Interval Scale — Definition & Example
An interval scale is a way of measuring where equal steps on the scale mean equal amounts, but zero does not mean 'none.' Temperature in degrees Fahrenheit is the classic example. You can subtract interval values, but you cannot say one is 'twice' another.
An interval scale is a level of measurement whose values are ordered, whose equal numerical differences represent equal differences in the attribute, and whose zero point is arbitrary. Such a scale is unique only up to a positive linear transformation with . As a result, differences and ratios of differences are meaningful. Ratios of the values themselves are not.
Worked Example
Problem: The Declaration of Independence was signed in 1776 and the U.S. Constitution in 1787. Calendar years are on an interval scale. Which comparison makes sense: "11 years apart" or "1787 is 1.006 times 1776"?
Step 1: Find the difference. Differences are meaningful on an interval scale.
Step 2: Measure the same two years from a different starting point, 1700. The calendar's starting point is a convention, not "no time", so any start is just as valid.
Step 3: Compare the ratios on the two counts. The difference stays 11, but the ratio changes with the starting point, so the ratio means nothing.
Answer: "11 years apart" makes sense; "1.006 times" does not. On an interval scale differences keep their meaning when the zero point moves, but ratios do not.
Why It Matters
In intro statistics, the level of measurement tells you which summaries are valid. For interval data such as temperatures or calendar years, the mean and standard deviation are fine. Statements like 'percent more' and the coefficient of variation are not, because those need a true zero (a ratio scale).
Common Mistakes
Mistake: Treating 0 on an interval scale as 'none' and comparing values by division. A typical error is saying 80°F is twice as hot as 40°F.
Correction: 0°F is just a point on the scale, not an absence of heat. Compare interval values by subtraction (80°F is 40°F warmer). To use ratios, you need a scale with a true zero, such as Kelvin.
