10.1 Measures of Central Tendency and Dispersion in Appraisal

10.1 Measures of Central Tendency and Dispersion in Appraisal

Descriptive statistics provide the mathematical foundation for both single-property valuation (such as extracting paired-sales adjustments or direct capitalization rates) and mass appraisal modeling. Without statistical rigor, an appraiser risks selecting unrepresentative market benchmarks or falling victim to extreme transaction outliers.

---

1. Measures of Central Tendency

A measure of central tendency is a single summary metric that identifies the center, midpoint, or typical value of a statistical distribution. In real estate appraisal, three primary measures of central tendency are utilized: the arithmetic mean, the median, and the mode.

+---------------------------------------------------------------------------------------------------+
|                             MEASURES OF CENTRAL TENDENCY AT A GLANCE                              |
+-------------------+-----------------------------------+-------------------+-----------------------+
| STATISTIC         | MATHEMATICAL DEFINITION           | OUTLIER IMPACT    | APPRAISAL SUITABILITY |
+-------------------+-----------------------------------+-------------------+-----------------------+
| Mean (x̄)          | Sum of all values divided by n    | Highly sensitive  | Homogeneous sets only |
| Median (Positional)| 50th percentile midpoint value   | Highly robust     | Skewed property data  |
| Mode (Frequency)  | Most frequently occurring value   | Resistant         | Clustering / Bimodal  |
+-------------------+-----------------------------------+-------------------+-----------------------+

1. The Arithmetic Mean

The arithmetic mean (commonly called the average) is calculated by summing all individual values in a data set and dividing by the total number of observations ($n$):

$\bar{x} = \frac{\sum_{i=1}^n x_i}{n} = \frac{x_1 + x_2 + \dots + x_n}{n}$

• Appraisal Role: The mean represents the mathematical center of gravity of a dataset. It is computationally straightforward and incorporates the exact numerical magnitude of every observation.
• Critical Limitation (Sensitivity to Outliers): The mean is exceptionally vulnerable to extreme high or low values. In commercial real estate, a single distressed foreclosure liquidation or a single institutional sale of a newly developed, trophy Class A asset will severely distort the arithmetic mean, pulling it away from the typical transaction level.

2. The Median

The median is the positional midpoint of a dataset when all observations are arranged in ascending or descending order. Exactly 50% of observations lie below the median, and 50% lie above it:

$\text{Position of Median} = \frac{n + 1}{2}$

• Calculation:
  • If $n$ is odd, the median is the single observation situated at position $\frac{n + 1}{2}$.
  • If $n$ is even, the median is the arithmetic average of the two middle observations situated at positions $\frac{n}{2}$ and $\frac{n}{2} + 1$.
• Appraisal Superiority (Robustness): The median is a resistant statistic; it is entirely unaffected by extreme outliers. Whether the most expensive warehouse in a sample sold for $250/SF or $1,500/SF, the positional median remains unchanged. Consequently, appraisers, assessors (the IAAO ratio-study standard prefers the median), and market data providers widely treat the median as the most reliable single measure of typical real estate pricing.

3. The Mode

The mode is the observation that occurs with the greatest frequency in a dataset:

• Appraisal Role: While less commonly used than the median for continuous dollar data, the mode is invaluable for analyzing discrete or categorical variables (such as typical building class, number of stories, or ceiling clear height).
• Multimodal Distributions in Commercial Real Estate: A commercial dataset may have no mode, a single mode (unimodal), or multiple modes. When an appraiser analyzes commercial rents or prices per square foot and identifies a bimodal distribution (two distinct frequency peaks), this strongly indicates that the sample contains two separate submarkets that have been inadvertently lumped together. For example, a dataset of flex warehouse rents exhibiting peaks at $12.00/SF and $22.00/SF typically reflects unrenovated 1980s industrial space versus newly converted, air-conditioned life-science or creative tech flex space.

---

2. Measures of Dispersion

Central tendency describes where data clusters, but it reveals nothing about how widely the data is scattered. An average rent of $20.00/SF could represent five leases ranging from $19.50/SF to $20.50/SF (tight, highly predictable market), or five leases ranging from $10.00/SF to $30.00/SF (volatile, heterogeneous market). Measures of dispersion quantify this spread and uncertainty.

1. Range

The range is the simplest measure of dispersion, representing the arithmetic difference between the highest and lowest values:

$\text{Range} = x_{\text{max}} - x_{\text{min}}$

• Appraisal Role: Establishes the absolute outer boundaries (brackets) of observed market transactions.
• Limitation: Relies entirely on the two most extreme observations. It ignores all interior data and can be heavily distorted by a single unverified transaction.

2. Sample Variance ($s^2$)

Variance measures the average squared deviation of each observation from the arithmetic mean. In appraisal practice, because we analyze sample data rather than an entire property universe, we utilize the sample variance ($s^2$):

$s^2 = \frac{\sum_{i=1}^n (x_i - \bar{x})^2}{n - 1}$

3. Sample Standard Deviation ($s$)

Because variance is expressed in squared units (e.g., squared dollars or squared square feet), it cannot be directly interpreted alongside the original data. The sample standard deviation ($s$) is the positive square root of the sample variance, returning the dispersion metric back to the original units of measurement (e.g., $ or $/SF):

$s = \sqrt{s^2} = \sqrt{\frac{\sum_{i=1}^n (x_i - \bar{x})^2}{n - 1}}$

• Interpretation: Standard deviation quantifies the typical distance that individual observations deviate from the arithmetic mean. A small standard deviation denotes a tight, homogeneous market; a large standard deviation denotes substantial variation in property quality, location, or transaction terms.

4. The Normal Distribution and the Empirical Rule

Many natural and economic phenomena approximate a symmetrical, bell-shaped normal distribution. When transactional data is normally distributed, the Empirical Rule (68-95-99.7% Rule) dictates the exact proportion of data bounded by standard deviation intervals:

• $\bar{x} \pm 1s$: contains approximately 68.27% of all observations
• $\bar{x} \pm 2s$: contains approximately 95.45% of all observations
• $\bar{x} \pm 3s$: contains approximately 99.73% of all observations

                           THE EMPIRICAL RULE (68 - 95 - 99.7%)
                                      Mean (x̄)
                                         |
                                    .---'''---.
                                  .'     |     '.
                                .'       |       '.
                              .'         |         '.
                            .'           |           '.
                         .-'             |             '-.
                     ..-'  |             |             |  '-..
                 _.-'      |             |             |      '-._
           .---''          |             |             |          ''---.
         --|---------------|-------------|-------------|---------------|--
          x̄ - 3s         x̄ - 2s        x̄ - 1s        x̄ + 1s          x̄ + 2s        x̄ + 3s
           | <----------------------- 68.27% --------> |               |
           | <----------------------------- 95.45% ------------------> |
           | <----------------------------------- 99.73% ----------------------------> |

• Chebyshev's Theorem (Non-Normal Datasets): When data cannot be assumed to be normal, Chebyshev's Theorem guarantees that for any distribution shape, the proportion of observations falling within $k$ standard deviations ($k > 1$) of the mean is at least: $\text{Proportion} \ge 1 - \frac{1}{k^2}$ For $k = 2$, at least $1 - \frac{1}{4} = 75\%$ of all market data must lie within $\bar{x} \pm 2s$, regardless of how skewed the real estate market may be.

---

3. Skewness in Real Estate Price Distributions

In classical statistics, distributions are often assumed to be symmetrical. However, commercial real estate sales prices, rental rates, and building sizes rarely form a symmetrical bell curve. Instead, they exhibit skewness—an asymmetry in the distribution curve.

+---------------------------------------------------------------------------------------------------+
|                                 DISTRIBUTION SKEWNESS ARCHITECTURES                               |
+---------------------+-------------------------------+---------------------------------------------+
| DISTRIBUTION TYPE   | MATHEMATICAL RELATIONSHIP     | VISUAL SHAPE & APPRAISAL CAUSE              |
+---------------------+-------------------------------+---------------------------------------------+
| Symmetric (Normal)  | Mean = Median = Mode          | Symmetrical bell; perfectly balanced data.  |
| Positive (Right)    | Mean > Median > Mode          | Tail stretches right; high trophy sales.    |
| Negative (Left)     | Mean < Median < Mode          | Tail stretches left; distressed sales/caps. |
+---------------------+-------------------------------+---------------------------------------------+

1. Positive Skewness (Right-Skewed Distribution)

A distribution is positively skewed when the long tail extends toward the higher values on the right side of the distribution scale:

$\text{Positive Skewness: } \text{Mean} > \text{Median} > \text{Mode}$

• The Real Estate Phenomenon: The vast majority of real estate transaction datasets are positively skewed. Property values and rental rates have a rigid physical lower boundary (they cannot fall below $0.00), but there is theoretically no upper ceiling. A commercial neighborhood may have 20 warehouse sales between $100/SF and $150/SF averaging $125/SF, but two institutional-grade sales at $350/SF pull the arithmetic mean of all 22 sales up to about $145/SF ($3,200 ÷ 22), while the median stays near $125/SF.
• Appraisal Implication: Relying on the mean in a positively skewed market will systematically overvalue typical properties. The appraiser must utilize the median to reflect true market typicality.

2. Negative Skewness (Left-Skewed Distribution)

A distribution is negatively skewed when the long tail extends toward the lower values on the left side of the scale:

$\text{Negative Skewness: } \text{Mean} < \text{Median} < \text{Mode}$

• Real Estate Causes: Negative skewness occurs in markets experiencing sudden distress, where a cluster of liquidations, foreclosure auctions, or distressed short sales sell well below prevailing market prices. It also occurs when analyzing commercial overall capitalization rates ($R_o$) in prime metropolitan cores during cyclical peaks, where a few aggressive institutional acquisitions close at sub-3.5% cap rates while the broader market clusters around 5.5% to 6.0%.

---

4. Practical Commercial Appraisal Applications: Worked Market Rent Study

To understand how measures of central tendency and dispersion operate in appraisal practice, consider an appraiser analyzing contract rental rates for modern flex/industrial warehouse space across a competitive submarket. The appraiser verifies seven recent arm's-length triple-net (NNN) leases per square foot of gross building area:

Sample Leases ($/SF NNN): $12.50, $13.00, $13.50, $14.00, $14.50, $15.00, $24.50

Upon initial inspection, the first six transactions cluster tightly between $12.50 and $15.00/SF, while Transaction 7 ($24.50/SF) represents a specialized lease with substantial tenant improvements (cleanroom/lab buildout).

Step 1: Compute Central Tendency Metrics

1. Sample Size ($n$): $7$
2. Sum of Observations ($\sum x_i$): $\sum x_i = 12.50 + 13.00 + 13.50 + 14.00 + 14.50 + 15.00 + 24.50 = 107.00$
3. Arithmetic Mean ($\bar{x}$): $\bar{x} = \frac{107.00}{7} \approx 15.2857 \approx 15.29 \text{ per SF}$
4. Median (Midpoint):
   • Position: $\frac{7 + 1}{2} = 4\text{th observation}$.
   • Sorted data: $12.50, $13.00, $13.50, $14.00, $14.50, $15.00, $24.50.
   • Median = $14.00/SF.
5. Mode: None (all values appear exactly once).

Step 2: Compute Dispersion Metrics

1. Range: $\text{Range} = \$24.50 - \$12.50 = \$12.00 / \text{SF}$
2. Sum of Squared Deviations ($\sum (x_i - \bar{x})^2$): Using $\bar{x} = 15.2857$:

1. Sample Variance ($s^2$): $s^2 = \frac{103.4284}{7 - 1} = \frac{103.4284}{6} \approx 17.2381$
2. Sample Standard Deviation ($s$): $s = \sqrt{17.2381} \approx \$4.1519 \approx \$4.15 / \text{SF}$

Step 3: Diagnostic Appraisal Synthesis

• Evaluating Skewness: Notice that Mean ($15.29) > Median ($14.00). The distribution is heavily positively skewed due to Transaction 7. If the appraiser concluded market rent based on the arithmetic mean of $15.29/SF, the conclusion would exceed 85% of all actual competitive leases in the submarket!
• Isolating the Outlier: If the appraiser investigates Transaction 7, discovers the specialized cleanroom buildout, and eliminates it as non-comparable to standard flex space, the remaining six transactions yield:
  • $\bar{x}_{\text{revised}}$ = $82.50 / 6 = $13.75/SF
  • $\text{Median}_{\text{revised}}$ = ($13.50 + $14.00) / 2 = $13.75/SF
  • Revised Standard Deviation ($s$) = $0.935/SF With the outlier removed, the mean and median converge at $13.75/SF, and the standard deviation drops from $4.15/SF to $0.94/SF, demonstrating a remarkably tight, cohesive market.