Measures of Central Tendency & Dispersion: Mean, Median, Mode, Range, Variance, Standard Deviation

TAKEAWAYS:

• Central tendency (mean, median, mode) identifies the "typical" value in a dataset, but each has unique advantages and limitations.
• Dispersion (range, variance, standard deviation) quantifies how spread out data points are, with standard deviation being the most widely used.
• Skewness and kurtosis describe the shape of distributions beyond just central tendency and dispersion.
• Real-world applications include calculating average loan interest rates (banks), delivery delays (Pathao), and stock price volatility (NEPSE).
• Exam focus: Be able to compute all measures manually, interpret their meaning, and justify which measure to use in a given scenario.

Core Concepts: Central Tendency

1. Definitions and When to Use Each

Central tendency measures identify the "center" of a dataset. The three main measures are:

Why it matters:

• Mean is affected by extreme values (e.g., a billionaire in a room of poor people skews the average income).
• Median is robust to outliers (e.g., house prices in Kathmandu).
• Mode is useful for categorical data (e.g., most popular phone brand in Nepal: Samsung).

2. Worked Example: Comparing Mean, Median, and Mode

Scenario: A farmer sells pumpkins at a farm stand. Instead of weighing each pumpkin, he estimates the "typical" size by looking at the central tendency of a sample. Here are the weights (in kg) of 10 pumpkins: 3.2, 4.1, 2.8, 5.0, 3.5, 4.1, 6.0, 3.8, 4.1, 2.9

Step 1: Calculate the Mean

\bar{x} = \frac{3.2 + 4.1 + 2.8 + 5.0 + 3.5 + 4.1 + 6.0 + 3.8 + 4.1 + 2.9}{10} = \frac{39.5}{10} = 3.95 \text{ kg}

Step 2: Find the Median

1. Order the data: 2.8, 2.9, 3.2, 3.5, 3.8, 4.1, 4.1, 4.1, 5.0, 6.0
2. Since $n=10$ (even), median = average of 5th and 6th values: $\text{Median}=\frac{3.8+4.1}{2}=3.95\text{ kg}$

Step 3: Identify the Mode

• The value 4.1 appears 3 times (most frequent).
• Mode = 4.1 kg

Interpretation:

• All three measures are close here (3.95, 3.95, 4.1), suggesting a symmetric distribution.
• If one pumpkin were 20 kg, the mean would spike, but the median and mode would stay stable.

3. Choosing the Right Measure

Use this flowchart to decide:

flowchart TD
    A["Is data symmetric?"] -->|"Yes"| B["Use Mean"]
    A -->|"No"| C["Is data skewed?"]
    C -->|"Right-skewed"| D["Use Median"]
    C -->|"Left-skewed"| D
    D --> E["Check for outliers"]
    E -->|"Outliers present"| F["Use Median"]
    E -->|"No outliers"| G["Mean or Median (both similar)"]
    G --> H["Use Mode if data is categorical"]

Real-world tie-in:

• Nepal Rastra Bank (NRB) uses the median to report "typical" household income because extreme wealth (e.g., business tycoons) would distort the mean.
• Pathao might use the median delivery time to report "average" wait times, as a few extremely delayed orders (e.g., traffic jams) would skew the mean.

Core Concepts: Measures of Dispersion

Dispersion tells us how spread out the data is. Key measures:

4. Worked Example: Calculating Variance and Standard Deviation

Scenario: A bank offers two loan schemes to customers. The interest rates (in %) for 10 customers in each scheme are:

Step 1: Calculate Mean for Both Schemes

• Scheme A: ${\bar{x}}_A=\frac{8.2+9.1+\cdots +9.3}{10}=8.59\%$
• Scheme B: ${\bar{x}}_B=\frac{7.5+8.0+\cdots +8.0}{10}=7.93\%$

Step 2: Calculate Variance for Scheme A

Sum of squared deviations = 2.1051 Variance (${\sigma }_A^2$) = $\frac{2.1051}{10}=0.21051$

Step 3: Calculate Standard Deviation for Scheme A

${\sigma }_A=\sqrt{0.21051}\approx 0.459\%$

Step 4: Repeat for Scheme B

Variance (${\sigma }_B^2$) ≈ 0.0205 Standard Deviation (${\sigma }_B$) ≈ 0.143%

Interpretation:

• Scheme A has a higher standard deviation (0.459% vs. 0.143%), meaning its interest rates are more spread out.
• Scheme B is more consistent (lower risk for the bank, but also lower potential for high profits).

5. Skewness and Kurtosis: Beyond Central Tendency

Skewness

Measures asymmetry in data:

• Positive skew: Tail on the right (e.g., income distribution).
• Negative skew: Tail on the left (e.g., exam scores where most students score high).
• Symmetric: No skew (e.g., heights of adults).

Formula: $$\text{Skewness}=\frac{n}{(n-1)(n-2)}\sum {\left(\frac{x_i-\bar{x}}{\sigma }\right)}^3$$

Real-world example:

• NEPSE stock prices often show positive skew: a few stocks surge while most remain stable.
• Pathao delivery times might be right-skewed: most deliveries are fast, but a few are delayed due to traffic.

Kurtosis

Measures "tailedness" of a distribution:

• Mesokurtic: Normal distribution (kurtosis = 0).
• Platykurtic: Flatter than normal (kurtosis < 0).
• Leptokurtic: Heavier tails than normal (kurtosis > 0).

Relation to standard deviation and fourth central moment: For a symmetric distribution with standard deviation $\sigma$, the fourth central moment ${\mu }_4$ determines kurtosis: $$\text{Kurtosis}=\frac{{\mu }_4}{{\sigma }^4}-3$$

Worked Example: Past exam question adapted. Given: Standard deviation $\sigma =7$. Find ${\mu }_4$ for:

1. Mesokurtic: Kurtosis = 0 $$0=\frac{{\mu }_4}{7^4}-3⟹{\mu }_4=3\times 7^4=3\times 2401=7203$$
2. Platykurtic: Kurtosis = -1 $$-1=\frac{{\mu }_4}{7^4}-3⟹{\mu }_4=2\times 2401=4802$$
3. Leptokurtic: Kurtosis = 2 $$2=\frac{{\mu }_4}{7^4}-3⟹{\mu }_4=5\times 2401=12005$$

1. Mesokurtic (normal distribution, moderate tails).
2. Platykurtic (flatter, shorter tails).
3. Leptokurtic (heavier tails, sharper peak)."**

In the Real World

1. eSewa and Khalti (Digital Payments)

   • Central Tendency: The median transaction amount is used to report "typical" user spending because a few large transactions (e.g., bill payments) would inflate the mean.
   • Dispersion: The standard deviation of transaction times helps eSewa identify delays (e.g., during Dashain when everyone sends money).

2. Daraz (E-commerce)

   • Range: Daraz might report the range of delivery times (e.g., "2–5 days") to set customer expectations.
   • Variance: High variance in delivery times could indicate logistics issues (e.g., traffic in Kathmandu vs. Pokhara).

3. NTC (Telecom Infrastructure)

   • Skewness: Internet speeds in Nepal are often right-skewed: most users have decent speeds, but a few experience extreme slowdowns due to infrastructure gaps.
   • Standard Deviation: NTC uses this to compare network reliability across regions (e.g., high SD in rural areas vs. low SD in cities).

4. Ncell (Mobile Network)

   • Coefficient of Variation (CV): Ncell might compare the CV of call drop rates across towers to identify which areas need upgrades.
     • Example: If Tower A has a mean drop rate of 2% and SD of 0.5%, while Tower B has a mean of 1% and SD of 0.3%, Tower B is more consistent (lower CV).

5. NEPSE (Stock Market)

   • Leptokurtic Returns: Stock returns in Nepal often show heavy tails (leptokurtic), meaning occasional extreme gains/losses are more likely than in a normal distribution.
   • Mean vs. Median: The median stock return is often used because a few high-performing stocks (e.g., NMB Bank) skew the mean.

Exam Tip

1. Always justify your choice:

   • If asked which measure to use, explain why (e.g., "Use the median because the data is skewed").
   • Example answer:

     "For the pumpkin weights, the mean and median are both appropriate because the distribution is roughly symmetric. However, if one pumpkin were much heavier (e.g., 20 kg), the median would better represent the 'typical' size."

2. Show all steps for calculations:

   • For variance/standard deviation, always:
     1. Calculate the mean.
     2. Compute deviations from the mean.
     3. Square the deviations.
     4. Sum and divide by $n$ or $n-1$.
   • Partial credit is often given for correct intermediate steps.

3. Watch for units:

   • Standard deviation has the same units as the original data (e.g., kg for pumpkin weights, % for interest rates).
   • Coefficient of variation is unitless (always a percentage).

4. Memorize key formulas:

   • Mean: $\bar{x}=\frac{\sum x_i}{n}$
   • Variance: ${\sigma }^2=\frac{\sum (x_i-\bar{x}{)}^2}{n}$ (population)
   • Standard deviation: $\sigma =\sqrt{{\sigma }^2}$
   • Kurtosis relation: ${\mu }_4=(\text{Kurtosis}+3){\sigma }^4$

5. Practice with real data:

   • Use datasets from Nepal’s CBS (Central Bureau of Statistics) or Nepal Rastra Bank reports to compute measures manually.
   • Example: Calculate the mean and standard deviation of Nepal’s GDP growth rates over the past 10 years.

6. Common pitfalls:

   • Forgetting to square deviations when calculating variance.
   • Using $n$ instead of $n-1$ for sample variance (unless specified).
   • Misinterpreting skewness (e.g., confusing positive skew with "better" data).

Final Visual Summary:

mindmap
  root((Measures of Central Tendency & Dispersion))
    Central Tendency
      Mean
        Formula: Sum/n
        Affected by outliers
      Median
        Middle value
        Robust to outliers
      Mode
        Most frequent
        Useful for categories
    Dispersion
      Range
        Max - Min
        Simple but sensitive
      Variance
        Average squared deviation
        Units: squared original units
      Standard Deviation
        Square root of variance
        Original units
      Coefficient of Variation
        SD/Mean * 100
        Compares dispersion across datasets
    Shape Measures
      Skewness
        Asymmetry
        Positive/negative
      Kurtosis
        Tailedness
        Mesokurtic/Platykurtic/Leptokurtic
    Real-World Applications
      Banking: Loan interest rates
      E-commerce: Delivery times
      Telecom: Network reliability
      Stock Market: Return volatility