Basic Statistical Measures

Why This Matters

In physical sciences, raw data means nothing until you can describe it. Whether you're analyzing radioactive decay counts, measuring reaction times, or comparing material strengths, statistical measures give you the language to communicate what your data actually shows. You're being tested on your ability to choose the right measure for a given situation—knowing when the mean misleads you, why standard deviation matters more than range, and how distribution shape affects your conclusions.

These measures fall into two fundamental categories: central tendency (where's the middle?) and dispersion (how spread out is it?). A third category—distribution shape—tells you whether your data is symmetric or skewed, which determines which other measures you can trust. Don't just memorize formulas; know when each measure breaks down and what it reveals about your experimental results.

---

Measures of Central Tendency

These statistics answer the question: what's a typical value in my data set? Each measure captures "typical" differently, and choosing the wrong one can misrepresent your results entirely.

Mean

• Calculated as the sum of all values divided by the count—mathematically, $\bar{x} = \frac{1}{n}\sum_{i=1}^{n} x_i$
• Sensitive to outliers, meaning a single extreme measurement can drag the mean away from where most data actually sits
• Best used when data is symmetric and free of extreme values; the go-to measure for normally distributed experimental results

Median

• The middle value when data is ordered—for even-numbered sets, average the two central values
• Robust against outliers, making it the preferred central tendency measure for skewed distributions
• Reveals when mean is misleading; if median differs significantly from mean, your data is likely skewed

Mode

• The most frequently occurring value in a data set—the only central tendency measure applicable to categorical data
• Can be unimodal, bimodal, or multimodal, with multiple modes often indicating distinct subpopulations in your sample
• Useful for identifying clustering in discrete measurements like photon counts or particle detection events

---

Measures of Spread (Dispersion)

Knowing the center isn't enough—you need to quantify how much variation exists. These measures describe whether your data clusters tightly or spreads widely, which directly impacts how confident you can be in your results.

Range

• Simply the maximum minus minimum value—the quickest but crudest measure of spread
• Completely determined by two extreme points, ignoring how the rest of your data is distributed
• Useful for quick assessment but inadequate for serious analysis; always supplement with other dispersion measures

Variance

• The average of squared deviations from the mean—calculated as $\sigma^2 = \frac{1}{n}\sum_{i=1}^{n}(x_i - \bar{x})^2$ for populations (use $n-1$ for samples)
• Squaring eliminates negative deviations and weights larger deviations more heavily than small ones
• Units are squared, making variance difficult to interpret directly but essential for mathematical derivations

Standard Deviation

• The square root of variance, $\sigma = \sqrt{\sigma^2}$, returning dispersion to the original measurement units
• Quantifies uncertainty in measurements—the standard way to report experimental precision in physical sciences
• For normal distributions, approximately 68% of data falls within $\pm 1\sigma$ of the mean, 95% within $\pm 2\sigma$

Quartiles

• Divide ordered data into four equal parts—$Q_1$ (25th percentile), $Q_2$ (median, 50th percentile), and $Q_3$ (75th percentile)
• Provide distribution landmarks without assuming any particular shape, making them useful for any data set
• Essential for box plots and for identifying potential outliers using the 1.5×IQR rule

Interquartile Range (IQR)

• Calculated as $IQR = Q_3 - Q_1$, capturing the spread of the middle 50% of data
• Robust against outliers because it ignores the extreme 25% on each end of the distribution
• Preferred dispersion measure for skewed data when standard deviation would be misleading

---

Measures of Distribution Shape

These statistics describe how your data is distributed, not just where it centers or how much it spreads. Distribution shape determines which other statistical measures and tests are appropriate.

Skewness

• Measures asymmetry around the mean—positive skewness means a longer right tail, negative skewness means a longer left tail
• Indicates mean-median relationship: positively skewed data has mean > median; negatively skewed has mean < median
• Affects which central tendency measure to trust; high skewness is a red flag that the mean may misrepresent typical values

Kurtosis

• Measures "tailedness" or peak sharpness relative to a normal distribution—not simply how peaked the center is
• High kurtosis (leptokurtic) indicates heavy tails with more outliers than expected; low kurtosis (platykurtic) indicates light tails
• Critical for risk assessment in physical sciences when extreme values have significant consequences (equipment failure, safety limits)

---

Quick Reference Table

---

Self-Check Questions

1. You measure the masses of 20 samples, but one sample was contaminated and has an extremely high mass. Which central tendency measure should you report, and why?

2. Compare and contrast variance and standard deviation. Why do physical scientists typically report standard deviation rather than variance when describing measurement uncertainty?

3. A data set has a mean of 45 and a median of 52. What does this tell you about the skewness of the distribution, and which measure better represents a "typical" value?

4. Which two measures of spread are most robust against outliers? Explain why this robustness matters when analyzing experimental data with potential measurement errors.

5. If an FRQ asks you to describe both the center and spread of a non-symmetric data set, which four statistics would you calculate and report? Justify your choices.