deviations: (50.024 - 50.026) = -0.002; (50.028 - 50.026) = +0.002; (50.022 - 50.026) = -0.004; (50.030 - 50.026) = +0.004; (50.026 - 50.026) = 0.000.
Sum = (-0.002 + 0.002 - 0.004 + 0.004 + 0) = 0.000 mm (Verified).

Question 8 · Pearson's Empirical Skewness5 Marks

In an endurance trial of ball bearings, the mean operating life was determined to be 12,400 hours and the median was 12,100 hours.
(a) Estimate the modal bearing life using Pearson's empirical formula: Mean - Mode ≈ 3(Mean - Median).
(b) State the direction of skewness of this distribution and explain what this implies for scheduled preventive maintenance.

[2 Marks] (a) Mean - Mode = 3 × (12,400 - 12,100) = 3 × 300 = 900 hours.
Mode = Mean - 900 = 12,400 - 900 = 11,500 hours.

[1 Mark] (b) Since Mode (11,500) < Median (12,100) < Mean (12,400), the distribution is positively (right) skewed.

[2 Marks] Engineering Implication: The mean life (12,400 hrs) is inflated by a tail of long-lived bearings. If maintenance is scheduled based on the mean, more than 50% of the bearings will have already failed! Maintenance should be scheduled before the median (12,100 hrs) or modal risk window.