Mathwords logoMathwords

Validity — Definition & Examples

Validity has two meanings. In logic, an argument is valid when its conclusion must be true whenever its premises are true. In statistics, validity is how well a study or measurement actually supports the conclusion you draw from it: a valid test measures what it claims to measure, and a valid study's conclusions are justified by how the data were collected.

**Logical validity:** an argument is valid if it is impossible for all of its premises to be true while its conclusion is false. Validity depends only on the argument's form, not on whether the premises are actually true; a valid argument whose premises are true is called sound. For example, "All squares are rectangles. All rectangles have four sides. So all squares have four sides" is valid. **Validity in research and measurement** (the general sense, used throughout statistics) is the degree to which evidence supports a specific interpretation of results. **Measurement (test) validity** is the extent to which an instrument measures the intended construct without systematic error (bias). **Internal validity** is the extent to which an observed difference in the response variable can be attributed to the explanatory variable rather than to confounding variables. Random assignment is the main tool for securing it. **External validity** is the extent to which results generalize to a larger population or other settings. Random sampling from that population is the main tool for securing it.

How It Works

To check a logical argument, suppose every premise is true and ask whether the conclusion could still be false. If it cannot, the argument is valid. To check a study, ask three questions. First, does the measurement capture the intended quantity, or is it systematically off? Second, could something other than the treatment explain the difference between groups? Third, who do the results apply to? In AP Statistics, random assignment justifies cause-and-effect claims (internal validity). Random sampling justifies generalizing to the population (external validity). A study can have one without the other.

Example

Problem: 200 student volunteers at one high school are randomly assigned to two groups of 100. Group A uses Study App A and Group B uses Study App B for a month. The mean test score is 82 for Group A and 76 for Group B, and the difference is statistically significant. Evaluate the internal and external validity of the conclusion "App A causes higher scores."
Step 1: Find the observed difference in mean scores.
xˉA−xˉB=82−76=6 points\bar{x}_A - \bar{x}_B = 82 - 76 = 6 \text{ points}
Step 2: Check internal validity. The students were randomly assigned to apps, so other factors such as motivation and prior grades should be roughly balanced between the groups. The difference can reasonably be attributed to the app.
Step 3: Check external validity. The students were volunteers from a single school, not a random sample of any larger population. Volunteers may be more motivated than typical students, so the results may not apply to students in general.
Answer: Internal validity is strong: App A caused about a 6-point higher mean score for these volunteers. External validity is limited: you can generalize only to students similar to these volunteers, not to all high school students.

Another Example

Problem: Decide whether each argument is valid, and whether it is sound. (a) If a number is divisible by 4, it is even. 12 is divisible by 4. So 12 is even. (b) If a number is divisible by 4, it is even. 10 is even. So 10 is divisible by 4.
Step 1: Argument (a) has the form "If PP, then QQ. PP. So QQ" (modus ponens). Whenever both premises are true, the conclusion must be true, so it is valid.
Step 2: Both premises of (a) are true: every multiple of 4 is even, and 12=4×312 = 4 \times 3. A valid argument with true premises is sound.
Step 3: Argument (b) has the form "If PP, then QQ. QQ. So PP" (affirming the consequent). Its premises are true, but its conclusion is false, because 10 is even and 10÷4=2.510 \div 4 = 2.5. True premises with a false conclusion show the form is invalid.
Answer: Argument (a) is valid and sound. Argument (b) is invalid, so it cannot be sound, even though both of its premises are true.

Why It Matters

AP Statistics free-response questions often ask you to state what conclusions a study supports. Full credit requires linking random assignment to causation and random sampling to generalization. Researchers in medicine and psychology build clinical trials and surveys around these same validity checks before any result is published.

Common Mistakes

Mistake: Calling an argument invalid because one of its premises is false.
Correction: Validity is about form: if the premises were true, would the conclusion have to be true? "All fish can fly. A shark is a fish. So a shark can fly" is valid but not sound, because its first premise is false.
Mistake: Treating reliability and validity as the same thing.
Correction: Reliability means consistent results. Validity means correct results for the intended purpose. A scale that always reads 2 lb high is reliable but not valid. Check for systematic bias, not just consistency.
Mistake: Saying random sampling allows cause-and-effect conclusions, or that random assignment lets you generalize to a population.
Correction: These are swapped. Random assignment supports causation (internal validity). Random sampling supports generalization (external validity). Name the specific random process that justifies each claim.

Related Terms